See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/301228848

Si-SiC based materials obtained by infiltration of silicon: study and

applications

Thesis · February 2015

DOI: 10.13140/RG.2.1.1617.1920

CITATION READS

1 7,310

1 author:

Giuseppe Claudio D'Amico

Politecnico di Torino
7 PUBLICATIONS 77 CITATIONS

SEE PROFILE

All content following this page was uploaded by Giuseppe Claudio D'Amico on 13 April 2016.

The user has requested enhancement of the downloaded file.

 Politecnico di Torino
Dipartimento Scienza Applicata e Tecnologia (DISAT)

Politecnico di Torino
DISAT (Department of Applied Science and Technology)

February 2015

PhD Thesis
in Material Science

Si-SiC based materials obtained by

infiltration of silicon: study and applications

D’Amico Giuseppe Claudio

Mentors: Claudio Badini (Full Professor), Fino Paolo (Full Professor) and
Biamino Sara (Associate Professor)

Department of Applied Science and Technology

Corso Duca degli Abruzzi 24, 10129 Torino, Italy
2
 Dedications

At the end of my thesis work I want to thank my tutors for helping and supporting me
during my research: Fino Paolo, Biamino Sara and Badini Claudio. I also want to thank
the following professors for their availability: Pavese Matteo, Deorsola Fabio, Milanese
Daniel and Ortona Alberto.
Thanks also goes to all the technicians who assisted me during these years: Fuso Laura,
Leo Maurizio, Raimondo Mauro, Favero Antonio, Aghem Giuseppe and everyone I have
forgotten.
Furthermore I would to thanks my former office-colleagues: Silvia, Monica, Cristina,
Clara, Ehsan, Tiziana, Ratna and Marco.
In this experience I have worked alongside many group-colleagues, that I can list here like
old colleagues (many of whom no longer work at the Polytechnic): Simona, Paolo E.,
Andrea A., Silvia, Emiliano, Dreidy, Wenshu, Xiang, Ramzan, Rao, Azhar, Simone,
Lorenzo, Veronica, Mathieu, Andrea C. (the Master), Andrea R. and Eleonora;
and new ones: Giorgio, Giulio, Alberta, Karol, Luca, Giovanna, Federico, Abdollah,
Oxana and Andrea C. All of them, who more (the majority) who less, left me something.
A special thought goes to Elisa, my actual office-colleague and mate of this Ph.D.
Last, but not least, I wish to say a big thank you to my parents and to all my friends: each
one of them knows why.

3
4
Table of contents
Preface 9
1. UHTCs (Ultra High Temperature Ceramics) ........................................................ 11
1.1 Introduction .......................................................................................................... 11
1.2 Bonding and structure of UHTCs ........................................................................ 15
1.3 Synthesis and densification methods ................................................................... 21
1.3.1 Hot pressing (HP) ......................................................................................... 23
1.3.2 Spark plasma sintering (SPS) ....................................................................... 25
1.3.3 Self propagating high temperature synthesis (SHS) ..................................... 28
1.3.4 Reactive hot pressing (RHP) ........................................................................ 29
1.3.5 Pressureless sintering (PS)............................................................................ 30
1.4 Properties ............................................................................................................. 33
1.4.1 Thermal properties ........................................................................................ 33
1.4.2 Mechanical behavior ..................................................................................... 36
1.5 Oxidation of transition metal diborides ............................................................... 39
1.5.1 Oxidation of diborides ceramics containing SiC .......................................... 43
1.5.2 Additives ....................................................................................................... 45
1.6 Siliconized silicon carbide (Si-SiC) ..................................................................... 51
1.6.1 Infiltration theory .......................................................................................... 52
1.7 References ............................................................................................................ 61
2. Si-SiC-ZrB2 ceramics by silicon reactive infiltration ............................................ 68
2.1 Introduction .......................................................................................................... 68
2.2 Experimental procedure ....................................................................................... 69
2.2.1 Materials ....................................................................................................... 70
2.2.2 Green forming............................................................................................... 71
2.2.3 Pyrolysis ....................................................................................................... 71
2.2.4 Silicon reactive infiltration ........................................................................... 72
2.2.5 Machining ..................................................................................................... 72
2.2.6 High temperature oxidation .......................................................................... 73
2.2.7 Characterisation ............................................................................................ 73
2.3 Results and discussion ......................................................................................... 73
2.3.1 Microstructure .............................................................................................. 74
2.3.2 Oxidation behavior ....................................................................................... 78
2.4 Conclusions .......................................................................................................... 80
2.5 References ............................................................................................................ 81

5
3. Short-term and long-term oxidation behavior of Si-SiC-ZrB2 composites
produced by SRI ............................................................................................................... 83
3.1 Introduction ......................................................................................................... 83
3.2 Experimental procedure ....................................................................................... 85
3.2.1 Materials ....................................................................................................... 85
3.2.2 Sample preparation ....................................................................................... 86
3.2.3 Short-term oxidation treatment .................................................................... 87
3.2.4 Long-term oxidation treatment..................................................................... 87
3.2.5 Characterisation methods ............................................................................. 88
3.3 Results and discussion ......................................................................................... 89
3.3.1 As produced composites............................................................................... 89
3.3.2 Oxidation behavior ....................................................................................... 93
3.3.3 Oxidation behavior after short-term oxidation at 1600°C............................ 97
3.3.4 Oxidation behavior after long-term oxidation at 1500°C........................... 100
3.4 Conclusions ....................................................................................................... 111
3.5 References ......................................................................................................... 113
4. Cellular ceramics .................................................................................................... 115
4.1 Introduction ....................................................................................................... 115
4.2 Replica technique .............................................................................................. 117
4.2.1 Synthetic templates .................................................................................... 118
4.2.2 Natural templates........................................................................................ 121
4.3 Sacrificial template method ............................................................................... 123
4.4 Direct foaming methods .................................................................................... 127
4.4.1 Stabilization with surfactants ..................................................................... 128
4.4.2 Stabilization with particles ......................................................................... 131
4.5 High temperature applications of Si-SiC cellular ceramics............................... 133
4.5.1 Introduction ................................................................................................ 133
4.5.2 Porous burners ............................................................................................ 137
4.5.3 Catalyst carrier ........................................................................................... 139
4.5.4 Structural application ................................................................................. 140
4.5.5 Solar radiation absorbers ............................................................................ 141
4.5.6 Electrical applications ................................................................................ 142
4.5.7 Reformers ................................................................................................... 143
4.6 References ......................................................................................................... 145
5. Characterisation of ageing Si–SiC foams in porous burners ............................. 150
5.1 Introduction ....................................................................................................... 150
5.2 Experimental procedure ..................................................................................... 152

6
 5.2.1 Aging campaign .......................................................................................... 153
5.3 Characterisation methods ................................................................................... 154
5.3.1 Bending and compression tests................................................................... 154
5.3.2 Morphological characterisation .................................................................. 155
5.3.3 Thermal characterisation ............................................................................ 155
5.4 Results and discussion ....................................................................................... 155
5.5 Conclusion ......................................................................................................... 161
5.6 References .......................................................................................................... 162
6. Si-SiC heat exchangers for recuperative gas burnerns with highly structured
surface elements: CEREXPRO project ........................................................................ 163
6.1 Introduction ........................................................................................................ 163
6.2 Design ................................................................................................................ 172
6.3 Processing .......................................................................................................... 176
6.3.1 Textile structure assembly .......................................................................... 176
6.3.2 Ceramisation process .................................................................................. 180
6.3.3 Ceramic characterisation ............................................................................ 187
6.4 Prototype ............................................................................................................ 202
6.4.1 Manufacturing ............................................................................................ 202
6.4.2 Field tests .................................................................................................... 207
6.4.3 Comparison with the state of the art ........................................................... 213
6.4.4 Alternative designs ..................................................................................... 218
6.5 References .......................................................................................................... 223
7. List of author’s publications .................................................................................. 224

7
8
Preface

Many efforts and investments have been made in the last decades in order to obtain
materials able to withstand extreme conditions in several technological fields such as
future military, industrial and space based projects. The present PhD thesis addresses this
problem dealing Ultra High Temperature Ceramics (UHTCs) and in particular focusing on
Si-SiC-ZrB2 composites obtained with a technique named silicon reactive infiltration
(SRI).
Each major chapter of the present thesis relates to an article either published or in press on
the journals indicated at the end in the seventh chapter (List of author’s publications).
The sequence of these chapters is obviously not random, but trying to follow the
streamline that leads from the physico-chemical and technological properties of the
material to its applications, in my case this is meant its use in industrial burners within an
european project: CEREXPRO, that was financially supported by the European
Commission in the FP7 (7th Framework Programme).
The reader is introduced to the current state of the art of UHTC (Chapter 1) having a
historical review of UHTC research carried out thus far. Since the most comprehensive
research has been focused on compositions containing HfB2 or ZrB2, they are the primary
materials discussed. The chapter concludes by illustrating a particular type of UHTC: the
siliconized silicon carbide (Si-SiC) and the infiltration method used to obtain it.
In the second chapter a methodology to produce Si-SiC-ZrB2 bulk ceramics, taking
advantage of the reactive infiltration technique, is proposed. Preforms with different
contents of SiC and ZrB2 were compacted with phenolic resin, pyrolysed and finally
infiltrated with molten silicon at temperature above its melting point.
A detailed study of the mutual interactions between the Si-SiC-ZrB2 constituent materials
during their processing and of their chemical activity with oxygen at high temperatures in
the prospect of using this material as a matrix of a continuous fiber reinforced composite
is addressed in Chapter 3. The study of the early stage of oxidation is the first part of this
chapter, while the second part is focused on the long-term oxidation of some selected
samples at a high temperature.
The aim of chapter four is to give an overview of the processing routes currently available
for the preparation of macro-porous ceramics, with particular emphasis on the processing–
microstructure–property relations inherent to each process. At the end, the chapter

9
presents several high temperature applications in which Si-SiC foams are currently applied
because of their superior thermo-mechanical properties. The Chapter 5, in fact, aims to
track the changes of Si-SiC foam properties during the first part of a long time aging
condition, typical of industrial burner applications.
The project “CEREXPRO” (CERamic heat EXchangers with enhanced materials
PROperties), discussed in the Chapter 6, was aimed to develop a new generation of
ceramic heat exchangers for high temperature heat recovery with the target of significantly
reducing the size and weight as well as also the price of such components by simplifying
the manufacturing process and allowing a higher flexibility in the heat exchanger
geometry. The use of precursors/template materials taken from the textile industries and a
subsequent ceramic conversion, via liquid silicon infiltration, is proposed as the main
technological path for reaching the above objectives. The main role of the Polytechnic
within this project was the characterization process of materials.
I hope you will enjoy reading this work. Please contact me for any inquiry at
giuseppe.damico@polito.it.
Torino, November 2014.
Giuseppe Claudio D’Amico

10
1. UHTCs (Ultra High Temperature Ceramics)

1.1 Introduction

Refractory compounds such as ceramic carbides, borides and nitrides are characterized by
high melting points, high hardness and good chemical inertness and oxidation resistance
[1-3]. These refractory compounds have been broadly termed Ultra High Temperature
Ceramics (UHTCs). Strong covalent bonding is responsible for the high melting points,
moduli and hardness of the UHTC family of materials [4, 5]. High negative free energies
of formation also give UHTCs excellent chemical and thermal stability under many
conditions [4, 5].
The bibliographical search with the use of the SciFinder (CAS) system based on keywords
that characterize this field of research made it possible to obtain a distribution of the
frequency of references to different chemical compounds (Fig. 1.1). As can be seen, the
most frequently mentioned compounds are ZrB2 and HfB2, as well as SiC, which are the
basic components of ultra-high-temperature ceramics (UHTCs).

Figure 1.1: Distribution (%) of the publications concerning the ultra-high temperature
materials indexed over the compounds indicated (according to the data of SciFinder, CAS)
[6].

11
Development of non-oxide ceramics has lagged behind that of oxides because they are
rarely found in nature, need to be processed in controlled reducing atmospheres and tend
to oxidize in air even at room temperature, especially when in fine powder form.
While UHTC compounds have been studied for over 100 years [7], historically, this
family of ceramic materials was first significant processing-structure-property investigated
between the 1950s and 1970s by Russian and U.S. laboratories [2, 8, 9]. At present, most
of the structural materials for use in high temperature oxidizing environments are limited
to SiC and Si3N4 based materials, oxide ceramics and C/C composites with thermal
protection. Silicon-based ceramics and protected C/C composites exhibit good oxidation
resistance, but only up to ~ 1600°C, and their thermal cycling lifetimes are modest.
The development of structural materials for use in oxidizing and rapid heating
environments at temperatures above 1600°C is therefore of great importance for
engineering.
Some of the earliest and most thorough work was performed in the 1960’s by ManLabs
under a research program funded by the Air Force Materials Laboratory (AFML) [1, 2].
Work on these materials was initiated to meet the need for high temperature materials that
would enable the development of maneuverable hypersonic flight vehicles with sharp
leading edges [2]. Around the same time NASA was also investigating high temperature
materials to meet its own needs [3].
At present, many groups in the U.S., Japan, China, India and in Europe (especially in
Italy) are studying UHTC systems to improve resistance to oxidation [10-13].
The need for high temperature materials that can operate with no or limited oxidation or
ablation at temperatures greater than 3000K has driven the development of UHTC
materials. The potential applications for UHTCs span a wide number of needs arising
from future military, industrial and space based projects. Potential industrial applications
for UHTCs include use in foundry or refractory processing of materials. Their chemical
inertness makes them ideal for molten metal crucibles, thermo-well tubes for steel
refining, cutting tools, furnace elements and as parts for electrical devices such as heaters
and igniters [14-20].
The military and aerospace applications for UHTCs range from rocket nozzle inserts and
air augmented propulsion system components to leading edges and nose caps for future
hypersonic reentry vehicles [14, 21-23]. Early space vehicle designs, such as the space
shuttle, were designed with a large radius, blunt body design to reduce aero
thermodynamic heating to maintain moderate temperature limits on all parts of the

12
vehicle. However, the larger the leading edge radii, higher the vehicles drag, which
reduces maneuverability and cross range during reentry. Increased maneuverability
requires laminar flow across the control surfaces, which, in turn, necessitates the use of
sharp leading edges; therefore, to improve performance, hypervelocity vehicle concepts
have proposed using slender aerodynamic shapes with sharp leading edges [24, 25].
Development of sharp body vehicles increases the lift-to-drag ratio thereby improving the
vehicles’ reentry cross range. A higher lift to drag ratio also has the potential to improve
the overall vehicle system safety in a number of ways. However, the temperature of the
leading edge is inversely proportional to the square root of the leading edge radius, i.e. as
the leading edge radius decreases the temperature increases [25]. Therefore the successful
design of a sharp hypersonic vehicle requires the development of new materials with
higher temperature capabilities than the current state-of-the-art materials can provide.
Ultra High Temperature Ceramics are a family of materials that are promising candidates
for meeting such requirements [24, 25].
In comparison to carbides and nitrides, the diborides tend to have higher thermal
conductivity, which gives them good thermal shock resistance and makes them ideal for
many high temperature thermal applications [5, 26]. The relatively good oxidation
resistance of refractory diboride compounds, compared to other refractory intermetallic
compounds (i.e. carbides, nitrides and silicides), has focused many research efforts into
detailed investigations of transition metal diborides of groups IV and V (Ti, Zr, Hf, Nb,
Ta). Of the transition metal diborides, hafnium diboride (HfB2) and zirconium diboride
(ZrB2) were identified as the most promising candidates for high temperature applications.
In particular, ZrB2 has the lowest theoretical density among the ultra-high temperature
ceramics, which makes it an attractive material for aerospace applications [21, 27, 28].
However, the use of the single-phase material for high-temperature structural applications
is limited by its poor oxidation and ablation resistance, as well as its poor damage
tolerance.
The composite approach has been successfully adopted in order to improve the
densification, mechanical properties, physical properties, as well as the oxidation and
ablation resistance of the ZrB2 ceramics [29-40]. Obviously, the mechanical and physical
properties of the ZrB2-based composites are closely linked with the densification process,
compositions, starting powder, microstructure and intergranular second phase.
Of large importance is also the fact that upon the oxidation of ZrB2 and HfB2 there arise,
apart from products that are liquid at temperatures higher than 8001200°C (forming

13
borosilicate glass, which impedes rapid diffusion of oxygen into the bulk of the material),
also refractory oxides (Tm ~ 2700°C for ZrO2 [6], 28002850°C for HfO2 [6]) possessing
a low vapor pressure (MO (g) at T = 2160°C is ~ 2.88*10-7 atm for ZrO2; ~ 1.15*10-7 atm
for HfO2 [6]) and a sufficiently high (for the expected operating conditions) mechanical
strength; this must ensure a maximally possible retention of the geometry of an article,
which influences its aerodynamic characteristics.
In the field of designing UHT materials, two directions, which are substantially different
in, first of all, the priority purposes, can be distinguished. The first of these is connected
with the modification of classical UHT materials (Cf/C and Cf/SiC composites) by
refractory compounds, such as ceramic matrices and/or protective (antioxidant) coatings,
with the purpose to significantly increase their oxidation resistance without a significant
loss of the excellent thermo-mechanical characteristics of the base materials [6].
An alternative direction is the creation of purely ceramic almost poreless materials
(possibly, with a certain their strengthening by the introduction of nanostructured
components [6] or of short fibers [6]) on the basis of substances with a high thermal
conductivity. This will allow an efficient heat removal from the regions adjacent to the
points or lines of the total flow stagnation, which will work as separate segments of a
complex composite detail (Fig. 1.2) [6], whose different parts will bear different types of
load-thermal, oxidation, shock, shear, etc. An additional argument in favor of the selection
of the design of the thermo-loaded parts of the flight vehicles in the form of an assembly
of several segments is that the UHTCs are brittle ceramic materials (the probability of
their destruction can be described in terms of the Weibull statistics); therefore, from the
viewpoint of repairability, and also to make the manufacturing easier, structures that have
a limited size or volume of components (like tiles or segments rather than a one-piece
continuous edge) are preferable and, thus, have a reduced probability of the appearance of
defects, which can decrease the strength.

Figure 1.2: Construction of the leading edge of a wing prepared from a UHTC [6].

14
As a matter of fact, the best way to use monolithic UHTC materials for manufacturing
sharp components seems to be to combine these materials with others (CMC notably).
Thus, the UHTC part would be placed as an insert in front of the structural component in
order to use only UHTC pieces of limited size. This allows for optimization of the element
involved (performance, weight and cost in particular). However, it is important to note
that the attachment design needs to be properly addressed to minimize thermal stresses.
For example, it would be possible to avoid the contact of the UHTC element with cooler
sub-components in the structure in order to reduce bearing stresses at the attachment
location.
Clearly, the potential applications for UHTCs span a wider number of needs arising from
future military, industrial and space based projects. While continued work has provided
valuable insight on the performance of UHTCs consistent ongoing research efforts are
required if future applications are to be realized.
Realization of the full potential of UHTCs as future high temperature materials requires a
thorough knowledge of their properties and behavior under diverse conditions of
temperature, environment and stress states [5]. The primary purpose of this chapter is to
provide a historical review of UHTC research carried out thus far and, where available,
to provide engineering data on UHTCs. Since the most comprehensive research has been
focused on compositions containing HfB2 or ZrB2, they will be the primary materials
discussed. Results from materials containing HfB2 or ZrB2 as the principal component
with selected additives designed to enhance one or more of the following: oxidation
resistance, mechanical properties and thermal stress resistance will also be presented.
When available, data on other UHTC compositions will also be discussed. The chapter
concludes by illustrating a particular type of UHTC: the siliconized silicon carbide
(Si-SiC) and the infiltration method used to obtain it.

1.2 Bonding and structure of UHTCs

Ceramic borides, carbides and nitrides all have very strong chemical bonds that give them
high temperature structural stability [5, 41-43]. As a result of the extremely strong bond
between carbon atoms, carbides follow the classical definition of a brittle ceramic.
They come in three general classes: ionic, covalent and interstitial. None of the ionic

15
carbides have engineering uses because of their extreme brittleness [41]. The two covalent
carbides of major importance, SiC and B4C, both are valuable for their extreme hardness
as well as excellent thermal and chemical stability. The largest class of carbides, those of
the interstitial type, includes carbides of the metals Hf, Zr, Ti and Ta. These materials
benefit from strong carbon networks and have some of the highest melting points of
known materials. They are also known to have high strengths at elevated temperatures [41,
42]. From this standpoint, carbides offer a tremendous benefit in many engineering
applications. Unfortunately, these materials are hard to fabricate because of their
refractoriness, and little has been achieved beyond the laboratory scale [41, 42].
Ceramic nitrides have many of the same properties as carbides and nitrides are also
difficult to fabricate, especially in the pure form, due to strong covalent bonding. Silicon
nitride and boron nitride are the primary materials in the family of nitrides to be developed
for engineering applications [43].
Then there are the ceramic borides, such as HfB2 and ZrB2. These materials also benefit
from very strong bonding between boron atoms, although their bonding is not typically as
strong as seen in the carbides thus these materials often have melting points below that of
the carbides [5, 41]. A unique feature of the electronic nature of the boron bonding in
these materials results in the borides having high thermal and electrical conductivities,
higher than typically found in carbides and nitrides, as well as low coefficients of thermal
expansion which combined gives the borides relatively good thermal shock resistance,
for ceramics [5].
Table 1.I lists a number of metallic elements that form binary diboride compounds, with
the AlB2-type structure, shown in figure 1.3 [44, 45]. The AlB2 structure contains
graphite-like layers of boron separated by hexagonal close-packed (h.c.p.) layers of metal
atoms. The diborides are comprised of rigid covalent boron lattices, such that the boron
atoms have a trigonal prismatic metal environment with three close boron neighbors.
The metal atoms coordinate twelve boron atoms, six metallic atoms in the same layer and
two metal atoms in the two adjacent layers (top and bottom) [44, 45]. The boron nets have
very strong covalent bonds that hinder an increase in the ao direction, though no such
hindrance occurs in the co direction, giving borides the ability to accommodate a wide
variety of metals [44].

16
 Table 1.I: UHTCs metallic elements [46].

Figure 1.3: A) Atomic projections of the AlB2-type structure showing top down and side
view. B) An illustration of metal atom deformation within the AlB2-type structure [44].

Spear calculated three chemical bonding parameters for borides of the AlB2-type, plotted
in figure 1.4 [44]. The parameter in the top graph is the ratio of the two elliptic axes of the
metal atom and provides a measure of the amount of metal atom distortion.
The middle parameter is a ratio of the volume of the elemental metal atom to the metal
atom in the diboride. Ratios less than unity indicate the metal is too small for the space,
while ratios greater than one indicate the metal atom is compressed in the structure.
Ratios different from unity suggest M-M and possibly M-B bonds will have less than
optimum strength. Finally, the parameter on the bottom is a measure of B-B bond
strength; it is a ratio of the average minimum free boron radius to that of boron in the

17
diboride. Smaller values are a result of stretched bonds and smaller bond energies [44].
From these calculations, Spear deduced that M-B bonding is likely the leading contributor
to the structural integrity of AlB2 type borides, more so than M-M or B-B bonding.
Typically, the less distortion there is to the unit cell the stronger are the bonds that hold it
together. As the degree of bonding increases so does the melting point, modulus and
hardness, in the diborides as well as for carbide and nitride ceramics [5, 26, 41].

Figure 1.4: A) View of the (110) plane in a diboride structure, illustrating the geometrical
crystal parameters as calculated by Spear. B) Table of bonding parameters as calculated
from geometrical parameters in A), metals are grouped according to their period in the
periodic chart [44].

Enthalpies of formation for several boride systems were reported by Samsonov and
Vinitiskii [47]. From those results it is clear that the stability of diborides decrease in the
order HfB2 > TiB2 > ZrB2 >> TaB2 > NbB2 >VB2, with VB2 the least stable of the
diborides, having the lowest energy of formation. To understand this trend, Guillermet and
Grimvall showed that the cohesive properties of these materials could be described by the
filling of electron bands, as shown by a plot of density of electron states (bonding or anti-
bonding states), N(E), versus the respective energy of the system, figure 1.5 [48].
Using ZrB2 as an example, the peaks P1 and P2 arise from 2s and 2p electron orbital
hybridization with zirconium d orbital electrons. The peaks P3 and P4 represent bonding
and then anti-bonding states between boron 2p and zirconium 4d electrons [5].

18
Hybridization thus reduces the strength of B-B bonds but creates strong M-B bonds as
predicted by Spear in his analysis. Because the Fermi level, EF: (energy level of the
highest filled band at 0K) for ZrB2 falls between the bonding and anti-bonding peaks,
it can be seen that any increase in electron density would reduce bond strength by further
filling of the anti-bonding states; similarly, any reduction in electron density lowers the
number of electrons filling the bonding levels, also reducing bond strength. With those
results in mind, the trend in boride stability is clearly shown when looking at a plot of
standard entropy, ES, vs. number of electrons per unit cell, ne, see figure 1.6(left) [48].
A plot of ES vs. Tf, melting temperature (Fig. 1.6(right)) shows the same trends as those
illustrated in the plot of ES vs. ne [48]. Those materials, with their standard entropy or
cohesive energy at a maximum, show the highest stabilities and melting points. Simply
put, the materials with a Fermi energy between P3 and P4 have the highest stabilities.
Listed in in order of bond strength, those species are materials formed with Ti, Zr and Hf.
As the number of electrons, ne, increases (i.e. V, Nb, Ta) more anti-bonding states are
filled which decreases the bond strength and the melting temperature of the compound.
Similarly, as ne decreases (i.e. Sc, Y), fewer electrons are available to fill the bonding
levels and compound stability decreases, also demonstrated by a reduction of the melting
temperature, as shown in figure 1.6(right) [48].

Figure 1.5: Plot of the general features of the

electronic density of the states, N(E) number
of electrons in that state (bonding or anti-
bonding) versus energy of the system, for an
AlB2-type compound [48].

19
Figure 1.6: Left, plot of characteristic energy, defined from entropy related Debye
temperature and plotted vs. the average number of valence electrons. Right, characteristic
energy plotted vs. melting temperature [48].

Following the previous description of the atomic structure of boride, carbide and nitride
ceramics, table 1.II lists physical crystalline structural differences of a variety of UHTCs
along with respective density and melting point [46]. Note that density increases with
increasing mass of the metal atom. Note also the differences in melting points between
materials whereby the carbides typically have the highest melting points, above borides or
nitrides of the same metal constituent.

Table 1.II [46].

20
Some of the thermodynamic properties for a few diborides, including HfB2 and ZrB2 are
listed in tables 1.III and 1.IV. The data in these tables are from Pankratz et al. who
reviewed available data in the early 80’s and included only the data they deemed to be
reliable [49]. Enthalpies of formation are strongly correlated with Gibbs free energies of
formation for borides, because the entropy terms are small. This also means that the free
energy is relatively insensitive to temperature [5].

Table 1.III [49].

Table 1.IV [49].

1.3 Synthesis and densification methods

The conventional method for densifying UHTCs is hot pressing, performed either with or
without sintering additives. Significant research has examined alternative processing
routes for UHTCs to reduce processing times and temperatures and, therefore, reduce the
cost associated with the techniques. Some of these alternative techniques also improve the
oxidation resistance of the material.
The presence of secondary phases in the microstructure has a detrimental effect on the
material high temperature capability by introducing grain boundary phases, which can

21
have lower melting temperatures and provide routes for oxygen diffusion into the material.
The presence of oxygen impurities can be detrimental to the densification ability of the
starting powder or cause rapid grain growth; it also contributes to the formation of
secondary phases. Nitrides and reducing additives have been added to the powders to
enhance sinterability [50-53], but these additives introduce secondary phases to the
material.
The principal source of elemental zirconium metal, in particular, is the zirconium silicate
mineral, zircon (ZrSiO4). The mineral baddeleyite, a natural form of zirconia (ZrO2), is a
secondary source of zirconium. Zircon is also the primary source of all elemental hafnium,
which is contained in zircon at a ratio of about 1 part hafnium to 50 parts zirconium [54].
During diboride production, the respective metal oxide is used as the metal source while
boron carbide, crystalline boron or a combination of boron oxide and carbon acts as the
boron source. Due to the limited demand for diboride powders of Zr and Hf, the processes
for manufacturing these powders have not been refined to the same degree as for other
common ceramics, such as SiC or Si3N4.
The main production techniques used for formation of UHTC materials are as follows:
1. Hot pressing (HP)
2. Spark-plasma sintering (SPS)
3. Self-propagating high-temperature synthesis (SHS)
4. Reactive hot pressing (RHP)
5. Pressureless sintering (PS)
Furthermore, methods of polymer impregnation and pyrolysis (PIP) are employed,
although on a substantially smaller scale (mainly for creating refractory protective
matrices and coatings on Cf/C and Cf/SiC composites), as well as chemical vapor
deposition (CVD) and often also combinations of these approaches. For the efficient
application of these methods, the development of precursors and starting reagents with
given properties is necessary [6].
Due to the hard brittle nature of the UHTC materials, diamond tooling is typically required
to machine components. Electron Discharge Machining (EDM) can be used as an
alternative to diamond grinding to machine the borides due to the high electrical
conductivity of this materials. This is an advantage over some of the carbide, nitride and
oxide materials that typically have higher resistances, which does not allow them to be
machined using EDM.

22
1.3.1 Hot pressing (HP)

The densification of ZrB2 powder generally requires very high temperatures [55], owing to
the covalent character of the bonding as well as to its low volume and grain boundary
diffusion rates. Typically, HP of ZrB2 required a temperature of 2100°C or above and
moderate pressure (2030 MPa) [56-58], or lower temperatures ( 1800°C) and extremely
high pressures (> 800 MPa) [59, 60]. These studies [56-60] showed that densification of
ZrB2 is a diffusion-controlled rate process. Results of HP studies on commercially
available ZrB2 powders are summarized in table 1.V, which includes details of the starting
powders, sintering additives, HP conditions, mixing method and final densities. It has
been found that HP of coarse ZrB2 powder (d ≈ 20 m) at 2000°C with a pressure of
20 MPa achieved only a relative density of  73% [57], whereas the relative density
of  91% was obtained for a finer ZrB2 powder (d ≈ 2.1 m) under the same HP
conditions [58]. Furthermore, the attrition-milled ZrB2 powder, with average particle size
of d ≤ 0.5 m, required HP at 1900°C and 32 MPa for 45 min to achieve full density [30].
The lower HP temperature was attributed to reduction of starting particle size from
microns (d ≈ 2.1 m) to submicrons (d ≤ 0.5 m) by attrition-milling.

Table 1.V: Starting powder size, hot pressing conditions and final densities of the hot-
pressed ZrB2 ceramics and ZrB2-based composites with various additives [61].

23
Oxygen impurities (B2O3 and ZrO2) present on the starting powder surfaces have been
shown to inhibit densification and to promote grain growth in the non-oxide ceramic
systems. Recently, metallic Ni [22, 62], SiC [29-35, 58], Si3N4 [52, 63], AlN [51, 64],
HfN or ZrN [50, 65] have been added to ZrB2, producing an intergranular secondary phase
and/or reducing oxygen content, both of which assists in the densification of ZrB2. Silicon
carbide is the most common additive for ZrB2 or HfB2 ceramics. The addition of SiC
improved the sinterability, inhibited grain growth and increased the oxidation and ablation
resistance of ZrB2 and HfB2 ceramics as well [29-35, 58]. The improvement of
densification upon addition of SiC was attributed to the formation of intergranular liquid
phases during hot-pressing, assisting in densification at lower temperatures [31, 32].
However, the improvement of densification upon addition of ultra-fine SiC is effective
only for a uniformly dispersed SiC-ZrB2 system [58]. The agglomeration of the ultra-fine
SiC particles led to reduced improvement in densification of ZrB2, even with nano-sized
SiC particles [58].
Nitrides are other effective additives for improving sinterability or enhancing densification
of ZrB2. The main reason for incorporating nitrides as additives is the propensity of
nitrides to consume the oxygen-bearing species on the diboride powder surfaces.
The reduction of oxygen results in higher boron activity, which is one of the conditions
favoring lattice diffusion and, therefore, densification [66]. The addition of ≥ 2.5 wt.%
Si3N4 results in almost fully dense ZrB2 (relative density (RD):  98%) after compactions
at 1700°C and 30 MPa for 15 min [52, 63]. Some grain boundary phases, including BN,
ZrO2, ZrSi2 and borosilicate glassy phase, were confirmed to be present in pockets at
multiple-grains junctions for the ZrB2-based ceramics with Si3N4, the result of a reaction
of an oxide impurity with Si3N4. That reaction results in elimination or decrease in the
oxide impurity on the ZrB2 particles surfaces, thereby promoting densification [52, 63].
Similar to Si3N4, the primary effect of an AlN additive is the depletion of the ZrB2
particles from the outer oxide layer that prevents the formation of highly dense compacts
[51]. Compared to AlN and Si3N4 additives, the ZrN and HfN showed the unexpected
advantage of limiting undesirable secondary phases that eventually become detrimental to
high temperature stability. The resulting composites showed a fine and homogenous
microstructure with secondary phases such as M(C, N), MO2 (M = Zr and/or Hf) and BN.
The formation of the secondary phases during sintering was traced back to the interactions
among ZrN or HfN, carbon, and oxides such as B2O3 and ZrO2 or HfO2.
These interactions accelerated the densification of ZrB2 or HfB2 ceramics by reducing the
24
oxygen content on the starting powder surfaces. The resulting intergranular secondary
phases possess higher refractoriness than those made with AlN or Si3N4 additives.
Transition metal disilicides have been found to be an alternative and effective sintering
additive because they improve sinterability and increase oxidation resistance of ZrB2
ceramics as well. In the early 1970s, Kinoshita et al. [57] systematically investigated
densification behavior of ZrB2-based composites with MoSi2. Recently, it has been
reported that HP of the ZrB2-based ceramics with MoSi2 required only a temperature
≤ 1800°C [36-38]. The higher sintering temperature required in the earlier study resulted
from larger ZrB2 (d ≈ 20 m) and MoSi2 (d ≈ 5 m) particles. More recently, Guo et al.
[40] found that addition of 10÷40 vol.% ZrSi2 could further lower the densification
temperature of ZrB2 to 1550°C or below. Furthermore, fully dense ZrB2-ZrSi2 composites
with a fine and homogeneous microstructure, using a two-step HP process, which
consisted of a first stage at 1400°C for 30 min and a second stage at 1550°C for 15 min at
a pressure of 30 MPa were obtained. Improvement of densification, resulting from
addition of disilicides, is attributed to two major causes. One is formation of an
intergranular Si-O-B liquid phase between MoSi2 or ZrSi2 and ZrB2 particles due to the
interaction of SiO2 and B2O3 that occurs on the surfaces of particles [67]. Another is the
ductile deformation of MoSi2 or ZrSi2 particles at high temperature (> 800°C) [68, 69].
This deformation could force soft MoSi2 or ZrSi2 particles to fill in the voids left by the
ZrB2 skeleton under pressure during sintering, thus improving densification [38, 40].

1.3.2 Spark plasma sintering (SPS)

SPS, or Field Assisted Sintering Technology (FAST), is one of the most recent advanced
processing techniques developed for densifying ceramic materials [70, 71] (Fig. 1.7).
Although SPS is similar to HP, in place of indirect heating, the applied electrical field
heats the die and the powder compact. One advantage of using SPS is to enhance
densification of poorly sinterable ceramics, by simultaneously applying a uniaxial load
and a direct or pulsed electric current to a powder compact. Another advantage is that the
grain growth of starting materials is restricted, since a considerable shorter sintering time
(within minutes) is needed compared to HP or hot isostatic pressing (HIP), thereby
retaining the fine and homogeneous grains. Previous investigations [72-74] of compaction

25
of oxide, nitride and carbide ceramics produced by SPS have shown that the sintering
time, heating rate and sintering temperature are the important factors controlling fine-
grained microstructure and densification.

Figure 1.7: Schematic of SPS equipment [75].

SPS has been shown to provide a processing route for these highly refractory materials
that does not require sintering aids. The technique has been particularly successful when
using starting powders fabricated using an SHS method [76].
Materials produced by SPS have displayed comparable and/or improved mechanical
properties and oxidation resistance when compared with hot pressed materials. The reason
for the superior oxidation resistance may be the removal of surface oxides during
sintering. The storage and processing of ZrB2 (and HfB2) in air results in the formation of
surface oxides on the powder particles [77]. These surface oxides not only inhibit the
sinterability of the diboride, but they also introduce oxides such as B2O3 into the bulk
material. The presence of these low-melting-point oxides has a detrimental effect on the
high-temperature mechanical properties and oxidation resistance. The pulsed electrical
current used during SPS is thought to cause thermal or electrical decomposition of
insulating surface oxides [78, 79], although the process by which the SPS method achieves
this and the rapid sintering rates is not yet fully understood.
Table 1.VI summarizes SPS conditions, final density and grain size of ZrB2-based
ceramics produced by SPS. Medri et al. [80] showed that 60ZrB2-30ZrC-10SiC (vol.%)
composition could be sintered to a relative density of  96% at 2100°C and 30 MPa for
2 min. Grain size measurement indicated that the grain growth (maximum grain size:

26
 3 m) was inhibited during SPS. Recently, the various ZrB2-ZrC-SiC compositions
could be sintered to the fully dense compacts with fine and homogenous microstructure at
1950°C and 30 MPa for 2 min, by using the SPS technique [81]. This discrepancy in the
sintering temperature is probably associated with starting powder size and SPS conditions.
In addition, extending soaking time from 3 min to 5 min can produce fully dense ZrB2-
ZrC-SiC composites at a lower temperature (1900°C) [64]. Furthermore, addition of
5 wt.% AlN results in complete densification at 1850°C and 30 MPa for 5 min, but
addition of 5 wt.% Si3N4 still required a temperature of 1900°C. The discrepancy in
densification temperature due to additions of AlN and Si3N4 is likely attributable to a
lower onset temperature of densification and a faster shrinking rate for the AlN addition as
compared to the Si3N4 addition [64]. For the ZrB2-15 vol.% MoSi2 [37], however, the
density and grain size measurements of the compacts consolidated at 1750°C showed that
SPS was not superior to HP. Soaking time and total sintering time were noticeably shorter
for SPS (7÷24 min) than for HP (20÷140 min).

Table 1.VI: Spark plasma sintering conditions, final density and grain size of ZrB2
ceramics and ZrB2 or HfB2-based composites produced by an SPS process [61].

Without sintering additives, full density has, historically, been achieved only by HP at
temperatures ≥ 2100°C [56, 57]. Densification and grain growth occurred simultaneously
during the sintering. As a result, it was difficult to obtain a full density ZrB2 compact.
SPS of ZrB2 ceramics required a temperature of 1900°C, a holding time of 3 min and a
heating rate of 200°C/min or above. Increasing the sintering temperature to 1950°C or
extending the holding time to 10 min or above, as well as lowering the heating rate below
200°C/min, led to coarsening of the grain size. In addition, SPS has been used for
applications in other transition metal diborides, such as HfB2-SiC [37], TiB2-WB2-CrB2
[82], TiC-TiB2 [83], HfB2-MoSi2 [84] and HfC and HfB2-based composites with MoSi2
additives [85]. The enhanced densification resulting from SPS was attributable to mass
27
transfer processes, which are significantly enhanced in the process, effectively promoting
densification. The mechanism in SPS technique that enhances densification - mainly
whether or how an electric discharge is involved in accelerating the densification and
grain growth - is still the subject of intense debate. However, it is suggested that
enhancement is most probably due to (i) an efficient heat transfer; (ii) the use of
comparatively high pressure; (iii) the presence of an electrical field (use of DC pulses);
and (iv) the presence of local spark discharges generated between the powders under high-
energy electrical pulses.

1.3.3 Self propagating high temperature synthesis (SHS)

SHS is not a densification method but uses solid-state combustion to produce materials by
using internally generated chemical energy from exothermic reactions. The characteristics
of this method include fast reaction times, low energy requirements, simple experimental
apparatus and high-purity products. A disadvantage is that the reactions are difficult to
control.
SHS can also be used to prepare ZrB2 powders using inexpensive raw materials [86].
When used for the production of powders, the high heating and cooling rates involved are
thought to introduce planar defects, such as stacking faults, and linear defects such as
dislocations whose associated strain fields increase the sinterability of the powders by
providing a driving force for rearrangement of atoms. ZrB2 powders can be formed by
SHS using zirconium and boron [87]. The powders are dry mixed and cold pressed to
form pellets. The pellets are ignited and the process is an explosive one accompanied by a
large gas release. As a consequence, the resulting material is too fragile to be used as a
bulk material, although X-ray diffraction indicates it is ZrB2.
SHS has several advantages over conventional hot pressing of UHTCs, such as control of
exaggerated grain growth and lower processing temperatures. However, several problems
with hot pressed materials have been overcome by addition of reinforcing phases and
refinement of starting powder size.

28
1.3.4 Reactive hot pressing (RHP)

RHP has been identified as a potential route to produce ZrB2 ceramics with low impurity
levels and high density at a lower temperature. There are two processes that occur in RHP,
in situ reaction of precursor powders and densification, which must be completed
simultaneously during heating and subsequent holding.
Recently, RHP has been used to produce ZrB2 and/or HfB2 dense compacts by using Zr
and/or Hf and B precursors as well as to fabricate the ZrB2-based composites with SiC
and/or ZrC by using Zr, Si and B4C precursors. Table 1.VII summarizes RHP sintering
conditions, precursors, grain size and final density of the ZrB2 ceramics and ZrB2-based
ceramics fabricated by RHP.

Table 1.VII: Reactive hot-pressing conditions, precursors, grain size and final density of
ZrB2 or HfB2 ceramics and ZrB2 or HfB2-based composites fabricated by a RHP method
[61].

The main advantage of RHP compared with SHS from a processing point of view is that
the displacement reactions can take place at much lower temperatures than those that
occur during SHS. The heating rate during reactive hot pressing must be sufficiently slow
(~ 10°C/min) to prevent spontaneous self-combustion. The starting powder size and
morphology has a strong effect on the microstructure in RHP ceramics.
The improvement of densification by RHP was attributed to the formation of nano-sized
ZrB2 particles during the reactive process because the fine crystalline size should enhance
the driving force for densification when the densification is driven by minimization of the
surface free energy.
Another application of RHP is to produce ZrB2-based composites with SiC and/or ZrC,
using Zr, Si and B4C powders as precursors. Zhang et al. [88] used RHP to fabricate ZrB2-
based composites with SiC, by reacting Zr, Si and B4C at 1800°C where the following
reaction (1.1) is thermodynamically favorable:

29
 2 Zr + Si + B4C = 2 ZrB2 + SiC (1.1)

The relative density of  98% was obtained by RHP of Zr, Si and B4C powder mixtures at
1900°C and 30 MPa for 60 min. Later, Wu et al. [89] also successfully consolidated ZrB2-
SiC-ZrC composites (RD:  97%) by RHP of Zr, Si and B4C powders at 1800°C and
20 MPa for 60 min in an argon atmosphere. They found that the reactions for producing
ZrB2, ZrC and SiC were not simultaneously induced during the sintering, but in steps.
ZrB2 and ZrC were first formed by the reaction of B4C with Zr at a low temperature, then
SiC was produced by reaction of Si with ZrC and the residual B4C at a higher temperature.
In addition, Zimmermann et al. [90] found that excess B4C and Si were necessary in the
ZrH2-B4C-Si system for obtaining ZrB2-SiC composites without oxide impurity as well as
for avoiding grain coarsening during the sintering process. Similarly, RHP is also used to
produce the HfB2-based composites. Monteverde [91] obtained a fully densified
HfB2-22 vol.% SiC-6 vol.% HfC at 1900°C and 50MPa for 60 min through reaction of a
mixture of Hf, B4C and Si powders.
Recently, RHP has also been utilized to produce HfB2 by reaction of Hf and B powders at
a low temperature, by using SPS. An almost fully densified HfB2 compact was achieved
by reacting Hf and B at 1700°C and 95 MPa for 10 min in vacuo using SPS [92], instead
of HP. For comparison, HfB2 ceramics produced from commercially available powder
could achieve only  62% and  87% densities at 1800°C with 3085 MPa for 10 min and
at 1900°C with 8095 MPa for 10 min [92], respectively. The reaction between Hf and B
occurred at 1100°C, while the completion of the reaction extended over a relatively wide
temperature range [92]. However, the association of the reaction with densification was
absent during the reactive sintering. Densification was observed only at a temperature
where the conversion to the diboride was complete. Thus, the simultaneous synthesis and
consolidation of the Zr, B, or B4C and Si precursor powders, i.e. reactive sintering, could
produce the densified ZrB2-based composites at a lower temperature by using either HP or
SPS, as compared with direct consolidation of commercially available powders.

1.3.5 Pressureless sintering (PS)

In studies that were conducted in the 1970s and earlier, densification of ZrB2 ceramics was
only accomplished by HP [27, 57, 59, 60]. Because of the extreme pressures required for

30
densification, pressureless sintering of ZrB2 was considered unlikely or impossible until
the late 1980s, when studies of pressureless sintering actually began to show results.
Compared with HP, the development of a PS process would enable almost-net-shape
processing of ceramic parts with complex geometries using standard powder-processing
methods, thus reducing processing costs. Various additives have been used to improve
densification of ZrB2. In general, the additives used can be divided into two main groups:
liquid phase formers and reactive agents. Table 1.VIII summarizes the PS conditions,
agents used, grain size and final density of the resulting ZrB2 ceramics.

Table 1.VIII: Pressureless sintering conditions, grain size and final densities of the
pressureless sintered ZrB2 ceramics with various additives [61].

Liquid phase formers include refractory metals, such as Ni, Fe, Co and Mo [93-95],
as well as disilicides of transition metals, such as MoSi2 [96, 97] and ZrSi2 [39]. Cech et
al. [93] found that addition of ≥ 2 wt.% of metals was required to bring about adequate
sintering, because formation and continuous action of a liquid phase occurred only at
higher contents of added metals. The additions are more efficient for producing adequate
sintering in an argon atmosphere than in vacuo because of loss of the added metals in the
vacuum from volatilization. Lattice parameter measurements showed a gradual decrease in
the crystal lattice dimensions during sintering, resulting from substitution of zirconium
atoms in the ZrB2 lattice by the atoms of the added metals. Obviously, the mode of action
of these metallic additives that influence sintering is associated with an appreciable
contraction of the ZrB2 crystal lattice. Presumably, the contraction of the ZrB2 crystal
lattice affected the surface free energy and consequently increased the driving force for
densification.
Kislui and Kuzenkova [94] found that with Mo additions up to 15 wt.% the energy of
activation of the densification process drops below 380 kJ/mol from 680 kJ/mol because

31
Mo was incorporated into a ZrB2 solid solution. As a result, the addition of Mo activates
diffusion processes during sintering, effectively promoting densification. This
densification behavior of Mo is also found in ZrB2-20 wt.% SiC mixture powder that
could be sintered without pressure to a relative density exceeding 97.7% at 2250°C for
120 min when 4 wt.% Mo was added [98]. Other studies found that the addition of
20 vol.% MoSi2 produced the almost fully densified ZrB2 at 1850°C for 30 min without
external pressure [96, 97]. Furthermore, the additions of ≥ 20 vol.% ZrSi2 further reduced
the densification temperature; the full density was obtained at 1650°C for 60 min [39].
Differing from the previously mentioned liquid phase formers, reactive agents act as
densification aids by reacting with the oxide impurities present on the surface of starting
particles (such as ZrO2 and B2O3) which inhibit densification. The main reactive agents
used recently include B4C [53, 99, 100], C [99-101] and/or WC [53, 102]. Chamberlain et
al. [102] showed that elimination of oxide impurities on ZrB2 particles surface by the
reactions of B4C or WC with ZrO2 was the key to densification. The above-mentioned
reactions are thermodynamically favorable at a temperature ≥ 1200°C for B4C,
but > 1500°C for WC. As a result, the discrepancy in densification temperature between
the two agent-doped ZrB2 ceramics is likely associated with their different onset
temperatures for the reactions.
Moreover, grain size measurement showed that excess B4C restricts the grain growth
during sintering. A similar densification and grain growth inhibition effect of B4C was
also reported in ZrB2-containing B4C and carbon, either alone or in combination [100].
In addition, the densification effect of B4C depended on the starting ZrB2 powder size
[100]. ZrB2 with a particle size of ∼ 2 µm allows sintering to a density of ∼ 95% at
2050°C for 120 min. For comparison, ZrB2 could be achieved with full densification at
1850°C for 60 min when the particle size was reduced to ∼ 0.5 µm by an attrition milling.
Furthermore, the densification is more effective for a combination of B4C and C than for
B4C alone. Using a combination of B4C and C, the same ZrB2 powder (∼ 2 µm) could be
sintered to almost full density at 1900°C for 120 min. However, the additional
densification effect of carbon does not appear in the reduced particle size ZrB2 powder
(∼ 0.5 µm), which could be sintered to a full density at 1850°C for 60 min using either
B4C or a combination of B4C and carbon.

