EFFECT OF CRYSTALLINE PHASES ON THE STRENGTH OF (quartz-feldspar-kaolin )PORCELAIN

Nyongesa F.W. and Aduda B.O.Department of Physics, University of Nairobi, P. 0. Box 30197,

Nairobi, Kenya.

Abstract. The effect of mullite and quartz phases on the strength of Kenyan tri-axial porcelainin the system Quartz-feldspar-kaolin has been investigated. It was found that both thecompressive and tensile strengths of porcelain decreased with excess quartz content as a resultof circumferential cracks around quartz grains that may constitute fracture initiating flaws. Theinterlocking effect of mullite was found to be more beneficial to the tensile strength than to thecompressive strength of porcelain.

1. INTRODUCTION

The mechanical properties such as flexuralstrength, elastic modulus, etc. of porcelain arestrongly dependent on the material'smicrostructural features such as the amountsand type of phases present especially quartz,mullite and the pore phases [1,2,3]. The role ofmullite and quartz phases on the strength ofporcelain is however still a subject of controversy.

Mullite evolves from the reaction between clayminerals particularly kaolinite and illite and hastwo different evolution paths: primary andsecondary. The exact source and temperature forthe formation of these two types of mullitecontinue to be debated. Generally, thetransformation of metakaolin into a nonequilibrium unstable spinel-type structure (Al-Sispinel) results in the crystallization of mullite andhberalization of amorphous free silica (Si02)

according to the reactions shown in Equations(la)-(2b) [4,5].

3 (A 1203 .2Si02 )950 -10000 C(meta/caolin)

. 0.28~/g (AIl3.33. 612.66 )032 +6SiO~(r-alu DUD ium) (silica)

(la)

or3(A 1203 .2Si02) 950 -10000 C

(metakaolin)

0.56Sig(AI1O.67. 615.33)°32 +3Si02(alu minosilicatespinel)

O.,?1S2AIg(AIl3.33612.66)o32 ~ 107SoC

(lb)

(2a)JAh 03 .2Si02 + 45)°2

(mullite )

Of

(2b)3AI203 .2Si02 + 4Si02

(mu1/ite)

In the above equations, E9 represents a vacancyAn aluminosilicate spinel (O.562Sis (AllO.67.E95.33

032) and occasionally, a y-alumina-type phas(0.282Als(A113.33.E92("s)032) are the predictereaction products. The exact structure of tbspinel-phase continues to be controversial arxauthors present conflicting evidence regarding tbexistence of either phase [4].

Since mullite is stronger than glass, the proposehypotheses concerning the strength enhancemenof porcelain by mullite phase has been attributeto the interlocking network of mullite crystals thaform within the kaolinite grains and the wealglassy matrix [6]. On the other hand, the prestress hypothesis [7,8] attribute strength 0

porcelain to the compressive pre-stress imposed i.J

the glassy phase by the quartz phase durirucooling of the ceramic. Quartz has higheexpansion coefficient compared to that of tbglassy matrix and also undergoes abrupexpansion changes during phase inversions (u-t(~-quartz and vice versa). Our study was devoteeto studying the effect of both mullite and quartphases on the strength of Kenyan tri-axiaporcelain with a view to solving the controverssurrounding the contnbution of each phase on tbstrength of porcelain.

Proceedings of the 5th KPS Regional Workshop, 18-2ff' September 2000 5

2.EXPE~NTALPROCEDURE

2.1 Materials and fabrication of test samples

The porcelain raw materials (silica, feldspar andkaolin) used in the present study are the same asthose used in Kenyan ceramic industry and wereobtained from Athi River Mining CompanyLimited. Test samples with an optimizedcomposition of 20wt>/o silica and feldspar tokaolin ratio of5:8 were used. In a previous study[Nyongesa and Aduda, in press], it wasreported that highest sintered density andstrength of Kenyan triaxial porcelain in thequartz-feldspar-kaolin system were obtainedwith this compositions. All the compositionscontained 20wt% bentonite clay for goodhandling (plasticity) of the clay bodies.

To prepare the test samples, the raw materialpowders were sieved to pass through a 63 J.l.ITl

sieve aperture to discard coarse particles.' Thepowders were then mixed and made into a softmass byadding 33wt% water. After aging for.24hours, the soft mass was formed into cylindricaltest bars (I5.00cm ± O.Olcm in length and 1.50 ±O.Olem in diameter) using an extruder fabricatedin our laboratory. For velocity measurements,cylindrical test samples (5.00 ± O.Olem indiameter and 5.00cm ± O.Olcm in length) wereprepared using a piston and cylinder. A smallweight of 50N was attached onto the piston toensure a uniform forming (compaction) pressurefor all samples.

