You are on page 1of 7

Construction and Building Materials 23 (2009) 2951–2957

Contents lists available at ScienceDirect

Construction and Building Materials


journal homepage: www.elsevier.com/locate/conbuildmat

Mechanical and microstructural characterization of an alkali-activated


slag/limestone fine aggregate concrete
Aaron R. Sakulich *, Edward Anderson, Caroline Schauer, Michel W. Barsoum
Department of Materials Science and Engineering, Drexel University, Philadelphia, PA, United States

a r t i c l e i n f o a b s t r a c t

Article history: Four limestone-based, alkali-activated slag fine aggregate concretes, two of which contained amorphous
Received 7 October 2008 silica in the form of diatomaceous earth, were fabricated using different activating solutions (NaOH/
Received in revised form 5 January 2009 waterglass or Na2CO3). Emphasis in this work was placed on using simple manufacturing methods and
Accepted 8 February 2009
widely available materials, to ensure that these formulae are practical as construction materials in the
Available online 14 March 2009
developing world. Although cured only at room temperature, these fine aggregate concretes have good
compressive strengths (45 MPa) and their tensile strengths increased from 2.6 MPa after 1 day of cur-
Keywords:
ing to 4 MPa after 28 day for the NaOH-activated formulae. Samples activated with Na2CO3 had negli-
Slag
Alkali-activated cement
gible tensile strengths after 1 day, increasing to 2.5 MPa after 28 day. The main cementing phase was
Environment shown to be calcium–silicate–hydrates in all formulae; those activated with Na2CO3 also showed the
C–S–H presence of hydrotalcite. No evidence of geopolymeric phases was found, though incorporation of Na
to form N–S–H that balance charges arising from Al substitution of Si in C–S–H is likely. Despite the short
(120 s) pot life of the strongest formula, NaCl was shown to be an effective retarding agent, which
reduced the strengths of different formulae, at worst, by less than 25% after 28 day of curing.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction There is some debate as to the exact nature of the reaction that
is occurring within AAS cements and concretes. Early reports [7]
Cements and concretes made of alkali-activated slag (AAS) were detailed the presence of calcium–silicate–hydrates (C–S–H),1 one
originally investigated in the former Soviet Union, Scandinavia, and of the main phases in OPC cement pastes, as well as a variety of sec-
Eastern Europe as early as the 1940s. A recent revival of interest ondary minerals. However, it has been reported that the C–S–H gel
[1–4] has shown that as a partial replacement [5] in ordinary Port- formed in AAS cements has a lower Ca/Si ratio than that found in
land cement (OPC) the slag, which contains high amounts of lime, OPC [5,13–16]. Later studies reported calcium–aluminate phases
silica, and alumina, is activated by the alkaline conditions arising [7], also commonly found in OPC, while others found zeolitic struc-
from cement hydration [2]. Some effort has gone into developing tures [1]. Consequently, opinion broadly falls into two groups: the
cements based solely on slag activated directly by alkali solutions alkaline-cement group and the alkali-activated (or geopolymer)
[1–4,6–9] – without the presence of OPC – focusing on NaOH, group. The former believes that after activation the slag forms prod-
waterglass, or Na2CO3 as activators [2]. Others use a two-step pro- ucts similar to those in OPC (C–S–H, etc.) and that the alkali ions
cess with both NaOH and waterglass [4]. (Na+, K+, Cs+, etc.) either move to the pore solution, take the place
The development and properties of AAS cements have been re- of Ca in C–S–H, or balance charges arising from Al replacement of
viewed in a number of papers [1–3], with the general consensus Si in C–S–H; they play no widespread structural role [15–18]. The
that alkali activation produces high early strength and quick set- geopolymer group believes that alkali ions are incorporated into
ting [2]. However, shrinkage, cracking during drying [10,11], and highly connected, 3-dimensional aluminosilicate structures and play
short pot lives [2] provide not inconsequential hurdles to be over- a critical role in charge balancing. The term ‘geopolymer’ is some-
come. A major benefit of AAS cements is that they produce very lit- times used broadly to describe any alkali-activated system (includ-
tle greenhouse gas compared with OPC, which relies on the ing those that produce C–S–H); however, it is used here solely to
calcination of limestone and produces around 5% of worldwide describe a highly connected alkali aluminosilicate network.
greenhouse gas emissions [12].

1
Cement chemists use single letters to represent oxides: C = CaO, S = SiO2,
A = Al2O3, F = Fe2O3, H = H2O and so on. Calcium–silicate–hydrates are abbreviated
* Corresponding author. C–S–H because a range of compositions are possible, and using exact stoichiometries
E-mail address: ars27@drexel.edu (A.R. Sakulich). would be inaccurate.