32
1.4 Properties

1.4.1 Thermal properties

Typical values for the coefficient of thermal expansion (CTE) and thermal conductivity of
some UHTC materials including HfB2, ZrB2 and these diborides with SiC are given in
table 1.IX [46]. In general, the CTE’s for these materials increase sufficiently with
temperature such that temperature dependent coefficients should be used during design.
To illustrate this, a comparison of CTE of pure HfB2, pure SiC and HfB2-20 vol.% SiC is
shown in figure 1.8(left).

Table 1.IX [46].

33
As previously indicated, the thermal conductivities of borides are typically higher in
comparison to many other ceramics and are a result of both a lattice and an electronic
contribution to phonon transport [5]. Figure 1.8(right) illustrates the large difference in
conductivities of the borides from the nitride and carbide ceramics of hafnium [103-105].
Although the conductivities of the borides initially drop with increasing temperature, they
eventually even out at levels still above that for the nitrides and carbides. For a leading
edge, high thermal conductivity reduces thermal stresses within the material, by reducing
the magnitude of the thermal gradients within the part. High thermal conductivity also
allows energy to be conducted away from the tip of the leading edge and reradiated out of
surfaces of the component with lower heat fluxes, reducing the surface temperature for a
given incident heat flux, compared to an insulating material.
The added benefits of high temperature materials with high conductivity can be seen when
comparing a sharp UHTC leading edge to a relatively blunt, low conductivity leading edge
such as the conventional RCC (Reinforced Carbon Carbon) thermal protection system,
used on the shuttle. Because of the low conductivity of conventional leading edge thermal
protection system (TPS) materials, the maximum surface temperature is determined by a
balance of the incident heat flux with the energy that is re-radiated out of the leading edge,
hence the need for materials with an emittance as high as possible. Due to the low thermal
conductivity of these materials there is little or no conduction of energy away from the tip
of the leading edge.
Currently the leading edges of operational space vehicles are relatively blunt, due to
temperature capability of the state of the art materials, which are typically based on SiC
and limited to ~ 1600°C, before active oxidation becomes an issue. As the leading edge
radius decreases the surface temperature increases. But the blunt leading edges
significant1y impact the vehicles performance, which effects overall vehicle safety,
as mentioned earlier. Therefore, in order to achieve maximum performance, materials are
needed that are both capable of withstanding the reentry environment at temperatures
greater than 2000°C and have a high thermal conductivity that will direct more energy
away from the tip of the leading edge allowing for even further improvements in vehicle
performance, i.e. faster velocities at lower altitudes.

34
Figure 1.8: Left, plots of CTE vs. temperature for HfB2, SiC and HfB2-20 vol.% SiC [104,
105]. Right, plots of conductivity vs. temperature for several UHTCs [103-105].

Figure 1.9 illustrates how the surface temperature for the sharp UHTC leading edges is
determined by an energy balance of incident heat flux, re-radiated energy and energy
pulled away from the leading edge tip and re-radiated out the sides of the component
where the incident heat flux is lower. In this sharp leading edge design, three dimensional
heat transfer plays an important role in understanding and achieving the full potential of
these materials. Thus the need for highly conducting, yet refractory materials is essential
in the design of sharp vehicles.

Figure 1.9: Comparison of the functionality of conduction for blunt and sharp leading
edge designs of a hypersonic space vehicle [25].

35
1.4.2 Mechanical behavior

The mechanical properties of a number of materials, including HfB2 and ZrB2 with and
without SiC additions, are given in table 1.X. Hardness values in these materials are
typically relatively high, due to their high degree of covalent bonding. The wide scatter in
the data is due to differences in processing of these materials resulting in different grain
sizes and porosity [3].
The Young’s modulus of HfB2 and ZrB2 ceramics ranges from ∼ 350 GPa to 530 GPa,
depending on porosity and additives; for a fully densified ZrB2 without additive, it is equal
to 498 GPa [30]. Historic studies have shown the Young’s modulus of the fully densified
hot-pressed polycrystalline ZrB2 to be equal to 500 GPa [5, 106]. The additions of Ni,
AlN, SiC, Si3N4, B4C and C affect the Young’s modulus of ZrB2 ceramics. The modulus
of HfB2 and ZrB2 with and without SiC addition begins to fall off above 800°C [47, 105,
107, 108]. The Young’s modulus of fully densified solids is determined principally by
interatomic forces, which decrease sharply with the interatomic distance [109].
The addition of Ni caused a decrease in the crystal lattice dimensions of ZrB2, as a result
of substitution of Zr atoms in the ZrB2 lattice by atoms of the added Ni [93]. Therefore,
the increase of Young’s modulus due to Ni addition is associated with the decrease in the
crystal lattice dimensions of ZrB2. B4C alone or a combination of B4C with C also led to
an increase in Young’s modulus. In contrast, AlN and Si3N4 additions led to a lower
Young’s modulus. The different changes with the additives are associated with the
different grain-boundary phase developed between ZrB2 grains [51, 63, 100], which result
from the interactions of the additives with impurities on the ZrB2 particles surfaces,
because the grain-boundary phase affects the Young’s modulus of ceramics [64, 110].
For a fully densified composite, Young’s modulus, Ec, may be described by the rule of
mixtures [111]

(1.2)

where Ei is Young’s modulus of ith constituent phase, Vi is volume fraction of ith

constituent phase and n is total number of constituent phases. A comparison between the
measured and predicted values found that Young’s modulus of pore-free ZrB2-based
composites obeys the rule of mixtures.

36
 Table 1.X [46].

It has been found that Young’s modulus of the pores-containing ZrB2-based composites is
mostly dominated by the porosity. For ceramic materials, the use of a linear empirical
dependence has been recommended. Assuming that the effect of pore structure and shape
on Young’s modulus is neglected, Young’s modulus, E, can be given by [112]

(1.3)

where E0 is Young’s modulus of pore-free materials, α is a constant and P is the volume

fraction of porosity in the material.
In addition, the effect of pore structure and shape on Young’s modulus have been
investigated by other researchers [113, 114]. These studies demonstrated that the local
elastic moduli decrease when the pore shape changed from spherical to oblate, as a result
of the increased stress concentration around the pore. Therefore, the effective elastic
modulus measured is lower for the case of non-spherical pores than for the case of
spherical pores. Also, it has been found that elastic properties have a different sensitivity
to porosity, regardless of pore shape. Bulk modulus is most affected by porosity; Young’s
modulus is the next most-affected modulus, followed by the shear modulus (which is

37
slightly less sensitive than Young’s modulus). In contrast, Poisson’s ratio is insensitive to
additives as well as to porosity and it remains almost constant for the various ZrB2-based
composites [38, 40, 64, 82].
Fracture toughness was in the range 2.44.8 MPa m1/2. For the ZrB2 ceramics with low
fracture toughness, intragranular cracking is common with the cracks propagated across
ZrB2 grains without being deflected along the grain boundaries. For the ZrB2 with high
fracture toughness, intergranular cracking is partially present. Thus, large grains and
intergranular cracking are required for increasing fracture toughness. Also, the fracture
toughness measured is larger in the ZrB2 with nitrides and/or carbides than with metallic
additives. This difference seems to be associated with a larger tensile residual thermal
stress in the ZrB2 ceramics with metallic additives than in the ZrB2 ceramics with carbide
and/or nitride ceramic additives.
Rezaie et al. [34] found that the fracture toughness of ZrB2-based composites with SiC is
dominated by the SiC particles size and distribution in the composites. Increased fracture
toughness produced by SiC addition is attributed to the crack deflection that occurs near
the SiC particles and/or at ZrB2/SiC interfaces. The interactions of the crack with the
microstructure are most likely controlled by the complex residual stress state that develops
during cooling from the processing temperature; that in turn is caused by the thermal
expansion mismatch between the ZrB2 and SiC particles. The contribution of crack
deflection to increasing fracture toughness depends on the total number of crack
deflections and the crack deflection angle, i.e. crack propagation path. Crack deflection is
enhanced in larger diameter grains. The increase of fracture toughness that accompanies
crack deflection is also associated with the elastic and/or frictional bridging mechanism of
grains. Elastic bridging is enhanced in larger diameter grains, while the frictional bridging
mechanism is in operation only when crack deflection and grain pullout occur, which is
prevalent in smaller diameter grains.
It was found that the flexural strengths range from 350 MPa to 580 MPa, depending on
grain size, additives and on the relative density as well. In general, finer grain size results
in higher strengths. Chamberlain et al. [30] showed that the addition of 10, 20 and
30 vol.% SiC led to the high room-temperature strengths of ZrB2, typically ∼ 1000 MPa,
with fine ZrB2 grain microstructure (average grain size: ∼ 2÷3 µm).
Recently, the effects of microstructure and SiC grain size on room-temperature flexural
strength were examined in hot pressed 30 vol.% SiC-containing ZrB2 ceramics by Zhu et
al. [33] and Rezaie et al. [34], both studies showed, based on the linear elastic fracture
38
mechanics, that the critical flaw size correlates strongly with SiC particle size.
Both studies concluded that the maximum SiC grain size in the ZrB2-SiC composites is
the strength-limiting factor and that the strength is not correlated with average ZrB2 grain
size. Although additives, such as Ni, SiC, nitrides and disilicides, could reduce the
densification temperature of ZrB2, the strength of the resulting composites is also
degraded at high temperature, as a result of softening of the intergranular amorphous
phase. For the ZrB2-based ceramics with Ni and Si3N4, the strength degraded significantly
above 800°C [55, 62]. For the ZrB2-based ceramics with the Ni additive, in particular,
the strength decreased sharply and dropped almost to zero at 1200°C. For comparison, the
ZrB2-based composites with MoSi2 retained the constant strength at temperatures
approaching 1200°C [37, 97]. However, the strength degraded rapidly above 1200°C, and
the degradation was more rapid than in the ZrB2 ceramics without additive.
High-temperature strength may be improved by increasing the refractory index of the
intergranular phase and minimizing impurities, as well as promoting crystallization of the
intergranular amorphous phase [115].

1.5 Oxidation of transition metal diborides

Historical studies concluded that the relative oxidation resistances of ZrB2 and HfB2 were
superior to those of other transition metal diborides [35]. The purpose of this section is to
critically review historical and recent research related to the oxidation behavior of ultra-
high temperature diboride ceramics with emphasis on ZrB2 and HfB2.
Zirconium and hafnium diborides undergo stoichiometric oxidation according to reactions
(1.4) and (1.5) [116]. The expressions for the change in standard state Gibbs’ free energy
with reaction (G°rxn) were calculated for the temperature range from room temperature
( 25°C) to  2000°C using data from the standard reference tables [117] (reaction (1.4))
and thermodynamic software (reaction (1.5)) [118].
ZrB2 (cr) + 5/2 O2 (g)  ZrO2 (cr) + B2O3 (l) G°rxn = - 1999500 + 374.4 T [J] (1.4)

HfB2 (cr) + 5/2 O2 (g)  HfO2 (cr) + B2O3 (l) G°rxn = - 2000460 + 375.7 T [J](1.5)

The reaction products, B2O3 and either ZrO2 or HfO2, show limited/no mutual solubility,
so the scales contain two distinct phases. As with metals and other non-oxide ceramics,

39
oxidation of these materials is favorable across a wide range of temperatures that includes
from room temperature to > 2000°C. The implication is that differences in oxidation
behavior depend on kinetic factors since a strong driving force exists across a wide
temperature range. As such, the oxidation behavior of the phase-pure diborides is divided
into two regimes based on whether the oxide that is formed is protective (at ‘low’
temperatures) or not (at ‘high’ temperatures).
Both ZrB2 and HfB2 exhibit mass gain kinetics consistent with diffusion limited processes
in the low temperature regime. The upper limit of this regime depends on factors such as
external pressure, oxygen partial pressure and gas flow-rate, but is generally considered to
be between 1100°C and 1200°C in static air. Below the transition temperature, a protective
oxide scale is formed on the surface of ZrB2 and HfB2 and both ceramics show parabolic
trends for mass gain, scale thickness and oxygen consumption as a function of time [116,
119]. Historical [120] and current (Fig. 1.10(a)) analyses of cross-sections of oxidized
specimens reveal a two layer oxide scale that consists of an outer layer of glassy B2O3 and
an inner layer that contains porous ZrO2 with the pores filled by glassy B2O3.
More recently, Parthasarathy et al. [121-123] developed an oxidation model for TiB2, ZrB2
and HfB2 showing that the oxidation rate is limited by the diffusion of oxygen through
B2O3 (i.e. transport of oxygen through ZrO2 is negligible). Based on the combination of
historical and recent experimental and modelling results, diborides exhibit passive
oxidation behavior with the formation of a protective oxide scale in the low temperature
regime.

Figure 1.10: Cross-section images of oxide scale on nominally pure ZrB2 oxidized in air at
(a) 900°C for 8h (oxide layer thickness,  10 m) and (b) 1500°C for 2h (oxide layer
thickness,  400 m) [124].

40
In the high temperature regime (i.e. above  1200°C), the oxidation behavior of the
diborides changes [119, 125-127]. Microstructural analysis (previous [120, 128] and
Fig. 1.10(b)) reveals that loss of protection for ZrB2 coincides with evaporation of B2O3
from the oxide scale. Thermodynamic models that employ either volatility diagrams [129,
130] or kinetic models, such as the one proposed by Parthasarathy et al. [121], support the
evaporation of B2O3 as the cause of the transition. Thermodynamic models predict vapour
pressures of the various gaseous boron oxides that form as a function of external
conditions such as temperature, oxygen partial pressure, etc. As shown in Fig. 1.11, B2O3
(g) is the predominant vapour species formed by evaporation of B2O3 in air at 1500°C.
Although B2O3 volatilises over a wide range of conditions, changes in the partial pressure
of oxygen in the external atmosphere or temperature at which oxidation occurs affect the
predominant species in the vapour phase. HfB2 exhibits the best oxidation resistance of the
diborides over this temperature range because the oxide layer that remains after B2O3
evaporation has a more equiaxed microstructure, which gives it greater resistance to
oxygen transport [116]. Although ZrB2 is inferior to HfB2, both have significantly better
oxidation protection than other diborides such as TiB2, TaB2 and NbB2 [46].

Figure 1.11: (a) vapour pressure of various B-O species as function of oxygen partial
pressure at 1500°C and (b) ZrB2 volatility diagram based on calculations [130].

Direct comparisons of historical and recent oxidation results for nominally pure diborides
is difficult due to lack of convention in reporting results. Whereas historical studies have
used a combination of mass gain, scale thickness, parabolic rate constant and oxygen
uptake as a function of temperature and time, more recent studies have focused on thermal

41
gravimetric analysis (TGA) to measure mass gain as a function of temperature and/or
time. In addition, differences in oxidation temperature and time also complicate direct
comparison, but table 1.XI provides an overview of some historical and recent reports
[131]. Where direct comparisons are possible for nominally phase-pure diborides,
the trends are consistent, but the quantitative values do not agree. No obvious differences
in density, composition or microstructure can be identified as the cause of the difference in
mass gain. Based on the model of Parthasarathy et al. [121], small changes in the fraction
of porosity in the ZrO2 have a significant effect on the oxidation rate. Trace impurities in
the diborides are one potential cause of the differences in oxidation behavior.

Table 1.XI: Summary of historical and recent oxidation results for nominally pure ZrB2
and HfB2 ceramics [124].

Furnace oxidation studies, such as those described above, do not reproduce conditions that
are representative of extreme environments such as the aerothermal heating encountered
during hypersonic flight. While the conclusions of these studies provide insight into
oxidation mechanisms and may be useful for screening candidate materials, hypersonic
flight produces higher heat fluxes, dissociated gaseous species and gas flow-rates that
cannot be duplicated in typical laboratory furnaces [46]. Specialised facilities such as arc
heaters [132], plasma wind tunnels [133], oxyacetylene torch testing [134] and inductively
coupled plasma facilities [135] have been developed to test materials in more realistic
environments.
The response of the refractory UHTC materials to high temperature oxidizing conditions
imposed by furnace heating has been observed to differ markedly from the behavior
observed in arc plasma facilities that provide a simulated reentry environment [136].
Furnace evaluations are normally performed for long times at fixed temperature and slow

42
gas flow with well defined solid/gas-reactant/product chemistry. Arc jet tests, on the other
hand, are carried out under high velocity gas flow conditions in which energy flux, rather
than temperature, is defined. Furthermore, furnace studies employ air at 1 atmosphere and
as diatomic species. But during a typical reentry profile for a manned space vehicle,
the pressures will generally be much less than 1 atmosphere and a significant portion of
the gas molecules will be dissociated into highly reactive monatomic species as they cross
the bow shock formed during reentry. The resulting monatomic species may recombine at
the surface giving up some of their energy to the material and, depending on the catalycity
of the substrate, this recombination can add a significant fraction to the overall heating of
the articles’s surface. Because of the differences between static or flowing air oxidation
experiments and experiments in the arc jet, correlation of material responses from the two
test situations is difficult, if not impossible in many cases, i.e. if material A performs better
in the furnace test than material B it does not necessarily hold that the same trend will
occur in the arc jet [137].
Oxidation rates during furnace tests were found to be much greater than oxidation rates
during arc jet testing [138]. This was explained by the formation of significant temperature
gradients (between the oxide surface and materials substrate) during arc jet testing, that
were otherwise absent during furnace tests. During arc jet testing the gradients maintained
the bulk material at a lower temperature than the surface, thus reducing the levels of
surface recession. The harsher conditions of the arc heater may have also resulted in a
thinner scale due to loss of material by evaporation and/or flow off of the surface during
testing. Further, it could be concluded that the more severe conditions encountered during
arc heater testing may promote the formation of an oxide scale (ZrO2 or HfO2) having a
higher density that is more protective than the scale formed in static laboratory furnace
tests.

1.5.1 Oxidation of diborides ceramics containing SiC

The relatively poor oxidation resistance of nominally phase-pure diborides at temperatures

above  1200°C motivated researchers to investigate a number of approaches to improve
oxidation resistance including solid solution additions, synthesising ternary diboride
compositions and adding second phases [35]. The most promising approach was found to

43
be the addition of SiC as a second phase, which reduced the thickness of the oxide scale
across a wide temperature range when compared to either a pure diboride or pure SiC [29,
35]. The improved oxidation resistance was attributed to the formation of a stable
borosilicate glass layer on the surface of the oxidized ceramics [120]. Since this
pioneering work, research on oxidation resistant ZrB2 and HfB2 based compositions with
silica scale forming additions has resulted in a significantly larger body of work on SiC
containing materials than other systems.
Several groups have studied the effects of SiC additions on the behavior of diborides when
exposed to air at elevated temperatures (Fig. 1.12). The reported behavior is summarized
in table 1.XII.

Figure 1.12: The layered structure that developed on ZrB2-SiC oxidized by heating to
1627°C for 10 cycles. The oxide scale consists of an outer layer of SiO2, a layer of ZrO2
and SiO2 and a layer of SiC depleted ZrB2 on top of unreacted ZrB2-SiC [139].

In addition to furnace oxidation and TGA studies that have been used to determine oxide
layer thicknesses and mass gains as functions of time and/or temperature, recent progress
has been made in the use of real time observations of growing oxide layers. Because
diborides are electrically conductive, direct heating of specimens is possible using
conventional furnace power supply and control systems [140, 141]. Direct heating allows
observation of the specimen surface during oxidation, which has led to discoveries related
to liquid convection and oxide particle transport during oxidation [142, 143]. Notably, this

44
technique has led to the identification of zirconia rich nodules that grow up and through
the liquid oxide (Fig. 1.13). The presence of these nodules has provided insight into
potential causes for the variability in thicknesses of the outer glassy oxide and underlying
partially oxidized regions noted by some authors [144-146].

Table 1.XII: Summary of oxidation behavior of SiC containing ZrB2 and HfB2 ceramics
[124].

Figure 1.13: Images of ZrB2-SiC after oxidation at 1550°C showing (a) lighter colored
ZrO2 rich nodules protruding through the darker glassy surface oxide layer [140] and (b)
the cross-section of a lighter colored ZrO2 rich nodule extending through the outer glassy
oxide layer [142].

1.5.2 Additives

The oxidation products of a material formed during exposure to an oxidizing environment

are largely responsible for the high-temperature performance of that material. This is

45
determined by the extent to which the oxidized layer can protect the bulk material from
subsequent oxidation. The physical and chemical processes that occur at the exposed
surface depend on the microstructure and composition of the oxidized material. It follows
that modification of the microstructure and composition can have a beneficial
(or detrimental) effect on the material’s oxidation resistance.
Additives can be used in several ways to improve the oxidation resistance of the UHTCs.
The main areas of interest are as follows:
A. Increasing viscosity of the borosilicate liquid
B. Inhibiting the ZrO2 polymorphic transformations
C. Using alternatives to SiC to introduce silicon
D. Forming protective refractory phases at high temperature
E. Modifying the microstructure of the ZrO2 scale

A. Increasing viscosity of the borosilicate liquid

Systems with higher viscosity and increased liquidus temperatures inhibit oxygen
diffusion to the unreacted bulk, retain the protective liquid at higher temperatures,
and suppress evaporation of boria from the glassy phase. Diffusivity is inversely
proportional to the viscosity of the liquid through which the diffusion is taking place,
which is shown in the Stokes-Einstein relationship [147]

(1.6)

where D is the diffusion constant, k is Boltzmann’s constant, T is temperature in Kelvin,

is viscosity and r is the spherical particle radius.
The viscosity of the borosilicate glass can be increased even more by adding certain
elements to the bulk material. The addition of tungsten, in 10 and 20 vol.% additions, to
ZrB2-SiC ceramics increased the viscosity of the borosilicate glassy phase significantly
but also reduced the material’s thermal shock resistance and structural stability at elevated
temperatures, which is highly undesirable for sharp, leading-edge materials [129].
The oxidation resistance of hot-pressed ZrB2 + 25 vol.% SiC composites has been
improved by the addition of diborides of Cr, Ti, Ta, Nb and V [148]. These additions
result in the production of the respective oxides in the glass. The improvement in
oxidation resistance comes from the fact that borate and silicate glasses containing oxides

46
of the elements listed (Group IV-VI transition metals) are immiscible and lead to phase
separation. Such systems contain compositions with high viscosity and liquidus
temperatures [147]. The immiscibility of the glass increases with increasing cation field
strength of the metallic oxide forming element z/r2 where z is the valence and r is the ionic
radius. ZrB2 ceramics with additions of 10 mol % CrB2, NbB2, TaB2, TiB2 and VB2 had
improved oxidation resistance, but most improvement was observed in the ZrB2-SiC +
10 mol % TaB2. The order in which the oxidation resistance was improved correlated well
with the cation field strength of the modifying additive.
Opila et al. [149] investigated the effect of tantalum additions on the oxidation
performance of zirconium diboride. They found that the addition of TaSi2 improved the
oxidation resistance of a ZrB2 + 20 vol.% SiC composite. The oxidation rate was reduced
by a factor of 10 at 1627°C. They concluded that more research was required to confirm
that the improvement in oxidation was a result of the tantalum addition and not from the
accompanying increase in silicon. It was suggested that the introduction of Ta resulted in
immiscibility of the liquid formed on oxidation, which increased the viscosity of the liquid
phase, providing a protective layer that was more resistant to volatilization.
The finer particles are more effective at trapping the liquid phase in the ZrO2 layer and
preventing oxygen transport through the liquid. The additives were effective only at small
concentrations (~ 3.32 mol %) and at larger concentrations were detrimental to the
oxidation resistance because of the formation of zirconia dendrites, which act as conduits
for oxygen transport into the bulk. The addition of yttria has been investigated by several
groups. Zhang et al. [150] found that adding 3 vol.% improved sinterability of the powders
and suppressed grain growth by reacting with oxides on the starting powder surfaces.
Grain size refinement improved the fracture toughness and flexural strength of the
material. Adding LaB6 to a ZrB2 + 20 vol.% SiC UHTC [151] resulted in significantly
higher fracture toughness compared with the same UHTC without the LaB6 (5.7 MPa m1/2
and 4.0 to 4.8 MPa m1/2, respectively) because of enhanced crack deflection and bridging
near SiC particles.

47
B. Inhibiting the ZrO2 polymorphic transformations

The integrity of the oxide scale can be improved by inhibiting the ZrO2 polymorphic
transformations and their associated volume changes. In low temperature systems, this is
achieved by the addition of stabilizing cations such as Mg, Ca and Y. However, these
cations are lost from the ZrO2 lattice at relatively low temperatures, and for UHTCs,
alternative cations have been sought. The addition of a cation such as Ta results in
substitution of the cation on the Zr site in ZrO2, thus reducing the concentration of oxygen
vacancies because of the higher valence of the cation (Ta forms Ta2O5 when oxidized).
This decreases oxygen diffusion through the scale and stabilizes the oxide phase,
increasing adhesion of the scale to the bulk ZrB2-SiC material. The cation must be of
higher valence and must form a refractory oxide scale. In addition to this, candidate
additives must be introduced as a refractory phase and form a refractory oxide. The two
best candidates are those based on niobium and tantalum, but tantalum is preferable as
Ta2O5 has a melting temperature of 1880°C (compared with 1520°C for Nb2O5) [148].
Tantalum can be added in elemental form or as a carbide, boride, or silicide.
The formation of intermediate phases should be considered. For instance, the addition of
Ta2O5 could form Ta2O5*6 ZrO2 with ZrO2. This phase has a lower melting temperature
than the pure oxides and could have a beneficial or detrimental effect on the oxidation
behavior of the composite.

C. Using alternatives to SiC to introduce silicon

Alternative methods of introducing Si to the system have been investigated. Ceramics in

the system ZrB2-Ta5Si3 have been investigated, as Ta5Si3 has a higher melting point than
SiC (2500°C and 2300°C, respectively). Ta5Si3 provides the tantalum to induce glass
immiscibility and silicon to form the protective borosilicate glass in the oxidized layer
[152]. ZrB2- and HfB2-TaSi2 composites have been formed by hot pressing [153],
resulting in formation of a solid solution as the tantalum entered the boride matrix.
The HfB2-TaSi2 displayed superior mechanical properties to the ZrB2-based material.
The substitution of Ti for SiC (e.g., a Zr-B-Ti system) has been suggested for making an

48
alternative UHTC system because of the significantly reduced vaporization rate displayed
by TiO2 compared with SiO2 [154].

D. Forming protective refractory phases at high temperature

Research into the introduction of additives to ZrB2-SiC ceramics largely focuses on using
the additives to alter the properties of the liquid phase formed at oxidation. A different
approach is to use additives to form a solid refractory scale at high temperatures, which
can resist oxidation at higher temperatures than the original material, thus providing
effective protection to the underlying bulk and preventing subsequent oxidation.
Candidate additives for this approach include those based on rare earth elements,
in particular those containing lanthanum.
Lanthanum has also been added to a ZrB2-SiC material as La2O3 [155], but resulted in the
formation of an amorphous grain boundary phase and substantial ZrB2 and SiC grain
growth. The same work found that additions of other rare earth oxides (Y2O3 and Yb2O3)
had beneficial effects on the densification, hardness and fracture toughness of the ZrB2-
SiC but did not investigate their effect on oxidation resistance.
Jayaseelan [156] investigated the addition of several rare earth (RE)-containing
compounds to ZrB2 + 20 vol.% SiC. Samples were prepared with 10 vol.% LaB6, La2O3,
or Gd2O3 and underwent oxidation testing at 1600°C. All samples successfully formed a
thick (> 100 m), dense layer of RE2Zr2O7 during oxidation (Fig. 1.14). These zirconates
have melting temperatures > 2300°C and will provide oxidation protection at temperatures
when the borosilicate phase has vaporized from the exposed surface. Also, the reaction of
the RE with ZrO2 is expansive and therefore fills voids at the oxidized surface created by
the removal of volatile species such as B2O3.

49
Figure 1.14: Secondary electron micrographs of oxidized scale on (a) ZrB2-20 vol.% SiC-
10 vol.% La2O3 and (b) ZrB2-20 vol.% SiC-10 vol.% LaB6 after oxidation for 1h at
1600°C showing (1) dense La-containing layer, (2) intermediate ZrO2-containing layer
and (3) unreacted bulk [156].

E. Modifying the microstructure of the ZrO2 scale

Another novel technique is the use of additives to alter the microstructure of the ZrO2
scale. By providing a liquid phase sintering route for the ZrO2, it is possible to decrease
the porosity of the scale and inhibit the subsequent transport of oxygen into the bulk
material. As ZrO2 has a melting point of 2715°C, a sufficiently dense scale would provide
effective oxidation resistance at temperatures above those at which the boria or
borosilicate phase is vaporized (~ 1600°C). This approach has been investigated by Zhang
et al. [157] with additions of W to ZrB2, which results in formation of a WO3-ZrO2
eutectic at ~ 1275°C. ZrB2 + 4 mol % WC ceramics underwent TGA at 10°C/min to
1500°C and isothermal oxidation studies at 1500°C or 1600°C for 1, 2, or 3 hours in
flowing air. The ZrB2 + 4 mol % WC had better oxidation resistance than ZrB2,
as indicated by the normalized mass gain after TGA heating to 1500°C (~ 4.5 mg/cm2 and
~ 14 mg/cm2, respectively). ZrB2 + WC also showed superior oxidation resistance in the
isothermal oxidation tests at 1500°C and 1600°C, and the reduced mass gain of the
ZrB2 + WC samples was more significant at longer oxidation times, supporting
densification of the ZrO2 scale. The addition of WC to ZrB2-SiC ceramics was also
investigated and the presence of W increased the oxidation resistance of these ceramics as
well. However, additional research is necessary to confirm that the improved oxidation
resistance is caused by liquid phase sintering of the ZrO2 scale.

50
1.6 Siliconized silicon carbide (Si-SiC)

Research on UHTCs slowed considerably until the early 1990’s, when interest in the
monolithic UHTC materials was renewed. The high costs of raw materials in addition to
the high temperatures and pressures required to hot press UHTCs powders has led many
new investigations into alternate starting materials and methods of fabricating UHTCs.
In addition to conventional methods, researchers are also looking at reactive hot pressing
and pressureless sintering by liquid infiltration and reaction [88, 116, 158]. These new
reaction-based processes share the near-net shape and near-net dimension capabilities of
gas-phase and liquid-phase reaction bonding as well as reduced processing temperatures
and times required for solid state sintering.
Siliconised silicon carbide (Si-SiC) ceramics were developed originally as a potential
nuclear fuel cladding material, now finding applications as submersible pump shaft seals.
They have undergone successful trials as high-temperature gas turbine components,
principally for the combustion chamber. Si-SiC is one of the promising CMC (Ceramic
Matrix Composite) materials currently being promoted in Ultra High Temperature Heat
Exchanger (UHTHE) systems for their improved heat transfer properties as well as their
operational durability. Under solid fuel combustion conditions, however, the formation of
ash deposition onto heat transfer surfaces of CMC tubes, reduces their envisaged thermal
efficiency and long time applicability.
Silicon infiltrated reaction bonded Si-SiC, is an important member of this SiC-ceramics
family, due to its suitability in oxidizing atmospheres at high temperature and the
passivating effect of the silica scale formed during exposure. It not only protects against
further oxidation, but also against corrosion by other gaseous species. To assure good
protection, a good cohesion within the scale and adhesion to the Si/SiC substrate is
required.
Si-SiC composites were developed in 1975 by Hillig and his collaborators. Their
production involved conversion of the surface of a graphite component to silicon carbide
by a vapour-phase reaction. In this way the easy machining of graphite is combined with
the wear characteristics of silicon carbide. The mechanical properties of Si-SiC
composites are similar to those of normal reaction-bonded silicon carbide.
Reaction-bonded (or self-bonded) silicon carbide covers a wide range of compositions and
manufacturing processes, some of which have been used since the 1960s. Typical

51
examples are the commercial materials REFEL and KT, and silicon-infiltrated sintered
silicon carbides.
In general, a plastic body is formed using a mixture of silicon carbide powder ( 5 m),
carbon or graphite and a plasticizer. In some variations of the process, silicon carbide
powder and a char-forming plastic binder is used. The plastic body is pressed, extruded,
injection moulded or otherwise formed into a green body. The plasticizers are then burned
off or converted to a porous char by pyrolysis. Silicon either as a liquid or vapour, is then
introduced into the body where it reacts with graphite or char to form silicon carbide in
situ. The reaction product crystallizes on and serves to bond the primary silicon carbide
powder. Excess silicon, about 10 vol.% is usually left to fill the pores, thus yielding a non-
porous body. It is completely impermeable and fine-grained in structure, consisting of
interlocking grains interspersed with a small proportion of free silicon and it has a very
high strength, which can be retained up to the melting point of silicon, about 1413°C.
These materials show good bending strength (at room temperature) in the order of
400600 MPa. When silicon melts, but skeleton is unaffected, the strength can be retained
up to 50%. The Young modulus of this material is approximately 415 GPa.

1.6.1 Infiltration theory

Infiltration (also known as sinter casting) is a process whereby a molten phase is wicked
into the open pores of a previously sintered porous structure [159]. The liquid is added
from an external source. For infiltration to occur, the structure to be infiltrated should have
pores which are open and interconnected. The preform to be infiltrated should have at
least 10% porosity [159]. Infiltration offers the advantages of sintering at lower
temperatures, shorter processing time, possibility of fully dense products, elimination of
porosity with minimal dimensional change and near net-shaping without necessarily
requiring application of an external pressure. It can also be successfully performed at low
pressures giving a dense product. Below is a schematic illustration of the infiltration
sequence (Fig. 1.15).

52
 Figure 1.15: Schematic infiltration
sintering, where the pore space in a
powder or sintered porous structure is
filled with molten phase by cycling to a
temperature between the melting points
between the two materials [159].

Infiltration can occur in two different ways, namely reactive infiltration and non-reactive
infiltration. For a reactive infiltration system, there is interaction (reaction) between the
melt (infiltrant) and the preform whereas for the non-reactive infiltration system, there is
no interaction (reaction) between the melt and the preform. The former system mainly
depends on the wetting behavior and the reaction between the melt and the preform
whereas the latter system depends on the wetting behavior of the melt and the preform
only. The infiltration process is sensitive to surface contamination therefore it requires
clean surfaces for wetting.

A. Non-reactive infiltration

For the non-reactive infiltration system, the wetting behavior of the liquid and the preform
material plays a major role. The wicking of the melt only occurs if the liquid wets the
solid phase. The wetting and spreading of a liquid on a solid surface (Fig. 1.16) depend on
short range molecular forces that can be modified by a monolayer coating. Wetting may
cause compressive forces between particles and liquid migration.
A system with  < 90° is considered as wetting and a system with 90° <  < 180° is
considered as non-wetting.
The wetting angle of the interfacial tensions at the liquid-vapour-solid juncture is
described by the Young equation ( > 0).

 (1.7) [160]

Where γSV, γSL, γLV are the effective interfacial tensions as indicated in Fig. 1.16(d).

53
 Figure 1.16: Wetting behavior between
a liquid and a solid showing a) good
wetting b) poor wetting c) complete
wetting and d) the balance between the
interfacial tensions for a liquid with
contact angle  [160].

Assuming cylindrical pore channels, the height of infiltration (height risen by the fluid in
the preform), that can be obtained from the capillary force is:
From:

  (1.8)

Thus
 
 (1.9) [160]

Where: H = height risen by the liquid; P = internal pressure; = density of the liquid;
g = acceleration due to gravity; Rc = pore radius; γLV = surface tension;  = contact
(wetting) angle.
It can be deduced from equation 1.9 that fine pores give rise to a higher liquid rise
(i.e. provide a greater suction) than larger pores.
Capillary pressure provides the driving force for infiltration and good wetting on the
surfaces by the liquid is a key requirement. The kinetics of infiltration is complex [161].
However, if the pore structure of the preform is simplified as a set of parallel cylinders,
the kinetics can be determined from Poiseuille’s equation of viscous flow through a tube.
The average laminar flow velocity () of a column of liquid in a horizontal cylindrical
capillary is therefore given by the Poiseuille equation.

 (1.10) [161]

54
Combining equations 1.8 and 1.10, thus,

 (1.11) [161]

Where = viscosity of the liquid; L = length filled by the liquid.

From equation 1.11, it can be seen that the rate of penetration of the liquid into a preform,
will be greater for a liquid of lower viscosity and higher surface tension and for pores of
larger radius [161].
If time is taken into consideration during infiltration, the depth of infiltration varies with
the square root of time. The height infiltrated in time t is then found to be:

(1.12) [162]

Where, h = height of infiltration; r = radius of the tube; γ = surface tension of the liquid;
 = contact (wetting) angle; t = time taken for infiltration; = viscosity of the fluid.
Equation 1.12 shows that the rate of infiltration is proportional to the pore size. Since the
pore size increases with the particle size, preforms made with larger particles are more
readily infiltrated [162]. Theoretically, it is possible to calculate the green density of the
preform required to produce a fully dense composite, but in practice it is necessary to
leave additional porosity to allow the infiltrating liquid to move freely.
In a non-wetting system, infiltration does not readily occur, as opposed to the wetting
system. In-fact a liquid that is non-wetting tends to force particles apart. For such a
system, for infiltration to take place, an external pressure has to be applied. This pressure
should be greater than the maximum capillary resistance of the preform to the penetration
of the liquid melt into the largest interfaces [162]. Darcy’s law states that:


 (1.13) [163]

Where, h = infiltration height; t = infiltration time; k = intrinsic permeability; Vp = particle

volume fraction of packed sample; μ = viscosity of liquid melt; P = pressure drop in the
liquid.
When a plot of h2 against pressure is done (Fig. 1.17), we get a straight line and this also
shows the threshold pressure (Po) that is needed to kick-start infiltration [163].

55
 Figure 1.17: An illustration of Darcy’s plot of infiltration height vs. pressure [164].

B. Reactive infiltration

Reaction sintering is a method whereby objects are made by reacting two components in a
mix in order to produce a third, ceramic phase. The high temperature needed to initiate the
reaction is sometimes further boosted by the heat generated from the enthalpy of the
reaction taking place. This high temperature further serves to sinter the particles of the
new phase generated into a generally dense ceramic body. However, for the reaction
sintering route, voids seem to be a problem if the reaction proceeds under low pressure
conditions, due to volume reduction occurring during the reaction of carbon with silicon to
form silicon carbide [165].
For a reactive infiltration system, the wetting behavior and the rate of reaction of the
system plays a major role. There are many factors that influence these parameters.
Good wetting is an important factor during reactive infiltration. A liquid possessing a low
contact angle readily wets the solid thus allowing easier wicking of the liquid into the
pores. Both carbon and SiC show good equilibrium wetting with liquid silicon.
Another important factor in such systems is the pore size distribution and porosity.
For infiltration to occur, the preform should have at least 10% porosity and pores which
are open and interconnected [159]. There is volume expansion of the solid phase during
the reaction in the carbon-SiC system, so this means that there exists an optimal pore
diameter that should allow full infiltration and at the same time accommodating this
volume expansion. In the carbon-SiC system, too large pores are undesirable because
there will be a high probability of ending up with free silicon in the final product.
One major advantage to the reactive infiltration technique is the avoidance of porosity

56
defects. Residual silicon (free silicon) of relatively low melting point (1410ºC) is
detrimental to high-temperature application and it impairs the mechanical properties, so it
has to be minimal in the final product [166].
For reactive infiltration to occur, temperature also plays a major role. Infiltration is carried
out at temperatures higher than the melting point of the infiltrant but less than the
softening or deformation temperature of the preform. Higher temperatures greater than the
melting point of the infiltrant also have an advantage of reducing the melt viscosity
thereby allowing easier wicking of the melt into the porous preform. The temperature has
to be strictly controlled though, because there is a need to establish the refractoriness of
the preform material so that maximum temperatures can be determined which will allow
the melt viscosity to be reduced without compromising the preform material.
In most cases, the reaction and infiltration occur simultaneously. Also sometimes, as in the
carbon-silicon system, the rate of product formation is very fast and this results in
blocking of the pore channels thus making further infiltration of the preform very difficult.
So in order to maintain an interconnected porosity, ways should be found that can retard
the rate of reaction (product formation) so that this phenomenon is curbed. If the system is
non-wetting, an external pressure is required to force the liquid into the pores. Even if the
system is wetting, the external pressure can also be used to accelerate the rate of
infiltration and also help the melt to reach the much smaller pores.
The reaction between carbon and silicon to form silicon carbide is mainly thought to be a
diffusion process though existing literature presents conflicting interpretations of the
mechanism of the reaction. The assessed equilibrium phase diagram of the Si-C system is
presented in Fig. 1.18.
Researchers, Olesinski and Abbaschian [167], studied the thermodynamics of the Si-C
system to determine the stability of various phases of the Si-C system. They concluded
that silicon carbide is a thermodynamically stable compound, with the cubic form (-SiC)
more stable than the hexagonal form (-SiC) at any temperature below the peritectic
point. The heat of formation from the elements is given by: H0298 = - (6973) kJ/mol,
which means a distinctive exothermal chemical reaction [168].
Studies on the kinetics and the reaction mechanism of the SiC formation from different
precursors have been reported and several theories have been put forward. There is a
general agreement that the rate controlling step of the reaction is the diffusion of carbon
atoms through the newly formed SiC with activation energies of 100400 kJ/mol [169].

57
 Figure 1.18: Equilibrium phase diagram of the carbon-silicon system [167].

Other suggestions pointed that the vacancy mechanism is responsible for the diffusion of
silicon and carbon through -SiC [170]. Carbon is the faster diffusing component in the
lattice of -SiC due to its smaller atomic size. The lowest energy migration route is
believed to be the direct jump to the nearest Si vacancy without the occupation of an
intermediate vacant octahedral position. The diffusion constants of carbon and silicon in
-SiC are given in Fig. 1.19.

Figure 1.19: Self-diffusion coefficients

of 14C lattice diffusion (Dlc), 14C grain-
boundary diffusion (Dbc) and 30Si
lattice diffusion (Dsi), in -SiC as a
function of 1/T [170].

58
Fitzer and Gadow [168] investigated the reactivity of different carbons with liquid silicon.
They found out that the microstructure and porosity of the original carbon determines how
the heterogeneous chemical reaction between carbon and liquid silicon takes place.
The reaction mechanism involves three stages: a silicon mass transfer by volume diffusion
caused by capillary effect, the complete wetting of molten silicon on pure carbon surfaces
and the grain boundary and bulk diffusion of silicon through the primarily formed SiC
layer. The structure of the SiC layer on the inner and outer surface of the sample
determines the effective diffusion coefficient. This further supports the theories that the
reaction between carbon and silicon is mainly a diffusion process. However, Hase and
Suzuki [171] found no continuous product layer between carbon particles and silicon.
They suggested that the reaction product quickly spalled owing to the volume misfit
between SiC and carbon leading to an interface-limited reaction directly between silicon
and carbon.
Sawyer and Page [172] and Ness and Page [173] also proposed a different reaction
mechanism. They proposed that the reaction mechanism is mainly a solution
re-precipitation mechanism. The SiC is precipitated from a supersaturated solution of
graphite in liquid silicon on to local nucleation sites. The rate of carbon dissolution is
accelerated by the heat evolved in the reaction and this allows the reaction to go to
completion. The highest dissolution of carbon in silicon is believed to be at the reaction
front since the reaction is highly exothermic and the supersaturation is believed to exist in
the cooler regions away from the reaction front [174]. The newly formed SiC grows as the
-polytype and nucleates either on the existing grits or less frequently, on the graphite or
even homogenously [173]. The finely dispersed -SiC forms during the local cooling
down of the liquid silicon immediately behind the reaction front and thus never
experiences sufficient thermal activation to transform to -SiC. Zhou and Singh [174]
supported this solution-precipitation mechanism. They found SiC precipitates in the
silicon phase and this was assumed that carbon had dissolved in the liquid phase and got
carried away into the silicon phase and finally precipitated as SiC during cooling.
The kinetics of the reaction between carbon and silicon has also been studied. Zhou and
Singh [174] studied the reaction kinetics of glass carbon and a pure silicon melt. In their
study, they found out that most of the SiC was formed during the early stage of the
reaction and the growth rate of the SiC layer decreased in the latter stage of the reaction.
A model which can explain the experimentally observed rate law was proposed, in which

59
a negative space charge was assumed to exist in the growing SiC layer. The diffusion of
the carbon-ion vacancy through this layer under an electric field is proposed as the rate
limiting step for the growth of the SiC layer [174]. A critical analysis of the data of Zhou
and Singh [174] would suggest that there is a change in the mechanism after the layer has
formed. Assuming such change in the reaction mechanism, there is no necessity for a forth
power rate law to describe the data. The time dependence of the growth of the SiC film
itself is difficult to determine due to the low growth rate and the relative low time of the
experiments.

Figure 1.20: Thickness of the SiC layer as a function of the reaction time for Si-C at
1430°C, 1475°C and 1510°C [174].

From Fig. 1.20, it can be observed that an increase in the reaction time gave rise to an
increase in the SiC thickness layer. There is a sharp increase in the SiC thickness at the
initial stages of the reaction and then it slows down as the reaction progresses.

60
1.7 References

[1] L. Kaufman and E. V. Clougherty, Investigation of Boride Compounds for Very High Temperature
Applications, RTD-TRD-N63-4096, Part III, ManLabs Inc., Cambridge, MA, (1966).
[2] E. V. Clougherty, D. Kalish and E. T. Peters, Research and Development of Refractory Oxidation
Resistant Diborides, AFML-TR-68-190, ManLabs Inc., Cambridge, MA, (1968).
[3] J. J. Gangler, NASA Research on Refractory Compounds, High Temp. High Press., 3 487-502 (1971).
[4] E. L. Courtright, H. C. Graham, A. P. Katz and R. J. Kerans, Ultra High Temperature Assessment
Study - Ceramic Matrix Composites, AFWAL-TR-91-4061, Wright Patterson Air Force Base, Ohio, (1992).
[5] R. A. Cutler, Engineering Properties of Borides, In Ceramics and Glasses, Engineering Materials
Handbook, vol. 4, ed. S. J. Schneider Jr. ASM International, Materials Park, OH, 787- 803 (1991).
[6] E. P. Simonenko, D. V. Sevast’yanov, N. P. Simonenko, V. G. Sevast’yanov and N. T. Kuznetsov,
Promising Ultra-High-Temperature Ceramic Materials for Aerospace Applications, Rus. J. Inorg. Chem.,
58 [14] 1669-93 (2013).
[7] S. A. Tucker and H. R. Moody, II. - The Production of Hitherto Unknown Metallic Borides, J. Chem.
Soc., 81 14-7 (1902).
[8] G. V. Samsonov, Refractory Transition Metal Compounds: High Temperature Cermets, Academic
Press, New York, (1964).
[9] P. Schwarzkopf and R. Kieffer, Refractory Hard Metals: Borides, Carbides, Nitrides, and Silicides,
New York, MacMillan, (1953).
[10] S. R. Levine, E. J. Opila, M. C. Halbig, J. D. Kiser, M. Singh and J. A. Salem, Evaluation of Ultra-
High Temperature Ceramics for Aeropropulsion Use, J. Eur. Ceram. Soc., 22 [14-15] 2757-67 (2002).
[11] J. Han, P. Hu, X. Zhang and S. Meng, Oxidation Behavior of Zirconium Diboride-Silicon Carbide
at 1800°C, Scripta Mater., 57 [9] 825-8 (2007).
[12] W. G. Fahrenholtz, Thermodynamic Analysis of Zrb2-SiC Oxidation: Formation of a SiC-Depleted
Region, J. Am. Ceram. Soc. 90 [1] 143-148 (2007).
[13] F. Monteverde and A. Bellosi, Oxidation of ZrB2-Based Ceramics in Dry Air, J. Electrochem. Soc.,
150 [11] B552-9 (2003).
[14] C. Mroz, Annual Mineral Review; Zirconium Diboride, Am. Ceram. Soc. Bull., 74 165-6 (1995).
[15] R. Telle, L. S. Sigl and K. Takagi: in ‘Handbook of ceramic hard materials’, (ed. R. Riedel),
Weinheim, Wiley-VCH, 802-945 (2000).
[16] B. Stucker, W. Bradley, P. T. Eubank, S. Norasetthekul and B. Bozkurt: Proc. Symp. on ‘Solid
freeform fabrication’, Austin, TX, USA, SAMPE, Vol. 1, 257-65 (1997).
[17] N. Kaji, H. Skikano and I. Tanaka, Taikabutsu Overseas, 14 39-43 (1992).
[18] J. Sung, D. M. Goedde, G. S. Girolami and J. R. Abelson, Remote-Plasma Chemical Vapor
Deposition of Conformal ZrB2 Films at Low Temperature: A Promising Diffusion Barrier for Ultralarge
Scale Integrated Electronics, J. Appl. Phys., 91 [6] 3904-11 (2002).
[19] Y. Murata, Cutting tool tips and ceramics containing hafnium nitride and zirconium diboride,
US Patent 3, 487 594 (1970).
[20] Z. J. Jin, M. Zhang, D. M. Guo and R. K. Kang, Electroforming of Copper/ZrB2 Composite Coating
and its Performance as Electro-Discharge Machining Electrodes, Key Eng. Mater., 291-2 537-42 (2005).
[21] K. Upadhya, J. M. Yang and W. P. Hoffmann, Materials for Ultrahigh Temperature Structural
Applications, Am. Ceram. Soc. Bull., 76 [12] 51-6 (1997).
[22] F. Monteverde, A. Bellosi and S. Guicciardi, Processing and Properties of Zirconium Diboride-
Based Composites, J. Europ. Ceram. Soc., 22 [3] 279-88 (2002).
[23] I. M. Low and R. McPherson, Fabrication of New Zirconium Boride Ceramics, J. Mat. Sci. Lett.,
8 [11] 1281-4 (1989).
[24] P. Kolodziej, J. Salute and D. L. Keese, First Flight Demonstration of a Sharp Ultra-High
Temperature Ceramic Nosetip, NASA TM-112215, (1997).
[25] D. A. Kontinos, K. Gee and D. K. Prabhu, Temperature Constraints at the Sharp Leading Edge
of a Crew Transfer Vehicle, AIAA 2001-886 (2001).
[26] A. F. Guillermet and G. Grimvall, Phase Stability Properties of Transition Metal Diborides,
Am. Inst. Phy. Conf. Proc., 231 423-31 (1991).
[27] W. G. Fahrenholtz, G. E. Hilmas, I. G. Talmy, and J. A. Zaykoski, Refractory Diborides
of Zirconium and Hafnium, J. Am. Ceram. Soc., 90 [5] 1347-64 (2007).
[28] A. S. Brown, Hypersonic Designs with a Sharp Edge, Aerospace Am., 35 [9] 20-1 (1997).
[29] W. C. Tripp, H. H. Davis and H. C. Graham, Effect of an SiC Addition on the Oxidation of ZrB2,
Bull. Am. Ceram. Soc., 52 [8] 612-6 (1973).

