Thermal analysis of geopolymer pastes synthesised from ve y ashes of

variable composition
William D.A. Rickard
a,
, Jadambaa Temuujin
b
, Arie van Riessen
a
a
Centre for Materials Research, Curtin University, PO Box U 1987, Perth, WA 6845, Australia
b
Laboratory of Materials Science and Technology, Institute of Chemistry and Chemical Technology, Mongolian Academy of Sciences, Ulaanbaatar 51, Mongolia
a b s t r a c t a r t i c l e i n f o
Article history:
Received 29 February 2012
Received in revised form 11 May 2012
Available online 12 June 2012
Keywords:
Geopolymer;
Fly ash;
Thermal properties;
Thermal expansion;
Density
This paper presents a study on the thermal properties of a range of geopolymers in order to assess their suit-
ability for high temperature applications such as thermal barriers, refractories and re resistant structural
members. Geopolymers were synthesised from ve different y ashes using sodium silicate and sodium alu-
minate solutions to achieve a set range of Si:Al compositional ratios. The thermo-physical, mechanical and
microstructural properties of the geopolymers are presented and the effect of the source y ash characteris-
tics on the hardened product is discussed, as well as implications for high temperature applications. The
amount and composition of the amorphous component (glass) of each of the y ashes was determined by
combining XRD and XRF results. It was found that the Si:Al ratio in the glass of the y ashes strongly
inuenced the thermal performance of the geopolymers. Geopolymers synthesised from y ashes with a
high Si:Al (5) in the glass exhibited compressive strength gains and greater dimensional stability upon ex-
posure to 1000 C, whereas geopolymers synthesised fromy ashes with low Si:Al (b2) in the glass exhibited
strength losses and reduced dimensional stability upon high temperature exposure.
2012 Elsevier B.V. All rights reserved.
1. Introduction
Geopolymers are a synthetic inorganic polymeric material
produced by the alkali activation of aluminosilicates [1]. They are a
broad class of binding material with applications that range fromcon-
ventional concrete to high-tech, light weight composites for use in
aviation. Geopolymers have also shown promise for use in high tem-
perature applications, such as re proof coatings, structural concrete
in re prone areas and thermal insulation for refractory type applica-
tions, due to their intrinsic thermal stability [2].
Another driver for the development of geopolymer based mate-
rials is due to their environmental credentials. Geopolymer concretes
are reportedly able to be produced with up to 80% less CO
2
emissions
than the equivalent ordinary Portland cement product [3], though re-
cent studies suggest this value is closer to a 50% reduction [4]. They
also have the potential to be made using industrial waste materials
[5,1,6], further enhancing their environmental credentials. In a polit-
ical climate where governments around the world are taxing CO
2
emissions and promoting sustainable and environmentally friendly
materials, geopolymers are emerging as an important material of
the future.
Typical aluminosilicate sources used for the synthesis of geo-
polymers are metakaolin or industrial waste products such as y
ash, slag and rice husk ash [7,1]. Metakaolin based systems are
regarded as an ideal system but most researchers agree that the
utilisation of waste precursor materials will enhance the environ-
mental and economic credentials of the end product. Signicantly,
this can be achieved without the loss of physical performance, and
in many cases, improves the performance when compared to equiv-
alent metakaolin systems [8].
Fly ash is produced from the coalescence of non-combustible
material in the ue gases of coal red power stations [6]. Fly ash is a
powdery material made up of small glass spheres, consisting primar-
ily of silicon, aluminium, iron, and calcium oxides [9]. The typical
particle size range is between 1 and 150 m [5]. Crystalline phases
commonly identied in y ashes are quartz, mullite and various
iron rich phases such as hematite [10]. Fly ash characteristics are
highly variable between power stations and are strongly dependent
on the coal source and burning conditions. Fly ash characteristics
that are important when considering their use in geopolymers are;
particle size (inuences reactivity) [11,12], glass content (sets the
amount of reactive material) [13], Si:Al ratio in the glass (controls
the mechanical strength and thermal resistance) [6], iron content
(important in high temperature applications) [14] and calcium con-
tent (affects setting time) [15,16]. Class F y ash, dened as having
SiO
2
+Al
2
O
3
+Fe
2
O
3
70 wt.%, is readily available around the world
and is produced from burning anthracite or bituminous coal [17].
It is critical to understand the characteristics of y ash in order to
produce geopolymers suitable for high temperature applications.
Previous research has found geopolymers to be intrinsically ther-
mally resistant due to their inorganic framework [2]. Kong et al. [8]
Journal of Non-Crystalline Solids 358 (2012) 18301839
Corresponding author. Tel.: +61 8 9266 4219; fax: +61 8 9266 2377.
E-mail address: w.rickard@curtin.edu.au (W.DA. Rickard).
0022-3093/$ see front matter 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.jnoncrysol.2012.05.032
Contents lists available at SciVerse ScienceDirect
Journal of Non-Crystalline Solids
j our nal homepage: www. el sevi er . com/ l ocat e/ j noncr ysol
compared the thermal resistance of y ash and metakaolin geo-
polymers and found that y ash geopolymers have a pore structure
which facilitates the escape of moisture without damaging the struc-
ture during heating, whereas metakaolin geopolymers do not. Other
researchers have assessed the thermal properties of y ash geo-
polymers [15,18,8,6,19] and observed low thermal shrinkage and
good strength retention. While most of these papers describe the
total chemical analysis of their y ash source/s, few have quantita-
tively assessed the chemical composition of the glass content, consid-
ered to be the only reactive component of the y ash during the
geopolymerisation process [13].
