Dry powder alkali-activated slag cements

Maxim Kovtun Julia Shekhovtsova

Researcher, Department of Civil Engineering, University of Pretoria, PhD Student, Department of Civil Engineering, University of Pretoria,
Pretoria, South Africa Pretoria, South Africa
Elsabe P. Kearsley
Head of Department of Civil Engineering, University of Pretoria, Pretoria,
South Africa

This paper reports on the results of an investigation into the possibility of producing dry powder alkali-activated slag
cements as a ready-to-use product which can be packed in bags and mixed with water to produce a concrete. The
cements were produced using a neutral granulated blast-furnace slag and sodium carbonate. To accelerate strength
development at ambient temperatures, a combination of silica fume and slaked lime was used as accelerating
admixture in the cement’s formulation. Powder sodium lingosulfonate was added into the formulation to reduce the
water demand of the cements; it also delayed setting and increased compressive strength. Alkali-activated slag
concretes were produced using the developed cements. Compressive strengths in the range from 30 to 85 MPa were
achieved for the concretes at 28 d. The accelerating admixture increased 1-d compressive strength for the alkali-
activated slag concretes cured at ambient temperature up to 25 MPa. The optimum concentration of sodium
carbonate was equivalent to 3 .5% and 4 .5% sodium oxide (Na2O) for precast and in situ application, respectively.
The accelerating admixture is recommended at 6% for dry powder alkali-activated slag cements. The accelerated
ageing test showed that the dry powder alkali-activated slag cement had a sufficient shelf life.

