ORI GI NAL ARTI CLE

Microstructural changes in alkali activated y ash/slag

geopolymers with sulfate exposure
Idawati Ismail

Susan A. Bernal

John L. Provis

Sinin Hamdan

Jannie S. J. van Deventer
Received: 8 June 2012 / Accepted: 26 June 2012 / Published online: 11 July 2012
RILEM 2012
Abstract Sulfate attack is recognized as a signicant
threat to many concrete structures, and often takes place
in soil or marine environments. However, the under-
standing of the behavior of alkali-activated and geo-
polymer materials in sulfate-rich environments is
limited. Therefore, the aimof this study is to investigate
the performance of alkali silicate-activated y ash/slag
geopolymer binders subjected to different forms of
sulfate exposure, specically, immersion in 5 wt%
magnesium sulfate or 5 wt% sodium sulfate solutions,
for 3 months. Extensive physical deterioration of the
pastes is observedduringimmersionin MgSO
4
solution,
but not in Na
2
SO
4
solution. Calcium sulfate dihydrate
(gypsum) forms in pastes immersed in MgSO
4
, and its
expansive effects are identied as being particularly
damaging to the material, but it is not observed in
Na
2
SO
4
environments. Alower water/binder (w/b) ratio
leads to a greatly enhanced resistance to degradation
by sulfate attack. Infrared spectroscopy shows some
signicant changes in the silicate gel bonding environ-
ment of geopolymers immersed in MgSO
4
, attributed
mostly to decalcication processes, but less changes
upon exposure to sodium sulfate. It appears that the
process of sulfate attack on geopolymer binders is
strongly dependent on the cation accompanying the
sulfate, and it is suggested that a distinction should be
drawn between magnesiumsulfate attack (where both
Mg
2?
and SO
4
2-
are capable of inducing damage in the
structure), and general processes related to the presence
of sulfate accompanied by other, non-damaging cations.
The alkali-activated y ash/slag binders tested here are
susceptible to the rst of these modes of attack, but not
the second.
Keywords Geopolymer Fly ash Slag
Alkali-activated Sulfate attack
1 Introduction
The chemical mechanisms that can lead to degradation
of concretes include leaching of paste components,
carbonation of calcium hydroxide and calcium silicate
I. Ismail S. A. Bernal J. L. Provis (&)
J. S. J. van Deventer
Department of Chemical & Biomolecular Engineering,
University of Melbourne, Melbourne, VIC 3010, Australia
e-mail: j.provis@shefeld.ac.uk
I. Ismail
Department of Civil Engineering, Universiti Malaysia
Sarawak, Kota Samarahan, Sarawak, Malaysia
Present Address:
J. L. Provis
Department of Materials Science and Engineering,
University of Shefeld, Shefeld, UK
S. Hamdan
Department of Mechanical Engineering, Universiti
Malaysia Sarawak, Kota Samarahan, Sarawak, Malaysia
J. S. J. van Deventer
Zeobond Pty Ltd., PO Box 210, Somerton, VIC 3062,
Australia
Materials and Structures (2013) 46:361373
DOI 10.1617/s11527-012-9906-2
hydrate (CSH) if these are present in the binder, and
paste deterioration through exposure to aggressive
chemicals such as mineral acids, organic acids, and
sulfates. Sulfate attack on concrete is a major concern
in some civil infrastructure applications, and can be
either internal (due to sulfates present in the binder or
aggregates), or external (induced by the environment
into which the concrete is placed) [1, 2]. External
sulfate attack is generally caused by the exposure of
concrete to the alkali metal or alkaline earth sulfates
which are present in soils or groundwater. The rate
and mechanism of external sulfate attack on concrete
depend on many factors, including the cation accom-
panying the sulfate ions. Several factors can deter-
mine the resistance of concrete to sulfates, including
permeability [determined by water/binder (w/b) ratio,
mix design and curing], paste chemistry (binder type,
and the use of organic or mineral admixtures), and
aggregate mineralogy.
Reactions with sulfate can lead to cracking, spalling,
expansion and loss of strength in concrete structures.
Sodium sulfate (Na
2
SO
4
) reacts with calcium-contain-
ing binder components to formgypsum(calciumsulfate
dihydrate, CaSO
4
2H
2
O) and/or ettringite (Ca
6
Al
2
(SO
4
)
3
(OH)
12
26H
2
O) [1]. The reaction between hydrated
Portland cement and magnesium sulfate solution tends
to lead instead to the formation of magnesium hydrox-
ide (brucite, Mg(OH)
2
) and gypsum. Gollop and Taylor
[3] tested OPC paste under sulfate exposure, and found
that the deterioration of the material is greater in
MgSO
4
compared to when the same materials are
immersed in Na
2
SO
4
. More decalcication of CSH
takes place in the presence of Mg
2?
, which replaces
Ca
2?
to produce serpentine or magnesium silicate
hydrate (M-SH), which are much less stable than
CSH. The low pH environment of a brucite-saturated
solution also destabilizes ettringite and thus prevents its
formation [4].
The addition of slag to Portland cement systems is
well known to enhance resistance to sulfate attack [5].
The lowCa/Si ratio in blended slag cement contributes
to less formation of gypsum [6], and slag-blended
binders are also known to generate higher imperme-
ability and durability [7]. The effect of water content is
also important in determining the rate of sulfate attack
on concrete. Generally, an increase in w/b ratio
increases the susceptibility to sulfate attack due to the
greater porosity of the material, which enables more
rapid transport of ionic species, but mineralogical
aspects of the binder, and especially of the calcium-
rich components, can also be highly signicant [8].
The alkali activation of industrial byproducts to
form alkali-activated (including geopolymer) materi-
als is a growing eld of research and application at
present [912]. Alkali-activated binders and concretes
can be based on raw materials including y ash [13
15], ground granulated blast furnace slag [1618], or a
combination of both [1921], with the blended
materials showing some particularly attractive tech-
nical properties. Microstructural and chemical analy-
sis has provided good insight into the reaction
products [20, 2226] and some durability-related
properties [2731] of alkali activated y ash/slag
blends, but the study of sulfate exposure in these
systems has to date been limited. Alkali-activated y
ash was reported by Bakharev [32] to perform better
than Portland cement during exposure to Na
2
SO
4
and
MgSO
4
, with less physical deterioration and little or
no loss of compressive strength. Bakharev et al. [33]
also conducted a study of sulfate resistance in alkali
activated slag systems; those samples did not expand,
but showed visible cracks and traces of gypsum.