Five class F y ashes from power stations around Australia were in-
vestigated in this study. The y ashes used for analysis were selected for
their diversity in terms of iron content, glass Si:Al ratio and calciumcon-
tent. The objective of this paper is to quantitatively characterise the y
ashes and resulting thermal properties of a range of geopolymers
synthesised from the different y ashes in order to gain a better under-
standing of the mechanisms that determine the observed properties.
The measured data on the ve y ashes and the resulting geopolymers
used in this study will be useful for researchers and engineers looking to
develop geopolymers from these sources in the future.
2. Experimental
2.1. Materials
Fly ash was sourced from Collie power station in Western Austra-
lia, Eraring and Bayswater power stations in New South Wales, Port
Augusta power station in South Australia and Tarong power station
in Queensland. The y ash was provided by Fly Ash Australia (FAA).
The sodium silicate solutions used for the Collie y ash series were
prepared by dissolving sodium hydroxide pellets from Univar Pty
Ltd and fumed silica from Cabosil in deionised water. The pellets
were allowed to dissolve for 24 h at 70 C prior to use. The sodiumsil-
icate solutions used for the Port Augusta y ash series were prepared
by dissolving sodium hydroxide pellets from Univar Pty Ltd and
deionised water into PQ-D (29.4 wt.% SiO
2
, 14.7 wt.% NaOH,
55.9 wt.% H
2
O) from PQ Corporation. Sodium aluminate solutions
were supplied by Coogee Chemicals who specied that the solution
contained 19 wt.% Al
2
O
3
, 25.5 wt.% NaOH and 55.5 wt.% H
2
O. Table 1
lists the amorphous content and Table 2 lists the crystalline content
of each of the y ashes as determined by x-ray uorescence (XRF)
and quantitative x-ray diffraction (QXRD). Further details on the
quantication of Collie, Eraring and Tarong y ashes can be found in
the paper by Rickard et al. [6]. The quantication of Port Augusta
and Bayswater y ashes was conducted using the same method but
has not been previously published.
2.2. Geopolymer synthesis
Samples were prepared with a range of compositional ratios as
shown in Table 3. Only the amorphous component (from Table 1) of
the aluminosilicates in the y ashes was used in the calculation as
the crystalline material was considered as an inert phase. Three
mixes were prepared for each y ash with designed Si:Al ratios of
2.0, 2.5 and 3.0. This Si:Al ratio range was selected as it is known to
produce high strength geopolymers [20]. The Na:Al and H:Si compo-
sitional ratios were kept reasonably constant and were only varied
where workability or solution chemistry demanded it. Geopolymers
were synthesised by mixing the y ash and the alkaline solution in
a planetary centrifugal mixer (ARE 250, Thinky, Japan) for 5 min at
1300 RPM. A further 30 s at 2100 RPM was used for defoaming. The
slurry was then cast into various sized moulds for testing. The moulds
were sealed and samples cured for 24 h at 70 C. Samples were
demoulded and stored at ambient temperature prior to testing.
2.3. Dilatometry
Dilatometry samples were cast into 1 mL syringes and cut to
length to achieve a cylinder of 15 mm length and 5 mm diameter.
Pushrod dilatometry was performed using a DI-24 Adamel Lhomargy
(France) dilatometer. Measurements were conducted according to
ASTM E831 [21] over a temperature range of 201000 C with a
heating rate of 5 C/minute. A preload of 100 mN was set to allow
the instrument to record data from shrinking samples. All presented
dilatometry results are an average of measurements from three
samples.
2.4. Optical microscopy
Low magnication imaging was performed on the samples using a
Nikon SMZ 800 stereo microscope. Sample fragments were mounted
Table 1
Amorphous (glass) composition of each of the y ashes. Uncertainties in brackets [6].
Oxide Collie
wt.%
Port Augusta
wt.%
Eraring
wt.%
Bayswater
wt.%
Tarong
wt.%
SiO
2
20.90 (65) 24.53 (67) 45.03 (65) 57.15 (77) 42.79 (59)
Al
2
O
3
15.39 (41) 11.14 (45) 7.67 (32) 6.47 (29) 4.11 (26)
Fe
2
O
3
9.11 (32) 1.03 (17) 2.49 (16) 1.36 (19) 0.64 (2)
CaO 1.74 (5) 4.60 (5) 1.59 (5) 1.37 (5) 0.08 (5)
K
2
O 0.90 (4) 1.34 (4) 1.68 (4) 0.55 (4) 0.53 (4)
TiO
2
1.47 (1) 1.42 (1) 0.84 (1) 0.61 (1) 1.28 (1)
MgO 1.41 (3) 2.53 (3) 0.51 (3) 0.29 (3) 0.17 (3)
Na
2
O 0.41 (5) 2.30 (5) 0.56 (5) 0.09 (5) 0.09 (5)
P
2
O
5
1.09 (2) 0.82 (2) 0.27 (2) 0.15 (2) 0.08 (2)
SrO 0.23 (1) 0.06 (1) 0.05 (1) 0.02 (1) 0.01 (1)
BaO 0.38 (1) 0.00 (1) 0.06 (1) 0.00 (1) 0.03 (1)
Other 1.15 (6) 0.46 (6) 1.67 (6) 0.11 (5) 0.96 (6)
Sum of amorphous aluminosilicates 36.29 (77) 35.67 (89) 52.70 (73) 63.62 (95) 46.90 (64)
SiO
2
/Al
2
O
3
1.36 (6) 2.20 (7) 5.87 (26) 8.83 (22) 10.42 (68)
Si/Al (molar) 1.15 (4) 1.87 (6) 4.98 (19) 7.49 (19) 8.84 (49)
Table 2
Crystalline phase composition of each y ash. Uncertainties in brackets [6].
