Applied Clay Science 20 Ž2001.

73–80
www.elsevier.comrlocaterclay

Thermal treatment of kaolin: the effect of mineralogy on the

pozzolanic activity
G. Kakali a , T. Perraki b, S. Tsivilis a,) , E. Badogiannis a
a
Chemical Engineering Department, Laboratories of Inorganic and Analytical Chemistry, National Technical UniÕersity of Athens,
9 Heroon Polytechniou St., 15773 Athens, Greece
b
Min. Engineering Department, National Technical UniÕersity of Athens, 9 Heroon Polytechniou St., 15773 Athens, Greece
Received 9 June 2000; received in revised form 15 January 2001; accepted 5 February 2001

Abstract

This paper reports an investigation on the effect of mineralogy on the pozzolanic activity of fired kaolin. Representative
samples of Greek kaolin ŽMilos Island. and a commercial kaolin of high purity are studied. The samples are tested by X-ray
diffraction ŽXRD., Differential Thermal Analysis ŽDTA. and Infrared ŽIR. Spectroscopy in order to determine their
mineralogical composition and structural differences. Calcination of samples is carried out at 6508C for 3 h. The
decomposition of kaolinite and alunite is recorded using methods of thermal analysis. The resultant products are identified
by XRD. The reactivity of the thermally treated samples is evaluated based on Chapelle test. It is concluded that the
pozzolanic activity of metakaolinite is strongly related to the crystallinity of the original kaolinite. Well-ordered kaolinite is
transformed into less reactive metakaolinite. q 2001 Elsevier Science B.V. All rights reserved.

Keywords: Kaolin; Mineralogy; Calcination; Metakaolinite; Pozzolanic activity

1. Introduction Si–O network remains largely intact and the Al–O

network reorganizes itself. While kaolinite is crys-
Fired clays, ground and mixed with lime, have talline, metakaolinite has a highly disordered struc-
been one of the first developed structural materials. ture and offers good properties as mineral additive.
Nowadays, fired clays are still widely discussed in Metakaolinite reacts particularly well with lime and
cement literature for their technical properties. forms in the presence of water hydrate compounds of
It is well known that during calcination Ž450– Ca and Al silicates. Therefore, it is considered to be
6008C., kaolinite loses the OH lattice water ŽGrim, a good synthetic pozzolana. The development of
1968. and is transformed into metakaolinite, a mate- pozzolanic properties in fired clays mainly depends
rial with some degree of order. In metakaolinite, the on the nature and abundance of clay minerals in the
raw material, on the calcination conditions and on
the fineness of the final product ŽSurana and Joshi,
)
Corresponding author. Tel.: q30-1-7723262; fax: q30-1-
1990; Dunster et al., 1993; He et al., 1995; Gruber
7723188. and Sarkar, 1996; Kostuch et al., 1996; Wild et al.,
E-mail address: stsiv@central.ntua.gr ŽS. Tsivilis.. 1996; Stroeven and Dau, 1999..

0169-1317r01r$ - see front matter q 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 9 - 1 3 1 7 Ž 0 1 . 0 0 0 4 0 - 0
74 G. Kakali et al.r Applied Clay Science 20 (2001) 73–80

