The carbothermal production of Si3Al3O3N5 from kaolin, its sintering and its properties

The carbothermal production of Si3Al3O3N5 from kaolin, its

sintering and its properties

Citation for published version (APA):

van Dijen, F. K. (1986). The carbothermal production of Si3Al3O3N5 from kaolin, its sintering and its properties.

Technische Universiteit Eindhoven. https://doi.org/10.6100/IR251692

DOI:

10.6100/IR251692

Document status and date:

Published: 01/01/1986

Document Version:

Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be

important differences between the submitted version and the official published version of record. People

interested in the research are advised to contact the author for the final version of the publication, or visit the

DOI to the publisher's website.

• The final author version and the galley proof are versions of the publication after peer review.

• The final published version features the final layout of the paper including the volume, issue and page

numbers.

Link to publication

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain

• You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:

www.tue.nl/taverne

Take down policy

If you believe that this document breaches copyright please contact us at: openaccess@tue.nl

providing details and we will investigate your claim.

THE CARBOTHERMAL PRODUCTION OF

Si

3

AI

3

0

3

N

5

FROM KAOLIN,

THE CARBOTHERMAL PRODUCTION OF

Si3AI303N

5

FROM KAOLIN,

lTS SiNTERING AND lTS PROPERTIES.

P.RQEFS<:;I;{

,

R,I

,

FT

ter

ve~krij~ing

van d,e

w~~d v~n ~oçtor a~n

de Technisçhe

Universit~it

.

Eindhoven, op ge

,

zag

van de rector rnagnificus, prof.dr

.

F.N:

Hooge,

voor een commissie aangewezen

~o~r

h

et

col'

e~e

van dekanen in het

openba~r

te

verpedjg~n

op

pinsdag 28

q

ktober te

1

6.90

uur

.

P()Or

FRANCISCUS KORNELI? VAN DIJEN

Dit proefschrift is goedgekeurd door de promotoren:

Prof.dr. R. Metselaar en

CONTENTS.

CHAPTER 1: INTRODUCTION. 1.1. Technica! ceramics.

1.2. Non oxide engineering ceramics. 1.2.1. State of the art of si 3N4 and

sialons.

1.2.2. Preparation techniques for non oxide ceramic powders based on silicon and aluminium.

1.3. Problems associated with the carbothermal production of Si 3Al 30 3N5 from

PAGE 01 01 02 02 08

kaolin, its sintering. and its properties. 12 1. 4. Aim of this study.

1. 5. Comment on the properties of Si 3Al 30 3N5

1. 6. Literature.

CHAPTER 2: RAW MATERIALS, REACTION EQUATIONS, THERMODYNAMIC ASPECTS, AND

EXPERIMENTAL SET UP. 2 .1. Introduet ion.

2.2. Raw materials. 2.2.1. Introduction.

2.2.2. The oxide materials. 2.2.3. The carbon.

2.2.4. The gas.

2.3. The reaction equations. 2.4. Thermadynamie calculations.

2.4.1. Thermadynamie data for the main compo-nents.

2.4.2. Thermadynamie data for the impurities and water. 12 14 16 18 18 18 18 19 20 22 22 24 24 29

2.5.4. Determination ~f the reaction rate and the reaction products.

2.5.5. Powder characterisation. 2.6. Literature.

CHAPTER 3: REACTION KINETICS FOR THE CARBOTHERMAL PRODUCTION AND CHARACTERISATION OF Si 3Al 3

o

_{3N5 .}

3 .1. Introduetion.

3.2. Factors which influence the overall reaction rate.

3.2.1. The influence of the reactor on reaction kinetics.

3.2.2. Physical phenomena influencing the reaction rate.

3.2.2.1. Diffusion.

3.2.2.2. The gas flow rate. 3.2.2.3. Thermal effects.

3.2.3. The chemical reaction rate.

3.2.4. A proposal for the reaction mechanism in the carbothermal production. 3.3. Experimental results.

3.3.1. General results.

3.3.2. Results concerning the physical aspects of the reaction rate. 3.3.3. Results concerning the chemical

aspects of the reaction rate. 3.3.4. Comparison between the results

obtained in this thesis and the

PAGE 38 39 43 45 45 46 46 48 48 51 66 67 69 72 72 78 82

results described in the literature. 84 3.4. The role of impurities on the chemicàl

reaction rate and their behaviour during the carbothermal production

3.5. Characterisation of the produced si 3A1 3o 3N5 powders.

3.5.1. Introduction. 3.5.2. Resu1ts.

3.5.3. Parameters which inf1uence the specific surface area of Si 3A1 30 3N5 powder. 3.6. Conc1usions.

3.7. Literature.

CHAPTER 4: THE PROCESS' PROSPECTS. 4.1. Upscaling.

4.1.1. Theoretica! ca1cu1ation of the tempera-ture distribution.

4.1.2. Measurements of the temperature distribution.

4.2. Heat effects. 4.3. F1exibi1ity.

4.4. Heat generation inside the reactor. 4.5. Literature.

CHAPTER 5: EVALUATION OF THE SINTERING

PAGE 88 88 88 90 93 94 96 96 96 105 106 109 112 115 BEHAVIOUR OF si 3A1 3

o

3N5 . 116 5.1. Introduction. 116 5.2. Powder characterisation. 116 5.3. Sintering and the use of sintering aids. 122 5.4. Experimenta1 aspects of the sintering

process.

5.5. Results of the sintering experiments. 5.5.1. Inf1uence of the type of

additives and the N/0 ratio. 5.5.2. Influence of time. temperature

124 126

PAGE

5.6. Observations with TEM. 137

5.6.1. Experimenta1 resu1ts. 137

5.6.2. Discussion of the data in the

1iterature. 138

5.7. Discussion and conclusions. 138

5.8. Literature. 140

CHAPTER 6: EVALUATION OF MICROSTRUCTURE AND PROPERTIES OF THE SINTERED MATERIAL. 6.1. Introduction.

6.2. The crystal structure of si 3A1 30 3N5 .

6.2.1. Theoretica! aspects.

6.2.2. Experimental resu1ts.

6.2.3. Comparison between data from the literature and experiments.

6.3. The microstructure of Si 3Al 30_{3N5 .}

6.3.1. Introduction.

6.3._{2. The microstructure of Si 3Al 3}

o

3N5 .

produced trom kaolin.

6.4. Properties of Si₃Al₃

o

3N5 .

6.4.1. Hardness and fracture toughness.

6.4.2. The strength.

6.4.3. The oxidation resistance.

6.4.4. The modulus of elasticity, Poisson•s

ratio, and shear modulus.

6.4.5. The heat capacity and thermal conduc

-141 141 14). 141 142 144 144 144 147 148 148 155 156 167 tivity of Si 3Al3 03N5 . 168

6.4.6. The electrical conductivity. 169

6.5. Conclusion about the sintered

Si 3Al 30_{3N5 material. produced}

from kaolin. 6.6. Literature.

170 171

CHAPTER 7: SUMMARY AND CONCLUSIONS. SAMENVATTING EN KONKLUSIES.

APPENDIX.

WOORD VAN DANK.

CURRICULUM VITAE. PAGE 173 178 183 188 189

1. INTRODUCTION.

1.1. Technica! ceramics.

Ceramic materials are often divided into groups on the basis of their applications: for instanee for buildings. refractories, sanitary ware. dinner ware. electric-. elec-tronic-, magnetic materials. and "engineering" materials. The electric-. electronic-. magnetic-. and "engineering" materials are often referred to as technica! ceramics. This work concerns "engineering" ceramics: ceramics with good mechanica! properties. More specifically we will be concerned with the material si 3Al 3o 3N5 • produced from kaolin.