Ongoing binder formation reactions in the alkaline
Na
2
SO
4
environment resulted in an increase in
strength while the materials were immersed, while a
decrease in strength was observed in MgSO
4
[33].
From other published studies, alkali activated slag
binder systems are generally reported to have better
resistance to sulfate attack than OPC [3437]. Most of
these conclusions have been drawn from changes in
physical appearance and compressive strength, but the
mechanisms controlling this behavior on a micro-
structural level are not yet well understood, and the
analysis of sulfate resistance of blended y ash/slag
binders has not previously been reported in detail.
Therefore, this paper presents analysis of the chem-
istry and microstructure of geopolymers derived from
y ash and slag precursors, when exposed to sulfate
environments.
2 Materials and methods
2.1 Materials
Australian ground granulated blast furnace slag and y
ash, supplied by Zeobond Pty Ltd., were used as the
main binder components in the geopolymers, in a 1:1
362 Materials and Structures (2013) 46:361373
mass ratio in all binder formulations. Table 1 shows the
chemical composition of these raw materials. Sodium
metasilicate (Na
2
SiO
3
), added at 8 wt% with respect to
the solid binder components, was dissolved in tap water
for use as the activator. Pastes were produced with
w/b ratios of 0.40, 0.50 and 0.60, where the binder is
dened as the y ash ? slag content. The highest
w/b ratio, 0.60, is intended to demonstrate the effect of
very excessive water content on geopolymer durability.
The pastes were molded in 25 mm diameter polypro-
pylene tubes and sealed for curing under Malaysian
ambient conditions, approximately 25 5 C. After
28 days, the paste were demolded, cut to a length of
25 mm, and stored in covered containers (with
a separate container for each sample), immersed in
50 g/L MgSO
4
or Na
2
SO
4
solutions for 3 months at
ambient conditions. After 3 months, the samples were
manually crushed and dried in a desiccator with silica
gel for 24 h prior to analysis.
2.2 Analytical methods
X-ray diffraction (XRD) analysis was performed on a
Bruker D8 Advance instrument, scanning from 5 to
65 2h, with a 0.02 step size and 2 s/step count time.
Fourier transform infrared (FTIR) analysis was per-
formed using the KBr pellet method (1 mg sample per
100 mg KBr) on a Bruker Tensor 27 spectrometer,
with 32 scans per sample collected from 4,000 to
400 cm
-1
at 4 cm
-1
resolution.
Thermogravimetry was conducted using a PerkinEl-
mer Diamond instrument. Samples were crushed, trans-
ferred immediately to an alumina crucible, held under
isothermal conditions for 60 min at 40 C to equilibrate
in a nitrogen environment (N
2
owing at 200 mL/min),
and then heated to 1,000 Cat 10 C/min in the same gas
environment. Scanningelectronmicroscopy(SEM), with
energy dispersive X-ray (EDX) analysis, was performed
using a JEOL JSM6390La instrument. Samples were
coated with platinum prior to SEM analysis, and EDX
was performed at an accelerating voltage of 15 kV. The
specimens analyzed by SEM analysis were taken from
fractured pieces of the samples. The samples exposed to
MgSO
4
could not be polished because the material was
too prone to disintegration. Therefore, EDXspectra were
obtained from the fracture surfaces. For each sample
studied, spectra were collected at 30 random locations
across the sample including gel areas, unreacted precur-
sor particles, and in exposed samples, crystalline degra-
dation products and residual gel areas.
Analysis of the solutions at the end of the testing
period was performed using a Thermo Scientic iCE
3500 atomic absorption spectrometer (AAS) for Ca
2?
ion concentrations. Sulfate concentrations were ana-
lyzed using gravimetric analysis by mixing the sulfate
solution with 10 % barium chloride to precipitate
BaSO
4
, which was then ltered and weighed.
3 Results and discussion
3.1 Analysis of the raw materials
and as-synthesized binders
3.1.1 X-ray diffraction
Unreacted y ash and slag were analyzed by XRD,
as shown in Fig. 1. The raw y ash shows peaks
consistent with quartz (SiO
2
, PDF # 01-079-1910),
mullite (Al
6
Si
2
O
13
, PDF# 00-015-0776), hematite
(Fe
2
O
3
, PDF# 00-033-0664), and maghemite (Fe
2
O
3
,
PDF# 00-039-1346). There is also a broad feature
(amorphous hump) between 20 and 35 2h attributed
to the glassy component of the y ash. The diffrac-
togram of the unreacted slag shows a predominantly
amorphous material, with small but detectable quan-
tities of crystalline calcite (CaCO
3
, PDF# 01-071-
3699) and gypsum (CaSO
4
, PDF# 00-033-0311).
Mineralogical characterization of the reaction prod-
uct of alkali activation of the 1:1 y ash/slag geopoly-
mer paste (Fig. 2) again shows crystalline quartz (Q),
mullite (M) and maghemite (Mg) resulting from the
unreactive nature of these phase supplied by the y ash.
The reaction between y ash, slag and the sodium
Table 1 Chemical composition of raw materials by X-ray uorescence. LOI is loss on ignition at 1000 C
Oxide (wt%) SiO
2
Al
2
O
3
Fe
2
O
3
CaO MgO SO
3
Na
2
O K
2
O Other LOI
Fly ash 62.9 24.9 5.2 \0.1 1.0 0.2 0.2 1.3 1.7 2.6
Slag 33.8 13.7 0.4 42.6 5.3 0.8 0.1 0.4 1.1 1.8
Materials and Structures (2013) 46:361373 363
silicate activator results in amorphous to partially
crystalline phases. The main phase detected is a poorly
crystalline C(A)SH phase (similar to riversideite,
Ca
5
Si
6
O
16
(OH)
2
, PDF # 00-029-0329), and calcite is
again detectable. Slight differences were found as a
function of w/b ratio, with a low intensity peak
observed at approximately 10 2h at w/b 0.50, attrib-
uted to hydrotalcite (Mg
6
Al
2
(CO
3
)(OH)
16
4(H
2
O),
PDF# 00-041-1428)), which is known to be a product
of alkaline activation of slags [20, 38, 39]. This phase is
more prominent in this sample than in the others. The
intensity of the peak assigned to the main reaction
product, C(A)SH, decreases with increasing water
content, as the addition of water reduces the alkalinity
of the sodium metasilicate and thus inuences the
binder development [40, 41].
