Minerals Engineering 16 (2003) 205–210

This article is also available online at:

www.elsevier.com/locate/mineng

Fire-resistant geopolymer produced by granulated blast furnace slag

T.W. Cheng *, J.P. Chiu
Department of Materials and Mineral Resources Engineering, National Taipei University of Technology, Taipei 106, Taiwan, ROC
Received 8 September 2002; accepted 18 December 2002

Abstract
This paper describes research into the use of granulated blast furnace slag as an active filler in the making of geopolymers. During
this work it was found that geopolymer setting time correlates well with temperature, potassium hydroxide concentration, meta-
kaolinite and sodium silicate addition. The physical and mechanical properties of the geopolymer also correlated well with the
concentration of alkaline solution and the amount of metakaolinite that is added. The highest compressive strength achieved was 79
MPa. For fire resistance tests, a 10 mm thick geopolymer panel was exposed to a 1100 °C flame, with the measured reverse-side
temperatures reaching less than 350 °C after 35 min. The products can be fabricated for construction purposes and have great
potential for engineering applications.
Ó 2003 Elsevier Science Ltd. All rights reserved.

Keywords: Environmental; Recycling; Waste processing

1. Introduction Deventer, 1999, 2000). Unlike conventional organic

polymers, glass, ceramic, or cement, geopolymers are
Blast furnace slag is formed in processes such as pig non-combustible, heat-resistant, formed at low temper-
iron manufacture from iron ore, combustion residue of atures, and fire/acid resistant. Three sources are needed
coke, and fluxes such as limestone or serpentine, and for geopolymer synthesis: raw materials, inactive filler,
other materials. If the molten slag is fast-cooled by high- and geopolymer liquor (Ikeda, 1998). Raw materials can
pressure water, a vitreous Ca–Al–Mg silicate fine grain be industrial wastes, such as fly ash, blast furnace slag,
glass is formed. Generally, the way to utilize granulated red mud, waste glasses, or some natural minerals and
blast furnace slag is in partially replacing Portland ce- rocks. Inactive filler, mainly supporting Alþ3 ions, can
ment. There are at least four million tons/year of gran- be kaolinite or metakaolinite. Geopolymer liquor in-
ulated blast furnace slag used in Taiwan. Granulated cludes a sodium silicate solution acting as binder, and
blast furnace slag is a non-toxic material, and can be a alkali hydroxide solution for the dissolution of raw
good raw material for making high-value geopolymers materials. The chemical process to form geopolymers
which can be utilized in fire resistant applications. involves two steps: (1) dissolution of raw materials in
Geopolymers are kinds of inorganic polymers that alkaline solution to form Si and Al gel on the materialÕs
have been gradually attracting world attention as po- surface, (2) polycondensation to form networked poly-
tentially revolutionary materials. Similar to natural meric oxide structures. However, the exact mechanism
zeolite minerals, geopolymer is a class of three-dimen- of geopolymer setting and how hardening occurs is not
sionally networked alumino-silicate materials, and first fully understood. The aim of this research is to try to
developed by Joseph Davidovits in 1978 (Comrie and fabricate a granulated blast furnace slag-based geo-
Davidovits, 1988; Davidovits et al., 1990). According to polymer for fire-resistance purpose and to better un-
previous research, a wide range of natural Al–Si min- derstand the mechanism of geopolymerisation.
erals, wastes, and slags could serve as potential source
materials for the synthesis of geopolymers (Davidovits, 2. Experimental methods
1999a; Van Jaarsveld et al., 1997, 1999; Xu and Van
2.1. Materials
*
Corresponding author. Tel.: +886-2-2771-2171; fax: +886-2-2731-
7185. Sodium silicate (weight ratio SiO2 :Na2 O ¼ 3:3, wt.%
E-mail address: [email protected] (T.W. Cheng). Na2 O ¼ 6–7% and SiO2 ¼ 23–25%) and KOH (analytical
0892-6875/03/$ - see front matter Ó 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0892-6875(03)00008-6
206 T.W. Cheng, J.P. Chiu / Minerals Engineering 16 (2003) 205–210

