Influence of thermal conditions on the combustion synthesis of Si2N2O phase

Article abstract of DOI:10.1016/j.materresbull.2007.02.032

This work is a contribution to the study of Si₂N₂O synthesis via Self-Propagating High-temperature Synthesis (SHS). The elaboration of Si₂N₂O was carried out using two kinds of experimental equipments and differences both in reaction temperatures and in products compositions were evidenced. In order to determine the reaction mechanisms, several compositions and thermal conditions of the reactant mixture were studied. Considering our results, two kinds of reaction mechanisms for the combustion synthesis of Si₂N₂O could be suggested. Furthermore, the results previously reported in the literature have been explained in the light of this work. The SHS samples were characterized by X-ray diffraction and scanning electron microscopy (SEM) analysis. Their quantitative compositions were determined by Rietveld refinement.

Full text of DOI:10.1016/j.materresbull.2007.02.032

Materials Research Bulletin 43 (2008) 463–472
www.elsevier.com/locate/matresbu
Influence of thermal conditions on the combustion synthesis
of Si N O phase
2
2
a
N. Pradeilles , M.C. Record , R.M. Marin-Ayral , A.V. Linde ,
b
a,
c
*
c
I.A. Studenikin , V.V. Grachev
c
a
Laboratoire de Physicochimie de la Mati e` re Condens e´ e, LPMC UMR 5617, Universit e´ Montpellier II cc003,
Place Eug e` ne Bataillon 34095 Montpellier Cedex 5, France
b
Laboratoire TECSEN, UMR 6122, cc251, UPCAM Aix-Marseille III,
Avenue de l’escadrille Normandie Ni e´ men 13397 Marseille Cedex 20, France
c
Institute of Structural Macrokinetics and Materials Science (ISMAN), Chernogolovka, Moscow Region 142432, Russia
Received 2 October 2006; received in revised form 4 January 2007; accepted 19 February 2007
Available online 23 February 2007
Abstract
This work is a contribution to the study of Si
The elaboration of Si O was carried out using two kinds of experimental equipments and differences both in reaction
2 2
N O synthesis via Self-Propagating High-temperature Synthesis (SHS).
2 2
N
temperatures and in products compositions were evidenced.
In order to determine the reaction mechanisms, several compositions and thermal conditions of the reactant mixture were studied.
2 2
Considering our results, two kinds of reaction mechanisms for the combustion synthesis of Si N O could be suggested.
Furthermore, the results previously reported in the literature have been explained in the light of this work.
The SHS samples were characterized by X-ray diffraction and scanning electron microscopy (SEM) analysis. Their quantitative
compositions were determined by Rietveld refinement.
#
2007 Elsevier Ltd. All rights reserved.
Keyword : A. Ceramics; B. Chemical synthesis; C. High pressure
1
. Introduction
Silicon oxynitride (Si N O) is the unique compound in the SiO –Si N quasi-binary [1]. Like Si N , Si N O is a
4
2
2
2
3
4
3
2 2
refractory material which exhibits good thermal, chemical and mechanical stability, as well as diffusive barrier and
dielectric properties [2].
However, while Si N can be encountered at room temperature and atmospheric pressure in two crystalline forms (a
3
4
and b), Si N O is a compound with a well-defined composition and structure [3]. Therefore, its constitution is constant
2
2
and the improvement and/or the control of its mechanical and electrical properties must be easier than for Si N [4,5].
3
4
The synthesis of silicon oxynitride is usually performed by reactive sintering of a Si N and SiO mixture in the
3
4
2
presence of a liquid phase obtained by the addition of Al O or Y O [6,7]. Nevertheless, because the additives are still
2
3
2 3
present in the product of this reaction, other routes have been investigated: nitridation of high purity silica in ammonia
*
Corresponding author. Tel.: +33 467143355; fax: +33 467144290.
E-mail address: ayral@lpmc.univ-montp2.fr (R.M. Marin-Ayral).
0025-5408/$ – see front matter # 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.materresbull.2007.02.032