3.1.2 Fourier transform infrared spectroscopy
The FTIR spectra in Fig. 3 indicate major bands in
alkali activated y ash/slag systems at approximately
3460, 1650, 1470, 970, 870 and 454 cm
-1
. The
structure of molecular water in the alkali activated
y ash/slag system is characterized by the OH
stretching band, from 3,200 to 3,700 cm
-1
, while the
bending of the chemically bonded HOHis located at
1,650 cm
-1
. The carbonate in the system is charac-
terized by absorption at 1,470 and 870 cm
-1
, which is
consistent with the presence of calcite [42]
.
The main
binder gel band appears at 970 cm
-1
, assigned to the
asymmetric stretching mode of the SiOT (T:
tetrahedral Si or Al) bonds within the reaction
products. The position of this band is consistent with
both the C(A)SH structure formed by the activa-
tion of slag in alkaline media [20, 43, 44], and the
NASH gels formed in geopolymer binder systems
derived from y ash [45]. The coexistence of
geopolymeric NASH gel with C(A)SH gel in
blended alkali-activated binders was also investigated
by NMR spectroscopy in another publication by our
research group [46], and the ndings obtained from
this technique are consistent with the observations
presented here. The effect of w/b ratio on the gel
nanostructure as displayed by the FTIR spectra in
Fig. 3 is rather limited.
Fig. 1 X-ray diffractograms of the precursor materials
Fig. 2 XRD data for alkali activated slag/y ash geopolymers
as a function of w/b ratio
Fig. 3 FTIR spectra of alkali activated y ash/slag pastes at
different w/b ratios as marked
364 Materials and Structures (2013) 46:361373
In alkali-activated materials solely based on slag
[47, 48] or y ash [49], this vibration band has usually
been identied between 950 and 1,100 cm
-1
, and it is
typically associated with the binding gels (C(A)SH
for slag and NASH for y ash), with a lower
wavenumber indicating a depolymerized or more
highly substituted silicate gel, and higher wavenum-
bers being due to more crosslinked and highly
siliceous gels. The observation of the band at
relatively low wavenumber here can be attributed to
a reduced content of calcium in the C(A)SH
formed from the activation of the slag, and increased
incorporation of Al into this gel, as the availability of
this element increases in the systems by the dissolution
of the y ash [50]. A shoulder at 1,100 cm
-1
is also
identied, which is assigned to the quartz in y ash.
These results suggest that it is likely the gel structure
of the blended system studied is strongly dominated
by the reaction products derived from the activation
of the slag, consistent with the results obtained by
X-ray microtomographic analysis of 1:1 y ash/slag
blends [25].
3.1.3 Thermogravimetry (TGA/DTG)
Figure 4 shows the TGA proles for all w/b ratios,
showing mass loss between 7 and 9 % during the
measurement period. However, the evaporation of free
water from the samples during the 60 min isothermal
hold period prior to the start of heating accounted for
approximately 7, 10 and 15 % additional mass loss in
the samples with w/b 0.40, 0.50 and 0.60 respectively.
Thus, in total, the mass loss is between 14 and 21 %,
increasing with w/b ratio, and similar to that of
Portland cement paste which normally ranges between
15 and 20 % depending on w/b [51, 52]. The mass loss
proles begin to reach a plateau after approximately
700 C. Differential thermograms (DTG) of the pastes
are shown in Fig. 5, where detectable peaks occur at
various temperatures, marked as AC. The rst peak
(A) is observed at a temperature of 107 C for all
w/b ratios. This is consistent with the removal of free
evaporable water which is present in the pores of the
geopolymer gel products, either C(A)SH type or
NASH (zeolite-like) gels [53, 54]. The intensity
varies with w/b ratio, being lower for the pastes
formulated with lower water content, as expected due
to the lower amount of free water in these pastes. A
large shoulder just below 200 C is attributed to the
dehydration of the calcium-rich silicate gel [51]. At
the location marked with B, a weak mass loss effect at
270 C in the sample with w/b 0.60 is assigned to the
dehydration of hydrotalcite [5557]; this phase was
identied by XRD in all samples, and apparently with
the highest degree of crystallinity in the w/b 0.50
sample, but seems to be most prominent in the
w/b 0.60 sample in the DTGdata, which are inuenced
much less by differences in crystallinity. At location
C, approximately 685 C, there are weak, broad mass
loss peaks observed in all samples, attributed to the
decomposition of the carbonate minerals as identied
Fig. 4 Themogravimetric data for alkali activated y ash/slag
pastes formulated with different w/b ratios as marked
Fig. 5 Differential thermograms of alkali activated y ash/slag
pastes formulated with different w/b ratios as marked
Materials and Structures (2013) 46:361373 365
by FTIR. The decomposition temperature of calcite is
reported to be between 505 and 600 C [58], while the
higher decomposition temperature observed in the
geopolymer samples may indicate the presence of
different structural forms of carbonates, in particular
poorly crystalline or amorphous forms, in the paste.
3.2 Reaction with sulfates
3.2.1 Visual inspection of exposed specimens
Figures 6 and 7 show photographs of the three paste
samples with different w/b ratios, after immersion in
5 wt% MgSO
4
and Na
2
SO
4
solutions for 3 months.
No color change was observed in the sulfate solutions.
The appearance of the pastes exposed to Na
2
SO
4
shows no evident physical change regardless of
w/b ratio, as shown in Fig. 6. It is likely that the
geopolymer binder continues to stabilize and develop
in the presence of the Na
2
SO
4
solution (noting that
Na
2
SO
4
has also been used as an activator in alkali
activated slag systems [59]), and the relatively high
pH and high Na content are helpful in maintaining the
pore solution chemistry required for good stability of
the geopolymer samples here.