Phase Collie
wt.%
Port Augusta
wt.%
Eraring
wt.%
Bayswater
wt.%
Tarong
wt.%
Amorphous
(glass)
54.00 (45) 50.63 (73) 62.74 (31) 68.39 (39) 50.82 (28)
Quartz 26.19 (39) 19.99 (75) 14.89 (30) 23.14 (37) 24.08 (27)
Mullite 15.80 (18) 26.83 (33) 20.88 (14) 7.39 (15) 25.10 (11)
Hematite 1.50 (64) 0.87 (38)
Magnetite 2.51 (83) 1.49 (52)
Maghemite-C 0.85 (27) 1.05 (69)
Rutile 0.74 (50)
1831 W.DA. Rickard et al. / Journal of Non-Crystalline Solids 358 (2012) 18301839
in epoxy resin and polished to a 1 mnish using progressively ner di-
amond paste on a Struers magnetic disk polishing system. The sample
blocks were cleaned in an ultrasonic bath to remove any polishing
debris.
2.5. SEM
Scanning electron microscopy (SEM) was conducted on a NEON
40EsB (Zeiss, Germany) eld emission SEM. Samples were out gassed
in a desiccator over a 48 h period and then coated with a 2 nmlayer of
platinum prior to imaging in the SEM. The accelerating voltage used
was 15 kV for backscattered electron (BSE) imaging.
2.6. Mechanical testing
Cylinders of 15 mm diameter and 30 mm height were prepared
for mechanical testing. Geopolymer samples were tested 28 days
after synthesis. Compressive strength testing was conducted on a
Lloyds universal tester EZ50. A load rate of 0.25 MPa/s was used to
closely comply with ASTM C39 (a test method for concrete speci-
mens) [22]. The stated strength values are the average of results
from 4 samples.
2.7. Density
The density of solid geopolymer samples was measured by the
simple method of dividing the mass by the volume. Cylindrical sam-
ples (15 mm diameter, 30 mm high) were used for density measure-
ments. Samples were carefully selected to ensure voids, cracks or
shape defects did not skew volume measurements. Dimensions
were measured using digital Vernier calipers (Kincrome, Australia).
Sample mass was measured using a laboratory scale (model AA-
200, Denver Instrument Company, USA). All reported results are the
average of 5 separate measurements.
2.8. X-ray diffraction
X-ray diffraction (XRD) patterns were collected on a D8 Advance
diffractometer (Bruker AXS, Germany) using a Cu K source. The
data was collected using a nominal 2 step size of 0.01, a count
time of 0.5 s per step and a 2 range of 10120.
3. Results
3.1. Thermal expansion
Fig. 1 details the thermal expansion of the geopolymers synthesised
from each y ash. For each y ash there is the thermal expansion curve
for samples with Si:Al ratios of 2.0, 2.5 and 3.0. A marked difference can
be observed between the different y ash geopolymers and in some
cases, with changing Si:Al ratio. Eraring, Bayswater and to a slightly
lesser extent Tarong y ash geopolymers exhibited impressive thermal
stability up to almost 800 C by having minimal shrinkage after the ini-
tial shrinkage due to the dehydration of free water (noted by a shrink-
age of 12% between 50 C and 200 C).
Collie and Port Augusta geopolymers exhibited less thermal stabil-
ity. Notably, Collie y ash exhibited increasing thermal expansion
with increasing Si:Al ratio (Fig. 1a). These y ashes had reactive Si:
Al ratios of 1.15 (Collie) and 1.87 (Port Augusta) and thus to achieve
geopolymers with Si:Al ratios between 2.0 and 3.0 the activating so-
lutions contained additional silicate. It is likely that a portion of the
activating solution remained unreacted or partially reacted during
the formation of the geopolymer leaving residual silicates in the sam-
ple. Thermal expansion during heating in geopolymers due to the
swelling of high silicate secondary phases has been described previ-
ously in the literature [23,24]. The thermal expansion observed in
the Collie y ash geopolymer samples is consistent with the swelling
of unreacted silicates with the magnitude proportional to the amount
of added silica. For example, the expansion in the Collie 3.0 geo-
polymer is much greater than the Collie 2.0 geopolymer due to the
greater amount of added silica in the activating solution. There was
considerably less swelling in the Port Augusta y ash geopolymer
samples due to the lower amount of added silica (as the Si:Al ratio
of the glass was closer to 2).
All samples exhibited a sharp shrinkage at high temperature. This
occurred at approximately 800 C for Eraring, Bayswater and Tarong
samples whereas for the Collie and Port Augusta geopolymers it oc-
curred between 550 and 850 C, depending on the Si:Al ratio of the
sample. This high temperature shrinking event has been observed
previously in geopolymers [14,25] and is due to the viscous ow of
the aluminosilicate material causing sintering and subsequent densi-
cation of the samples.