61
 [30] A. L. Chamberlain, W. G. Fahrenholtz, G. E. Hilmas and D. T. Ellerby, High-Strength Zirconium
Diboride-Based Ceramics, J. Am. Ceram. Soc., 87 [6] 1170-2 (2004).
[31] F. Monteverde, Beneficial Effects of an Ultra-Fine -SiC Incorporation on the Sinterability and
Mechanical Properties of ZrB2, Appl. Phys. A, 82 [2] 329-37 (2006).
[32] S. S. Hwang, A. L. Vasiliev and N. P. Padture, Improved Processing and Oxidation-Resistance
of ZrB2 Ultra-High Temperature Ceramics Containing SiC Nanodispersoids, Mater. Sci. Eng. A, 464 [1-2]
216-24 (2007).
[33] S. Zhu, W. G. Fahrenholtz and G. E. Hilmas, Influence of Silicon Carbide Particle Size on the
Microstructure and Mechanical Properties of Zirconium Diboride-Silicon Carbide Ceramics, J. Eur. Ceram.
Soc., 27 [4] 2077-83 (2007).
[34] A. Rezaie, W. G. Fahrenholtz and G. E. Hilmas, Effect of Hot-Pressing Time and Temperature on
the Microstructure and Mechanical Properties of ZrB2-SiC, J. Mater. Sci., 42 [8] 2735-44 (2007).
[35] E. V. Clougherty, R. L. Pober and L. Kaufman, Synthesis of Oxidation Resistance Metal Diboride
Composites, Trans. TMS AIME, 242 [6] 1077-82 (1968).
[36] D. Sciti, F. Monteverde, S. Guicciardi, G. Pezzotti and A. Bellosi, Microstructure and Mechanical
Properties of ZrB2-MoSi2 Ceramic Composites Produced by Different Sintering Techniques, Mater. Sci.
Eng. A, 434 [1-2] 303–309 (2006).
[37] A. Bellosi, F. Monteverde and D. Sciti, Fast Densification of Ultra-High Temperature Ceramics by
Spark Plasma Sintering, Int. J. Appl. Ceram. Technol., 3 [1] 32-40 (2006).
[38] S.-Q. Guo, T. Nishimura, T. Mizuguchi and Y. Kagawa, Mechanical Properties of Hot-Pressed
ZrB2-MoSi2-SiC Composites, J. Eur. Ceram. Soc., 28 [9] 1891-8 (2008).
[39] S.-Q. Guo, Y. Kagawa, T. Nishimura and H. Tanaka, Pressureless-Sintering and Physical
Properties of ZrB2-Based Composites with ZrSi2 Additive, Scripta Mater., 58 [7] 579-82 (2008).
[40] S.-Q. Guo, Y. Kagawa and T. Nishimura, Mechanical Behavior of Two-Step Hot-Pressed ZrB2-
Based Composites with ZrSi2, J. Eur. Ceram. Soc., 29 [4] 787-94 (2009).
[41] P. T. B. Shaffer, Engineering Properties of Carbides, in ASTM Engineered Materials Handbook,
Vol. 4 - Ceramics and Glasses, S. J. Schneider, technical chairman, 804-11 (1991).
[42] Battelle Columbus Laboratories, Engineering Property Data on Selected Ceramics,Vol. 2: Carbides.
Metals and Ceramics Information Center, Battelle Columbus Laboratories, Report MCIC-HB-07-Vol. 2
(1979).
[43] S. Hampshire, Engineering Properties of Nitrides, in ASTM Engineered Materials Handbook,
Vol. 4 - Ceramics and Glasses, S. J. Schneider, technical chairman, 812-20 (1991).
[44] K. E. Spear, Chemical Bonding in AlB2-Type Borides, J. Less-Common Metals, 47 [C] 195-201
(1976).
[45] J. K. Burdett, E. Canadell and G. J. Miller, Electronic Structure of Transition Metal Borides with the
AIB2 Structure, J. Am. Chem. Soc., 108 [21] 6561-8 (1986).
[46] M. J. Gasch, D. T. Ellerby and S. M. Johnson, Ultra High Temperature Ceramic Composites,
Handbook of Ceramic Composites, ed. N. P. Bansal, Kluwer Academic Publishers, (2005).
[47] G. V. Samsonov and I. M. Vinitskii, Handbook of Refractory Compounds, Plenum Press, (1980).
[48] A. F. Guillermet and G. Grimvall, Bonding Properties and Vibrational Entropy of Transition Metal
MeB2(AlB2) Diborides, J. Less-Common Metals, 169 [2] 257-81 (1991).
[49] L. B. Pankratz, J. M. Stuve and N. A. Gokcen, Thermodynamic Data for Mineral Technology,
Bulletin 677, U.S. Bureau of Mines, 98-102 (1984).
[50] F. Monteverde and A. Bellosi, Development and Characterization of Metal-Diboride-Based
Composites Toughened with Ultra-Fine SiC Particulates, Solid State Sci., 7 [5] 622-30 (2005).
[51] F. Monteverde and A. Bellosi, Beneficial Effects of AlN as Sintering Aid on Microstructure and
Mechanical Properties of Hot-Pressed ZrB2, Adv. Eng. Mater., 5 [7] 508-12 (2003).
[52] F. Monteverde and A. Bellosi, Effect of the Addition of Silicon Nitride on Sintering Behaviour and
Microstructure of Zirconium Diboride, Scripta Mater., 46 [3] 223-8 (2002).
[53] S. Zhang, G. E. Hilmas, and W. G. Fahrenholtz, Pressureless Densification of Zirconium Diboride
with Boron Carbide Additions, J. Am. Ceram. Soc., 89 [5] 1544-50 (2006).
[54] J. B. Hedrick, Zirconium and Hafnium, U.S. Geological Survey Minerals Yearbook, 86.2-86.8,
(1999).
[55] M. Pastor, Metallic Borides: Preparation of Solid Bodies, Sintering Methods and Properties of Solid
Bodies, In Boron and Refractory Borides, ed. V. I. Matkovich. Springer, New York, 457-93 (1977).
[56] G. A. Meeson and A. F. Gorbunow, Activated Sintering of Zirconium Boride, Inorg. Mater.,
4 267-70 (1968).
[57] M. Kinoshita, S. Kose and Y. Hamano, Hot-Pressing of Zirconium Diboride-Molybdenum Disilicide
Mixtures, Yogyo-Kyokai-Shi, 78 [2] 32-41 (1970).

62
 [58] S.-Q. Guo, J.-M. Yang, H. Tanaka and Y. Kagawa, Effect of Thermal Exposure on Strength of ZrB2-
Based Composites with Nano-Sized SiC Particles, Compos. Sci. Technol., 68 [14] 3033-40 (2008).
[59] D. Kalish and E. V. Clougherty, Densification Mechanisms in High-Pressure Hot-Pressing of HfB2,
J. Am. Ceram. Soc., 52 [1] 26-30 (1969).
[60] D. Kalish, E. V. Clougherty and K. Kreder, Strength, Fracture Mode, and Thermal Stress
Resistance of HfB2 and ZrB2, J. Am. Ceram. Soc., 52 [1] 30-6 (1969).
[61] S.-Q. Guo, Densification of ZrB2-Based Composites and their Mechanical and Physical Properties:
A Review, J. Eur. Ceram. Soc., 29 [6] 995-1011 (2009).
[62] J. J. Melendez-Martinez, A. Dominguez-Rodriguez, F. Monteverde, C. Melandri and G. De Portu,
Characterization and High Temperature Mechanical Properties of Zirconium Boride-Based Materials,
J. Eur. Ceram. Soc., 22 [14-15] 2543-9 (2002).
[63] F. Monteverde, S. Guicciardi and A. Bellosi, Advances in Microstructure and Mechanical
Properties of Zirconium Diboride Based Ceramics, Mater. Sci. Eng. A, 346 [1-2] 310-9 (2003).
[64] S.-Q. Guo, Y. Kagawa, T. Nishimura and H. Tanaka, Elastic Properties of Spark Plasma Sintered
(SPSed) ZrB2–ZrC–SiC Composites, Ceram. Int., 34 [8] 1811-7 (2008).
[65] F. Monteverde and A. Bellosi, Efficacy of HfN as Sintering Aid in the Manufacture of Ultra-High-
Temperature Metal Diborides-Matrix Ceramics, J. Mater. Res., 19 [12] 3576-85 (2004).
[66] S. Baik and P. F. Becher, Effect of Oxygen Contamination on Densification of TiB2, J. Am. Ceram.
Soc., 70 [8] 527-30 (1987).
[67] T. J. Rockett and W. R. Foster, Phase Relations in the System Boron Oxidesilica, J. Am. Ceram.
Soc., 48 [2] 75-80 (1965).
[68] Y.-L. Jeng and E. J. Lavernia, Review: Processing of Molybdenum Disilicide, J. Mater. Sci., 29 [10]
2557-71 (1994).
[69] R. Rosenkranz and G. Frommeyer, Microstructures and Properties of the Refractory Compounds
TiSi2 and ZrSi2, Z. Metalkd., 83 [9] 685-9 (1992).
[70] M. Tokida, Trends in Advanced Spark Plasma Sintering System and Technology, J. Soc. Powder
Technol. Jpn., 30 [11] 790-804 (1993).
[71] M. Nygren and Z. Shen, On the Preparation of Bio-, Nano- and Structural Ceramics and
Composites by Spark Plasma Sintering, Solid State Sci., 5 [1] 125-31 (2003).
[72] Z. Shen, M. Johnsson, Z. Zhao and M. Nygren, Spark Plasma Sintering of Alumina, J. Am. Ceram.
Soc., 85 [8] 1921-7 (2002).
[73] K. A. Khor, K. H. Cheng, L. G. Yu and F. Boey, Thermal Conductivity and Dielectric Constant
of Spark Plasma Sintered Aluminum Nitride, Mater. Sci. Eng. A, 347 [1-2] 300-5 (2003).
[74] F. Guillard, A. Allemand, J. D. Lulewicz and J. Galy, Densification of SiC by SPS-Effects of Times,
Temperature and Pressure, J. Eur. Ceram. Soc., 27 [7] 2725-28 (2007).
[75] M. Stuer, Z. Zhao, U. Aschauer and P. Bowen, Transparent Polycrystalline Alumina Using Spark
Plasma Sintering: Effect of Mg, Y and La Doping, J. Eur. Ceram. Soc., 30 [6] 1335-43 (2010).
[76] R. Licheri, R. Orrù, C. Musa, A. M. Locci and G. Cao, Spark Plasma Sintering of UHTC Powders
Obtained by Self-Propagating High-Temperature Synthesis, J. Mater. Sci., 43 [19] 6406-13 (2008).
[77] A. Bellosi, F. Monteverde, D. D. Fabbriche and C. Melandri, Microstructure and Properties
of ZrB2-Based Ceramics, J. Mater. Process. Manuf. Sci., 9 [2] 156-70 (2000).
[78] J. Groza and A. Zavaliangos, Sintering Activation by External Electrical Field, Mater. Sci. Eng. A,
287 [2] 171-7 (2000).
[79] J. Groza, M. Garcia and J. A. Schneider, Surface Effects in Field-Assisted Sintering, J. Mater. Res.,
16 [1] 286-92 (2001).
[80] V. Medri, F. Monteverde, A. Balbo and A. Bellosi, Comparison of ZrB2-ZrC-SiC Composites
Fabricated by Spark Plasma Sintering and Hot-Pressing, Adv. Eng. Mater., 7 [3] 159-63 (2005).
[81] S.-Q. Guo, Y. Kagawa, T. Nishimura, D. Chung and J.-M. Yang, Mechanical and Physical
Behavior of Spark Plasma Sintered ZrC–ZrB2–SiC Multiphase Composites, J. Eur. Ceram. Soc., 28 [6]
1279-85 (2008).
[82] H. Kaga, E. M. Heian, Z. A. Munir, C. Schmalzried and R. Telle, Synthesis of Hard Materials by
Field Activation: the Synthesis of Solid Solutions and Composites in the TiB 2-WB2-CrB2 System, J. Am.
Ceram. Soc., 84 [12] 2764-70 (2001).
[83] A. M. Locci, R. Orru, G. Cao and Z. A. Munir, Simultaneous Spark Plasma Synthesis and
Densification of TiC–TiB2 Composites, J. Am. Ceram. Soc., 89 [3] 848-55 (2006).
[84] D. Sciti, L. Silvestroni and A. Bellosi, Fabrication and Properties of HfB2-MoSi2 Composites
Produced by Hot Pressing and Spark Plasma Sintering. J. Mater. Res., 21 [6] 1460-66 (2006).
[85] D. Sciti, S. Guicciardi and M. Nygren, Densification and Mechanical Behavior of HfC and HfB 2
Fabricated by Spark Plasma Sintering, J. Am. Ceram. Soc., 91 [5] 1433-40 (2008).

63
 [86] A. K. Khanra, L. C. Pathak, S. K. Mishra and M. M. Godkhindi, Self-Propagating-High-
Temperature Synthesis (SHS) of Ultrafine ZrB2 Powder, J. Mater. Sci. Lett., 22 [17] 1189-91 (2003).
[87] N. Bertolino, M. Monagheddu, A. Tacca, P. Giuliani, C. Zanotti, F. Maglia and U. A. Tamburini,
Self-Propagating High-Temperature Synthesis of Functionally Graded Materials as Thermal Protection
Systems for High-Temperature Applications, J. Mater. Res., 18 [2] 448-55 (2003).
[88] G.-J. Zhang, Z.-Y. Deng, N. Kondo, J.-F. Yang and T. Ohji, Reactive Hot Pressing of ZrB2-SiC
Composites, J. Am. Ceram. Soc., 83 [9] 2330-32 (2000).
[89] W.-W. Wu, G.-J. Zhang, Y.-M. Kan and P.-L. Wang, Reactive Hot Pressing of ZrB2-SiC-ZrC Ultra
High-Temperature Ceramics at 1800°C, J. Am. Ceram. Soc., 89 [9] 2967-9 (2006).
[90] J. W. Zimmermann, G. E. Hilmas, W. G. Fahrenholtz, F. Monteverde and A. Bellosi, Fabrication
and Properties of Reactively Hot Pressed ZrB2-SiC Ceramics, J. Eur. Ceram. Soc., 27 [7] 2729-36 (2007).
[91] F. Monteverde, Progress in the Fabrication of Ultra-High-Temperature Ceramics: In Situ
Synthesis, Microstructure and Properties of a Reactive Hot-Pressed HfB2-SiC Composite, Compos. Sci.
Technol., 65 [11-12] 1869-79 (2005).
[92] U. Anselmi-Tamburini, Y. Kodera, M. Gasch, C. Unuvar, Z. A. Munir, M. Ohyanagi and S. M.
Johnson, Synthesis and Characterization of Dense Ultra-High Temperature Thermal Protection Materials
Produced by Field Activation Through Spark Plasma Sintering (SPS): 1. Hafnium Diboride, J. Mater. Sci.,
41 [10] 3097-104 (2006).
[93] B. Cech, P. Oliverius and J. Sejbal, Sintering of Zirconium Boride with Activating Additions,
Powder Metall., 8 [15] 142-51 (1965).
[94] P. S. Kislui and M. A. Kuzenkova, Regularities of Sintering of Zirconium Diboride-Molybdenum
Alloys, Sov. Powder Metall. Met. Ceram., 5 360-5 (1966).
[95] M.-A. Einarsrud, E. Hagen, G. Pettersen and T. Grande, Pressureless Sintering of Titanium
Diboride with Nickel, Nickel Boride, and Iron Additives, J. Am. Ceram. Soc., 80 [12] 3013-20 (1997).
[96] D. Sciti, M. Brach and A. Bellosi, Oxidation Behavior of a Pressureless Sintered ZrB2-MoSi2
Ceramic Composites, J. Mater. Res., 20 [4] 922-30 (2005).
[97] D. Sciti, S. Guicciardi, A. Bellosi and G. Pezzotti, Properties of a Pressureless-Sintered ZrB2-MoSi2
Ceramic Composite, J. Am. Ceram. Soc., 89 [7] 2320-2 (2006).
[98] Y. Yan, Z. Huang, S. Dong and D. Jiang, Pressureless Sintering of High Density ZrB 2-SiC Ceramic
Composites, J. Am. Ceram. Soc., 89 [11] 3589-92 (2006).
[99] S. Zhu, W. G. Fahrenholtz, G. E. Hilmas and S. C. Zhang, Pressureless Sintering of Zirconium
Diboride Using Boron Carbide and Carbon Additions, J. Am. Ceram. Soc., 90 [11] 3660-63 (2007).
[100] W. G. Fahrenholtz, G. E. Hilmas, S. C. Zhang and S. Zhu, Pressureless Sintering of Zirconium
Diboride: Particle Size and Additive Effects. J. Am. Ceram. Soc., 91 [5] 1398-404 (2008).
[101] S. Zhu, W. G. Fahrenholtz, G. E. Hilmas and S. C. Zhang, Pressureless Ssintering of Carbon-
Coated Zirconium Diboride Powders, Mater. Sci. Eng. A, 459 [1-2] 167-71 (2007).
[102] A. L. Chamberlain, W. G. Fahrenholtz and G. E. Hilmas, Pressureless Sintering of Zirconium
Diboride, J. Am. Ceram. Soc., 89 [2] 450-6 (2006).
[103] M. M. Opeka, I. G. Talmy, E. J. Wuchina, J. A. Zaykoski and S. J. Causey, Mechanical, Thermal
and Oxidation Properties of Refractory Hafnium and Zirconium Compounds, J. Europ. Ceram. Soc.,
19 [13-14] 2405-14 (1999).
[104] E. V. Clougherty, K. E. Wilkes and R. P. Tye, Research and Development of Refractory Oxidation
Resistant Diborides, Part II, Vol. V: Thermal, Physical, Electrical and Optical Properties, AFML-TR-68-
190, ManLabs lnc., Cambridge, MA, (1969).
[105] R. G. Munro, Material Properties of a Sintered -SiC, J. Physical and Chemical Reference Data,
26 [5] 1195-203 (1997).
[106] D. E. Wiley, W. R. Manning and O. Hunter Jr., Elastic Properties of Polycrystalline TiB2, ZrB2
and HfB2 From Room Temperature to 1300 K, J. Less-Common Met., 18 [2] 149-57 (1969).
[107] H. E. Exner, Physical and Chemical Nature of Cemented Carbides, Int. Metall. Rev., 24 [4] 149-73
(1979).
[108] W. H. Rhodes, E. V. Clougherty and D. Kalish, Research and Development of Refractory
Oxidation Resistant Diborides, Part II, Vol. IV: Mechanical Properties, AFML-TR-68-190, ManLabs lnc.,
Cambridge, MA, (1970).
[109] R. Pampuch, Ceramic Materials: An Introduction to Their Properties, Elsevier Scientific
Publishers/PWN-Polish Scientific Publishers, New York/Warsaw, Poland, 172-90 (1976).
[110] S.-Q. Guo, N. Hirosaki, Y. Yamamoto, T. Nishimura and M. Mitomo, Hot-Press Sintering Silicon
Nitride with Lu2O3 Addition: Elastic Moduli and Fracture Toughness, J. Eur. Ceram. Soc., 23 [3] 537-45
(2003).

64
 [111] R. W. Rice, Particle Effects on Elastic Properties, Crack Propagation, and Fracture Toughness
of Ceramic Composites at ∼ 22°C, Mechanical Properties of Ceramics and Composites. Marcel Dekker,
Inc., New York, 457 (2000).
[112] E. A. Dean and J. A. Lopez, Empirical Dependence of Elastic Moduli on Porosity for Ceramic
Materials, J. Am. Ceram. Soc., 66 [5] 366-70 (1983).
[113] C. W. Bert, Prediction of Elastic Moduli of Solid with Oriented Porosity, J. Mater. Sci., 20 [6]
2220-4 (1985).
[114] H. Jeong and D. K. Hsu, Quantitative Estimation of Material Properties of Porous Ceramics by
Means of Composite Micromechanics and Ultrasonic Velocity, NDT Int., 29 [2] 95-101 (1996).
[115] S.-Q. Guo, N. Hirosaki, Y. Yamamoto, T. Nishimura and M. Mitomo, Improvement of High-
Temperature Strength of Hot-Pressed Sintering Silicon Nitride with Lu2O3 Addition, Scripta Mater., 45 [7]
867-74 (2001).
[116] L. Kaufman, E. V. Clougherty and J. B. Berkowitz-Mattuck, Oxidation Characteristics of Hafnium
and Zirconium Diboride, Trans. TMS-AIME, 239 458-66 (1967).
[117] M. W. Chase Jr, NIST-JANAF Thermochemical Tables, 4th edn, Woodbury, New York, American
Chemical Society and the American Institute of Physics, (1998).
[118] FactSage 6 1, Thermfact and GTT-Technologies, Montreal, QC, Canada, (2009).
[119] J. B. Berkowitz-Mattuck, J. Electrochem. Soc., 113 908-14 (1966).
[120] H. C. Graham, H. H. Davis, I. A. Kvernes and W. C. Tripp, Ceramics in Severe Environments,
ed. W. W. Kriegel and H. Palmour III, New York, Plenum Press, 35-48 (1971).
[121] T. A. Parthasarathy, R. A. Rapp, M. Opeka and R. J. Kerans, A Model for the Oxidation of ZrB2,
HfB2 and TiB2, Acta Mater., 55 [17] 5999-6010 (2007).
[122] T. A. Parthasarathy, R. A. Rapp, M. Opeka and R. J. Kerans, A Model for Transitions in Oxidation
Regimes of ZrB2, Mater. Sci. Forum, 595-8 PART 2 823-32 (2008).
[123] T. A. Parthasarathy, R. A. Rapp, M. Opeka and R. J. Kerans, Effects of Phase Change and Oxygen
Permeability in Oxide Scales on Oxidation Kinetics of ZrB2 and HfB2, J. Am. Ceram. Soc., 92 [5] 1079-86
(2009).
[124] W. G. Fahrenholtz and G. E. Hilmas, Oxidation of Ultra-High Temperature Transition Metal
Diboride Cceramics, Intern. Mater. Rev., 57 [1] 61-72 (2012).
[125] A. K. Kuriakose and J. L. Margrave, J. Electrochem. Soc., 111 827-31 (1966).
[126] C. B. Bargeron, R. C. Benson, R. W. Newman, A. N. Jette and T. E. Phillips, Oxidation
Mechanisms of Hafnium Carbide and Hafnium Diboride in the Temperature Range 1400 to 2100°C, John
Hopkins APL Technical Digest (Applied Physics Laboratory), 14 [1] 29-36 (1993).
[127] W. C. Tripp and H. C. Graham, Solid State Sci., 118 1195-9 (1971).
[128] R. J. Irving and I. G. Worsley, The Oxidation of Titanium Diboride and Zirconium Diboride at
High Temperatures, J. Less Common Met., 16 [2] 103-12 (1968).
[129] M. M. Opeka, I. G. Talmy and J. A. Zaykoski, Oxidation-Based Materials Selection for 2000°C +
Hypersonic Aerosurfaces: Theoretical Considerations and Historical Experience, J. Mater. Sci., 39 [19]
5887-904 (2004).
[130] W. G. Fahrenholtz, The ZrB2 Volatility Diagram, J. Am. Ceram. Soc., 88 [12] 3509-12 (2005).
[131] S. C. Zhang, G. E. Hilmas and W. G. Fahrenholtz, Improved Oxidation Resistance of Zirconium
Diboride by Tungsten Carbide Additions, J. Am. Ceram. Soc., 91 [11] 3530-5 (2008).
[132] M. Gasch, D. Ellerby, E. Irby, S. Beckman, M. Gusman and S. Johnson, Processing, Properties
and Arc Jet Oxidation of Hafnium Diboride/Silicon Carbide Ultra High Temperature Ceramics, J. Mater.
Sci., 39 [19] 5925-37 (2004).
[133] F. Monteverde, R. Savino, M. De Stefano Fumo and A. Di Maso, Plasma Wind Tunnel Testing
of Ultra-High Temperature ZrB2-SiC Composites Under Hypersonic Re-Entry Conditions, J. Eur. Ceram.
Soc., 30 [11] 2313-21 (2010).
[134] A. Paul, D. D. Jayaseelan, S. Venugopal, E. Zapata-Solvas, J. Binner, B. Vaidhyanathan, A.
Heaton, P. Brown and W. E. Lee, UHTC Composites for Hypersonic Applications, Am. Ceram. Soc. Bull.,
91 [1] 22-29B (2011).
[135] J. Marschall and D. G. Fletcher, High-Enthalpy Test Environments, Flow Modeling and In Situ
Diagnostics for Characterizing Ultra-High Temperature Ceramics, J. Eur. Ceram. Soc., 30 [11] 2323-36
(2010).
[136] R. Perkins, L. Kaufman and H. Nesor, Stability Characterization of Refractory Materials Under
Velocity Atmospheric Flight Conditions, Experimental Results of High Velocity Cold Gas/Hot Wall Test,
Part III, Vol. II, AFML-TR-68-84, ManLabs Inc., Cambridge, MA, (1969).
[137] L. Kaufman and H. Nesor, Stability Characterization of Refractory Materials Under High Velocity
Atmospheric Flight Conditions, Part II, Vol. II, AFML-TR-69-84, ManLabs Inc., Cambridge, MA, (1969).

65
 [138] E. V. Clougherty, E. T. Peters and D. Kalish, Diboride Materials, Candidates for Aerospace
Applications, Proc. 15th SAMPE Symp. on ‘Materials and processes for the 1970s’, Los Angeles, CA, USA,
SAMPE, 297-308 (1969).
[139] E. J. Opila and M. C. Halbig, Oxidation of ZrB2-SiC, Ceram. Eng. Sci. Proc., 22 [3] 221-8 (2001).
[140] S. Gangireddy, S. N. Karlsdottir, S. J. Norton, J. C. Tucker and J. W. Halloran, In Situ Microscopy
Observation of Liquid Flow, Zirconia Growth, and CO Bubble Formation during High Temperature
Oxidation of Zirconium Diboride-Silicon Carbide, J. Eur. Ceram. Soc., 30 [11] 2365-74 (2010).
[141] S. N. Karlsdottir and J. W. Halloran, Rapid Oxidation Characterization of Ultra-High Temperature
Ceramics, J. Am. Ceram. Soc., 90 [10] 3233-8 (2007).
[142] S. N. Karlsdottir, J. W. Halloran and C. E. Henderson, Convection Patterns in Liquid Oxide Films
on ZrB2-SiC Composites Oxidized at a High Temperature, J. Am. Ceram. Soc., 90 [9] 2863-7 (2007).
[143] S. N. Karlsdottir, J. W. Halloran and A. N. Grundy, Zirconia Transport by Liquid Convection
during Oxidation of Zirconium Diboride-Silicon Carbide, J. Am. Ceram. Soc., 91 [1] 3652-8 (2008).
[144] O. N. Grigoriev, B. A. Galanov, V. A. Lavrenko, A. D. Panasyuk, S. M. Ivanov, A. V. Koroteev
and K. G. Nickel, Oxidation of ZrB2-SiC-ZrSi2 Ceramics in Oxygen, J. Eur. Ceram. Soc., 30 [11] 2397-405
(2010).
[145] C. M. Carney, P. Mogilvesky and T. A. Parthasarathy, Oxidation Behavior of Zirconium Diboride
Silicon Carbide Produced by the Spark Plasma Sintering Method, J. Am. Ceram. Soc., 92 [9] 2046-52
(2009).
[146] F. Monteverde, The Thermal Stability in Air of Hot-Pressed Diboride Matrix Composites for uses
at Ultra-High Temperatures, Corros. Sci., 47 [8] 2020-33 (2005).
[147] D. R. Uhlmann and N. J. Kreidl, Glass: Science and Technology, vol. 3: Viscosity and Relaxation,
Academic Press, New York, 412 (1986).
[148] I. G. Talmy, J. A. Zaykoski, M. M. Opeka and S. Dallek, High Temperature Corrosion and
Materials Chemistry III, 144-58 (2001).
[149] E. Opila, S. Levine and J. Lorincz, Oxidation of ZrB2- and HfB2-Based Ultra-High Temperature
Ceramics: Effect of Ta Additions, J. Mater. Sci., 39 [19] 5969-77 (2004).
[150] X. Zhang, X. Li, J. Han, W. Han and C. Hong, Effects of Y2O3 on Microstructure and Mechanical
Properties of ZrB2-SiC Ceramics, J. Alloys Compd., 465 [1-2] 506-11 (2008).
[151] A. Spring, W.-M. Guo, G.-J. Zhang, P.-L. Wang and V. D. Krstic, Fabrication and
Characterization of ZrB2-Based Ceramic Using Synthesized ZrB2-LaB6 Powder, J. Am. Ceram. Soc., 91 [8]
2763-5 (2008).
[152] I. G. Talmy, J. A. Zaykoski, M. M. Opeka and A. H. Smith, Properties of Ceramics in the System
ZrB2-Ta5Si3, J. Mater. Res., 21 [10] 2593-9 (2006).
[153] D. Sciti, L. Silvestroni, G. Celotti, C. Melandri and S. Guicciardi, Sintering and Mechanical
Properties of ZrB2-TaSi2 and HfB2-TaSi2 Ceramic Composites, J. Am. Ceram. Soc., 91 [10] 3285-91 (2008).
[154] J. C. A. Bronson and J. Chessa, An Evaluation of Vaporizing Rates of SiO2 and TiO2 as Protective
Coatings for Ultrahigh Temperature Ceramic Composites, J. Am. Ceram. Soc., 91 [5] 1448-52 (2008).
[155] W. Guo, J. Vleugels, G. Zhang, P. Wang and O. Van der Biest, Effects of Re2O3 (Re = La, Nd, Y
and Yb) Addition in Hot-Pressed ZrB2-SiC Ceramics, J. Eur. Ceram. Soc., 29 [14] 3063-8 (2009).
[156] D. D. Jayaseelan: unpublished research, Imperial College London, 2010.
[157] S. C. Zhang, W. G. Fahrenholtz and G. E. Hilmas, Oxidation of ZrB2 and ZrB2-SiC Ceramics with
Tungsten Additions, ECS Trans., 16 [44] 137-45 (2009).
[158] S. K. Woo, C. H. Kim and E. S. Kang, Fabrication and Microstructural Evaluation of ZrB 2/ZrC/Zr
Composites by Liquid Infiltration, J. Mat. Sci., 29 [20] 5309-15 (1994).
[159] R. German, Sintering Theory and Practice, John Wiley and Sons Inc., New York, 385-9 (1996).
[160] M. N. Rahaman, Ceramic Processing and Sintering, Marcel Dekker Inc., New York, 526-30
(1995).
[161] L. R. Radovic, P. M. Walker and P. A. Thrower, Chemistry and Physics of Carbon: A Series
of Advances, Marcel Dekker Inc., New York, 149-61 (1965).
[162] M. N. Rahaman, Ceramic Processing and Sintering, 2nd Edition, Marcel Dekker Inc., New York,
15-7 (2003).
[163] J. Narciso, E. Louis and C. Garcia-Cordovilla, Pressure Infiltration in a Reactive System: Packed
SiC Particulates Infiltrated by Pure Silver with Dissolved Oxygen, Acta Mater., 45 [12] 5111-8 (1997).
[164] K. Mlungwane, The Development of a Diamond-Silicon Carbide Composite, PhD thesis,
University of the Witwatersrand, South Africa, (2008).
[165] W. B. Hillig, Making Ceramic Composites by Melt Infiltration, Am. Ceram. Bull., 73 [4] 56-62
(1994).
[166] L. Hozer, J.-R. Lee and Y.-M. Chiang, Reaction-Infiltrated, Net-Shape SiC Composites, Mater.
Sci. Eng., 195 [C] 131-43 (1995).

66
 [167] R. W. Olesinski and G. J. Abbaschian, The C-Si (Carbon-Silicon) system, Bulletin of Alloy Phase
Diagrams, 5 [5] 486-9 (1984).
[168] E. Fitzer and R. Gadow, Investigations of the Reactivity of Different Carbons with Liquid Silicon,
Proc. Of International Symposium on Ceramic Components for Engine, Japan, S. Somiya, 561-72 (1983).
[169] C. Pantea, G. A. Voronin, T. W. Zerda, J. Zhang, L. Wang, Y. Wang, T. Uchida and Y. Zhao,
Kinetics of SiC Formation during High P-T Reaction between Diamond and Silicon, Diamond & Related
Materials, 14 [10] 1611-5 (2005).
[170] M. H. Hon, R. F. Davis and D. E. Newbury, Self-Diffusion of 30Si in Polycrystalline -SiC,
J. Mater. Sci., 15 [8] 2073-80 (1980).
[171] T. Hase, H. Suzuki and T. Iseki, Rise in Temperature of SiC Pellet Involving Reaction Sintering,
J. Nucl. Mater., 59 [1] 42-8 (1976).
[172] G. R. Sawyer and T. F. Page, Microstructural Characterization of “REFEL” (Reaction-Bonded)
Silicon Carbides, J. Mater. Sci. I, 13 [4] 885-904 (1978).
[173] J. N. Ness and T. F. Page, Microstructural Evolution in Reaction-Bonded Silicon Carbide, J. Mate.
Sci., 21 [4] 1377-97 (1986).
[174] H. Zhou and R. N. Singh, Kinetics Model for the Growth of Silicon Carbide by the Reaction
of Liquid Silicon with Carbon, J. Am. Ceram. Soc., 78 [9] 2456-62 (1995).

67
2. Si-SiC-ZrB2 ceramics by silicon reactive
infiltration

2.1 Introduction
Molten silicon reactive infiltration is a manufacturing technique in which a carbon bound
porous body, usually called “preform”, is infiltrated with silicon in the liquid or vapor
phase which, through an in situ chemical reaction, converts into silicon carbide. Si-SiC
ceramics were first obtained by Hillig et al. [1] by infiltrating carbonaceous material with
molten Si in vacuum (10-2 mbar) at temperatures ranging from 1450°C and 1600°C.
The related chemical reaction [2] is:
Siliquid + Csolid = SiCsolid [H = -73 kJ mol-1] (2.1)

At these temperatures molten Si first infiltrates the porous carbon body by capillarity then
reacts with it. In Ceramic Matrix Composite (CMC) manufacturing, this process is usually
known as Melt Infiltration (MI) or Liquid Silicon Infiltration (LSI) [3].
Si-SiC CMC, are well known for their good thermo-mechanical properties [3] and are one
of the few CMC “mass products” (e.g. ceramic brakes, furnace parts) on the market.
The “price to pay” is for the presence of unreacted silicon (515 vol.%) percolated into the
structure. Indeed, in order to allow molten Si to fully infiltrate the preform in reasonable
time, the volume of its porosity must be higher than the volume of silicon necessary for
the stoichiometric reaction with carbon [4]. Si-SiC are fully dense materials and thus are
oxidation resistant, but their use is limited by their working temperatures [5, 6].
For applications at higher temperatures other ceramics, usually known as ultra high
temperature ceramics (UHTCs), are used including the transition metal diborides of group
IV [7]. ZrB2 and HfB2 have been widely studied for their high melting points, hardness,
chemical and thermal stabilities [8]. Zirconium diboride can be of great interest in some
applications (e.g. aerospace) not only because of its lower density with respect to hafnium
diboride. ZrB2 powders have been densified into compact bodies by hot pressing,
pressureless sintering, reactive routes and spark plasma sintering [9, 10]. In general these
processes involve the use of expensive equipment, because of the processing conditions,
and present limitations in product shaping. ZrB2 has also been processed with additives to
tailor its properties; among these, silicon carbide is one of the most widely employed
[9, 10]. The flexural strength and thermal conductivity of ZrB2 ceramics can be increased
68
by adding SiC particles [10, 11]. ZrB2-SiC ceramics can maintain the “protective” passive
oxidation regime over a wide range of temperatures. Adding SiC (up to 30 vol.%) to ZrB2
not only extends the oxidation temperature range, thanks to the protective behavior of the
formed mixed oxide scale [12], but it also allows the ceramic to rapidly regain its
protective behavior if scale detaches during excessive temperature exposure [10]. Peng et
al. [13] showed that higher concentrations of SiC resulted in thinner glassy surface layers,
with lower concentrations of embedded zirconia, and thinner zirconium oxide under layers
which could be beneficial or detrimental, depending on the application.
There are few works in the literature which try to produce ZrB2-SiC based ceramics with
alternative routes and they are mainly targeted at producing ceramic matrices for CMCs
[14, 15].
To the authors knowledge, literature reports two studies that explore “mild processes” in
order to obtain dense Si-SiC-ZrB2 composites at relatively a low temperature (1500°C)
and under pressureless conditions. Blum et al., using molten silicon reactive infiltration
[16], developed thick UHTC protective coatings on the top of lightweight C-SiC, SiC-SiC
and C-C fiber reinforced ceramics. Zhou et al. produced coatings on C/C composites by
vapor silicon infiltration [17].
In this chapter a methodology to produce Si-SiC-ZrB2 bulk ceramics, taking advantage of
the reactive infiltration technique, is proposed. Preforms with different contents of SiC
and ZrB2 were compacted with phenolic resin, pyrolysed and finally infiltrated with
molten silicon at temperature above its melting point. The process was intentionally kept
as simple as possible in order to maximize its cost efficiency.
Following the historical path of Si-SiC ceramics our long-term goal, starting from this
work, is to develop Si-SiC-ZrB2 ceramic matrix composites by silicon reactive infiltration.

2.2 Experimental procedure

In order to study how the SiC/ZrB2 volume ratios affect composite manufacturing steps
and the resulting microstructure, three different constituent compositions were prepared
for this study. The compounds were compacted, pyrolysed in an inert atmosphere, and
further infiltrated in vacuum with molten silicon.

69
2.2.1 Materials

The ceramic powders employed in this work were:

 -Silicon carbide (Grade UF15 Stark Ag, D) with an average particle size d50 of
0.55 m by Laser Diffraction and a specific area of 14÷16 m2/g.
 Zirconium diboride (grade A Stark Ag D) average particle size d50 of 0.35.0 m
by Laser Diffraction.
The powders were pre-mixed to obtain a bimodal particle size distribution [11].
To produce the green bodies, a micronized ( 64 m diameter) phenolic novolac powder
(Hexion, USA) was mixed with the powders.
Samples were mixed maintaining a constant ratio between the solid charge (powders) and
the plastic binder (phenolic resin). For the solid charge, three SiC-ZrB2 powder mixtures
were prepared (Tab. 2.I). Pyrolysed samples presented a similar porosity volume because
the ratio of ceramic powders and phenolic resin was kept equal. A volume of around 50%
of phenolic resin allowed to the compacted enough green and to the self-standing
pyrolysed porous body with a sufficient amount of glassy carbon [18, 19] to be prepared.

Solid content (vol.%) Binder (vol.%)

51.44 48.56
Phenolic resin
Sample SiC (vol.%) ZrB2 (vol.%)
(vol.%)
On solid On On solid On
content total content total
S100 100.00 51.44 0.00 0.00 48.56
S50Z50 50.00 25.72 50.00 25.72 48.56
S25Z75 25.00 12.86 75.00 38.58 48.56
Table 2.I: SiC ( = 3.21 g/cm3), ZrB2 ( = 6.10 g/cm3) and phenolic resin ratios for the
produced samples.

70
2.2.2 Green forming

The powders and phenolic resin were first dried separately in a vacuum oven for 30 min at
40°C and 50 mbar residual pressure. First the solid charge was dry mixed for 30 min by
slow ball milling in a sealed container, then milling was continued for a further 30 min
with phenolic powder. The compound was immediately hot pressed with a uniaxial press
for 150 s at 180°C and at a pressure of 56 MPa. In plane dimensions the die and thus the
sample was 90 x 40 mm2. The amount of powder put into the die was calculated in order
to produce green preforms of 45 mm thick.
This first set of samples was over packed to be successfully infiltrated with Si.
After pyrolysis, silicon was not able to infiltrate these bodies even for long dwelling times.
In a second set, samples were then pressed under the same conditions with the exception
of the temperature, which was lowered to 60°C. At this temperature phenolic resin
becomes more plastic than liquid. This allowed the compaction of green bodies which,
after pyrolysis, were porous enough to be fully silicon infiltrated.

2.2.3 Pyrolysis

Green samples were pyrolysed in an electrical furnace inside a steel retort under inert
atmosphere. During pyrolysis argon was injected into the retort at 300 SCCM. The heat
ramp is reported in table 2.II.
After pyrolysis, samples were then measured, weighted and new volume fractions
calculated (Tab. 2.III). The plain phenolic resin carbon yield was 70% while the other
powders were unaffected by this thermal treatment. An overall volumetric shrinking was
similar for all the samples (57%). The apparent density was then calculated dividing the
new mass by the new volume. These values were then used to calculate samples porosity.
It is worth noting that the bodies did not crumble during pyrolysis, keeping high
mechanical resistance after this processing step.

71
 Temperature [°C] Temperature rate [°C/h] Ramp Time [h] Total Time [h]

21÷500 40 12 12
500÷1000 60 8 20
1000 dwell 1 21
1000÷21 Power OFF ~ 24 ~ 45

Table 2.II: Pyrolysis heat ramp.

ZrB2 SiC Res. carbon Porosity

Sample
(vol.%) (vol.%) (vol.%) (vol.%)
S100 0.0 38.3 15.0 46.7
S50Z50 18.8 19.2 15.9 46.3
S25Z75 31.6 10.6 19.4 38.2
Table 2.III: Constituent materials calculated volume fractions after pyrolysis.

2.2.4 Silicon reactive infiltration

After pyrolysis, samples were placed on a boron nitride painted graphite plate. Carbon
wicks were placed between the samples and the plate, to drain excess molten silicon.
Silicon flakes were placed onto the samples. The amount of silicon was much higher than
needed to react with phenolic derived carbon and to infiltrate the remaining porosity.
However, based on our experience, this is necessary to fully infiltrate the preforms.
Silicon infiltration was performed at 1500°C with a 10-2 mbar residual pressure.
The furnace was quickly brought to process temperature and held at that temperature for
2h. The excess silicon reacted with the wicks or remained outside the sample
as per Fig. 2.1.

2.2.5 Machining

Sample plates (Fig. 2.1) were first ground with a diamond mill on the outer surfaces, their
thickness was consequently reduced in order to be sure to have a genuine Si-SiC-ZrB2
system on the outer surface. The samples were then cut with a diamond saw into several
pieces for characterization.

72
2.2.6 High temperature oxidation

In order to compare oxidized Si-SiC-ZrB2 with other works [11, 16, 17, 20], specimens
were oxidized in a resistance heated tubular furnace at atmospheric pressure. They were
placed in an alumina boat which was then positioned at the centre of an alumina tube.
The furnace was brought to 1500°C with an heating rate of 10°C/min and held at that
temperature for 2h. A dry air flow of 200 SCCM was maintained in the furnace during the
tests.

2.2.7 Characterisation

Samples were characterized by several techniques.

Vickers micro-hardness measurements were carried out with a micro-durometer (Leitz
Wetzlar D). Microstructure evaluations on both oxidized and non-oxidized samples were
performed with a Field Emission Scanning Electron Microscope (FESEM Zeiss Supra
25 D). Phase analysis was performed, during micrographs acquisition, with a Scandium
5.0 Soft Imaging System GmbH. XRD data were collected with a Philips PW3830 X-ray
generator system.

2.3 Results and discussion

The following are the results of an initial characterization study aiming at evaluating
Si-SiC-ZrB2 properties and comparing them with a Si-SiC (S100) compound produced
with the same methodology and with literature data. Si-SiC-ZrB2 oxidation behavior was
compared with similar systems adopted as protective coatings [16, 17] and with ZrB2-SiC
[12, 21, 22].

73
 Figure 2.1: Samples during processing steps.

2.3.1 Microstructure

Sample fracture surfaces after pyrolysis were analyzed by SEM. Well dispersed ZrB2
powder is clearly visible as well as residual carbon around the big pores (Fig. 2.2(a)).
Much finer SiC powders, in an agglomerate state (Fig. 2.2(b)) are present, while they are
hardly discernible inside the pyrolysed carbon. Samples show a twofold pore size: bigger
pores (Fig. 2.2(c)) correspond to the volume formerly occupied by the phenolic powders,
smaller pores to the intra particle porosity. It was this bimodal porosity which allowed the
silicon infiltration of rather thick (45 mm) samples. During the pyrolysis step the
phenolic decomposed byproducts filled the intra particle porosity, keeping the ceramic
body compact.

Figure 2.2: S25Z75 sample after

pyrolysis (fracture surface). Beside the
dispersed ZrB2 which is visible
throughout the fracture surface, the
micrograph shows (magnification:
1000X, voltage: 15 kV, working
distance: 6 mm, aperture size: 30 m):
(a) glassy carbon, (b) SiC powders
agglomerates and (c) big pores.

74
The patterns in Fig. 2.3 show the crystalline phases detected by the instrument in the Si
infiltrated samples. As expected, in S100 both silicon and silicon carbide are present.
S50Z50 and S25Z75 samples show three crystal phases, namely: silicon, silicon carbide
and zirconium diboride. Their presence is confirmed by micro hardness measurements
whose data are reported in table 2.IV. Average results are lower than ZrB2-SiC systems
[23, 24], due to the presence of silicon, while the hardness of each phase is in good
agreement with literature data [25].

Figure 2.3: XRD patterns of S100, S50Z50 and S25Z75 sample, respectively.

Fig. 2.4 shows the microstructures of S100, S50Z50 and S25Z75 samples after reactive
silicon infiltration. Micrographs confirm the tendency of the SiC powders to agglomerate,
due to the powder size and the fabrication method. Agglomeration is the cause of the
small size pores clearly visible in Fig. 2.4 (a and c).
The micrographs were taken in back-scattered electrons (BSE) mode to enhance the
contrast between each phase. In Fig. 2.4(a) two phases with different gray tones are
distinguishable, while in Fig. 2.4 (b and c) a third white phase appears. An Energy-
dispersive X-Ray spectroscopy (EDS or EDX) analysis on these three phases revealed that
the light gray zones are made of Si, the dark gray zones of SiC and the white grains of
ZrB2. A further confirmation of the phases present in the samples comes from the
elemental mapping reported in Fig. 2.5 for the S25Z75 sample. Furthermore, the interfaces

75
between the different phases, especially for ZrB2 compared with the rest of the matrix, are
well defined.

Multiphase indentations Single phase indentations

Sample Average value SD Si SiC ZrB2
[GPa] [GPa] [GPa] [GPa] [GPa]

S100 16.2 3.2 9.9 24.0 -

S50Z50 17.6 0.8 9.5 26.1 15.6

S25Z75 14.2 2.4 9.3 24.2 14.0

Table 2.IV: Vickers micro-hardness (load 300 g).

Figure 2.4: SEM images of S100 (a), S50Z50 (b) and S25Z75 (c) obtained with back
scattered electrons.

XRD patterns in Fig. 2.3 and the micrographs in Fig. 2.4 (b and c) also show that, during
molten infiltration, silicon did not react with ZrB2 although it formed SiC with the
polymer derived carbon.
Silicon infiltrated the porous preforms regardless what SiC to ZrB2 amounts was into the
solid charge. Phenolic derived carbon was the real “neck bottle” in terms of wetting since
it presents poor wetting properties before reacting with silicon and forming SiC.
When SiC is formed, all Si contact angles, measured on bulk specimens, are quite similar
(Tab. 2.V). There are no data available on silicon wetting onto ZrB2. The suggestion [26]
of assuming a wetting behavior similar to TiB2 for ZrB2 (Tab. 2.V) was carried out.
This hypothesis was confirmed by the weight gains during reactive infiltration and by the
morphology analysis in Fig. 2.4 (b and c) in which it is evident that silicon wrapped the
ZrB2 particles. Infiltration was somehow hindered by SiC powder agglomeration (Fig. 2.4
(a and c)) since intra particle gaps were smaller inside the agglomerates.