In this work, the mineralogy of four representa- remained for 3 h. The optimum conditions of thermal
tive Greek kaolins and one commercial product is treatment were determined in a previous work
studied. Special attention is paid to the effect of ŽKaloumenou et al., 1999.. The sample with the
kaolinite structure on the pozzolanic reactivity of the higher percentage of alunite ŽK3., was also heated at
relative metakaolinite. Instrumental techniques, such 5508C, 6508C, 7508C, 8508C and 9508C in order to
as X-ray diffraction, Differential Thermal Analysis observe the intermediate products of kaolinite and
and Infrared Spectroscopy, are used for the charac- alunite decomposition.
terization of raw materials. The pozzolanic activity is The pozzolanic activity of thermally treated
evaluated on the basis of Chapelle test. This work Ž6508C, 3 h. samples was measured according to
belongs to a research project aiming at the exploita- Chapelle test ŽLargent, 1978; Kostuch et al., 1996..
tion of Greek kaolins in concrete technology. A gram of metakaolinite is mixed with 1 g of
CaŽOH. 2 and 100 ml of boiling water. The suspen-
sion is boiled for 16 h and the free CaŽOH. 2 is
2. Experimental determined by means of sucrose extraction and titra-
tion with a HCl solution.
2.1. Materials
2.3. Instrumental techniques
Four Greek kaolins, having varying chemical and
mineralogical composition, are examined. K1 and Mineralogical analyses in the raw and thermally
K2 are typical samples with an average kaolinite treated samples were carried out by X-ray diffraction
ŽXRD. using a Siemens D5000 diffractometer ŽCu
content and low SO 3 content. K3 has an average
kaolinite but high SO 3 content while K4 has the K a radiation, Ni Filter..
highest and the lowest content of kaolinite and alu- An infrared ŽIR. spectrophotometer, Perkin Elmer
nite, respectively. Kaolins having similar composi- 880, was also used in order to confirm the identifica-
tion are widely used in Greek industry. In addition, a tion of the constituents in raw and treated samples.
commercial sample K–C, a kaolin of high purity, is IR spectroscopy is generally used in order to give
also examined. Table 1 presents the chemical com- information as far as the composition of a sample, its
position of the samples. structure and its characteristic bonds are concerned.
In addition, IR spectroscopy is applied on clay min-
2.2. Procedure erals in order to study the nature of isomorphic
substitutions of cations in octahedral and tetrahedral
The thermal treatment of the kaolins was carried sheet of the lattice ŽStubican and Rustum, 1961. and
out in a laboratory programmable furnace. The sam- the degree of crystallinity.
ples Ž m s 140 g. were thermally treated at a constant Thermogravimetric analysis ŽTGA. and Differen-
rate Ž108Crmin. from ambient to 6508C, where they tial Thermal Analysis ŽDTA. were applied in order
to observe the reactions taking place during the
thermal treatment of the samples. A Mettler Toledo
TGArSDTA 851 instrument was used. Powdered
Table 1
samples were heated from ambient to 12008C at a
Chemical composition of kaolins Ž% wrw.
rate of 108Crmin under static atmospheric condi-
K1 K2 K3 K4 K–C
tions. a-Al 2 O 3 was used as reference material.
SiO 2 73.45 72.47 38.92 65.92 47.85
Al 2 O 3 18.04 18.40 35.38 22.56 38.20
CaO 0.40 0.35 0.54 0.36 0.32
3. Results and discussion
MgO 0.03 0.03 0.06 – –
Fe 2 O 3 – – 0.60 – 0.30
K 2O 0.80 0.80 2.51 0.57 0.27 3.1. Mineralogy
L.O.I. 8.10 8.00 21.50 8.60 12.30
SO 3 3.00 3.12 10.03 2.00 – Fig. 1 presents the XRD patterns of the raw
samples. According to the relative patterns, the sam-
 G. Kakali et al.r Applied Clay Science 20 (2001) 73–80 75

Fig. 1. XRD patterns of raw samples Ž1: kaolinite, 2: K-alunite, 3: cristobalite, 4: quartz, 5: illite..

ples mainly consist of kaolinite and K-alunite and Tucker, 1988. in combination with the bulk
ŽKAl 3 ŽSO4 . 2 ŽOH. 6 .. They also contain quartz and chemical analysis of the samples ŽTable 1..
cristobalite. In addition, the sample K–C contains The determination of alunite is based on the SO 3
detectable amount of illite, while little amount of content of the samples, since there is no other de-
illite is also present in sample K4. tectable mineral which may contain S. The K 2 O
The crystallinity of kaolinite in the samples can content, after extracting the K 2 O in alunite, is used
be evaluated on the basis of the XRD background in for the determination of illite since there is no other
the range 2 u s 20–308 ŽHe et al., 1994. and the K-containing mineral detected in the samples. The
width of the Ž002. diffraction peak Ž d s 3.58 A ˚ . at average content of K 2 O in illite and alunite is con-
half the maximum height ŽBrindley and Brown, sidered to be 9% ŽGrim, 1968; Kelepertsis et al.,
1984.. As observed, the sample K3 contains well- 1990. and 11.4%, respectively.
crystallized kaolinite Žbackground: 98.34, width:
0.1798. while kaolinite in the other samples are less
ordered Žbackground: 122.56 – 153.92, width:
Table 2
0.2059–0.3861.. This is also confirmed by the obser-
Mineralogical composition of kaolins Ž% wrw.
vation of the Ž131. diffraction peak Ž d s 2.29 A ˚ .,
K1 K2 K3 K4 K–C
which is related to the number of defects along c
axis ŽCuinier, 1956.. Kaolinite 38 39 65 52 96
Alunite 7 7 22 5 –
The semi-quantitative mineralogical estimation is QuartzqCristobalite 55 54 8 41 –
presented in Table 2. The estimation is based on the Illite – – – – 3
characteristic XRD peaks of each mineral ŽHardy
76 G. Kakali et al.r Applied Clay Science 20 (2001) 73–80