464
N. Pradeilles et al. / Materials Research Bulletin 43 (2008) 463–472
at 1120 8C [8] and Self-Propagating High-temperature Synthesis (SHS) from desert sand and silicon mixture under
nitrogen pressure [9]. In the first case, the formation of Si N O always goes along with the formation of undesirable
phases: Si N , SiN O and SiN O . In the SHS product, a further phase b-Si N is also evidenced though in a minor
2
2
3
4
3
2
2
3 4
amount.
Therefore, the Self-propagating High-temperature Synthesis seems to be a good alternative.
In this process, a strong exothermic reaction propagates spontaneously through the reactant mixture, converting it
into product.
Recently, some of us [10] have also investigated the elaboration of Si N O from a Si + SiO mixture under nitrogen
2
2
2
pressure using the SHS process. However, in this case, the SHS device was not a conventional one. The product of the
reaction was constituted of a-Si N , b-Si N and residual silicon; Si N O was not observed. The only way to get
3
4
3
4
2 2
Si N O with this kind of apparatus was to use colloidal silica as raw material. The rate of formation was then very high
2 2
(
97 wt.% at PN ¼ 2 MPa) and no extra phase was found in the SHS product.
2
The objective of this work was to understand the combustion synthesis of Si N O with two kinds of apparatus in
2
2
order to explain the differences between the results previously reported [9,10]. In this aim, we studied the influence of
the SHS device and more generally of the thermal conditions on the formation of Si N O.
2
2
This work results from an international collaboration between the LPMC Laboratory (Laboratoire de Physico-
chimie de la Matiere Condensee, Universite Montpellier II, France) and the ISMAN Institute (Institute of Structural
Macrokinetics and Materials Science, Chernogolovka, Russia).
2
. Experimental procedure
2
.1. Experimental devices
Two kinds of SHS device were used in this study; they mainly differ by the ignition system and the thermal
conditions.
2
.1.1. LPMC device
In the LPMC device, the reactant mixture is compacted, then placed into a furnace under a high nitrogen pressure
and heated at a high speed. The crucible which contains the compact is boron nitride.
Because of the existence of a thermal gradient (the furnace only consists in one heating zone), the reaction is ignited
on one side of the sample and its direction of propagation is controlled. The temperature profiles are determined by
means of three thermocouples (T , T , T ) and recorded during the reaction. A schematic diagram of the SHS furnace
1
3
4
is given in Fig. 1.
The compacting pressure (Pcomp = 300 MPa) and the heating rate (90 8C/min) have been selected in order to get
both the ignition and the complete propagation of the SHS reaction. If these experimental parameters are too low (for
the heating rate) or too high (for the compacting pressure), thermal losses are higher than the provided energy and the
reaction cannot be ignited. The nitrogen pressure ðPN ¼ 2 MPaÞ has been selected to be in the same order of
2
magnitude as the nitrogen pressure used by Radwan et al. [9].
Fig. 1. Schematic diagram of the SHS furnace (LPMC).

N. Pradeilles et al. / Materials Research Bulletin 43 (2008) 463–472
465
2
.1.2. ISMAN device
In the ISMAN device, the reactant mixture is used without compaction at room temperature.
It is placed into a container filled with boron nitride powder to minimize heat losses. The reaction ignition, under
high nitrogen pressure, of the top of the sample is mediated by the combustion of titanium powder. A tungsten spiral
heated by electric current is used to initiate titanium powder combustion. The temperature is determined by means of
two thermocouples WRe5/WRe20 of 0.2 mm in diameter. This device is presented on Fig. 2.
The value of 10 MPa for the nitrogen pressure was also selected in this case in order to compare all the results.
2
.2. Green mixtures
In our previous work [10], it was shown that the reaction of formation of Si N O can be optimised with a better
2
2
contact between the reactant powders. Therefore, the present study was carried out using four kinds of raw mixtures.
The reactant proportions in all these mixtures were defined according to the stoichiometry of the following reaction:
3
2
1
2
SiðsÞ þ SiO2ðsÞ þ N2ðgÞ ! Si2N2OðsÞ:
2
.2.1. LPMC device
The first one (labelled A LPMC) was constituted of silicon (>99 wt.% of Si, 0.28 wt.% of Fe, 0.15 wt.% of Ag, and
-
0.11 wt.% of Al, Aldrich, ꢀ45 mm in size) and silica (Fontainebleau sand, 260 mm) powders. In order to improve the
reactivity of the reactant mixture, the Fontainebleau sand was ball-milled for 15 h.
-
The second one (labelled B LPMC) was prepared using a colloidal silica solution LUDOX HS40 (39.5 wt.% of SiO2,
.4 wt.% of Na O and 60.1 wt.% H O, Dupont Chemical). Silicon powder was added to the LUDOX silica under
2 2
0
stirring. After a total homogenisation (45 min), this solution was heated (70–80 8C) to remove the liquid phase. And
a further thermal treatment, performed at 110 8C for 2 h, led to a homogeneous powder.
2
.2.2. ISMAN device
-
The third one (labelled A ISMAN) has the same composition as the first green mixture (A LPMC). It is constituted of
crystalline silicon and silica. However, its preparation was realized from ISMAN’s powders by mixing them into a
planetary ball-milling machine during 7 min.
-
The fourth one (labelled B ISMAN), contains, like the mixture B LPMC, silicon powder and amorphous silica. But
the preparation is also different in this case. It was made from ISMAN powders, the amorphous silica was heat
Fig. 2. Schematic diagram of the SHS device (ISMAN).