The samples were dried at room temperature andthereafter, oven dried at 11O°C for 8 hours at aheating rate of 10K min-I. The dried sampleswere then fired in an electric furnace from room.temperature to lOOO°Cat a rate of 5.0 K min-i.Above this temperature, they were heated at a rateof 2.0 K min-I and soaked for one hour at fixedtemperatures of 1000°C, 1050°C, 1100°C,1150°C, 1200°C and 1250°C respectively beforefurnace cooling.

2.2 Sample characterization

Identification and quantitative analysis ofcrystalline phases was done by X-ray diffraction

(XRD). A few chippings from test pieces wereground using a pestle and mortar, ground andsieved through a 100 mesh (150 urn) sieve. Thefine powder was then loaded into an X-ray Phillips1711 diffractometer. The quantity of crystallinephases was analyzed using Ni filtered CuI<..radiation with a PW i820/00 vertical goniometerand a PW 1710 microprocessor based control andmeasuring system. A scanning range (28) of 10-60° was used. The\ diffractograms weresmoothened.

The linear thermal expansion of firedrectangular samples of dimensions 10.0x4.0x35.0mm samples was measured in a high-temperature horizontal dilatometer (type 402E)with a heating rate of 5.O°CKi up to 1120°C.

The modulus of rupture (MOR) wasdetermined according to the standard procedure[9] and measurements were made using anInstron 1195 Universal testing machine. with across-head speed of 3 mm.min". The reportedvalues for the modulus of rupture is the mean ofat least 10 samples. The elastic modulus wasevaluated from the longitudinal velocity ofultrasound through the samples [10]. Prior tovelocity measurements, the samples werepolished to a smooth pair of parallel opposingsides with silicon carbide paper to 600 gritfinish. The longitudinal velocity was obtained,through the transmission method, by measuringthe transit time of ultrasonic pulse through thesample using a model Pundit instrument (CNSElectronics Ltd.). In this method, a test samplewas sandwiched between the transmitting andthe receiving transducers and coupled on eitherside using a thin layer of medium gradepetroleum based grease. The reportedlongitudinal velocity is the average from at least'five samples,

Selected samples were prepared for (SEM)analysis by cutting, polishing to 600 grit and thenetching with dilute hydrofluoric acid (IS vol-'%HF) for 60s to dissolve the glassy matrix andenhance the crystalline morphologies. Themicrostructure of the samples was observed usinga JEOL TSM T-330A scanning electronmicroscope at 10KV accelerating voltage. To

Proceedings of the 5th KPS Regional Workshop, 18-2dh September 2000 52

avoid charging effects from the non-conductive(electrical) ceramic surface, a thin gold coatingwas applied under low pressure argon gas using a'cool' Edwards S150B sputter coater. A layer of20-4Onm was sufficient to avoid the charging.problem.

3. RESULTS AND DISCUSSION

3.1 X-ray diffraction analysis

The X-ray diffraction patterns (28 versusrelative intensity) for porcelain samples sinteredto various temperatures are shown in Figure 1.The diffraction peaks show that the main mineralspresent in sintered porcelain were quartz (Si02),anorthoclase (potassium sodium aluminum silica;(Na.Kj-O. Al203.6Si02), albite (sodium aluminumsilica; Na2.Ab03.6Si02), mullite (3Al203.2Si02),o-alumina and illite (potassium aluminium silicatehydroxide; KAllSi~lO)(OHh).

Al- a.-elumina,M- rnnllite ,An- anorthoclase ,1- Illite, Q - Quam

AI

10 20 30 40 50 60

Diffraction angle Zq (degrees)Fig: 1. X-Ray (CuKa radiation) versus diffraction angle

2Ofor porcelain sample sintered at (a) JOOO'C,(b) 1150'C and (c) 1200't:.

From Figure 1, it is clear that sintering attemperatures from l050°C to I200°C produced aseries of different structures mainly characterizedby the decrease in the peak intensities of quartzand anorthoclase and the formation of somemullite. Very little mullite was present at lowsintering temperatures (lOOO°C). At 1150°C,

higher peak intensities of mullite were observedwhile at further higher temperatures of 1200°C, aninsignificant further change in the mullite peakintensities was noted. This mullite is thought toform from kaolinite. Its formation at lowertemperature than that predicted from phasediagrams (1547-1587'C) [11], could be attributedto the presence of fluxing oxides such as K20,Na20, CaO, and MgO in the porcelain rawmaterials used in this study [12] which lower themullite formation temperature [13].