Introduction manufacturing process involves large volumes of alkaline liquids

Alkali activation of slag has been known for over 70 years (Van Deventer et al., 2012). This is one of the key barriers to the
(Purdon, 1940). Intensive research and development of alkali- widespread adoption of these materials.
activated slag (AAS) cements and concretes has been carried out
in the USSR by Glukhovsky and co-workers since the 1950s There are four main activators usually used for alkali activation
(Glukhovsky, 1959; Glukhovsky et al., 1983). During the last two of slags, namely silicates (M2O.nSiO2), hydroxide (MOH), sulfate
decades AAS cements and concretes have been intensively studied (M2SO4) and carbonate (M2CO3), where M is usually sodium or
worldwide (Bakharev et al., 1999a, 2000; Ben Haha et al., 2011; potassium (Wang et al., 1994). There are also three methods of
Collins and Sanjayan, 1999b; Douglas et al., 1991; Douglas and adding the alkaline component to slag: in an aqueous solution, in
Brandstetr, 1990; Fernández-Jiménez et al., 1999; Wang et al., the solid state by mixing dry alkali powder with slag or in the
1994, 1995; Živica, 2007). Interest in these cements is driven not solid state where dry alkali powder is ground together with the
only by the need of human society to reduce the carbon dioxide slag (Moseson et al., 2012; Wang et al., 1994; Yang et al., 2008).
emissions, but also by several advantages that AAS cements can
offer. These advantages include low heat of hydration, rapid Dry powder AAS cements can be produced by grinding together
hardening, high ultimate strength, resistance to chemical attack, a raw slag with an alkali and additives if necessary. Little
heat resistance, low permeability, freeze–thaw performance and research has been published on grinding slag with an alkali
reduced sensitivity to aggregate quality (Bakharev et al., 2003; (Wang et al., 1994). However, producing a binder that can be
Shi and Stegemann, 2000; Shi, 1996; Wang et al., 1995; Xu et al., used with the addition of water only will expand the use of AAS
2008). It is also known that AAS cements and concretes have cements (Van Deventer et al., 2012).
several disadvantages, including fast setting, relatively high
shrinkage (in comparison to ordinary Portland cement (OPC) Sodium silicate and sodium hydroxide or combinations of these
concretes) and the formation of efflorescence (Atiş et al., 2009; two are usually the first choice for alkali activation of slags (Ben
Bakharev et al., 1999a, 1999b; Collins and Sanjayan, 1999a, Haha et al., 2011; Douglas and Brandstetr, 1990; Douglas et al.,
1999b; Živica, 2007). The resistance of AAS concretes to 1991; Turner and Collins, 2013; Wang et al., 1994; Yang et al.,
carbonation and alkali–silica reaction was questioned by Bakharev 2008). Nevertheless, these alkalis are difficult to use in the
et al. (2001a, 2001b). Another disadvantage of AAS concretes is production of dry powder AAS cements owing to the high
that they cannot be routinely used on construction sites, as the hygroscopicity of sodium hydroxide and the poor solubility of
1
sodium silicate, as well as the possible absorption of moisture by in Table 1. The basicity coefficient (Kb ¼ (CaO + MgO)/(SiO2 +
sodium silicate with silicate modulus (Ms) below 1 .2 (Wang et Al2O3)), activity coefficient (Ka ¼ Al2O3 /SiO2), and quality coef-
al., 1994). A recent study by Turner and Collins (2013) also ficient (Kq ¼ (CaO + Al2O3 + MgO)/(SiO2 + TiO2)) based on the
showed that the use of sodium hydroxide and silicate contributes chemical composition are 0 .92, 0 .41 and 1 .67, respectively. X-ray
significantly to the total carbon dioxide emissions associated with diffraction (XRD) analysis shows that the slag is mostly amor-
alkali-activated concrete production. In contrast to sodium hydro- phous matter, and that it contains minor amounts of gehlenite.
xide and sodium silicate, sodium carbonate is not hygroscopic, The specific gravity of the slag was 2900 kg/m3.
and the associated carbon dioxide emission is expected to be
much lower (Moseson et al., 2012). The biggest disadvantage of Sodium carbonate powder (99% purity) was used as an activator.
using sodium carbonate in dry powder AAS cement is the slow A combination of silica fume and slaked lime was used as an
strength development with neutral and acid slags at ambient accelerating admixture to promote the strength development of
temperatures (Wang et al., 1994). To overcome this problem, dry powder AAS cements at ambient temperatures. Slag was
different accelerating admixtures can be used (Glukhovsky, 1959; replaced by the admixture at 2, 6 and 10 wt%. The chemical
Wang et al., 1994). compositions of silica fume and slaked lime are presented in
Table 1. Sodium lignosulfonate powder was used as a plasticiser.
This paper reports on the results of an investigation into the
possibility of producing dry powder AAS cements as a ready-to- Crushed dolomite stone and dolomite sand were used as aggre-
use product which can be packed in bags and mixed with water gates. The specific gravity of the aggregates was 2860 kg/m3. The
to produce a concrete. Eliminating the need for an alkaline coarse aggregate had a maximum size of 9 .5 mm. The fineness
solution gives the following advantages over traditional AAS modulus of the sand was 3 .86 with 9% passing the 75 ìm sieve.
concrete technology, where an alkali is introduced to the concrete
mixture in the form of an aqueous solution. Preparation of dry powder AAS cements
The dry powder AAS cements were produced by grinding
j Safety on the construction site: aqueous solutions of alkalis together the slag with sodium carbonate, the accelerating admix-
used for activation are normally highly corrosive mediums ture and plasticiser in a laboratory steel ball mill. Compositions
and utilising them is hazardous. of the cements produced are shown in Table 2. The duration of
j No need for a ‘two packages’ product. milling was 3 h 30 min and 4 h 30 min to produce dry powder
j Quality control: in the case of an aqueous activator solution, AAS cements with Blaine finenesses in the region of 390 m2/kg
two parameters should be controlled in the production process and 430 m2/kg, respectively. Fineness of the cements was meas-
– the amount of liquid activator and its density (concentration). ured with a Blaine apparatus according to BS EN 196-6:2010
Technologically it becomes even more difficult to control AAS (BSI, 2010). A Malvern Mastersizer 2000 apparatus was used to
concrete mix design when aggregates are piled in open areas. determine a particle size distribution of the dry powder AAS
Water from aggregates will dilute the activator solution, cements.
deteriorating many characteristics of AAS concrete. Keeping
in mind the limits of solubility of typical activators in water at Production of AAS concretes
certain temperatures, it is obvious that in some cases it would During the second stage of this research, AAS concretes were
not be possible to achieve the correct mix proportions. These produced. The binder (slag + silica fume + slaked lime) content
hurdles do not exist for dry powder AAS cement. was kept constant of 500 kg/m3. Thus, the total AAS cement
content with activator concentrations of 3 .5, 4 .0 and 4 .5%
sodium oxide was 530, 534 and 538 kg/m3, respectively. The
Experimental procedures
binder content was chosen based on the following factors.
Materials
A granulated blast-furnace slag (GBFS) was used to produce j Good workability at low water/binder (W/B) ratio due to a
AAS cements. The chemical composition of the slag is presented substantial amount of AAS paste. Workability cannot be

Material Silicon Titanium Aluminium Iron oxide Magnesium Calcium Sodium Potassium Sulfur Loss on
dioxide dioxide oxide (Fe2O3) oxide oxide oxide oxide trioxide ignition
(SiO2) (TiO2) (Al2O3) (MgO) (CaO) (Na2O) (K2O) (SO3) (LOI)