76
Figure 2.5: Maps of the elements (b) silicon = red, (c) zirconium = white, (d) boron =
green present on the detail SEM image of sample S25Z75.

Surface material Equilibrium contact angle  [°] of molten silicon

Unreacted glassy carbon >120 [27, 28]
a)
Reacted glassy carbon 10÷40 [28] 40÷50 [29] 35÷40 [27]
a) b)
Reacted graphite 3÷35 [28] 5÷15 [29]
c) d) d) e)
SiC 38,5 ± 2 [30] 30 [29] 38 , 41.5 [31]
f)
ZrB2 15 [26]
a) Depending on surface roughness, b) pyrolitic graphite, c) -SiC, d) sintered, e) -SiC, f) datum is referred to TiB2.

Table 2.V: Literature contact angles of molten silicon on constituent materials.

Table 2.VI shows the phase volume percentages calculated from Fig. 2.4. SiC volume was
calculated as the sum of the -SiC powders and the SiC formed by reaction bonding
between glassy carbon and Si. Values in table 2.VI take into account the fact that the
porosity, after silicon infiltration, was completely filled by unreacted Si.
Considering that, during reactive infiltration, carbon and Si became SiC as per Eq. (2.1),
given the densities of SiC, ZrB2 (Tab. 2.I) and Si ( = 2.33 g/cm3), it was possible to
calculate Si-SiC-ZrB2 theoretical density and, after samples real density measurements,
to estimate porosity.

77
 Theoretical Experimental
Sample SiC phase ZrB2 phase Si phase theor. exp.
Porosity (%)
(vol.%) (vol.%) (vol.%) [g/cm3] [g/cm3]
S100 61.7 0.0 38.3 2.87 2.75 4.2%
S50Z50 43.8 18.8 37.4 3.43 3.30 3.8%
S25Z75 41.0 31.6 27.4 3.88 3.85 0.8%

Table 2.VI: Calculated phase volume fractions and overall densities on a ideally fully
infiltrated sample. Measured density and estimated porosity.

2.3.2 Oxidation behavior

Fig. 2.6 shows the SEM analysis of the Si-SiC specimens (S100). From a morphological
point of view, a continuous 2030 m external layer can hardly be detected, but it
becomes evident if oxygen mapping is performed (Fig. 2.6(b)). Even at higher
magnifications (Fig. 2.6(d)) the interface between oxidized and un-oxidized material is
undistinguishable without the help of the oxygen map (Fig. 2.6(e)), meaning that the
external oxidized layer is well adherent. The composition of the oxidized layer is
primarily SiO2, according to EDS analysis.
Fig. 2.7 shows a SEM analysis of the specimen with the highest amount of ZrB2
(S25Z75). Similarly to Fig. 2.6, it is rather difficult to distinguish an oxidized layer
(Fig. 2.7(a)), but one does become evident with the help of an oxygen map (Fig. 2.7(b)).
Scale thickness is around 2030 m and it is composed of two phases; ZrO2 (phase 3) and
SiO2 (phase 4). In the same figure the labels phase 1 and phase 2 are referred to ZrB2 and
Si-SiC, respectively. It is possible to point out that the difference between ZrB2 and ZrO2
is distinguishable also by a morphological point of view: ZrB2 particles have smooth
contours while ZrO2 particles in the oxidized material have irregular and broken contours.
Boron distribution is difficult to characterize due to the low sensitivity of energy-
dispersive spectroscopy to light elements but, as suggested by the literature on UHTCs
[11, 13, 20], boron is oxidized to B2O3 which together with SiO2 forms a borosilicate glass
that is reported to significantly increase the resistance to oxygen diffusion through the
surface.
Several authors have reported that oxidation of ZrB2-SiC at 1500°C in air produces a
structure that consists of four layers [12, 21, 22, 32]:

78
 1) continuous SiO2 on the outer surface;
2) ZrO2 embedded in amorphous SiO2;
3) SiC-depleted and ZrB2;
4) unaffected ZrB2-SiC.

Figure 2.6: SEM analysis of specimen S100 after oxidation. (a) General view with:
(b) oxygen map, (c) silicon map. (d) Details of the interface between oxidized and un-
oxidized material with: (e) oxygen map and (f) silicon map.

Figure 2.7: SEM analysis of specimen S25Z75 after oxidation. (a) General view with: (b)
oxygen map, (c) silicon map, (d) zirconium map. (e) Details of the interface between
oxidized and un-oxidized material with: (f) oxygen map, (g) silicon map and (h)
zirconium map.

79
Conversely to these findings, our oxidized Si-SiC-ZrB2 specimens showed two layers
(Fig. 2.7):
1) ZrO2 embedded in SiO2 and B2O3;
2) unaffected Si-SiC-ZrB2.
These results are comparable with other Si-SiC-ZrB2 systems employed as a protective
coating [16, 17] which were oxidized in similar conditions. Si-SiC-ZrB2 systems differ
from ZrB2-SiC by the relative amount of SiC, which is much higher in Si-SiC-ZrB2 and,
of course, by the presence of unreacted silicon. An oxidation campaign is currently
ongoing, in order to better understand Si-SiC-ZrB2 oxidation behavior.

2.4 Conclusions

This work shows that Si-SiC-ZrB2 composite bulks can be produced via silicon reactive
infiltration. Although the process was kept as simple as possible in order to prove its cost
efficiency, some features, like powder size selection and mixing, need to be improved in
order to avoid powder agglomeration. ZrB2 powders did not react with molten silicon and
remained a separate phase within the composite system. Preform porosity of samples after
pyrolysis was intentionally kept at  50% to ease Si infiltration. These high values affect
the corresponding quantity of unreacted silicon after reactive infiltration. Further work
will focus on reducing preform porosity and Si quantity.
Oxidized Si-SiC-ZrB2 presented a double layer structure:
1) ZrO2 embedded in SiO2 and B2O3;
2) unaffected Si-SiC-ZrB2.
The SiO2 and B2O3 layer seems to act as a better oxygen barrier for the unaffected
material. The typical SiC depleted region, present in oxidized ZrB2-SiC materials, was not
observed in this study.

80
2.5 References

[1] W. B. Hillig, R. L. Mehan, C. R. Morelock, V. J. DeCarlo and W. Laskow, Silicon-Silicon Carbide

Composites, Am. Ceram. Soc. Bull., 54 [12] 1054-6 (1975).
[2] M. W. Chase Jr., NIST-JANAF Thermo-chemical Tables, fourth edition, J. Phys. Chem. Ref. Data,
(Monograph 9) 1-1951 (1998).
[3] B. Heidenreich, Melt Infiltration Process in Ceramic Matrix Composites, in: W. Krenkel (Ed.),
Wiley-VCH, 113-9 (2008).
[4] Y. Wang, S. Tan and D. Jiang, The Effect of Porous Carbon Preform and the Infiltration Process on
the Properties of Reaction-Formed SiC, Carbon, 42 [8-9] 1833-9 (2004).
[5] N. S. Jacobson, Corrosion of Silicon-Based Ceramics in Combustion Environments, J. Am. Ceram.
Soc., 76 [1] 13-28 (1993).
[6] C. E. Ramberg and W. L. Worrell, Oxygen Transport in Silica at High Temperatures: Implications of
Oxidation Kinetics, J. Am. Ceram. Soc., 84 [11] 2607-16 (2001).
[7] R. A. Cutler, Engineering Properties of Borides, in ASTM Engineered Materials Handbook, vol. 4
Ceramics and Glasses, Schneider, S.J., technical chairman, 787-803 (1991).
[8] M. J. Gasch, D. T. Ellerby and S. Johnson, Ultra High Temperature Ceramic Composites, in: N. P.
Bansal (Ed.), Handbook of Ceramic Composites, Springer US, 197-224 (2005).
[9] W. G. Fahrenholtz, G. E. Hilmas, I. G. Talmy and J. A. Zaykoski, Refractory Diborides of Zirconium
and Hafnium, J. Am. Ceram. Soc., 90 [5] 1347-64 (2007).
[10] S.-Q. Guo, Densification of ZrB2-Based Composites and their Mechanical and Physical Properties:
A Review, J. Eur. Ceram. Soc., 29 [6] 995-1011 (2009).
[11] E. Eakins, D. D. Jayaseelan and W. E. Lee, Toward Oxidation-Resistant ZrB2-SiC Ultra High
Temperature Ceramics, Metall. Mater. Trans. A, 42 [4] 878-87 (2011).
[12] W. G. Fahrenholtz, Thermodynamic Analysis of ZrB2-SiC Oxidation: Formation of a SiC-Depleted
Region, J. Am. Ceram. Soc., 90 [1] 143-48 (2007).
[13] F. Peng, Y. Berta and R. F. Speye, Effect of SiC, TaB2 and TaSi2 Additives on the Isothermal
Oxidation Resistance of Fully Dense Zirconium Diboride, J. Mater. Res., 24 [5] 1855-67 (2008).
[14] N. Padmavathi, K. K. Ray, J. Subrahmanyam, P. Ghosal and S. Kumari, New Route to Process Uni-
Directional Carbon Fiber Reinforced (SiC + ZrB2) Matrix Mini-Composites, Mater. Sci., 44 [12] 3255-64
(2009).
[15] H. Hu, Q. Wang, Z. Chen, C. Zhang, Y. Zhang and J. Wang, Preparation and Characterization of
C/SiC-ZrB2 Composites by Precursor Infiltration and Pyrolysis Process, Ceram. Int., 36 [3] 1011-16 (2010).
[16] Y. D. Blum, J. Marschall, D. Hui and S. Young, Thick Protective UHTC Coatings for SiC-Based
Structures: Process Establishment, J. Am. Ceram. Soc., 91 [5] 1453-60 (2008).
[17] H. Zhou, L. Gao, Z. Wang and S. Dong, ZrB2-SiC Oxidation Protective Coating on C/C Composites
Prepared by Vapor Silicon Infiltration Process, J. Am. Ceram. Soc., 93 [4] 915-9 (2010).
[18] S. J. Wheeler, J. Sibold and I. Reimans, The Effect of Carbon on the Processing of SiC/SiCf
Composites, in: J.P. Singh, N.P. Bansal, M. Singh (Eds.), Advances in Ceramic Matrix Composites VIII,
John Wiley & Sons, 43-53 (2002).
[19] C. A. Nannetti, A. Borello, D. A. de Pinto, D. Carbone, A. Licciulli and A. Ortona, C Fiber
Reinforced Ceramic Matrix Composites by a Combination of CVI, PIP and RB, in: W. Krenkel, R. Naslain,
H. Schneider (Eds.), High Temperature Ceramic Matrix Composites, Wiley-VCH Verlag GmbH & Co.
KGaA, Weinheim, 368-74 (2006).
[20] A. Rezaie, W. G. Fahrenholtz and G. E. Hilmas, Evolution of Structure during the Oxidation of
Zirconium Diboride-Silicon Carbide in Air up to 1500°C, J. Eur. Ceram. Soc., 27 [6] 2495-501 (2007).
[21] E. J. Opila, S. Levine and J. Lorincz, Oxidation of ZrB2- and HfB2-Based Ultra High Temperature
Ceramics: Effect of Ta Additions, J. Mater. Sci., 39 [19] 5969-77 (2004).
[22] M. Gasch, D. Ellerby, E. Irby, S. Beckman, M. Gusman and S. Johnson, Processing, Properties,
and Arc Jet Oxidation of Hafnium Diboride/Silicon Carbide Ultra High Temperature Ceramics, J. Mater.
Sci., 39 [19] 5925-37 (2004).
[23] S. C. Zhang, G. E. Hilmas and W. G. Fahrenholtz, Mechanical Properties of Sintered ZrB2-SiC
Ceramics, J. Eur. Ceram. Soc., 31 [5] 893-901 (2011).
[24] S. C. Zhang, G. E. Hilmas and W. G. Fahrenholtz, Pressureless Sintering of ZrB2-SiC Ceramics, J.
Am. Ceram. Soc., 91 [1] 26-32 (2008).
[25] http://www.schunk-sik.com/en/schunk01.c.9111.de/all_productarticle.
[26] G. V. Samsonov, A. D. Panasiuk and M. S. Boriskova, Wettability of Group IV Metal Borides by
Molten Metals, Smachivaemost Poverrkh. Svoistva Rasplavov Tverd. Tel-Naumka Dumka-Kiev, (1972).

81
 [27] O. Dezellus, S. Jacques, F. Hodaj and N. Eustathopulos, Wetting and Infiltration of Carbon by
Liquid Silicon, J. Mater. Sci., 40 [9-10] 2307-11 (2005).
[28] J.-G. Li and H. Hausner, Reactive Wetting in the Liquid-Silicon/Solid-Carbon System, J. Am.
Ceram. Soc., 79 [4] 873 (1996).
[29] T. J. Whalen and A. T. Anderson, Wetting of SiC, Si3N4 and Carbon by Si and Binary Si Alloys,
J. Am. Ceram. Soc., 58 [9-10] 396 (1975).
[30] C. Rado, S. Kalegeropoulou and N. Eustathopulos, Wetting and Adhesion in Metal-Silicon Carbide
Systems: the Effect of Surface Polarity of SiC, Scripta Mater., 42 [2] 203-8 (2000).
[31] P. Nikolopoulos, S. Agatho Pou Los, G.N. Angelopoulos, A. Naoumidis and H. Grűbmeier,
Wettability and Interfacial Energies in SiC-Liquid Metal Systems, J. Mater. Sci., 27 [1] 139-45 (1992).
[32] E. J. Opila and M. C. Halbig, Oxidation of ZrB2-SiC, Ceram. Eng. Sci. Proc., 22 [3] 221-28 (2001).

82
3. Short-term and long-term oxidation behavior of
Si-SiC-ZrB2 composites produced by SRI

3.1 Introduction

Dense silicon carbide materials are commonly employed in air at high temperatures
because their outer SiO2 scale, under passive oxidation conditions, has the lowest
permeability to oxygen among common oxides [1]. By “dense SiC” one usually refers to
ceramics without open/close porosity produced by techniques such as sintering
(pressureless or pressure assisted), chemical vapour deposition (CVD), self-propagating
high-temperature synthesis (SHS) or spark plasma sintering (SPS). The drawback of these
techniques is that they are expensive, in fact the high melting points and the strong
covalent interactions present in UHTCs make it difficult to achieve uniform densification
in these materials. Densification is only achieved at temperatures above 1800°C once
grain boundary diffusion mechanisms become active [2, 3]. Furthermore these techniques
present limitations in product thickness or shape and, in the case of pressureless sintering,
may contain sintering aids (typically carbon [4] and boron [4]) which, under particular
operating conditions, decrease oxide scale viscosity and thus enhance oxygen diffusion
[5]. Silicon carbide ceramics produced by silicon reactive infiltration (SRI) were first
obtained by Hillig et al. [6]. They infiltrated carbonaceous material with molten Si under
10-2 mbar residual pressure at temperatures ranging from 1450°C to 1600°C. This process
allows the shaping and consolidation of preforms made of ceramic powders or fibers that
are bound by a polymer with a high carbon yield after pyrolysis. Reaction-bonded SiC
ceramics have low synthesizing temperatures, short processing times and minimal
dimension changes after the densification process (near-net shaping) in comparison with
pressure-less sintered SiC and hot-pressed sintered SiC. Morover, fine, high purity SiC
powders are not required for the fabrication of reaction-bonded SiC ceramics, which
results in a significant reduction in cost. The major drawback of this technique is that SiC
grains are interpenetrated with a continuous phase made of unreacted silicon which melts
at 1423°C. Liquid silicon may reduce the thermo-mechanical properties of the composite
and their resistance to oxygen. This is why the application of these multiphase materials is
currently foreseen as a coating or as a matrix in fiber reinforced composites in a specific

83
range of temperatures (14001600°C) or in applications where no mechanical loads are
applied. In TPS (thermal protection system), however, high mechanical strength is not
required so that for this application have been proposed materials such as SiC, ZrB2 and
their composites that in the conditions of reentry into Earth's atmosphere may give rise to
oxidizing vitreous phases that become progressively more fluid with increasing
temperature.
SiC has been added to transition metal diborides (ZrB2 and HfB2) for aerospace
applications to increase the relative poor oxidation resistance of pure ZrB2 and HfB2 at
temperatures above 1600°C [7]. At these temperatures several researchers reported the
formation of an oxide scale of borosilicate glass layer (BSZ) containing SiO2, ZrO2 and
B2O3 and the formation of an intermediate region, between the oxides and the unreacted
material, characterized by SiC depletion [8-10]. Recently Willams et al. [11] studied the
oxidation performance of ZrB2-SiC composites with a significantly higher amount of SiC
than the standard 2030 vol.%. They showed that oxidized samples containing > 50 vol.%
of SiC did not present a SiC depleted region because their larger SiC reservoir did not
allow SiO (g) to be transported from the SiC depleted region to the SiO 2 layer.
In the general case, the mechanism of an interfacial reaction of composite oxidation can be
represented as the formation of a reaction interface and the subsequent interaction of a
solid phase with products of melt dissociation and afterwards gaseous by-products.
The rate of the process is governed by the diffusion of reaction products in the growing
scale layer and the three-dimensional decrease in the reactant volume due to the
propagation of the reaction interface [12]. The oxidation processes of refractory
carbides/borides have been shown to be the combined processes of oxygen inward or
metal ion outward diffusion and gaseous (or liquid at relatively lower temperatures)
by-product outward diffusion through oxide scale. Therefore, the oxidation resistances of
carbides/borides are largely influenced by the formation and escape of gaseous
by-products (such as CO, CO2, B2O3) during the oxidation processes which are
significantly different from those of their metal counterparts.
In the previous chapter we presented Si-SiC-ZrB2 composites produced by SRI [13].
Both works showed similar oxide layers and the absence of a SiC depleted region.
The main difference, mainly due to the presence of silicon, stands in the thickness of the
outer borosilicate layer. This chapter presents a detailed study of the mutual interactions
between the Si-SiC-ZrB2 constituent materials during their processing and of their

84
chemical activity with oxygen at high temperatures in the prospect of using this material
as a matrix of a continuous fiber reinforced composite.
The first part of this chapter is devoted to the study of the early stage of oxidation (up to
2 hours at 1600°C), while the second part is focused on the long-term oxidation of some
selected samples at a high temperature (1500°C) for dwelling times of up to 48 hours.
The microstructural features after long-term oxidation were reported and discussed,
together with the weight gains and the average thicknesses of the resulting scale.
The oxidation process was then analyzed on the basis of elemental maps and XRD
patterns taken on the surface of the specimens. In addition, the effect of oxidation time on
the room-temperature flexural strength was also studied.
As silicon is the more reactive component in Si-SiC-ZrB2 composites, we also developed
a technique to quantify its amount from the analysis of the silicon XRD peaks. Si wt.%
was inferred from a curve that was drawn on the basis of data from XRD acquisitions of
Si-SiC powder samples with a known composition.

3.2 Experimental procedure

3.2.1 Materials

The composites were produced by mixing ceramic powders with a plastic binder.
The ceramic powders and the plastic binder employed in this chapter were respectively:
 -silicon carbide (Grade UF 05,. Stark Ag, Goslar, D) with an average particle size
d50 of 1.4 m and a specific area of 4÷6 m2/g;
 zirconium diboride (grade A, Stark Ag, Goslar, D), the particle size d50 is
0.35.0 m;
 micronized (~ 64 m diameter) phenolic novolac powder (Momentive, Columbus,
OH, USA).
Different compositions, (Tab. 3.I), were prepared in order to study the effect of the
constituent materials on the ceramic microstructure before and after their oxidation.
Four vol.% compositions were selected for the ceramic part of the mix (Tab. 3.I).
Each one of these compositions was mixed with the phenolic resin in three different
volume fractions. The first three ceramic compositions have been studied in the short-term

85
oxidative treatment, while for the long-term oxidative study only the batches with the 40
vol.% of phenolic resin were chosen. In the short-term oxidation, the data were acquired
also on Si-SiC composites produced by SRI and on SiC and ZrB2 samples pressureless
sintered for comparison.

Vol.% in the ceramic content Vol.% in the mix

ZrB2 SiC ZrB2 SiC Ph. Res.
70.00 0.00 30.00
100 0 60.00 0.00 40.00
50.00 0.00 50.00
52.50 17.50 30.00
75 25 45.00 15.00 40.00
37.50 12.50 50.00
35.00 35.00 30.00
50 50 30.00 30.00 40.00
25.00 25.00 50.00
- - -
0 100 0.00 60.00 40.00
- - -

Table 3.I: Volume fractions of constituent materials in the ceramic powders and in the mix
green before forming by cold uniaxial pressing.

3.2.2 Sample preparation

Samples were produced improving on the process adopted in the previous chapter,
by performing wet instead of dry milling. The powders were mixed with distilled water
(70 wt.%) for 60 minutes by ball milling in a sealed container, then the phenolic powder
was added and milling continued for a further 60 minutes. The compound was dried in a
vacuum oven at 60°C for 24 hours. Ambient temperature pressing was then performed
with an uniaxial press. The holding time and pressure were respectively 150 seconds and
56 MPa. The amount of the mix to put into the die was calculated in order to produce,
after pressing, 45 mm thick green preforms.
The green samples were pyrolysed at 1000°C under inert atmosphere and the pyrolysis
details can be found in the previous chapter. After pyrolysis, samples were measured and
weighed. They were then positioned on carbon wicks placed on a BN painted graphite
plate. Silicon flakes were put onto the samples. The set up was heated at 1500°C with a

86
10-2 mbar residual pressure in a vacuum furnace which was quickly brought to the
dwelling temperature and held for 1h (for details see chapter 2). The furnace was then
slowly brought to environment temperature by natural cooling.
For the TGA measurements, to avoid the influence of the sample surface area, the
specimens were milled and cut in 3 x 3 x 3 mm3 (± 0.01 mm) cubes with diamond tools.
For the bending tests plates with dimensions of 90x40x4 mm3 were milled and cut with
diamond tools to produce 25x5x2 mm3 bars.

3.2.3 Short-term oxidation treatment

With the aim to compare oxidized Si-SiC-ZrB2 with other studies in literature for similar
systems [14, 15] and for SiC materials [16, 17], specimens were oxidized at 1600°C in a
resistance heated furnace (HT64/17, Nabertherm GmbH, Lilienthal, Germany) in static air
at atmospheric pressure. To avoid contact with the furnace, the samples were placed in an
alumina boat. As this work intends to study their early stage of oxidation, each set of
samples was brought to 1600°C with a heating rate of 10°C·min-1 and held at that
temperature for 5, 30 and 120 minutes.

3.2.4 Long-term oxidation treatment

With the purpose of comparing oxidized Si-SiC-ZrB2 with other works [2, 12, 18-24] and
sintered SiC, specimens were oxidized at 1500°C with a resistance heated furnace
(HT64/17, Nabertherm GmbH, Lilienthal, Germany) in static air at atmospheric pressure
for 4, 8, 24 and 48 hours. The eight-hour time was achieved with two cycles of 4 hours,
while 48 hours were obtained with two cycles of 24 hours. The furnace was slowly heated
and cooled to avoid thermal shock on the samples and preserve the oxide layer. To prevent
contact with the furnace in order to evaluate the weight gain of the samples, silicon
carbide fibers were employed as supports for the specimens. The fibers were suspended
between two alumina plates. This solution allowed the contact surface of the samples to be
minimized and thus the weight variations of the specimens to be more accurately
evaluated.

87
3.2.5 Characterisation methods

Samples were characterized by several techniques.

Thermogravimetry analysis was carried out using TGA/SDTA 851eModule, (Mettler
Toldedo Inc., Im Langacher Greifensee CH) from RT to 1600°C using a heating rate of
10°C min-1. The atmospheric conditions were dry air and different contents of oxygen in
argon (respectively 0.25%, 1%, 10% and 21% of O2 in Ar) with all of them flowing at
50 ml min-1. The sample density was measured with a He pycnometer (ASAP 2010,
Micromeritics, Norcross, GA, USA) and by water pycnometry.
Microstructure evaluations were performed with an optical microscope (Leica DMLM,
Wetzlar, D) and scanning electron microscopy (SEM-FEG Assing SUPRA 25, Zeiss,
Germany) coupled with energy dispersive spectroscopy (EDS Oxford INCA X-Sight
analyzer). Phase quantification was performed, during micrograph acquisition, with the
Gimp 2.6.11 (GNU Image Manipulation Program). XRD analysis was performed with a
X-ray generator system (PW3830, Philips, Eindhoven, Netherlands).
A key factor in this typology of composite material is represented by the amount of
unreacted silicon present in the material. To evaluate this amount, an ad-hoc procedure
was developed. It consisted in preparing five standards covering the range of the expected
phase composition of the samples. The standards were made by mixing powders of silicon
(≥ 99.7% of purity, average particles size < 150 µm; Schuchardt, Műnchen, D) with
powders of silicon carbide (grade UF-10, Stark Ag, Goslar, D); with the following weight
percentages of silicon: 3%, 5%, 10%, 15% and 25%.
A calibration curve was then drawn correlating the integrated areas of the XRD diffraction
pattern peaks of Si (111) for each standard.
Calibration was performed by using the peak areas. In principle this is more accurate than
exploiting their intensities (peak heights) or their width at half maximum (FWHM).
The intensities are in fact conditioned by the percentage of crystallinity of the material,
while the width is affected by the size of the crystallites. Using the areas, the two above
mentioned effects can be minimized. The peak considered was the peak at 28.5°
(corresponding to the plane (111)), the most intense for the crystalline silicon.

88
Before and after long-term oxidation, the specimens were weighed using an analytical
balance (Bc, Orma, Milan, Italy) with an accuracy of 0.1 mg. The final mass change rates
of the specimens were averaged on six samples for 4h, 24h and on three samples for 8h,
48h.
The flexural strength of the as produced and oxidized bars was determined via three-point
bending with a span of 20 mm and a crosshead speed of 0.2 mm/min (Sintech 10/D, MTS,
Eden Prairie, Minnesota USA). For each composition and oxidation time, at least three
specimens were tested.

3.3 Results and discussion

3.3.1 As produced composites

Figure 3.1 shows the morphology of the samples produced in this work. They are fully
dense composites bound by an unreacted silicon matrix (gray phase). Well dispersed ZrB2
powders (large white particles) are clearly visible in the microstructures as well as the
-SiC (larger black particles) and -SiC smaller black particles. This newly produced
phase is the result of the reaction between the pyrolyitic carbon (from the phenolic resin)
with silicon. Their fine grained feature depends on the poor mechanical integrity of
pyrolytic carbon which was further shredded by the highly exothermic reaction with liquid
silicon [25]. As expected, in column 100-0 of figure 3.1, -SiC powder is not visible,
the visible large black areas correspond to voids left by ZrB2 powders popping up during
sample polishing.
An Energy-dispersive X-Ray spectroscopy (EDS or EDX) analysis on these three phases
confirmed that the gray zones are made of Si, the black ones of SiC and the white areas of
ZrB2.

89
Figure 3.1: Micrographs of the samples with different content of ceramic powder (top
labels) and phenolic resin (left labels) after liquid silicon reactive infiltration.

Table 3.II summarizes the open porosity calculations inferred from the different density
measurements. The He density measurements show good agreement with H2O ones.
The open porosity, evaluated by the difference between H2O or He density and geometric
density, shows very low values confirming the effectiveness of the reactive silicon
infiltration, with the exception of the samples 75-25-50 and primarily the 50-50-50,
demonstrating that in those cases full infiltration was not accomplished. This is probably
due to the higher amount of phenolic resin and consequent worse wetting behavior of the
silicon on pyrolytic carbon [26, 27].
The patterns in figure 3.2 show the crystalline phases detected by the instrument in the
samples with 40% of phenolic resin after the SRI (the samples chosen for a more thorough
characterization). All the three samples display the same three crystal phases, namely:
silicon, silicon carbide (both polymorphs: α-SiC and β-SiC) and zirconium diboride.
The shape of the peaks indicates a high degree of crystallinity and the absence of

90
amorphous phases. It is clear from the XRD patterns that during molten infiltration, silicon
did not react with ZrB2 and formed -SiC with the polymer derived carbon.

Density [g/cm3]
Samples
He picnometry H2O picnometry Geometric
SiC sintered 3.2 3.0 2.9
ZrB2 sintered 6.1 5.4 5.1
100-0-50 - 4.0 3.9
100-0-40 4.1 4.1 4.0
100-0-30 - 4.2 4.1
75-25-50 - 3.7 3.3
75-25-40 3.8 3.8 3.7
75-25-30 - 3.9 3.8
50-50-50 - 3.5 2.8
50-50-40 3.5 3.5 3.3
50-50-30 - 3.6 3.5
Table 3.II: Density of the references and samples.

Figure 3.2: XRD Patterns of 100-0-40, 75-25-40 and 50-50-40 sample, respectively.

Table 3.III shows the data used to construct the calibration curve (Fig. 3.3) by fitting the
experimental points with a Lorentzian function. This was then used to retrieve the silicon
wt.% by including the corresponding peak area in the graph.

91
 Peak Area of Si Wt.% of Si
Standard
(2θ ~ 28.5°) (Standards)
1 407.02 25
2 169.49 15
3 98.30 10
4 28.32 5
5 27.13 3
Peak Area of Si Wt.% of Si calculated from peak areas of Si
Sample
(2θ ~ 28.5°) (2θ ~ 28.5°)
50-50-40 164.06 14.6
75-25-40 116.56 11.8
100-0-40 93.53 10.2
Table 3.III: Values of the integrated peak areas (from XRD) and weight percentages of
free silicon. In the case of the standards they were imposed, in the case of the samples they
were obtained intersecting the calibration curve with the measured value of the peak area.

Figure 3.3: Calibration curve on the experimental points of the standards with the peak
area of Si for the samples. These weight percentages were subsequently converted to
volume percentages in table 3.IV using the following densities: 2.3 g/cm3, 3.18 g/cm3 and
6.2 g/cm3 for Si, SiC and ZrB2, respectively.

At the bottom of table 3.III the percentages of free silicon in the samples are given.
It should be noted that the sample with the greatest amount of unreacted silicon is the
50-50-40, approximately 15% by weight. Finally, table 3.IV shows the distribution of the
three phases (vol.%) in the samples from the batch with 40% of phenolic resin.
The percentages of Si and SiC added and ZrB2 have been deduced from the image

92
analysis. This was done by greatly increasing the contrast between the phases, on different
sections and obtaining negative images on which the percentages were computed.
Table 3.IV shows large differences between the calculated and the measured phase
composition. These differences can be ascribed to the different shrinking of the samples
during thermal treatments, i.e. pyrolysis and high temperature treatment (Tab. 3.IV).
During pyrolysis, the binding polymer experienced a mass loss which was accompanied
by a volume reduction (Tab. 3.IV).
Taking for granted the complete inertness of the ceramic powders, one can assume that the
diverse Si contents can also be ascribed to the different carbon yield of the binder caused
by its different distribution within the powders, i.e. coarse (ZrB2) or fine (SiC).

After LSI After pyrolysis

Sample Si SiC ZrB2 Linear Weight loss
[vol.%] [vol.%] [vol.%] shrinking [%] [wt.%]
100-0-40 13.3 53.0 33.7 1.17 3.01
75-25-40 15.4 58.7 25.9 1.94 2.93
50-50-40 18.8 60.0 21.2 2.14 3.39
Table 3.IV: Percentage volume distribution of the phases in the samples with 40% of
phenolic resin after SRI obtained from calibration with X-ray and image analysis.
Linear shrinking and weight loss after pyrolysis.

3.3.2 Oxidation behavior

Most of the current understanding on the oxidation of SiC, ZrB2 and ZrB2-SiC composites
is based on thermo-gravimetric analysis, oxygen consumption studies and ex situ
evaluation of microstructural changes. While thermo-gravimetric methods provide useful
information based on weight changes during oxidation, they record the simultaneous
occurrence of multiple events, including weight gain due to formation of oxide phases,
such as ZrO2, B2O3 and SiO2, weight loss due to vaporization of CO, B2O3 and perhaps
SiO. As a result, the TGA data needs to be deconvoluted and requires careful
interpretation [7]. The TGA in air curves of figure 3.4A shows three different oxidation
behaviors.
The SiC and Si-SiC thermograms in figure 3.4A confirm that both oxidation kinetics of
SiC and Si in these composites are rather low [28], in fact small changes in weight were
monitored.

93
Figure 3.4: A) TGA in air of reference materials; B) TGA in air of the samples with 30
vol.% of phenolic resin (PF); C) TGA in air of the samples with 40 vol.% of PF; D) TGA
in air of the samples with 50 vol.% of PF.

According to the literature [29, 30], the temperature range and the composition of the
oxidizing atmosphere adopted in this work should cause passive oxidation of SiC and Si
according to the reactions:

2 SiC (cr) + 3 O2 (g) → 2 SiO2 (s) + 2 CO (g) (3.1)

Si (cr) + O2 (g) → SiO2 (s) (3.2)

Only the atmosphere containing 0.25 vol.% of oxygen in Ar used for one TGA test
represents a limit conditions since it could cause active oxidation of Si but passive
oxidation of SiC [30].
Both these reactions result in a significant weight gain. On the other hand the relevant
thermograms in figure 3.4A show that the mass gain for SiC is lower than 0.5% and that
of Si/SiC is only a bit higher, which is consistent with the higher mass gain occurring
during silicon oxidation (reaction 3.2). After the formation of a thin silica scale on the
surface of these materials the oxidation progresses very slowly because of the low solid
state diffusion of oxygen through pure silica (10-21 m2/s). The oxygen diffusing through
the passive layer can still react when reaching the interface between the passive layer and
94
the substrate. In principle at this interface the oxygen activity could become low enough to
cause active oxidation with formation of gaseous reaction products like SiO and CO.
Nevertheless, to the authors knowledge, the literature does not report any example of
active oxidation of SiC under the conditions adopted in the present work. The lack of
porosity below the passive layer and the growing trend of TGA curves confirm that
passive oxidation occurred. The considerable gain in weight of the pure ZrB2, which
already begins at around 700°C, is due to the starting reaction of oxidation (reaction 3.3);
it can be explained not only by the great porosity of the sample (around 17%) but also by
the microstructure of the oxide layer.
2 ZrB2 (cr) + 5 O2 (g) → 2 ZrO2 (cr) + 2 B2O3 (l) (3.3)

Below 1100°C pure ZrB2 shows parabolic oxidation kinetics due to the presence of a
dense glassy oxide layer (B2O3) that controls the oxygen diffusion because ZrO2 is a
porous skeleton that does not form a protective scale. In addition the diffusion coefficient
of oxygen through B2O3 is significantly lower than that of ZrO2 (10-12 m2/s for B2O3 and
10-10 m2/s for ZrO2 at 1500°C) [11]. These components of the oxide layer grant some
protection against further oxidation even if they are obviously less effective than silica.
From 1100°C to 1400°C paralinear kinetics can be observed resulting from the weight
gain due to the formation of B2O3 and ZrO2 and the weight loss due to the evaporation of
B2O3 (equilibrium 3.4).

B2O3 (l)  B2O3 (g) (3.4)

Above 1400°C the volatilization of B2O3 is higher than the production of B2O3 in a regime
of rapid linear kinetics with weight gain, in this case it can be concluded that the weight
increase due to the ZrO2 formed is greater than the weight reduction related to ZrB2
consumption and to the evaporation of B2O3.
In figures 3.4 B, C, D the thermograms of the Si-SiC-ZrB2 composites are grouped
according to the different content of phenolic resin. The oxidative behavior of the different
samples turned out to be quite similar with final weight gains that range from at 1.5% to
2.5%. Each of the previous chemical reactions occurred, but the temperature at which the
curves show sharp increase in weight gain is 1400°C. This transition was clearly visible in
the Si-SiC composite, but it was even more evident in all the Si-SiC-ZrB2 samples.
ZrB2 thus dramatically influences sample oxidation behavior at this temperature.
This temperature corresponds to the melting temperature of Si and may trigger reaction
3.2. This behavior is in accordance with Ramberg et al.’s [28] previous findings on Si-SiC

95
materials. In that work [28] it was demonstrated that at temperatures above 1400°C the
parabolic rate constant (kP) for siliconized SiC oxidation increases owing to the transition
from molecular-interstitial to network-dominant oxygen transport through amorphous
SiO2 scale.
In any case when SiC and ZrB2 contemporaneously suffer oxidation the microstructure of
the oxide scale becomes more and more complex as well as the oxidation mechanism
[9, 10, 31, 32]. A glassy borosilica forms on the surface, surrounding crystals of primary
zirconia. Borosilica at rather low temperature grants protection from oxidation since the
oxygen diffusion coefficient through this glass (1,7*10-14 m2/s for a B2O3-21 mol % SiO2)
is lower than those for diffusion in boria and zirconia. Borosilica glass dissolves zirconia
and forms channels that connect the interface between the oxide layer and the not oxidized
substrate to the surface, oxygen permeation through borosilicate glass channels is believed
to occur and also convection currents and liquid flow can be significant during oxidation.
Therefore this BSZ glassy phase can act as a carrier of gaseous species that can diffuse in
it (molecular O2, CO, CO2 and SiO); for this reason it has been proposed that glassy phase
can provide the oxygen for the oxidation of the ZrB2-SiC material bulk [31, 32].
In addition the microstructure of the oxide scale changes with the temperature increase
because of the progressive volatilization of boria. The evaporation of boria increases the
viscosity of the glassy phase and decreases the solubility of zirconia, then secondary
zirconia precipitates on the external part of the scale. Conclusively the oxide scale must be
considered as a multi-phase layer showing composition, physical properties (like
viscosity) and relative amount of phases changing with the temperature. The oxidation rate
is controlled by the oxygen diffusion that can happen according to several mechanisms,
and the dominant mechanism likely changes with the temperature variation. This might
explain the steeper TGA curves around 1400°C. However, it has to be kept in mind that
TGA recorded simultaneous oxidation events.
The only sample that showed an anomalous weight gain was the one labelled 50-50-50,
which started at about 600°C. This behavior was in agreement with its higher porosity due
to an incomplete SRI. It is, however, worth noting that at 1600°C its cumulative weight
change was similar to that of the others.
The increase of both free silicon and SiC content in ZrB2-SiC-Si specimens seems to
improve oxidation resistance, allowing the formation of more silica, as showed by the
comparison of Si wt.% (Tab. 3.III) and TGA curves in figure 3.4C.

96
TGA tests with different mixtures of oxygen in argon showed an oxidizing effect on the
composites which was similar to that of dry air, regardless of the ratio of the two gases.

3.3.3 Oxidation behavior after short-term oxidation at 1600°C

Due to the similar oxidative behavior of the samples during the TGA analysis, only the
sample 50-50-40 was chosen for a more in-depth characterization after the short-term
oxidation. Figure 3.5 shows FESEM photos taken on the edge of the section of the sample
after an oxidation time of 30 minutes at 1600°C. Localized EDS analysis revealed a
structure made of three layers : (a) a borosilicate layer rich in silica, (b) an intermediate
layer containing a BSZ glassy phase and ZrO2 particles, (c) the bulk un-oxidized material
[13]. From a morphological standpoint it can be seen how the silica scale thickness is
100÷150 μm, while the thickness of the layer below is around 40÷80 μm (the boundaries
between this area and the non-oxidized part and the outer layer are marked with a dashed
black line). We could observe, particularly in this sample, that the layer of silica is very
porous and with non uniform thickness, presumably because of CO bubble formation [31].
From XRD patterns taken on the surface, the formation of crystalline silica was observed
already after 30 minutes, although the shape of the peaks suggests that amorphous silica is
also present. Comparing these values with those found in the previous chapter, it can be
observed how the main difference is in the higher thickness of the silica layer, while the
second layer is comparable with the outer layer found in the chapter 2. As observed in the
previous chapter, it was possible to discern between ZrB2 and ZrO2 from a morphological
point of view: ZrB2 particles have smooth contours while ZrO2 particles in the oxidized
material have irregular and broken contours. This is also confirmed by Energy-Dispersive
X-Ray Spectroscopy (EDS or EDX) analysis on these two different zones.
Fahrenholtz et al. [33], following a model analysis of the volatility diagram for the
ZrB2-SiC system, attributed the formation of a SiC depleted region to a p O2 gradient that
develops underneath the surface during oxidation in air and results in low oxygen potential
at the interface between the oxide scale and the substrate. The SiC depletion relies on
active oxidation of SiC (reaction 3.5),

SiC (cr) + O2 (g) → SiO (g) + CO (g) (3.5)

97
Figure 3.5: FESEM images of 50-50-40, oxidized for 30 minutes at 1600°C (collected
with back scattered electrons at lower magnification (500X, left) and zoomed (2000X,
right)), on the section of the oxidized layer.

On the other hand a recent study on the oxidation of ZrB2-SiC compounds with high SiC
contents, specifically over the 50 vol.% [11], reports that the formation of a SiC-depleted
layer above the ZrB2-SiC bulk composite was not observed. This result is well consistent
with the lack of SiC-depletion layer observed in the present work, which can be related
most likely to the low mobility of oxygen in the outer SiO2 layer. However, a combination
of compositional and microstructural factors are likely to control the final microstructure
that develops in the oxide scale, i.e. SiC depletion formation or not.
The TGA thermograms on the 50-50-40 sample oxidized for 5 minutes, 30 minutes and
2 hours at 1600°C are shown in figure 3.6. The graphs show that passivation is very
effective, in fact, after only 5 minutes of oxidation, the weight gain is drastically reduced
compared to the non-oxidized sample. The thermograms do not vary significantly when
the oxidation times are increased. Comparing this oxidative behavior with the reference
materials, it can be observed how it is similar to that of the SiC (which passivates
immediately), while ZrB2 continues to oxidize significantly even after 2 hours of
oxidation. A further confirmation of this passivation characteristic, can be seen by
matching the two SDTA (simultaneous difference thermal analysis) curves related to the
un-oxidized sample and after 2 hours of oxidation, respectively (Fig. 3.7). The melting of
the free silicon can be clearly noticed in the un-oxidized sample in which the endothermic
peak at about 1350°C is very pronounced, as well as the final exothermic peak, which is
due to the oxidation of SiC (reactions 3.1, 3.2 and 3.5) with the release of gas bubbles at
scale/boride interface.

98
After being subjected to 2 hours of oxidation, the peak relating to the silicon melting is
less pronounced (showing that less unreacted silicon is present) and the exothermic
oxidation peak has disappeared.

Figure 3.6: TGA in air of the sample 50-50-40 after different times of oxidation at 1600°C
as produced, after 5 minutes, after 30 minutes and after 2 hours.

Figure 3.7: SDTA of the as-produced sample 50-50-40 during the first TGA in air and
after 2h of oxidation at 1600°C.

99
3.3.4 Oxidation behavior after long-term oxidation at 1500°C

In this section, the nomenclature of the samples studied was changed and has been
referred to them using the volume percentage fraction of zirconium diboride calculated
after SRI (see Tabs. 3.IV and 3.V).

New sample labels ZrB2 [vol.%] Previous sample labels

ZrB34 33.7 100-0-40
ZrB26 25.9 75-25-40
ZrB21 21.2 50-50-40
ZrB0 0.0 0-100-40
Table 3.V: New sample names for the long-term study.

Figure 3.8: Overview of the samples’ surface after oxidation treatments. The top labels
indicate the sample compositions, the left labels refer to the dwelling oxidation time.

Figure 3.8 shows the surface of the samples oxidized in this study for different times.
Even at low magnifications, the different oxidation behavior of the four samples is
evident. On the ZrB21 surface, just after 4 oxidation hours, large ZrO2 islands appear.
The presence of zirconia dendrites adjacent to the ZrO2 islands is an evidence that the
secondary zirconia is precipitating from a saturated borosilicate (BSZ) liquid solution
100
[18]. Indeed boria and, to a lesser extent, silica can evaporate from the BSZ liquid on the
external surface, changing the composition of the remaining liquid. Boron oxide is much
more volatile, with a vapor pressure of 233 Pa at 1500°C [33]. The vapor pressure of silica
is only 3*10-4 Pa at 1500°C [34]. On the surface, essentially all of the boria will evaporate.
Depending on the temperature and atmosphere, some of the silica might also evaporate,
but not zirconia. As boria evaporates, the remaining BSZ liquid becomes richer in silica,
and zirconia must precipitate from the BSZ liquid. Thus the formation of secondary
zirconia precipitates, located near the site of boria evaporation, is expected. The secondary
zirconia will be precipitated from a liquid, and might have a different morphology from
the primary zirconia, suggesting that it might be possible to distinguish secondary zirconia
within the microstructure. The zirconia precipitates will either remain at the location
where they formed, or be carried as a dispersed particle with the flowing BSZ liquid,
increasing, most likely, its viscosity. It should be stressed, however, that it is also possible
that some ZrO2 grains form upon cooling, as the solubility of ZrO2 in the borosilicate glass
will decrease rapidly with temperature.
XRD patterns (Fig. 3.9) show several oxides present on the exposed surface, namely:
zirconia (both monoclinic and tetragonal), silica (cristobalite, tridymite, quartz), zirconium
silicate (ZrSiO4) and three zirconium oxide (Zr3O). These oxides are predicted by the state
diagrams (Figs. 3.10, 3.11, 3.12) and were present already after 4 hours of oxidation.
B2O3 was not detected due to its volatilization that starts above 1100°C. The main
chemical reactions of oxidation that occur up to 1500°C in all Si-SiC-ZrB2 systems (in
addition to those already mentioned, see reactions from 3.1 to 3.4) are:

ZrO2 (cr) + SiO2 (s) → ZrSiO4 (s) (3.6)

3 ZrB2 (cr) + 5 O2 (g) → Zr3O (cr) + 3B2O3 (l) (3.7)

After 8h, for the ZrB21 sample, the amount of silica has increased and the amorphous
phase has disappeared, furthermore the un-oxidized bulk material below the scale is
undetectable by XRD. This indicates that SiC, Si and ZrB2 are thoroughly oxidized during
the exposure. The formation of crystalline SiO2 is not well-understood but related to Si
concentration in amorphous silica. The precipitation of crystalline SiO2 from amorphous
silica when Si concentration reached saturation is to be expected [35]. After ZrB2 and SiC
were oxidized, the ZrO2 and amorphous silica coexisted in the silica-rich glass scale layer
on the surface of samples. This results in a further reaction between ZrO2 and amorphous
silica, as shown in reaction (3.6). An early study [36] in ZrO2 and amorphous silica

101
showed that interstitial silicon diffuses and dissolves into crystalline ZrO2 until the
solution limit is reached when ZrO2 and amorphous silica coexist, thereafter causing
precipitation of ZrSiO4. The formation of this phase, called zircon [21], has been reported
by several authors [12, 21, 23, 24, 35, 37-40].

Figure 3.9: XRD sample patterns at 0h, 4h, 8h, 24h and 48h of oxidation at 1500°C.

With regard to ZrB34 the external oxide surface does not present the zirconia islands, but
a smooth surface with bubbles and craters. The amorphous glassy phase is confirmed by
the XRD, with a considerable increase starting from 24 hours of oxidation. Bubbles are
caused by gasses development of the following species: CO and B2O3. SiO is not expected
to form because it is based on the active oxidation of SiC (reaction 3.5). A passive to
active transition of SiC oxidation occurs between 1873 K and 1973 K (1600°C and
1700°C) in air under atmospheric pressure [41, 42] so that, above these temperatures,
any protection afforded by a glassy oxide layer is largely lost.

102
The presence of different polymorphic phases of silica and zirconia can be explained with
equilibrium transformations that happen during cooling (reactions 3.8 and 3.9).

ZrO2 (tetragonal)  ZrO2 (monoclinic) (3.8)

SiO2 (cristobalite)  SiO2 (tridymite)  SiO2 (quartz) (3.9)

Figure 3.10: Calculated binary phase diagram for the system ZrO2-B2O3, showing the
solubility of zirconia in boron oxide liquid as a function of temperature [43].

Figure 3.11: The ZrO2-SiO2 phase diagram [44].

The fields of stability of the different polymorphs are shown in their respective state
diagrams (Figs. 3.10, 3.11, 3.12). The presence of phases of high temperature at RT is
explained by the fact that these are metastable phases. In fact, the state diagrams refer to

103
equilibrium situations; our case, on the other hand, due to the presence of impurities, is a
metastable situation. Each one of these phases is characterized by its own structure and by
a specific density. Three phases are known for ZrO2: monoclinic < 1170°C, tetragonal
11702370°C, and cubic > 2370°C, generally showing higher symmetry at higher
temperatures. SiO2, more than almost any material, exists in at least 14 crystalline forms
[45], while the most known and stable at room temperature is quartz.

Figure 3.12: State diagram of silica.