3.2. Structural evolution
The geopolymers in this study were compressive strength tested
before and after exposure to high temperature in order to assess
their strength retention. Fig. 2 compares the compressive strength
of the geopolymers from each y ash before and after ring to
1000 C.
Initial compressive strengths ranged from an impressive 143 MPa
to a very low 6 MPa. Collie and Port Augusta y ashes produced the
strongest samples. Tarong and Eraring y ashes produced moderate
strength samples, whereas Bayswater y ash produced very weak
geopolymers. The geopolymers made with a sodium aluminate acti-
vating solution showed little difference in strength with change in
Si:Al ratio. However, for the sodium silicate activated samples there
was a signicant variation in strength with change in Si:Al. For the
Collie y ash geopolymers the strength decreased with increasing in
Si:Al, while for Port Augusta y ash geopolymers a peak in strength
was achieved for Si:Al of 2.5. With the exception of Bayswater y
ash geopolymers, the compressive strength of all samples was at
least 25 MPa, sufcient for most high temperature applications
where thermal resistance rather than mechanical strength is most
critical.
Post ring compressive strengths varied dramatically. In gener-
al, Eraring, Tarong and Bayswater y ash geopolymers exhibited
compressive strength increases after ring whereas Port Augusta
and Collie y ash geopolymers exhibited strength losses. Eraring
y ash geopolymers had the most signicant strength gains with
up to a 5 fold strength increase. Geopolymers with initial compres-
sive strengths greater than 40 MPa exhibited strength losses after
ring, whereas samples with lower initial compressive strength
exhibited strength gains.
Table 3
Elemental ratios of the geopolymers prepared in this study.
Fly ash Si:Al Na:Al H:Si Sample name
Collie 2.0 1.25 5.0 C2B
Collie 2.5 1.25 4.5 C2.5B
Collie 3.0 1.25 6.0 C3B
Port Augusta 2.0 1.25 7.9 P2D
Port Augusta 2.5 1.25 5.5 P2.5D
Port Augusta 3.0 1.25 5.5 P3D
Eraring 2.0 1.25 5.0 E2B
Eraring 2.5 1.25 4.0 E2.5B
Eraring 3.0 1.25 4.5 E3B
Bayswater 2.0 1.6 6.0 B2C
Bayswater 2.5 1.47 5.5 B2.5 C
Bayswater 3.0 1.33 5.5 B3C
Tarong 2.0 1.32 6.4 T2B
Tarong 2.5 1.25 5.5 T2.5B
Tarong 3.0 1.25 5.4 T3B
1832 W.DA. Rickard et al. / Journal of Non-Crystalline Solids 358 (2012) 18301839
Fig. 3 compares the post ring compressive strengths as a percent-
age of the room temperature strengths with the percentage of Si or Al
introduced via the sodium aluminate or sodium silicate activating
solution. The trendlines in Fig. 3 suggest that reducing the amount
of Si or Al added via activating solution improves the compressive
strength gain/retention after ring. The effect is more pronounced
in the sodium aluminate activated samples which exhibited strength
gains of almost ve fold where 40% of the total Al was added via the
activating solution. Given there is no indication that the trend has
reached a maxima, it is likely that lower Al concentrations would
result in even greater strength increases though this would require
high Si:Al ratio samples which may have low as cured compressive
strengths.
3.3. Density evolution
Thermal expansion measurements indicated a densication of the
samples due to the sintering of the aluminosilicates at temperatures
between 550 C and 850 C. Other effects, such as the dehydration
of water in the sample and the swelling of the high silicate phases
will reduce the samples density by reducing the mass or expanding
the volume. Thus the measured changes in density after high temper-
ature exposure are a convolution of the various effects. Density is
known to affect the mechanical strength of geopolymers [26] and as
such its quantication provides useful additional insight into the
thermally induced changes in the samples after heating to 1000 C.
Table 4 lists the density of the samples in this study before and
after ring to 1000 C. The as cured densities were observed to be
consistent with the compressive strength results. The lower strength,
sodium aluminate geopolymers (from Eraring, Tarong and Bayswater
y ashes) were measured to have densities ranging from 1.51 to
1.71 g cm
3
, whereas the higher strength sodium silicate activated
geopolymer (from Collie and Port Augusta y ashes) had higher den-
sities, ranging from 1.86 to 2.11 g cm
3
. Bayswater y ash geo-
polymers exhibited a higher density than was expected based on
their low compressive strength. This may be due to a higher y ash
particle density relative to the other y ashes (i.e. less porous
spheres) which would increase the measured total density. The den-
sity results for Eraring, Tarong and Bayswater y ashes were consis-
tent with a comparable study by Andini et al. [27] that measured
Fig. 1. Thermal expansion of geopolymers made from (a) Collie y ash, (b) Port Augusta y ash, (c) Eraring y ash, (d) Bayswater y ash and (e) Tarong y ash with Si:Al =2.0, 2.5
and 3.0. Note (a) has a different y-axis scale.
1833 W.DA. Rickard et al. / Journal of Non-Crystalline Solids 358 (2012) 18301839
the density of y ash geopolymers to range between 1.48 and 1.74 g
cm
3
.
The post ring densities are consistent with the measured changes
in compressive strengths after ring, though the percentage change
was much smaller. The densities of the Port Augusta y ash geo-
polymers only reduced by a few percent after ring, though the com-
pressive strengths after ring reduced by more than 50%. This
suggests that other strength reducing (but not density reducing) ef-
fects, such as cracking, were more dominant in these samples than
the reduction in the density of the binding geopolymer phase.
Bayswater y ash geopolymers were once again an anomaly with a
slight reduction in density even though they exhibited increased
post ring compressive strengths. It is likely that the increased parti-
cle connectivity due to sintering had a more dominant effect on the
samples strength than any contribution of strength decrease due to
reduced sample density.