A bistorical review. up to 1983, of the development of engineering ceramics based on silicon and aluminium is gi-ven by P. Popper (1). So far, the following materials based on Si and Al are known: sio2 • Al 2o 3 . 3Al 2o 3 .2Sio2 . Si 2N20, Si 3N4 , AlN, SiC, y-alon. a•-sialon (Si 6 _zAlzOzNB-z

(Z<4.2)). a'-sialon (Mx(Si.Al) 12 (0,N) 16 (x<2),

si 2 _xAlxN2 _xol+X or 0'-sialon with x<0.2, SiC-AlN solid solutions. polytypes. X-phase, and Si-C-Al-0-N (2,3,4,5). Of all these compounds, only the oxides. Si 3N4 and SiC are well developed.

Applications for these materials include engine com

-ponents, heat exchangers. wear-resistant materials. cut

-ting tools, substrates for integrated circuits. refrac-tories. armour-platings etc. Si 3N4 and SiC-alloys look particularly attractive. Materials like partially sta-bilised zro2 • Al 2Tio5 and AlN are also worth studying. At present, the problems associated with engineering eera

-mies relate to economie mass production of products (6) with little variatien in their mechanica! properties.

-1-In the early Seventies, the aim of most research projects in this field was to develop a ceramic gas turbine. Since a gas turbine is a very complicated and expensive product, the cost of the overall production line. especially the cost of raw materials, is less important. In the Eighties. research focussed on the less sophisticated applications mentioned earlier. In this kind of application. the cost of the raw materials and the overall production line must not be neglected.

1.2. Non oxide engineering ceramics.

1.2.1 State of the art for Si 3N4 and sialons.

In the early Seventies _{Si 3N4 could only be sintered to}

sufficiently high densities by applying mechanica! pres-sure (prespres-sure sintering}. In the late Seventies, sin-tering si 3N4 without the application of mechanical

pressure was successful. The problem with the sintering is the nature of the chemical bond in Si 3N4 . The bonding between the atoms in materials like Si 3N4 , AlN, and SiC has a covalent nature. This results in low atomie

diffusion coefficients in these materials. An increase of temperature increases the ditfusion coefficient; however, at temperatures where the mobility of the atoms is sig-nificant, sublimatien of the material is also significant.

In order to sinter covalent materials. the use of proper

sintering aids is necessary. For instance, sintering of Si 3N4 can be achieved by the addition of a

thermodynamically stable oxide which farms an eutectic of low melting point with the surface silica of the si 3N4 particles. During sintering, this oxide reacts with the

surface _{silica and with some of the Si 3N4 , and forms a}

sintering occurs at a temperature where sublimation of the material is not critical. After the sintering process. the liquid generally forms a glassy phase at the grain

boundaries and the triple points. By applying the proper heat treatment, this glassy phase crystallises in some cases. By selecting the correct composition. the sintering aids can be incorporated into the Si 3N4 crystal structure. i.e. solid solutions are formed. A special class of solid solutions is obtained by replacing part of the Si by Al and simultaneously substituting 0 for N. These compounds are known under the acronym •sialons•.

The properties of si 3N4 . include toughness~ strength and oxidation resistance at elevated temperatures, all de-pend to a large extent on the physical and chemical pro

-perties of the grain boundary phase. The advantages of si 3N4 as a structural ceramic are its relatively low specific density, its high strength at elevated tem-peratures and its thermal shock resistance. Problems as-sociated with si 3N4 (and other ceramic materials) are variation in its strength and its brittleness. The latter is demonstrated by its KIC value of about 8

MPa.m~.

see also table 1.1.

Two types of sialons which are related to si 3N4 are: the

a'- and the ~·-sialons. The a'-sialons have a similar crystal structure to a-Si 3N4 : however. the dimensions of the unit cell of a•-sialons are somewhat larger. The structure is stabi1ised by incorporating ions like ca, Li, or Y in the crystal structure. Theoretically, the

a'-sialons are represented by the formula

Mex(Si.Al) 12 (0,N) 16 where x < 2. It must be noted that a unit cell cannot contain more than one atom of oxygen (3).

-Fig. 1.1. Projection on (001) of the B'-sialon structure with spaeegreup P6 3 /m.

Si:Si,Al sites; 1: N.O. 6h sites. 2: N,O 2c sites. all atoms at fractional heights z = ~. %.

A

c

Cnml

Cnml

0.772

0.304

0

.

770

0

.

302

0.768

0.300

0.766

0.298

0.764

0.296

0.762

0.294

0.760

0

.

292

0

2

3

4

5

6

z

Fig. 1.3. The phase diagram of the system Si 3N4 -Al 20 3 -Si02 -AlN. idealized (4).

~13Aip1AIN)

-5-Youngs modulus 320

Bend strength (20°C) 1000

Bend strength (10oo•c) 1000

Theemal conductivity (2o•c) 35 Theemal conductivity (1ooo•c) 18 Poisson•s ratio

Hardness (HV 15.625 kg) Specific heat (2o•c)

0.27 15 630

Specific heat (10oo•c) 1255

Electrical conductivity (2o•c) <1o- 14 Dielectric constant (9375 GHz) 8.5 Tan b (9375 GHZ) 0.003 8 230 700 700 6 6 0.285 15 720 1240 <10-14 7 0.003 6

The 8'-sialons have a similar crystal _{structure to a-si 3N4} and the space group is P6 3/m (Z=2), see tigure 1.1. In the S'-sialon structure, Al replaces Si and 0 replaces N

resulting in increased unit cell

dimensions due to the increased Al and 0 contents. see Figure 1.2. The amount of si and N replaced by Al and 0 is found from the forrnula si_{6 _}2Al202N8 _ 2 (z<4.2). The

stability region of these solid solutions is shown in the phase diagram of the system Si02 -Al 2

o

3-si 3N4 -AlN, see figure 1.3. Due to incorporation of Al and o in the

B-Si 3N4 structure. the chemical bond becomes more ionic in character. Although S'-sialons and other non oxide ceramic materials are considered to be covalent materials, they

Unit GPa MP a MP a W m- 1 K- 1 W m-1K- 1 GPa J kg-1K-l J kg-lK-1

s m

-

1 MPa m~

are Bornetimes tteated as ionic materials. The "ionic"

radii of the •ions" are (4): N3 -: 0.150 nm;

o

2 -: 0.138 nm; Al3_{+: 0.037 nm; si}4_{+: 0.025 nm. These radii refer to ions}

with a coordination number of 4. Table 1.2 shows the bond length, the difference in electronegativity, and the percentage ionic characteristic for each bond, assuming a coordination number of 4.

Table 1.2. Character of the different bond types in a•-sialon.

bond Si-N Si-0 Al-N Al-O bond length (nm) ionic character (~) 0.0175 0.163 0.187 0.175 difference in electronegativity 30 1.2 51 1.7 43 1.5 63 2.0

Due to the incorporation of Al and 0 in the crystal struc-ture, the combination of properties of a·-sialons with

zjo, _{is different from that of si 3N4 . For instance, with}

an increasing z-value for a·-sialon. a marked de-crease in the thermal conductivity, the coefficient of thermal expansion and the modulus of elasticity is observed, see table 1.1.

When Si 3N4 is sintered with Al 2o 3 and AlN. theoretically. a material without a glassy phase at the grain boundaries should be obtained. However, in the literature no

a•-sialon material without a glassy phase at the grain boundaries has been reported. The same holds for tt'-sialons.

At present. the sialons have not been studled as much as si 3N4 (or SiC); In this thesis. attention is focus~ed on Si 3Al 30 3N5 • a a•-sialon with a high z-value.

-7-1.2.2. Preparatien techniques for non oxide ceramic pow-ders based on silicon and aluminium.

Ceramic powders form the foundation of a ceramic product and it could be expected that powders for non-oxidic en-gineering ceramics based on Al and/or Si would be rela-tively cheap. due to the fact that their elements: Si. Al,

o.