The reference ZrB0 material (Fig. 3.9) showed a very different oxidation behavior.
Even after 48 hours of thermal treatment the peaks related to the un-oxidized bulk are well
defined and the only oxide crystalline phase is silica. SiO2, initially present in both
amorphous and crystalline phase, as oxidation continues, tends to become totally glassy.
This behavior is also evident from the photos of the surface oxidized (first column in
Fig. 3.8). Only after 24 hours of oxidation several craters left by the explosion of bubbles
due to evaporation of CO can be seen, but after 48 hours these craters have visually
disappeared and the surface is completely smooth and shiny. X-ray evidence is in good
agreement with the weight gain measurements (Fig. 3.13). Figure 3.14, shows how the
oxide layer for each composite increases with the increase of the dwelling time. The only
exception is represented by the sample ZrB21 oxidized for 48h. This is due to the

104
excessive thickness and weight of the scale that partially collapsed between 24 and 48h
also owing to its porosity. The behaviors of ZrB26 and of ZrB34 are very similar.
The thickness of each layer was measured by averaging several values at various
locations. A large standard deviation in the oxide thicknesses was found due to the
irregularity of the scales. After 24 hours of oxidation the thickness of the sample ZrB21 is
about double compared to the ZrB26 and ZrB34 samples. It is worth noting that these high
values can be achieved because the air in the furnace is stagnant. The different atmosphere
and time conditions are also the reason why the sample ZrB21 showed a contrasting
behavior compared to the TGA reported before (see Fig. 3.4). The overall trend of the
graph in figure 3.14 is in good agreement with the values of specific weight gain shown in
figure 3.13.

Figure 3.13: Normalized weight gain of the samples and reference materials for different
oxidation times.

The oxidation rate of these Si-SiC-ZrB2 materials, evaluated in terms of weight gain, is
lower in comparison with ZrB2-SiC systems [12, 19, 22, 23, 38], but higher than the
values measured on sintered SiC. From figure 3.13 it is clear that even after 48 hours the
process of oxidation is not terminated, in fact all the three Si-SiC-ZrB2 samples continue
to gain weight. A stable passivation oxide layer is achieved only for samples ZrB0 and
sintered SiC. Moreover, the trend of the measured thicknesses of oxide on the sample
ZrB0 appear much less high than the respective gains in weight, because in this sample,

105
as can be seen from the elemental maps (Figs. 3.15, 3.16), the oxide layers are more
compact, homogeneous and almost devoid of porosity. In the reference material, indeed,
the thickness of the scale, even after 48 hours of treatment, does not exceed 30 microns.
After 8 hours, however, a parabolic oxidation behavior is observed for each curve.
This suggests that after the external glassy scale formed at the initial stage of oxidation,
diffusion, more specifically the outward diffusion of constituent elements from the bulk
material to the oxidized region and the inward diffusion of oxygen through the scale layer,
is the rate controlling mechanism for oxidation, as observed for both ZrB2-MoSi2 and
ZrB2-SiC composites [9, 33, 46].

Figure 3.14: Average thickness of the oxide layer for different oxidation times.

Figures 3.15 and 3.16 report the FESEM images taken on the edge section for each
specimen at different dwelling oxidation times. Every micrograph is matched with the
corresponding elemental map of oxygen (green), silicon (red) and zirconium (white).
Boron is difficult to characterize due to the low sensitivity of EDS to light elements.
A first characteristic, common to all samples, is the change in feature of the oxide layer
with increasing oxidation time. At the first stages, 4 and 8 hours, the scales appear
compact and still rather well-fitting to the underneath bulk. After 24 and 48 hours, the
scales are very porous and irregular; still well adherent to the un-oxidized substrate.
The thickness of the BSZ layer and the underlying ZrO2-containing layer are not uniform
over the specimens surface at all temperatures. This may be due also to wetting

106
characteristics or other local variations such as composition, surface topology or surface
cracks that might enhance the local oxidation rate. Moreover, frequently the
ZrO2-containing layer is thicker in the areas where the silica-rich layer is thinner,
indicating less effective oxidation protection in those regions. The boundary between the
oxidized zone and the un-oxidized bulk can also be easily identified visually: the zirconia
particles have jagged and irregular edges, while the particles of ZrB2 are more regular and
smooth.

Figure 3.15: FESEM images and elemental maps of sample scales after 4h at 1500°C in
air.

The high porosity can be explained not only with the volatilization of some phases.
Volume changing can also produce porosity and cracks. Some cracks are observed around
the nodules of zirconia (Fig. 3.8). This cracking behavior is associated either with
different thermal expansion between ZrO2 and SiO2 or with the volumetric change
accompanying transformations (3.8) and (3.9). The scales are mainly made of silica and
particles of zirconia as deduced from the elemental maps. The silica phase is wrapped
around the ZrO2 particles. For the samples ZrB21 and ZrB26 the oxidation mechanism

107
firstly proposed by Karlsdottir and Halloran seems to work [20]. They suggested a
convection mechanism in which the driving force for borate flow is the very large volume
increase upon oxidation of ZrB2. The molar volume of the condensed oxides is different to
the molar volume of the substrate. In the literature for oxidation of metals, the oxide/metal
volume ratio is known as the Pillings–Bedworth Ratio (PBR). When the SiC is oxidized to
silicon dioxide, the molar volume of SiC (12.47 cm3/mole) is much smaller than the molar
volume of SiO2 (27.43 cm3/mole). The carbon oxidizes to CO gas. The ratio of the molar
volume of silica to the molar volume of SiC is the PBR of SiO2, which is 2.2. For ZrB2,
we can define partial PBRs for B and for Zr, and a total PBR, using the molar volumes of
ZrB2 (18.54 cm3/mole), B2O3 (55.2 cm3/ mole) and ZrO2 (20.92 cm3/mole) [47]. The B
partial PBRB is 2.98, forming a liquid that can flow. The zirconium partial PBRZ is 1.13,
with the zirconium oxide being a rigid solid. Due to the large increase in volume, it
follows that the liquid oxide must be squeezed out onto the surface. It is likely that the
volume increase upon oxidation is the driving force for liquid flow. A schematic sketch of
the proposed mechanism is shown in Fig. 3.17 [2]. The main differences between this
model and our samples are represented by the lack of boron fingering on the surface and
by the absence of the porous ZrO2 layer under the external layer. The absence of boron
oxide on the surface is justifiable given the long oxidation times at 1500°C which cause
the complete evaporation. The second aspect, the absence of the columnar structure of
zirconia, is confirmed in other works, in which it is demonstrated that it is formed only
with percentages of ZrB2 above 50% [7, 9, 11, 48] as in the study of Karlsdottir and
Halloran [20, 21]. From the elemental maps of ZrB34, this sample seems not to follow the
convection cells model, indeed no zirconia particles can be detected in the scale.
This behavior is in good agreement with what can be visually observed on the surface
(Fig. 3.8). XRD patterns (Fig. 3.9) show how zirconia and zircon are present on the
external oxide layer of this sample, but the intensity of the related peaks is lower.
The lower intensity of the peaks relating to the zirconia, may accordingly be explained by
the presence of many small crystallites not appreciable at low magnifications.
It is therefore possible to hypothesize that this mechanism also takes place in this type of
sample, but to a lesser extent.
In all micrographs the SiC depletion layer in agreement with what was previously pointed
out by Williams et al. [11] is not observed. In this study it was found that, for isothermal
exposure, the SiC-depleted layer can be reduced by increasing the volume fraction of SiC,
and is undetectable in composites with at least vol.% SiC > 50 vol.%.

108
Figure 3.16: FESEM images and elemental maps of sample scales after 48h at 1500°C in
air.

Figure 3.17: Schematic of convection

cell features on the surface of oxidized
ZrB2-SiC UHTCs. BSZ is a liquid oxide
solution containing boria, silica and
zirconia [21].

The strength of non-oxide ceramics is affected by many factors [49, 50]. In our study, to
better drive high-temperature engineering applications of non-oxide ceramics, an
understanding of the effects of oxidation on the room-temperature flexural strength of the
material is essential. Considerable studies have reported strength retention for SiC and
Si3N4 ceramics after oxidation [51, 52], but only limited experimental work involves
strength retention of ZrB2-based ceramics after oxidation [53, 54]. Oxidation influenced
the strength of Si-SiC-ZrB2 ceramics (Fig. 3.18). As oxidation times progressed,
the retained flexural strength of the samples with ZrB2 had similar trends that gradually

109
decreased. Nevertheless, the strengths after 48 hours of oxidation were between seventy
and ninety percent compared to the un-oxidized samples. The same trend can be observed
in the sample ZrB0, albeit with slightly lower values, but consistent with literature data
[55, 56] due to the higher content of free silicon. Eventually, the flexural strength of the
ZrB2-based ceramics was significantly increased by adding second phase particles. It must
be underlined, however, how the dispersion in this kind of measurements is very high.
By way of sample, the fractured surfaces of sample ZrB26 failed at room temperature,
after 48 hours of oxidation at 1500°C, are shown in Fig. 3.19. The fracture mode, owing to
the lack of cleavage facets, smooth and planar, and to the presence of holes due to the
pullout of particles, is intergranular failure in the inner area of the sample (Fig. 3.19, left);
the fracture surface of the oxide, however, appears smooth and flat (Fig. 3.19, right).
Other works have reported an increase in strength after oxidation for silicon based
ceramics such as SiC and Si3N4 [51, 52] but for lower temperatures and lower times of
oxidation. This increase has been attributed to the formation of a thin, dense oxide layer
that could heal the surface flaws resulting from sample processing and machining. Surface
flaw healing was effective only when the oxide layer was dense, very thin and well
adherent to the bulk. When the oxide layer became thicker and porous, the flaw healing
effect was counterbalanced by the generation of new defects, either within the oxide scale
or at the interface between the oxide scale and bulk materials. Guo [50] noted how the first
behavior is typical of the nano-sized SiC-ZrB2 composites, while the second one is
characteristic of micron-sized SiC-ZrB2 composites, as in our case.

Figure 3.18: Flexural strength of the samples at different oxidation times.

110
Figure 3.19: Fractured surface of sample ZrB26 after 48h of oxidation at 1500°C in air;
left: inner zone; right: oxide scale (outer zone).

3.4 Conclusions

This study demonstrates that it is possible to obtain fully dense Si-SiC-ZrB2 composite
bulks produced via silicon reactive infiltration. A detailed explanation of all the stages of
oxidation with the different reactions and phase transformations that occur at varying
temperatures has been given in this work with the aim of interpreting the final
microstructure in the oxidized samples. The oxidation behavior observed proved to be
very similar to that of the silicon carbide obtained via SRI. In agreement with the previous
chapter and other recent publications [11], a SiC-depleted layer was not observed.
Si-SiC-ZrB2 system after short-term oxidation presented a three layer structure:
a) Silica;
b) BSZ with ZrO2 particles;
c) un-oxidized bulk material.
Another important point to stress is how the TGA measurements showed clearly that this
scale effectively passivates these materials already after a few minutes of oxidation.
The content of phenolic resin seems not to affect the oxidation behavior of the composites.
With regard to long-term oxidation already after 4 oxidation hours all the samples with
ZrB2 showed passivating oxidation layer, notwithstanding, as is evident from the specific
weight gain, even after 48 hours the oxidation process was not finished. Several oxides
were detected on the surface: zirconia (monoclinic and tetragonal), silica (cristobalite,
tridymite, quartz), zircon (ZrSiO4) and three zirconium oxide (Zr3O). Boria was not
detected. The starting combination of powders without -SiC, among those that contain

111
ZrB2, seemed to offer the best long-term oxidation behavior. By the authors’ judgement,
this study lends support to the oxidation model proposed by Karlsdottir and Halloran [20,
21]: the so called “convection cells mechanism” in which the volume increase upon
oxidation is the driving force for BSZ liquid to squeeze out onto the surface. Regarding
mechanical tests, oxidation influenced the strength of Si-SiC-ZrB2 ceramics. All the
samples provide the same tendency of keeping from 70% to 90% of the original flexural
strength.

112
3.5 References

[1] N. S. Jacobson, Corrosion of Silicon-Based Ceramics in Combustion Environments, J. Am. Ceram.

Soc., 76 [1] 13-28 (1993).
[2] E. Eakins, D. D. Jayaseelan and W. E. Lee, Toward Oxidation-Resistant ZrB2-SiC Ultra High
Temperature Ceramics, Metall. and Mater. Trans. A, 42 [4] 878-87 (2011).
[3] J. K. Sonber and A. K. Suri, Synthesis and Consolidation of Zirconium Diboride: Review, Adv. Appl.
Ceram., 110 [6] 321-34 (2011).
[4] L. Stobierski and A. Gubernat, Sintering of Silicon Carbide: I. Effect of Carbon; II. Effect of Boron,
Ceram. Int., 29 [3-4] 287-92 355-61 (2003).
[5] J. A. Costello, R. E. Tressler and I. S. T. Tsong, Boron Redistribution in Sintered α-SiC during
Thermal-Oxidation, J. Am. Ceram. Soc., 64 [6] 332-5 (1981).
[6] W. B. Hillig, R. L. Mehan, C. R. Morelock, V. J. Decarlo and W. Laskow, Silicon-Silicon Carbide
Composites, Am. Ceram. Soc. Bull., 54 [12] 1054-6 (1975).
[7] W. G. Fahrenholtz and G. E. Hilmas, Oxidation of Ultra-High Temperature Transition Metal
Diboride Ceramics, Int. Mater. Rev., 57 [1] 61-72 (2012).
[8] E. V. Clougher, R. L. Pober and L. Kaufman, Synthesis of Oxidation Resistant Metal Diboride
Composites, T Metall. Soc. Aime., 242 [6]1077-82 (1968).
[9] A. Rezaie, W. G. Fahrenholtz and G. E. Hilmas, Evolution of Structure during the Oxidation of
Zirconium Diboride-Silicon Carbide in Air Up to 1500°C, J. Eur. Ceram. Soc., 27 [6] 2495-501 (2007).
[10] E. J. Opila and M. C. Halbig, Oxidation of ZrB2-SiC, Ceram. Eng. Sci. Proc., 22 [3] 221-8 (2001).
[11] P. A. Williams, R. Sakidja, J. H. Perepezko and P. Ritt, Oxidation of ZrB2-SiC Ultra-High
Temperature Composites Over a Wide Range of SiC Content, J. Eur. Ceram. Soc., 32 [14] 3875-83 (2012).
[12] I. B. Ban’kovskaya and V. A. Zhabrev, Kinetic Analysis of the Heat Resistance of ZrB2-SiC
Composites, Glass Phys. and Chem., 31 [4] 482-8 (2005).
[13] A. Ortona, P. Fino, C. D’Angelo, S. Biamino, G. D’Amico, D. Gaia and S. Gianella, Si-SiC-ZrB2
Ceramics by Silicon Reactive Infiltration, Ceram. Int., 38 [4] 3243-50 (2012).
[14] Y. D. Blum, J. Marschall, D. Hui and S. Young, Thick Protective UHTC Coatings for SiC-Based
Structures: Process Establishment, J. Am. Ceram. Soc., 91 [5] 1453-60 (2008).
[15] H. J. Zhou, L. Gao, Z. Wang and S. M. Dong, ZrB2-SiC Oxidation Protective Coating on C/C
Composites Prepared by Vapor Silicon Infiltration Process, J. Am. Ceram. Soc., 93 [4] 915-9 (2010).
[16] S. Biamino, V. Liedtke, C. Badini, G. Euchberger, I. H. Olivares, M. Pavese and P. Fino, Multilayer
SiC for Thermal Protection System of Space Vehicles: Manufacturing and Testing Under Simulated Re-
Entry Conditions, J Eur Ceram Soc., 28 [14] 2791-800 (2008).
[17] C. M. V. Bolivar, A. Antonini, S. Biamino, M. Pavese, P. Fino and C. Badini, Oxidation resistance
of Multilayer SiC for Space Vehicle Thermal Protection Systems, Adv. Eng. Mater., 12 [7] 617-22 (2010).
[18] S. N. Karlsdottir and J. W. Halloran, Oxidation of ZrB2-SiC: Influence of SiC Content on Solid and
Liquid Oxide Phase Formation, J. Am. Ceram. Soc., 92 [2] 481-6 (2009).
[19] M. M. Opeka, I. G. Talmy, E. J. Wuchina, J. A. Zaykoski and S. J. Causey, Mechanical, Thermal
and Oxidation Properties of Refractory Hafnium and Zirconium Compounds, J. Eur. Ceram. Soc., 19 [13-
14] 2405-14 (1999).
[20] S. N. Karlsdottir and J. W. Halloran, Convection Patterns in Liquid Oxide Films on ZrB2-SiC
Composites Oxidized at a High Temperature, J. Am. Ceram. Soc., 90 [9] 2863-7 (2007).
[21] S. N. Karlsdottir and J. W. Halloran, Zirconia Transport by Liquid Convection during Oxidation of
Zirconium Diboride-Silicon Carbide, J. Am. Ceram. Soc., 91 [1] 272-7 (2008).
[22] W.M. Guo and G.-J. Zhang, Oxidation Resistance and Strength Retention of ZrB2-SiC Ceramics,
J. Eur. Ceram. Soc., 30 [11] 2387-95 (2010).
[23] I. B. Ban’kovskaya and D. V. Kolovertnov, Effect of the Thermal Treatment Mode on the
Composition and Structure of ZrB2-SiC System Composites, Glass Phys. and Chem., 39 [5] 579-88 (2013).
[24] F. Tao, L. He-Jun, S. Xiao-Hong, Y. Xi and W. Shao-Long, Oxidation and Ablation Resistance of
ZrB2-SiC-Si/B-Modified SiC Coating for Carbon/Carbon Composites, Corros. Sci., 67 292-7 (2013).
[25] M.W. Chase Jr., NIST-JANAF Thermochemical Tables, 4th edition, J. Phys. Chem. Ref. Data,
1-1951 (Monograph 9) (1998).
[26] J. G. Li and H. Hausner, Reactive Wetting in the Liquid-Silicon/Solid-Carbon System, J. Am.
Ceram. Soc., 79 [4] 873-80 (1996).
[27] O. Dezellus, S. Jacques, F. Hodaj and N. Eustathopoulos, Wetting and Infiltration of Carbon by
Liquid Silicon, J. Mater. Sci., 40 [9-10] 2307-11 (2005).

113
 [28] C. E. Ramberg and W. L. Worrell, Oxygen Transport in Silica at High Temperatures: Implications
of Oxidation Kinetics, J. Am. Ceram. Soc., 84 [11] 2607-16 (2001).
[29] C. Wagner, Passivity during the Oxidation of Silicon at Elevated Temperatures, J. Appl. Phys., 29
[9] 1295-97 (1958).
[30] E. A. Gulbrasen and S. A. Jansson, The High-Temperature Oxidation, Reduction and Volatilization
Reactions of Silicon and Silicon Carbide, Oxid. Met., 4 [3] 181-201 (1972).
[31] S. Gangireddy, S. N. Karlsdottir, S. J. Norton, J. C. Tucker and J. W. Halloran, In Situ Microscopy
Observation of Liquid Flow, Zirconia Growth, and CO Bubble Formation during High Temperature
Oxidation of Zirconium Diboride-Silicon Carbide, J. Eur. Ceram. Soc., 30 [11] 2365-74 (2010).
[32] T. A. Parthasarathy, R. A. Rapp, M. Opeka and M. K. Cinibulk, Modeling Oxidation Kinetics of
SiC-Containing Refractory Diborides, J. Am. Ceram. Soc., 95 [1] 338-49 (2012).
[33] W. G. Fahrenholtz, Thermodynamic Analysis of ZrB2-SiC Oxidation: Formation of a SiC-Depleted
Region, J. Am. Ceram. Soc., 90 [1] 143-8 (2007).
[34] W. G. Fahrenholtz, The ZrB2 Volatility Diagram, J. Am. Ceram. Soc., 88 [12] 3509-12 (2005).
[35] S.-Q. Guo, T. Mizuguchi, M. Ikegami and Y. Kagawa, Oxidation Behavior of ZrB2-MoSi2-SiC
Composites in Air at 1500°C, Ceram. Int., 37 [2] 585-91 (2011).
[36] C. Veytizou, J. F. Quinson, O. Valfort and G. Thomas, Zircon Formation From Amorphous Silica
and Tetragonal Zirconia: Kinetic Study and Modeling, Solid State Ionics, 139 [3-4] 315-23 (2001).
[37] C. R. Wang, J.-M. Yang and W. Hoffman, Thermal Stability of Refractory Carbide/Boride
Composites, Mat. Chem. and Phys., 74 [3] 272-81 (2002).
[38] M. Mallik, K. K. Ray and R. Mitra, Oxidation Behavior of Hot Pressed ZrB2-SiC and HfB2-SiC
Composites, J. Eur. Ceram. Soc., 31 [1-2] 199-215 (2011).
[39] V. O. Lavrenko, A. D. Panasyuk, O. M. Grigorev, O. V. Koroteev and V. A. Kotenko, High
Temperature Oxidation of ZrB2-SiC and ZrB2-SiC-ZrSi2 Ceramics Up to 1700°C in Air, Powder Metall. and
Met. Ceram., 51 [3-4] 217-21 (2012).
[40] Y. Wang, L. Luo, J. Sun and L. An, ZrB2-SiC(Al) Ceramics with High Resistance to Oxidation at
1500°C, Corrosion Science, 74 154-8 (2013).
[41] J. Wang, L. Zhang, Q. Zeng, G. L. Vignoles and A. Guette, Theoretical Investigation for the Active-
to-Passive Transition in the Oxidation of Silicon Carbide, J. Am. Ceram. Soc., 91 [5] 1665-73 (2008).
[42] M. J. Balat, Determination of the Active-to-Passive Transition in the Oxidation of Silicon Carbide in
Standard and Microwave-Excited Air, J. Eur. Ceram. Soc., 16 [1] 55-62 (1996).
[43] W. C. Beard, Research on Phase Equilibria between Boron Oxides and Refractory Oxides,
Including Silicon and Aluminum Oxides, Quarterly Progress Report [9], (1961), Technical Report, Ohio
State University Research Foundation, Columbus, OH, (1962).
[44] A. S. Berezhnoi, Multicomponent Oxide Systems, Naukova Dumka, 216 (1970).
[45] A. F. Holleman and E. Wiberg, Inorganic Chemistry, San Diego: Academic Press, ISBN
0-12-352651-5 (2001).
[46] D. Sciti, M. Brach and A. Bellosi, Oxidation Behavior of a Pressureless Sintered ZrB2-MoSi2
Ceramic Composite, J. Mater. Res., 20 [4] 922-30 (2005).
[47] S. Fujino, C. Hwang and K. Morinaga, Density, Surface Tension, and Viscosity of PbO-B2O3-SiO2
Glass Melts, J. Am. Ceram. Soc., 87 [1] 10-6 (2004).
[48] S. R. Levine, E. J. Opila, M. C. Halbig, J. D. Kiser, M. Singh and J. A. Salem, Evaluation of Ultra-
High Temperature Ceramics for Aeropropulsion Use, J. Am. Ceram. Soc., 22 [14-15] 2757-67 (2002).
[49] S. C. Zhang, G. E. Hilmas and W. G. Fahrenholtz, Mechanical Properties of Sintered ZrB2-SiC
Ceramics, J. Eur. Ceram. Soc., 31 [5] 893-901 (2011).
[50] S.-Q. Guo, Densification of ZrB2-Based Composites and their Mechanical and Physical Properties:
A Review, J. Eur. Ceram. Soc., 29 [6] 905-1011 (2009).
[51] H.-W. Kim, H.-E. Kim, H. Song and J. Ha, Effect of Oxidation on the Room Temperature Flexural
Strength of Reaction-Bonded Silicon Carbides, J. Am. Ceram. Soc., 82 [6] 1601-4 (1999).
[52] H. Park, H.-W. Kim and H.-E. Kim, Oxidation and Strength Retention of Monolithic Si3N4 and
Nanocomposite Si3N4-SiC with Yb2O3 as a Sintering Aid, J. Am. Ceram. Soc., 81 [8] 2130-4 (1998).
[53] D. Sciti, M. Brach and A. Bellosi, Long-Term Oxidation Behavior and Mechanical Strength
Degradation of a Pressurelessly Sintered ZrB2-MoSi2 Ceramic, Scripta Mater., 53 [11] 1297-302 (2005).
[54] S.-Q. Guo, J.-M. Yang, H. Tanaka and Y. Kagawa, Effect of Thermal Exposure on Strength of ZrB2-
Based Composites with Nano-Sized SiC Particles, Compos. Sci. Technol., 68 [14] 3033-40 (2008).
[55] Q. W. Huang and L. H. Zhu, High-Temperature Strength and Toughness Behaviours for Reaction-
Bonded SiC Ceramics Below 1400°C, Mater. Lett., 59 [14-15] 1732-5 (2005).
[56] http://www.schunk-sik.com/en/schunk01.c.9111.de/all_productarticle.

114
4. Cellular ceramics

4.1 Introduction

Materials containing tailored porosity exhibit special properties and features that usually
cannot be achieved by their conventional dense counterparts. Therefore, porous materials
find nowadays many applications as end products and in several technological processes.
Macro-porous materials are used in various forms and compositions in everyday life,
including for instance polymeric foams for packaging, aluminum light-weight structures in
buildings and airplanes, as well as porous ceramics for water purification.
Contrary to metallic and polymeric porous structures, pores have been traditionally
avoided in ceramic components because of their inherently brittle nature. However,
an increasing number of applications that require porous ceramics have appeared in the
last decades, especially for environments where high temperatures, extensive wear and
corrosive media are involved. Such applications include for example the filtration of
molten metals, high-temperature thermal insulation, support for catalytic reactions,
filtration of particulates from diesel engine exhaust gases and filtration of hot corrosive
gases in various industrial processes [1-3]. The advantages of using porous ceramics in
these applications are usually the high melting point, tailored electronic properties, high
corrosion and wear resistance in combination with the features gained by the replacement
of solid material by voids in the component. Such features include low thermal mass,
low thermal conductivity, controlled permeability, high surface area, low density, high
specific strength and low dielectric constant.
These properties can be tailored for each specific application by controlling the
composition and microstructure of the porous ceramic. Changes in open and closed
porosity, pore size distribution and pore morphology can have a major effect on a
material’s properties. All of these microstructural features are in turn highly influenced by
the processing route used for the production of the porous material.
The most straightforward processing route for the preparation of porous ceramics is the
partial sintering of initially porous powder compacts or the sintering of powder mixtures
which undergo solid state reactions that lead to pore formation [4, 5]. This method often

115
results in a relatively low porosity (< 60 vol.%), with pores homogeneously distributed
within the microstructure.
In addition to such a straightforward approach, many novel methods for the preparation of
porous ceramics with controlled microstructure have been developed in response to the
increasing number of new potential applications for cellular ceramics [2]. Versatile
techniques that allow one to deliberately tune the porosity, pore morphology and size
distribution, and that can additionally be applied to ceramic materials of many different
chemical compositions are especially demanded. Novel applications where specific
chemical compositions and tailored microstructures are required include electrodes and
supports for batteries and solid oxide fuel cells, scaffolds for bone replacement and tissue
engineering, heating elements, chemical sensors, solar radiation conversion, among others
[1, 2, 6].
The main processing methods used for the preparation of cellular ceramics such as foams,
honeycomb structures and interconnected rods, fibers and hollow spheres were recently
reviewed by Colombo [7].
The aim of this chapter is to compare some of the processing routes currently available for
the preparation of macro-porous ceramics (i.e. pore size > 50 nm), with particular
emphasis on the processing–microstructure–property relations inherent to each process.
The chapter focus mainly on macro-porous ceramics that exhibit total porosity higher than
50 vol.%. Here is adopted the nomenclature recommended by the IUPAC [8], according to
which macro-porous materials are those exhibiting pore width (d) larger than 50 nm.
Recent articles on the processing of micro- (d < 2 nm) and mesoporous (2 nm < d < 50
nm) ceramic materials using a number of sol-gel, templating, and other chemistry routes
can be found in the literature [9-11].
The processing routes described in this chapter have been classified into replica, sacrificial
template and direct foaming methods, as schematically illustrated in Fig. 4.1.
The processing features of each of these methods are thoroughly discussed and compared
with regards to the versatility and ease of fabrication, as well as their influence on the
microstructure and mechanical strength of the final macro-porous ceramics.
At the end, the chapter presents several high temperature applications in which Si-SiC
foams are currently applied because of their superior thermo-mechanical properties.

116
Figure 4.1: Scheme of possible processing routes used for the production of macro-porous
ceramics [12].

4.2 Replica technique

The replica method is based on the impregnation of a cellular structure with a ceramic
suspension or precursor solution in order to produce a macro-porous ceramic exhibiting
the same morphology as the original porous material (Fig. 4.1(a)). Many synthetic and
natural cellular structures can be used as templates to fabricate macro-porous ceramics
through the replica technique. This paragraph is focused mainly on those processes that
allow the fabrication of bulk ceramic structures.

117
4.2.1 Synthetic templates

The replica technique is in fact considered as the first method deliberately used for the
production of macro-porous ceramics. The original invention dates back to the early
1960s, when Schwartzwalder and Somers [13] started using polymeric sponges as
templates to prepare ceramic cellular structures of various pore sizes, porosities
and chemical compositions. Since then the sponge replica technique has become the most
popular method to produce macro-porous ceramics and is today extensively used in
industry to prepare ceramic filters for molten metal filtration [3] and other applications.
This success is primarily attributed to the simplicity and flexibility of the method.
In the polymer replica approach, a highly porous polymeric sponge (typically
polyurethane) is initially soaked into a ceramic suspension until the internal pores are
filled in with ceramic material. The impregnated sponge is then passed through rollers to
remove the excess suspension and enable the formation of a thin ceramic coating over the
struts of the original cellular structure. At this stage, the slurry has to be sufficiently fluid
to be partially removed under the shearing conditions applied by the rollers, but the
remaining ceramic wet coating should be viscous enough to avoid dripping.
Therefore, ceramic suspensions exhibiting shear-thinning behavior are needed to
efficiently coat the polymeric template. Suspensions exhibiting a viscosity decrease from
10÷30 Pa*s at a shear rate of 5 s-1 down to 1÷6 Pa*s at a shear rate of 100 s-1 have been
shown to be suitable for this process [14-16]. This shear-thinning behavior is achieved
using thixotropic and thickening additives such as clays, colloidal silica, carboxymethyl
cellulose and polyethylene oxide in combination with conventional dispersants [16-20].
The ceramic-coated polymeric template is subsequently dried and pyrolysed through
careful heating between 300°C and 800°C [18]. Heating rates usually lower than 1°C/min
are required in this step to allow for the gradual decomposition and diffusion of the
polymeric material, avoiding the build-up of pressure within the coated struts [18].
Binders and plasticizers are added to the initial suspension in order to provide ceramic
coatings sufficiently strong to prevent cracking the struts during pyrolysis. Typical binders
used are colloidal aluminum orthophosphate, potassium and sodium silicates, magnesium
orthoborate, hydratable alumina, colloidal silica, polyvinyl butyral with polyethylene
glycol as plasticizer and polymerizable monomers [16-18, 21]. After removal of the

118
polymeric template, the ceramic coating is finally densified by sintering in an appropriate
atmosphere at temperatures ranging from 1100°C to 1700°C depending on the material.
Macro-porous ceramics of many different chemical compositions have been prepared
using the sponge replica technique, as illustrated in table 4.I.

Table 4.I: Examples of replica methods reported in the literature [12].

The great flexibility of the method is partly because of the fact that it is applicable to any
ceramic material that can be appropriately dispersed into a suspension.
Porous ceramics obtained with the sponge replica method can reach total open porosity
levels within the range 4095% and are characterized by a reticulated structure of highly
interconnected pores with sizes between 200 µm and 3 mm, as outlined in Figs. 4.2 and
4.3(a). The high pore interconnectivity enhances the permeability of fluids and gases
through the porous structure [22], making these reticulated materials very suitable for high
through-put filtration.
The minimum cell size of replica-derived porous ceramics is however limited to
approximately 200 µm (Fig. 4.2), because of the difficulty with impregnating polymeric
sponges with excessively narrow cells [19]. Predominantly open porous structures are
produced with this method, as the original cellular sponge has to be accessible for the
impregnation of the ceramic suspension or precursor. However, the ratio of open to closed

119
pores in the final ceramic foam may be adjusted to a certain extent by controlling the
suspension viscosity and shear thinning behavior.
A disadvantage of the sponge replica technique is the fact that the struts of the reticulated
structure are often cracked during pyrolysis of the polymeric template (Fig. 4.3(b)),
markedly degrading the final mechanical strength of the porous ceramic [23]. Figure 4.4
shows that the strut flaws reduce the compressive strength of replica-derived porous
ceramics to levels usually lower than the strength theoretically predicted for open cell
structures [24]. Many attempts have been made to avoid this shortcoming by, for instance,
improving the wetting of the suspension on the sponge with the help of additives [18, 25],
performing a second impregnation step to fill the cracks in the ceramic struts [14-16, 19]
and introducing fibers [26, 27] or reactive compounds [25] to enhance the material’s
integrity. In contrast to ceramic suspension-derived reticulated structures, cellular
materials obtained from preceramic polymers have crack-free struts due most likely to the
improved wetting on the sponge and the partial melting of the cross-linked polymer during
pyrolysis [28, 29].

Figure 4.2: Typical porosity and average pore size achieved via the replica, sacrificial
templating and direct foaming processing routes [12].

120
Figure 4.3: Microstructures of macro-porous ceramics produced via the replica technique.
(a) Alumina-based open-cell structure obtained using polyurethane sponge templates [22],
(b) detail of a strut of a cellular ceramic produced from polymeric sponges, illustrating the
typical flaws formed upon pyrolysis of the organic template [30], (c) transversal view of a
highly-oriented SiC porous ceramic obtained after infiltration of a wood template with Si
gas (the longitudinal view is shown in the inset) [31], (d) macro-porous hydroxyapatite
obtained from a coral structure [32].

4.2.2 Natural templates

In addition to synthetic polymer foams, other cellular structures have been used as
templates for the fabrication of macro-porous ceramics through the replica approach,
as outlined in table 4.I. Cellular structures available in nature are particularly interesting as
natural replica templates, due mainly to their special pore morphology and intricate
microstructures, which might be difficult to produce artificially.
Corals have long been applied as a natural template for the preparation of macro-porous
ceramic materials, particularly for the production of scaffolds for tissue engineering and
bone replacement [33, 34]. In the 1970s, White et al. [34] employed a lost-wax method
named ‘‘replamineform’’ to replicate the structure of corals and other marine invertebrate
skeletons. In this technique, the coral is first impregnated with wax under vacuum to
obtain a negative form of the cellular structure. After hardening the wax, the calcium

121
carbonate of the coralline skeleton is leached out using a strong acidic solution.
The macro-porous ceramic is obtained by impregnating the negative wax model with a
ceramic suspension and subsequently removing the organic material by pyrolysis.
This approach has been successfully used to produce macro-porous PZT (lead zirconium
titanate) ceramics with tailored piezoelectric properties [35, 36]. The cellular structure of
corals has also been directly converted into macro-porous hydroxyapatite scaffolds by
submitting the marine skeleton to hydrothermal treatments at high temperatures and
pressures [33] (Fig. 4.3(d)).
The most common approach to transform wood structures into cellular ceramics is to first
prepare a carbon cellular preform by heat treating the wood structure under inert
atmosphere at temperatures within 6001800°C. The carbon preform is subsequently
infiltrated with gases or liquids at high temperatures to obtain the macro-porous ceramic
(Fig. 4.3(c)). Alternatively, the carbon preform is infiltrated at room temperature with
liquid sols and afterwards oxidized to render the cellular ceramic. Table 4.I shows
examples of reactants used for infiltration and the resulting composition of the wood-
derived macro-porous ceramic.
The pore size distribution of wood-derived porous ceramics is determined by the
microstructure of the original wood template used. As wood structures may exhibit cell
sizes down to a few micrometers, porous ceramics obtained from wood templates display
pore sizes markedly lower than those achieved with the polymer replica technique,
typically within the range 10÷300 µm (Fig. 4.2). The porosity of these structures is
predominantly open and can vary from 25% up to 95% depending on the amount of
material impregnated into the template, as illustrated in Fig. 4.2.
As a result of their highly oriented structures, the mechanical properties of wood-derived
ceramics are markedly anisotropic. The mechanical strength in the axial direction (along
the oriented pores) is considerably higher than that in the tangential/perpendicular
direction, as indicated in Fig. 4.4. As in the case of the sponge replica method, the walls of
the wood-derived structures might contain flaws/pores generated from the pyrolysis of
residual carbon in oxidizing environments. Such pores degrade the mechanical strength of
the cellular structure, particularly at high porosity levels (Fig. 4.4) [37, 38]. Structures
with high mechanical strength can only be achieved when the carbon preform is totally
converted into the ceramic phase [39]. Another disadvantage of the wood-replica approach
is the several time-consuming steps involved, which might add considerable costs to the
process.
122
Figure 4.4: Relative compressive strength as a function of the relative density of macro-
porous ceramics produced via replica, sacrificial template and direct foaming methods.
The relative strength was calculated from the ratio between the compressive strength of
the porous structure (σc) and the bending strength of its dense counterpart (σb0). The data
were obtained from porous ceramics of different chemical compositions, assuming a
bending strength of 400, 450, 450, 400, 110 and 320 MPa for the alumina [40], silicon
carbide [41], silicon oxycarbide [42], titania [43], hydroxyapatite [44] and mullite
[45, 46] dense materials, respectively. The y-scales on the right hand side indicate the
corresponding absolute compressive strength for alumina, silicon carbide/oxycarbide and
hydroxyapatite porous structures. The labels a and t indicate the strength data measured
axially and tangentially, respectively, to the aligned pores of wood-derived structures.
The relative strength theoretically expected for open- and close-cell porous ceramics
according to the model from Gibson and Asbhy [24] are indicated by the dashed lines.

4.3 Sacrificial template method

The sacrificial template technique usually consists of the preparation of a biphasic

composite comprising a continuous matrix of ceramic particles or ceramic precursors and
a dispersed sacrificial phase that is initially homogeneously distributed throughout the
matrix and is ultimately extracted to generate pores within the microstructure (Fig. 4.1(b)).
This method leads to porous materials displaying a negative replica of the original

123
sacrificial template, as opposed to the positive morphology obtained from the replica
technique described above.
The biphasic composite is commonly prepared by (a) pressing a powder mixture of the
two components [47], (b) forming a two-phase suspension that is subsequently processed
by wet colloidal routes such as slip, tape or direct casting [48], or (c) impregnating
previously consolidated preforms of the sacrificial material with a pre-ceramic polymer or
ceramic suspension [49].
The way that the sacrificial material is extracted from the consolidated composite depends
primarily on the type of pore former employed. A wide variety of sacrificial materials
have been used as pore formers, including natural and synthetic organics, salts, liquids,
metals and ceramic compounds. Table 4.II illustrates some examples from the recent
literature of possible sacrificial templates.

Table 4.II: Examples of sacrificial template methods reported in the literature [12].

124
Synthetic and natural organics are often extracted through pyrolysis by applying long
thermal treatments at temperatures between 200°C and 600°C [47, 50, 51]. The long
periods required for complete pyrolysis of the organic component and the extensive
amount of gaseous by-products generated during this process are the main disadvantages
of using organic materials as sacrificial phase.
The mismatch in thermal expansion coefficient between the organic and inorganic phases
can also induce cracks within the porous structure during pyrolysis.
Most of these drawbacks can be partially overcome by applying liquid pore formers such
as water and oils or a solid phase that can be easily sublimated (Tab. 4.II). Even though
the extraction process is also time consuming, liquids and volatile oils can be evaporated
or sublimated at milder conditions without generating undesired toxic gases and excessive
stresses during removal of the pore former [52-58].
Sacrificial materials such as salts, ceramic and metallic particles, on the other hand, are
usually extracted by chemical rather than thermal means. The extraction of salts has been
easily accomplished by repeatedly washing the composite with water [49, 59]. Ceramic
and metallic particles or fibers require more aggressive agents and are in most cases
removed by acidic leaching [60-62].
In all of these processes, the continuous matrix phase has to be partially consolidated
before removal of the sacrificial material, so that the porous structure does not collapse
during the extraction step. When the continuous phase is a suspension of colloidal ceramic
particles, consolidation is usually achieved with the help of setting agents and binders [63]
or through the formation of a stiff attractive network of particles throughout the matrix
[48, 58, 64].
In the case of pre-ceramic polymers, consolidation is achieved by cross-linking the
macromolecules through a curing process that takes place at temperatures slightly lower
than that used for the pyrolysis step [47, 49, 65, 66].
The use of relatively volatile oils as a sacrificial phase in aqueous [67-69] or non-aqueous
[57] emulsions is an interesting alternative for the fabrication of porous ceramics by this
template method. The main advantages of this approach are (a) the fact that the template is
easily incorporated into the continuous phase by simple agitation or mixing, (b) the very
small droplet/pore sizes that can be achieved when using immiscible liquids exhibiting
low interfacial energy and (c) the mild conditions required for removal of the template.
The incorporation of a third phase into the emulsions, as for example water in non-
aqueous systems and air in aqueous emulsions, has allowed for the fabrication of porous

125
ceramics with a well-defined hierarchical pore size distribution (Fig. 4.5(a)) [58, 67].
The sacrificial template technique is quite flexible with respect to the possible chemical
compositions that can be used in the fabrication process. A wide number of different
oxides have been used to fabricate porous ceramics using starch particles as sacrificial
template. The main advantages of this approach are its simplicity, the ready availability of
starch particles of various morphologies and sizes (2100 m), and the fact that it can be
applied to any material which can be dispersed in an aqueous suspension [48, 70].
Non-oxide porous ceramics, on the other hand, have been often produced using pre-
ceramic polymers in combination with various different template materials (Tab. 4.II).
Figure 4.2 shows that the range of porosity and pore sizes that can be achieved with this
technique is very broad (2090% and 1700 m, respectively), as they only depend on the
volume fraction and size of the sacrificial template used. Porous materials exhibiting
mono-modal pore size distribution at various porosity levels can for instance be easily
produced using different initial amounts of mono-disperse templating particles
(Fig. 4.5(b)). Pores with special morphologies have also been produced by a freeze casting
approach using water or a high-melting point oil as sacrificial material. By tuning the
growth of ice/oil crystals during the unidirectional freezing of suspensions, dendritic or
highly oriented porous structures has been obtained after sublimation of the templating
crystals (Fig. 4.5(c)) [52-56, 71].
Since in this method the ceramic component corresponds to the negative of the original
template, the removal of the sacrificial phase does not lead to flaws in the struts as in the
case of the positive replica techniques described earlier. Therefore, the mechanical
strength of sacrificial-templated porous structures is usually considerably higher than that
of porous materials produced via positive replica, as clearly indicated in Fig. 4.4.

126
Figure 4.5: Microstructures of macro-porous ceramics produced with the sacrificial
template method. (a) TiO2 foam exhibiting hierarchical porous structure produced via
emulsion templating [67], (b) ordered macro-porous SiO2 obtained using polystyrene
beads as templates [50], (c) highly oriented SiO2 honeycomb structure achieved via the
unidirectional freeze-drying of silica gels [54], (d) macro-porous Al2O3 exhibiting
dendrimer-like pores obtained using camphene as sacrificial template [71].

4.4 Direct foaming methods

In direct foaming methods, porous materials are produced by incorporating air into a
suspension or liquid media, which is subsequently set in order to keep the structure of air
bubbles created (Fig. 4.1(c)). In most cases, the consolidated foams are afterwards sintered
at high temperatures to obtain high-strength porous ceramics.
The total porosity of directly foamed ceramics is proportional to the amount of gas
incorporated into the suspension or liquid medium during the foaming process. The pore
size, on the other hand, is determined by the stability of the wet foam before setting takes
place. Wet foams are thermodynamically unstable systems which undergo continuous
Ostwald ripening and coalescence processes in order to decrease the foam overall free
energy. These destabilization processes significantly increase the size of incorporated
bubbles, resulting in large pores in the final cellular microstructure. Therefore, the most

127
critical issue on direct foaming methods is the approach used to stabilize the air bubbles
incorporated within the initial suspension or liquid media.

4.4.1 Stabilization with surfactants

Several long-chain amphiphilic molecules and biomolecules such as lipids and proteins
can be used as surface-active agents to stabilize wet aqueous foams. These molecules slow
down the coalescence and disproportionation of bubbles by adsorbing at the air bubble
surface and reducing the air-water interfacial energy. However, due to the low adsorption
energy of surfactants at the gas-liquid interface, long-chain surfactants and biomolecules
cannot prevent the long-term destabilization of foams. Wet foams stabilized with long-
chain surfactants collapse within a few minutes after foaming, whereas those stabilized by
proteins exhibit bubble disproportionation within a few hours.
Therefore, direct foaming based on surfactants require a setting agent to consolidate the
foam microstructure before extensive coalescence and disproportionation take place.
The ultimate pore size of the porous ceramic depends on a balance between the kinetics of
bubble disproportionation and the speed of liquid/suspension setting. Small pore sizes
( 50 m) can only be achieved by using efficient surfactants and by rapidly setting the
wet foam.
Here, the focus is mainly on techniques which are flexible in terms of possible chemical
compositions and which can be deliberately tuned to control the final porosity, pore
morphology and size distribution. Table 4.III depicts the main techniques that fulfill these
conditions and have been thus extensively investigated in recent years. The fabrication
processes differ mainly with respect to the method of air incorporation, foam setting and
possible chemical compositions (Tab. 4.III).
Some of these direct foaming methods were adapted from conventional techniques used
for the production of polymer foams.
The in situ incorporation of gas occurs through the nucleation and growth of gas bubbles
upon heating (physical blowing) or chemical reaction (chemical blowing) (Tab. 4.III).
By this method, the ceramic particles are distributed within the polymeric phase, avoiding
the formation of hollow struts that usually degrade the foam final mechanical strength
after pyrolysis (Fig. 4.4) [72].

128
Foam consolidation has not been restricted only to polymeric setting systems, but has been
imparted also by inorganic materials exhibiting sol-gel phase transitions in liquid medium.

Table 4.III: Examples of direct foaming methods reported in the literature [12].

Even though the preparation of porous ceramics by sol-gel can in principle be extended to
many other chemical compositions, the technique has been predominantly applied to
silica-based systems.
In order to produce porous ceramics with a wider chemical compositional range, several
techniques using polymeric setting systems instead of the silica-based sol-gel transitions
have been developed since the 1990s. The amount of organics applied in these so-called
gel-casting methods is, however, significantly lower than that used in previous polymer

129
blowing techniques (Tab. 4.III), reducing the need of an extensive pyrolysis step before
sintering and allowing for the formation of porous ceramics with controlled degree of pore
interconnectivity [23].
The drawbacks of the first gel-casting system were partially [73] or totally [74, 75]
overcome by replacing the original organic additives by less harmful polymerizable
monomers [73] or by applying non-toxic cross-linking reactants which do not require
oxygen-free atmosphere [74, 75]. In the last years, several alternative methods that apply
environmental-friendly setting agents from the food industry have been developed to
circumvent the shortcomings of the original gel-casting technique. Many of these
alternative-setting methods are considerably slower than the original polymerization
reaction. Consequently, the speed of the setting reaction is another important criterion,
besides toxicity and atmosphere sensitivity, when selecting direct foaming methods for
fabrication of porous ceramics of small pore size (< 200 m).
By controlling the foam stability and the setting kinetics, pore sizes within the range
of 35 m to 1.2 mm have been achieved using the above surfactant-based direct foaming
methods (Fig. 4.2). Figure 4.2 also shows that the porosity of cellular structures produced
via surfactant-based direct foaming can be tuned from approximately 40% up to 97%.
The pores obtained with this method are typically spherical and can be either closed or
opened depending on the foam wet processing (Fig. 4.6). Open pores exhibiting
interconnecting windows (Fig. 4.6(b)) are obtained if particles segregate at the plateau
borders of the foam because of bubble disproportionation. Closed pores, on the other
hand, are typically achieved when the particles are distributed uniformly around the gas
bubbles upon setting (Fig. 4.6(a)). These conditions can be controlled in the process by
adjusting the foam stability, air content, particle concentration and setting kinetics.
As opposed to the replica techniques, the direct foaming methods usually lead to dense
flawless struts after sintering (Fig. 4.6(a)), which markedly increase the mechanical
strength of the porous ceramic in comparison with the replica-derived structures (Fig. 4.4).
For porosities higher than 90% (relative density < 10%) the cell walls are markedly thin,
leading to lower mechanical strength than that theoretically estimated for open-cell
structures (Fig. 4.4). However, by decreasing the porosity (increasing density) pores
gradually change from a highly open to a completely closed morphology, increasing the
strength towards the level predicted for closed-cell structures.

130
 Figure 4.6: Microstructures of macro-porous
ceramics produced by direct foaming with
long-chain surfactants as foam stabilizers
[22, 23, 30]. (a) Shows a closed-cell foam
produced from pre-ceramic polymers and
polyurethane precursors [30], highlighting
the dense struts typically obtained with direct
foaming methods. (b) Depicts an open-cell
ceramic foam produced through the in situ
polymerization of monomers shortly after air
incorporation [23].