Slag 34 .87 0 .72 14 .38 0 .89 8 .03 37 .05 ,0 .01 0 .72 1 .96 0 .16
Silica fume 84 .00 0 .03 0 .75 1 .96 1 .08 2 .22 0 .17 3 .43 0 .09 5 .56
Slaked lime 4 .01 0 .03 0 .37 0 .36 1 .67 67 .80 0 .00 0 .08 0 .11 25 .0

Table 1. Chemical composition of materials used

2
Cement ID Components: wt% The coarse to fine aggregate ratio of 65:35 was kept constant in
all mixes. The W/B ratio was 0 .30 or 0 .35. Concrete mix designs
Slag Silica Slaked Activator (sodium oxide are shown in Table 3.
fume lime eq. of slag mass)
The concrete preparation procedure consisted of the mixing of
SC35-390 100 3 .5 dry aggregates with dry powder AAS cement in a pan mixer, the
SC40-390 100 4 .0 addition of water and then further mixing until a homogeneous
SC45-390 100 4 .5 fresh concrete mix was produced. After workability testing, the
SC35-430 100 3 .5 fresh concrete was immediately placed in 100 mm plastic cube
SC40-430 100 4 .0 moulds and vibrated. Two initial curing regimes were used: 258C
SC45-430 100 4 .5 at . 90% relative humidity and steaming at 858C. Steamed
SF2-500 98 1 1 4 .5 samples were pre-cured at 258C for 2 h (which was found in
SF6-590 94 3 3 4 .5 previous research to have a positive effect on the properties of
SF10-570 90 5 5 4 .5 AAS concretes (Bakharev et al., 1999b)) and then placed into
SF6-610 94 3 3 4 .5 steam baths. The temperature was increased to 858C at a rate of
SF6L-620a 94 3 3 4 .5 108C/h, with the total duration of steam curing of 16 h. No
gradual cooling was used, and samples were de-moulded immedi-
a
AAS cement contains sodium lingosulfonate plasticiser at 1 .5 wt% ately after steaming. After de-moulding, the samples were stored
of slag mass.
in a room at 258C and 55% relative humidity up to the day of
testing.
Table 2. AAS cement compositions

Testing methods
The AAS cements’ setting times were measured according to BS
increased simply by adding water because it will reduce the EN 196-3:2005 (BSI, 2005) using a Vicat apparatus. The W/B
pH of the system due to the diluting effect and will ratio was kept constant at 0 .25, as the procedure of a standard
significantly impair the technical properties of AAS concrete paste consistency normally used for OPC cements (BSI, 2005)
(Shi and Day, 1996; Shi and Li, 1989). would influence the sodium carbonate/water ratio in the AAS
j Durability of AAS concretes increases with increased binder pastes, changing the ionic and anionic concentration of the liquid
content (Bernal et al., 2011). media (Shi and Day, 1996; Shi and Li, 1989). Scanning electron
j Optimal compressive strength of AAS concretes is achieved microscope (SEM) investigations of AAS cements were done
at about 25% binder content by volume (Ravikumar et al., with a Zeiss Ultra Plus SEM at 1 kV. Sticky carbon tape was
2010). submerged into the AAS cement; glued cement particles were

Concrete ID Components: kg/m3 Slump:

mm/
Cement Stone Sand Water W/B ratio Vebe: s

SC35-390 530 1189 646 175 0 .35 30/–

SC40-390 534 1186 645 175 0 .35 30/–
SC45-390 538 1182 643 175 0 .35 35/–
SC45-390 aged 538 1182 643 175 0 .35 10/–
SC35-390 530 1235 672 150 0 .30 –/6
SC40-390 534 1232 670 150 0 .30 –/7
SC45-390 538 1229 668 150 0 .30 –/5
SC35-430 530 1235 672 150 0 .30 –/13
SC40-430 534 1232 670 150 0 .30 –/11
SC45-430 538 1229 668 150 0 .30 –/10
SF2-500 538 1181 641 175 0 .35 –/16
SF6-590 538 1178 639 175 0 .35 –/18
SF10-570 538 1168 634 175 0 .35 –/17
SF6-610 538 1178 639 175 0 .35 –/18
SF6L-620 538 1178 639 175 0 .35 –/10