SiO 2 , which is not bounded by the kaolinite and 1975.. The intensities of these bands are decreased
illite, is found as quartz and cristobalite. The ratio of according to the alunite content of the samples in the
quartz to cristobalite in the studied natural samples order K3 ? K2 ) K1 ) K4, while no band is ob-
follows the order K3 ? K1 ) K2 ) K4: quartz is served in sample K–C.
the main SiO 2 mineral in K3 while cristobalite is the
main SiO 2 mineral in K1, K2 and K4. 3.2. Thermal decomposition of kaolins
Fig. 2 presents the IR spectra of untreated sam-
ples. The characteristic bands of kaolinite Ž; 3690, Fig. 3 presents the DTA curves of the studied
3620, 1100, 1032, 1008, 913, 694, 539, 471 and 430 samples. The main changes pointed out by TG and
cmy1 . show wide variations in the intensities, espe- DTA, during the heating of the samples are the
cially at the higher wave numbers ŽVan der Marel, following.
1976.. Bands of large intensities at this region 1. T - 1008C: low temperature release of ab-
Ž3690–3620 cmy1 . correspond to well-crystallized sorbed water in pores, on the surfaces, etc.
kaolinite ŽK3. while the weakest ones are for the less 2. ; 100–4008C: weight loss that can be corre-
ordered samples. The bands at 3485 and 475 cmy1 lated with a pre-dehydration process, which takes
are associated with alunite ŽOlphen Van and Fripiat, place as a result of the reorganization in the octahe-

Fig. 2. IR spectra of raw samples.

 G. Kakali et al.r Applied Clay Science 20 (2001) 73–80 77

Fig. 3. DTA curves of the samples.

dral layer, first occurring at the OH of the surface 5. ; 10008C: formation of mullite, being indi-
ŽBalek and Murat, 1996.. cated by an exothermic peak. The crystallinity of the
3. ; 400–6508C: dehydroxylation of kaolinite and samples can be evaluated on the following TG-DTA
formation of metakaolinite according to the reaction: data:
Al 2 Si 2 O5 Ž OH . 4™Al 2 Si 2 O5 Ž OH . x O 2yx
Ži. The weight loss in stage 2. The greater the
q Ž 2 y xr2 . H 2 O weight loss, the less the crystallinity of the
with a low value of x Žabout 10% of residual samples ŽSmykatz-Kloss, 1974; Worral, 1975..
hydroxyl groups in metakaolinite.. Žii. The starting temperature of kaolinite dehy-
4. ; 500–9008C: decomposition of alunite in two droxylation. In well-crystallized samples the de-
stages according to the reactions: composition of kaolinite starts at higher temper-
ature ŽMurat, 1983..
2KAl 3 Ž SO4 . 2 Ž OH . 6™2KAl Ž SO4 . 2 q 2Al 2 O 3
Žiii. The slope of the DTA curve corresponding
q 6H 2 O Ž 480–6208C . to the dehydroxylation process. Decomposition
2KAl Ž SO4 . 2™K 2 SO4 q Al 2 O 3 DTA curves are generally sharper Žgreater abso-
lute value of slope. in well-crystallized samples.
q 3SO 3 Ž 770–9008C.
The two stages of alunite decomposition are Table 3 presents the above data for the studied
clearly recorded in TG-DTG curves but they cannot samples. According to Table 3, K3 has the greater
be distinguished on the basis of DTA curves. In Fig. crystallinity of all samples. K1, K2, K4 and K–C are
3, we can see the increase of the peak in the range less ordered, without considerable differences among
550–6008C Žsample K3. where there is an overlap-
ping of kaolinite and alunite decomposition. Accord-
ing to these reactions, alunite losses 13% of its Table 3
weight during the first step and 29% during the DTA-TG data related to the crystallinity of the samples
second step, which corresponds to a 75% release of TG-DTA data K1 K2 K3 K4 K–C
sulfates. According to the literature, 100% of sulfates Weight loss of 1.05 1.18 0.58 1.46 0.74
are removed when alunite is mixed with kaolinite pre-dehydration Ž%.
and quartz ŽPiga, 1995.. However, our study does Initial temp. of 403 416 422 406 403
dehydroxylation Ž8C.
not confirm this, since there was, still, detectable Slope of DTA y0.168 y0.186 y0.258 y0.156 y0.186
sulfate content in sample K3, after thermal treatment curve
at 9508C.
78 G. Kakali et al.r Applied Clay Science 20 (2001) 73–80