466
N. Pradeilles et al. / Materials Research Bulletin 43 (2008) 463–472
treated at 600 8C in order to increase the specific surface, and the mixture was prepared into a planetary ball-milling
machine during 7 min.
2
.3. Methods of characterization
The starting materials were characterized by means of laser granulometry (Mastersizer 2000-Malvern), scanning
electron microscopy (SEM) coupled with an EDX detector (Leica S260), gas adsorption (BET) and differential
thermal analysis (DTA).
X-ray diffraction was used to characterize the SHS products. The diffraction patterns were recorded on a Phillips
Expert diffractometer (with a copper K a anticathode) in the [10–1308] 2u range, with a step size of 0.008368 and a
1
step time of 500 s. The phase quantification was performed by Rietveld refinement (FULLPROF suite) [11,12].
Morphologies and compositions of the SHS products were identified by scanning electron microscopy coupled with
an EDX detector (Leica S260).
3
3
3
. Results and discussion
.1. Characterization of the starting materials
.1.1. Mixture A LPMC (Si + Fontainebleau sand)
In the original Fontainebleau sand, the grains are round shaped with a mean size approximately equal to 260 mm.
After the ball-milling, the grain size is reduced to 8.4 mm.
The BET specific surface areas measured for the silicon, the original Fontainebleau sand and the ball-milled one are
given in Table 1. As expected, the specific surface area of the sand is higher after milling.
The DTA profile recorded for this mixture under argon up to 1200 8C is presented on Fig. 3. An endothermic
transformation is observed at 575 8C on the heating curve. This phenomenon which is reversible could correspond to
the phase transition between low quartz (rhombohedral) and high quartz (hexagonal) [13]. This result is in agreement
with the X-ray diffractograms recorded before and after the differential thermal analysis. In both cases, the silica is
crystalline with a low quartz structure.
3
.1.2. Mixture B LPMC (Si + SiO LUDOX)
2
The grain distribution of this mixture is monomodal with a mean size equal to 50 mm (Fig. 4). This result is in
agreement with SEM observations which present silicon grains surrounded by silica (Fig. 5).
The DTA profile recorded for this mixture under argon up to 1200 8C is presented on Fig. 6.
An exothermic phenomenon is observed on the heating curve at 950 8C. Since the cooling curve does not show any
endothermic event, this phenomenon is not reversible. Therefore, we assume that it corresponds to the crystallization
of an amorphous silica phase. The exothermic phenomenon observed at 210 8C on the cooling curve might correspond
to the phase transition between high and low cristobalite [13].
Table 1
Specific surface areas (BET) of the reactants in LPMC and ISMAN mixtures
2
Specific surface area (m /g)
Reactant
Average size of particles (mm)
LPMC powders
Silicon
0.921
0.025
1.830
–
260
8.4
Fontainebleau sand
Milled sand
ISMAN powders
Silicon
4.2
0.6
1.12
5.04
–
Milled quartz
Amorphous silica
310

N. Pradeilles et al. / Materials Research Bulletin 43 (2008) 463–472
467
Fig. 3. DTA thermogram of Fontainebleau sand: (1) heating; (2) cooling.
Fig. 4. Particle size distribution of the reactant mixture B (LPMC).
The X-ray diffractograms recorded before and after the DTA corroborate these assumptions (Fig. 7): crystalline
silica is only present after the DTA (b), it is evidenced as cristobalite. A slight deviation of the base line in the angular
range [20–408] can be noticed on the diffractogram recorded before DTA (a).
2
The BET specific surface area of this mixture B is relatively high; it amounts to 109 m /g.
Fig. 5. SEM micrograph of the reactant mixture B (LPMC).