grade) were obtained from L.K. Chemical Co. Ltd., chanical property measurements. Table 2 summarizes
Taiwan. Granulated blast furnace slag samples were the compositions of the geopolymers under discussion.
collected from China Steel in Kaoshiung, Taiwan. Table The geopolymers setting time measurements were
1 shows the chemical composition of the slag sample. conducted by Vicat needle (refer to CNS 786 and ASTM
After being ground to )200 mesh using a roll crusher, C191-01 standard test methods). For SEM and energy
the particle size of the slag ranged from 11.5 to 103.4 lm dispersive (EDS) examinations, a Hitachi S-4100 scan-
with D50 of 44.6 lm. Kaolinite was obtained from the ning electron microscope, linked with an EDS attach-
Standard Reagent Co. Ltd., USA. Metakaolinite was ment, was used to examine the geopolymers. The testing
prepared by calcining the above kaolinite at 750 °C for methods and formulas for physical property tests such
6 h as described by Davidovits (1999a). The particle size as water absorption rate and volumetric density were
of the metakaolinite ranged from 2.7 to 19.9 lm with evaluated according to the Archimedes method. Com-
D50 of 6.3 lm. Chemical compositions of the meta- pressive strength tests were carried out after 1, 7, 14, and
kaolinite are also shown in Table 1. All experiments 28 days by a Shimadzu UEH-30 compressive strength
were carried out using the same batches of starting analysis apparatus. For fire resistant tests, a 10 mm
materials and reagents. thick geopolymer panel was exposed to an 1100 °C
flame. The reverse-side temperature of the panel was
2.2. Methods measured and recorded (Davidovits, 1999b).

A mechanical mixer firstly mixed metakaolinite and

KOH for 10 min. Sodium silicate solution and ground 3. Results and discussion
blast furnace slag were then added. Following a further
5 min of mixing, the samples were cast in 50-mm cube 3.1. Setting time measurements
molds and vibrated for 5 min in order to release bubbles.
The molds were sealed and set at 60 °C for 3 h in a A Vicat needle was used to measure the geopolymer
sample-drying oven. After being removed from the setting time. By recording the time of a 1 mm needle
moulds, the samples were kept at room temperature for penetration in the softening specimen after pouring the
another 21 h, before the beginning of physical and me- paste into a mould, the time of a penetration of 25 mm
was determined. This is the initial setting time. The final
setting time is when the needle does not sink visibly into
Table 1 the paste. Fig. 1 shows the effects of KOH concentration
Chemical composition of granulated blast furnace slag and meta- on the setting time at room temperature. It can be seen
kaolinite that setting time is a function of KOH concentration.
Chemical composition Granulated blast furnace Metakaolinite The higher the KOH concentration added, the longer
wt.% slag was the setting time. The initial setting time and final
SiO2 34.39 52.26 setting time for K1, K2, K3 samples were 35, 45, 89 and
Al2 O3 14.47 42.83 63, 80, 142 min, respectively. Fig. 2 shows the effect of
Fe2 O3 0.63 1.01
KOH concentrations on setting time at 60 °C. The initial
CaO 41.67 0.02
MgO 6.49 0.09 setting time and final setting time for K1, K2, K3
Na2 O 0.22 0.02 samples at 60 °C were 18, 30, 35 and 29, 38, 44 min,
K2 O 0.36 1.56 respectively. It is obvious that the geopolymer setting
TiO2 0.53 0.13 time is faster at 60 °C than that at room temperature.
ZnO <0.01 <0.01
The effect of the addition of metakaolinite on setting

Table 2
Compositions of the geopolymers prepared from granulated blast furnace slag
Sample KOH concentration (N) SiO2 /Al2 O3 mole ratio SiO2 /K2 O mole ratio Al2 O3 /K2 O mole ratio K2 O/H2 O mole ratio
K1 5 3.46 2.69 0.78 0.024
K2 10 3.46 1.36 0.39 0.036
K3 15 3.46 0.91 0.26 0.069
WG1 10 3.26 1.28 0.39 0.022
WG2 10 3.36 1.32 0.39 0.030
WG3 10 3.46 1.36 0.39 0.036
WG4 10 3.56 1.39 0.39 0.041
M1 10 4.32 3.50 0.81 0.077
M2 10 3.85 1.97 0.51 0.052
M3 10 3.46 1.36 0.39 0.036
M4 10 3.16 1.05 0.33 0.026
 T.W. Cheng, J.P. Chiu / Minerals Engineering 16 (2003) 205–210 207

Fig. 1. The effect of KOH concentrations on setting time at room Fig. 4. The effect of the addition of sodium silicate on setting time at
temperature. 60 °C.

silicate on the setting time at 60 °C was also found in a

similar manner, as shown in Fig. 4.
From the results of the setting time measurement, it
can be seen that the geopolymer setting time was very
short, 15–45 min at 60 °C, due to the water loss in-
creasing the rate. As shown in Table 2, the setting time is
directly related to the potassium oxide (K2 O) content; it
can be found that the ratio of SiO2 /K2 O is very impor-
tant for setting time. The lower the SiO2 /K2 O ratio in
the pulp, the longer the time needed for setting. The
reason is that when the KOH solution is mixed in so-
dium silicate solution, it can lower the viscosity of the
pulp; therefore, the system needs more time for setting.