N,

c.

are readily available at low prices. However. the

raw materials must be very fine (submicron) powders. Ideally, these powders will be spherical. uniform sized granules that are free from agglomerates and have a high purity. The production methods for such powders, the small quantities. and the high purity required make these pow-ders expensive. A review of production methods and the

properties of non-oxidic powders was given by Schwier (7). As the most important powder production methods we mention:

a) _{gas phase reactions. e.g. 3SiC1 4 + 4NH3} 4 si 3N4 +

12HC1; SiC1 4 + _CH4 4 SiC + 4 HCl; A1Cl 3 + NH3 4

AlN + 3 HCl.

b) reactions between the elements e.g. 3Si + _2N2 4

Si 3N4 ; Si + C 4 SiC; 2Al + _N2 ~ 2AlN.

c) reactions of a gas _{with a liquid e.g. SiC1 4} + _2NH3

~ _{Si(NH) 2} + _{4HC1 with 3Si(NH) 2} 4 Si 3N4 + 2NH 3 .

d) _{carbothermal reactions e.g. Si02} + 3C 4 SiC + 2CO+

3Sio2 + 6C + _{2N 2} 4 Si 3N4 + 6CO.

The reducing agent carbon can be in a gaseous form. for

instanee _{as CH4 .} or in a solid ferm, for instance, as

carbon black, Instead of carbon, hydrogen is sometimes

~sed as the reducing agent.

Let us cernpare these methods in more detail. The usual

si-licon souree is quartz sand and the usual aluminium souree

is alumina which in turn. is usually produced from

-reaction step, whereas the other powder production methods require more than one reaction step. For instance. sicl 4 can be made by the reaction of sio2 with carbon and chlorine: the elements Al and Si are usually produced by electrochemical reaction between the oxide and carbon. So, the carbothermal production process is, in theory, the most economie powder production method. In comparison with other powder production processes for non oxide powders. it is also a good process from the environmental point of view. For instance. no chlorine or HCl are produced in the carbothermal production process.

Yet, the method has attracted relatively 11ttle attention for the production of sialons. Usually, these materials are made by reaction sintering. To this end, Si 3N4 . AlN and Al 2o 3 powders are first thoroughly mixed. This mixture of powders is then shaped and, subsequently. i t reacts and sinters in one heat treatment. Using ihe

carbo-thermal process to produce e•-sialon. a mixture of sili

-ca. alumina, and carbon is heated in nitrogen. It should be noted that powder production and sintering are two se

-parate steps in this case.

To avoid the step of homogeneously mixing the starting oxide powders one can sometimes use a chemical compound: for instanee aluminiumsilicates which are compounds of aluminium, silicium, oxygen and, often, •crystal water•. The aluminium, silicium and oxygen are mixed at atomie

level; therefore, these silicates are a good raw material for the carbothermal production of 6'-sialon. Besides, in the form of minerals, they are relatively cheap. How-ever. the presence of impurities in the silicates may in-fluence the production of the powder and the properties of the sintered ceramic articles, in such a way that the use of minerals cannot be considered very practical.

For the production of solid solutions in the system. Si02 -Al 2o 3-si 3N4 -AlN, several minerals can be used; e.g. kaolin or andalusite.

-9-In this way, compounds of Si 6 _zAlzOzNB-z are ob-tained with different values of z.

This work is mainly concerned with the production of ~·-sialon _{from kaolin, 2Sio2 .Al 2o 3 .2H2o. leading toa} composition with z;J according to the reaction: 3(2Si02 .Al 2o 3 .2H20) + _{SN 2 • lSC} 4 2Si 3Al 30 3 N5 + _6H20 +

lSCO.

The mineral kaolin is readily available, and has rela-tively high purity. Moreover, this mineral has been stu-dled by several other investigators too, which makes i t possible to campare the dif(erent results reported in the literatuce.

The ditference between kaolin and kaolinite can be ex-plained in the next way: kaolinite is the clay mineral with the chemical _{formula (OH) 8si 4Al 4o 10 and with a} spec1tic crystal structure, kaolin is a powder which mainly consists of kaolin minerals e.g. nacrite, dickite, halloysite and kaolinite. In the literature t i l l 1986, on-ly kaalins which mainly consist of the mineral kaolinite have been used for the carbothermal production of

a•-sialon.

Lee and Cutler (B) were the first to describe the for-mation of ~·-sialon by the carbothermal conversion of kaolin.

This was achieved by mixing kaolin with carbon powder, pelletizing the mixture and heating the pellets in a ver-tical tube turnace at temperatures below l4SO•c. Paris and Grollier-Baron (9) were not able to repeat the work of Lee and Cutler; however, they succeeded in producing

si 3Al 3o 3N5 powder from kaolin with the use of porous bricks made from kaolin, carbon powder. an inert filler like saw dust, and nitrogen, at temperatures of about 1600 •c. This is contrary to the findings of Lee and

Cutler who stated that si 3Al 3o 3N5 was only

pro-duced at temperatures below 1450°C. Several other people looked into this reaction. for instanee Baldo et. al. (10). Gavrish et. al. (11). and Yoshimatsu et. al. (12).

The carbothermal production of other non-oxide ceramic powders has been studled too (13-17). However. all these studies present insufficient information for the design of a reactor. for optimisation of the reaction. or for selec-tion of the raw materials.

The carbolhermal production of non-oxide ceramic materials was already studied in the nineteenth century. At this

time the Acheson process for the production of SiC (18). as well as. the process for the production of cac 2 (18). were developed. Also. in the nineteenth century, the first attempts to produce nitrides by the carbothermal produc-tion process were made (19). It must be noted that in those days there was no interest in the production of non-oxide ceramic articles.

During later years, a small but constant interest in the production of non-oxide ceramic materials for different purposes occurred; e.g. hard metals, wear resistent coa-tings, fertilizers, the production of aluminium from AlN. etc. The research initiated in the fifties in Great Bri-tain and in the early seventies in West Germany, sweden and the ijSA. in the field of ceramics as an engineering material. also stimulated research into the production of non-oxidic ceramic powders.

At present. no carbothermal production process for fine non oxide ceramic powders based on Si and/or Al is com

-mercially available.

ll-1.3. Problems associated with the cachothermal production of

si

_{3Al 3}

o

_{3N5 from kaolin. its sintering. and}

its properties.

From paragcaph 1.2, it becomes clear that the design cri-teria for a commercial reactor are unknown at present. Criteria for the selection of raw materials and for op-timlsation of the ceaction are lacking. Descrepancies exist concerning the carbothermal production of fine. non-oxidic, ceramic powders in the literature.

The sintering of Si 3A1 30 3N5 powder rather than the reaction sintering of a mixture of si 3N4 • AlN. and Al 2o 3 powder is rarely reported in the literature. The influence

of impurities. which may originate from the

car-bon or the kaolin on the sintering process. is not

des-cribed at all.

The properties of a B'-sialon material with a. high z-value

are not well-known, this is especially so for Si 3Al 3

o

_3N5 produced from kaolin. The relationship between the

properties and microstructure of such a

mate-rial is not fully understood yet.

Therefore, no conclusion can be drawn in order to say whether the carbothermal production of fine. non oxide powders is a good production method or not, on an in-dustrial scale. It is not known either. whether, the use

of minerals instead of synthetic oxides is justified or

not. No conclusion can be drawn either to say if i t is

worth producing Si 3Al 3

o

3N5 products rather than products

made from si 3N4 or SiC. 1.4. Aim of this study.

to optimise the carbothermal production of si 3Al 3o 3N5 powder from kaolin,

to study the role of impurities and characteristics of the powder,

to investigate the sintering behaviour of the powder with and without sintering aids,

Therefore, this study will include some research into the structure and properties of 6'-sialon and on shaping the powder too.