Conversely, for the samples exposed to MgSO
4
,
two distinct layers were observed in the samples, with
a less-damaged black layer visible close to the surface
in the w/b 0.40 sample, and present underneath the
outer layer in the other samples. In the w/b 0.40 and
0.50 samples, delamination of the exposed outer layer
produced scales of crystalline components which are
identiable as gypsum from electron microscopy
analysis, as will be discussed in Sect. 3.2.4. This
degradation is more notable at higher w/b ratio, so that
specimens formulated with a w/b ratio of 0.6 were
covered with a thick white layer of crystals, and
became very soft and disintegrated upon touching. The
difference between the effects of exposure to MgSO
4
or Na
2
SO
4
indicates that the interaction between the
magnesium sulfate solution and the binder actually
involves the magnesium (Mg
2?
) ion as a major
contributor to the binder degradation, rather than
being purely sulfate attack on the C(A)SH phase.
3.2.2 X-ray diffraction
Figure 8 shows the XRD patterns for the three pastes
after immersion in Na
2
SO
4
. The main phases that can
still be detected, as in Fig. 1, are the poorly crystalline
C(A)SH, mullite, quartz and maghemite. A small
quantity of hydrotalcite is still visible in the w/b 0.50
sample, although its peak intensity appears to be lower
than in the undamaged material. The calcite is still also
visible after exposure in Na
2
SO
4
, and the metastable
calcium carbonate polymorph vaterite (CaCO
3
, PDF#
00-033-0268) is also formed, probably resulting from
reactions involving atmospheric CO
2
, and consistent
with the fact that sulfate is known to favor the
stabilization of vaterite for longer periods of time [60].
Fig. 6 Physical condition of samples: A w/b 0.40; B w/b 0.50;
and C w/b 0.60, after immersion in Na
2
SO
4
for 90 days
Fig. 7 Physical condition of samples: A w/b 0.40; B w/b 0.50;
and C w/b 0.60, after immersion in MgSO
4
for 90 days
366 Materials and Structures (2013) 46:361373
The diffractograms in Fig. 9, however, present a
different picture when examining MgSO
4
attack on the
pastes. The outer, white layer of the paste samples was
taken for the XRD analysis. Gypsum crystals show a
much more intense presence in these systems, regard-
less of the w/b ratio. The C(A)SH type phase is not
observable in Fig. 9, and while calcite is still shown, it
is present at a greatly reduced intensity compared to
the unexposed samples. This suggests that decalci-
cation or decomposition of the C(A)SH phase in
the pastes has supplied the calcium required for the
formation of gypsum. Mg
2?
ions can also induce the
partial dissolution of calcite [61] which explains
the reduced intensity of this phase. Unlike in Portland
cement systems, as discussed in the introduction, no
ettringite was detected here, consistent with the low
Ca/Si ratio of these systems. The Al will be expected
to participate in the formation of NASH type gels,
and is thus less available for ettringite formation than
in Portland cement. Similar ndings for alkali acti-
vated slag mortars were reported by Bakharev [33],
and it is evident that MgSO
4
reacts more aggressively
than Na
2
SO
4
in exposure tests of alkali activated slag/
y ash geopolymers.
3.2.3 Fourier transform infrared spectroscopy
Figure 10 shows the FTIR spectra of the three pastes
after exposure to Na
2
SO
4
, where no distinct difference
is seen between the pastes formulated with different
w/b ratios. The water component at 3,4003,450 cm
-1
and the chemically bonded water at 1,650 cm
-1
were
not altered by the exposure to Na
2
SO
4
. The asymmet-
ric stretching vibration mode at 950970 cm
-1
also
shows no signicant shift after exposure. This suggests
that Na
2
SO
4
exposure does not signicantly alter the
chemical structures of the main binder products in the
y ash/slag geopolymer studied here.
Conversely, the reaction of the alkali activated y
ash/slag geopolymer with magnesium sulfate shows
marked decomposition of the C(A)SH phase in the
paste microstructure, and Fig. 11 shows distinct
differences between the spectra obtained for the
exposed specimens when compared with the unex-
posed specimens shown in Fig. 3. The decomposition
of the C(A)SH gel is associated with the
Fig. 8 X-ray diffractograms of samples exposed to Na
2
SO
4
as a
function of w/b ratio as marked
Fig. 9 X-ray diffractograms of samples exposed to MgSO
4
as a
function of w/b ratio as marked
Fig. 10 FTIRspectra of pastes exposed to Na
2
SO
4
as a function
of the w/b ratio as marked
Materials and Structures (2013) 46:361373 367
disappearance of the main SiOTband, along with the
formation of two new bands at 1,115 and 1,018 cm
-1
in the samples, which are consistent with the presence
of gypsum [62]. Gypsum also shows peaks around 600
and 665 cm
-1
[63], which are also visible in the spectra
in Fig. 11. Comparing the effect of the cations included
in the sulfate salt used for the attack tests, it is identied
that the presence of Mg
2?
promotes the decomposition
of the Ca-rich phases and therefore the erosion of
the paste. The disappearance of the calcite band at
1,470 cm
-1
, and appearance of a low intensity band
at 1,385 cm
-1
during MgSO
4
exposure, is also
consistent with the degradation of the binder, as the
new band is assigned to the presence of a hydrous
sodium carbonate compound [63]. This is likely to be
the result of the release of Na as the gel binder phases
are decomposed, which can then react with carbonates
provided either by dissolution of calcite, and/or from
the atmosphere.