3.4. Microstructural evolution
Low magnication, large eld of view, microscopy was conducted
to observe changes in the microstructure at a moderate length scale.
Fig. 4 compares polished fragments of Collie and Tarong y ash geo-
polymers (Si:Al =2.5) before and after ring to 1000 C. The Collie
y ash geopolymer can be seen to have reduced homogeneity after
ring with an increase in porosity and regions of increased iron oxi-
dation (seen as red veins running along the cracks in the sample).
In contrast, the Tarong y ash geopolymer exhibits a homogenous
structure with no visible macro porosity after ring. This is consistent
with the mechanical strength results as the Tarong sample with less
thermally induced defects exhibited greater strength retention.
Higher magnication microscopy was conducted with a eld emis-
sion SEM to further investigate the changes to the microstructure
after ring. Fig. 5 compares the microstructure of the Si:Al =2.5 series
geopolymers from each y ash. The as cured microstructure was typi-
cal for y ash geopolymers consisting of geopolymer gel amongst
unreacted y ash particles. Collie and, in particular Port Augusta y
ash, geopolymers exhibited the greatest portion of geopolymer gel
which was expected given their high compressive strengths. However,
they both also exhibited a high degree of cracking. It is believed that
this is an artefact of sample preparation rather than an existing sample
characteristic. The higher density (from Table 4) and greater portion of
geopolymer gel (Fig. 5) in these samples reduces their permeability and
Table 4
Density of the y ash geopolymers before and after ring to 1000 C. Uncertainties in
brackets.
Si:Al As cured density
(g.cm
3
)
Post ring density
(g.cm
3
)
% of as cured density
Collie 2.0 1.99 (5) 1.69 (6) 85%
2.5 2.11 (10) 1.56 (4) 74%
3.0 2.04 (7) 1.45 (3) 71%
Port
Augusta
2.0 1.86 (8) 1.80 (3) 97%
2.5 1.89 (7) 1.86 (1) 98%
3.0 1.86 (4) 1.75 (17) 94%
Eraring 2.0 1.58 (6) 1.67 (3) 106%
2.5 1.61 (12) 1.71 (4) 106%
3.0 1.57 (4) 1.71 (3) 109%
Bayswater 2.0 1.70 (6) 1.51 (2) 89%
2.5 1.71 (16) 1.58 (11) 93%
3.0 1.68 (2) 1.68 (4) 100%
Tarong 2.0 1.55 (7) 1.55 (1) 100%
2.5 1.58 (5) 1.64 (2) 104%
3.0 1.51 (5) 1.80 (2) 119%
Fig. 2. Compressive strength before and after ring to 1000 C for geopolymers from
each y ash. Note: the data for Collie, Eraring and Tarong y ashes has been adapted
from a previous publication [6].
Fig. 3. Comparing the effect of the amount of the X (Si or Al) added via the activating solution on the post ring compressive strengths (as a percentage of room temperature
strengths).
1834 W.DA. Rickard et al. / Journal of Non-Crystalline Solids 358 (2012) 18301839
as such inhibits the dehydration of free water during drying (prior to
SEManalysis), causing cracking. This is not the case in the lower density
samples where water can escape without damaging the structure. The
red samples are completely dehydrated and as such are not subjected
to dehydration stresses when placed in a vacuum.
After ring the microstructure changed considerably. In all sam-
ples there is a reduced portion of discernible unreacted y ash parti-
cles and a better interparticle bonding due to sintering and viscous
ow of aluminosilicates during heating. This is likely the main mech-
anism which leads to the increase in compressive strength after ring
in some samples. There was also an apparent increase in porosity in
the size range of 5 to 20 m. An increase in porosity would be
expected to lead to reduced compressive strength though there was
no clear trend in the SEM analysis of increasing post ring porosity
with lower residual compressive strength. Thus it is likely the mech-
anisms that reduce compressive strength, as such cracking, occur on
longer length scales than observed by the SEM analysis.
3.5. Phase changes
The crystalline phase composition of the as cured geopolymers
reected their source y ash by having the same phase composition
(as the crystalline phases are inert during the geopolymerisation
process). The XRD patterns also had a broad hump, characteristic of
an amorphous component (Fig. 6). The sodium aluminate activated
geopolymers (Eraring, Tarong and Bayswater) also contained various
zeolitic phases such as hydrosodalite (Na
6
(AlSiO
4
)
6
8(H
2
O)) and
faujasite-Na (Na
1.88
Al
2
Si
4.8
O
13.54
9H
2
O). These samples were found
to have a low amount of geopolymer gel and a high portion of
unreacted source material (Fig. 5). This suggests that only a portion
of the activating solution was consumed during the formation of the
samples and the zeolitic phases are likely to have crystallised out of
the residual activating solution and the dissolved aluminosilicates.
The peak intensity of the zeolitic phases was detected to increase as
the Si:Al ratio reduced indicating that there was a greater concentra-
tion of these phases in the samples with more aluminate added via
the activating solution.
The Tarong y ash geopolymers were found to contain gibbsite
(Al(OH)
3
). Gibbsite is likely to have precipitated out of the sodium
aluminate activating solution either prior to sample synthesis or
during the geopolymerisation process when the pH of the solution
dropped as the sodium was consumed by the geopolymer phase.
After ring the samples retained their characteristic amorphous
hump though there were changes in the crystalline phase composition.