Table 3. AAS concrete mix designs and workability

3
coated with carbon. Workability (slump, Vebe test) of AAS addition of sodium lignosulfonate resulted in an increase in specific
concrete mixes was tested following the standard procedure of surface area of AAS cement from 606 to 624 m2/kg (Table 4), but
BS EN 12350:2009 (BSI, 2009a). Concrete samples were tested this change is not deemed to be significant.
for compressive strength using 100 mm cubes according to BS
EN 12390-3:2009 (BSI, 2009b) at 1 d (2 d if strength at day 1 Setting times of AAS cements
was not enough for samples to be de-moulded), 28 d and 91 d. Setting times of the AAS cements are presented in Table 4. An
increase in alkali content (from SC35-390 to SC45-390) and
To estimate the shelf life of dry powder AAS cements, the AAS fineness (from SC45-390 to SC45-430 and from SF6-590 to SF6-
cement SC45-390 was aged in an opened bag at 258C and 99% 610) shortens the initial setting time. The final setting time seems
humidity for 9 weeks after manufacturing. A detailed study of to be sensitive to fineness of the AAS cements, with significantly
prehydration of the AAS cements was not the main aim of this shorter final setting times recorded with an increase in fineness.
research and a rather practical approach was used, ageing the The final setting is not significantly affected by alkali content,
AAS cement in a bag instead of a thin layer of a few millimetres which correlates with the findings made by Živica (2007).
(Whittaker et al., 2013), assuming that AAS cement would be
stored in bags as OPC normally is. Cement generally would not The addition of silica fume and slaked lime does not influence
be stored in such conditions, but 99% humidity was used to the initial setting time but delays final setting (SF6-590 compared
intensify the ageing process, thus evaluating the performance of to SC45-390, Table 4). The plasticiser not only decreases water
the cement if it would be stored for a long time (to replicate a demand of dry powder AAS cement (see the later section entitled
worst-case scenario). After ageing, the AAS cement was studied ‘Workability of AAS concretes’) but also prolongs setting times
using SEM to determine any differences in particle morphology (SF6-610 compared to SF6L-620, Table 4), which is in agreement
in comparison to the fresh AAS cement. with other work (Bakharev et al., 2000). It is well known that
plasticisers act as retarders with OPC (Neville, 1995). Very often
the influence of plasticisers on OPC setting times is seen as a
Results and discussion
negative effect and accelerators are used in some cases to shorten
Fineness of AAS cements setting time (Neville, 1995). This side-effect of using plasticisers
The fineness of the dry powder AAS cements varied, depending on can, however, be very useful and important for producing dry
milling time and the composition of the cements as indicated in powder AAS cements because setting times of these cements are
Table 4. AAS cements containing only slag and sodium carbonate critical due to possible fast setting, especially with increasing
were divided into two groups, cements with Blaine fineness values fineness (Wang et al., 1995).
of about 390 m2/kg and about 430 m2/kg. Higher fineness values
were found in AAS cements containing silica fume and slaked lime. Ageing of AAS cement SC45-390 significantly delays both initial
This increase in fineness could be explained by the softness and and final setting times. The delay is most probably caused by the
high initial specific surface area of slaked lime and silica fume. The formation of a crust of hydration products (see next section) on slag
Blaine fineness of AAS cements containing silica fume and slaked particles, which act as a barrier between slag particles and water
lime varied between 504 and 624 m2/kg depending on milling time, (activator solution, after dissolution of sodium carbonate contained
amount of the admixture and presence of the plasticiser. The in the AAS cements) decelerating hydration of the AAS cement.

Cement ID Time of milling: Particle size range: d10: ìm d50: ìm Blaine: Initial set: Final set:
min ìm m2/kg h:min h:min

SC35-390 210 0 .3–168 2 .3 20 .5 372 1:05 2:10

SC40-390 210 0 .3–168 2 .4 21 .7 393 — —
SC45-390 210 0 .3–168 2 .2 21 .5 387 0:50 2:15
SC45-390 aged 210 0 .3–168 2 .2 21 .5 387 3:10 6:35
SC35-430 270 0 .3–138 1 .5 17 .6 435 — —
SC40-430 270 0 .4–138 1 .9 18 .4 434 — —
SC45-430 270 0 .4–125 1 .7 17 .5 435 0:45 1:05
SF2-500 210 0 .4–138 1 .8 20 .1 504 0:50 2:00
SF6-590 210 0 .4–138 1 .5 19 .3 589 0:45 2:45
SF10-570 210 0 .4–168 1 .5 19 .9 573 1:00 2:55
SF6-610 270 0 .4–138 1 .6 19 .4 606 0:30 1:40
SF10L-620 270 0 .4–152 1 .3 17 .3 624 3:45 8:30