0950-0618/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.conbuildmat.2009.02.022
2952 A.R. Sakulich et al. / Construction and Building Materials 23 (2009) 2951–2957

There is evidence to support both sides [1,11,13,14,17–21], and XRD diffractograms of the slag showed only a broad hump centered at roughly
30°–2h; no crystalline impurities were detected. The same analysis on DE showed a
what is actually occurring is likely to be somewhere in between. In
somewhat less broad hump centered around 23°–2h; a pair of small peaks around
fact, recently a binder containing both geopolymer- and C–S–H 28°–2h were unidentifiable, however, in the SEM occasional particles of an Al-rich
products has been produced through the alkali activation of a impurity were observed.
metakaolin/blast furnace slag mixture [22]. Powder reactants (slag, aggregate, DE, and NaCl) were combined in 2 l nalgene
This study focuses on AAS-based fine aggregate concretes incor- bottles and mixed on a rotary mixer at 28 rpm for 5 min while NaOH and water
were mixed separately on a stir plate for 5 min. After the powders were mixed,
porating a fine limestone aggregate and activated by either a two-
waterglass was added (formulae A and B only) and mixed by hand for 3 min.
step addition of NaOH solution/waterglass (waterglass first, then The NaOH solution was then added and mixed using a mechanical mixer for
NaOH solution) or a one-step addition of a Na2CO3 solution. In 2 min.
our preliminary work developing geopolymer-like materials [23], Formulae C and D did not use waterglass. In formula C, Na2CO3 solution (made
by mixing Na2CO3 and water on a stir plate for 5 min, to a pH of 11.3) was added
limestone proved a better aggregate material for making fine
to the powders and mixed using a mechanical mixer for 3 min. In formula D, a DE/
aggregate concretes than sand, and is thus used here. Amorphous Na2CO3 solution was added to the dry powders and mixed by hand for 2 min.
silica in the form of diatomaceous earth (DE) was also used in Then, more Na2CO3 solution was added and mixed using the mechanical mixer
one formula along with Na2CO3 to attempt to replace the water- for an additional 3 min. Exact amounts of reactants used in each formula can be
glass, which is regarded as the best activator [2]. found in Table 1.
After mixing, the slurries were poured into 5  10 cm cylindrical polypropylene
All formulae were also investigated with the addition of NaCl,
molds (Jatco Industries, Union City, CA), then sealed with plastic wrap and placed
which has been shown to be an effective retarder [3] that works on a shelf to cure at ambient conditions (20 ± 5 °C, 60 ± 10% relative humidity) for
by lowering the pH of the system, thus slowing the reaction. 1, 7 or 28 days before testing.
Though any industrially available superplasticizer would work as After curing, samples were removed from the molds and tested in groups of six
on a compression tester (Instron 5700, Norwood, MA). Three samples were tested in
a retardant, NaCl was chosen here to minimize the ‘carbon foot-
standard compression format (ASTM C39); the other three samples were tested to
print’ of the formulae. NaCl is available all over the world, thus determine their splitting tensile properties (ASTM C496). When needed, the ends of
making it ideal if AAS materials are to be used for construction in the samples were ground to ensure parallel surfaces.
impoverished regions. While it may even be possible to use seawa- Powders for X-ray diffraction (XRD) (Siemens D500, Karlsruhe, Germany)
ter in the activating solution, the salt content would likely be too experiments were obtained by drilling into the bulk of the AAS paste samples.
While examining the pure paste (which contained less water) may have affected
low, and other impurities (i.e. organic material) may have deleteri-
hydration, the effect of this was negligible. The XRD scans (Cu Ka source,
ous effects. Using seawater would be especially beneficial in arid k = 0.154 nm) were run with a step size of 0.02° and a dwell time of 1 s between
coastal regions, but remains speculative and was not investigated 10° and 90°–2h. As no peaks appeared on the diffractograms above 70°–2h, that data
here. While we are fully cognizant of the deleterious effect of NaCl was cropped out of the figures.
on rebar, not all construction, especially in the third world, is rein- For SEM analysis, samples were mounted in epoxy, polished, and sputter coated
with a Pt/Pd layer (Cressington Scientific Instruments, Watford, England). A scan-
forced with steel; further, previous research [3] has shown that ning electron microscope, SEM, (Zeiss Supra 50VP, Thornwood, NY) with an electron
although the addition of NaCl slows hydration, it does not alter energy dispersive spectrometer, EDS, (Oxford Inca X-Sight, Oxfordshire, UK), was
the final products and mostly recrystalizes in the pore solution. used for elemental analysis. Image analysis was performed using ImageJ software,
Therefore strength deterioration after extended periods is unlikely, averaging the results from four different SEM micrographs (500 magnification,
20 KV beam strength, 30 lm aperture, 12 mm working distance, 1024  768
even if NaCl is leached out of the fine aggregate concrete.
resolution).
Mechanical properties, in the form of compressive and splitting Fast Fourier infrared spectroscopy, FTIR, experiments were performed using a
tensile strengths, weight loss and setting time, as well as micro- Varian Excalibur FTS 3000MX microspectrometer. 10 mg of powdered sample was
structural characteristics determined through elemental analysis, mixed with 300 mg of KBr and compressed for 5 min with 9 tonnes of force to cre-
scanning electron microscopy, infrared spectroscopy, and X-ray ate a solid pellet approximately 1 mm thick. Spectroscopy experiments were run in
transmission mode using a pellet holding accessory and the resulting spectra were
diffraction are presented. smoothed to facilitate peak identification using the 10-point adjacent averaging
method in Origin PRO 7.0 software. Identification of peaks was based on the liter-
2. Experimental procedure ature [24,25].