4.4.2 Stabilization with particles

Solid particles with tailored surface chemistry have lately been shown to efficiently
stabilize gas bubbles upon adsorption at the air-water interface [76-79]. Particles adsorbed
at the gas-liquid interface were observed to impede the destabilization mechanisms
responsible for bubble coalescence and disproportionation for several days, as opposed to
the few minutes typically required for the collapse of foams prepared with long-chain
surfactants [79]. Based on this remarkable long-term stability, this direct foaming method
uses colloidal particles as foam stabilizers in order to produce macro-porous ceramics with
smaller pore sizes than those obtained with surfactant-based foaming techniques [79, 80].
In this method, the attachment of colloidal particles at the air-water interface is promoted
by deliberately changing the wettability of the particle upon adsorption of short-chain
amphiphilic molecules on the surface. The amphiphiles initially added to the suspension
render the particle partially hydrophobic by adsorbing with its polar anchoring group on
the surface and leaving a short hydrophobic tail (up to six carbons) in contact with the
aqueous phase.

131
A minimum amphiphile concentration is required to impart enough hydrophobicity on the
particle surface and induce their attachment at the air-water interface.
After the surface modification of particles, air can be easily incorporated by mechanical
frothing, injection of gas stream, or initiation of a chemical reaction that releases gaseous
by-products directly into the initially fluid suspension.
The outstanding stability achieved with this novel direct foaming method relies on the
irreversible adsorption of colloidal particles at the air-water interface of the gas bubbles, as
opposed to the continuous adsorption-desorption experienced by conventional surfactant
molecules [76-79]. Particles irreversibly adsorbed at the interface sterically impede the
coalescence of neighboring bubbles and also form a coating layer that restrict bubble
shrinkage and expansion in such a way that Ostwald ripening is strongly hindered [79].
Because of its remarkable stability, particle-stabilized foams do not necessarily require a
setting step and can thus be directly dried and sintered to obtain the macro-porous
ceramic. In the absence of a setting system, particle rearrangement within the foam might
still take place during drying, leading to a volumetric shrinkage of approximately 35%.
In this case, unidirectional drying is recommended in order to allow for the homogeneous
shrinkage of the foam. This avoids the formation of pores inside the body because of the
shrinkage constraints imparted by the outer dried layer formed at early drying stages.
Figure 4.2 shows that the outstanding stability of particle stabilized foams enables the
preparation of porous ceramics with more than 80% porosity exhibiting pore sizes four- to
fivefold smaller than those achieved with surfactant-based foaming methods. The porosity
of foams produced with this method vary typically between 40% and 93%, whereas the
average pore sizes can be tuned from approximately 10 to 300 m. As the air bubbles in
the wet foam can be completely covered with a layer of surface modified particles, porous
ceramics with closed pores can be easily prepared with this method (Fig. 4.7(a)).
Figure 4.4 shows that macro-porous ceramics produced from particle-stabilized foams
exhibit remarkable mechanical strength at very high porosity levels (8790%).
This strength level is achieved by tailoring the foaming conditions to produce porous
structures with closed cells and flawless walls (Fig. 4.7). Porous ceramics exhibiting open
porosity (Fig. 4.7(b)) has also been prepared with this technique by simply decreasing the
concentration of stabilizing particles or by adding minor amounts (< 1 wt.%) of a
sacrificial phase into the initial suspension.

132
 Figure 4.7: Microstructures of macro-
porous ceramics produced by direct
foaming with particles as foam stabilizers.
(a) Displays a closed-cell foam produced
from 35 vol.% suspensions containing
alumina particles surface hydrophobized
with valeric acid. The inset in (a) shows
that the cell wall of these structures can be
thinned down to the width of single grains
after drying and sintering the wet foams
[80]. (b) Depicts an open-cell ceramic foam
produced from 20 vol.% suspensions
containing alumina particles surface
modified with propyl gallate. In this case,
0.05 wt.% alginate (based on the powder)
was used as gelling agent.

4.5 High temperature applications of Si-SiC cellular

ceramics

4.5.1 Introduction

Ceramic foams are a class of porous materials characterized by a peculiar distribution of

matter within the space. In nature foams have evolved to maximise their surface area and
minimise weight. Foams are also called cellular materials because their structure can be
represented by a lattice of a repeatable unit called “cell”. Mimicking nature, plastic and
metallic foams have been widely utilised in several technological fields, but in high
temperature applications (> 1000°C) only cellular ceramics can be employed because of
their superior thermo-physical properties [2] like: heat capacity, surface area, porosity,
heat transfer, thermal stability and mechanical stability.
High temperature applications of both open and closed cells ceramic foams include:
filtration (molten metals, particulate); radiant heaters; catalyst supports; kiln components;
insulating components; supports for space mirrors; solid oxide fuel cells ancillary
133
components; lightweight sandwich structures; heat sinks; electrodes and heat exchangers.
Ceramic foams can be virtually of any ceramic material, carbon, carbide and oxide based
foams are mostly employed (Tab. 4.IV).
Alumina is a durable, high temperature, hard and chemically resistant solution at relatively
low cost. However, it is not employed in thermo-mechanical applications because of its
relatively low thermal conductivity and poor thermal shock resistance. Zirconia-based
materials (more expensive than Alumina) have excellent high temperature resistance,
but very low thermal conductivity and thermal shock resistance. Carbon (glassy phase) has
outstanding thermo-physical properties and is widely employed in several applications,
but its use at high temperatures is limited to oxygen free environments because it oxidizes
at temperatures higher than 500600°C. Silicon carbide foams (Fig. 4.8) are a an attractive
solution due to their bulks excellent heat conductivity and very good thermal shock
resistance. However, its corrosion resistance under extreme conditions depends on the
oxidation conditions of the formed outer silica layer [81].

Figure 4.8: CT (computed tomography) reconstruction of reticulated Si-SiC foam (source:

PREC ETH).

Silicon carbide foams’ bulk materials properties strongly depends on the adopted
manufacturing methods [7, 12]. The most used is the replication [13] which is based on
the impregnation of a polyurethane foam with a ceramic slurry or polymeric solution in
order to realize a macro porous cellular ceramic exhibiting the same morphology of the
original template foam. Si-SiC foams are produced infiltrating with molten silicon a

134
pyrolysed foam produced by replication [82]. In liquid silicon infiltration a micro-porous
body, containing SiC powders bonded by glassy carbon, is impregnated by liquid silicon
at high temperatures and, through an in situ chemical reaction (eq. 4.1), carbon is
converted into silicon carbide. Since the amount of silicon is always higher than needed
for the stoichiometric reaction with carbon, these materials are commonly named Si-SiC
ceramics. Si-SiC ceramics were first produced infiltrating carbonaceous material with
molten Si in vacuum (10-2 mbar) at temperatures ranging from 1450°C and 1600°C [83].
The related chemical reaction [84] is:

Siliquid + Csolid = SiCsolid H = - 73 kJ mol-1 (4.1)

Si-SiC is mainly used because of its chemical resistance and its outstanding thermo-
mechanical properties, being able to withstand long oxidative exposing conditions with
low material degradation [85, 86].
Si-SiC foams are high costs materials when compared to other ceramic foams, but their
utilization is necessary for those applications where both thermo-mechanical properties
and long lasting oxidation resistance at high temperatures are required. Open-cell foams
are normally described in terms of linear pore density by measuring the number of Pores
Per Inch (PPI). Ceramic foams, also called macro-porous cellular foams because the width
of their pores is in the range of millimeters or from 10 to 40 PPI, used for high-
temperature applications (Fig. 4.9) have porosities in the range 8090% and a free silicon
content from 5 wt.% to 30 wt.%. Typical Si-SiC foams properties are shown in table 4.V.
Si-SiC foams are nowadays employed in many industrial sectors like: catalysis, porous
combustion, reforming, structural application, etc. Their high-purity and microstructure
makes them the better performers.

135
 Figure 4.9: Shapes of Si-SiC foams (ErbiSiC foams, source: ERBICOL SA).

Yttria-Stabilized Glassy
Si-SiC Alumina
Zirconia Carbon
3
Density [g/cm ] 2.83 3.96 6.02 1.5
Flexural stress [MPa] 203 350 1240 200
Young’s modulus [GPa] 264 340 210 30
-1
CTE, linear (at RT) [K ] 4.30 5.50 10.30 3.50
Specific heat capacity (at RT)
0.94 0.75 0.40 0.50
[J/(g*K)]
Thermal Conductivity (at RT)
85.10 46.00 2.20 4.60
[W/(m*K)]
Maximum Service
1450 1750 1500 500
Temperature, Air [°C]

Table 4.IV: Si-SiC (trade name ErbiSiC) material properties compared to Alumina, Yttria-
Stabilized Zirconia and Glassy Carbon (Si-SiC data source: ERBICOL SA; other
materials: matweb).

3
Foam Density [g/cm ] 0.323
Normalized Density 0.114
Macro-porosity 0.87
2 3
Surface Area [m /m ] 552
Average Strut Thickness [mm] 0.9
Flexural Strength [MPa] 4
Compression Strength [MPa] 2.5
Thermal Conductivity [W/(m*K)] 8÷10

Table 4.V: Si-SiC foam properties (ErbiSiC foam, 10 PPI, source: ERBICOL SA).

136
4.5.2 Porous burners

Combustion processes are fundamental for the production of energy. To allow combustion
to occur, different burner types are available depending on the type of fuel and the
operating conditions. There are several porous burner technologies which have been
developed over the years [87-89]. A relatively new concept, developed in the 1990’s at the
LSTM-Erlangen, is based on the stabilization of the combustion process within a porous
medium where a flameless combustion takes places in its empty spaces. Combustion
stabilization is achieved at the transition zone between two porous regions with different
pore sizes (dotted region in Fig. 4.10). The combustion propagation into this porous body
is determined by its critical pores dimension: if it is smaller or bigger than critical pore
size, combustion front stops or propagates. Since part of the heat generated by the
combustion zone is transferred backward to the flame trap (blue region in Fig. 4.10) by
conduction and radiation, the temperature at the interface between the two porous media is
higher than that of the air-fuel mixture. Local overheating is avoided by incoming gasses
heat removal [88, 89]. Compared to conventional combustion processes with free
premixed flames, stabilized combustion in porous media presents several advantages [88].
Environmentally wise it allows an impressive decrease in NOx and CO emissions [88, 90,
91]. Thermal efficiency was found to be inversely proportional to the equivalence ratio of
the fuel-air mixture and the CO and NOx emissions were respectively from 9 to 16 ppm
and 0.2 ppm. It is worth to remember that the emissions in a conventional burner are from
50 to 225 ppm for CO and from 2 to 7 ppm for NOx. In addition, this more efficient
combustion reduces fuel consumption. These porous burners are applied in: environment
and water heating [92], metal heat-treating [90], coating and paint drying [2, 89], glass and
chemical processing [93], paper and wood drying [2, 89], food processing [88] and pilot
flame in stationary gas turbines [90]. Because of the pollution concerns, combustion in
porous media has recently been the subject of a significant amount of research and
development [88, 94]. The flame stabilization concept imposes severe thermo-mechanical
loads at the interface of the two porous materials employed because of the temperature
levels and gradients. A great issue is then to select the proper material for such
components. Moreover, foams are exposed to high temperature oxidation while the burner
is operating [81] and to a severe thermal shock during power ON/OFF cycles which turns
out to be detrimental during the very first cycles [86]. Si-SiC foams demonstrated to be a

137
key element for flame stabilization technology, because they can guarantee the system
functionality in oxidative conditions at high temperatures (1400°C) for several thousand
hours [85, 86].

Figure 4.10: Porous burner [95].

With the use of Si-SiC foams, it’s possible to reach thermal loads up to 5 MW/m2, even
though industrial capacity is mostly limited to 1 MW/m2. Concerning the Si-SiC capability
to withstand oxidation during operating conditions, if temperatures and partial oxygen
pressure fall into a passive oxidation region the following reaction between SiC and
oxygen takes place:

SiC (s) + 2O2 (g)  CO2 (g) + SiO2 (s) (4.2)

Part of SiC converts to SiO2 producing an adherent scale which protects the unaffected
Si-SiC underneath. Scale formations results in foams’ weight increase. Hock is the other
phenomenon that happens in a porous burner during burner operation and especially the
very first ON/OFF cycles. Foams are immediately brought to burner working temperature
by gasses combustion inside their porosity and even quickly cooled during system shut-
down, while fuel is interrupted and cold plain air flows through them. In these transients,
foams’ structure suffers localized damages because of cracks formation, most likely
provoked by microscopic accumulations of metallic silicon which has a different
coefficient of thermal expansion than SiC. Due to micro cracks, foams’ mechanical

138
properties lower after the first ON/OFF cycles, to remain almost unchanged up to 1000
operating hours and many ON/OFF cycles [7].

4.5.3 Catalyst carrier

Very often a catalytic reaction takes place inside a porous body, where reactants and
catalyst come in contact. The porous body has the function of a support, being the catalyst
deposited on it [96]. To be efficient in terms of volume, the catalytic support needs to have
a high surface area. Available porous structures for catalysis are: packed beds,
honeycombs and foams [96]. Foams offer several advantages compared to the other two
structures: higher radial mass flow, higher radial heat exchange, higher tortuosity of the
flow resulting in more interesting fluid dynamics, lower pressure drop and a higher
geometrical porosity [2]. The use of ceramic foams is of advantage in cases of heat-
transfer limitation, pore-diffusion limitation or processes with selectivity-control problems
[2]. The use of Si-SiC foams, compared to other materials, is justified for reactions that
take place in acid or basic conditions or reactions for which high thermal conductivity of
the catalyst carrier is necessary. Furthermore, since catalyst carriers are often inserted into
a reactor tube which is cooled externally, the low coefficient of thermal expansion of
Si-SiC is suitable in order to avoid stress concentrations. A recent study examined ceramic
foams-to-reactor fit during catalytic reactions [97]. The presence of annular gap flows and
the consequent inability to transfer heat to the reactor wall during exothermic reactions is
minimized for Si-SiC foams, because of their low coefficient of thermal expansion and
thermal conductivity. Another study compared Si-SiC foams, mullite foams and steatite
spheres as catalyst carriers for the oxidation of o-xylene to phtalic anhydride [98].
The catalyst carriers were coated with a vanadia/titania catalyst and fitted in a tube reactor.
During the exothermic reaction, which takes place in the temperature range 350450°C,
Si-SiC foams proved best because of the higher thermal conductivity, giving a smoother
temperature profile and increasing conversion. The study also demonstrated that higher
surface areas, i.e. foams with higher PPI numbers, have a beneficial effect on heat transfer
properties. Still another work demonstrated the superiority of SiC foams compared to
alumina as catalyst carrier for the Fischer-Tropsch synthesis [99]. The ability of the foam

139
to evacuate the heat generated during the cobalt catalyzed reaction leads to a superior
efficiency at higher temperatures [99].

4.5.4 Structural application

Because of their low density and the ability to withstand high thermal loads, Si-SiC foams
are good candidates for thermo-structural applications. This is the case of reusable thermo-
protective shields for reentry vehicles, where sandwiches made of a Si-SiC foam core and
Ceramic Matrix Composites (CMCs) skins can been employed (Fig. 4.11). Early studies
on sandwich structures with a SiC foam core were performed in 1985 [100]. CMC skins
were joined to a Si-SiC foam through Chemical Vapour Infiltration (CVI) [101].
In alternative to the CVI process, another method of fabricating sandwich structures has
been recently developed [102]. In this process, CMC skins and foam core are assembled,
before CMC densification, using a polymer derived ceramic bonding layer. The sandwich
was then densified by Polymer Impregnation and Pyrolysis (PIP). Using this technique,
foam struts facing the CMC resulted completely enveloped into the bonding layer,
allowing a strong mechanical joint between skins and core. Bonding is strictly related to
skins and core coefficient of thermal expansion matching. In bonded sandwich ultimate
strengths is higher than that of the equivalent plain foam. After reaching a maximum stress
all samples did not experience a catastrophic failure. In this sandwich configuration failure
was observed into the ceramic core mainly because of shear.
SiC foam CMC sandwiches are also good thermal insulators, protecting the vehicle
interiors by the very high temperatures produced by the friction with the atmosphere, but
to resist thermal cycling skins and core must match their coefficients of thermal expansion
[103].

140
Figure 4.11: Sandwich structure made of a Si-SiC core foam and two CMC skins (source:
SUPSI).

4.5.5 Solar radiation absorbers

Concentrated Solar Power (CSP), and in particular solar towers, are a promising way to
generate electricity from solar radiation [104]. Solar tower systems use many large
individually tracking reflectors (heliostats) to concentrate the sun’s radiation onto a central
receiver located at the top of a tower 80÷120 meters high [105]. Air is usually the transfer
medium which absorbs the solar energy and generates steam passing through a heat
exchange unit. Once generated, steam is expanded in a turbine to convert heat into work
(Rankine cycle). This huge amount of energy (1 GW for the present CSP plants) is
collected by a solar radiation absorber/receiver. A consolidated system is the open
volumetric receiver [106]. “Volumetric” means that the receiver temperature at the
irradiated side is lower than the temperature of the medium leaving it [105].
A good volumetric receiver should have a high open porosity to allow the radiation to
penetrate deeply into the structure and let the heat exchange fluid flow through it with
minimal pressure losses. The open volumetric receiver is an important element since it
drives the performance of the overall plant; therefore it greatly influences the final cost per
kWh produced. Some volumetric receivers, under certain material and boundary
conditions, lead to flow instabilities which can be avoided by using suitable receivers like
ceramic foams [106]. Indeed a Si-SiC foam presents a high heat conductivity, a quadratic
behavior of the pressure drop correlation and a high ratio of viscous permeability to
inertial permeability [106]. In alternative to a receiver made of a continuous porous body,
an interesting solution is made by discrete porous elements placed in order to allow solar
energy absorption and better heat exchange and fluid flow characteristics [107]. To absorb
solar radiation and to transfer heat to the medium, the porous body should have

141
appropriate absorptivity and thermal conductivity. At an industrial level a temperature
resistance up to 1200°C, a thermal shock resistance (> 200K/min air outlet temperature),
a thermal cycle resistance, a high thermal conductivity, a low pressure drop and a low
reflectivity are also needed. Further requirements are lower cell size to achieve large
surface areas necessary to transfer heat from the material to the gas. Several works
focused on the research of new materials to improve the efficiency of the solar radiation
absorption, and foams proved to be good candidates [105, 108-110]. Comparisons
between some materials and structures for their use in solar absorption were carried out in
a 20KWsolar installation capable to generate concentrated radiation of up to 5 MW/m2
peak flux [109]. Tested materials included a fibre mesh, a Si-SiC catalyst carrier with 2
mm parallel channels (honeycomb), a 20 PPI Si-SiC foam and a multiple porosity Si-SiC
foam with 80 PPI porosity at the front, being responsible for absorption and heat transfer,
coupled with a 20 PPI Si-SiC foam on the back. The fibre mesh and the 80 PPI/20 PPI Si-
SiC foam resulted best performers. Although foams and mesh proved best, Si-SiC
honeycombs were finally selected because ceramic honeycombs are a mass product being
employed in the automotive sector for particulate filtering. As previously mentioned, an
advantage of Si-SiC foams over Si-SiC honeycombs is that foams allow 3-dimensional
fluid dynamic, compared to the one dimensional flow within honeycomb channels
avoiding thus flow instabilities. Finally, foams porosity can be tuned in order to optimize
volumetric absorption [106, 110] and thus final system efficiency. Another application in
CSP is for the hydrogen production [111]. Among the different techniques the Hybrid
Sulphur Cycle (HyS) is considered as one of the most promising cycles. The key
components, necessary for the high temperature part of the process, are a ceramic foam for
the vaporisation of liquid sulphuric acid and a ceramic honeycomb for the catalytic
dissociation of sulphur trioxide [112]. In both structures Si-SiC based materials, withstood
CSP conditions and kept their structural integrity over a substantial period of time
(1000 hrs).

4.5.6 Electrical applications

Si-SiC foams can be heated by an applied electrical power (Fig. 4.12) thanks to the
semiconductivity of silicon carbide. Si-SiC foams produced using the replica method and

142
Liquid Silicon Infiltration (LSI) show enhanced electrical conductivity [2]. This is due to
the free silicon content inside the foam (2040 wt.%) which is a better electrical conductor
than silicon carbide [112-114]. In addition, the effect can be enhanced by filling the
hollow ligaments of the foams with silicon [2]. A study analyzed the influence of cell size,
expressed in terms of Pores Per Inch (PPI) and content of free silicon on electrical
resistance [2]. The foams were heated up to 900°C within 4 minutes by applying a
15 V/50 A. The experiment provides evidence that the content of free silicon inside the
foam is negatively correlated to the electrical resistance. The pore size only has an indirect
effect, because it affects infiltration behavior, with smaller diameter ligaments being more
effectively infiltrated by silicon [2]. SiC heating element are already employed for
electrical heating, Si-SiC foams offer several advantages because of their low specific
weight, which permits rapid heating and cooling and because of their large surface area
which leads to a high heat transfer to the flowing medium. Beside these advantages Si-SiC
foams can bear high thermal loads and thermal shocks. When lower voltages are applied,
Si-SiC foams can be also employed as sensors. Electrical resistance was used to monitor
sandwich foams integrity during bending, demonstrating a correlation between the
formation of a crack inside the foam with an increase of its electrical resistance [115].

Figure 4.12: Si-SiC foam used as electrical heater at about 1000°C (source: ERBICOL SA).

4.5.7 Reformers

The growing demand for green energy in the last years fostered the development of Solid
Oxide Fuel Cells (SOFC), where H2 and CO can be used as fuel. These synthesis gas

143
(syngas) can be produced from diesel, natural gas, heating oil, gasoline or LPG (liquid
petroleum gas) by steam reforming, catalytic or thermal partial oxidation and auto-thermal
oxidation. The partial oxidation offers several advantages compared to other processes, as
it is catalyst free and doesn’t need any external heat source. The reaction can be performed
inside an inert porous body.
A recent work performed within the European Project FlameSOFC [116] compared the
performance of a Si-SiC foam and an Al2O3 fiber static mixer structure as reformers [117].
In this experiment, air was preheated (up to 700°C) and mixed with methane and then
passed through the reformer (Fig. 4.13). This is a special reactor with a conical and a
cylindrical section. In the diverging conical section the oxidation zone is stabilized.
Downstream in the cylindrical section reforming reactions take place. Temperatures at
several points in the reformer were measured and outlet gas samples composition was
analysed. The study concluded that for the partial oxidation of methane, the Si-SiC foam
demonstrated higher heat recuperation and a better performance concerning soot-point.

Figure 4.13: Schematic representation of the TPOX (thermal partial oxidation) reformer
(source: TU BAF).

144
4.6 References

[1] R. W. Rice, Porosity of Ceramics, pp. 539. Marcel Dekker Inc., New York, (1998).
[2] M. Scheffler and P. Colombo, Cellular Ceramics: Structure, Manufacturing, Properties and
Applications, WILEY-VCH Verlag GmbH, Weinheim, Germany, (2005).
[3] L. J. Gauckler, M. M. Waeber, C. Conti and M. Jacobduliere, Ceramic Foam for Molten-Metal
Filtration, J. Metals, 37 [9] 47-50 (1985).
[4] S. Y. Shan, J. F. Yang, J. Q. Gao, W. H. Zhang, Z. H. Jin, R. Janssen and T. Ohji, Porous Silicon
Nitride Ceramics Prepared by Reduction-Nitridation of Silica, J. Am. Ceram. Soc., 88 [9] 2594-6 (2005).
[5] I. H. Arita, V. M. Castano and D. S. Wilkinson, Synthesis and Processing of Hydroxyapatite Ceramic
Tapes with Controlled Porosity, J. Mater. Sci. - Mater. Med., 6 [1] 19-23 (1995).
[6] J. Will and L. J. Gauckler, Ceramic Foams as Current Collector in Solid Oxide Fuel Cells (SOFC):
Electrical Conductivity and Mechanical Behaviour, in Proceedings of the Fifth International Symposium on
Solid Oxide Fuel Cells (SOFCV), Aachen, Germany. Edited by U. Stimming, S. C. Singhal, H. Tagawa, and
W. Lehnert. The Electrochemical Society Inc., Pennington, NJ, (1997).
[7] P. Colombo, Conventional and Novel Processing Methods for Cellular Ceramics, Philos. Trans. Roy.
Soc. A, 364 [1838], 109-24 (2006).
[8] K. S. W. Sing, D. H. Everett, R. A. W. Haul, L. Moscou, R. A. Pierotti, J. Rouquerol and
T. Siemieniewska, Reporting Physisorption Data for Gas Solid Systems with Special Reference to the
Determination of Surface-Area and Porosity (Recommendations 1984), Pure Appl. Chem., 57 [4] 603-19
(1985).
[9] J. Fricke and A. Emmerling, Aerogels, J. Am. Ceram. Soc., 75 [8] 2027-36 (1992).
[10] S. Mann, S. L. Burkett, S. A. Davis, C. E. Fowler, N. H. Mendelson, S. D. Sims, D. Walsh and
N. T. Whilton, Sol-Gel Synthesis of Organized Matter, Chem. Mater., 9 [11] 2300-10 (1997).
[11] U. Ciesla and F. Schuth, Ordered Mesoporous Materials, Microporous Mesoporous Mater., 27 [2-3]
131-49 (1999).
[12] A. R. Studart, U. T. Gonzenbach, E. Tervoort and L. J. Gauckler, Processing Routes to
Macroporous Ceramics: A Review, J. Am. Ceram. Soc., 89 [6] 1771-89 (2006).
[13] K. Schwartzwalder and A. V. Somers, Method of Making Porous Ceramic Articles, US Pat. No.
3090094, May 21, (1963).
[14] X. P. Pu, X. J. Liu, F. G. Qiu and L. P. Huang, Novel Method to Optimize the Structure of
Reticulated Porous Ceramics, J. Am. Ceram. Soc., 87 [7] 1392-4 (2004).
[15] D. D. Brown and D. J. Green, Investigation of Strut Crack Formation in Open Cell Alumina
Ceramics, J. Am. Ceram. Soc., 77 [6] 1467-72 (1994).
[16] X. W. Zhu, D. L. Jiang, S. H. Tan and Z. Q. Zhang, Improvement in the Strut Thickness of
Reticulated Porous Ceramics, J. Am. Ceram. Soc., 84 [7] 1654-6 (2001).
[17] L. Montanaro, Y. Jorand, G. Fantozzi and A. Negro, Ceramic Foams by Powder Processing, J. Eur.
Ceram. Soc., 18 [9] 1339-50 (1998).
[18] J. Saggio-Woyansky, C. E. Scott and W. P. Minnear, Processing of Porous Ceramics, Am. Ceram.
Soc. Bull., 71 [11] 1674-82 (1992).
[19] F. F. Lange and K. T. Miller, Open-Cell, Low-Density Ceramics Fabricated from Reticulated
Polymer Substrates, Adv. Ceram. Mater., 2 [4] 827-31 (1987).
[20] J. M. Tulliani, L. Montanaro, T. J. Bell and M. V. Swain, Semiclosed-Cell Mullite foams:
Preparation and Macro- and Micromechanical Characterization, J. Am. Ceram. Soc., 82 [4] 961-8 (1999).
[21] H. R. Ramay and M. Q. Zhang, Preparation of Porous Hydroxyapatite Scaffolds by Combination of
the Gel-Casting and Polymer Sponge Methods, Biomaterials, 24 [19] 3293-302 (2003).
[22] M. D. M. Innocentini, P. Sepulveda, V. R. Salvini, V. C. Pandolfelli and J. R. Coury, Permeability
and Structure of Cellular Ceramics: A Comparison between Two Preparation Techniques, J. Am. Ceram.
Soc., 81 [12] 3349-52 (1998).
[23] P. Sepulveda, Gelcasting Foams for Porous Ceramics, Am. Ceram. Soc. Bull., 76 [10] 61-5 (1997).
[24] L. J. Gibson and M. F. Ashby, Cellular Solids: Structure and Properties, p. 510. 2nd edition,
Cambridge University Press, Cambridge, (1997).
[25] J. Luyten, I. Thijs, W. Vandermeulen, S. Mullens, B. Wallaeys and R. Mortelmans, Strong Ceramic
Foams from Polyurethane Templates, Adv. Appl. Ceram., 104 [1] 4-8 (2005).
[26] M. X. Wu and G. L. Messing, SiC-Whisker-Reinforced Cellular SiO2 Composites, J. Am. Ceram.
Soc., 73 [11] 3497-9 (1990).
[27] A. E. M. Paiva, P. Sepulveda and V. C. Pandolfelli, Processing and Thermomechanical Evaluation
of Fibre-Reinforced Alumina Filters, J. Mater. Sci., 34 [11] 2641-9 (1999).

145
 [28] M. R. Nangrejo, X. Bao and M. J. Edirisinghe, Processing of Ceramic Foams from Polymeric
Precursor-Alumina Suspensions, Cellular Polymers, 20 [1] 17-35 (2001).
[29] M. R. Nangrejo, X. Bao and M. J. Edirisinghe, The Structure of Ceramic Foams Produced Using
Polymeric Precursors, J. Mater. Sci. Lett., 19 [9] 787-9 (2000).
[30] P. Colombo and J. R. Hellmann, Ceramic Foams from Preceramic Polymers, Mater. Res.
Innovations, 6 [5-6] 260-72 (2002).
[31] E. Vogli, H. Sieber and P. Greil, Biomorphic SiC-Ceramic Prepared by Si-Vapor Phase Infiltration
of Wood, J. Eur. Ceram. Soc., 22 [14-15] 2663-8 (2002).
[32] B. Ben-Nissan, Natural Bioceramics: From Coral to Bone and Beyond, Curr. Opinion Solid State
Mater. Sci., 7 [4-5] 283-8 (2003).
[33] D. M. Roy and S. K. Linnehan, Hydroxyapatite Formed from Coral Skeletal Carbonate by
Hydrothermal Exchange, Nature, 247 [5438] 220-2 (1974).
[34] R. A. White, E. W. White and J. N. Weber, Replamineform - New Process for Preparing Porous
Ceramic, Metal, and Polymer Prosthetic Materials, Science, 176 [4037] 922 (1972).
[35] R. E. Newnham and G. R. Ruschau, Smart Electroceramics, J. Am. Ceram. Soc., 74 [3] 463-80
(1991).
[36] D. P. Skinner, R. E. Newnham and L. E. Cross, Flexible Composite Transducers, Mater. Res. Bull.,
13 [6] 599-607 (1978).
[37] A. Herzog, R. Klingner, U. Vogt and T. Graule, Wood-Derived Porous SiC Ceramics by Sol
Infiltration and Carbothermal Reduction, J. Am. Ceram. Soc., 87 [5] 784-93 (2004).
[38] C. Zollfrank, N. Travitzky, H. Sieber, T. Selchert and P. Greil, Biomorphous Si-SiC/Al-Si Ceramic
Composites Manufactured by Squeeze Casting: Microstructure and Mechanical Properties, Adv. Eng.
Mater., 7 [8] 743-6 (2005).
[39] C. R. Rambo and H. Sieber, Novel Synthetic Route to Biomorphic Al2O3 Ceramics, Adv. Mater.,
17 [8] 1088 (2005).
[40] I. Fukuura and T. Asano, Fabrication and Properties of Some Oxide Ceramics: Alumina, Mullite
and Zirconia, pp. 165-74 in Fine Ceramics. Edited by S. Saito. Elsevier, New York, (1985).
[41] Y. W. Kim, M. Mitomo, H. Emoto and J. G. Lee, Effect of Initial Alpha-Phase Content on
Microstructure and Mechanical Properties of Sintered Silicon Carbide, J. Am. Ceram. Soc., 81 [12] 3136-
40 (1998).
[42] P. Greil, Active-Filler-Controlled Pyrolysis of Preceramic Polymers, J. Am. Ceram. Soc., 78 [4]
835-48 (1995).
[43] J. Li, S. Forberg and L. Hermansson, Evaluation of the Mechanical-Properties of Hot Isostatically
Pressed Titania and Titania Calcium-Phosphate Composites, Biomaterials, 12 [4] 438-40 (1991).
[44] M. Akao, H. Aoki and K. Kato, Mechanical-Properties of Sintered Hydroxyapatite for Prosthetic
Applications, J. Mater. Sci., 16 [3] 809-12 (1981).
[45] M. Mizuno, Microstructure, Microchemistry, and Flexural Strength of Mullite Ceramics, J. Am.
Ceram. Soc., 74 [12] 3017-22 (1991).
[46] S. Kanzaki, H. Tabata, T. Kumazawa and S. Ohta, Sintering and Mechanical-Properties of
Stoichiometric Mullite, J. Am. Ceram. Soc., 68 [1] C6-7 (1985).
[47] P. Colombo, E. Bernardo and L. Biasetto, Novel Microcellular Ceramics from a Silicone Resin,
J. Am. Ceram. Soc., 87 [1] 152-4 (2004).
[48] O. Lyckfeldt and J. M. F. Ferreira, Processing of Porous Ceramics by Starch Consolidation, J. Eur.
Ceram. Soc., 18 [2] 131-40 (1998).
[49] T. J. Fitzgerald, V. J. Michaud and A. Mortensen, Processing of Microcellular SiC Foams. 2.
Ceramic Foam Production, J. Mater. Sci., 30 [4] 1037-45 (1995).
[50] Y. Hotta, P. C. A. Alberius and L. Bergstrom, Coated Polystyrene Particles as Templates for
Ordered Macroporous Silica Structures with Controlled Wall Thickness, J. Mater. Chem., 13 [3] 496-501
(2003).
[51] J. M. Bouler, M. Trecant, J. Delecrin, J. Royer, N. Passuti and G. Daculsi, Macroporous Biphasic
Calcium Phosphate Ceramics: Influence of Five Synthesis Parameters on Compressive Strength, J. Biomed.
Mater. Res., 32 [4] 603-9 (1996).
[52] T. Fukasawa, M. Ando, T. Ohji and S. Kanzaki, Synthesis of Porous Ceramics with Complex Pore
Structure by Freeze-Dry Processing, J. Am. Ceram. Soc., 84 [1] 230-2 (2001).
[53] T. Fukasawa, Z. Y. Deng, M. Ando, T. Ohji and Y. Goto, Pore Structure of Porous Ceramics
Synthesized from Water-Based Slurry by Freeze-Dry Process, J. Mater. Sci., 36 [10] 2523-7 (2001).
[54] S. R. Mukai, H. Nishihara and H. Tamon, Formation of Monolithic Silica Gel Microhoneycombs
(SMHs) Using Pseudosteady State Growth of Microstructural Ice Crystals, Chem. Commun., [7] 874-5
(2004).

146
 [55] T. Fukasawa, Z. Y. Deng, M. Ando, T. Ohji and S. Kanzaki, Synthesis of Porous Silicon Nitride
with Unidirectionally Aligned Channels Using Freeze-Drying Process, J. Am. Ceram. Soc., 85 [9] 2151-5
(2002).
[56] D. Koch, L. Andresen, T. Schmedders and G. Grathwohl, Evolution of Porosity by Freeze Casting
and Sintering of Sol-Gel Derived Ceramics, J. Sol-Gel Sci. Technol., 26 [1-3] 149-52 (2003).
[57] A. Imhof and D. J. Pine, Ordered Macroporous Materials by Emulsion Templating, Nature, 389
[6654] 948-51 (1997).
[58] L. J. Gauckler, U. T. Gonzenbach, A. R. Studart, E. Tervoort and I. Akartuna, Porous Ceramics
derived from particle stabilized foams and emulsions, CFI, 84 [6] E36 (2007).
[59] E. S. Toberer, J. C. Weaver, K. Ramesha and R. Seshadri, Macroporous Monoliths of Functional
Perovskite Materials Through Assisted Metathesis, Chem. Mater., 16 [11] 2194-200 (2004).
[60] H. Wang, I. Y. Sung, X. D. Li and D. Kim, Fabrication of Porous SiC Ceramics with Special
Morphologies by Sacrificing Template Method, J. Porous Mater., 11 [4] 265-71 (2004).
[61] H. Kim, C. da Rosa, M. Boaro, J. M. Vohs and R. J. Gorte, Fabrication of Highly Porous Yttria-
Stabilized Zirconia by Acid Leaching nickel from a Nickel-Yttria-Stabilized Zirconia Cermet, J. Am. Ceram.
Soc., 85 [6] 1473-6 (2002).
[62] N. Miyagawa and N. Shinohara, Fabrication of Porous Alumina Ceramics with Uni-Directionally-
Arranged Continuous Pores Using a Magnetic Field, J. Ceram. Soc. Jpn., 107 [7] 673-7 (1999).
[63] K. M. Lindqvist and E. Carlstrom, Indirect Solid freeform Fabrication by Binder Assisted Slip
Casting, J. Eur. Ceram. Soc., 25 [16] 3539-45 (2005).
[64] N. W. Androff, L. F. Francis and B. V. Velamakanni, Macroporous Ceramics from Ceramic-
Polymer Dispersion Methods, A. I. Ch. E. J., 43 [11] 2878-88 (1997).
[65] P. Colombo and E. Bernardo, Macro- and Micro-Cellular Porous Ceramics from Preceramic
Polymers, Compos. Sci. Technol., 63 [16] 2353-9 (2003).
[66] T. J. Fitzgerald and A. Mortensen, Processing of Microcellular SiC Foams. 1. Curing Kinetics of
Polycarbosilane in Air, J. Mater. Sci., 30 [4] 1025-32 (1995).
[67] A. Imhof and D. J. Pine, Preparation of Titania Foams, Adv.Mater., 11 [4] 311-4 (1999).
[68] C. Ritzoulis, N. Scoutaris, K. Papademetriou, S. Stavroulias and C. Panayiotou, Milk Protein-Based
Emulsion Gels for Bone Tissue Engineering, Food Hydrocolloids, 19 [3] 575-81 (2005).
[69] B. P. Binks, Macroporous Silica from Solid-Stabilized Emulsion Templates, Adv. Mater., 14 [24]
1824-7 (2002).
[70] L. M. Rodriguez-Lorenzo, M. Vallet-Regi and J. M. F. Ferreira, Fabrication of Porous
Hydroxyapatite Bodies by a New Direct Consolidation Method: Starch Consolidation, J. Biomed. Mater.
Res., 60 [2] 232-40 (2002).
[71] K. Araki and J. W. Halloran, Porous Ceramic Bodies with Interconnected Pore Channels by a Novel
Freeze Casting Technique, J. Am. Ceram. Soc., 88 [5] 1108-14 (2005).
[72] H. X. Peng, Z. Fan, J. R. G. Evans and J. J. C. Busfield, Microstructure of Ceramic Foams, J. Eur.
Ceram. Soc., 20 [7] 807-13 (2000).
[73] F. S. Ortega, P. Sepulveda and V. C. Pandolfelli, Monomer Systems for the Gelcasting of Foams,
J. Eur. Ceram. Soc., 22 [9-10] 1395-401 (2002).
[74] F. S. Ortega, F. A. O. Valenzuela, C. H. Scuracchio and V. C. Pandolfelli, Alternative Gelling
Agents for the Gelcasting of Ceramic Foams, J. Eur. Ceram. Soc., 23 [1] 75-80 (2003).
[75] J. K. Park, J. S. Lee and S. I. Lee, Preparation of Porous Cordierite Using Gelcasting Method and
its Feasibility as a Filter, J. Porous Mater., 9 [3] 203-10 (2002).
[76] Z. P. Du, M. P. Bilbao-Montoya, B. P. Binks, E. Dickinson, R. Ettelaie and B. S. Murray,
Outstanding Stability of Particle-Stabilized Bubbles, Langmuir, 19 [8] 3106-8 (2003).
[77] E. Dickinson, R. Ettelaie, T. Kostakis and B. S. Murray, Factors Controlling the Formation and
Stability of Air Bubbles Stabilized by Partially Hydrophobic Silica Nanoparticles, Langmuir, 20 [20]
8517-25 (2004).
[78] B. P. Binks and T. S. Horozov, Aqueous Foams Stabilized Solely by Silica Nanoparticles,
Angewandte Chemie-Int. Ed., 44 [24] 3722-5 (2005).
[79] U. T. Gonzenbach, A. R. Studart, E. Tervoort and L. J. Gauckler, Ultrastable Particle-Stabilized
Foams, Angewandte Chemie-Int. Ed., 45 [21] 3526-30 (2006).
[80] U. T. Gonzenbach, A. R. Studart, E. Tervoort and L. J. Gauckler, Macroporous Ceramics from
Particle-Stabilized Wet Foams, J. Am. Ceram. Soc., 90 [1] 16-22 (2007).
[81] R. A. Mach, F. Von Issendorff, A. Delgado and A. Ortona, Experimental Investigation of the
Oxidation Behavior of Si-SiC Foams, in Advances in bioceramics and porous ceramics: ceramic engineering
and science proceedings, 29 [7] 299-311 (2009).
[82] J. Adler, M. Teichgraber, G. Standke, H. Jaunich, H. Stover and R. Stotzel, Open Cell Expanded
Ceramic with a High Level of Strength, and Process for the Production Thereof, E. P. 0 907 621, (1999).

147
 [83] B. W. Hillig, R. L. Mehan, C. R. Morelock, V J. DeCarlo and W. Laskow, Silicon-Silicon Carbide
Composites, Am. Ceram. Soc. Bull., 54 [12] 1054-56 (1975).
[84] M. W. Chase Jr., NIST-JANAF themochemical tables, fourth edition, J.Phys. Chem. Ref. Data,
1-1951 (Monograph 9) (1998).
[85] A. Füssel, D. Böttge, J. Adler, F. Marschallek and A. Michaelis, Cellular Ceramics in Combustion
Environments, Adv. Eng. Mater., 13 [11] 1008-15 (2011).
[86] A. Ortona, S. Pusterla, P. Fino, F. R. Mach, A. Delgado and S. Biamino, Aging of Reticulated Si-SiC
Foams in Porous Burners, Adv. Appl. Ceram., 109 [4] 246-51 (2010).
[87] M. M. Kamal and A. A. Mohamad, Combustion in Porous Media, Proc. Inst. Mech. Engrs. Part A,
220 [5] 487-508 (2006).
[88] D. Trimis and F. Durst, Combustion in a Porous Medium - Advances and Applications, Combust.
Sci. Technol., 121 [1-6] 153-68 (1996).
[89] M. Abdul Mujeebu, M. Z. Abdullah, M. Z. Abu Bakar and A. A. Mohamad, Development of
Premixed Burner Based on Stabilized Combustion Within Discrete Porous Medium, J. Porous Media, 14
[10] 909-17 (2011).
[90] S. Mobauer, O. Pickenacker and D. Trimis, Proceedings of the Fifth International Conference on
Technologies Combustion for a Clean Environment (Clean Air V), 519 (1999).
[91] N. Djordjevic, P. Habisreuther and N. Zarzalis, A Numerical Investigation of the Flame Stability in
Porous Burners Employing Various Ceramic Sponge-Like Structures, Chem. Eng. Sci., 66 [4] 682-88
(2011).
[92] T. Brehmer, F. Heger, K. Lucka, J. von Schloss, Y. Abu-Shareck, D. Trimis, A. Heeb, G. Kob, T. C.
Hayashi, J. C. F. Pereira, M. Founti, D. Kolaitis, M. Molinari, A. Ortona, J.-B. Michel and P. Theurilliat, 7th
International Conference on Energy for a Clean Environment, Lisbon, pp 11 (2003).
[93] F. Von Issendorff, V. Stamatov, A. Mach and D. Trimis, Proceedings of the Eighth International
Conference on Technologies Combustion for a Clean Environment (Clean Air VIII), 321 (2005).
[94] S. A. Leonardi, J. P. Gore and R. Viskanta, Experimental Investigation of Partially-Premixed
Combustion in a Novel Porous Radiant Burners, in Proceedings of the National Heat Transfer Conference,
1 915-21 (2001).
[95] A. Mohammadi and A. Jazayeri, Numerical Simulation of Combustion in Porous Media, Numerical
Simulation - From Theory to Industry, Edited by Dr. Mykhaylo Andriychuk (2012).
[96] M. V. Twigg and J. T. Richardson, Fundamentals and Applications of Structured Ceramic Foam
Catalysts, Ind. Eng. Chem. Res., 46 [12] 4166-77 (2007).
[97] P. Mlheims, S. Marz, S. P. Mller and B. Kraushaar-Czarnetzki, Chem. Ing. Tech., 83 286 (2011).
[98] P. Mlheims and B. Kraushaar-Czarnetzki, Ind. Eng. Chem. Res., 50 9925 (2011).
[99] M. Lacroix, L. Dreibine, B. de Tymowski, F. Vigneron, D. Edouard, D. Bgin, P. Nguyen, C. Pham,
S. Savin-Poncet, F. Luck, M.-J. Ledoux and C. Pham-Huu, Appl. Catalysis A, General (2010).
[100] R. Fisher, C. Burkland and W. Busmante, Ceram. Eng. Sci. Proc., 6 806 (1985).
[101] F. I. Hurwitz, W. Steffier, Koenig and J. Kiser J, NASA Tech Briefs, 36 (2009).
[102] A. Ortona, S. Pusterla and S. Gianella, An Integrated Assembly Method of Sandwich Structured
Ceramic Matrix Composites, J. Eur. Ceram. Soc., 31 [9] 1821-26 (2011).
[103] A. Ortona, S. Pusterla and S. Gianella, Conference proceedings HTCMC7, 694 (2010).
[104] M. Roeb and H. Mller-Steinhagen, Science 329 773 (2010).
[105] M. Romero, R. Buck and J. E. Pacheco, An Update on Solar Central Receiver Systems, Projects
and Technologies, J. Sol. Ener. Eng., Transactions of the ASME, 124 [2] 98-108 (2002).
[106] M. Becker, Th. Fend, B. Hoffschmidt, R. Pitz-Paal, O. Reutter, V. Stamatov, M. Steven and
D. Trimis, Theoretical and Numerical Investigation of Flow Stability in Porous Materials Applied as
Volumetric Solar Receivers, Sol. Energy, 80 [10] 1241-48 (2006).
[107] H. Cafri, Method for Manufacturing a Solar Radiator Absorber, US Patent Application
20110308514 (2011).
[108] M. Becker, S. Cordes and M.Bohmer, Proc. of 6th Int. Symp. on Solar Thermal Concentrating
Tech. Edited by CIEMAT, 2 945 (1992).
[109] T. Fend, High Porosity Materials as Volumetric Receivers for Solar Energetic, Opt. Appl., 40 [2]
271-84 (2010).
[110] T. Fend, B. Hoffschmidt, R. Pitz-Paal, O. Reutter and P. Rietbrock, Porous Materials as Open
Volumetric Solar Receivers: Experimental Determination of Thermophysical and Heat Transfer Properties,
Energy, 29 [5-6] 823-33 (2004).
[111] L. Xia, S.-Y. Wu, Y.-R. Li, Renew. Energ., 41 1 (2012).
[112] M. Roeb, Proceedings of 13th Solar PACES International Symposium, 20 (2006).
[113] J. Adler, G. Standke and Z. Keram,55 694 (2003).
[114] J. Adler, G. Standke and Z. Keram,55 786 (2003).

148
 [115] A. Ortona, C. D’Angelo and G. Bianchi, Monitoring Sandwich Structured SiC Ceramics Integrity
with Electrical Resistance, NDT and E Int., 46 [1] 77-82 (2012).
[116] S. Voss, D. Trimis and J. Valldorf, 18th World Hydrogen Energy Conference 2010 - WHEC 2010,
Parallel Sessions Book 6: Stationary Applications/Transportation Applications, (2010).
[117] Z. Al-Hamamrea, S. Vo and D. Trimis, Int J Hydrogen Ener, (2009).