them. The size of the exothermic peak in the range and K 2 SO4 at 7508C, mullite, g-Al 2 O 3 and K 2 SO4
900–10008C, which is connected to the formation of at 9508C. This indicates that alunite is firstly decom-
mullite, also confirms this. In well-crystallized kaoli- posed into KAlŽSO4 . 2 , and then into K and Al
nite, greater amount of mullite is formed ŽWorral, sulfates. The latter is decomposed at 7708C, accord-
1975.. ing to the literature, and the released Al 2 O 3 , in the
The above remarks have also been confirmed by form of g-Al 2 O 3 , is gradually incorporated in mul-
XRD and IR observations. lite.
Fig. 5 presents the IR spectra of the sample K3
3.3. Characterization of the thermally treated kaolins after thermal treatment at 8508C and 9508C, as well
as the IR spectrum of the untreated sample. After the
Fig. 4 presents the XRD patterns of the sample thermal treatment of sample, the transformation of
K3 after thermal treatment at 5508C, 6508C, 7508C, kaolinite to metakaolinite is confirmed by the ab-
8508C and 9508C. As it is observed, the background sence of detectable Al–O–H bands at 913 cmy1 .
in the range 2 u s 20–308 is increased by 130% at The reduction of band at 539 and 913 cmy1 and the
6508C and by 78% at 9508C, compared to the un- appearance of a new band at 800 cmy1 can be
treated sample. This indicates that the disordered connected with the change from octahedral coordina-
metakaolinite which has been formed at 6508C is tion of Al 3q in kaolinite to tetrahedral coordination
gradually converted into more ordered forms at higher in metakaolinite. The bands at 1100 and 1200 cmy1
temperatures. are assigned to amorphous SiO 2 . The characteristic
Alunite and kaolinite are still present at 5508C but bands of alunite have totally been disappeared after
they are almost disappeared at 6508C. The identified the thermal treatment over 7508C. A new band at
products of the alunite decomposition are: KAlŽSO4 . 2 2800–3000 cmy1 is observed in the sample treated
and little Al 2 ŽSO4 . 3 at 6508C, KAlŽSO4 . 2 , Al 2 ŽSO4 . 3 at 8508C and is reduced as the treatment temperature

Fig. 4. XRD patterns of kaolin K3, treated at different temperatures Ž1: KAlŽSO4 . 2 , 2: Al 2 ŽSO4 . 3 , 3: mullite, 4: g-Al 2 O 3 , 5: kaolinite, 6:
quartz, 7: K-Alunite, 8:K 2 SO4 ..
 G. Kakali et al.r Applied Clay Science 20 (2001) 73–80 79

Fig. 5. IR spectra of kaolin K3 Ž1: raw sample, 2: thermal treatment at 8508C, 3: thermal treatment at 9508C..

is increased. This band is probably due to K–Al evaluate the reactivity of metakaolinite. As shown,
sulfate whose formation and decomposition in this although the commercial product K–C seems to be
temperature range has been also confirmed by XRD. the more reactive one, when its reactivity is ex-
pressed on the basis of metakaolinite, it is consider-
3.4. Pozzolanic actiÕity of treated samples ably reduced due to the elevated metakaolinite con-
tent. Comparing the samples K1–K4, it is observed
Table 4 presents the amount of CaŽOH. 2 con- that the less ordered kaolinites exhibit the higher
sumed per gram of thermally treated kaolin Žmeta- reactivity while K3, which has been characterized as
kaolin. during the pozzolanic reaction ŽChapelle test.. well ordered, is the less reactive.
It must be noticed that the content of metakaolinite,
the only pozzolanic constituent, is not constant in all
samples. Therefore, the consumed CaŽOH. 2 is also
expressed per gram of metakaolinite in order to 4. Conclusions

The following conclusions can be drawn from the

Table 4 present study.
Reacted CaŽOH. 2 per gram of metakaolin and metakaolinite The pozzolanic activity of metakaolinite is
Sample CaŽOH. 2 CaŽOH. 2 strongly related to the crystallinity of the original
code Žgrg of metakaolin. Žgrg of metakaolinite.
kaolinite. Well ordered kaolinite is transformed in
K1 0.759 2.19 less reactive metakaolinite.
K2 0.704 1.98 The crystallinity of kaolinite can be evaluated on
K3 0.741 1.18
K4 0.815 1.70
the basis by X-ray diffraction, Differential Thermal
K–C 0.833 0.91 Analysis and Infrared Spectroscopy studies of raw
kaolin.
80 G. Kakali et al.r Applied Clay Science 20 (2001) 73–80