The reactants used for the fine aggregate concrete were: NaOH pellets and NaCl
(Alfa Aesar, Ward Hill, MA), DE (PermaGuard, Inc., Albuquerque, NM), type N solu-
3. Results and discussion
ble silicate (PQ Corporation, Valley Forge, PA), anhydrous sodium carbonate, Na2CO3
(Dharma Trading Co., San Rafael, CA) and granulated ground blast furnace slag (St.
Lawrence Cement Co., Camden, NJ). The Blaine number of the latter was 5585 cm2/ Formulae A and B (Table 1) had smooth, glossy surfaces while C
g. The limestone aggregate (OldCastle Stone Products, Charlotte, NC, 91% calcium and D had dull, rough surfaces, regardless of NaCl content. Formula
carbonate equivalent) was obtained from a local home improvement store. Sieving
A was dark green, bordering on black, becoming lighter with the
showed the following particle size distribution: 23 wt.% below 75 lm, 24 wt.% be-
tween 75 lm and 150 lm, 19 wt.% between 150 lm and 300 lm and 32 wt.% above addition of NaCl. Formula B was a lighter shade of green, with a
300 lm. As determined by X-ray fluorescence and reported by the manufacturer, beige band at the top surface; the addition of NaCl resulted in
the slag composition was: 42.1% CaO, 34.6% SiO2, 11.7% Al2O3, 6.6% MgO, 1.3% the appearance of beige patches throughout the body of the sam-
SO2, 1.0% TiO2, and 0.51% Fe2O3. ple. Formula C had a similar band at the top surface, with a body
that varied from light green to beige. The addition of NaCl resulted
in a uniform light green color throughout. Formula D had bands of
darker green and beige in a marble-like pattern. The addition of
NaCl had no effect on its appearance. As the unreacted slag powder
Table 1
Composition of four formulae investigated, in grams. Samples with NaCl as a is light gray in color, the reason for the dark green coloration is un-
retarding agent also contain 79.3 g NaCl. clear. The beige bands at the tops of the samples could be the result
of carbonation due to atmospheric exposure in the case of inade-
A (g) B (g) C (g) D (g)
quately sealed molds; however, as formulae C and D had beige col-
Slag 396.8 396.8 396.8 396.8
ors throughout their bodies, and not solely at the sealed end, this is
Limestone 480 480 480 480
Water 124.8 174.8 224.8 224.8 not certain. The exact reason for the variation in colors is unclear
Waterglass 144 144 – – but should be pursued further, as it may provide a simple qual-
DE – 49.5 – 32 ity-control method wherein properly produced materials may have
NaOH 41.6 41.6 – –
a different coloration than those where mixing was insufficient,
Na2CO3 – – 31 31
curing parameters were non-optimal, and so on.
A.R. Sakulich et al. / Construction and Building Materials 23 (2009) 2951–2957 2953

Whether anyone would want to use a (literally) green building tion in an effort to make the fine aggregate concrete as simple as
material is a question whose answer changes based on region and possible to work with. The set time and compressive strengths at
aesthetic preferences, however, in developing areas a strong fine one week of each batch (Fig. 1) showed that 20 wt.% NaCl was
aggregate concrete is not likely to be ignored solely based on col- the minimum amount that provided a meaningful increase in set
oration. Even in hot regions where a dark colored building might time without negatively impacting the compressive strengths.
retain too much heat to make a comfortable home, a wide variety Though this quantity is more than used by other researchers [3]
of options (from painting the fine aggregate concrete to using it and may not be optimal for all compositions, it was used through-
only in supplementary structures such as the walls of animal pens, out this work. The reason for this discrepancy in the amounts of
floors in homes, patios, or roadways) are available. NaCl used is likely related to the fact that here the NaCl was di-
The NaCl-free samples shrank slightly; after one day of curing, rectly added to the powder reactants, rather than added to the acti-
formulae A and B could be removed from their molds without cut- vating solutions. Adding NaCl slowed the setting times of all
ting the latter. Formulae C and D had not solidified after a day, but formulae; formulae A and B were malleable for up to 30 min; for-
could be removed after one week. For all samples the addition of mulae C and D remained so for at least 2 h.
NaCl reduced shrinkage, and the samples had to be cut from their
molds, while NaCl was clearly visible as small crystals on the sur- 3.2. Mechanical characterization
faces of formulae A and B, but not C and D.
Though the samples shrank slightly, the weights did not change Compressive strength data for all formulae are shown in Fig. 2.
to any great degree: even after one month of curing at room tem- Formula A reached a strength of 45 ± 2 MPa within one day and
perature, none of the samples lost more than 5 wt.% of their initial then leveled off, making it the formula which gained strength most
water content, implying that the water is either chemically bound rapidly. Formula B was the strongest overall, increasing from
or trapped in pores. 30 ± 3 MPa after 1 day to 50 ± 0.3 MPa after 1 week. As noted
above, after one day formulae C and D were too soft to remove
3.1. Effects of NaCl addition from their molds. After a month their strengths were 37 ± 2 MPa,
with 75% of this strength gained during the first week.
Formulae A and B were subjectively determined to set in <3 min The formulae containing NaCl cured more slowly. After 1 day,
while the two formulae with Na2CO3, C and D, were malleable for formulae A and B achieved strengths of only 20 ± 1 and
at least 45 min. To determine the amount of NaCl that should be 18 ± 1 MPa, respectively (Fig. 2). These are 50% and 65% of the
used as a retarding agent, samples of formula A with varying strengths of samples formulated without NaCl over the same time
amounts of NaCl added to the powder reactants (5, 10, 15, and period. However, formula A returned to full strength after one
20 wt.% of the amount of slag) were formulated. The NaCl was month, while formula B increased to only 34 ± 2 MPa, compared
added to the powder reactants rather than to the activating solu- to the 50 ± 0.3 MPa without NaCl. Formulae C and D also hardened
more slowly; the average strength was 20 ± 3 MPa after one week,
and 25 ± 4 MPa after a month. These values were 75% of the
50 35 strengths achieved in the salt-free samples.
45 Table 2 summarizes the tensile test results. In terms of tensile
30
Compressive Strength