The hydrated phases that were detected in the as cured geopolymers
were destroyed upon ring. Sodium based feldspars were detected in
all the samples after ring. Nepheline (NaAlSiO
4
) was detected in all of
the red geopolymers. Other formed phases were albite (NaAlSi
3
O
8
)
(Collie and Port Augusta geopolymers) and tridymite (SiO
2
) (all except
Port Augusta geopolymers). These high temperature phases are reported
to improve the thermal resistance of the geopolymers due to their high
melting points (nepheline 1257 C, albite 1118 C [28] and tridymite
1670 C [2931].
The peak intensity of the mullite and the quartz phases were
found to reduce in all geopolymers after ring. Mullite has a melting
point of approximately 1830 C, varying slightly with mullite compo-
sition [32]. As such it was not expected that the mullites would melt
or dissolve during ring to 1000 C. It is believed that the presence of
free sodium in the geopolymers causes the low temperature dissolu-
tion/melting of the mullite into the geopolymer as alkali's are known
to be excellent sintering agents [31]. The reduction of quartz peak in-
tensity is likely due to the partial phase change to tridymite.
Phase changes were also observed in the iron oxides though they
were difcult to analyse due to their low peak intensity and excessive
peak overlap with other phases. Maghemite-C (Fe
2
O
3
) was destroyed
or detected with reduced peak intensity after ring. Magnetite (Fe
3
O
4
)
(Collie and Eraring geopolymers) peak intensity was not observed to
change after ring. Hematite (Fe
2
O
3
), however, exhibited increased
peak intensity, especially in the Collie y ash geopolymers (due to
their higher concentration of iron). Hematite is likely to have
crystallised from the amorphous iron in the y ash (observed previ-
ously by Rickard et al. [14]) as well as from the transformation of
maghemite-C, known to occur between 370 C and 600 C [33].
Port Augusta and Bayswater y ash geopolymers were found to
Fig. 4. Optical microscope images comparing Collie (top) and Tarong (bottom) geopolymers before and after ring to 1000 C.
1835 W.DA. Rickard et al. / Journal of Non-Crystalline Solids 358 (2012) 18301839
Fig. 5. Backscattered SEM micrographs of the Si:Al =2.5 geopolymers before (left images) and after ring to 1000 C (right images). (a) Collie, (b) Port Augusta, (c) Eraring, (d)
Bayswater and (e) Tarong, y ash geopolymers.
1836 W.DA. Rickard et al. / Journal of Non-Crystalline Solids 358 (2012) 18301839
contain small maghemite-C peaks after ring. It has been reported
that substitution in maghemites, as is likely in y ash phases, can re-
tard the phase transformation [34] explaining their presence after
ring.
4. Discussion
Fly ash based geopolymers are a highly complex material due to
their many phases and inhomogeneous microstructure. Physical mea-
surements on these samples are convoluted by the varying inuence
of each of these phases and as such the interpretation of the results is
inherently difcult. The results in this paper are complex though use-
ful information was able to be obtained by observing general trends.
It was found that the reactive Si:Al ratio of the y ashes strongly
inuenced the as cured and post ring compressive strengths as it
inuenced the composition and type of the activating solution. Collie
and Port Augusta y ashes had reactive Si:Al ratios less than 2 (1.2 and
1.9, respectively) and as such were activated with sodium silicate solu-
tions to achieve the desired geopolymer Si:Al ratios of 2.0, 2.5 and 3.0,
producing high strength geopolymers. Eraring, Tarong and Bayswater
y ashes had reactive Si:Al ratios greater than 3 (5.0, 8.8 and 7.5, respec-
tively) and as such were activated using sodium aluminate solutions to
Fig. 6. XRD patterns from the geopolymers from each of the y ashes. Blue=as cured, red=red. (a) Collie, (b) Port Augusta, (c) Eraring, (d) Bayswater and (e) Tarong, y ash
geopolymers.
1837 W.DA. Rickard et al. / Journal of Non-Crystalline Solids 358 (2012) 18301839
achieve the same set of compositional ratios, producing geopolymers
with low to moderate compressive strengths. The post ring compres-
sive strength testing of the geopolymers produced opposite results
where the sodium aluminate activated samples exhibited moderate to
high compressive strengths and the sodium silicate activated samples
exhibited low compressive strengths.
Microstructural investigations of the as cured geopolymers showed
that the sodium silicate activated samples had a greater portion of geo-
polymer gel than the sodium aluminate activated samples suggesting
that this activating solution was more successful in converting the
amorphous aluminosilicates in the y ashes to form the binding
phase. The sodium silicate activated geopolymers were also found to
be denser than the sodium aluminate samples. This is likely due to the
sodium silicate activated samples being able to be synthesised with
marginally lower water content, as a soluble silicate was used to con-
centrate the activating solutions. The greater portion of binding phase
and higher density of the sodium silicate activated geopolymers lead
to them having greater compressive strengths than the sodium alumi-
nate activated geopolymers.
After ring, geopolymer compressive strength were affected by
the varying inuence of phase changes in the non geopolymer
phase as well as damage due to dehydration during heating. Sintering
was the main factor that leads to increased compressive strength in
the geopolymers after ring as it released more of the aluminosilicate
material from the unreacted y ash particles into the binding phase
and improved inter-particle connectivity. This effect was evident
and reasonably consistent in all samples and as such the cause for
the variable post ring strengths was due to the varying inuence
of other strength reducing effects.