In this first chapter an introduetion to the engineering ceramics based on Si and Al is given as well as an in-troduction to this thesis. In the second chapter, the pro-perties of the raw materials, the reaction equations, and some thermadynamie calculations for the carbothermal pro

-duction of si 3Al 3o 3N5 from kaolin are given.

Chapter 3 contains a model describing the physical and chemical aspects of the reaction rate. It contains the ex-perimental part of the carbothermal production of

Si 3Al 3o 3N5 from kaolin and a discussion of the factors which influence the partiele size of the product.

The fourth chapter deals with sealing up the process and with the carbothermal production of other chemica! com

-pounds. This results in a comparison between the carbo

-thermal production of si 3Al 3o 3N5 and SiC.

Chapter five contains a description of the powder before the sintering process and the factors which influence the sintering process. A discussion is given of the sintering process including the influence of impurities, sintering aids, and the oxygen/nitrogen ratio.

Data in the literature concerning the sintering process is incorporated into the discussion. Also this chapter deals with sintering experiments.

In chapter six, microstructures of sintered Si 3Al 3o 3N5 are shown. The crystal structure is described in detail and, on the basis of the literature, a review is given of the relationship between the properties, on one hand, and the crystal structure and microstructure, on the other hand.

-13-This chapter _{also contains the properties of si 3Al 3}

o

_3N5 which was made according to the production methad described in this thesis. A summary and a conclusion of the results are given in chapter seven.

In this section a survey is made of a number of properties of Si 3Al 3o 3N5 . Table 1.1 summarizes the available

literature data. According to Gauckler et al (20), the X-ray density of Si 3Al 3o 3N5 is 3.1 g cm- 3 . The relatively low value of the modulus of elasticity. compared to the value (or si 3N4 • is probably due to the presence of Si-0 bands. lts hardness is similar to that _{of Si 3N4 . The low}

thermal conductivity o( _{Si 3Al 3}

o

3N5 , compared to the

thermal conductivity of Si 3N4 is due to pbonon scattering which is a normal phenómenon for solid solutions. The luwer value of the coefficient of thermal expansion. for

si

3Al 3

o

3N5 • in comparison with that tor si 3 N4 • is due to Lhe lower X-_{ray density of si 3Al 30 3N5 .}

The resistance to fracture initiatien by thermal shock,

R'. can be predicted by the application of formula ~:

R'

where a is the bend strength. ~ is the thermal con-ductivity, ~ is Poisson's ratio. E is Young's modulus and a is the coefficient of thermal expansion.

The resistance to fracture initiatien by thermal shock of si 3Al 3

o

3N5 is nat as good as that of si 3N4 . This is mainly due to the much lower thermal conductivity. The lower modulus of elasticity and the lower coefficient of thermal expansion of Si 3Al 30 3N5 . compared to those _{of si 3N3 ,} can

thermal shock is concerned. The fracture toughness of si 3Al 30 3N5 is lower in general. than the fracture

toughness of si 3N4 . The fracture toughness of Si 3Al 30 3N5 is about 6 MPa

m~.

The reason for this is not known. however more attention has been paid to optimizing the microstructure of Si 3N4 than to optimizing the

microstructure of Si 3Al 30 3N5 .

Si 3Al 3

o

_{3N5 is a non-conducting material. from the}

electrical point of view. The dielectric properties of Si 3N4 and si 3Al 3

o

3N5 were published by Lowell (21). The relative dielectric constant for Si₃Al₃

o

₃N₅ is 7.1 at 9375

GHZ for temperatures between 25°C and 12006_C. The loss tangent ranges from 0.0002 at 25°C to 0.004 at 1200°C.

-15-1.6. Literature.

1. Popper P .• Trans. J. Br. eer. Soc., 82, 187 (1983).

2. Mceauley J.W .. earbin N.O., in Progress in Nitrogen

eeramics, Riley F.L .. ed., Martinus Nijhoff Pu-blishers, Boston, 111 (1983).

J. Jack K.H .. i.n Progress in Ni tragen eeramics, Riley

F.L .. ed., Martinus Nijhoff Publishers. Boston, 45

(1983).

4. Jack K.H., J. Mat. Sci.,

.u..

1135 (1976).

5. Rafaniello W., Plichta M.R., Virkar A.V., J. Am. eer.

soc .. 66, 272 (1983).

6. Langer M., Siebels J., Heinrich H .• in Keramische

Komponenten für Fahrzeug-Gasturbinen lil,

Sprin-ger-Verlag, Berlin. 419, (1984).

7. Schwier G., in Progress in Nitrogen eeramics. Riley

F.L., ed., Martinus Nijhoff Publishers, Boston. 157, (1983).

8. Lee J.G .. eutler I.B., Am. eer. soc. Bull.,

2!.

869

(1979).

9. Paris R.A .. Grollier-Baron T .. Eur. Pat. Appl. 23869,

(1981).

10. Baldo J.B .. Pandolfelli v.e .• easarini J.R., in

eera-mie Powders, Vincenzini P .• ed., Elsevier Scientific

Publishing eompany, Amsterdam, 437 (1983).

ll. Gavrish A.M., Boyarina I.L., Degtyareva E.V., Puchkov

A.B .• Zhockova Z.D., Gul'ko N.V., Tarasova L.A .• In

-org. Mat., ~. 46 (1982).

l2. Yoshimatsu H .. Mitomo M., Mihashi H., Ohmori s.,

Yabuki T .. Yogyo-Kyokai-Shi.

21.

443 (1983).

13. Li W.L .. Huang L.P., Huang X.Z., Kuang G., Tan S.H.,

~·wu S.R., Yen T.S., in eeramic Powders, Vincenzini

P .• ed., Elsevier Scientific Publishing eompany, Am

14. Sweda A., Hendry A., Jack K.H., in Specia~ Ceramics 7, Taylor D., Popper ?., eds., The British Ceramic Society, Shelton, Stokeon Trent, 107. (198l). 15. Wei G.C .. Comm. Am. Cer. Soc.,~. C-111 (1983). 16. Mori M .. Inoue H .• Ociai T .. in Progress in Nitrogen

ceramics. Riley F.L .• eds .. Martinus Nijhoff Pu-blishers. Boston, 149 (1983).

17. Zhang s.c .. Cannon R.W .. J. Am. eer. Soc .. 67. 691 (1984).

18. Kirk-Othmer, Encyclopedia of Chemica! Techno1ogy, John Wiley & Sons. Inc .. (1979).

19. Gmelins Handbuch der Anorqanischen Ch~mie. Aluminium Teil B, l_i, Verlag Chemie GmbH, Berlin (1984).

20. Gauckler L.J., Prietze1

s.,

Boderoer G., Petzow G .. in Nitrogen Ceramics. F.L. Riley ed., Noordhoff, Leiden. 529 (1977).

21. Lowell R.F.,

22. Wild S., Elliot H., J. Mat. Sci.,

.u_,

1769 (1978).

-17-2. RAW MATERIALS, REACTION EQUATIONS, THERMODYNAMIC ASPECTS AND EXPERIMENTAL SET UP.

2.1. lntroduction.

In this chapter, the raw materials which are needed for the carbothermal production of si 3Al 3

o

3 N5 or other

non-oxide ceramic powders are described. It continues with a discussion of the reaction equations, which explain the

importance o( the carbon:oxide ratio in the carbothermal production process. The following paragraph deals with the thermadynamie aspects of the carbothermal production of si 3Al 30 3N5 from kaolin, carbon and nitrogen. It will appear that the thermodynamic data which is avai

-lable, is not accurate enough to explain all the reactions which can occur and are described in the literature.

In the last paragraph of this chapter, the preparatien of the raw materials and the pellets, as well as the powder characterisation methods, the determination of the reaction rate and the reaction products, are described.