40 strengths, the NaOH/waterglass-activated samples were again


25
Set Time (min.)

35 stronger than the others, though the values were not particularly
30 20 high (4.2 MPa for formula A being the best). In the period between
(MPa)

25 1 day and 1 month, the tensile strengths generally increased


20 15
slightly (the exception being Formula B, which showed a slight de-
15 10 crease). That the compressive strengths of the different formulae
10
5 were evolving by much greater amounts during this period ac-
5
counts for the lack of a clear trend when tensile strength is viewed
0 0
as a percent of compressive strength (some increase over time,
0% 5% 10% 15% 20% 25%
some decrease, and the addition of NaCl makes some formulae in-
NaCl Concentration
crease, then decrease).
Fig. 1. Effect of NaCl concentration on 7 days compressive strengths and set time of The mechanism of alkali activation of slag has been shown
formula A. All formulae followed a similar trend. (at early stages) to be a dissolution–precipitation reaction that is

60.0 1 Day
7 Days
Compressive Strength (MPa)

50.0 28 Days

40.0

30.0

20.0

10.0

0.0
A A+NaCl B B+NaCl C C+NaCl D D+NaCl
Formula

Fig. 2. Compressive strengths of all formulae with and without NaCl as a retarder after 1, 7, and 28 days of room temperature curing.
2954 A.R. Sakulich et al. / Construction and Building Materials 23 (2009) 2951–2957