149
5. Characterisation of ageing Si–SiC foams in
porous burners

5.1 Introduction

Combustion processes are fundamental for the production of energy. To allow combustion
to occur, different burner types are available depending on the type of fuel and the
operating conditions. Nowadays, a matter of concern is the efficiency of the burner and
the reduction of its emissions. Volumetric porous burners can significantly improve these
features because this technology allows combustion at comparably low emissions and high
surface loads. The combustion is stabilised in an inert open porous ceramic structure that
permits an internal recuperation of heat, a homogeneous temperature field and a compact
design. The basic set-up of such a burner is shown in Fig. 5.1. The porous burner
technology [1-3], applied in many different fields of energy conversion [4, 5], is strongly
dependent on high temperature resistant porous components.
Different structures and materials have been tested as porous medium in volumetric
burners. Reticulated foams typically have porosities in the range 7090% and pore sizes in
the range 0.45 mm [6]. Pore size plays a fundamental rule in foam selection because it
influences the combustion process via its effect on heat transport. For this application
(e.g. radiative burner), high pores sizes ( 5 mm) are more effective [7].
Ceramic foams are produced mainly through the following methods [8]:
a) direct foaming of a liquid slurry;
b) burnout of fugitive pore formers;
c) replication of a sacrificial foam template.
In the direct foaming of a liquid slurry, a gas is inflated into a suspension or liquid media
and subsequently set in order to block bubbles. In most cases, the consolidated foam is
sintered at high temperatures to obtain a strengthened porous ceramic.
Burnout of fugitive pore formers, also known as sacrificial template method, consists in
the preparation of a ceramic green body with a homogeneously dispersed sacrificial phase
which is extracted, generally during heat treatment, to generate pores in the bulk.
Replication technique consists in the impregnation of a cellular structure with a ceramic

150
suspension or liquid polymer precursor in order to produce a macro-porous ceramic
exhibiting the same morphology of the original porous material [9, 10]. These techniques
allow the manufacturing of foams with the higher values of porosity and pore size [11],
suitable for infrared heating applications.
The most used replication technique is the polymer replica, in which a highly porous
polymeric foam is initially soaked into a ceramic suspension, shaken in order to eliminate
the excess slip and pyrolysed by heating between 300°C and 800°C. The ceramic coating
can be sintered or, in the case of Si-SiC, infiltrated with molten Si. A modification of the
polymer replica method consists in pyrolysing a high carbon yield polymeric foam
(e.g. phenolic) into vitreous carbon and coating it using chemical vapour deposition
(CVD) [12].
A recently developed technique consists in electrospraying a foam template with ethanol
based nanohydroxyapatite suspension, drying, pyrolysing at 400°C and sintering [13].
Ceramics are among the most suitable bulk materials for porous burners because of their
high working temperatures, chemical stability and resistance to erosion and wear [6].
The most commonly used ceramics are alumina (Al2O3), silicon carbide (SiC) and zirconia
(ZrO2).
SiC and Si-SiC ceramics do oxidize at temperatures above 600°C, but if formed silica
layer remains stable, they can be used up to 14001600°C [14]. Compared to oxides, they
have higher thermal conductivity and emissivity, a lower coefficient of thermal expansion
and very good thermal shock resistance.
Characterisation of aged Si-SiC foams has been already performed [7, 15-17]. Foams
behavior and limitations under different partial oxygen pressures and temperatures have
been experimentally investigated [6]. This chapter aims to track the changes of Si-SiC
foam properties during the first part of a long time aging condition, typical of industrial
applications.

Figure 5.1: Basic set-up of porous burner [18].

151
5.2 Experimental procedure

ErbiSiC R foams from ERBICOL SA Switzerland were used. Manufacturing of ceramic

foams is based on commercially available polyurethane foams which are used as templates
[9]. The process starts with an -SiC powder slurry which is applied as a dip coating on
the polymer network (ligaments), followed by pyrolysis. To improve the strength of the
ligaments ceramic network, a second polymer infiltration and pyrolysis is also performed.
Finally, silicon melt infiltration is used to obtain a reaction bonded -SiC-Si matrix which
binds the dispersed -SiC powder. The final ceramic is an isotropic three-dimensional
network of hollow Si-SiC ligaments (bulk density: 2.83 g/cm3) with an open porosity of
> 87%. The silicon content of as received ErbiSiC foams is  30 wt.% [18]. Pore sizes are
about 47 mm and are equivalent to those of commercial reticulated foams with a pore
size of  10 pores per inch (PPI). As prepared foams (185 x 139 x 15 mm) were cut with
diamond saw in seven pieces (139 x 25 x 15 mm) (Fig. 5.2). The first two dimensions
have tight tolerances (± 0.05 mm) because of machining, while the third dimension was
kept as produced and thus with a ± 1 mm tolerance. Foams were thus carefully
reassembled and placed onto the burner rig.

Figure 5.2: Samples preparation before tests: (left) as produced, (right) cutting scheme.

152
5.2.1 Aging campaign

The porous medium burner rig consists of 16 porous medium burners, each with a
methane gas supply, a mixing chamber, a blower, a fuel control valve and a control unit
(Fig. 5.3). The system can operate for 24h, 7 days per week with different cycles, powers
and air/fuel ratios. Data loggers collect the data for each burner, such as operating time
and number of cycles. For the present investigations, four burners were used. The power
and the excess air ratio of the system were adjusted to 25 kW and 1.3 respectively.
These operating conditions (1400°C) fall into the passive oxidation regime. The samples
under investigation were reassembled and placed as combustion zone in one of the porous
medium burners. The investigations were carried out for 1, 5, 10, 50, 100 and 1000h with
a 20 min heating and 1 min cooling cycle. Details are displayed in table 5.I.

Figure 5.3: Set-up of porous medium burner test rig (source: CERPOR project).

Mass change
Foam no. N° cycles Aging [h]
after test %

32 1 0.33 0.08

34 3 1.00 0.01

35 15 5.00 0.14

36 35 11.50 0.15

37 153 51.00 0.22

38 298 99.50 0.10

39 3000 1000.00 nd

Table 5.I: Main data of foam aging.

153
5.3 Characterisation methods

5.3.1 Bending and compression tests

Three point bending tests (Fig. 5.4) were performed with a universal testing machine
(Z050; Zwick/Roell Ulm, Germany) using a span of 100 mm. Samples dimensions (185 x
139 x 15 mm) were chosen in order to have three cells engaged along the specimen
thickness [19]. Since it was not possible to have reliable strain measures, only force was
acquired during tests. Crosshead speed was 0.01 mm/s.

Figure 5.4: Foam sample under three-point bending showing foam cells engaged along the
thickness.

Tests for resistance to compression were performed by applying a crosshead speed of

0.5 mm/min (Sintech 10/D, MTS, Eden Prairie, Minnesota USA) using a growing load to
the specimens until these come to fracture. The samples dimensions were 25 x 25 x
15 mm.

154
5.3.2 Morphological characterisation

SEM/EDS measurements have been carried out to evaluate the chemical and
morphological stability of the SiC foams using a scanning electron microscopy (Leo 1450
VP; Leo, Oberkochen, Germany) equipped with EDS (model 7353; Oxford Instrument,
Abingdon, UK). Differently aged samples have been observed either on fracture surfaces
or in polished zones, with the purpose to evaluate the material behavior during aging
treatments.
Porosimetry measurements have been carried out using a quicksilver intrusion apparatus
Porosimeter 2000 coupled to a Macro-pores Unit 120 (Fisons Instruments, San Carlos,
CA, USA) to detect any surface modifications during aging. The portions of the material
to be subjected to infiltration of mercury were inserted in an oven at 200°C for a few hours
before the test, in order to make the pores completely dry.

5.3.3 Thermal characterisation

Thermal stability and oxidation resistance of the samples in all the aging states were
compared using a thermogravimetric analyser (TG/SDTA851; Mettler Toledo,
Schwerzenbach, Switzerland) coupled with a mass spectrometer (MS Balzers, Balzers,
Liechtenstein), being able to detect the gaseous products arising from sample
decomposition or oxidation. Specimens of 3 x 3 x 3 mm were used for thermal analysis.

5.4 Results and discussion

Figure 5.5 shows stress data calculated from the bending tests. Each value was calculated
over the 7 samples. Foam 38 value (Tab. 5.I) was calculated over 4 samples because three
of them accidentally broke during transportation. Figure 5.5 shows a reduction of the
bending strength of the foams after 5h. This reduction could be ascribed to crack
formations during the first ON/OFF cycles provoked by thermal shock [15, 18]. Because
of the passive oxidation conditions of these experiments [18], for longer periods,

155
the formation of silica could have a healing effect on cracks, causing a slight increase in
bending strengths. Indeed when aging foams in these conditions, they tend to experience a
mass gain for the first  1000h due to the growth of a silica scale.
Within the burner, central samples (no. 4 in Fig. 5.6) in long aged foams seem to have
lower mechanical properties than lateral ones (nos. 1 and 7 in Fig. 5.6) probably due to
local active oxidation phenomena [18], but in general, sample position within the burner
does not seem to be more significant than other factors affecting foams bending strengths,
such as cell shape and orientation, ligament thickness, ligament cross-section and ligament
cavity sharpness [10, 20]. These detrimental effects are evident in figure 5.7 where three-
point bending strength of unaged Si-SiC foams are compared with those calculated from
the Gibson-Ashby model [21] which does not consider local defects. The Gibson-Ashby
equation for the three-point bending of foams is

(5.1)

where is the three-point foam bending strength, 0 is the dense material three-point
bending strength (taken as 420 MPa for Sapco Si-SiC), / 0 is the relative density and C is
a constant taken as C = 0.18 [22].

Figure 5.5: Bending strength in function of aging time.

156
Figure 5.6: Bending strength in function of samples position into the burner for three
different aging times.

Figure 5.7: Comparison of theoretical bending strength values from Gibson-Ashby model
with experimental values on as produced foam samples.

157
In figure 5.8 the compression tests data are reported (3 specimens for the as produced as
well as for the sample aged for 1000 h). When a ceramic foam is subjected to a
compression test, the stress-strain has a first area with resulting linear elastic behavior for
small deformations. [23]. After reaching the breakdown load in correspondence of which
there is a propagation of macroscopic cracks and resulting in a load drop, there is a second
zone, named plateau; the plateau is connected to an elongation rather accentuated with
constant effort, due to a continuous collapse of the cells to brittle fracture of the bridges.
Finally, for high strain when the porous structure of the material appears to be completely
destroyed the load grow back quickly. Due to the foams geometry, these are subjected to
different stress states, including the cutting effort, depending on the direction of the strut
subjected to stress. Comparing the two graphs, it can be seen as the mechanical
characteristics of the foams do not vary substantially with increasing thermal stress.

0 hrs

400

300
A
Load [N]

200 B
C
100

0
0 0,01 0,02 0,03 0,04
Strain [mm]

1000 hrs

400

300
A
Load [N]

200 B
C
100

0
0 0,01 0,02 0,03 0,04
Strain [mm]

Figure 5.8: Compression tests for as produced sample and for 1000 hrs aged sample.

158
Porosimetry results are compared in figure 5.9 (as produced material and aged material,
1000 h). The result of the porosimetry shows the volume in cubic millimeters of voids
with a given radius, whether as regards the macro-pores and as regards the micro-pores.
In both graphs it can be observed that the volume of voids associated with each radius is
very low, less than three cubic millimeters. The as produced sample has an average pore
radius equal to 1.028 microns while that aged 1000 hours a mean radius of 0.066 microns;
these values are comparable with those obtained by Ding et al., who found an average
pore radius of 1.9 microns by analyzing a sample of porous silicon carbide obtained by
reaction bonding in air at 1450°C for 4 hours [24].

Figure 5.9: Mercury intrusion porosimetry (left) as produced, (right) after aging of
1000 hrs.

Figure 5.10 shows the morphology and the chemical composition of two typical foam
sectors, one from the as produced foam and the second one from the 1000h aged one. In
both cases it is possible to observe in the core zone the presence of pure metallic Si coated
with a thick wall of dense Si-SiC. The Si-SiC layer does not show any structural
modification during aging, neither along the thickness nor in the inner interface between
Si-SiC and Si. In the case of aged samples, it is also possible to note a thin superficial
layer of SiO2 derived by the passive oxidation of the substrate.
Thermal stability tests results are summarized in figure 5.11. The thermo-gravimetrical
pattern is coupled with the gases mass spectrometer analysis from room temperature to
1550°C. They show limited CO2 emissions from 700°C to 800°C for the as produced
sample and just a little spot at 1200°C for the aged one. The curve which indicates the
trend of the carbon monoxide can not be analyzed because, since the measurement is done
in air, having the same atomic weight of nitrogen N2, i.e. 28, the small weight changes
would not be visible. No significant TG patterns variations occur in the temperature range
between RT and 1400°C. This behavior appears either for untreated and aged foams,

159
without any appreciable difference. With the mass analyser, it was possible to observe that
for the untreated foam oxidation starts at low temperature and produces a SiO2 passivated
superficial layer.
As aging proceeds, passive oxidation becomes less important after 15h. Low CO2
emission (mass 44 in Fig. 5.11) probably indicates the onset of active SiC oxidation.
The SiO2 layer could not be completely coherent with the SiC substrate and during
thermal cycling, part of SiO2 coating could collapse probably because of the low aging
temperature, which is not sufficient to produce a continuous and adhesive film. For this
reason, the thickness of the silica layer does not exceed 45 m and it is quite constant
during the whole oxidation process.

Figure 5.10: SEM images and Si maps of as produced and 1000h aged foam.

160
Figure 5.11: Thermogravimetrical analyses coupled with mass spectrometry for the
untreated foam (left) and the 1000h aged foam (right).

5.5 Conclusion

Si-SiC foams were aged under lean methane combustion for different times. Samples were
mechanically and morphologically tested. Mechanical strength seems to decrease by
roughly 30% during the very first hours of operation. This is due to ligament crack
formation arising from shocks caused by the ON/OFF aging procedure. These cracks
could be healed during silica formation as aging proceeds. Besides the formation of a
silica layer, no significant variations were detected during aging up to 1000h.

161
5.6 References

[1] D. Trimis and F. Durst, Combustion in Porous Medium - Advances and Applications, Combust. Sci.
Technol., 121 [1-6] 153-68 (1996).
[2] D. Trimis, Porenbrennertechnologie - ein űberblick, GWF Gas Erdgas, 2 92-9 (2006).
[3] S. Wood and A. T. Harris, Porous Burners for Lean-Burn Applications, Prog. Energy Combust. Sci.,
34 [5] 667-84 (2008).
[4] S. Diezinger, P. Talukdar, F. von Issendorff and D. Trimis, Verbrennung von Niederkalorischen
Gasen in Porenbrennern, Gaswärme Int., 3 192-7 (2005).
[5] M. M. Kamal and A. A. Mohamad, Combustion in Porous Media, Proc. Inst. Mech. Eng. A, J. Power
Energy, 220 [5] 487-508 (2006).
[6] J. R. Howell, M. J. Hall and J. L. Ellzey, Combustion of Hydrocarbon Fuels Within Porous Inert
Media, Prog. Energy Combust. Sci., 22 [2] 121-45 (1996).
[7] M. Scheffler and P. Colombo, Cellular Ceramics: Structure, Manufacturing, Properties and
Applications, WILEY-VCH Verlag GmbH, Weinheim, Germany, (2005).
[8] P. Colombo, Conventional and Novel Processing Methods for Cellular Ceramics’, Philos. Trans. R.
Soc. Lond. A, 364 [1838] 109-24 (2006).
[9] K. Schwartzwalder and A. V. Somers, Method of Making Porous Ceramic Articles, US Patent No.
3090094, (1963).
[10] R. Cahn, P. Haasen and E. J. Kramer, Material Science and Technology: A Comprehensive
Treatment’, Structure and properties of ceramics, Weinheim, Wiley-VCH, 11 (1999).
[11] A. R. Studart, U. Gonzenbach, E. Tervoort and L. J. Gauckler, Processing Routes to Macroporous
Ceramics: A Review, J. Am. Ceram. Soc., 89 [6] 1771-89 (2006).
[12] A. J. Sherman, R. H. Tuffias and R. B. Kaplan, Refractory Ceramic Foams: A Novel, New High
Temperature Structure, Am. Ceram. Soc. Bull., 70 1025-29 (1991).
[13] A. Muthutantri, J. Huang and M. Edirisinghe, Novel Method of Preparing Hydroxyapatite Foams,
J. Mater. Sci., Mater. Med., 19 [4] 1485-90 (2008).
[14] O. Pickenäcker, K. Pickenäcker, K. Wawrzinek, D. Trimis, W. E. C. Pritzkow, C. Műller,
P. Goedthe, U. Papenburg, J. Adler, G. Standke, H. Heymer, W. Tauscher and F. Jansen, Innovative
Ceramic Materials for Porous Medium Burners, II’ Interceram, 48 [6] 424-33 (1999).
[15] J. Adler, G. Standke, M. Jahn and F.Marschallek, Cellular Ceramics Made of Silicon Carbide
Ceramics for Burner Technology’, in Advances in bioceramics and porous ceramics: ceramic engineering
and science proceedings, Hoboken, NJ, Wiley, 29 [7] 270-86 (2009).
[16] X. Zhu, Y. Sakka, S. Tan and D. Jiang, Bend Deformation Behavior of Silicon Carbide Reticulated
Porous Ceramics, Mater. Sci. Forum, 534-36 [PART 2] 941-44 (2007).
[17] J. Zhang, Y. Fu and X. Zeng, Compressive Properties of Open Cell Ceramic Foams, Trans. Nonferr.
Met. Soc. Chn, 16 [SUPPL] s453-s56 (2006).
[18] R. A. Mach, F. von Issendorff, A. Delgado and A. Ortona, Experimental Investigation of the
Oxidation Behavior of Si-SiC Foams’ in Advances in bioceramics and porous ceramics: ceramic engineering
and science proceedings, Hoboken, NJ, Wiley, 29 [7] 299-311 (2009).
[19] Standard Test Method for Flexural Strength of Advanced Ceramics with Engineered Porosity
(Honeycomb Cellular Channels) at Ambient Temperatures, C1674-08, ASTM, West Conshohocken, PA,
USA, (2008).
[20] J. Luyten, I. Thijs, W. Vandermeulen, S. Mullens, B. Wallaeys and R. Mortelmans, Strong Ceramic
Foams From Polyurethane Templates, Adv. Appl. Ceram., 104 [1] 9-21 (2005).
[21] L. Gibson and M. Ashby, Cellular Solids: Structure and Properties, Oxford, Pergamon, (1988).
[22] R. Brezny, Mechanical Behavior of Open Cell Ceramics, PhD thesis, Penn State University, State
College, PA, USA, (1990).
[23] S. Elbir, S. Yilmaz, A.K. Toksoy and M. Guden, SiC-Particulate Aluminum Composite Foams
Produced by Powder Compacts: Foaming and Compression Behaviour, J. Mater. Sci., 38 [23] 4745-55
(2003).
[24] S. Ding, Y. P. Zeng and D. Jiang, In Situ Reaction Bonding of Porous SiC Ceramics, Mater.
Charac., 59 [2] 140-3 (2008).

162
6. Si-SiC heat exchangers for recuperative gas
burnerns with highly structured surface
elements: CEREXPRO project

6.1 Introduction

Industrial production is a major consumer of energy. The energy balance of the EU 27

(Fig. 6.1) in 2005, as reported by Eurostat, shows a total end-energy consumption in the
industry sector of more than 13000 PJ representing 28% of the overall end-energy
consumption in all sectors. Based on data of the same year for Germany (Source: German
Federal Ministry of Economics and Technology) one can asses that about 60% of the
industrial end-energy use is spent for industrial process heating (total energy consumption
in the German industrial sector of 2550 PJ and 1562 PJ for process heat), while more than
80% of process heat is covered by fossil fuels. Assuming that the German distribution is
representative and can be extrapolated to the European dimension, a total of more than
6000 PJ are spent for process heat generated by burning of fossil fuels on the European
level, representing more than 12% of the total end-energy consumption of EU-27 and at
the same time approx. 9.7% of the total fossil fuel consumption.

Figure 6.1: Development of end-use energy by countries in EU27 (source: CECOF).

Consequently, an efficiency increase in process heating of only a few percent would result
in a significant reduction of energy consumption and CO2-emissions at the European level.

163
Industrial branches with high energy consumption are for example iron and steel making,
metal processing and ceramics production. Most of the processes are operated at a high
temperature level. Heat recovery at this high temperature level is essential in terms of
efficiency. The most common way for heat recovery is preheating of air or other gas flows
by using the energy of the hot waste gas flows. The so-called recuperative or regenerative
heat exchanger systems [1], which may be integrated in the burner assemblies,
are commonly used for this purpose.
Among the different heat recovery systems, centralized systems like central recuperators
show the lowest efficiency increase in practice. This is due to the long collection and
distribution pipes needed and the associated heat losses at the high temperature level.
The use of burners equipped with recuperators results in significant efficiency increase.
However, their efficiency is proportional to the heat exchanger surface which is limited by
the overall volume and burner length restrictions and by the required heat exchanger and
its material limitations.
The highest efficiency can be reached using regenerative burners. However, the
complexity of regenerative burner systems is higher, due to the discontinuous alternately
mode of operation utilizing several regenerator matrices.
In all cases, it can be presumed that a 100 K decrease in the flue gas temperature after heat
recovery results in an estimated reduction in fuel consumption of about 5% (Fig. 6.2).

Figure 6.2: Combustion efficiency with respect to the exhaust gasses temperature before
heat recovery; experimental results for different types of heat recovery [2].
164
The heat exchangers are manufactured in industrial numbers of pieces using technologies
of primary shaping. Two different material groups are used. One group is highly alloyed
steel grades. These materials can be casted and then joined by welding processes. A great
variety in the shape and structure of the surface of the heat exchangers is possible.
The metallic recuperators are limited in the application temperatures to about 1050°C.
The use of alloys with higher application temperature is avoided in industry due to the
significant increase of the materials price by an only marginal increase in application
temperature. A more recent development are recuperators using the effect of gap flow.
These heat exchangers exhibit an increased heat recovery but are also limited in
application temperature.
Ceramic recuperators used nowadays in industrial burners are tubular (Fig. 6.3). The parts
are formed using a slip casting process. This technology allows a moderate structuring of
the outer surface of the heat exchangers. Due to the deposition conditions the structures at
the inner surface of the heat exchanger are less sharp defined. Joining of flanges or other
additional parts can only be done in the green unburned state of the ceramic parts.
The parts are subsequently transformed into ceramics and infiltrated with liquid silicon.
A mechanical treatment of the hard ceramic parts is extremely difficult connected with
great expenses and thus reduced to a minimum in industrial production. The ceramic is
usually called Si-SiC – silicon infiltrated silicon carbide. It contains some percent of
metallic silicon what is limiting the application temperature to about 1350°C.

Figure 6.3: Burners with ceramic heat exchanging parts; left: recuperative (source:
Aichelin), right: regenerative (source: WS).

165
The use of ceramic materials brings the following advantages:
 operation and heat recovery at the higher temperatures up to 1300°C;
 higher corrosion resistance in many harsh environments;
 no scale formation;
 no creeping at high temperatures.
On the other hand, they present the following drawbacks:
 lower amount of heat can be recovered in comparison with the metal recuperator,
due to manufacturing restrictions;
 fragile mechanical behavior, which implies that components are always over-
dimensioned, very simple surface structures and difficulty in machining and in
joining;
 restrictions due to dust transported with the gas flows;
 expansive manufacturing process
 heat recovery devices should be directly integrated into the thermal processing
facility, to minimize thermal losses and utilize the energy in an immediate manner.
Despite the obvious benefits of ceramic heat exchanger components for heat recovery at
high temperature applications, the penetration of such technologies in industrial furnaces
is restricted by comparatively higher prices and larger size of such components. Relatively
simpler ceramic heat exchanger geometries with large dimensions are used up to now,
due to several manufacturing and operational obstructions. Increasing their efficiency will
lead to decrease their dimension overcoming the above-mentioned economical problems.
Common recuperator lengths are in the order of 0.5 m and hence longer than the thickness
of a typical furnace wall. Heat enhancement with structured surfaces is limited by the
associated ceramic production technologies. Given these technological obstructions and
the dimensional limitations, the level of heat recovery is restricted. State of the art
recuperative burners show an air preheating and flue gas temperature level after the
recuperator in the range of 500700°C resulting in heat losses through the flue gas of
about 2535% [2]. Regenerative burners (typical nominal power range around 200 kW)
operate with higher preheating temperatures of the combustion air. Due to the larger
surface the regeneration temperature reaches 8595% of the flue gas temperature.
The project “CEREXPRO” (CERamic heat EXchangers with enhanced materials
PROperties) was aimed to develop a new generation of ceramic heat exchangers for high
temperature heat recovery. CEREXPRO project has been constituted by eight partners

166
(Tab. 6.I); the role of each partner is reported in table 6.II. The project duration is been of
48 months, from 01-04-2009 to 31-03-2013. CEREXPRO is been an FP7 (7th Framework
Programme) collaborative project financially supported by the EUROPEAN
COMMISSION in the programme “Energy.2008.8.1.1: Energy efficiency of industrial
heat exchangers and boilers”.

Partecipant Partecipant legal Organisation

Short name Country
no. name type
Technische
Universität
1 University TU BAF DE
Bergakademie
Freiberg
Scuola Universitaria
2 Professionale della University SUPSI CH
Svizzera Italiana
3 Politecnico di Torino University POLITO IT
Institut Français du
Research
4 textile et de IFTH FR
Institute
l'habillemen
5 ERBICOL S.A. SME ERBICOL CH
WS
6 Wärmeprozesstechnik SME WS DE
GmbH
7 Aichelin Ges.m.b.H. Industry AICHELIN AT
8 NOXMAT GmbH Industry NOXMAT DE
Gesellschaft zur
Förderung des European
9 CECOF DE (EU)
Maschinenbaues Committee
mbH
Table 6.I: Partnership of CEREXPRO.

The project aims to develop a new generation of ceramic heat exchangers for high
temperature heat recovery with the target of significantly reducing the size and or weight
of these components. A reduction of production price of these components is aspired by
simplifying the manufacturing process and allowing a higher flexibility in the heat
exchanger geometry. The fundamental idea is the manufacturing of ceramic heat
exchangers with a highly structured surface to increase heat transfer by permanently
disturbing the formation of flow profiles of the gas flows (combustion air and flue gas).
The main resistance to heat transfer from a contact fluid to a solid surface (or vice versa)
is due to slowly moving fluid layers adjacent to the wall. In order to increase the heat
transfer coefficient without the drawback of high pressure losses of a turbulent flow at
high velocities, artificially generated turbulent boundary layers, or generally boundary
layers with very low thickness are generated by breaking them apart. Generally, a high
heat transfer rate can be achieved by increasing the heat transfer surface area or the heat

167
transfer coefficient or both of them simultaneously. The most effective heat transfer
enhancement can be achieved by using fins (plate, interrupted or pin-fins) as elements for
heat transfer area extension, by the fin geometry and arrangement and also increasing the
heat transfer coefficient. A variety of fin element designs and associated technologies are
used for heat transfer enhancement like plate fin heat exchangers, tube fins, wavy fins,
offset strip fins, louvered fins and also pin fins. These technologies allow high area
densities typically in the range of 1500 m2/m3 up to over 5000 m2/m3. Fins are effective in
heat transfer enhancement only if they exceed the boundary layer thickness, resulting in a
major part of the heat transfer area being exposed to a free fluid stream.
The heat transfer coefficient on the extended surface may be lower or higher than that on a
not finned surface, e.g. plain fins increase the heat transfer surface but may result in a
reduced heat transfer coefficient, while interrupted fins (strip, louvered) or pin fins
(staggered or in-line arrangement) provide both an increased surface and increased heat
transfer coefficient.

Partner Role
- Experimental and numerical evaluation of heat transfer enhancing
structure;
TU BAF - Heat exchanger simulations;
- Design and performance tests;
- Alternative applications for heat exchangers.
- Material and process development of ceramic conversion;
SUPSI
- Evaluation of textile templates and development in slurry composition.
- Detailed characterization of material properties in manufacturing
POLITO
before and after use.
- Selection of fiber/filament materials;
IFTH - Development of textile structures;
- Development of associated manufacturing procedures.
- Development of production processes (joining, slurry infiltration,
ERBICOL transformation to ceramics, silicon infiltration);
- Manufacturing of ceramic samples and prototypes.
- Specifications of burners with integrated air preheating;
WS - Integration of new heat exchangers prototypes for application;
- Testing and evaluation of prototypes.
- Specifications of burners with integrated air preheating;
- Integration of new heat exchangers prototypes for application;
AICHELIN & NOXMAT
- Testing and evaluation of prototypes;
- Testing of alternative applications of new heat exchangers.
- Market study concerning burners with air preheating;
CECOF - Organization and coordination of dissemination actions;
- Guidance of end-users.
Table 6.II: Role of CEREXPRO’s partner.

The use of precursors/template materials taken from the textile industries gives a great
degree of freedom in the design of the surface structures. The first technical challenge is

168
the production of three dimensional textiles with a very stable geometry. This must be
combined with the use of fibres or filaments based on a polymer material which can be
easily worked and with a high materials yield into ceramics. The structures must not lose
their shape during the whole manufacturing process till they are finally converted into
silicon carbide ceramics. The ceramic heat exchangers must be gas tight for application.
The use of structured geometries typical of the textile industry followed by their
conversion into a ceramic is the main innovation of this project. Although the principle is
not new, what makes it innovating is the unusual combination of such different production
fields which in turn leads to a high design flexibility of the heat exchanger [3].
To facilitate the development, this approach was applied on an existing robust Si-SiC heat
exchanger component successfully applied in industrial furnaces (Fig. 6.4). This approach
guaranteed application safety and robustness. Due to the improved efficiency, we proved
that a significant size reduction or, alternatively, an increase in the heat recovery level can
be achieved because of the higher heat transfer rate of the heat exchange elements.

Figure 6.4: Concept of the CEREXPRO burner: schematic cross-section of the

recuperator.

The size and shape of the structures are defined taking into account limitation of the
manufacturing process and of the geometry of the targeted applications. The structures are
examined intensively by numerical and physical experiments. The numerical
investigations are used for calculating the influence of different patterns of the structures
on the thermal and fluid dynamical properties of the heat exchanger elements.
The technical efforts can be minimized by using commercial numerical simulation
software. The aim of these calculations is a maximum increase in heat transfer connected
with a moderate increase in pressure loss over the structures. The numerical models are

169
reduced to single cells reproducing the smallest repeating units that will create the heat
transferring surfaces. The results are extended to calculate greater structures.
Physical models of structured surfaces were produced using steel materials. This gives the
opportunity to realise samples with a very good defined shape and pattern. In a second
step ceramic samples are used. These samples are produced during the development of the
ceramic conversion processes. The performance of the samples is examined in a wind
tunnel at moderate temperatures. The results serve as proof of the numerical calculations
and the extension of results for single cells to the behaviour of greater parts.
The transformation process of the textile templates into ceramic bodies can be performed
following several technological paths. The technologies are based on industrial
technologies for production of SiC foams: Schwarzwalder + Silicon Infiltration (SI),
Chemical Vapor Deposition (CVD) and Polymer Impregnation and Pyrolysis (PIP).
These techniques are adapted and optimized in order to obtain ceramic structures from
weaved and/or knitted polymer fibers and filaments. Depending on the fiber employed,
mostly because of its behavior during thermal treatments, one method will result more
efficient than the other. For instance, if chosen fibers will be made of a polymer which
disappears during pyrolysis, Schwarzwalder technique followed by Silicon Infiltration will
be most likely employed.
Based on the above described methods a technology is developed which is cost, time and
material efficient. The textiles are glued on ceramic parts. The tests for material and
technology development are performed on smaller and plain parts. First tests are carried
out only with textile based structures fixed on one surface of the plates. The final heat
exchangers will be tube shaped and must have structured surfaces on the inner and outer
surface of the cylinder barrel. All samples are intensively characterized using standard
technologies in materials science. The results help to improve the manufacturing process.
The samples are treated also in conditions similar to the final applications.
Final target is the development of ceramic heat exchangers with highly structured surfaces
for industrial gas burners. A long-term market study is performed to identify the market
potential and the economic limitations of the innovative burners. The innovative heat
exchangers will be integrated in existing industrial gas burners. A partial redesign of the
burners is necessary. Performance tests of the burners shall demonstrate the increased
efficiency of the heat exchangers in applications. The prototypes are operated in testing
facilities of industrial burner manufacturers. The temperatures of flue gas and combustion
air are measured at the input and output of the recuperator. Also the pressure drops in both

170
ducts is continuously monitored. The formation of harmful gas components is monitored
and compared to the emissions of existing burners and legislative limitations.
The successful market introduction of new technologies is always related to the associated
investment costs. For the case of the CEREXPRO target products (recuperative or
regenerative industrial burners and additionally further future innovative alternative
application schemes) a reference comparison to nowadays most advanced burner
technology with integrated waste heat recovery is adequate. Intensive market studies on
industrial heating technology and energy costs are performed. On the basis of recent
economical data the improvement in efficiency due to the application of the new heat
exchanger is significant. The calculated pay back times are below 2 years.
The calculations are performed for new installations as well as for refurbishment of
existing industrial furnaces. Especially the replacement of older burners is a great market
in Europe. Depending on the degree of preheating of combustion air the energy
consumption and therefore the CO2 emissions could be reduced drastically.
During the course of the project a great number of dissemination activities are performed.
The first activity is the launch of a project web site (http://www.cerexpro.org/).
The project was presented worldwide at 9 trade fairs. Oral presentations were given at 8
scientific conferences and meetings of engineer associations. Furthermore the European
Committee of industrial furnace and heating equipment associations CECOF has already
co-operation agreements and established co-operation with several non-European
associations in the process-heating sector like the US Industrial Heating Equipment
Association (IHEA) and the Japan Industrial Furnace Manufacturers Association
(JIFMA). A bilateral participation as a guest at the respective annual delegate assemblies
established, providing an overview of the last year activities to the foreign associations.
On these occasions CECOF regularly reported on the progress of CEREXPRO.
In March 2013 a dissemination work shop was held in Frankfurt/Main. Referents from all
beneficiaries explained their contribution to the project and the main results gained.
The invitation was sent to companies active in the branch of thermal processing, in
ceramic technology and heat exchanger production. The companies are mostly situated in
Europe. Also engineer associations and associations of companies representing different
industrial branches were invited. The participants of the workshop were – as planned and
expected – employees of manufacturers of thermo process plants and burners and their
components from Europe, with a focus on Germany. The participants were either head of
research and development departments or project managers. The future applications were

171
from particular interest for the participants. A main part of the workshop was dedicated to
an open discussion between the participants from the different customer industries.
The work shop was held in an open atmosphere. The response from the audience was very
positive and encouraging. The aim of the work shop was fully achieved.
This chapter is divided into three sections. The thermo fluid dynamic behavior of the
structured surface is studied in section ‘Design’. Section ‘Processing’ deals with the textile
structure development and production and its conversion process from textile-derived
materials into a thermal shock resistant ceramic with characterization of the material.
Finally, section ‘Prototype’ presents a prototype of Si-SiC heat exchanger for recuperative
gas burners with Si-SiC structured surface elements manufacturing and field testing.

6.2 Design

This part of the project was mainly developed by the University of Freiberg (TU BAF).
The design of the highly structured Si-SiC surface elements must meet the requirements of
robustness and adaptability to an existing heat exchanger. It was also taken into account
what textile and ceramic state of the art technology could provide. An arc (or loop) of
180° was selected as the basic element for the structured surface. The aim was to reach a
higher heat transfer coefficient by increasing the transferring surface and also by utilizing
the bigger dispersion and thermal boundary layer structures due to the resulting turbulent
flow characteristics. Figure 6.5 shows different possible arrangements of such loops on a
flat surface.

Figure 6.5: Different basic designs/arrangements of heat enhancing elements (radius R = 8

mm and diameter D = 1 mm) (source: TU BAF).

Numerical simulations using commercial CFD (computational fluid dynamics) packages

were performed for geometries in Fig. 6.5 to calculate the flow characteristics and analyze

172
the increase in heat transfer and pressure drop, that is, friction factor coefficient. The
numerical simulation of the surface structures is performed in the Reynolds number range
from 2000 to 20000. The commercial software FLUENT is used to solve the governing
equations. To reduce computation time, a planar arrangement was selected as reference
geometry, instead of a pipe, which is the typical application case. To simulate the fluid
flow and heat transfer process, a typical repeating cell was defined. This cell was used to
represent the entire structure by geometrical symmetry to further decrease the
computational efforts (Fig. 6.6).

Figure 6.6: Path from ceramic structure to one repeating single cell (source: TU BAF).

The governing equations were solved using finite volume methods assuming a forced
convection situation. There was also a need to take care about the minimum dimensions of
design elements to guarantee robust manufacturing and long-term stability. For each side
of the cubic cell a boundary condition is required. Boundary conditions of the cubic cell
were as follows: constant temperature at the bottom surface, which must be lower than the
inlet temperature of the air flow to generate a heat flux from the hot air to the cooled heat
exchanger surface (Fig. 6.7). It is important to work with a high number of elements near
the walls to implement special wall functions and reflect the no-slip condition on the wall
(Fig. 6.8). Based on this numerical experiment with a single cell the heat transfer
coefficient of the reference heat enhancing element was calculated.

Figure 6.7: Boundary conditions on one single cell used for the simulation (source:
TU BAF).

173
 Figure 6.8: Structured grid (source: TU BAF).

The reference surface for comparison and normalization purposes is the smooth
cylindrical surface at the middle plane between the two streams. A smooth tube at typical
recuperator operating conditions (1200°C flue gas temperature) shows a heat transfer
coefficient of approximately α = 56 W/(m2 K). Considering only the effect of increased
surface area, the heat transfer coefficient of the wavy tube (see Fig. 6.9) rises to
α = 90 W/ (m2 K) on the basis of the plane tube surface. This number is used as reference
(benchmark geometry) for evaluation of the new structured surfaces, while the operating
conditions are kept constant.
Fig. 6.10 shows the heat flux and the velocity field of a single cell for the layout with
3600 loops/m2. At the bottom and the top the velocity is less than in the free cross section
of the model because of the “no-slip” boundary condition. Such simulations for many
other cases were solved.

Figure 6.9: Current recuperator design on the left side and smooth cylindrical surface on
the right side (source: TU BAF).

174
Figure 6.10: Results of a simulation with 3600 loops/m2, wall heat flux (on the left) and
field of velocity (on the right) (source: TU BAF).

Pressure drop, Nusselt number and heat transfer coefficient were calculated at different
Reynolds numbers for every solution. Figure 6.11 shows the ratio of the expected
performance in comparison with the reference conventional recuperator burner at nominal
conditions average (Reynolds number 5200). With increased loop density a higher heat
transfer but also a higher pressure drop is found. No clear loop density emerges as more
suitable than any other for general applications and must be determined on a case by case
basis.
These results indicate that the targeted recuperator design will result in a significantly
higher heat recovery and slightly higher pressure losses, given the same burner size.
On the other hand, keeping the same heat recovery, the burner length and the pressure
losses can be significantly reduced.

Figure 6.11: Results of different layouts (source: TU BAF).

175
6.3 Processing

6.3.1 Textile structure assembly

This section of the project was entirely developed by IFTH. The aim of the textile
development is to utilize materials (as fibers, filaments), which are suitable for the
subsequent ceramic conversion and can be processed through textile processes.
Technologies like 3D weaving or knitting or converting technologies which allow to
assembly different fabrics (woven, knitted or non-woven fabrics) in a 3D way that match
with the objectives of the project, are employed. These technologies allow to mount three-
dimensional textiles structures, controlling the thickness, the packing and, overall, the
orientation of fibers in the different dimensions of the final product. These parameters are
key issues for a complete mastering of final innovative structures to be used in heat
exchangers. Indeed, the introduction of fibers coupled with their precise repartition and
orientation in the space allow to master and orient the heat fluxes aiming at an optimal
configuration of heat exchanges. Fibers orientation have also impact on the final
composites mechanical properties.
During the project’s development three kinds of materials have been selected:
monofilaments, multifilaments or staple fibers yarns and grids.

A) Monofilaments
A specific research of available monofilaments and providers has been realised.
The following materials have been tested:
- Teijin (JPN) PET (diameter 0.9 mm and 0.7 mm);
- Luxilon (BE) LDPE (diameter 1 mm);
- Sider Arc (IT) Crinlene PP (diameter 1 mm);
- PP Multifilament for weaving the base fabric.

B) Multifilaments and yarns

To obtain an additional source of carbon for ceramization process, the following polymers
were chosen: polypropylene charged with silicon carbide (PP+SiC), polyester (PET) and

176
oxidized polyacrylonitrile (PANox). PP+SiC and PET yarn are multifilaments and PANox
yarns are spun with staple fibers.
Trials are been made on the following materials:
- Basic PET (yarn for loops: 200 Tex (Tex is a unit of measure for the linear mass density
of fibers and is defined as the mass in grams per 1000 meters), yarn for base fabric: 70
Tex);
- PANox (yarn for loops: 400 Tex, yarn for base fabric: 80 Tex), this material which is
commonly available in roving form (around 320 000 filaments per yarn around 22 400
Tex), has to be cracked prior to spinning process;
- PP + SiC (UF-15 from H. C. Starck) multifilaments (yarn for loops: 150 Tex, yarn for
base fabric: 50 Tex).

C) Grids
Polymer grids, like polypropylene, polyethylene or polyester grids, produced through
extrusion process, thank to a cutting and assembly operations, were also tried.

Textile refers to any material made of interlacing fibers. Fabric refers to any material
made through weaving, knitting, spreading, crocheting, or bonding that may be used in
production of further goods in textile industry [4]. The processes analyzed to realize a
suitable structure are summarized in table 6.III.

Technology Example Description

Knitting Knitting is a way to make a fabric by
interlocking series of loops of one or more
yarns (Fig. 6.12). The various geometries of
the loops require a specific knitting loom;
it allows to control:
 loop height;
 space between loop;
 form of the base of the loop.
Two different types of knitting technologies
exist: weft and warp knitting.
Weft Knitting: a common type of knitting,
in which one continuous thread runs
crosswise in the fabric making all of the loops
in one course. Different trials showed the
interest to use in the building of the structure
an additional yarn with thermo soluble
properties (Fig. 6.13(b)).
Warp Knitting: a type of knitting in which the
yarns generally run lengthwise in the fabric.
The yarns are prepared as warps on beams
with one or more yarns for each needle.

177
Braiding Braiding is a method of intertwining three or
more strands to make a cord. The strands
form a regular diagonal pattern down the
length of the cord. There is no machine
presently available to braid 3D structures.
The development of such a machine would
require important resources which were not in
adequacy with the project.
Trimming These are narrow fabrics, often used for
trimming or binding, and which are
manufactured by plaiting several strands of
yarn. The fabric is formed by interfacing the
yarns diagonally to the production axis of the
material. To be used in heat exchanger,
the trimming should have to be subsequently
wounded and fixed on the exchanger ceramic
tube.
Grids The grids are made with a method of
extrusion and stretching. Polymer is extruded
through a circular spinneret which has holes
only at the rim; a translation of the spinneret
allows the formation of a ‘weft’ yarn.

Nonwoven An assembly of textile fibres held together by

mechanical interlocking in a random web or
mat, by fusing of the fibres (in the case of
thermoplastic fibres), or by bonding with a
cementing medium such as starch, glue,
casein, rubber, latex, or one of the cellulose
derivatives or synthetic resins. Initially, the
fibres may be oriented in one direction or may
be deposited in a random manner. This web
or sheet of fibres is bonded together by one of
the methods described above. Nonwoven
was immediately abandoned because the
proposed design could not be obtained with
this technology.
Weaving Weaving is made by interlacing two yarns of
similar materials so that they cross each other
at right angles to produce woven fabric.
The warp yarns run lengthwise in the fabric
and the filling threads (weft) run from side to
side. Trials were performed and samples
produced on rapier weaving looms (Fig. 6.14)
that were adapted to match the specification
of the design (Fig. 6.5). Two weave drawings
have been set corresponding to the different
densities of loops, and thread count for base
fabric has been determined according to the
diameter of loops. A rapier loom was adapted
to reproduce a sort of velvety texture
(Fig. 6.13(a)). Velvet process uses several
warp beams to balance the difference of
absorbed warp length between base fabrics
and loops and performs a weft thread or rails
to maintain the loops height.
Table 6.III: Explored textile technologies.

178
Figure 6.12: Terry warp knitting machine (left), flat weft machine (right) (source: IFTH).

Figure 6.13: Textile structures (geometry 01) obtained by (a) weaving and (b) knitting
(source: IFTH).

The best performing material, both for weaving and ceramization, was a multifilament
yarn made of polyethylene terephthalate (PET) fibers. Textile structure assembly had
many objectives to comply with:
 structure geometry should comply with the design constraints;
 being the textile structure a scaffold on which to deposit the ceramic layer, the
thermo-physical behavior of its material under pyrolysis should be tuned properly;
 elastic modulus as well as yarn thickness and sizing should be sufficient to keep
their geometry while replicating;
 another extremely important requirement is that, as discussed later, the base fabric
in contact with the Si-SiC tube walls should be as loose and light as possible to
minimize the formation of porosity onto the Si-SiC tube (Fig. 6.15). Porosity is
detrimental because it reduces heat transfer and acts as a crack initiator during
thermal cycling.

179
The base fabric was assembled using PET yarns made of 120 filaments with a dtex 668.
The loop structure was obtained with PET yarns made of 660 filaments with dtex 2004.
As fibers selection was also dictated by the ceramization requirements, further details will
be given in the next section. Indeed, there were opposite needs for the two technologies,
for example, carbon fibers are interesting because they can be converted to SiC during
LSI, but due to their rigidity, they are difficult to weave.

Figure 6.14: Weaving looms (source: IFTH).

Figure 6.15: 3D rendering, obtained by X-ray computed tomography, of highly packed

base fabric, impregnated and processed, showing high internal porosity (source: SUPSI).

6.3.2 Ceramisation process

The ceramisation process was optimized by SUPSI together with ERBICOL.

In combustion environments, silicon carbide ceramics are, for their outstanding thermal

180
and mechanical properties, among the best materials [5]. In fact, they are commonly
applied into high temperature furnaces because, if passive oxidation conditions are met
[5], they withstand operating conditions for long time [6].
Three processes were envisaged to fabricate highly structured Si-SiC surface elements
(Fig. 6.16):
I. Schwarzwalder + Liquid silicon infiltration (LSI);
II. Chemical vapor deposition (CVD);
III. Polymer impregnation and pyrolysis (PIP).

Figure 6.16: Outline of the ceramisation technique investigated (from top to down):
Schwarzwalder + LSI, CVD and PIP (source: SUPSI).

The Schwarzwalder process is the oldest production technique to make replica ceramic of
preforms. A starting preform polymer is deepened into a slurry (made of ceramic powders,
solvent and binders) and shook to remove excess material. The preform is kept into a
conditioned ambient for some time to evaporate slurry’s solvents. It is then heated up at
8001000°C in an inert atmosphere for binders’ removal by pyrolysis and infiltrated with
molten silicon, in a vacuum furnace at a temperature above the melting point of silicon
(1423°C). Binder choice is fundamental for component integrity during the first thermal
cycle because its coefficient of thermal expansion should match that of the parent foam to
avoid cracks formation in the ceramic body.

181
In CVD the pyrolysed preform (at the same conditions as before) is placed into a reactor
and coated with a SiC layer obtained from the decomposition at high temperatures
( 1200°C) and low pressures ( 50 mbar) of a ceramic organometallic precursor
(e.g. methyltrichlorosilane CH3SiCl3). For this kind of application, the purpose is to
increase preform stiffness and to enhance bulk material oxidation resistance. CVD is very
expensive but, because effective, it should represent a benchmark for the other techniques.
In Polymer Impregnation and Pyrolysis (PIP) the fibrous preform is soaked into a liquid
polymer. Preform is drained and the remaining polymer onto it, cross-linked.
Some polymers (Silanes) after thermal treatments yield amorphous SiC which should be
later nano-crystallized with a high temperature treatment in inert atmosphere at around
1600°C; adding ceramic powders, process efficiency can be further improved.
Several process tests were performed, at the end, because of its industrial maturity and
also because the commercial Si-SiC tube is made with a similar technology (proven to be
the best performer in a combustion chamber environment), LSI was adopted as
manufacturing procedure.
Few polymers are available on the market fulfilling the project requirements to make the
slurry with the appropriate characteristics. They have been identified as:
1) Ceraset PSZ 20 (KION a Clariant businnes, Switzerland);
2) Starfire SMP 10 (Starfire Systems, US).
To reduce bulk material shrinking and have an already stable phase, polymers were mixed
with -SiC powders from:
1) UF-15 HC Starck, Germany;
2) F 800 Washington Mills, US.
To evaluate the influence of the composition on the slurry characteristic, ceramic powders
(6 μm grain size) UF-15 (HC Starck) and coarse SiC powders (8.6 μm grain size) F800
microgrit (Washington Mills) were mixed with different commercially available
preceramic polymers Ceraset PSZ 20 and Starfire SMP 10 (Tab. 6.IV). Slurry
Components have been mixed by ball milling in a glass container for 15h using a Turbula
shaker (WAB, Switzerland).
Viscosity measurement showed slurries tixotropic behavior (Fig. 6.17(right)). It was also
observed that the polymers employed significantly influence their rheological
characteristics; slurries containing Kion Ceraset PSZ 20 show higher viscosity at the same
SiC charge and powder grit size in respect of those containing Starfire SMP 10
(Fig. 6.17(left)). In slurries containing the same preceramic polymer, the viscosity
182
increases at the increasing of the SiC content in weight, while it decreases with the
increasing of the grit size.