3.2.4 Thermogravimetry (TGA/DTG)
Figures 12 and 13 show the TGA and DTG data
respectively, for the samples after 3 months of expo-
sure in Na
2
SO
4
. The mass loss values shown for the
samples are between 10 and 12 %, slightly higher than
for the unexposed samples (Fig. 4). Accounting for the
water loss in the isothermal hold period, the total mass
loss of each sample falls in the range of 1115 %,
which is similar to the unexposed samples. Decom-
position of phases is shown in the DTG plot, showing
the release of water between 110 and 135 C, which is
much higher compared to the unexposed samples, and
indicates that the Na
2
SO
4
solution may have had a
further activation effect as proposed above. The
sample with w/b 0.40 shows this peak at 115 C,
while the samples with w/b 0.50 and 0.60 show
asymmetric peaks at 135 C. The difference indicates
that there might be some additional gel formation or
evolution in the samples with w/b 0.50 and 0.60 that
increases this peak temperature, and this is consistent
with the increased intensity of the peak assigned to
C(A)SH type phases in Fig. 8 at higher w/b ratio,
which indicates a more ordered gel structure. Hydro-
talcite and carbonates are again observable, as noted in
Fig. 11 FTIRspectra of pastes exposed to MgSO
4
as a function
of the w/b ratio as marked
Fig. 12 Thermogravimetry curves of alkali-activated y ash/
slag geopolymers with w/b ratios as marked, after immersion in
Na
2
SO
4
for 3 months
Fig. 13 Differential thermograms of alkali-activated y ash/
slag geopolymers with w/b ratios as marked, after immersion in
Na
2
SO
4
for 3 months. The low-temperature region is expanded
in the inset
368 Materials and Structures (2013) 46:361373
Sect. 3.1.3, and in agreement with the XRD and FTIR
data.
Figures 14 and 15 show the TGA and DTG data for
the samples after exposure in MgSO
4
solutions for
3 months. The mass loss observed is much higher in
all samples after exposure to MgSO
4
, when compared
with undamaged pastes or specimens exposed to
Na
2
SO
4
. The mass loss is similar for all samples,
between 13 and 15 % during heating and 12 % during
the isothermal hold period, in each sample. The DTG
data in Fig. 15 show a sharper and more intense mass
loss peak at location (A), centered at 145 C, for all
samples exposed in MgSO
4
. From XRD and FTIR as
discussed above, it is evident that gypsum is the
dominant product in the samples exposed to MgSO
4
.
Gypsum starts to decompose between 110 and 150 C
[64, 65], indicating that this is the most likely
attribution for this peak, because based on the XRD
and FTIR results, the C(A)SH phase has been
largely removed from the samples through expo-
sure to MgSO
4
and is unlikely to contribute nota-
bly to the mass loss. Gypsum is known to dehydrate
via an intermediate calcium sulfate hemihy-
drate (CaSO
4
0.5H
2
O) phase, to anhydrite (CaSO
4
)
at 150200 C [66]. This phase transformation is
difcult to distinguish in the DTG data here due to
overlap and because of other phases present in the
paste. Peak B in Fig. 15, near 400 C, corresponds to
the presence of a small amount of sodium carbonates
as seen from FTIR and XRD.
3.2.5 Scanning electron microscopy
SEM images of samples after exposure to Na
2
SO
4
showed no signicant formation of any new distinct
phases or specic changes in morphology, and are
not presented here. However, the situation following
MgSO
4
exposure was quite different, and Fig. 16
shows an SEM micrograph of the sample with w/b 0.4
after exposure to MgSO
4
. This micrograph shows
formation of a gypsum crystal (marked as A),
embedded in a phase which is shown by EDX to be
rich in Mg, Si and Al (marked as B), and which
probably represents the residual, decalcied binder
gel. All SEM micrographs taken from the samples
with w/b 0.5 and 0.6 also after exposure in MgSO
4
showthe formation of gypsum, with very little residual
paste visible. EDX analysis of the delaminated pieces
fromthe samples also conrmed that these precipitates
were dominated by gypsum.
In general, as the sulfate attack progresses, the
C(A)SH decalcies, and therefore the Ca/Si molar
ratio decreases. Table 2 shows the elemental ratios
(Mg/Si and Ca/Si) calculated from SEM/EDX data
collected before and after sulfate exposure for the
samples with 0.40 and 0.50 w/b ratios. The sample
with w/b 0.60 was too damaged to be reliably
analyzed, and SEM/EDX analysis shows only gypsum
was formed in the sample. In general, the Ca/Si molar
ratios of the samples before sulfate exposure are quite
variable, with values of up to 2.0 observed at some
Fig. 14 Thermogravimetry curves of alkali-activated y ash/
slag blend geopolymers with w/b ratios as marked, after
immersion in MgSO
4
for 3 months
Fig. 15 Differential thermograms of alkali-activated y ash/
slag geopolymers with w/b ratios as marked, after immersion in
MgSO
4
for 3 months. The low-temperature region is expanded
in the inset
Materials and Structures (2013) 46:361373 369
points on the sample, with a mean of 0.50. In alkali
activated slag systems without y ash, this mean Ca/Si
ratio tends to be higher, 0.801.2 [67], depending on
the choice of activator and the composition of the raw
materials. The inclusion of y ash in this study has
reduced this ratio in the geopolymer paste product.
This ratio is also slightly lower than observed in a
binder formulated with a 1:1 ratio of y ash and slag
activated using sodium hydroxide [20], due to the
additional silica added in the activator. The sample
exposed in MgSO
4
generally has a lower Ca/Si ratio in
the residual binder, with a mean of around 0.2,
compared to the sample exposed in Na
2
SO
4
, which has
maintained a mean of 0.5. This shows that there is no
notable decalcication taking place in Na
2
SO
4
.
3.3 Sulfate solution analysis
Table 3 shows the measured Ca
2?
and SO
4
2-
con-
centrations in both sulfate solutions at the end of the
testing period. In almost all cases, the sulfate concen-
tration in the solution has been reduced from its initial
value, which is attributed to precipitation during the
exposure test. The slight increase in one sample may
have been due to release of sulfate (and or release and
later oxidation of sulde) from the slag particles
present in the sample. The reduction is much more
marked in the MgSO
4
solutions than in Na
2
SO
4
. The
highest measured Ca
2?
concentration in the sulfate
solutions is associated with the sample with w/b 0.60
exposed in MgSO
4
. This is consistent with the high
porosity of this sample, which eases the migration of
ionic species through the pore structure of the sample
and gives the highest extent of degradation during the
test duration. In this process, Mg
2?
enters the sample
and decalcies the C(A)SH phase, leading to the
release of Ca
2?