Table 2 3.3. Microstructural characterization


Tensile strengths (MPa and percentage of compressive strength). Standard deviation
is approximately 1%.
3.3.1. X-ray diffraction
Formula 1 day (%) 1 week (%) 1 month (%) As noted above, the as-received slag powder produced a broad
A 2.6 ± 0.2 MPa–(5.9) 3.1 ± 1.5 MPa–(7.0) 4.2 ± 0.6 MPa–(9) hump centered around 25–35°–2h (Fig. 3, lowest curve) indicative
B 2.7 ± 0.1–(8.7) 3.0 ± 1.4–(5.9) 2.3 ± 0.1–(4.8) of the short-range order in the glassy CaO–MgO–Al2O3–SiO2 struc-
C 0.0 1.4 ± 1.0–(4.8) 2.2 ± 0.3–(5.9) tures within the slag. Samples of fine aggregate concrete were
D 0.0 2.0 ± 0.8–(6.8) 2.2 ± 0.4–(5.6)
investigated using XRD after one week of room temperature cur-
A + NaCl 1.7 ± 0.2–(8.6) 2.5 ± 0.2–(6.3) 3.5 ± 0.6–(7.9) ing; the resulting diffractograms showed strong peaks correspond-
B + NaCl 1.8 ± 0.5–(9.8) 2.4 ± 0.4–(7.2) 2.4 ± 0.5–(8.2)
C + NaCl 0.0 1.8 ± 0.2–(9.6) 2.0 ± 0.2–(6.9)
ing to the presence of aggregate while peaks belonging to the
D + NaCl 0.0 1.7 ± 0.1–(7.7) 2.9 ± 0.6–(12.4) crystalline phases in the cementing phase (discussed below) were
weaker and more difficult to identify.
Samples of AAS pastes were examined after curing at room tem-
perature for 7, 28 and 55 days (formulae A and B, Fig. 3; formulae C
affected by pH, while at later stages a more complex solid-state and D, Fig. 4). After 7 days the XRD diffractograms had changed
reaction may be taking place [17]. Hunhai et al. [26] reported that substantially from those of pure slag, with all formulae showing
the degree of hydration of slag increases with pH, due to the ability a strong peak at about 30°–2h corresponding to the (110) reflection
of OH- ions to attack the slag structure, creating intermediary reac- of poorly crystallized C–S–H gel and weak peaks near 50 and 55°–
tants that later precipitate into C–S–H and other final products. 2h corresponding to the (020) and (310) reflections. Overall, the
Therefore, that NaOH/waterglass activation (pH 13.5) provides only peaks detected in formulae A and B were those of C–S–H
higher strengths than Na2CO3 activation (pH 11.3) is unsurprising gel. Further, for the Na2CO3-activated formulae peaks at 11 and
and due to the fact that, as a stronger alkali, the degree of hydra- 23°–2h were clearly visible (Fig. 4). These peaks correspond to
tion in the slag is greater. Na2CO3 may, however, be a more practi- the (003) and (006) reflections of hydrotalcite, a magnesium-alu-
cal activator, especially in developing regions, as it is widely minum hydroxycarbonate, Mg6Al2CO3(OH16)4H2O.
available, less caustic than NaOH, and easier to handle and cheaper The C–S–H (110) peaks of formulae C and D were at all times
than waterglass. sharper than those found in A and B, while the peaks correspond-
The addition of DE resulted in slightly higher compressive ing to the (020) and (310) reflections are too small to make a rea-
strengths after 28 days for formulae B and D compared to ‘neat’ sonable comparison. Between 7 and 55 days all peaks became
formulae (A and C). This is likely due to dissolution of the DE, sharper and somewhat more intense, indicating that the produc-
which presumably provides more Si for the formation of C–S–H. tion of C–S–H was continuing and that it was becoming more
At high pH, however, amorphous Si is known to produce silicic crystalline.
acid, which would act as a retardant and slow the reaction, leading The use of XRD to determine the presence of C–S–H in AAS
to a delay in strength gains associated with DE dissolution. pastes activated by a variety of alkali or alkali-silicate solutions
For formulae A and C, the addition of NaCl initially acts as a has been noted by numerous authors. Hydrotalcite has also been
retardant but, as NaCl precipitates in pore solution, this effect is detected in AAS fine aggregate concretes including fly ash [14]
minimized, the hydration reaction continues, and strengths after and geothermal silica waste [27] as additives, although C–S–H
28 days either equal (formula A) or are approaching (formula C) was not present in the former. Puertas et al. detected hydrotalcite
the strengths of formulae without NaCl. The reason for the drop in AAS systems activated solely by NaOH, but not in systems acti-
in strength at 28 days for formula B with NaCl is unclear. In for- vated by waterglass [15]. C–S–H and CaCO3 were also present. Our
mula D strength levels off at relatively low levels (25 MPa) after results most closely match those of Scrivener and Wang, who
one week, likely due to the NaCl lowering the pH of the system to investigated NaOH- and waterglass-activated AAS pastes and
levels where the hydration reaction takes place only weakly. found only C–S–H and hydrotalcite [17].

Fig. 3. XRD patterns of formulae A and B without NaCl or limestone aggregate after Fig. 4. XRD patterns of formulae C and D without NaCl or limestone aggregate after
7, 28, and 55 days of curing showing development of poorly crystalline CSH gel. 7, 28, and 55 days of curing showing development of poorly crystalline CSH gel and
Spectrum of as-received slag shown at bottom for comparison. hydrotalcite. Spectrum of as-received slag shown at bottom for comparison.
A.R. Sakulich et al. / Construction and Building Materials 23 (2009) 2951–2957 2955