Table 4. AAS cements fineness and setting times

4
SEM observation of AAS cements Investigation of aged AAS cement with SEM showed that
Figure 1 shows SEM images of the AAS cement containing the conglomerates of the cement particles were formed (Figure 2(a)),
accelerating admixture (SF6-610). Silica fume particles and slaked and some reaction, induced by water vapours in the fog room,
lime crystals can be observed on the surface of a slag particle took place (Figures 2(b)–2(d)). The fresh AAS cement does not
(Figure 1(a)). As can be seen, inter-grinding provides close contact form the conglomerates (Figure 3(a)) or large amounts of reaction
between slag, slaked lime and silica fume particles, which is products (Figure 3(b)) similar to that which cover particles of the
important for early and long-term hydration because the early aged AAS cement (Figures 2(c) and 2(d)). Similar formations of
hydration process of AAS cement occurs through nucleation and reaction products on the surface of prehydrated OPC particles
growth mechanisms (Hubler et al., 2011), whereas the long-term were observed by Dubina et al. (2010) and Whittaker et al.
reaction may continue through partially solid-state mechan-isms (2013). These products effectively increase the surface area of the
(Wang and Scrivener, 1995). In both cases close contact between aged AAS cement and form a barrier between water and the
the particles should facilitate the hydration process. In Figure 1(b), unreacted part of the slag (Whittaker et al., 2013). The increased
more fine particles deposited on the slag surface can be seen. Some surface area increases the water demand of the cement, decreas-
particles went through amorphisation induced by the milling ing the workability of AAS concrete (see next section). The
process (Scian et al., 1991; Tang et al., 2012; Tromans and Meech, hydration products also result in rougher particle interlocking
2001). The amorphisation of the particle surface favours leaching (Whittaker et al., 2013), contributing to the loss of the work-
(Tang et al., 2012) and enhances dissolution (Tromans and Meech, ability. The layer of hydration products could also be one of the
2001), potentially increasing the hydraulic activity (Paya´ et al., reasons for decelerated hydration of aged AAS cement (Whittaker
1997; Scian et al., 1991) of the AAS cement. et al., 2013), which resulted in delayed setting times (Table 4).

Workability of AAS concretes

Workability of AAS concretes (as measured with Vebe and slump
tests) was low, as indicated in Table 3. AAS concrete containing
SC45-390 had a slump of 35 mm, while workability of concrete
containing aged SC45-390 cement was 10 mm slump. The
reduction in workability could be caused by the increase in surface
area of cement due to the formation of hydration products during
the ageing process (Figure 2(d)) and the interlocking effect caused
by increased roughness of the cement particles (Whittaker et al.,
2013). AAS concretes containing SF6-610 and SF6L-620 cements
had Vebe workability of 18 and 10 s, respec-tively. Sodium
lignosulfonate increased workability of the AAS concrete,
indicating that water demand of the dry powder AAS cement
100 nm SF6L-620 was reduced, which is critical in AAS concrete
technology. The AAS concretes could be produced at lower W/B
(a) ratios. High W/B ratios impair strength development of AAS
concretes by increasing porosity of the concretes due to the increase
in water content (Wang et al., 1994). The increase in W/B ratio also
changes the ionic and anionic concentration of the alkaline solution
due to the effect of diluting (Shi and Day, 1996; Shi and Li, 1989)
and, as a result, the pH of the solution decreases as the amount of
alkali in the AAS cement is fixed. Reduced pH would significantly
decelerate strength development of AAS cement at early age (Roy
et al., 1992). It is even more critical for sodium carbonate as the
initial pH of its aqueous solution is relatively low in comparison to
most activators usually used in AAS concrete production. As a
result of reduced pH, initial hydration and early strength could be
significantly impaired by the additional water introduced at high W/
B ratios.
100 nm

Compressive strength
(b)
Strength development of the dry powder AAS cements is
Figure 1. SEM images of AAS cement containing accelerating sensitive to activator content, especially when AAS concretes are
admixture of silica fume and slaked lime cured at ambient temperature. An increase in activator content
from 3.5 t o 4 .5% sodium oxide results in a 47% increase
in
5
10 μm 1 μm

(a) (b)

3 μm 200 nm

(c) (d)