468
N. Pradeilles et al. / Materials Research Bulletin 43 (2008) 463–472
Fig. 6. DTA thermogram of the reactant mixture B: (1) heating; (2) cooling (LPMC).
Fig. 7. X-ray diffractograms of the reactant mixture B: (a) before DTA, (b) after DTA.
3
.1.3. Mixture A and B ISMAN
The characteristics of the ISMAN powders are given in Table 1. For mixture B ISMAN, the specific surface of silica
2
2
is 266 m /g before the heat treatment at 600 8C and 310 m /g after that. It is still amorphous after the heat treatment.
3
3
.2. Combustion synthesis
.2.1. LPMC device
When the studied mixture was constituted of silicon and Fontainebleau sand (grain size = 8.4 mm), the reaction was
ignited and propagated all along the sample. The corresponding thermal profiles are depicted in Fig. 8. The ignition
temperature is 1416 8C that corresponds to the melting point of silicon. The maximum temperature is around 1500 8C.
The duration of the phenomenon is approximately 20 s.
Fig. 8. Thermal profiles recorded for the reactant mixture A (LPMC device).

N. Pradeilles et al. / Materials Research Bulletin 43 (2008) 463–472
469
Fig. 9. Thermal profiles recorded for the reactant mixture B (ISMAN device).
For the mixture B LPMC, the thermal profiles have similar shapes to the previous ones. The reaction is also ignited
at the melting point of silicon but in this case the maximum temperature is slightly lower (1485 8C) than the previous
one and the duration of the phenomenon is slightly longer (35 s).
3
.2.2. ISMAN device
For the mixture A ISMAN the reaction is ignited and propagated all along the sample. The corresponding thermal
profiles are depicted in Fig. 9. The maximum temperature is around 2000 8C, very close to the maximum temperature
reported in the elaboration of the silicon nitride [14] with the same kind of SHS device.
The speed of combustion in the ISMAN device (0.3 mm/s) is much lower than the speed reported for the LPMC
apparatus (2.6 mm/s). This difference could be due to the preheating of the sample in the latter device.
The mixture B ISMAN was also ignited. The thermal profiles have similar shapes to the previous ones. The
maximum temperature is almost the same, but combustion velocity is 1.1 mm/s.
Table 2 compares the velocity of combustion and the maximum temperature obtained with each kind of device. The
LPMC device leads to a lower maximum temperature and a higher velocity of combustion. The mixture B LPMC was
also studied with the ISMAN apparatus, but the reaction could not be ignited.
3
3
.3. Characterization of the SHS samples
.3.1. Reactant mixture (A) (LPMC, ISMAN)
The results of the Rietveld refinement are given in Table 3. When the ISMAN device is used, the SHS product is
made of Si N O, a-Si N and b-Si N . No residual silicon was observed.
3 4
2
2
3
4
With the LPMC device, the SHS samples contain a-Si N , b-Si N and residual silicon.
3
4
3 4
Table 2
Velocity of combustion and maximum temperature obtained from reactant mixtures A and B
Device
Velocity of combustion (mm/s)
Maximum temperature (8C)
A (ISMAN)
A (LPMC)
B (ISMAN)
B (LPMC)
0.3
2.6
1.1
2.0
2160
1520
2133
1485