Many other phases that are described in the literature [7,20], be found in Table 3. Of the samples cured for 28 days, all spectra
such as Strätlingite (C2ASH8), Merwinite, hematite, Ca(OH)2, CaCO3, featured a peak at 470 cm 1 due to the deformation of SiO4 tetra-
and (C,M)4AH13, were not observed here. The first four simply did hedra. Peaks centered at 960 cm 1 indicate the Si–O vibrations of
not appear in the XRD diffractograms, while the second two are silica tetrahedra of various coordination states; these peaks have
difficult to identify as their peaks overlap closely with the peaks shifted to slightly higher numbers in the paste samples, indicating
of C–S–H or hydrotalcite. The presence of CaCO3 is suggested by an increase in the polymerization of the C–S–H gel that is being
the FTIR results (see below), but neither FTIR nor SEM/EDS de- formed. All formulae also show a slight peak around 700 cm 1,
tected (C,M)4AH13, putting us again in agreement with Scrivener the origin of which is unclear. This peak is so weak as to be essen-
and Wang. tially negligible, however. A peak at 1460 cm 1 shows CO23 ions,
formed from CO2 incorporated in the cement as it was mixed in
3.3.2. FTIR air or during later carbonation. All four formulae show the exis-
Samples of AAS paste cured at room temperature for 7 (Fig. 5) tence of a peak corresponding to out of plane Si–O bending cen-
and 55 (not shown; 55 day and 7 day data show only minor differ- tered around 530 cm 1, though in formulae C and D this peak
ences) days were investigated using FTIR spectroscopy. The major appears more as a shoulder than a distinct peak. A peak around
peaks shown in these spectra, as well as the origin of the peaks, can 1640 cm 1 found in all formulae is caused by the bending vibra-
tions of water; while peaks in the region between 2800 and
3700 cm 1 (here found in the form of a broad peak around 3480)
are generally due to the presence of hydrogen bonds of varying
strengths, in this case likely those of O–H stretching in water or
other weakly-bonded OH-bearing molecules.
In FTIR analyses performed after 55 days of curing, all four spec-
tra showed only minor differences. The peaks at 960 cm 1 shifted
to slightly higher wavenumbers and the peak at 3480 cm 1 was
not as deep, implying that most of the product-forming reactions
are finished or thoroughly slowed after one week and that some
water had left the system.
These FTIR spectra are almost identical to data reported [28] by
Delgado et al. for pure C–S–H gel, and show some substantial dif-
ferences from spectra reported for geopolymers.

3.3.3. SEM analysis


Fracture surfaces of each of the four fine aggregate concrete for-
mulae (with and without NaCl) were examined in the SEM. Typical
secondary SEM micrographs of formula A fine aggregate concrete
are shown in Fig. 6. At low magnification, the aggregate and the
Fig. 5. FTIR spectra of cement pastes of formulae A, B, C, and D after 7 days (without cementing phase were clearly visible (Fig. 6a). NaCl was found in
NaCl) of curing with labeled major peaks. Spectrum of as-received slag is shown spherical regions surrounded by the cementing phase (Fig. 6b),
(top) for comparison. likely due to crystallization of NaCl from solution in pores. The
NaCl particles were easily found in formulae A and B, and some-
what less frequently in formulae C and D. As previously mentioned,
Table 3
the NaCl has been shown to be relatively stable in the pores, and is
Major peaks detected by FTIR in unreacted slag and formulae A, B, C, and D (without unlikely to contribute to long-term deleterious effects.
NaCl) including origin of the signal. At higher magnifications, the cementing phase and grains of
Slag A B C D Origin
unreacted slag are clearly visible in the AAS paste samples (and
are described further below). Formulae that contained DE con-
450 486 486 486 486 Si–O–Al bending
532 522 n/a 543 n/a Out of plane SiO
tained some undissolved diatoms, but in all other respects the
724/870 n/a n/a n/a n/a Six member Si ring microstructures of the formulae were quite similar. Some surface
953/1045 990 1017 925/1054 980 SiO2 stretching microcracking is also evident. SEM analysis of fine aggregate con-
1438 1448 1448 1485 1457 CO3 cretes mounted after compression testing shows both limestone
1656 1685 1676 1675 1679 Water
aggregate and unreacted slag, both of which exhibit numerous
3454 3536 3536 3536 3536 Water
intragranular fractures.

Fig. 6. SEM micrographs of representative structures from formula A. (A) Fine aggregate concrete without NaCl. (B) Cement paste containing NaCl as a retarding agent
(arrow). (C) Cement paste at higher magnification showing grains of unreacted slag (arrow).
2956 A.R. Sakulich et al. / Construction and Building Materials 23 (2009) 2951–2957

Table 4 these explanations is more likely than the formation of a


Data from image analysis of non-NaCl containing cement pastes. geopolymer.
Formula Unreacted slag (vol.%) Unreacted slag particle size (lm) The EDS analyses performed on the cementing phases of all four
A 26.5 4.4 formulae showed their compositions to be roughly consistent (the
B 23.5 4.5 amount of any particular element in any one formula was within
C 41.8 1.8 5% of the amount found in any other formula) except that the
D 30.3 3.3 Na2CO3-activated formulae contained some carbon and less Na
than the NaOH-activated formulae.