2.2. Raw Materials.

2.2.1. Introduction.

For the carbothermal production of a chemical compound, usually a carbon source, an oxide compound and often a gas iike nitrogen is needed. The gas is often needed to remove reaction products from the reactor. Should i t be necessary to produce a nitride or an oxynitride, nitrogen will be needed to obtain the desired chemical compound. In this paragraph, the raw materials used for the experiments are described.

2.2.2. The oxide materials.

Bath for the reaction and for the prope~ties of the re-sultant powders and/or sintered products, the purity and partiele size of the starting oxides are of great im-portance. In our experiments. we have used mineral kaolin, with the chemica! formula: 2Sio2 .Al 2o 3 .2H20.

In comparison with synthetic oxides, minerals are gene

-rally less pure. Impurities commonly encountered in kaolin are quartz, titania, calcia, magnesia, ironoxide,

sodiumoxide and potassiumoxide. The amounts of these impurities depends on the origin of the mineral. Kaalins from three different sourees were used in this study: Monarch, Kick and Provins. The chemical analyses of these kaalins are shown in table 2.1.

Experiments were also performed with quartz sand. The re

-sults of the chemical analyses of this mineral are also shown in table 2.1. The table also gives the specific sur-face area measured by nitrogen adsorption according to the one-point BET methad {"Ströhlein"). Por kaolins, the spe-cific surface area may vary between 5 and 50 m2 ;g. Prom the data in table 2.1, i t can concluded that Provins is a very fine kaolin and Monarch is a relatively pure kaolin. Por a mineral, the quartz can be considered to be rela-tively pure too. Prom X-ray diffraction measurements (Philips PW1009 with Ni-filtered CuKa radiation). it was found that the main impurities were anatase in Monarch, a-quartz and anatase in Provins and a-quartz and

K-feldspar in Kick. Upon heating, the kaolin will loose its "crystal water" at app~oximately

sso•c

and will farm SiO and mullite at about 1"S0°C.

L

-19-Table 2 . l . Properties of Monarch kaolin -- ·- ··- - -- ·· Si02 45.6x Al 2o 3 38.6 Fe_{2o 3} 0.34 Ti0 2 1. 37 cao MgO K20 0.06 Na 2o loss on 13.9 ignition

_..

B.E.T. 6.7 the minerals Provins kaolin 46.6 34.9 l.B l.B 0.65 0.2 0. l 0.2 14.0 45.2 used. Kick kao lin 52.5 34.9 0.3 0.3 0.3 0.1 2.1 0.2 11.7 5.7 Brunssumroer sand (milled) (Si02 ) 99.96 0.01 0.03 0.7

x _{wt\ on "dry" basis. Here, "dry• means in equilibrium} with the moisture in air.

specific surface area in m2/g, as determined by the BET method.

Data supplied by N.V. Koninklijke Sphinx, Maastricht.

2.2.3. The carbon.

A1though gases like c3 H8 can be selected as a carbon souree for the carbothermal production process. this study was limited to solid sourees with a high carbon content. Many of the experiments were carried out with carbon

Several types are commercially available. In this work different types of carbon black (Cabot BV. Rotterdam) were used. Characteristic data on these products is given in table 2.2.

Other possible sourees of solid carbon are graphite. coal and coke. For the carbothermal production. synthetic grap-hite is comparable with carbon black. Natural grapgrap-hite i i similar to a coal such as anthracite. In general. a coal will be coarser and less pure than carbon black. Espe-cially. it will contain more sulphur. volatile matter and ash. Coke, produced from coal. will have similar proper-ties to coal, but the amount of sulphur and volatile mat-ter will be lower (1). Petroleum coke is also relatively coarse and i t contains sulphur and volatile matter too, but its ash content is generally less than 1 wt\ (2). For comparison with the carbon black in this study, a petroleum coke (SSM Nederland BV, Rotterdam) was used, ha-ving the properties specified in table 2.2.

Cokes and coal are of special interest since they are much cheaper than carbon black.

Table 2.2. Properties of some carbon materials.

volatile matter x (wt\) ash content (wt\) sulphur (wt\) fixed carbon (wt\) ash constituents BET (m2/g) x heated in N2 at 1400°C. Fixed carbon is estimated.

125 <1 <0.1 <0.5 98 27 -21-Elftex 475 575 <1 <1 <0.1 <0.1 <0.5 <0.5 98 98 75 110 Petroleum 675 coke <1 .12.5 <0.1 0.4 <0.5 0.5 98 86 Si02 140 7.2

2.2.4. The gas.

For the carbolhermal production, the gas serves a dual

purpose. In the first place, it is needed to remove CO

from the reactor in case the CO pressure is less than the operating pressure of the reactor. For this purpose. an

inert gas like argon can be used.

When nitrides or oxynitrides are wanted. nitrogen is nee-ded for the reaction. then. i t can then perform both func-tions.

lmpurities like water. oxygen or co2 may influence the

reaction. The nitrogen used in the experiments contained 6 ppm

o

_{2 .} measured with an oxygen gauge. As the nitrogen was

obtained by the cold destillation of air, the amount of

H2o or co2 was less then l ppm due to their high freezing

or sublimation point compared with the boiling point of

o

₂

or N2 .

2.3. The reaction equations.

Starting with kaolin. carbon and nitrogen. many reactions are possible. We shall not discuss the reaction mechanism or reaction rate here. only giving possible overall reac-tion equareac-tions.

Thc formation of _{Si 3Al 3}

o

_{3 N5 is described by the equation}: J(2Si0₂.Al₂o₃.2H₂0)~SN₂+15C ~ _{2Si 3Al3}0_{3 N5 +15C0+6H2}0 2.1

Lee and Cutler (3) stated that above l450°C, a mixture of Sic plus AlN is formed instead of Si 3Al 30 3N5 :

It can be seen that for reaction 2.2. the ratio car-bon:kaolin is much higher (9:1) than for reaction 2.1 (5:1). When a stoichiometrie amount of carbon is used for the formation of si 3Al 3o 3N5 . a product consisting of only SiC and AlN cannot be obtained whatever the tem

-perature. The boundary temperature of 1450°C is determined by thermadynamie considerations, see the next paragraph. Other carbothermal reactions with the kaolin may occur: 2(Sio2 .Al 2o 3 .2H20)+6C ~ _{Al 20 3+2SiC+4CO+H2o} 2.3 3(2Sio2 .Al 2o 3._2H20)+21C+_7N2 ~ 2Si 3N4 +6AlN+21C0+_6H2o 2.4 3(2Sio2 .Al 2o 3._{2H20)+12C+4N2 )} ~ _{2Si 3N4 +3Al 20 3 +12C0+6H20 2.5}

For each type of reaction. a different ratio carbon:kaolin will be needed when stoichiometrie amounts of carbon are required. Reactions which produce SiO are not considered here.

In the next section. the investigation concerns reac -tions which are likely to occur at a given temperature. In the system Si-Al-0-N-C, without impurities obtained from the raw materials, solid solutions of si 2N2o with Al and 0, of Si 3N4 with Al and 0, and of SiC with AlN exist. Yet more compounds are described in the literature, see chapter 1. As no thermadynamie data is available for many compounds in the system Si-Al-0-N-C. it will not be possible to describe the system in a thermadynamie way. so. in this thesis, the main system is divided into two subsystems: the system sio2 -c-N2 and the system

Al 2o 3-C-N2 . The description of the two subsystems proved to be adequate to explain the reactions observed. Even these two subsystems are rather complex and not fully understood. For instanee two crystal modifications exist for Si 3N4 and several crystal modifications exist for SiC. It is known that ~-sic can contain several wt~ of nitrogen (10).