470
N. Pradeilles et al. / Materials Research Bulletin 43 (2008) 463–472
Table 3
Composition of the SHS products (from mixture A and mixture B) determined by Rietveld refinement ðPN2 ¼ 10 MPaÞ
Device
Si (wt.%)
Si
2
N
2
O (wt.%)
a-Si
3
N
4
(wt.%)
3 4
b-Si N (wt.%)
Mixture A
LPMC
29
–
–
36
27
35
30
ISMAN
43
Mixture B
LPMC
5
92
55
3
–
ISMAN
7.5
26
11.5
Fig. 10. X-ray diffractogram of an SHS product from mixture B (LPMC).
3
.3.2. Reactant mixture (B) (LPMC, ISMAN)
The results of the Rietveld refinement are given in Table 3. The SHS product obtained with the LPMC device is
constituted of Si N O, a-Si N , and residual silicon. As the silica phase has not been evidenced and because we can
2
2
3 4
notice a slight deviation of the base line in the angular range [20–408] (Fig. 10), we can assume that this phase is not
crystallized but is present in an amorphous form.
The ISMAN product contains Si N O, a-Si N , b-Si N and residual silicon. Once more, we can notice a slight
2
2
3
4
3 4
deviation of the base line in the angular range [20–408] and no silica in the SHS product. Therefore, the residual silica
might be in an amorphous form.
4
. Mechanisms of formation
The main results of this study are summarized in Table 4. They clearly show that the percentage of Si N O, the
2
2
velocity of combustion, the ignition and the maximum temperature depend on the device. Therefore, we can
reasonably assume that the mechanism of reaction differs from one apparatus to another.
Considering our results, we can suggest the following mechanisms.
With the LPMC device, the reaction is ignited at the melting point of silicon.
For the mixture A, nitrogen can easily reach the silicon grains, then it can be dissolved into the melted silicon and
when the composition of the liquid corresponds to the stoichiometry of Si N , the reaction takes place that yields this
3
4
Table 4
Main results of this study
Device
Mixture
Velocity of combustion (mm/s)
Ignition temperature (8C)
Maximum temperature (8C)
2 2
Si N O (wt.%)
LPMC
ISMAN
LPMC
ISMAN
B
B
A
A
2.0
1.1
2.6
0.3
1416
–
1485
2133
1520
2160
92
55
0
1416
–
43

N. Pradeilles et al. / Materials Research Bulletin 43 (2008) 463–472
471
compound. As it was previously suggested for the elaboration of Si N [15], this reaction might lead to the formation
3
4
of b-Si N . a-Si N might be the product of a direct nitridation which takes place in the coldest part of the compact.
3
4
3 4
Indeed, this phase was evidenced at the rim of the sample.
The SHS samples are then constituted of a-Si N , b-Si N , residual Si and amorphous SiO .
3
4
3
4
2
In the mixture B, the nitrogen cannot reach the silicon grains. When silicon melts, it dissolves silica which
surrounds it and comes in contact with the nitrogen gas. Then this liquid containing silicon and oxygen dissolves
nitrogen and when the stoichiometry is reached, the formation of Si N O occurs. This phenomenon of dissolution
2
2
results in a wider thermal profile.
The presence of a-Si N is also evidenced in this case. It was previously reported that the direct reaction between
silicon and nitrogen is made possible by the use of a high nitrogen pressure which breaks the coating of silica. At lower
3
4
pressure ðPN ¼ 2 MPaÞ, only Si N O, residual silicon and amorphous silica were evidenced in the SHS product [10].
2
2 2
The maximum temperature is around 1500 8C in both cases.
2
3
With the ISMAN device, the reactant powders are heated from room temperature at a very high speed (10 –10 K/
s). Therefore, we can assume that in these conditions silicon is sublimed instead of being melted. This assumption is
corroborated by the fact that the reaction was not ignited with this apparatus when the silicon grains are surrounded by
silica (mixture B LPMC).
Thus, the sudden rise in temperature appearing in the thermal profiles is due to a highly exothermic reaction which
takes place between gaseous silicon and nitrogen. Part of the released energy is then used to melt the remaining solid
silicon which could not be sublimed so far and this liquid can either react with nitrogen to form b-Si N or dissolve
3
4
silica to subsequently react with nitrogen to form Si N O.
2
2
The mechanism of formation of Si N O by SHS has never been studied so far. The results available from literature
2
2
only concern the elaboration of Si N in a conventional SHS device. Two kinds of mechanism have been suggested:
4
3
according to Mukasyan and Borovinskaya [16], nitrogen reacts with silicon by a vapour–liquid–crystal (VLC)
mechanism; by contrast, according to Ge et al. [17], nitrogen reacts with gaseous silicon.
The mechanism suggested in this work combines these two kinds of reaction with nitrogen dissolution into the
melted silicon. As shown in Fig. 11, the thermal profiles can be considered as the superposition of two phenomena: the
first one (a) corresponds to a highly exothermic reaction between nitrogen and gaseous silicon, and the second one (b)
corresponds to the reaction between nitrogen and melted silicon.
The previously reported results on the combustion synthesis of Si N O [9] can be explained by our suggestion.
2
2
The maximum temperature reached by this system is around 1500 8C, and only two phases have been evidenced in
the SHS product, namely Si N O and b-Si N .
If a-Si N does not appear, we can assume that the reaction between gaseous silicon and nitrogen does not occur.
2
2
3 4
3
4
Indeed it seems difficult for silicon to be sublimed with the configuration used in this study. The SHS device is a
conventional one but the ignition system is located at the bottom of the crucible (between the crucible and the
powders).
Fig. 11. Decomposition of a thermal profile recorded on the ISMAN device. (a curve) Reaction between N
and Si(l).
2 2
and Si(g); (b curve) reaction between N