4. Summary and conclusions


Image analysis was performed using the Image J program on
BSE images of each of the AAS pastes that did not include NaCl. Slag was activated using either a waterglass/NaOH mixture or a
In BSE mode, the NaCl was difficult to distinguish from unreacted Na2CO3 solution, and concretes were prepared through the addi-
slag based on contrast alone. The shapes of the particles made their tion of a fine limestone aggregate. These concretes had good com-
composition clear (grains of NaCl are almost uniformly round in pressive strengths, which were slightly elevated through the
SEM images and somewhat larger than grains of slag) but this addition of DE, and splitting tensile strengths that varied but were
could not be accounted for by the Image J. At least 3000 particles slightly lower than those reported for OPC-based fine aggregate
of unreacted slag were measured for each formula. concretes, which have tensile strengths of 10% of their compres-
The analysis showed that the amount of unreacted slag in the sive strengths. The largest drawback to the use of these formulae is
formulae varied from 20 to 40 vol.%. This data, as well as the the short pot lives, but this can be overcome through the use of
average size of the particles of unreacted slag, can be found in Ta- NaCl as a retarding agent and is somewhat longer in Na2CO3 acti-
ble 4. Lower amounts of unreacted slag for formulae A and B are in vated formulae.
line with their higher strengths, however, the reason that formula Emphasis in this work is placed on the fact that no high-temper-
C has the largest amount of unreacted slag but the smallest average ature curing (all samples cured at room temperature) or compli-
particle size is unclear. cated fabrication techniques (autoclave curing, humidity
chamber curing, etc.) were used, making these formulae both prac-
3.3.4. EDS analysis tical for large-scale usage and of reduced environmental impact.
Elemental analysis was performed on the cementing phases of XRD and FTIR revealed the presence of C–S–H gel as the binding
the four formulae that did not contain NaCl. phase in all formulae; those activated by Na2CO3 also showed the
The EDS results were compared with the chemistry that would formation of hydrotalcite. The presence of the C–S–H phase (with
be expected based on the starting reactants (Table 5). Most values a Ca/Si ratio lower than that found in OPC) was verified through
were reasonably close, considering the mixing methods used, ex- use of EDS. No compelling evidence for the presence of zeolites,
cept for Ca, which was consistently about half of the expected geopolymers, or crystalline phases (aside from hydrotalcite) was
value. found.
Typically, five points from between unreacted slag grains were Even the weakest of these four formulae would be acceptable
investigated. In formula A, one of these areas contained no Na and for use in the majority of building applications. The greatly reduced
is likely to be pure C–S–H gel. This gel had a CaO/SiO2 ratio of 1.2, environmental impact, the simplicity of manufacture, and the use
which is lower than the average of between 1.5 and 2.0 found in of natural reactants (Na2CO3, DE, and NaCl) are all reasons for fur-
OPC, and closer to the ratio of 1.0 reported in the literature for ther investigating these materials. Preliminary work suggests that
AAS cements [2,15,17,19,22]. This area, however, was anomalous the tuning of the Si/Al ratio through addition of clays or alumina
and was observed only once. The other four points contained powders during the dry mixing phase can further increase
12 ± 0.6 at.% Na and a lower (3.1 ± 0.5%) quantity of Ca, resulting strengths and will be reported on later.
in the average chemistries listed above.
While the roughly homogenous distribution of Na may at first Acknowledgements
appear to indicate the production of geopolymers, there are a num-
ber of other explanations. First, if Si4+ in the C–S–H is being substi- The authors would like to thank Paul Lessard of Graniterock,
tuted by Al3+, Na ions would have to appear in the interlayer space Inc., and Denny Whitesell of Graymont for donating granite and
to balance the charge. Second, Na can substitute for Ca in the C–S– lime. The help of Dr. Eva Jud Sierra and Sean Miller of Drexel Uni-
H structure to produce a similar product sometimes called N–S–H. versity is also appreciated. This work was partially funded by NSF
Considering that the presence of C–S–H is unambiguous, either of (DMR 0503711).

Table 5
Theoretical and actual (measured by EDS) elemental compositions of cementing regions in non-NaCl containing limestone-based concrete.

A B C D
Measured Theoretical Measured Theoretical Measured Theoretical Measured Theoretical
H 34 ± 1 30.1 33 ± 1 30.1 31 ± 1 29.1 29 ± 1 26.5
O 40 ± 1 35.1 43 ± 1 36.5 46 ± 1 40.4 47 ± 1 40
Na 7±3 8.0 6.4 ± 0.5 7 3 ± 0.5 2.4 3 ± 0.5 2.3
Mg 2.0 ± 0.5 2.1 1.5 ± 0.5 1.8 2 ± 0.5 4 2 ± 0.5 2
Al 2.5 ± 0.5 4.1 2 ± 0.5 3.6 3 ± 0.5 2.1 2 ± 0.5 4
Si 8±1 10.5 10 ± 0.5 12.6 6±1 8.6 8±1 10.4
Ca 7±3 10.1 4.5 ± 0.5 8.4 5.5 ± 1 10.9 5±1 12.7
C 0 0 0 0 3 ± 0.5 2.5 5±1 2.1
A.R. Sakulich et al. / Construction and Building Materials 23 (2009) 2951–2957 2957