The magnitude of the difference between the reactive Si:Al ratio in
the y ash to the designed Si:Al ratio of the geopolymers (between
2.0 and 3.0) affected the compressive strength of the samples as it
inuenced the composition of the activating solution. In the sodium
silicate activated samples it was found that increasing the amount
of silicon added via the activating solution lead to reduced post ring
compressive strengths. It is likely that the samples with a large por-
tion of Si added via the activating solution did not incorporate all
the added material into the geopolymer structure, leaving residual
silicates. A high silicate phase was not detected in the XRD analysis
of sodium silicate activated samples though this is most likely due
to it being x-ray amorphous.
Thermal expansionmeasurements indicated that there was swelling
of a high silicate phase at high temperatures in the sodium silicate acti-
vated samples. It is likely that swelling of this phase contributed to the
strength losses in these samples. This is supported by the fact that the
post ring compressive strengths of the Collie and Port Augusta y
ash geopolymers reduced signicantly as the silica content in the acti-
vating solution increased (i.e. as the Si:Al increased from 2.0 to 3.0).
SEM analysis also indicated that the sodium silicate activated samples
were susceptible to dehydration damage, due to their high concentra-
tion of geopolymer gel, which would also have contributed to strength
losses in these samples.
The amount of aluminium added via the activating solution had
little effect on the as cured compressive strengths of the sodium alu-
minate activated samples, though it did inuence the post ring
strengths. From Fig. 3 it can be seen that reducing the amount of alu-
minium added via the solution increased the post ring compressive
strength gains. XRD analysis of these samples found that the sodium
aluminate activated samples contained crystalline zeolitic phases,
with increasing peak intensity with reducing Si:Al ratio in the geo-
polymer (i.e. increasing aluminium added via activating solution).
These phases were destroyed upon ring which may have had a sim-
ilar effect on the strength of the geopolymers as the destruction of the
high silicate phase in the sodiumsilicate activated geopolymers, though
to a much lesser extent as it was not associated with any signicant vol-
ume changes in the thermal expansion measurements.
4.1. Thermally induced changes in non geopolymer phases
Previous studies [6] have suggested that high iron content in the
source y ash can cause cracking and subsequent strength losses in
geopolymers. The results in this study suggest that there is little effect
in low iron concentrations (b3 wt.%) as there was no observable cor-
relation of thermal performance with iron content in the y ashes
other than Collie. Collie y ash contains 13.2 wt.% [6] iron oxide and
samples synthesised from this y ash exhibited the poorest strength
retention and dimensional stability suggesting that high concentra-
tions of iron in the y ash precursor is detrimental to a geopolymer's
thermal performance.
XRD phase analysis revealed interesting information about the role
of the secondary phases in the geopolymer. It was not possible to deter-
mine if the formed phases (such as nepheline) were derived from
changes in the original crystalline material or crystallisation of amor-
phous material either from the unreacted y ash or the geopolymer
gel. Previous research has found the geopolymer gel phase remains to-
tally amorphous when red up to 1200 C [2], whereas other studies
have observed crystallisation starting as low as 600 C [1]. It is believed
by the authors that the bulk of the crystallisation observed in the
samples in this study derives from the free Na, Si and Al species after
the decomposition of the zeolitic phases that are present in the as
cured samples. This is supported by the amorphous hump in the
diffraction patterns remaining largely unchanged after ring suggest
the geopolymer phase maintained its amorphous structure.
The phase composition after ring was largely independent of Si:
Al ratio as the phases for the three samples for each y ash geo-
polymer was the same. However, it was observed that the peak inten-
sity of the formed phases varied with Si:Al ratio. In all samples the
peak intensity of the nepheline phase reduced with increasing Si:Al
ratio. This suggests that low silica content is preferable for nepheline
crystallisation in geopolymers red to 1000 C. The trend for the
albite was not consistent across the y ashes. In the Collie y ash
geopolymers the peak intensity of the albite phase decreased with in-
creasing Si:Al ratio, whereas in the Port Augusta geopolymers the
peak intensity increased. Albite was not detected in the geopolymers
from the other y ashes.
5. Conclusions
Physical characteristics of geopolymers synthesised from ve dif-
ferent y ashes before and after ring to 1000 C have been pres-
ented. The results varied greatly with y ash source and the most
inuential y ash characteristic was the reactive Si:Al ratio (amor-
phous Si:Al ratio). Fly ashes with a high reactive Si:Al ratio (5)
were sodium aluminate activated and produced geopolymers with
low to moderate as cured compressive strengths but exhibited ex-
cellent dimensional stability during heating and greater compressive
strengths after heating. Fly ashes with a low reactive Si:Al ratio (b2)
were sodium silicate activated and produced geopolymers with high
as cured compressive strengths but exhibited poor dimensional sta-
bility during heating and greatly reduced compressive strengths after
heating.
All samples exhibited strength improving microstructural changes
such as improved inter-particle bonding due to sintering after ring.
However, the instability of non geopolymer phases during high tem-
perature exposure led to strength losses in some samples depending
on the type and composition of the activating solution. The sodium
silicate activated geopolymers were more susceptible to damage by
non geopolymer phases (such as unreacted high silicate phases) as
they were associated with volume expansion which caused cracking.
XRD analysis found the as cured geopolymers were largely amor-
phous though they still contained the crystalline phases initially pre-
sent in the y ash precursor. Some of the geopolymers, in particular
the low strength geopolymers, contained poorly ordered zeolitic
1838 W.DA. Rickard et al. / Journal of Non-Crystalline Solids 358 (2012) 18301839
phases as well. After ring the geopolymers remained largely amor-
phous though they all contained feldspars that likely crystallised
from the non geopolymer phases.