Figure 2. SEM images of aged AAS cement SC45-390

compressive strength at 28 d from 29 to 42 MPa (Figure 4). The increase in strength can be achieved by steam curing of the AAS
sensitivity of the strength development to activator content is concretes (Figure 4). The high sensitivity of AAS concretes to
significantly reduced if AAS concretes are steamed. Only 16% heat treatment has also been reported by other researchers
difference in compressive strength at 28 d can be observed (Bakharev et al., 1999a, 1999b).
between AAS concretes containing 3 .5 and 4 .5% sodium oxide
(Figure 4). The difference in strength drops even further to 14% As mentioned above, it is interesting to note that the relative
when the W/B ratio is reduced from 0 .35 to 0 .30, and to 13% for strength difference between concretes containing AAS cements
cements with increased fineness (Figure 5). with different concentration of sodium carbonate decreases sig-
nificantly with heat treatment. The AAS cements are more
Results presented in Figure 4 also show that the strength develop- sensitive to activator content when hardening at ambient tempera-
ment of concretes containing the dry powder AAS cements, made tures. The best concentration of the activator in terms of com-
from neutral GBFS and sodium carbonate, is slow at ambient pressive strength gained by AAS concrete will thus depend on the
temperature. Samples could not be demoulded after 1 d and application of AAS cement; that is, the cements for precast
strength testing was carried out after 2 d, although hydration production can be produced with lower activator content, which
progresses with time, providing reasonable strengths at 28 d will reduce the cost of AAS cements and reduce the possibility of
(Figure 4). Increased curing temperature significantly improves efflorescence formation on the surfaces of concrete products
the performance of AAS concretes by accelerating strength (Wang et al., 1995).
development, especially at early age (Figures 4 and 5). Compres-
sive strengths in the range from 57 to 65 MPa were obtained at A reduction in water content leads to a significant increase in
28 d depending on the alkali concentration. More than 50% compressive strength of AAS concretes containing binders of the
6
 90
80

Compressive strength: MPa

70
60
50
40 SC35-390
SC40-390
30 SC45-390
20 SC35-430
10 SC40-430
SC45-430
0
0 7 14 21 28 35 42 49 56 63 70 77 84 91
Time: d
10 μm
Figure 5. Strength development of AAS concretes with W/B ratio
(a)
0.30 steamed at 858C

390 series (Figure 5 compared to Figure 4); with more than 25%
strength increase immediately after steaming and about 20% at
28 d. This is a big advantage of dry powder AAS cements over
traditional AAS technology, where the alkali is introduced into
the concrete mixture as an aqueous solution. The amount of
alkaline solution cannot be reduced to a comparative extent
because it will reduce the alkali concentration in AAS concrete
due to a limit of sodium carbonate solubility in water at a certain
temperature. This advantage increases the flexibility of AAS
concretes containing dry powder AAS cements: the concretes can
1 μm be designed to have the same strength properties with increased
workability or to have higher strength performance with the same
(b)
workability of fresh concrete mix, in comparison to traditional
Figure 3. SEM images of fresh AAS cement SC45-390 AAS concretes.

Increased fineness of dry powder AAS cement improved the

compressive strength of AAS concretes by only 7–10% at 28 d
SC35-390
(Figure 5). The increase is even less at 91 d, varying between 0
SC40-390
SC45-390 and 9%. AAS concretes containing SF3-590 and SF3-610 have
SC35-390 steamed almost no difference in compressive strength at 28 and 91 d
90 SC40-390 steamed (Figure 6), although initial strength at 1 d differs substantially.
SC45-390 steamed This confirms the findings of other researchers that increased
80
SC45-390 (aged) steamed
Compressive strength: MPa

fineness of slag mostly affects early-age compressive strength and

70
does not necessarily improve the strength of AAS concretes in
60
the long term (Shi and Li, 1989; Wang et al., 1994).
50
40 As shown in Figure 4, heat treatment is essential for strength
30 development of AAS cements containing the neutral GBFS and
sodium carbonate. Heat treatment very often cannot be done on a
20
construction site. To accelerate hardening of dry powder AAS
10 cements at ambient temperatures, a combination of silica fume
0 with slaked lime was used (Figure 6). It can be seen that the
0 7 14 21 28 35 42 49 56 63 70 77 84 91
optimum amount of the accelerating admixture was 6 wt%.
Time: d
Although 2 wt% concentration of the admixture provides the
Figure 4. Strength development of AAS concretes with W/B ratio highest strength in the long term, it does not accelerate strength
0 .35 development at early age sufficiently to be practical. In contrast
10 wt% concentration of combined silica fume and slaked lime
7
 90 SF2-500 prehydration of OPC for only 28 d at 85% relative humidity
80 SF6-590
Compressive strength: MPa