The following reactions may be considered: Si02 +3C+2/3N2 ~ Si0+C0+2C+_2/3N2

Si02 +3C+2/3N2 ~ _{Si+2CO+C+2/3N2} Si02 +3C+2/JN2 ~ _{SiC+2C0+2/JN2}

Si02 +3C+2/3N2 ~ _{l/3Si 3N4 +2CO+C}

Si02+3C+2/3N2 ~ _{l/2Si 2N20+3/2C0+3/2C+l/6N2} -23 -2.6 2.7 2.8 2.9 2.10

Al 20 3 +9/2C+N2 ~ _{Al 20+2C0+5/2C+N2}

Al 20 3+9/2C+N2 ~ _{Al 20 2 +C0;7/2C+N2}

Al 20 3 +9/2C+N2 ~ _{2Al+3C0+3/2C+N 2}

Al 20 3 •9/2C;N2 ~ ~Al₄c₃.JCO+N₂

Al 2

o

3 • 9/2C+·N2 ~ 2AlN+lCO+l/2C

In this work. experiments were performed at temperatures between lJOo•c and 17oo•c. According to the phase diagram of McCauley and Corbin (4). no aluminium oxynitirdes will farm below 170o•c. Therefore. these reactions are left out of the discussion. In addition to the alumina or the si1i-ca, also the formed products can react, and a reaction can

be considered like:

2SiC+Si0 2 2N2 ~ Si_{3N4 +2CO}

Si0+·2C ~ SiC<CO

Si 2N20+C<1/3N2 ~ _{2/3Si 3N4 +C0} 3SiC+Al 20 3 +3N2 ~ _{Si 3N4 +2AlN+3CO}

These equations suggest that the amount of carbon plays an

important role (5) as confirmed by Lee. Since the reactions take place at temperatures higher than lOOo•c. the formation of

co

_{2 insteadof CO is unlikely (6). The}

formation of SiO. Al 2

o

and Al 2

o

_{2 is highly undesirable.}

because these oxides. which are volatile at high

temperatures. will cause a loss of silicon or aluminium.

2.4. Thermadynamie calculations.

2.4. l. Thermadynamie data for the main components.

Thermodynamics may be used to predict whether a reaction

can occur or not. This is done by calculating the change

in the standard Gibbs energy of a reaction, according to

2 . l l 2.12 2. 13 2.14 2.15 2.16 2.17 L:....!.!!_ 2.19

This formu1a is simplified by assuming that

T o T o

J

₂₉₈ ~CP . dT- T .

J

₂₉₈ ~CP/T.dT equa1s zero (7). For our purpose this assumption is certainly justified since the error introduced in this way is less than 1 kJ/mole. so. equation 2.20 can be written as

0 0 0

~GT ~ ~H298 - T.~S298"

Also. no discriminatien between the different crystal structures was made in the ca1culations. The change in Gibbs energy, as a function of the temperature for the reactions 2.6 to 2.19, is given in table 2.3. Reliable thermodynamic data for the calculation of the change of the standard Gibbs energy as a function of the temperature is available for most of the reactions ~ to 2.19. An exception is the change of the standard Gibbs energy as a function of the temperature for the formation of si 2N2o from the elements. Experimental thermodynamic data for si 2N2o is given by Hendry (8) and Fegley (9). Hendry's relationship for the formation of si 2N2o from the

elements. below 14l2°C, is: ~G~ = - 658.3 + 0.131 T kJ. This expresslon differs considerably from Fegley's relationship for the formation of si 2N2o from the elements. below l4l2°C: ~G~ = -947.7 + 0.28 . T kJ. Finally, according to the theoretica! model of Dörner (11). in which Si 2N2o is treated as a mixture of si 3N4 and sio2 • the following relationship was found for the

formation of si 2N2o from the elements. below l412°C: ~G~ -815.5 + 0.245 . T kJ.

-25-It should be noted that, at the relevant temperatures, discrepancies between the models decrease. si 2N2o seems exceptionally stable according to Hendry's model compared

with _{si 3N4 . Since Dörners model falls between Hendry's and}

Fegley's data and because the theoretica! model according to Dörner agrees better with the data from Fegley than with the data from Hendry, this model was used in our

calculations.

Table 2.3. Change in Gibbs energy per male Sio2 or Al 2o 3 or an equivalent*). Reaction ll.Go T (J) ref. -- - -· ··-.?~--~ 448000-256.3 T 11 _?_,_fi_ 687000-343.8 T 7 ~,_2 677000-347.1 T (T < 1685 K) 7 728000-377.1 T {T > 1685 K) 7 .?~ 604000-339.4 T 7 L~ 435000-242.0 T 7 .?_,_10 328000-182 T 11

LU

1288000-547.7 T 7 .?~ 1102000- 383.7 1' 7

L.U

1344000--584.1 T 7 2. 14 1212000-536.6 T 7 2 .1~ 690000-353.1 T 7 _?~,___!__§_ 33000-15.8 T 7 2.17 -83000+4.4 T 7 2. lA 219000-111 T 11 2.19 183000-61.0 T 7

Fig. 2.1. Standard Gibbs energy changes in the system Si02-_{c-N2 as a function of the temperature. The}

sym-bols correspond with the reaction equations given in this chapter. ~: SiO ~: Si L.!!_: SiC ~: Si3 N4 2.10: Si 2N20

4

3

2

0

1000

_K

2000

Fig. 2.2. Standard Gibbs energy changes in the system Al 2o 3 -C-N2 as a function of the temperature. The symbols correspond with the reaction equations given in this chapter. 2.11: Al 20 t.Go

₈

T

_2.13

2.12: Al 2

o

₂ (J.10 5 ) 2.13: Al 2.14: A4C3

6

2.15: AlN

4

2

0

1000

_K

2000

-27-Figure 2.3. Standard free energy changes for some reae -lions as a tunetion of the temperature. The symbols cor-respond with the reaction equations given in this chapter.

2.16: 2SiC<Si02 +2N2 ~ _{Si 3N4 +2CO} ?~17: si0+2C ~

sic.co

;2~: _{Si 2N20+li3N2 .. 2/3Si 3N4}<-C0 2.19: 3SiC+Al 203+3N2 .. Si 3N4+2AlN+3CO ;2_,_!_: Si₃Al_{30 3N5} 2 0

-1

1000

_K

2000

The change in the standard Gibbs energy of reaction ~

and the formation of si 3Al 3o 3N5 from kaolin. can be calcu1ated because the change in enthalpy and entropy for the formation of si 3Al 3o 3N5 from si 3N4 • AlN and Al 2o 3 or from si 3N4 and Al 3o 3N is described in the literature (11,12). This result is also included in Table 2.3. In the calculation, the removal of crystal water and the formation of mullite and quartz were not included because these reactions take place befere the reaction temperature for the carbothermal production is reached. i.e. in the calculation kaolin was simulated as ~(4Sio₂ +

3Al 2o 3 .2Si02 ).

The results of these calculations are represented in the figures 2.1, 2.2 and 2.3. From these figures, it can be seen that the formation of volatile oxides is not likely; further AlN is more stable than Al 4c 3 . while above 1450°C, the formation of SiC instead of Si 3N4 may occur.

2.4.2. Thermodynamic data for the impurities and water.

Minerals such as kaolin and sand may contain a variety of impurities. Normal impurities are: Fe, Ti. ca. Mg, Na and K. Kaolin also contains •crystal water• (13). Carbon will contain impurities like ash and sulphur. The ash consists of elements similar to these present in kaolin. Knowledge is required of the behaviour of the impurities during the reaction, their influence on the reaction kinetics. the reaction products, the sintering and the properties of the sintered product in order to select the proper raw mate-rials. According to Ullmann. water will react with carbon at temperatures above 700°C (6). This reaction will take

-29-Fig. 2.4. Standard Gibbs energy changes for the system Mg0-_{C-N 2}

-

si

3N4 . E : 2Mg0~2C ~ 2C0+2Mg àG~-1230000-579.7 J/mol

6

5

4

3

2

N : 2Mg0•2C+2/3N2 ~ _{2C0+2/3Mg 3N2 •} AG~-666000-251.4 T J/mol

S : 2Mg0•2C+l/3Si_3N4 ~ _{2CO+Mg 2Si}+2/3N 2

~G~=ll82000-495.4 T J/mol

Fig. 2.5. Standard Gibbs energy changes for the system Ca0-C-N2 -si 3N4 .