. The Ca
2?
concentrations measured in
the solutions will not represent the full extent of
Fig. 16 SEM micrograph of the sample with w/b 0.40 after exposure to MgSO
4
, showing (A) gypsum crystals and (B) MgAlSi-rich
gel
Table 2 Mean Ca/Si and Mg/Si atomic ratios before and after
MgSO
4
exposure
Sample
w/b ratio
Atomic ratio
Ca/Si Mg/Si
Before
exposure
After
exposure
Before
exposure
After
exposure
0.40 0.50 0.20 0.35 0.83
0.50 0.50 0.20 0.20 0.72
0.60 0.48 Not tested 0.18 Not tested
370 Materials and Structures (2013) 46:361373
leaching of this ion from the geopolymer materials, as
the gypsum precipitation has removed much of the
calcium from solution, as seen by the reduction in
sulfate concentration compared to the initial solution
compositions. The degradation of the C(A)SH
phase in the presence of MgSO
4
is faster than in
Na
2
SO
4
due to the low water solubility of Mg(OH)
2
and the low pH of the solution in equilibrium with this
phase [68].
However, the pattern in the Ca
2?
concentrations
when the samples were exposed in Na
2
SO
4
is reversed,
and the Ca
2?
concentration is higher in lower
w/b samples. The reaction with Na
2
SO
4
appears to
lead to ion exchange with Ca
2?
compounds during the
gel evolution, releasing some Ca
2?
into the solution
without any further reaction to form gypsum. This also
agrees with the XRD and FTIR analysis, where the
CSH phase is maintained in the paste. The results
obtained here correspond well with the discussion of
Bellmann et al. [69], who mentioned that if the alkali
concentration is too high, gypsum will not precipitate,
but alkali-calcium sulfate double salts will form
instead. In alkaline solutions such as Na
2
SO
4
, the
extra alkalinity supplied by the geopolymer pore
solution, which is highly alkaline [70], will thus
prevent the formation of gypsum.
4 Conclusions
The structural evolution of alkali activated y ash/slag
geopolymer pastes exposed to sodium and magnesium
sulfate environments has been investigated. The
different sulfate salt solutions react differently with
the geopolymer paste; magnesium sulfate is more
aggressive to geopolymer paste than sodium sulfate.
The presence of magnesium leads to decalcication of
the Ca-rich gel phases present in the blended ash/slag
geopolymer system, causing degradation of the binder
system and the precipitation of gypsum. The products
of magnesium sulfate attack are poorly cohesive and
expansive, leading to dimensional instability and a
loss of mechanical performance. Conversely, immer-
sion of geopolymer pastes in Na
2
SO
4
does not lead to
any apparent degradation of the binder, and no
conversion of the binder phase components into
sulfate-containing precipitates is observable. The
w/b ratio also appears to contribute to the resistance
to degradation, as the renement of the pore system of
the geopolymer paste at low w/b densies the binder
and reduces the rate of attack on the binder system.
The results obtained here provide strong indications
that the key factor which determines the rate and
effects of the processes commonly described as
sulfate attack in geopolymer systems is actually the
nature of the cation which accompanies the sulfate
anion. Magnesium sulfate can have severely deleteri-
ous effects on these materials, while sodium sulfate at
the same concentration appears to have little or no
effect on the structure of the material.
Acknowledgments The authors acknowledge the Malaysian
Ministry of Higher Education for funding of this project through
the Melbourne-Malaysia Split Ph.D. program. This study has also
been supported by the Australian Research Council, including
some support through the Particulate Fluids Processing Centre,
through a Linkage Project grant co-sponsored by Zeobond
Research, and via the Discovery Grants program. We thank
Dr. Zainal Abidin Talib from Department of Physics, Universiti
Putra Malaysia (UPM), Selangor, Malaysia for the XRD of raw
materials and undamaged samples.
References
1. Neville A (2004) The confused world of sulfate attack on
concrete. Cem Concr Res 34(8):12751296
2. Cohen MD, Mather B (1991) Sulfate attack on concrete.
Research needs. ACI Mater J 88(1):6269
3. Gollop RS, Taylor HFW (1992) Microstructural and
microanalytical studies of sulfate attack. I. Ordinary Port-
land cement paste. Cem Concr Res 22(6):10271038
4. Aye T, Oguchi CT (2011) Resistance of plain and blended
cement mortars exposed to severe sulfate attacks. Constr
Build Mater 25(6):29882996
5. Gollop RS, Taylor HFW (1996) Microstructural and
microanalytical studies of sulfate attack. IV. Reactions of a
slag cement paste with sodium and magnesium sulfate
solutions. Cem Concr Res 26(7):10131028
Table 3 Chemical analysis of sulfate solutions after leaching
Sample w/b ratio Sulfate type Concentration (g/L)
Ca
2?
SO
4
2-
0.40 Sodium 2.78 23.6
Magnesium 1.26 12.2
0.50 Sodium 1.48 34.0
Magnesium 1.70 9.26
0.60 Sodium 1.29 28.5
Magnesium 2.79 10.3
Sulfate concentrations before leaching (5 wt% salt solutions):
Na
2
SO
4
33.8 g/L, MgSO
4
39.9 g/L
Materials and Structures (2013) 46:361373 371
6. Gollop RS, Taylor HFW (1996) Microstructural and
microanalytical studies of sulfate attack. V. Comparison of
different slag blends. Cem Concr Res 26(7):10291044
7. Elahi A, Basheer PAM, Nanukuttan SV, Khan QUZ (2010)
Mechanical and durability properties of high performance
concretes containing supplementary cementitious materials.
Constr Build Mater 24(3):292299
8. Sahmaran M, Kasap O, Duru K, Yaman IO(2007) Effects of
mix composition and watercement ratio on the sulfate
resistance of blended cements. Cem Concr Compos 29(3):
159167
9. Provis JL, van Deventer JSJ (eds) (2009) Geopolymers:
structure, processing, properties and industrial applications.