SF10-570
caused around 13% loss in compressive strength of mortars at
70 28 d. A loss of less than 25% in compressive strength during
SF6-610
60 SF6L-620 storage in extreme conditions (99% relative humidity) for
50 2 months could be a good indicator of sufficient shelf life for the
40 dry powder AAS cements.
30
Conclusions
20
The following conclusions can be drawn from the results of this
10
research.
0
0 7 14 21 28 35 42 49 56 63 70 77 84 91
Time: d j Sodium carbonate concentration in dry powder AAS cements
can be varied depending on the cement application. The
Figure 6. Strength development of AAS concretes with W/B ratio optimum concentration is 3 .5% sodium oxide for precast
0.35 at 258C accelerated with combinations of silica fume and production with steam curing. For cast in situ application and
slaked lime curing at ambient temperatures, 4 .5% sodium oxide should be
considered the optimum.
j Heat curing is essential for strength development of dry
effectively accelerates strength development at early age but the powder AAS cements containing neutral GBFS and sodium
long-term compressive strength is significantly impaired. As carbonate, but these cements can also be used for in situ
opposed to 2 and 10 wt% dosage, 6 wt% of the accelerating applications where early strength is not critical.
admixture gives a good balance between early and long-term j Combinations of silica fume and slaked lime accelerate
compressive strength for AAS concretes (Figure 6). The compres- strength development of dry powder AAS cements containing
sive strength of AAS concretes increases from 2 .5 MPa at 2 d for neutral GBFS and sodium carbonate to the extent that they
SC45-390 to 10 .3 MPa at 1 d for SF6-590 (Figure 4 compared to can be used for in situ applications without special heat
Figure 6). The effectiveness of the accelerating admixture goes treatment on construction sites. For formulation of dry
up even more with increasing fineness of AAS cement, as powder AAS cements, 6 wt% of the accelerating admixture
illustrated by the fact that AAS concrete containing SF6-610 can be recommended.
gained 24 .5 MPa in compressive strength in 1 d (Figure 6). j The addition of sodium lignosulfonate not only reduces water
Accelerated strength development can be explained by the in- demand, but also prolongs setting times and increases
crease in pH upon dissolution of the slaked lime and by calcium– strength performance of dry powder AAS cements.
silicate–hydrate (C–S–H) formed during rapid reaction between j Potentially, dry powder AAS cements have a sufficient shelf
silica fume and slaked lime. The calcium–silicate–hydrates act as life, but long-term testing is required to confirm this.
crystallisation centres (substrate) for new portions of calcium– j Further research is needed to estimate different engineering
silicate–hydrate produced by the promoted slag hydration (Hubler properties of AAS concretes containing dry powder AAS
et al., 2011; Shi and Day, 1996). The acceleration could also be cement.
explained by the reaction of slaked lime with sodium carbonate,
resulting in the formation of sodium hydroxide that increases Acknowledgements
early-age compressive strength, but the reaction is not possible at The authors would like to thank André Botha and Antoinette
258C (Gibbs energy is 55 .8 kJ/mol). Buys of the Laboratory for Microscopy and Microanalysis of the
University of Pretoria for their assistance in conducting the SEM
The addition of sodium lignosulfonate significantly increases the studies.
compressive strength of dry powder AAS cement with the
accelerating admixture at the same W/B ratio (Figure 6). This REFERENCES
phenomenon could be explained by the improved structure of the Atiş CD, Bilim C, Çelik Ö and Karahan O (2009) Influence of
hydration product caused by a slower hydration process during activator on the strength and drying shrinkage of alkali-
the initial stage (Neville, 1995) and by a possible reaction activated slag mortar. Construction and Building Materials
between slag and the additional sodium brought into the system 23(1): 548–555.
with the plasticiser. Bakharev T, Sanjayan JG and Cheng YB (1999a) Alkali activation
of Australian slag cements. Cement and Concrete Research
After the ageing procedure, dry powder AAS cement SC45-390 29(1): 113–120.
was used to produce concrete with a compressive strength of Bakharev T, Sanjayan JG and Cheng YB (1999b) Effect of
49 MPa after 28 d (Figure 4) in comparison to the strength of elevated temperature curing on properties of alkali-activated