6

5

4

3

2

1000

E : 2Ca0+2C ~ 2C0+2Ca ~G~=l052000-388.7 T J/mol N : _{2Ca0+2C+2/3N2} ~ _{2C0+2/3Ca 3N2} ~G~=744000-240.B T J/mol C : 2Ca0+6C ~ _{2C0+2Cac 2} ~G~=931000-441.2 T J/mol

S : _{2Ca0+2C+l/3Si 3N4} ~ _{2CO+Ca 2Si+2/3N2}

~G~=ll52000-518.6 T J/mol

Fig. 2.6. Standard Gibbs energy changes for the system Fe 20 3 -C-N 2 -si 3N4

0

-1

-2

-3

1000

E : 2/3Fe 20 3 +2C ~ 2C0+4/3Fe, l\G~c3l4000-338.7 T J/mol N : 2/3Fe 20 3 t2C+l/6N2 • l/3Fe 4N, l\G~~303000--315.4 T J/mol

K

~ _{: 4/9Fe 3C+4/9Si 3N4} ~ _{4/3FeSi+8/9N 2 +4/9C,} l\G~~210000-l40.7 T J/mol

Fig. 2.7. Standard Gibbs energy changes for the system Ti02 -C-N2 -Si 3N4 .

3

-1 E : Ti0 2 +2C ~ 2CO+Ti ~G~=712000-349.1 T J/mol N : Ti0 2 +2C+l/2N2 ~ 2CO+TiN ~G~=376000-255.9 T J/mol C : Ti0 2 +3C ~ 2CO+TiC ~G~=527000-336.6 T J/mol

S : TiN+l/5Si 3N4 ~ _{l/5Ti 5si 3 +9/4N2} ~G~=364000-156.3 T J/mol

-33-place according to the following equation: H2o + c 4 co +

H2 . "Crystal water" is released from the kaolin at approximately sso•c, therefore no reaction between the "crystal water" and carbon can be expected in the reaction with a sufficiently high gas flow.

Sulphur in carbon black and petroleum coke is not released until temperatures of about l6oo•c are reached (2). The sulphur is probably part of the carbon structure. Due to the reaction of the oxide with carbon. this structure is braken down and the following reaction will take place:

0

25 + C ~ _{cs 2 . ÖGT} ~ - 11422-6.49 T J.

With water or hydrogen, H2 s will be formed from cs 2 at high temperature (14).

For the impurities Fe, Ti, Ca, Mg, Na and K. the formation

of the elements, the carbides, the nitrides and the

sili-cides was investigated, as far as possible. Thermadynamie data was taken from Turkdogan, Grievesen and Ondracek (7,15,16).

In the figures 2.4 to 2.7, the change in Gibbs energy is shown as a function of the temperature for each element. According to these figures, titanium will be present as TiN, calcium as cao. magnesium as MgO and iron as Fes i x above approximately 122o•c. af ter the reaction. The

pre-sence of Fe 3c after the reaction is unlikely, because 6G0

T for a reaction like the formation of FeS i plus co

negative. Magnesium carbide is not stable at tem-peratures higher than approximately 6oo•c (14).

is

For the reactions of sodium and potassium. also the

for-~ation of suboxides and cyanides have to be considered. No thermadynamie data was found which could be used for the predietien of the reactions of sodium and potassium.

Ac-cording to Turkdogan (7) and Gmelin (17), sodium and po

-tassium will be present as metallic vapour during the reaction. So, they are expected to be blown out of the hot

2.5. Experimental.

2.5.1. The reactor.

As discussed in chapter 3 a packed bed reactor was chosen to study the carbothermal production of powders. As such a horizontal tube turnace was used. The reactor tube was· made of alumina (Degussa, 99.7 % Al₂

o

3 • impurities Ca and

Mg). Most experiments were carried out in a tube with a diameter of 25 mm. The length of the heated zone was 100 mm. The heating elements were outside of the reactor tube. The raw materials were mixed and pelletized and about 30 ml of pellets were packed between alumina fibers. In order to examine sealing up of the reactor and to produce larger quantities of powder, a number of experiments were per-formed in tube turnaces with an inner diameter of 60 mm ar 110 mm.

2.5.2. The production of pellets.

The raw materials were dried. Sametimes they were milled after drying. Then, the carbon and the dry oxide powders were mixed in a ball mill for at least 1 hour. The ball mill comprised a polyethene flask with porcelain balls. After the mixing process. the balls were removed and water ar a polyvinylalcohol salution (1\) was added. For each kilogram of powder, 250 g of water ar salution was added according to the formula:

1/x 1 + (1-c)ps

x is the _{weight fraction liquid, p1} is the density of the liquid, ps is the density of the solid and c is the porosity of the pellets (18). The dry powders and the

35-Fig. 2.8. Partiele size distribution of milled petroleum cokes and carbon black. Weight percent versus partiele size in microns.

100

{ 0

/o)

50

0

100

50

10

5

{}Jm}

A Milled in the hammer mill. ball milled for 72 hours.

B Milled in the hammer mill. ba 11 milled for 48 hours.

c

Milled in the hammer mill. screened over a 50 )A.m

screen.

D carbon black EHtex 675.

E carbon black Elft ex 125.

1

These analyses were carried out by use of a Malvern 2600/3600 pactiele sizer at Morgan Materials Technology Ltd .. Stourport-on-Severn and at the Eindhoven University of Technology.

water or solution were mixed in a polyethene flask overnight, the lumps were crushed in a mortar and mixed again overnight in the flask. The pellets of the ap-propriate size were screened and used for the experiment.

2.5.3. Milling the raw materials.

The Provins kaolin had to be milled after drying which· was done in a "hammermill" (Retsch GmbH SKl). The petroleum coke was also milled in the hammermill after drying. Then. it was either wet screened over a 50 ~m screen or milled for 48 or 72 hours in a hall mill with alumina balls (95~

alumina). The partiele size distribution of the petroleum coke and two types of carbon black after milling is given in fig. 2.8.

When the reacted pellets are taken form the reactor they may be considered as large agglomerates. So, they have to be milled. The Si 3Al 30 3N5 pellets were hall milled. The ball mill was made of a polyethene flask and alumina (95~

Al 2o 3 • 5\ Si02 ) were used. Mill wear was generally less than 2.5 wt\ increase of the batch. To compensate for mill wear the powder was produced with a slight nitrogen

excess. This is achieved by the use of a carbon black to kaolin ratio of 0.26 instead of the

theo-retica! 0.24 ratio. As a result some 15R phase is present in the powder. Instead of alumina halls it seems better to use siliconnitride balls. We also milled the pellets with pressureless sintered siliconcarbide balls. Milling was performed in an organic liquid. free of water. like propa-none. When water was used, ammonia was smelt after mil-ling. which indicates a reaction. After milling the powder was screened over a 63 ~ screen, filtered off and dried at l05°C in air.

-37-Table 2. 4. Calculated and measured xrd-pattern for Si₃Al_{30 3N5 .}

hk.l d(A) relative hkl d{A) relative

intensity intensity

calcu- me a- calcu-

mea-lated sured lated sured

LOO 6.66 )4 .1 36 320 1.527 18.0 110 ).84 35. 1 35 002 1.490 20.7 200 3.33 93.6 91 102 1.453 12.1 lOt 2.72 100 100 401 l . 453 12.1 210 2.52 94.1 92 410 l . 453 12.1 111 2.355 10.9 10 112 1.389 2.7 300 2.220 36.5 36 202 l . 360 66.0 201 2.220 36.5 36 321 1.360 66.0 211 1.922 13.9 12 500 1.332 5.0 220 l . 922 13.9 12 411 1.306 28.4 310 l . 84 6 13.7 14 212 l . 281 30.2 JOl 1.780 42.8 36 330 l . 281 30.2 400 1.664 0.2 221 l . 615 16.4 18 311 1.570 6.3 6

-2.5.4. Determination of the reaction rate and the reaction products.