ID Slurry Preceramic Polymer % wt. SiC

PSZ_UF15_16% Kion Ceraset PSZ 20 16% UF15
PSZ_UF15_30% Kion Ceraset PSZ 20 30% UF15
PSZ_UF15_44% Kion Ceraset PSZ 20 44% UF15
PSZ_UF15_60% Kion Ceraset PSZ 20 60% UF15
SF_UF15_30% Starfire System 30% UF15
SF_UF15_44% Starfire System 44% UF15
SF_UF15_60% Starfire System 60% UF15
SF_UF15_80% Starfire System 80% UF15
PSZ_UF15_F800_30% Kion Ceraset PSZ 20 15% UF15 + 15% F800
PSZ_UF15_F800_44% Kion Ceraset PSZ 20 22% UF15 + 22% F800
PSZ_UF15_F800_60% Kion Ceraset PSZ 20 30% UF15 + 30% F800
SF_UF15_F800_44% Starfire System 22% UF15 + 22% F800
SF_UF15_F800_60% Starfire System 30% UF15 + 30% F800
SF_UF15_F800_80% Starfire System 40% UF15 + 40% F800
PSZ_F800_60% Kion Ceraset PSZ 20 60% F800
SF_F800_60% Starfire System 60% F800
Table 6.IV: Slurries composition.

Figure 6.17: Left: slurry viscosity vs. wt.% SiC highlighting the influence of the
preceramic polymer; right: viscosity measurement of slurries with different wt.% of SiC
showing the tixotropic behaviour of slurries as well as the influence of the wt.% of SiC
powder (source: SUPSI).

To evaluate the ability of different slurries to form a uniform green body, samples (wire
polyester porous bodies (Colbond, Netherlands)) were dipped in the slurries; the excess
slurry removed by centrifugation (speed 1000 s-1 for 30 sec.). Green bodies were then
fired using a high temperature oven (KEOS, Italy) under Argon flux (300 nl/h). Several
heat ramps were tested, the adopted one is reported in figure 6.18.

183
 Figure 6.18: Selected heat ramp.

The coating thickness was evaluated by optical microscopy of the strut cross sections
using a Leica DMLM Optical microscope. Measurements on migrographs showed the
dependance of the coating thickness on the slurry viscosity; higher thicknesses were
observed in sample impregnated with more viscous slurries.
During the evaluation of the fired samples the following features were taken in occount:
presence of cracks and pores in the coating, integrity after pyrolysis and the ability of
preform to withstand small mechanical loads (i.e. the specimen can be handled). Table 6.V
summarizes the observation results (red color or green color are used for a bad or good
behavior, respectively).

Slurry Cracks Porosity Survive Small loads

pyrolysis
PSZ_UF15_16% NO NO NO NO
PSZ_UF15_30% NO NO NO NO
PSZ_UF15_44% YES YES YES NO
PSZ_UF15_60% YES YES YES YES
SF_UF15_30% YES YES NO NO
SF_UF15_44% YES NO YES NO
SF_UF15_60% NO YES YES NO
SF_UF15_80% YES YES YES NO
PSZ_UF15_800_30% YES YES YES NO
PSZ_UF15_800_44% YES YES YES NO
PSZ_UF15_800_60% YES YES YES YES
SF_UF15_800_44% YES YES YES NO
SF_UF15_800_60% YES YES YES NO
SF_UF15_800_80% YES YES YES NO
PSZ_F800_60% NO NO YES NO
SF_F800_60% YES YES YES NO
Table 6.V: Coating observations summary.

184
According to table 6.V only two specimens survive the heating threatment and can support
small loads; other specimens resulted to be too fragile and thus usless for our purpose.
Figure 6.19 shows cracks formation, it should be a consequence of the preceramic
polymer contraction during the transformation from polymer to silicon carbide into the
thinner part of the coating. Thicker slurries experienced uneven coating.

Figure 6.19: Micrography of a PSZ_UF15_44% strut after pyrolysis, magnitude 50x,

showing defects (micro-cracks (1) and pores (2)) and uneven coating (source: SUPSI).

According to the reported results, two slurry compositions have the required properties
(those marked on the yellow background in the table 6.V): the slurry labeled
PSZ_UF15_800_60% (see Fig. 6.20) was chosen.

Figure 6.20: KION Ceraset Polysilazane 20.

The textile structure was therefore dipped into the slurry and hanged to drain the excess
liquid. As the solvent was partially evaporated and the structure still tacky, it was placed
on the Si-SiC substrate as per Fig. 6.21 and further dried and cured in a oven at 130°C for
15 min in Ar atmosphere. Dicumyl peroxide was added in 0.01 wt.% to allow KION cross
linking. The plate was then placed into a furnace and slowly pyrolyzed up to 1000°C (see
Fig. 6.18), under inert atmosphere. The textile structure experienced a minor shrinking
because of the mass loss during pyroloysis, and some cracks could be observed. Cracks

185
minimization was worth a long optimization work in which textile architecture of the base
fabric and amount of slurry employed were carefully adjusted. After pyroloysis, the object
was placed in contact with silicon flakes and LSI was performed at 1500°C under a 10-2
(mbar) residual pressure. The furnace was quickly brought to the process temperature,
held at that temperature for 1h, and left cooling naturally. A final flat sample is shown in
Fig. 6.21.

Figure 6.21: Steps in the production of prototype heat exchanger structures (source:
ERBICOL).

A great effort was devoted to the selection of proper fibers, further called fugitive
(Fig. 6.22(left)) or permanent (Fig. 6.22(right)), meaning that in one case, they would
produce, respectively, a hollow or full space within the loop. Examples of fugitive fibers
are polypropylene (PP) and PET. On the other end, permanent fibers are those containing
already a ceramic phase (e.g., C, SiC) (Fig. 6.23).

Figure 6.22: 3D rendering, obtained by X-ray computed tomography, of Si-SiC loops.

Left: hollow red region inside the loop formerly occupied by the fugitive yarn; right: full,
that is, permanent yarn (source: SUPSI).
186
Figure 6.23: Examples (from left to right) of fugitive fibers: Panox and PET, and non
fugitive fibers: PP + 25% SiC (source: IFTH).

Fugitive, PET fibers were selected because of their:

1. low cost;
2. easiness to be weaved;
3. similar behavior to the slurry (weight loss, shrinking) during pyroloysis resulting
in less cracks formation.
Their main drawback is due to their low rigidity which could not guaranty to keep the loop
shape during processing.

6.3.3 Ceramic characterisation

The ceramic characterisation was entirely worked out by POLITO. To characterize the
skeleton material of the loops, another manufacturing technique was used to obtain
samples with suitable dimensions. Green bulk plates of the same constituent materials
were produced by hot pressing, pyrolyzed, and silicon infiltrated. This solution allowed to
produce monolithic pieces complying with the geometries necessary for characterization.
Microscopic characterization confirmed that the base material of the loops is very similar
to that of the Si-SiC ERBISIC foams (ERBICOL SA, Balerna CH). Indeed, this material
applied by replica on reticulated foams has proven to withstand long lasting severe
oxidative environments typical of the porous burners technology [7].
Properties were utilized to refine system design. They also confirmed to be quite similar to
other commercial Si-SiC products.
Samples were characterized by several techniques.

187
Samples were cut (TR 100 S, Remet, Bologna, Italy; Isomet 4000, Buehler, Illinois, USA)
and polished (Mecatech 234, Presi, Oxford, UK). Vickers microhardness measurements
were carried out with a microdurometer (Leitz Wetzlar D). Microstructure evaluations on
both oxidized and non-oxidized samples were performed with a Field Emission Scanning
Electron Microscope (FESEM Zeiss Supra 25 D) and using a scanning electron
microscopy (FEI, QUANTA INSPECT 200 LV) equipped with EDS (EDAX, detector
SUTW with SW GENESIS). XRD data were collected with a Philips PW3830 X-ray
generator system. The sample density ( ) was measured by water picnometry using
Archimede's balance for the the hydrostatic density, while geometric density was found
using a palmer (Mitutoyo, Japan) and an analytical balance (Orma model bc). The elastic
modulus was measured using the impulse excitation technique with a Mk5 apparatus
(Grindo Sonic Leuven B), in flexure mode. Three point bending tests were performed with
a Sintech 10/D universal testing machine (MTS, Eden Prairie, MN) using a span of
40 mm. Crosshead speed was 0.5 mm/min. Thermal conductivity  was calculated
according to the relationship:  =  cp . Thermal diffusivity () together with specific
heat capacity (cp) was measured with a Flashline 4010 (Anter, New Castle, DE).
Thermogravimetry analysis was carried out using TGA/SDTA 851eModule, (Mettler
Toldedo Inc., Im Langacher Greifensee CH) from RT to 1600°C using a heating rate of
10°C min-1. The atmospheric conditions were dry air flowing at 50 ml min-1. To avoid the
influence of the sample surface area cubes of 3 x 3 x 3 mm, cut with diamond tools, were
used for TGA analysis. The linear coefficient of thermal expansion was measured using
Setaram Instrumentation's Setsys Evolution TMA with a fixed rate of 5°C/min and from
RT to 1500°C. In this case the purge gas used during the measurement was argon.
To determine the amount of un-reacted free silicon present in the loops, seven standards
have been prepared, thus covering the range of estimated composition. The standards were
made by mixing powders of silicon (≥ 99.7% of purity, average particles size < 150 µm;
Schuchardt, Műnchen, D) and powder of silicon carbide (grade UF-10, Stark Ag, Goslar,
D); these standards contained the following percentages by weight of silicon: 3%, 5%,
10%, 15%, 25%, 45% and 65%. To estimate the percentage of free silicon present in the
loops by using the standards a calibration curve on the basis of the ratio between the
integrated areas of the peaks of SiC (110) and Si (111) was constructed. Calibration
performed by using the areas in principle is more accurate than that exploiting the
intensities (peaks’ heights) or the width at half maximum (FWHM). The intensities are in

188
fact conditioned by the percentage of crystallinity of the material, while the width is
affected by the size of the crystallites. Using the areas, the two above mentioned effects
are minimized.
The peaks considered were:
- for the silicon the peak at 28.5° (corresponding to the plane (111));
- for the silicon carbide the peak at 60°(plane (110)).
This last peak was chosen, even if it is not the greatest in intensity, because in this case
there is not any overlapping with other peaks, and then a better integration can be reached.

A. Bulk: microstructure and thermo-mechanical properties

Si-SiC was manufactured by cold pressing a mixture of SiC powder, as provided by

ERBICOL, and phenolic resin, as binder. The weight composition of green plates (Fig.
6.24) was: 78 % of SiC powders and 22 % of phenolic resin. Produced plates were then
fired and infiltrated with molten silicon (Fig. 6.25). The following are the steps of
synthesis:

1. SiC powder and phenolic resin were mixed in a glass container with glass milling
spheres for 12 h;
2. drying in a vacuum oven, at 60°C for 12 h;
3. powder pressing, p = 200 Bar, at 60°C for 60 s;
4. green body pyrolysis under inert atmosphere at 1000°C;
5. molten Si infiltration at 1500°C.

The samples were produced in the form of buttons and rods, depending on the tests to
which they were subjected.

Figure 6.24: Green plates (source: ERBICOL).

189
 Figure 6.25: Final bulk specimens: rods and buttons (source: ERBICOL).

From micrographs (Fig. 6.26) the bulk microstructure appears to have two kinds of
porosity. The first family is in the range of tens of microns, the second family of pores is
in the range of microns. As can be seen from table 6.VI the two densities, geometric and
hydrostatic, are comparable if the standard deviation is taken into consideration. From this
can be asserted that the SiC samples do not have open porosity. The density value is lower
than that reported in literature [8], that is 3.09 g/cm3. This is due to the presence of
un-reacted free silicon, which has a density of 2.33 g/cm3, and by the presence of closed
porosity that decreases the average density. At higher magnification the microstructure
results not homogeneous because there are areas of different gray tone. On the points
marked with a blue cross in figure 6.26 an EDS analysis has been made : the light gray
zone is poor in carbon, nearly all Si (89 at.% of silicon), while the dark gray zone is richer
in carbon, that is SiC (50 at.% of silicon and 50 at.% of carbon). No other phases, like
residual carbon, were detected during the EDS analysis. The XRD pattern on bulk
powders (Fig. 6.27) confirms that the only phases present are silicon and silicon carbide.
The crystallinity is high and the presence of amorphous phases was not detected.

Figure 6.26: Bulk microstructure at lower magnification (left), taken with secondary
electrons, and higher magnification (right), taken with back-scattered electrons (source:
POLITO).

190
 Figure 6.27: Pattern XRD of bulk powders (source: POLITO).

Property Average value RSD %

3
Geometric density 2.62 g/cm 3%
3
Hydrostatic density 2.68 g/cm 2%
Vickers microhardness 15.5 GPa (1585 HV) 17%
(Load 300 g) Silicon: 10.9 GPa
SiC: 24.6 GPa
Vickers microhardness 16.0 GPa (1631 HV) 16%
(Load 500 g)
Young's modulus: E 266 GPa 3%
MOR (3-point flexural test): σf 196 MPa 7%

Table 6.VI: Material mechanical properties measured on the Si-SiC bulk material (source:
POLITO).

As far as the Vickers micro-hardness is concerned, the measurements show (Tab. 6.VI),
with both loads, a large dispersion around the average value, that is the samples do not
have a homogeneous hardness. From literature [8] for pure Si and SiC the micro-
hardnesses values are 9 GPa and 25 GPa, respectively. Our data are in good agreement
with the tabulated ones. The macro-indentation test cannot be made because of the
sample's fragility (fracture and chipping observed).
The values of Young's modulus in the rods show a narrow distribution around the average
value (RSD 3 %). The measured value is 266 GPa, significantly lower than that tabulated
[8] that is about 360 GPa. Also in this case, the explanation is given by the presence of
free silicon, which has an elastic modulus of about 160 GPa.
From the three-point flexural test (Tab. 6.VI), the average flexural stress (σf) of the rods is

about 200 MPa. The dispersion in these measurements is quite high as can be seen from
RSD index. This is common in these kinds of measurements for ceramic materials because
of their stiffness. Another important factor that strongly influences this kind of

191
measurement is the presence of internal microcracks that promote fracturing. From [8] the
flexural resistance for 4-point loading test is given equal to 280 MPa.
The figure 6.28 shows the thermomechanical analysis in which the displacement has a
linear trend to almost 1400°C. From literature [8] it’s known that SiC has a very low
coefficient of thermal expansion, CTE is about 4.9*10−6°C-1 in the range from RT to
1000°C, and experiences no phase transitions that would cause discontinuities in thermal
expansion. The CTE (Coefficient of Thermal Expansion) values, after the stabilization
reached around 300°C, display a constant trend. Only near the 1400°C it is possible to
notice a little variation in CTE values due to the melting point of silicon (1410°C).

Figure 6.28: TMA measurements: linear coefficient of thermal expansion (left) and
displacement as functions of temperature (right) (source: POLITO).

Table 6.VII shows the results obtained for conductivity, diffusivity and specific heat as
function of temperature. Graphite was the standard reference used to calculate the
conductivity and the specific heat. From figures 6.29 and 6.30 can be noted how the
conductivity and the diffusivity have a hyperbolic trend with the rise in temperature, while
in Fig. 6.31 a linear monotone trend in the specific heat can be observed.

Temperature Conductivity Diffusivity Specific heat

[°C] [W/(m*K)] [cm2/s] [J/(kg*K)]
193 85.08 0.2867 944.4
501 59.66 0.1639 1158.8
1004 41.68 0.0946 1402.9
1504 37.49 0.0668 1786.7

Table 6.VII: Thermal properties at different temperatures (source: POLITO).

192
Figure 6.29: Conductivity trend as a function of temperature (source: POLITO).

Figure 6.30: Diffusivity trend as a function of temperature (source: POLITO).

Figure 6.31: Specific heat trend as a function of temperature (source: POLITO).

193
The values reported in literature [8] on conductivity and on specific heat are for the first
one in the range from 160 W/(m*K), at 100°C, to 24 W/(m*K), at 1200°C, and from
600 J/(kg*K), at room temperature (RT), to 1200 J/(kg*K), at 1300°C, for the second one.
These values are in fairly good agreement with the data found.

B. Loops: microstructure

From SEM image (Fig. 6.32(left)) is clear that the polymer template used in the loop is
fugitive textile kind (like PET in our case) so that a hollow internal structure can be
observed. As in the case of bulk, the microstructure of the loop shows only two phases:
dark gray zones and light gray zone. By EDS analysis (Fig. 6.32(right)) these phases were
found to be silicon carbide and silicon, respectively. In addition to the hollow structure can
be observed how the porosity is very high; loop cross-section is not round due to the slurry
draining before solidification. The XRD analysis (Fig. 6.33) gives another confirmation on
the phases present: either silicon and SiC are highly crystalline and no amorphous phases
are detected. Both silicon carbide polymorphs are detected: α-SiC, which is the most
commonly encountered polymorph and has a hexagonal crystal structure, and β-SiC that
has a zinc blende crystal structure.

Figure 6.32: SEM image and EDS analysis on the loop cross-section (source: POLITO).

194
Figure 6.33: Pattern XRD of loops powders: for silicon phase the reflecting planes are
indexed, for SiC phase the various polytypes for each peak are labeled (source: POLITO).

Table 6.VIII shows all the data that have been used to construct the calibration curve and
the data related to loops. In figure 6.34 the experimental points of the standards are fitted
with an exponential decay fit of the first order, while the value of the loops (ratio between
peak areas) is superimposed to the calibration curve. By using this curve, it is possible to
see how the percentage of free silicon present in the loops is about 34 wt.%. Considering
equal the theoretical density of bulk and loops, and using this estimate percentage of free
silicon to calculate the loops density, a value of 2.83 g/cm3 is calculated, slightly higher
than the density value obtained for the hydrostatic bulk. This fact is perfectly in agreement
with the consideration that the theoretical density (as is the one that is obtained from the
powders) is always greater than the hydrostatic density because the first one eliminates the
internal porosity to the material. In conclusion, must be said that quantitative analyses
done with X-rays have a sensitivity of 5%, so this estimate on the percentage of free
silicon present in the loops should be read bearing this fact in mind.

Percentage of Si Standards: Standards: Standards:

free Peak Area of Si Peak Area of SiC Ratio of Peak
(wt.%) (2θ ~ 28.5°) (2θ ~ 60°) Areas SiC/Si
65% 1547.88 222.93 0.14402
45% 1032.19 364.79 0.35341
25% 470.02 477.49 1.0159
15% 169.49 549.69 3.2432
10% 98.30 550.85 5.604
5% 28.32 310.95 10.98
3% 27.13 309.25 11.40
Loops Powders 1326.65 657.49 0.496
Table 6.VIII: Values of peak areas and of ratios between areas for standards and loops
(source: POLITO).

195
Figure 6.34: Calibration curve resulting from the experimental points of the standards with
the ratio of peak areas (SiC/Si) for the loops (source: POLITO).

C. Oxidation behavior

With the aim to study the long-term oxidation behavior, specimens were oxidized at
1600°C in a resistance heated furnace (HT64/17, Nabertherm GmbH, Lilienthal,
Germany) in static air at atmospheric pressure. To avoid contact with the furnace, the
samples were placed in an alumina boat kept suspended with fibers of silicon carbide.
SiC fibers, not reacting with the samples, have allowed an accurate estimate of the weight
gain. Each set of samples was brought to 1600°C with a heating rate of 10°C·min-1 and
held at that temperature for 4h, 8h, 24h and 48 hours.
The thermal treatment on the loops was of 1500°C in air (flow of 200 SCCM) in an
alumina furnace for 2h, 4h and 6hours. The thermal shock test was performed with a
prototype system for cyclic oxidation test and heat shock called “Burner Rig” in which an
aggressive environment designed to simulate the working conditions can be created.
The operating parameters of the test were: ambient pressure, methane as combustion gas
and quenching using room temperature air stream at 5 bar pressure. The machinery used to
do the non standard compression test on loops was a Sintech 10/D; the crosshead speed
of 0.5 mm/min.
The thermogravimetric analysis on specimens coming from used and new burners is
shown in figure 6.35. The sample from the new burner shows a final increase in weight of
about 1.5 % and a mild oxidation that begins at about 1100 °C with the oxidation of SiC
(reaction 6.1)

196
 SiC (cr) + 3/2 O2 (g) → SiO2 (s) + CO (g) (6.1)

Accompanied by the oxidation of pure Si (reaction 6.2)

Si (cr) + ½ O2 (g) → SiO (g) (6.2)

Which is passive as long as there is the Si/SiO2 equilibrium [9] in the form (6.3)

Si (cr) + SiO2 (cr)  2SiO (g) (6.3)

The weight of the sample from used burner remains practically constant over the entire
range of temperatures, from RT to 1400°C. Above 1400°C a slight increment in weight
due to starting oxidation process of the free silicon can be observed (reaction 6.4).

Si (cr) + O2 (g) → SiO2 (s) (6.4)

At the end of the oxidation process the increase in mass is very low: about 0.2 % by
weight.

Figure 6.35: TGA of new and used bulk burner specimen (source: POLITO).

This different behavior can be explained by a passivation occurred on the used burner
specimen due to his working conditions.
The standard temperature range of working burner is between 9001200°C in oxidation
atmosphere. To simulate tens of thousands of work hours, it was decided to do a process
of accelerated oxidation going to higher temperatures and using shorter timeframes.
Figure 6.36 gives the weight gain behavior of the bulk material in static air at atmospheric

197
pressure at 1600°C. After 8 hours of thermal treatment the maximum weight gain is
reached, which results to be of approximately 3 wt.%. After this timing, a stabilization in
weight due to the passivation of the external silica layer can be observed. The slight loss in
weight after 24 and 48 hours of oxidation can be explained by active oxidation of SiC
(reaction 6.5),

SiC (cr) + O2 (g) → SiO (g) + CO (g) (6.5)

and by the evaporation of silica from scale (reaction 6.6).

SiC (cr) + 2 SiO2 (s) → 3 SiO (g) + CO (g) (6.6)

Figure 6.36: Long-term oxidation at 1600°C: percentage weight gain (left) and weight
gain normalized on area (right) (source: POLITO).

The previous explanations about the oxidation behavior can be observed on the
morphology in Fig. 6.37. The silica scale layer that passivate the sample is clearly visible
on the edge due to its different feature and varies from 30 µm to 50 µm of thickness.
Besides the smooth aspect, to definitively confirm the composition of the outer layer local
EDS analysis was made. From the micrograph can also be seen how the flake has a certain
porosity, due to the formation of CO and SiO gases, and as the continuity on the surface is
interrupted. This is ascribed to the different steps of sample preparation as embedding,
cutting and polishing. At this stage of oxidation, 4 hours, the scale is still growing up as
can be inferred from the graphs in figure 6.36.

198
Figure 6.37: SEM images, taken with secondary electrons, of the specimen oxidized for 4h
at 1600°C: sample edge at low magnification (left) and zoom on the silica scale (right)
(source: POLITO).

To test the stiffness of the loops as produced and after working use, a technological proof
was made performing a no-standard compression test on the loops with also an accelerated
oxidation ageing that simulates the working conditions of a burner. Furthermore, it was
decided to do this type of thermal treatment also to confront it with data in the literature.
In figure 6.38 there are different photos of the loops after several oxidation time: as
produced, after 2h, after 4h and after 6h of oxidation. The total number of loops tested was
forty: ten for each type of heat treatment time. The different aspect of the surface, with
increasing time of oxidation, is due to the fused silicon (m.p. 1410°C) that filled the loop’s
pores and to the formation of external silica passivation layer.
From energy-dispersive X-ray spectroscopy (EDS) analysis (Fig. 6.39) made on the
surface, there is the confirmation about the composition of the external loop’s layer, in
fact, it can be observed how the oxygen peak is much higher at the end of the thermal
treatment. After the oxidation, it is possible to see also peaks of aluminum, magnesium
and calcium probably due to the alumina furnace.

Figure 6.38: Loops after different oxidation time at 1500°C in air (source: POLITO).

199
Figure 6.39: EDS on the surface of the loop as produced (left) and after 6 hours of
oxidation (right) (source: POLITO).

After the no-standard compression test on the loops, three different kinds of fracture are
observed (Fig. 6.40). The labels given to different types of fracture were: “on the base”,
“on the tip” and “on tip/base”. In figure 6.42 the maximum load applied versus the
oxidation time is plotted. The sections of the loops were not homogeneous and vary from
loop to loop, so that it was impossible to plot the pressure versus time and not all the loops
were straight but many were crooked, which implies that the load is not evenly distributed.
The graph shows a very large dispersion of the data, but a slight increase in the average
maximum force applied with the increasing of oxidation is observed. This phenomenon is
probably due to the decrease in porosity with the increasing time of oxidation owing to the
formation of a silica layer that acts as a supporting element.
The cyclic oxidation tests were performed between 400°C and 1000°C. The profiles of
heating-cooling are shown in figure 6.41. The warming-up rate results in the order of
80100°C/min, while the cooling rate is in the range of 400600°C/min. Non-standard
compression tests on the loops were performed after the thermal shock treatment
(Fig. 6.42). Several loops have been tested on the following oxidation cycles: 10, 20, 30,
50 and 100 cycles. Even after one hundred cycles, the mechanical performances of the
loops are unchanged and result comparable to the oxidation time data. Must be stressed
how the dispersion of the data is an intrinsic feature of the tested samples. Concluding it
can be stated as the thermal shock test and the accelerated oxidative aging process do not
affect significantly the mechanical strength of the loops.

200
Figure 6.40: The three different types of fracture obtained after the compression test
(source: POLITO).

Figure 6.41: Theoretical oxidation cycles (left) and real oxidation cycles (right) (source:
POLITO).

Figure 6.42: Graph of the maximum applied load on the loops versus oxidation time data
and thermal shock data, respectively (source: POLITO).

201
6.4 Prototype

6.4.1 Manufacturing

The manufacturing method developed and optimized in section “Processing” was adopted
to fabricate two complete heat exchanger units with the two reference geometries: that is
geometry 01 and geometry 02 of figure 6.5. Textiles and Si-SiC tubes (Schunk
Ingenieurkeramik Willich-Műnchheide, D) were combined in the ceramization process,
and silicon infiltrated to obtain the final article. Textile structure placing inside and
outside the tube walls was optimized, given the small radiuses of curvature of the tube
(Fig. 6.43).
The entire procedure was optimized by ERBICOL on cylindrical Si-SiC tubes produced
by Schunk (Fig. 6.44). It was then applied on a real recuperator.

Figure 6.43: Fabrication of prototype heat exchanger structures (source: ERBICOL).

The first tubes were cut in shorter units. Textiles were impregnated in the ceramic slurry
and cut to dimension. These patches were placed inside and outside the tube sections.
Several trials were performed to obtain tube sections complying with the design
requirements. The pieces were pyrolized and infiltrated with molten silicon according to
the ceramization process defined in subsection “Ceramisation process”. Once validated the

202
entire procedure, two complete recuperators for an 80 kW burner were produced. Results
for the two different geometries are shown in Fig. 6.45.

Figure 6.44: History prototypes of manufacturing process (source: ERBICOL).

Figure 6.45: Two Si-SiC heat exchangers with different loops distribution:
(a) geometry 01, 3600 Loops/m2, Loop: R = 8 mm / D = 1 mm and (b) geometry 02,
7200 Loops/m2, Loop: R = 8 mm / D = 1 mm (source: ERBICOL).

A new recuperator with new heat exchanger geometry had to be integrated in a reference
burner. This burner prototype should be based on an existing burner of NOXMAT K-
RHGB Series (Figs. 6.46, 6.47).

203
The main requirements were to define the burner size and to analyze the burner
components. It had to be figured out which can be reused, modified slightly or developed
newly.
The product range of NOXMAT was analyzed by using characteristic numbers of burner
dimensions. The main focus lays on the gap-width between the burner tube and the
recuperator and accordingly between the recuperator and the jacket tube or insulation of the
waste gas chamber.

Figure 6.46: Burners K-RHGB 160 and K-RHGB 80 (source: NOXMAT).

Figure 6.47: Free gap-width of K-RHGB 160 and K-RHGB 80 (source: NOXMAT).

204
Figure 6.48 shows the section view of new burner with a loop recuperator. In comparison
to the burners with corrugated recuperator the section of flow is higher and as a result the
flow velocity is lesser.
The burner with the new loop recuperator is a combination of a K80 and a K160 burner of
the ceramic burner series of NOXMAT (Figs. 6.49, 6.50). Because of this combination it
is necessary to modify existing and to design new burner components. Table 6.IX shows
the existing parts for the new burner and table 6.X shows the new and redesigned
components of the new burner.

Figure 6.48: Concept of the CEREXPRO burner with a loop recuperator.

Existing parts of a K-RHGB 80 burner Existing parts of a K-RHGB 160 burner

Burner tube Recuperator flange
Swirl plate Insulating ring
Seal ring burner tube Seal ring recuperator
Gas part Air part
Gas supply Waste gas chamber
Combustion air supply Retainer device (recuperator)
Cooling air supply Gasket of the gas chamber
Gaskets Gasket of the recuperator flange
Table 6.IX: Existing parts of a K-RHGB 80 and K-RHGB 160 burner (source:
NOXMAT).

205
 New components Redesigned components
Recuperator with loops Electrode
Insulation of the waste gas chamber Gas lance
Additional adapter burner tube Retainer device (burner tube)
Additional adapter burner tube fixation
Distance tube for the burner tube
Additional adapter gas part
Table 6.X: New and redesigned components of the new burner (source: NOXMAT).

Figure 6.49: Exploded view of the reference burner with the new loop recuperator (source:
NOXMAT).

Figure 6.50: Section view of the reference burner with the new loop recuperator (source:
NOXMAT).

206
6.4.2 Field tests

The tests were carried out by NOXMAT.

The recuperators tested in this work at different power levels were as follows:
 plain recuperator without loops and with waste gas guiding tube (Fig. 6.51(a));
 structured recuperator with loops (geometry 01) and with waste gas guiding tube
(Fig. 6.51(b)).
Both recuperators were assembled in the same test burner and operated in indirect heating
using a jacket tube ø 200 mm. To characterize the burner, many different measuring points
were assembled at the burner and the jacket tube. Figure 6.52 shows the points where
several parameters were acquired.

Figure 6.51: Burner with different recuperators: (a) plain recuperator and (b) loop
recuperator with geometry 01, 3600 Loops/m2, Loop: R = 8 mm / D = 1 mm (source:
NOXMAT).

Figure 6.52: Overview of all measuring points (red for waste gas, yellow for natural gas,
blue for combustion air) (source: NOXMAT).

207
The following criterions had to be used for evaluating the new loop structure on the
recuperator:
 Waste gas outlet temperature;
 Air preheating;
 Firing efficiency;
 NOx-emissions in waste gas;
 Average temperatures at inlets and outlets as well as in the jacket and flame tube;
 Pressure drops in combustion air and waste gas;
 Temperature profiles throughout the recuperator;
 Condition of the surface before and after the tests.
The temperatures and pressures of both media had to be measured at the different
measuring points on their way through the system. The combustion air was determined
with the inlet and outlet conditions and the temperature trend above the recuperator.
The inlet conditions of the fuel (natural gas) measured. Both burners were tested with this
concept. All data were recorded over the whole test duration. All pressures and waste gas
measurements were documented by hand in a separate protocol.
The following edge conditions were applied to all tests:
 Air ratio λ = 1.15;
 All tests were heating tests. This means that the test furnace was heated by the
burner and the radiant tube from 100200°C (residual heat from the previous test)
to 10001150°C continuously.
The firing efficiency was calculated by the use of waste gas mixture, the inlet and outlet
temperatures of combustion air and waste gas and the formula of SIEGERT (formula 6.7)

(6.7)

With:
 ηf: firing efficiency in %;
 tAA: waste gas outlet temperature in °C;
 tLE … combustion air inlet temperature in °C;
 A: constant, for natural gas 0.66;
 O2,A: oxygen content in dry waste gas in %;
 B: constant, for natural gas 0.009.
Figure 6.53 and table 6.XI show the firing efficiency of the loop recuperator compared to
the plain recuperator. Both burners were assembled with a waste gas guiding tube.

208
Figure 6.53: Firing efficiency of a recuperative gas burner with different capacities and
recuperators (source: NOXMAT).

Waste gas inlet Plain recuperator Loop recuperator

temperature
800°C 75% 79%
900°C 71% 76%
1000°C 67% 73%
1100°C 64% 70%
1200°C 60% 67%
1300°C 56% 64%
Table 6.XI: Comparison of the firing efficiency of plain and loop recuperator with
P = 80 kW (source: NOXMAT).

A considerable increase of firing efficiency could be achieved by using the loop

recuperator. This increase depends on the waste gas inlet temperature: the higher waste
gas inlet temperature, the higher is the raise of the firing efficiency. In the normal
operating range of a ceramic burner (8001300°C) the increase in efficiency is about
4% to 8%.
The increase in efficiency causes a decrease of the waste gas outlet temperature and a
reduction of the heat loss. The waste gas outlet temperature was measured in the waste gas
fitting of the burner. Figure 6.54 and table 6.XII show a comparison of the waste gas
outlet temperatures of plain and loop recuperator assembled with a waste gas guiding tube.

209
Figure 6.54: Waste gas outlet temperature of a recuperative gas burner with different
capacities and recuperators (source: NOXMAT).

Waste gas outlet Plain recuperator Loop recuperator

temperature
800°C 590°C 460°C
900°C 675°C 550°C
1000°C 760°C 625°C
1100°C 850°C 700°C
1200°C 930°C 780°C
1300°C 1020°C 850°C
Table 6.XII: Comparison of the waste gas outlet temperature of plain and loop recuperator
with P = 80 kW (source: NOXMAT).

The decrease of the waste gas outlet temperature depends also on the waste gas inlet
temperature. In a temperature range from 800°C to 1300°C the decrease is between 130 K
and 170 K. This correlates with the rise of the firing efficiency.
Because of the higher firing efficiency a higher air preheating was expected. The
preheating was measured close to the combustion chamber. The following figure 6.55 and
table 6.XIII document the rise of the air preheating temperature by using a loop
recuperator assembled with waste gas guiding tube. The air preheating could be increased
by 135 K to 150 K. The differences between higher and lower waste gas inlet temperatures are
not as clear as they are in other characteristics like firing efficiency.

210
Figure 6.55: Air preheating temperature of a recuperative gas burner with different
capacities and recuperators (source: NOXMAT).

Waste gas outlet Plain recuperator Loop recuperator

temperature
800°C 125°C 260°C
900°C 200°C 330°C
1000°C 275°C 410°C
1100°C 350°C 490°C
1200°C 430°C 560°C
1300°C 500°C 650°C
Table 6.XIII: Comparison of air preheating of plain and loop recuperator with P = 80 kW
(source: NOXMAT).

The higher air preheating causes higher polluting emissions. The emissions were
measured at the same measuring point as the waste gas outlet temperature, the waste gas
fitting. Figure 6.56 and table 6.XIV show a comparison of NOx-emissions of the plain and
the loop recuperator, both assembled with a waste gas guiding tube. The increase of the
NOx-emissions is almost constant on the whole temperature range and is about 50 mg/m3.
At lower capacities the rise of the emissions is higher (about 10110 mg/m3) and depends
on the waste gas inlet temperature.

211
Figure 6.56: NOx-emissions of a recuperative gas burner with different capacities and
recuperators (source: NOXMAT).

Waste gas outlet Plain recuperator Loop recuperator

temperature
3 3
800°C 110 mg/m 170 mg/m
3 3
900°C 150 mg/m 205 mg/m
3 3
1000°C 200 mg/m 250 mg/m
3 3
1100°C 250 mg/m 300 mg/m
3 3
1200°C 310 mg/m 360 mg/m
3 3
1300°C 375 mg/m 425 mg/m
Table 6.XIV: Comparison of NOx-emissions of plain and loop recuperator with
P = 80 kW (source: NOXMAT).

212
Figure 6.57 shows the pressure drop about the recuperator of the burner with and without
loops. On the waste gas side the pressure drop was 5 mbar without and 9 mbar with loops.
On the air side we have 15 mbar with and 11 mbar without loops on the recuperator.
That is a difference of about 45 mbar per side.

Figure 6.57: Pressure drop about the plain and loop recuperator at furnace temperature
1000°C, 80kW (source: NOXMAT).

6.4.3 Comparison with the state of the art

The nominal capacity of the tested burner is 80 kW. A comparable and serial burner of the
NOXMAT product range is the K-RHGB 80-535 (Fig. 6.58) with a corrugated
recuperator. This burner and the loop burner are assembled with a fitting length of
535 mm. The differences between the two burners are:
 the geometry of the recuperator component (outer diameter);
 the waste gas part;
 the air part;
 other parts like gas part, electrode etc.
A comparison is only limitedly possible.

213
Figure 6.58: Ceramic burner with recuperator K-RHGB 80-535 (State of the art) (source:
NOXMAT).

The firing efficiency was calculated by using the formula of SIEGERT. Figure 6.59 and
table 6.XV show a comparison between the loop and corrugated recuperator of a serial
burner. Both descriptions show an increase of the efficiency of about 3.0% to 5.5% by
using a loop recuperator instead of a corrugated recuperator. This increase also depends on
the waste gas inlet temperature.

Figure 6.59: Comparison of the firing efficiency of a loop und corrugated recuperator
(source: NOXMAT).

214
 Waste gas outlet Loop recuperator Corrugated recuperator
temperature
800°C 81.5% 76.0%
900°C 77.0% 73.0%
1000°C 74.5% 69.8%
1100°C 70.0% 67.5%
1200°C 66.8% 63.5%
1300°C 63.5% 60.5%
Table 6.XV: Comparison of the firing efficiency of loop and corrugated recuperator
P = 80 kW (source: NOXMAT).

According to the increase of efficiency a decrease of the waste gas losses was expected.
Figure 6.60 and Table 6.XVI show a comparison of the waste gas outlet temperatures of
both burners.

Figure 6.60: Comparison of the waste gas outlet temperature of a loop und corrugated
recuperator (source: NOXMAT).

Waste gas outlet Plain recuperator Corrugated recuperator

temperature
800°C 460°C 550°C
900°C 550°C 630°C
1000°C 625°C 700°C
1100°C 700°C 775°C
1200°C 780°C 850°C
1300°C 860°C 920°C
Table 6.XVI: Comparison of waste gas outlet temperatures of loop and corrugated
recuperator P = 80 kW (source: NOXMAT).

215
By using a loop recuperator the waste gas outlet temperature falls by 60 K to 90 K
depending on the waste gas inlet temperature. These data correlate with the increase of the
firing efficiency.
Figure 6.61 shows the NOx-emissions of both burners. The values are related to 5% O2 in
the waste gas.

Figure 6.61: Comparison of the NOx-emissions of a loop und corrugated recuperator

(source: NOXMAT).

As shown in the chart the characteristic curves of loop and corrugated recuperator
(with power capacity of 80 kW) do not differ from each other. This should not be sò. The
increase of the firing efficiency causes a higher air preheating. As a consequence of this
increase, combustion temperature rises and in order to that the NOx-emissions rise too.
Nevertheless the shown values were measured. Further tests for clearing up this problem
were done. After further tests lower NOx-emissions were measured at the loop recuperator
than at the corrugated recuperator. Operating at capacities of 40 kW and 60 kW the
NOx-emissions of the loop recuperator are about 100 mg/m3 lower than those of the
corrugated recuperator. This decrease depends on the waste gas inlet temperature.
At a capacity of 80 kW there was no difference between the loop and the corrugated
recuperator. The reason for this effect is the different geometry of the air gap. This causes
another division of the combustion air into primary and secondary air. The gap between

216
burning tube and loop recuperator is bigger, the air preheating of the loop recuperator is
higher. The primary air drillings of the loop and the corrugated recuperator have the same
diameter. Because of the higher air preheating and the higher volumetric flow, less of the
combustion air flows into the combustion chamber. The lower primary air flow causes a
colder combustion temperature and the NOx-emissions decrease.
The loop recuperator had to be mounted and dismounted several times. The reasons for
that were the control and readjustment of different measuring points as well as mounting
the plain recuperator or the waste gas guiding tube. By doing these necessary activities
many loops were damaged or broken. After finishing the tests the surface of the loop
recuperator had many surface defects.
Due to the increased surface area of the tube with highly structured surface elements,
oxidation phenomena will be enhanced. In any case, if passive oxidation conditions are
met, they should not reduce the performance of the components, as previously
demonstrated on SiSiC reticulated ceramics [7].

Figure 6.62: Current serial burner K-RHGB 80 and the new burner with the loop
recuperator (source: NOXMAT).

Summary:
 The loop recuperator was tested in indirect heating with and without waste gas
guiding and at different capacities.
 A plain recuperator was tested at the same conditions.
 In dependence of the waste gas inlet temperature an increase of the firing
efficiency of about 4% to 8% could be achieved by using a loop recuperator.
 According to this the waste gas outlet temperature decreases by 130 K to 170K.
 Depending on the waste gas inlet temperature the air preheating increases by
130 K to 150 K.
217
  In consequence of the higher air preheating temperature the NOx-emissions rise by
50 mg/m3.
 A comparison between a burner with a corrugated recuperator and the loop
recuperator is only limitedly possible.
 In comparison to the state of the art some improvements in firing efficiency, waste
gas outlet temperature and air preheating could be achieved.
 There were better NOx-emissions of the loop recuperator than the corrugated one.
 By redesigning the test rig some loops on the inner and outer surface were broken
and lost.

Outlook:
 A better approach and reproducibility of the planned geometry should be aimed.
 Activities for protecting of the loops are necessary. The assembling and
maintenance could be simplified and damages could be prevented.
 The heights of the air and waste gas gap can be optimized and a better heat transfer
can be achieved. In order to that new and optimized loop geometries could be
possible.
 Further work is currently ongoing to study the overall behavior of these structures
along the entire lifespan of the component, in particular, to study the deposition of
dust and or soot onto the loops.

6.4.4 Alternative designs

The alternative designs include the development and fabrication of heat transferring
structures for innovative industrial applications. The heat exchangers will be tested under
laboratory conditions.
A new furnace component will be developed by using high efficient heat exchangers,
which will lead to a prototype at a laboratory scale. The aim is to create a high convective
gas flow for enhanced convective heating of protective furnace atmospheres. For a
successful use in industrial high temperature processes all heat exchanger components
must be completely gas-tight like ceramic heat exchangers are.

218
The preheater component will be designed and dimensioned based on the varying
consumption of protective gas of an industrial furnace. Scaling of these heat exchanger
applications is scheduled. The temperature required is defined by the industrial
application.
A first possible application is a preheater for protective gas (Fig. 6.63). Industrial furnaces
consume a considerable amount of protective gas. Preheating of process gas and injecting
the gas into the furnace will increase both efficiency and performance. An injection of
cold protective gas leads to a certain temperature inhomogenity within the furnace
atmosphere and increases the energy consumption in order to heat up the protective gas
using electrical heating elements.

Figure 6.63: Preheater for protective gas.

The layout of the preheating device is similar to the developed recuperator application.
An external burner is used for heating up the cold protective gas via a structured pipe with
heat exchange enhancing elements inside the pipe. The dimensioning is already done for
the recuperator application and could be scaled up or adapted according with the various
specific applications. For this scaling the ratio length to diameter of the structured heat
transferring pipe should remain constant.
The second possible application is the use of the caloric enthalpy of the furnace
atmosphere during the change of the atmosphere. This application is only in some details
of the design different to the pipe preheater of the first application and is also related to the
developed recuperator heat exchanger. However, in this application no additional burner
will be used. If a change of the process gas atmosphere is required, for example from

219
endothermic gas to nitrogen, the heat from the exhaust gas will be used for preheating the
incoming gas. Figure 6.64 shows a preheater transferring heat from flue gas to new
protective gas. In this arrangement the hot flue gas flows through the outer ring gap and
the protective gas gets heated up is in the middle of the heat exchanger.

Figure 6.64: Pipe heat exchanger for flue gas.

Another possible arrangement is to invert the gas flow arrangement. In this case the cold
protective gas is located at the outer ring gap and the hot flue gas in the middle of the heat
exchanger pipe (Fig. 6.65).

Figure 6.65: First detailed design of the prehaeter for flue gas, installed at a furnace wall.

220
To arrange the loops geometry in a plate heat exchanger also seems to be promising. Such
an application provides many advantages. A plate heat exchanger assembly could be
easily expanded with a various number of plates and therefore it is useful for different
performance categories. In general such a plate heat exchanger is useful for various
applications with hot media in many different fields of industry. It could be adopted for
higher temperatures than steel heat exchangers (above 1150°C) and it is corrosion
resistant. The higher heat transfer coefficient leads to higher heat recovery or same heat
recovery in smaller cross section and finally in significant higher energy savings.
First design studies were made with different arrangements concerning flow direction.
One possibility is the counter current flow of cold and hot media (Figure 6.66). Therefore
a comparatively high amount of heat could be transferred but the manufacturing efforts are
much higher than with other arrangements.

Figure 6.66: Plate heat exchanger with counter current flow.

A cross flow (Figure 6.67) would be the easiest way both to achieve gas tightness and for
manufacturing the periodic repeated elements. Only one element is needed to be stacked
rotated by an angle of 90° on top of each other. This makes it very easy to customize the
heat exchanger for different applications.
During literature research a patent was found which claims such a ceramic plate heat
exchanger with heat enhancing elements.

221
 Figure 6.67: Plate heat exchanger with cross flow.

222
6.5 References

[1] R. K. Shah and D. E. Metzger, American Society of Mechanical Engineers, American Society of
Mechanical Engineers, Heat Transfer Division and American Society of Mechanical Engineers, Gas Turbine
Power Division, Regenerative and Recuperative Heat Exchangers, ASME, New York, (345 East 47th ST.,
New York 10017), 86 (1981).
[2] J. G. Wűnning and A. Milani, Handbuch der Brennertechnik fur Industrieőfen, Vulkan Verlag,
Essen, (2007).
[3] D. Trimis, V. Uhlig, R. Eder, A. Ortona, S. Pusterla, E. P. Ambrosio, P. Fino, P. Rumeau,
C. Chazelas, S. Gianella, J. G. Wűnning, H. Altena, F. Beneke, M. Debier and T. Grämer, New Ceramic
Heat Exchangers with Enhanced Heat Transfer Properties for Recuperative Gas Burners, Heat. Process.,
9 [2] 183-7 (2011).
[4] Wikipedia, “Textile” (2012).
[5] N. S. Jacobson, Corrosion of Silicon-Based Ceramics in Combustion Environments, J. Am. Ceram.
Soc., 76 [1] 3-28 (1993).
[6] F. Mach, F. Issendorff, A. Delgado and A. Ortona, Experimental Investigation of the Oxidation
Behaviour of SiSiC Foams, Bioceramics and Porous Ceramics: Ceramic Engineering and Science
Proceedings, eds., R. Narayan and P. Colombo. John Wiley & Sons, Inc. Hoboken, NJ, 29 [7] 299-311
(2009).
[7] A. Ortona, S. Pusterla, P. Fino, F. Mach, A. Delgado and S. Biamino, Aging of Reticulated Si-SiC
Foams in Porous Burners, Adv. Appl. Ceram., 109 [4] 246-51 (2010).
[8] http://www.schunk-sik.com/en/schunk01.c.9111.de/all_productarticle;
[9] C. Wagner, Passivity during the Oxidation of Silicon at Elevated Temperatures, J Appl Phys, 29 [9]
1295-7 (1958).

223
 7. List of author’s publications

Articles

1) G. C. D’Amico, A. Ortona, S. Biamino, P. Fino, C. Badini and C. D’Angelo,

Oxidation Behavior at 1600°C of Si-SiC-ZrB2 Composites Produced by Si Reactive
Infiltration, Adv. Eng. Mat., 16 [2] 176-83 (2014).
2) A. Ortona, D. Trimis, S. Gianella, P. Fino, E. Boulet, V. Uhlig, R. Eder, C. Chazelas,
T. Grämer, G. C. D’Amico, E. Cresci, J. G. Wünning, H. Altena, F. Beneke and M.
Debier, SiSiC Heat Exchangers for Recuperative Gas Burners with Highly Structured
Surface Elements, Int. J. Appl. Ceram. Technol., (Article in press) 1-11 (2013).
3) A. Ortona, P. Fino, C. D’Angelo, S. Biamino, G. C. D’Amico, D. Gaia and S.
Gianella, Si-SiC-ZrB2 Ceramics by Silicon Reactive Infiltration, Ceram. Int., 38 [4] 3243-
50 (2012).
4) W. S. Yang, S. Biamino, E. Padovano, M. Pavese, S. Marchisio, G. C. D’Amico, S.
Ceresa Mio, X. Chen, P. Fino and C. Badini, Thermophysical Properties of SiC Multilayer
Prepared by Tape Casting and Pressureless Sintering, Compos. Structures, 96 469-75
(2013).

Presentations & Proceedings

A) G. C. D’Amico, A. Ortona, P. Fino, C. D'Angelo, S. Biamino, D. Gaia, S. Giannella,

Synthesis and characterization of Si-SiC-ZrB2 composite by LSI, in Carbon-based nano-
structured materials: graphene as a new player. Technologies, characterization,
applications, XVIII Scuola Nazionale di Scienza e Tecnologia dei Materiali, 16-20 July
2014, Ischia, Italy.
B) G. C. D’Amico, A. Ortona, P. Fino, C. D'Angelo, S. Biamino, D. Gaia, S. Giannella,
Synthesis of Si-SiC-ZrB2 composite by LSI, in Symposium B4.III - Composite Materials
and Systems for High Temperature Use, EUROMAT (European Congress on Advanced
Materials and Processes), 8-13 September 2013, Sevilla, Spain.

224

View publication stats