3.2. Geopolymer microstructure characterization

Fig. 2. The effect of KOH concentrations on setting time at 60 °C.
Scanning electron microscopy investigations were
conducted in order to get a better understanding of the
time at 60 °C is illustrated in Fig. 3. It can be seen that morphology of the geopolymer. The microstructure of
the more metakaolinite added in the system, the slower geopolymers for the K2 sample is shown in Fig. 5, and
was the setting time. The effect of the addition of sodium the X-ray microanalytical mapping images for Al, Si, K,

Fig. 3. The effect of the addition of metakaolinite on setting time at

60 °C. Fig. 5. SEM micrograph of geopolymer for K2 sample.
208 T.W. Cheng, J.P. Chiu / Minerals Engineering 16 (2003) 205–210

Fig. 6. The elemental distribution micrograph for sample K2 (a) Al, (b) Si, (c) K and (d) Na.

and Na elements are shown in Fig. 6. It can be seen that Table 3

the microstructure is not homogeneous, and the basic Various properties of the geopolymer materials
particle size of the K2 geopolymer is less than 50 nm. It Sample Density (g/cm3 ) Porosity (%) Water absorption (%)
can also be found from Fig. 6 that Si is the most widely K1 1.68 33.61 19.96
distributed element in the geopolymer, and also the K2 1.71 34.23 19.98
K3 1.82 30.74 16.94
distributions of Al, Si, K, and Na elements in K2 sample
WG1 1.77 31.58 17.88
are not homogeneous. The reason could be due to in- WG2 1.74 33.16 19.02
sufficient mixing time for making the geopolymer, so WG3 1.72 34.43 20.08
that the dispersion is not uniform. WG4 1.71 35.13 20.58
M1 1.83 28.31 15.49
M2 1.81 30.71 16.99
3.3. Physical and mechanical properties of the geopolymer
M3 1.72 34.43 20.08
M4 1.67 35.75 21.44
Various physical properties of geopolymer obtained
from granulated blast furnace slag at different condi-
tions are listed in Table 3. Densities of the geopolymer the structure formed. Generally, the geopolymers have
were in the range of 1.67–1.83 g/cm3 , and increased with earlier high strength. Of three different KOH concen-
increasing KOH concentration. However, while in- tration tests, the K2 sample (10 N KOH) shows the best
creasing sodium silicate or metakaolinite, the densities strength (more than 60 MPa) from 1 to 28 days. With
of geopolymers decreased. Porosities and water ab- increasing KOH concentration to 15 N, the compressive
sorptions of the geopolymer were in the range of 28.31– strength decreased which could be due to too much Kþ
35.63% and 15.49–21.44%, respectively. Both of them ions in the framework cavities. It is also clear from the
increase with increased sodium silicate and metakaoli- results obtained from Fig. 8 that the amount of meta-
nite. kaolinite added could play a significant role in the
Fig. 7 shows the geopolymer compressive strength strength. The maximum compressive strength was found
test results for different KOH concentrations and curing to reach 79 MPa, and the strength increases with the
days. From the results obtained, it is clear that the increasing addition of metakaolinite. Similar results had
concentration of KOH has an effect on the strength of also been found elsewhere (Swanepoel et al., 1999). The
 T.W. Cheng, J.P. Chiu / Minerals Engineering 16 (2003) 205–210 209

Fig. 7. Geopolymer compressive strength test results for different

Fig. 9. Compressive strength test results for different additions of so-
KOH concentration and curing days.
dium silicate.

Fig. 10. Geopolymer fire resistance test results for different KOH
concentrations.