Woodhead, Cambridge
10. Juenger MCG, Winnefeld F, Provis JL, Ideker J (2011)
Advances in alternative cementitious binders. Cem Concr
Res 41(12):12321243
11. van Deventer JSJ, Provis JL, Duxson P (2012) Technical
and commercial progress in the adoption of geopolymer
cement. Miner Eng 29:89104
12. Shi C, Krivenko PV, Roy DM (2006) Alkali-activated
cements and concretes. Taylor & Francis, Abingdon
13. Fernandez-Jimenez A, Palomo A, Criado M (2006) Alkali
activated y ash binders. A comparative study between
sodium and potassium activators. Mater Constr 56(281):
5165
14. Criado M, Palomo A, Fernandez-Jimenez A, Banll P
(2009) Alkali activated y ash: effect of admixtures on paste
rheology. Rheol Acta 48(4):447455
15. Fernandez-Jimenez A, Palomo A, Criado M (2005) Micro-
structure development of alkali-activated y ash cement: a
descriptive model. Cem Concr Res 35(6):12041209
16. Douglas E, Bilodeau A, Brandstetr J, Malhotra VM (1991)
Alkali activated ground granulated blast-furnace slag concrete:
preliminary investigation. Cem Concr Res 21(1):101108
17. Fernandez-Jimenez A, Palomo JG, Puertas F (1999) Alkali-
activated slag mortars: mechanical strength behaviour. Cem
Concr Res 29(8):13131321
18. Brough AR, Atkinson A (2002) Sodium silicate-based,
alkali-activated slag mortars: Part I. Strength, hydration and
microstructure. Cem Concr Res 32(6):865879
19. Kumar S, Kumar R, Mehrotra S (2010) Inuence of gran-
ulated blast furnace slag on the reaction, structure and
properties of y ash based geopolymer. J Mater Sci 45(3):
607615
20. Puertas F, Fernandez-Jimenez A (2003) Mineralogical and
microstructural characterisation of alkali-activated y ash/
slag pastes. Cem Concr Compos 25(3):287292
21. Escalante Garcia JI, Campos-Venegas K, Gorokhovsky A,
Fernandez A (2006) Cementitious composites of pulverised
fuel ash and blast furnace slag activated by sodium silicate:
effect of Na
2
O concentration and modulus. Adv Appl
Ceram 105(4):201208
22. Smith MA, Osborne GJ (1977) Slag/y ash cements. World
Cem Technol 1(6):223233
23. Puertas F, Mart nez-Ram rez S, Alonso S, Vazquez E(2000)
Alkali-activated y ash/slag cement. Strength behaviour and
hydration products. Cem Concr Res 30:16251632
24. Lloyd RR, Provis JL, van Deventer JSJ (2009) Microscopy
and microanalysis of inorganic polymer cements. 2: the gel
binder. J Mater Sci 44(2):620631
25. Provis JL, Myers RJ, White CE, Rose V, van Deventer JSJ
(2012) X-ray microtomography shows pore structure and
tortuosity in alkali-activated binders. Cem Concr Res
42(6):855864
26. Izquierdo M, Querol X, Phillipart C, Antenucci D, Towler
M (2010) The role of open and closed curing conditions on
the leaching properties of y ash-slag-based geopolymers.
J Hazard Mater 176(13):623628
27. Sugama T, Brothers LE, Van de Putte TR (2005) Acid-
resistant cements for geothermal wells: sodium silicate
activated slag/y ash blends. Adv Cem Res 17(2):6575
28. Goretta KC, Chen N, Gutierrez-Mora F, Routbort JL, Lukey
GC, van Deventer JSJ (2004) Solidparticle erosion of a
geopolymer containing y ash and blast-furnace slag. Wear
256(78):714719
29. Gordon LE, Provis JL, van Deventer JSJ (2011) Durability
of y ash/GGBFS based geopolymers exposed to carbon
capture solvents. Adv Appl Ceram 110(8):446452
30. Lloyd RR, Provis JL, van Deventer JSJ (2012) Acid resis-
tance of inorganic polymer binders. 1. Corrosion rate. Mater
Struct 45(12):114
31. Guerrieri M, Sanjayan JG (2010) Behavior of combined y
ash/slag-based geopolymers when exposed to high tem-
peratures. Fire Mater 34(4):163175
32. Bakharev T (2005) Durability of geopolymer materials in
sodium and magnesium sulfate solutions. Cem Concr Res
35(6):12331246
33. Bakharev T, Sanjayan JG, Cheng YB(2002) Sulfate attack on
alkali-activated slag concrete. CemConcr Res 32(2):211216
34. Rodriguez E, Bernal S, Mej a de Gutierrez R, Puertas F
(2008) Alternative concrete based on alkali-activated slag.
Mater Constr 58(291):5367
35. Deja J, Malolepszy J (2003) Some aspects of alkali activated
slag binders durability. Ann Chimie 28:5158
36. Kukko H, Mannonen R (1982) Chemical and mechanical
properties of alkali-activated blast furnace slag (F-con-
crete). Nord Concr Res 1:16.1-16.16
37. Douglas E, Bilodeau A, Malhotra VM(1992) Properties and
durability of alkali-activated slag concrete. ACI Mater J
89(5):509516
38. Oh JE, Monteiro PJM, Jun SS, Choi S, Clark SM(2011) The
evolution of strength and crystalline phases for alkali-acti-
vated ground blast furnace slag and y ash-based geopoly-
mers. Cem Concr Res 40(2):189196
39. Ben Haha M, Lothenbach B, Le Saout G, Winnefeld F
(2011) Inuence of slag chemistry on the hydration of
alkali-activated blast-furnace slag - Part I: effect of MgO.
Cem Concr Res 41(9):955963
40. Krizan D, Zivanovic B (2002) Effects of dosage and mod-
ulus of water glass on early hydration of alkali-slag cements.