The decrease in the peak intensities of quartz andanorthoclase with increase in sinteringtemperature indicates the decomposition ortransformation of these clay minerals to otherforms. Quartz transforms most probably intocristobalite and/or tridymite phases whileanorthoclase, being a flux, forms a liquid phase ata temperature of about 11OO°c.At 1150°C mostof the anorthoclase have decomposed.

3.2 Differential thermal analysis

The results of the linear thermal expansion ofporcelain are shown in Figure 2. The thermalexpansion curve of a dense (98%) alumina sampleis also shown in the same figure as a referencecurve.

Dense(98%) ~~ ~.:«>:--

poxcela.in

250' 500' 750" 1000'Temperature ("C)

Fig:2. Linear thermal expansion curves of porcelain anddense (98 %) alumina.

The initial shrinkage of porcelain at the onset offiring could be attributed to the liberation of waterof crystallization (dehydroxylation of the lattice

Proceedings of the 5th KPS Regional Workshop, 18-2rJh September 2000 53

water in the clay mineral's structure) and theburning out of organic matter present in the bodyincluding the oxidation of carbon to carbondioxide. The crystal structure of kaolinite containshydroxyl groups, and the dehydroxylation of thesegroups to form rnetakaolin (Al203.2Si02) occursat temperatures of about 400 - 550°C via thereaction [5]

Al203 .2Si02.2H 20 400 - 5500(kaolinite)

(3)AI203.2Si02+2H2 t

(metakaolin) (steam)

This dehydroxylation process is endothermic andis accompanied by a reorganization ofoctahedrally coordinated aluminum in kaolinite tomore tetrahedrally coordinated aluminium inmetakaolin.

The abrupt increase in the linear thermalexpansion in the porcelain between a temperatureof 550°C and 625°C, could be attributed to thetransformations of residual quartz (a- to B-quartztransformations). The rapid shrinkage at atemperature above 1000°C is attributed to onsetof liquid state sintering, the transformation ofmetakaolin into a spinel-type structure (Al-Sispinel), the crystallization of mullite and theliberalization of amorphous free silica (Si02)

according to Equations (la)-(2b) These reactionscould account for the observed decrease (fromXRD patterns shown in Figure 1 in the a -alumina content and an increase in the mullitecontent.

3.3 Young's modulus and modulus of rupture

Figure 3 shows the variation of the strength ofporcelain samples with mullite content. Themullite content in the samples was measured asthe height of the XRD peaks from thedifliactograms in figure 1. Both the Young'smodulus and the modulus of rupture increaseswith increasing mullite content and the Young'smodulus shows a higher dependence on mullitecontent compared to the modulus of rupture. Forexample, for an increase in the mullite content bya factor of 0.12, the Young's modulus increasesby a factor of 0.502 while the modulus of ruptureincreases by a factor of 0.456.

•

'.i)' 60r-- 1600 ~a so o 1500 ei 40 o c 400 ~~ 0 ~'8 30 ~ 300 ~~ 20 .------:--~- 200 '0'1l <) _ Young's Modulus ~..§' 10 <> - Modulw of :Rapture 100 "aO ~o>- llJ---.L.-----1---'---- ~

0.45 0.55 0.65Mulli te pe ak intensity

Fig:3. Variationofporcelain strength withmullite content

The above observation supports the mullitehypothesis and further indicates that mulliteincreases the stiffness of porcelain most probablyby its interlocking effect. Figure 4 shows thisphenomena .: Since mullite is stronger than glass,fracture will prefer the easiest path by avoiding themullite needles instead of crossing them. Mullitecrystals thus act as reinforcing agents in porcelain.

Notably, high sintering temperatures may leadto coarsening of mullite needles as shown infigure 5 leading to less interlocking effect. Thussintering temperature and generating thecorrect amount of properly sized mulliteneedles are vital in achieving strengthenhancement.

In figure 6, it is observed that both the Young'smodulus and the modulus of rupture decreaseswith increasing quartz content. In general, thisbehaviour can be attributed to the compressivepre-stresses imposed in the glassy phase due tothe difference in the expansion coefficientsbetween the glassy phase and the quartz alongwith the abrupt expansion changesaccompanying phase transitions (Figure 2).These compressive pre-stresses may introducecircumferential cracks around the quartzparticles (Figure 7) which on acting alone orcombined with large pores may constitutefracture- initiating flaws and thus weaken theporcelain [7].