Fig. 8. Compressive strength test results for different additions of

metakaolinite. when a 10 mm thick panel of geopolymers is exposed to
an 1100 °C flame, the measured reverse-side temperature
reason could be that the more metakaolinite added, the is a function of KOH concentration, and only reaches
more Al gel forms in the system, therefore, giving a
higher degree of geopolymer reaction. The compressive
strength test results for different additions of sodium
silicate are shown in Fig. 9. After the maximum com-
pressive strength reaches 70 MPa, the strength gradually
decreased with the increase in the amount of sodium
silicate. This is because excess silicate hinders water
evaporation and structure formation. Generally, when
the SiO2 /Al2 O3 ratio is in the range of 3.16–3.46, the
geopolymer has better compressive strength character-
istics, however, when the SiO2 /Al2 O3 ratio went up to
3.85, the compressive strength decreased.

3.4. Fire resistance tests of the geopolymer

Fig. 10 shows the geopolymer fire resistant tests at Fig. 11. Geopolymer fire resistance test results for different addition of
different concentrations of KOH. It can be seen that metakaolinite.
210 T.W. Cheng, J.P. Chiu / Minerals Engineering 16 (2003) 205–210

found that the K2 O content in the system plays an im-

portant role. With increasing K2 O content, the setting
time can be increased, the compressive strength can be
raised, and the fire resistance characteristics can also be
improved. It has therefore been established that granu-
lated blast furnace slag-based geopolymers can be fab-
ricated for construction purposes, and that these
geopolymers have great potential for engineering ap-
plications.

References

Comrie, D.C., Davidovits, J., 1988. Long term durability of hazardous

toxic and nuclear waste disposals. Geopolymer Õ88, First European
Conference on Soft Mineralurgy, Compiegne, France. pp. 125–134.
Fig. 12. Geopolymer fire resistance test results for different addition of
Davidovits, J., Comrie, D.C., Paterson, J.H., Ritcey, D.J., 1990.
sodium silicate.
Geopolymeric concretes for environmental protection. Concrete
International. pp. 30–40.
Davidovits, J., 1999a. Chemistry of geopolymeric systems. Terminol-
240–283 °C after 35 min. The fire resistant characteris- ogy, Geopolymer Õ99 Proceedings. Second International Confer-
tics can be improved by increasing the KOH concen- ence, France. pp. 9–39.
tration. The same fire resistant tests have been Davidovits, J., 1999b. Fire proof geopolymer cements. Geopolymer Õ99
Proceedings. Second International Conference, France. pp. 165–
conducted using different amounts of metakaolinite and
169.
sodium silicate, as shown in Figs. 11 and 12. When in- Ikeda, K., 1998. Consolidation of mineral powders by the geopolymer
creasing the amount of added metakaolinite, the fire binder technique for materials use. Journal of the Mining and
resistance characteristics can be improved. However, fire Materials Processing Institute of Japan 114, 497–500 (in Japanese).
resistance does not correlate well with the amount of Swanepoel, J.C., Strydom, C.A., Smit, J.P., 1999. Safe disposal of
brine water in fly-ash geopolymeric material. Geopolymer Õ99
sodium silicate added. The reverse-side temperatures of
Proceedings. Second International Conference, France. pp. 253–
test panels in all experiments were less than 350 °C after 267.
35 min, which demonstrated that granulated blast fur- Van Jaarsveld, J.G.S., Van Deventer, J.S.J., Lorenzen, L., 1997. The
nace slag based geopolymer is a good fire resistant ma- potential use of geopolymeric materials to immobilise toxic metals:
terial, which can be used in engineering applications. Part I. Theory and applications. Minerals Engineering 10 (7), 659–
669.
Van Jaarsveld, J.G.S., Van Deventer, J.S.J., Lorenzen, L., 1999. The
potential use of geopolymeric materials to immobilise toxic metals:
4. Conclusions Part II. Material and leaching characteristics. Minerals Engineering
12 (1), 75–91.
The geopolymerisation behavior, physical, mechani- Xu, H., Van Deventer, J.S.J., 1999. The geopolymerisation of natural
cal properties, and fire resistance characteristics of alumino-silicates. Geopolymer Õ99 Proceedings. Second Interna-
tional Conference, France. pp. 43–59.
granulated blast furnace slag-based geopolymer, are Xu, H., Van Deventer, J.S.J., 2000. The geopolymerisation of alumino-
strongly dependent upon the chemical composition in silicates minerals. International Journal of Mineral Processing 59,
the reaction system. In making the geopolymer, it was 247–266.