Cem Concr Res 32(8):11811188
41. Altan E, Erdogan ST (2012) Alkali activation of a slag at
ambient and elevated temperatures. Cem Concr Compos
34(2):131139
42. Mozgawa W, Deja J (2009) Spectroscopic studies of alka-
line activated slag geopolymers. J Mol Struct 924926:
434441
43. Puertas F, Palacios M, Manzano H, Dolado JS, Rico A,
Rodr guez J (2011) Amodel for the CASHgel formed in
alkali-activated slag cements. J Eur Ceram Soc 31(12):
20432056
372 Materials and Structures (2013) 46:361373
44. Garcia-Lodeiro I, Palomo A, Fernandez-Jimenez A, Mac-
Phee DE (2011) Compatibility studies between NASH
and CASH gels. Study in the ternary diagram Na
2
O
CaOAl
2
O
3
SiO
2
H
2
O. Cem Concr Res 41(9):923931
45. Rees CA, Provis JL, Lukey GC, van Deventer JSJ (2007)
Attenuated total reectance Fourier transform infrared
analysis of y ash geopolymer gel aging. Langmuir 23(15):
81708179
46. Provis JL, Bernal SA, Walkley B, San Nicolas R, Gehman
JD, Brice DG, Kilcullen AR, Duxson P, van Deventer JSJ
(2012) Accelerated carbonation testing of alkali-activated
y ash and y ash/slag binders: nanostructural and micro-
structural effects. To be submitted
47. Fernandez-Jimenez A, Puertas F, Sobrados I, Sanz J (2003)
Structure of calcium silicate hydrates formed in alkaline-
activated slag: inuence of the type of alkaline activator.
J Am Ceram Soc 86(8):13891394
48. Bernal SA, Provis JL, Mej a de Gutierrez R, Rose V (2011)
Evolution of binder structure in sodium silicate-activated
slag-metakaolin blends. Cem Concr Compos 33(1):4654
49. Fernandez-Jimenez A, Palomo A (2005) Mid-infrared
spectroscopic studies of alkali-activated y ash structure.
Microporous Mesoporous Mater 86(13):207214
50. Hajimohammadi A, Provis JL, van Deventer JSJ (2011)
Time-resolved and spatially-resolved infrared spectroscopic
observation of seeded nucleation controlling geopolymer
gel formation. J Colloid Interf Sci 357(2):384392
51. Alarcon-Ruiz L, Platret G, Massieu E, Ehrlacher A (2005)
The use of thermal analysis in assessing the effect of tem-
perature on a cement paste. Cem Concr Res 35(3):609613
52. Chaipanich A, Nochaiya T (2010) Thermal analysis and
microstructure of Portland cementy ashsilica fume
pastes. J Therm Anal Calorim 99(2):487493
53. Nochaiya T, Wongkeo W, Pimraksa K, Chaipanich A
(2010) Microstructural, physical, and thermal analyses of
Portland cementy ashcalcium hydroxide blended pastes.
J Therm Anal Calorim 100(1):101108
54. Bernal SA, Rodr guez ED, Mej a de Gutierrez R, Gordillo
M, Provis JL (2011) Mechanical and thermal characterisa-
tion of geopolymers based on silicate-activated metakaolin/
slag blends. J Mater Sci 46(16):54775486
55. Vagvolgyi V, Palmer SJ, Kristof J, Frost RL, Horvath E
(2008) Mechanism for hydrotalcite decomposition: a con-
trolled rate thermal analysis study. J Colloid Interf Sci 318(2):
302308
56. Pesic L, Salipurovic S, Markovic V, Vucelic D, Kagunya
W, Jones W (1992) Thermal characteristics of a synthetic
hydrotalcite-like material. J Mater Chem 2(10):10691073
57. Ben Haha M, Le Saout G, Winnefeld F, Lothenbach B
(2011) Inuence of activator type on hydration kinetics,
hydrate assemblage and microstructural development of
alkali activated blast-furnace slags. Cem Concr Res 41(3):
301310
58. Frost RL, Hales MC, Martens WN (2009) Thermogravi-
metric analysis of selected group (II) carbonate minerals
implication for the geosequestration of greenhouse gases.
J Therm Anal Calorim 95(3):9991005
59. Rashad AM, Bai Y, Basheer PAM, Collier NC, Milestone
NB (2012) Chemical and mechanical stability of sodium
sulfate activated slag after exposure to elevated tempera-
ture. Cem Concr Res 42(2):333343
60. Fernandez-D az L, Fernandez-Gonzalez A

, Prieto M (2010)
The role of sulfate groups in controlling CaCO
3
polymor-
phism. Geochim Cosmochim Acta 74(21):60646076
61. Ruiz-Agudo E, Putnis CV, Jimenez-Lopez C, Rodriguez-
Navarro C (2009) An atomic force microscopy study of
calcite dissolution in saline solutions: the role of magnesium
ions. Geochim Cosmochim Acta 73(11):32013217
62. Farmer VC (1974) The infrared spectra of minerals. Bar-
tholomew Press, London
63. Gadsden JA (1975) Infrared spectra of minerals and related
inorganic compounds. Butterworths, London
64. Borrachero MV, Paya J, Bonilla M, Monzo J (2008) The use
of thermogravimetric analysis technique for the character-
ization of construction materials: the gypsum case. J Therm
Anal Calorim 91(2):503509
65. Paulik F, Paulik J, Arnold M(1992) Thermal decomposition
of gypsum. Thermochim Acta 200:195204
66. Yu QL, Brouwers HJH (2012) Thermal properties and
microstructure of gypsum board and its dehydration prod-
ucts: a theoretical and experimental investigation. Fire
Mater. doi:10.1002/fam.1117
67. Gruskovnjak A, Lothenbach B, Holzer L, Figi R, Winnefeld
F (2006) Hydration of alkali-activated slag: comparison
with ordinary Portland cement. Adv Cem Res 18(3):
119128
68. Skalny J, Marchand J, Odler I (2002) Sulfate attack on
concrete. Modern concrete technology. Spon Press, New
York
69. Bellmann F, Moser B, Stark J (2006) Inuence of sulfate
solution concentration on the formation of gypsumin sulfate
resistance test specimen. Cem Concr Res 36(2):358363
70. Lloyd RR, Provis JL, van Deventer JSJ (2010) Pore solution
composition and alkali diffusion in inorganic polymer
cement. Cem Concr Res 40(9):13861392
Materials and Structures (2013) 46:361373 373
Reproducedwith permission of thecopyright owner. Further reproductionprohibited without permission.