Proceedings of the 5th KPS Regional Workshop. IB-2(/' September 2000 54

(a)

(b)

(c)

Fig:4. SEJ.,f micrograph showing formation of mullitecrystals (a) sintered at l05ifC showing mainlyplate-like kaolin particles, (b) at 1J5(fC showingneedle shaped muilite needles (c) at 115(fC,showing interlocking mullite needles.

Fig: 5. SEM micrograph of porcelain sample fired at120(fC showing coarsening of mullite needles

«S'60 600 'Ci'~~50 c

500~..•..•.

~40 •400 ~0

•300 "

'83Ot-3:Ii:: ....

.~20 IQ - Youog's Modulus I 200 0§1O ~• - Modulus of Rapture 100 :3

e 130

10.50 ::=:9.5 11.5

QU8l'12 peakin~nsity

Fig.6: Variation of strength of porcelain with quartzcontent.

Fig: 7. SEM micrograph showing circumferentialcracks around quartz grains

Unlike in figure 3, in figure 6 Young's modulusshows a greater rate of decrease with increasingquartz content than the modulus of rupture

Proceedings of the 5Ut KPS Regional Workshop, 18-2r1' September 2000 55

----------

indicating that the effect of compressive pre-stresses on tensile strength is more significant.than on the modulus of rupture. Compressivepre-stresses may act as barriers to thepropagation of flaws thus improvingcompressive strength (modulus of rupture).

4. CONCLUSIONS

In this study it was found that:1. Properly sized mullite needles that are vital for

strength enhancement were generated at atemperature of 1150°C.

2. Mullite formation was more beneficial totensile strength than compressive strength

3. Whereas compressive pre-stresses set up bythe quartz grains may in general weaken theporcelain, these stresses may on the contrary,strengthen porcelain by acting as barriers forthe propagation of flaws.

REFERENCES

1. Pickup P., 1997. Effect of Porosity on Young'sModulus of a Porcelain, British ceramictransactions, % [3], pp.96-98.

t Kobayashi Y, Osamu 0, Ohashi Y, and KatoE.,1994 Composition for Strengthening PorcelainBodies in Alumina-Feldspar-Kaolin systems,British Ceramic Transactions 93(2), pp.49-52.

3. Coble RL. and Kingery W. D., 1956. Effect ofPorosity on Physical Properties of SinteredAlumina, Journal of American CeramicSOCiety, 39 [11],377-85.

5. Sonuparlak R, Sarikaya M and Aksay L., 1987.Spinel Phase Formation During the 980DCExothermic Reaction in the Kaolinite -to-MulliteReaction Series. Journal of the American CeramicSociety, 70[11], pp.831-42.·

~. Chakraborty A K and Ghosh D. H., 1977.Reexamination of the Kaolinite-to-MulliteReaction Series, Journal of the Am-erican CeramicSociety, 61[3-4], pp.l70-173.

). Manyasovszky-Zsolnay, L., 1957. MechanicalStrength of Porcelain, Journal of AmericanCeramic Society, 40[9], pp.299-306.

ACKN"OWLEDGEMENT§

The authors wish to thank DAAD and theUniversity of Nairobi for providing the grants tocarry out the study and they also sincerelyacknowledge the assistance of Mr. Chimtawi ofthe International Centre for Insect Physiology andEcology (ICIPE) with taking the SEMmicrographs.

7. Ohya Y and Takahashi Y, 1999. Acoustic Emissionfrom a Porcelain Body during Cooling, Journal ofAmerican Ceramic Society, 82[2], pp.445-48.

8 ZuokaiK, 1990. Fracture-Initiating flaws inAluminous Electrical Porcelain, CeramicBulletin, 69[3], pp.380-390.

9. Japanese Industrial Standards., lIS. R2213, Tokyo. 1-4,1987.

10. Nyongesa F.W. and Aduda RO., 1999. UltrasonicAttenuation in Kenyan Clay Refractories, BritishCeramic Transactions, [8], pp.l-o.

11. Levin E. M, Robbins C.R and Me Murdie H.F., 1979.Phase Diagrams for Ceramists, Edited by RESERMK, American Ceramic Society, 1,pp.156-158.

12. NyongesaF.w. and AdudaB.n, Effect of SilicaAdditions on Strength and Elastic modulus ofKenyan Industrial Clay Ceramics, East AfricanJournal of Science, in press.

13. Papargyris A D. and Cooke RD., 19%. Structure andChemical Properties of Kaolin Based Ceramics,British ceramic transactions, 95[3], pp.107-120.

"roceedings of the 5th KPS Regional Workshop, JB-2rJh September 2000 56