Acknowledgements Mineral and chemical compositions of kaolins from Milos

lsland, Greece—procedure of kaolinite enrichment. Appl. Clay
Sci. 5, 277–293.
The authors gratefully acknowledge Titan Cement Kostuch, J.A., Walters, V., Jones, T.R., 1996. High performance
for its help with the evaluation of the poor quality concretes incorporating metakaolin: a review. In: Dhir, R.K.,
Greek Kaolins and Mrs. K. Choupa for her contribu- Jones, M.R. ŽEds.., Concrete 2000: Economic and Durable
Construction Through Excellence. E&FN SPON, London, pp.
tion in the Chapelle tests.
1799–1811.
Largent, R., 1978. Estimation de l’ activite pouzzolanique. Bull.
Liasons Lab. Pont Chausees 93, 61–65.
Murat, M., 1983. Hydration reaction and hardening of calcined
References clays and related minerals: II. Influence of mineralogical
properties of the raw-kaolinite on the reactivity of metakaolin-
Balek, V., Murat, M., 1996. The emanation thermal analysis of ite. Cem. Concr. Res. 13 Ž2., 511–518.
kaolinite clay minerals. Thermochim. Acta 282–283, 385–397. Olphen Van, H., Fripiatb, J.J., 1975. Data Handbook for Clay
Brindley, G., Brown, G., 1984. Crystal Structures of Clay Miner- Materials and Other Non-metallic Minerals. 1st edn. Perga-
als and their X-ray Identification. The Mineralogical Society, mon, London.
London. Piga, L., 1995. Thermogravimetry of a kaolinite–alunite ore.
Cuinier, A., 1956. Theorie et Technique de la Radiocristallogra- Thermochim. Acta 265, 177–187.
phie. Dunod, Paris. Smykatz-Kloss, W., 1974. Differential Thermal Analysis: applica-
Dunster, A.M., Parsonage, J.R., Thomas, M.J.K., 1993. Poz- tions and results in mineralogy. Springer-Verlag, Berlin.
zolanic reaction of metakaolinite and its effects on Portland Stroeven, P., Dau, P., 1999. Effect of blending with kaolin or
cement hydration. J. Mater. Sci. 28, 1345–1350. diatomite on characteristics of Portland cement paste and
Grim, R., 1968. Clay Mineralogy. McGraw-Hill, New York. mortar. In: Dhir, R.K., Dyer, T.D. ŽEds.., Modern Concrete
Gruber, K.A., Sarkar, S.L., 1996. Exploring the pozzolanic activ- Materials: Binders, Additions and Admixtures. Thomas
ity of high reactivity metakaolin. World Cem. 27 Ž2., 78–80. Telford, London, pp. 139–149.
Hardy, R., Tucker, M., 1988. X-ray diffraction. Techniques in Stubican, V., Rustum, R., 1961. Isomorphous substitution and
Sedimentology. Blackwell, Oxford. Infra-red spectra of the layer lattice silicates. Am. Mineral. 46,
He, C., Macovicky, E., Osbaeck, B., 1994. Thermal stability and 32–51.
pozzolanic activity of calcined kaolin. Appl. Clay Sci. 9, Surana, M., Joshi, S.N., 1990. Estimating reactivity of pozzolanic
165–187. materials by a spectrophotometric method. Adv. Cem. Res. 3
He, C., Osbaeck, B., Macovicky, E., 1995. Pozzolanic reactions of Ž10., 81–83.
six principal clay minerals: activation, reactivity assessments Van der Marel, H., 1976. Atlas of Infrared Spectroscopy of Clay
and technological effects. Cem. Concr. Res. 25 Ž8., 1691– Minerals and Their Admixtures. Elsevier, Amsterdam.
1702. Wild, S., Khatib, J.M., Jones, A., 1996. Relative strength, poz-
Kaloumenou, M., Badogiannis, E., Tsivilis, S., Kakali, G., 1999. zolanic activity and cement hydration in superplasticised con-
The effect of kaolinite particle size on its conversion to crete. Cem. Concr. Res. 26 Ž10., 1537–1544.
metakaolinite. J. Thermal Anal. 56, 901–907. Worral, W., 1975. Clays and Ceramic Raw Materials. Applied
Kelepertsis, A., Economou, K., Skounakis, S., Porfyris, S., 1990. Science Publ., London.