472
N. Pradeilles et al. / Materials Research Bulletin 43 (2008) 463–472
Without this highly exothermic reaction, the maximum temperature might not reach 2000 8C, in contrast to the
ISMAN device.
Nevertheless, the silicon was melted and could react directly or after the silica dissolution with nitrogen, leading
respectively to b-Si N and Si N O. These reactions were possible because a refractory compound (Si N O) was
3
4
2
2
2 2
added to the mixture, and because of the location of the ignition system, the heat losses were reduced.
The low amount of b-Si N can be explained by a preliminary ball milling (for 2 h) of the reactant powders. As with
3
4
the mixture B in the LPMC device, the direct reaction between liquid silicon and nitrogen was prevented by a better
contact between the reactant powders.
5
. Conclusion
The objective of this work was to explain the differences observed between previously reported results [9,10] on
combustion synthesis of Si N O. Because these differences seemed to be due to the experimental conditions, an
2
2
international collaboration between the LPMC laboratory (France) and the ISMAN institute (Russia) has been
established in order to carry out comparative investigations using two different kinds of SHS devices. These apparatus
mainly differ by the ignition system and the thermal conditions.
Our results allowed us to understand the reactions occurring within both two kinds of apparatus during the
combustion synthesis of Si N O. Furthermore, the results previously reported in the literature have been explained in
the light of this work.
2
2
Acknowledgments
We are grateful to the French embassy in Russia for financial support, PICS (Grant No. 3462) and RFBR (Grant No.
6-03-22000).
0
References
[
[
[
[
[
1] L. Kaufman, Calphad 3 (4) (1979) 275–291.
2] D.A. Buchanan, IBM J. Res. Dev. 43 (3) (1999) 245–247.
3] S.R. Srivinasa, L. Cartz, J.D. Jorgensen, T.G. Worlton, R.A. Beyerlein, M. Billy, J. Appl. Cryst. 10 (1977) 167–170.
4] R. Roucka, J. Tolle, D.J. Smith, P.A. Crozier, I.S.T. Tsong, J. Kouvetakis, Appl. Phys. Lett. 79 (18) (2001) 2880–2882.
5] R. Roucka, J. Tolle, A.V.G. Chizmeshya, P.A. Crozier, C.D. Poweleit, D.J. Smith, I.S.T. Tsong, J. Kouvetakis, Phys. Rev. Lett. 88 (20) (2002)
206102.
[
[
[
[
6] Z.K. Huang, P. Greil, G. Petzow, Ceram. Int. 10 (1) (1984) 14–17.
7] B. Bergman, H. Heping, J. Eur. Ceram. Soc. 6 (1) (1990) 3–8.
8] R. Vanweeren, E.A. Leone, S. Curran, L.C. Klein, S.C. Danforth, J. Am. Ceram. Soc. 77 (10) (1994) 2699–2702.
9] M. Radwan, T. Kashiwagi, Y. Miyamoto, J. Eur. Ceram. Soc. 23 (2003) 2337–2341.
[
[
[
[
[
[
[
[
10] N. Pradeilles, M.C. Record, R.M. Marin-Ayral, J. Eur. Ceram. Soc. 26 (13) (2006) 2489–2495.
11] H.M. Rietveld, J. Appl. Crystallogr. 2 (1969) 65–71.
12] J. Rodriguez-Carvajal, Abstract of the Satellite Meeting on Powder Diffraction of the XV Congress of the IUCr, Toulouse, France, 1990.
13] C.A. Jouenne, Trait e´ de c e´ ramiques et mat e´ riaux min e´ raux, ed. Septima, 1975.
14] K. Hirao, Y. Miyamoto, M. Koizumi, Adv. Ceram. Mater. 2 (4) (1987) 780–783.
15] N. Pradeilles, M.C. Record, R.M. Marin-Ayral, Int. J. SHS 15 (2) (2006) 157–167.
16] A.S. Mukasyan, I.P. Borovinskaya, Int. J. SHS 1 (1) (1992) 55–77.
17] C.C. Ge, J.T. Li, Y.L. Xia, Int. J. SHS 5 (2) (1996) 107–117.