References [15] Puertas F, Fernández-Jiménez A, Blanco-Varela MT. Pore solution in alkali-


activated slag cement pastes. Relation to the composition and structure of
calcium silicate hydrate. Cem Concr Res 2004;34(1):139–48.
[1] Roy DM. Alkali activated cements: opportunities and challenges. Cem Concr
[16] Fernández-Jiménez A, Puertas F, Sobrados I, Sanz J. Structure of calcium silicate
Res 1999;29(2):249–54.
hydrates formed in alkaline-activated slag: influence of the type of alkaline
[2] Brough AR, Atkinson A. Sodium silicate based, alkali activated mortars part I:
activator. J Am Ceram Soc 2003;86(8):1389–94.
strength, hydration and microstructure. Cem Concr Res 2002;32(6):865–79.
[17] Wang SD, Scrivener KL. Hydration products of alkali activated slag cement.
[3] Brough AR, Holloway M, Sykes J, Atkinson A. Sodium silicate based alkali
Cem Concr Res 1995;25(3):561–71.
activated slag mortars part II: the retarding effect of additions of sodium
[18] Puertas F, Fernández-Jiménez A. Mineralogical and microstructural
chloride or malic acid. Cem Concr Res 2000;30(9):1375–9.
characterization of alkali-activated fly ash/slag pastes. Cem Concr Comp
[4] Zhang Y, Sun W, Chen Q, Chen L. Synthesis and heavy metal immobilization
2003;25(3):287–92.
behaviors of slag based geopolymer. J Haz Mater 2007;143(1):206–14.
[19] Davidovits J. 30 Years of successes and failures in geopolymer applications.
[5] Garci-Juenger MC, Monteiro PJM, Gartner EM. In situ imaging of ground
Market trends and potential breakthroughs. In: Lukey GC, editor. Geopolymer
granulated blast furnace slag hydration. J Mater Sci 2006;41(21):7074–81.
2002 conference. Siloxo PTY Ltd; 2002.
[6] Cheng TW, Chiu JP. Fire resistant geopolymer produced by granulated blast
[20] Escalante-García JI, Fuentes AF, Gorokhovsky A, Fraire-Luna PE, Mendoza-
furnace slag. Minerals Eng 2002;16(3):205–10.
Suárez G. Hydration products and reactivity of blast-furnace slag activated by
[7] Roy A, Schilling PJ, Eaton HC, Malone PG, Brabston WN, Wakeley LD. Activation
various alkalis. J Am Ceram Soc 2003;86(12):2148–53.
of ground blast-furnace slag by alkali–metal and alkaline–earth hydroxides. J
[21] Lecomte I, Henrist C, Liegéois M, Maseri F, Rulmont A, Cloots R. (Micro)-
Am Cer Soc 1992;75(12):3233–40.
structural comparison between geopolymers, alkali-activated slag cement and
[8] Shi C. Strength, pore structure, and permeability of alkali-activated slag
Portland cement. J Eur Ceram Soc 2006;26(16):3789–97.
mortars. Cem Concr Res 1996;26(12):1789–99.
[22] Yip CK, Van Deventer JSJ. Microanalysis of calcium silicate hydrate gel formed
[9] Zosin AP, Priimak TI, Avsaragov KB. Geopolymer materials based on magnesia-
within a geopolymeric binder. J Mater Sci 2003;38(18):3851–60.
iron slags for normalization and storage of radioactive wastes. Atomic Energy
[23] Sakulich AR, Ivošević Ž, Barsoum MW. Development of a diatomaceous earth,
1998;85(1):78–82.
fine limestone aggregate alkali-activated concrete. In: Proceedings of the 11th
[10] Gifford PM, Gillott JE. Alkali–silicate reaction (ASR) and alkali–carbonate
Canadian Masonry Symposium; 2009.
reaction (ACR) in activated blast furnace slag cement (ABFSC) concrete. Cem
[24] Colthup NB, Wiberley SE, Daly LH. Introduction to infrared and Raman
Concr Res 1996;26(1):21–6.
Spectroscopy. New York: Academic Press; 1975.
[11] Douglas E, Bilodeau A, Malhotra VM. Properties and durability of alkali-
[25] Silverstein RM, Webster FX. Spectrometric identification of organic
activated slag concrete. ACI Mater J 1993;89(5):509–16.
compounds. New York: John Wiley & Sons, Inc.; 1998.
[12] Gartner E. Industrially interesting approaches to Low CO2 cements. Cem Concr
[26] Huanhai Z, Zuequan W, Zhongzi X, Mingshu T. Kinetic study on hydration of
Res 2004;34(9):1489–98.
alkali-activated slag. Cem Concr Res 1993;23(6):1253–8.
[13] Richardson IG, Groves GW. Microstructure and microanalysis of hardened
[27] Escalante-García JI, Palacios-Villanueva VM, Gorokhovsky AV, Mendoza-
cement pastes involving ground granulated blast-furnace slag. J Mater Sci
Suárez G, Fuentes AF. Characteristics of a NaOH-activated blast furnace slag
1992;27(22):6204–12.
blended with a fine particle silica waste. J Am Ceram Soc 2002;85(7):1788–92.
[14] Puertas F, Martinez-Ramirez S, Alonso S, Vazquez T. Alkali-activated fly ash/
[28] Delgado AH, Paroli RM, Beaudoin JJ. Comparison of IR techniques for the
slag cement strength behaviour and hydration products. Cem Concr Res
characterization of construction cement minerals and hydrated products. Appl
2000;30(10):1625–32.
Spectrosc 1996;50(8):970–6.

You might also like