Acknowledgements
This study was supported by the R&D Center for Valuable
Recycling (Global-Top Environmental Technology Development Pro-
gram) funded by the Ministry of Environment, South Korea (Project
No.:11-E28-2R). The authors would like to thank Ash Development
Association of Australia (ADAA) and Fly Ash Australia (FAA) for
their assistance in acquiring the y ash used in this study. The authors
acknowledge the facilities, scientic and technical assistance of the
Curtin University Electron Microscopy Laboratory, a facility funded
by the university, state and Commonwealth governments.
References
[1] P. Duxson, A. Fernandez-Jimenez, J.L. Provis, G.C. Lukey, A. Palomo, J.S.J. van
Deventer, J. Mater. Sci. 42 (2007) 29172933.
[2] V.F. Barbosa, K.J. MacKenzie, Mater. Res. Bull. 38 (2003) 319331.
[3] J. Davidovits, World Resour. Rev. 6 (1994) 263278.
[4] B.C. McLellan, R.P. Williams, J. Lay, A. van Riessen, G.D. Corder, J. Clean. Prod. 19
(2011) 10801090.
[5] E.E. Berry, V.M. Malhotra, J. Am. Concr. Inst. 77 (1980) 5973.
[6] W.D.A. Rickard, R.P. Williams, J. Temuujin, A. van Riessen, Mater. Sci. Eng. A 528
(2011) 33903397.
[7] S. Detphan, P. Chindaprasirt, Int. J. Miner. Metall. Mater. 16 (2009) 720726.
[8] D. Kong, J. Sanjayan, K. Sagoe-Crentsil, Cem. Concr. Res. 37 (2007) 15831589.
[9] R.W. Goodwin, Combustion Ash/Residue Management: An Engineering Perspec-
tive, Noyes Publications, New Jersey, 1993.
[10] T. Matsunaga, J.K. Kim, S. Hardcastle, P.K. Rohatgi, Mater. Sci. Eng. A 325 (2002)
333343.
[11] N.W. Chen-Tan, A. van Riessen, V.L.Y. Chi, D.C. Southam, J. Am. Ceram. Soc. 92
(2009) 881887.
[12] A. Fernandez-Jimenez, A. Palomo, M. Criado, Cem. Concr. Res. 35 (2005) 12041209.
[13] R.P. Williams, A. van Riessen, Fuel 89 (2010) 36833692.
[14] W.D.A. Rickard, A. van Riessen, P. Walls, Int. J. Appl. Ceram. Technol. 7 (2010)
8188.
[15] K. Dombrowski, A. Buchwald, M. Weil, J. Mater. Sci. 42 (2007) 30333043.
[16] J.G.S. van Jaarsveld, J.S.J. van Deventer, G.C. Lukey, Mater. Lett. 57 (2003)
12721280.
[17] O.E. Manz, Fuel 78 (1999) 133136.
[18] T. Bakharev, Cem. Concr. Res. 36 (2006) 11341147.
[19] J. Temuujin, A. Minjigmaa, W. Rickard, M. Lee, I. Williams, A. van Riessen, J. Hazard.
Mater. 180 (2010) 748752.
[20] M. Rowles, B. O'Connor, J. Mater. Chem. 13 (2003) 1161.
[21] ASTM E831, Standard test method for linear thermal expansion of solid materials
by thermo mechanical analysis, , 2006.
[22] ASTM C39, Standard Test Method for Compressive Strength of Cylindrical
Concrete Specimens, , 2005.
[23] R.A. Fletcher, K.J.D. MacKenzie, C.L. Nicholson, S. Shimada, J. Eur. Ceram. Soc. 25
(2005) 14711477.
[24] J.L. Provis, C.Z. Yong, P. Duxson, J.S.J. van Deventer, Colloids Surf. A 336 (2009)
5763.
[25] P. Duxson, G.C. Lukey, J.S.J. van Deventer, J. Non-Cryst. Solids 352 (2006)
21862200.
[26] H. Wang, H. Li, F. Yan, Synthesis and mechanical properties of metakaolinite-
based geopolymer, Colloids Surf. A 268 (2005) 16.
[27] S. Andini, R. Ciof, F. Colangelo, T. Grieco, F. Montagnaro, L. Santoro, Waste
Manag. 28 (2008) 416423.
[28] J. Schairer, N. Bowen, Am. J. Sci. 254 (1956) 129195.
[29] W.A. Deer, R.A. Howie, J. Zussman, An Introduction to The Rock Forming Minerals.
2nd Edition, Essex, England, 1996.
[30] V.F. Barbosa, K.J. MacKenzie, Mater. Lett. 57 (2003) 14771482.
[31] G. Kovalchuk, P.V. Krivenko, in: J.L. Provis, J.S.J. van Deventer (Eds.), Geopolymers:
Structures, Processing, Properties and Industrial Applications, Woodhead,
Cambridge, UK, 2009, pp. 227266.
[32] H. Schneider, J. Schreuer, B. Hildmann, J. Eur. Ceram. Soc. 28 (2008) 329344.
[33] R.M. Cornell, U. Schwertmann, The Iron Oxides: Structure, Properties, Reactions,
Occurrence and Uses, VCH, Weinheim, 1996.
[34] P.S. Sidhu, Clays Clay Miner. 36 (1988) 3138.
1839 W.DA. Rickard et al. / Journal of Non-Crystalline Solids 358 (2012) 18301839