Due to the reaction. more gas will leave the reactor than enters the reactor. For each molecule of N2 which reacts.

l molecules of

co

will enter the gas flow during the

formation of nitrides or oxynitrides. For the

for-mation of carbides. it is obvious tbat the amount of gas

which enters the reactor will be lower than the amount of

16 14 10 10 10 1 49 49 6 24 22 22

As only 100~ conversion of the free carbon is of interest for this work, the amount of free carbon was determined by heating the product after the reaction at 600°C in air. The weight loss measured was an indication of the amount of free carbon (19). The determination of free carbon was an extra check to see whether the thermal conductivity cell worked properly.

2.5.5. Powder characterisation.

All samples were examined with x-ray diffraction, using a Philips PW 1009 spectrometer, CU Ka-radiation and a Ni filter. For indentification the JCPDS system was used, 6-Si 3N4 (19-2132), a-Si 3N4 (9-250), Mullite (15-776), a-Al 2o 3 (10-173), y-Al 2o 3 (10-425), Sio2 (11-695), Si02

(5-0490), AlN (25-1133) and si 2N2o (18-1171). For the indentification of X-phase, polytypes and y-AlON a report by K.H. Jack was available (20). For Si 3Al 30 3N5 the JCPDS card 25-1492 is available. However, this card not only contains an error in the chemical formula, but the intensities are also far from comparable with those of 6-si 3N4 . Therefore the intensities were calculated on basis of the crystal structure (21), using data from Rietveld (22) and from Henry and Londsdale (23). Results are shown in table 2.4. As expected the calculated pattern closely agrees with the data for 6-si 3N4 and with the observed diffraction pattern of si 3Al 3o 3N5 . The presence of SiC as a second phase in the si 3Al 3o 3N5 powder was determined by the ratio of the peak area or the peak heights of the 200 and 210 reflections. The presence of SiC causes an increase in the intensity ratio

1(210)/1(200).

-39-As evident from figure 1.2 the composition (z-value) of a•-sialon can be obtained from the lattice constants. Besides XRD several techniques were used for chemica! ana-lysis. Raw materials and Si 3Al 3

o

3N5 powder were analyzed with x-ray fluorescence (JEOL 733 Superprobeat the Eindhoven University of Technology). The kaalins and the Si 3Al 3D3N5 powder after oxidation were also analyzed by means of partiele ihduced gamma emission (PIXE) using the

equipment in the Physics Faculty at the Eindhoven

University of Technology. _{Sintered Si 3Al 3}

o

3N5 samples were also analyzed with an electron microprobe.

The kaolins. the quartz and the oxidized si 3Al 30 3N5 powder were also analyzed by wet chemical analyzing techniques. The Al content was determined by AAS (24). Si was measured

as sio2 according to a gravimetrie procedure after HF treatment (25). The chemical analysis of the raw materials was also carried out by the NV Koninklijke Sphinx in Maastricht.

Oxygen analysis of the si 3Al 3o 3N5 powder were carried out by the so-called inert fusion methad (LECO. Ubach-over Warms). The amount of nitrogen of si 3Al 3

o

3N5 was found by melting the powder in LiOH followed by back titration of the ammonia formed according to the reaction:

si 3Al 3o 3N5 + lSLiOH ~ _5NH3 + _{Li 15si 3Al 3}

o

₁₈.

An alumina crucible was used to melt the powder in LiOH dnd nitrogen was used to remave the ammonia from the fused silica crucible in which the alumina crucible was placed.

Free _{Cdrbon in the Si 3Al 3}

o

3N5 powder was determined from iravimetric analysis. The powder was heated at

6oo•c

in air for 30 minutes. Under these circumstances no oxidation of the si 3Al 3D3N5 powder takes place. All weight loss had to be accounted for by free carbon. The resolution of the metbod is about 0.2 wt' of free carbon.

The physical aspects of the powder are of interest. namely the specific surface area. the partiele size distribution and the crystallite size of the powder.

The specific surface area of powders was measured ac-cording to the BET method by using a Ströhlein Areameter. The partiele size distribution of Si 3Al 30 3N5

pow-der was measured according to the sedimentation method with a Micromeritecs SediGraph. However, as the reacted pellets were in fact large agglomerates. they had to be milled before use. The influence of milling on the partiele size distribution of the powder is described in chapter 5. The crystallite size of the powder can be stu-died from the way that the peaks broadened in the X-ray diffractogram {26). The crystallite diameter t. can be calculated according to the formula:

t -_{- B cose} !L:..Q.L.y {nm)

where: 0.09 is a shape factor for spherical crystallites, e is the diffraction angle. y is the wave length of the x-ray radiation, and B is the peak broadening. Peak broadening can be caused by crystallite size. when the size is smaller than 100 nm. it may be caused by stresses within the crystallites. So, Equation 2.21 can only be used to calculate the crystallite size when there are no stresses within the crystallites and when the crystallite size is smaller than 100 nm. The peak broadening. B. is found by cernparing the peaks for the small crystallites with the peaks for coarse, larger than 100 nm, crystal-lites. This comparison is done with the aid of equation 2.22:

-41-B

where: B₅ is the FWHM radians on the 29 scale for the crystallites which are larger than 100 nm and Bm is the Sdme as B but for the smdll crystaliites. Their size

s

according to formula 2.30 should be similar to the size calculated from the specific surface area,

s.

according to the formula:

s

6/(p.t)

whece: p is the x-ray density of the material. Further physicdl characterisatlon of the powder is described in chapter 5.

2.6. Literature.

1. Kirk-Othmer. Encyclopedia of Chemical Technology. ~

and ~. J. Wiley

&

Sans Inc .• New York. (1976). 2. Materials and Technology. ~. Longman Group Ltd .•

London. (1971).

3. Lee J.G .• Cutler I.B .• Am. Ceram. Soc. Bull. ~. 869 (1979).

4. McCauley J.W .• Corbin N.D. in Progressin Nitrogen Ceramics. F.L. Riley ed .• Martinus Nijhoff Pu-blishers. Boston. 111 (1983).

5. Lee J .G .• Thesis, University of Utah. (1976). 6. Ullmanns Encyklopädie der Technischen Chemie. 10,

Verlag Chemie GmbH. Weinheim. (1975).

7. Turkdogan E.T., Physical Chemistry of High Tem

-perature Technology. Academie Press. New York. (1980).

8. Hendry A. in Nitrogen Ceramics. F.L. Riley ed .. Noordhoff. Leiden. 183 (1977).

9. Fegley M.B .• Comm. Am. Ceram. Soc .. ~. C-124 (1981). 10. MacDonald-Patience M .• Thesis, University of

New-castle upon Tyne. 1984.

11. Dörner P .• Gauck1er L.J .. Krieg H .• Lukas H.L .. Pet2ow G., Weiss J .. CALPHAD, ~. 241 (1979). 12. Kaufman L., CALPHAD, ~. 275 (1979).

13. Van Olphen H .• Fripiat J.J. eds .• Data handbaak for clay materials and other non-metallic minerals. Per -gaman Press. Oxford. (1979).

14. Remy H .• Lehrbuch der anorganischen Chemie. Aka-demische Verlag-geselischaft Geest & Portig K.-G .• Leip2ig, (1970).

15. Grieveson P. in Nitrogen Ceramics, F.L. Riley ed., Noordhoff. Leiden. 153 (1977).