AAS concrete containing fresh AAS cement SC45-390, which slag concrete. Cement and Concrete Research 29(10): 1619–
yielded 65 MPa after 28 d. Whittaker et al. (2013) showed that 1625.
8
Bakharev T, Sanjayan JG and Cheng YB (2000) Effect of strength of alkali activated slag paste. Cement and Concrete
admixtures on properties of alkali-activated slag concrete. Research 41(8): 842–846.
Cement and Concrete Research 30(9): 1367–1374. Moseson AJ, Moseson DE and Barsoum MW (2012) High volume
Bakharev T, Sanjayan JG and Cheng YB (2001a) Resistance of limestone alkali-activated cement developed by design of
alkali-activated slag concrete to alkali-aggregate reaction. experiment. Cement and Concrete Composites 34(3): 328–336.
Cement and Concrete Research 31(2): 331–334. Neville AM (1995) Properties of Concrete, 4th edn. Longman,
Bakharev T, Sanjayan JG and Cheng YB (2001b) Resistance of Harlow, UK.
alkali-activated slag concrete to carbonation. Cement and Payá J, Monzó J, Borrachero MV, Peris E and González-López E
Concrete Research 31(9): 1277–83. (1997) Mechanical treatments of fly ashes. Part III: Studies
Bakharev T, Sanjayan JG and Cheng YB (2003) Resistance of on strength development of ground fly ashes (GFA) –
alkali-activated slag concrete to acidic attack. Cement and Cement mortars. Cement and Concrete Research 27(9):
Concrete Research 33(10): 1607–1611. 1365–1377.
Ben Haha M, Le Saout G, Winnefeld F and Lothenbach B (2011) Purdon AO (1940) The action of alkalis on blast furnace slag.
Influence of activator type on hydration kinetics, hydrate Journal of the Society of Chemical Industry 59: 191–202.
assemblage and microstructural development of alkali Ravikumar D, Peethamparan S and Neithalath N (2010)
activated blast-furnace slags. Cement and Concrete Research Structure and strength of NaOH activated concretes
41(3): 301–310. containing fly ash or GGBFS as the sole binder. Cement and
Bernal SA, Gutierrez RM, Pedraza AL et al. (2011) Effect of Concrete Composites 32(6): 399–410.
binder content on the performance of alkali-activated slag Roy A, Schilling PJ, Eaton HC et al. (1992) Activation of ground
concretes. Cement and Concrete Research 41(1): 1–8. blast-furnace slag by alkali-metal and alkaline-earth
BSI (2005) BS EN 196-3: 2005. Methods of testing cement – hydroxides. Journal of the American Ceramic Society 75(12):
Part 3: Determination of setting times and soundness. BSI, 3233–3240.
London, UK. Scian AN, Porto López JM and Pereira E (1991)
BSI (2009a) BS EN 12350: 2009: Testing fresh concrete. BSI, Mechanochemical activation of high alumina cements –
London, UK. hydration behaviour. I. Cement and Concrete Research 21(1):
BSI (2009b) BS EN 12390-3: 2009: Testing hardened concrete – 51–60.
Part 3: Compressive strength of test specimens. BSI, London, Shi C (1996) Strength, pore structure and permeability of alkali-
UK. activated slag mortars. Cement and Concrete Research
BSI (2010) BS EN 196-6: 2010. Methods of testing cement – 26(12): 1789–1799.
Part 6: Determination of fineness. BSI, London, UK. Shi C and Day RL (1996) Some factors affecting early hydration
Collins F and Sanjayan JG (1999a) Strength and shrinkage of alkali-activated slag cements. Cement and Concrete
properties of alkali-activated slag concrete containing porous Research 26(3): 439–447.
coarse aggregate. Cement and Concrete Research 29(4): 607– Shi C and Li Y (1989) Investigation on some factors affecting the
610. characteristics of alkali-phosphorous slag cement. Cement
Collins FG and Sanjayan JG (1999b) Workability and mechanical and Concrete Research 19(4): 527–533.
properties of alkali activated slag concrete. Cement and Shi C and Stegemann JA (2000) Acid corrosion resistance of
Concrete Research 29(3): 455–458. different cementing materials. Cement and Concrete Research
Douglas E and Brandstetr J (1990) A preliminary study on the 30(5): 803–808.
alkali activation of ground granulated blast-furnace slag. Tang A, Su L and Li C (2012) Effect of dry grinding on the
Cement and Concrete Research 20(5): 746–756. physicochemical properties of silica materials prepared from
Douglas E, Bilodeau A, Brandstetr J and Malhotra VM (1991) kaolin residue. Powder Technology 218(March): 86–89.
Alkali activated ground granulated blast furnace slag Tromans D and Meech JA (2001) Enhanced dissolution of
concrete: preliminary investigations. Cement and Concrete minerals: stored energy, amorphism and mechanical
Research 21(1): 101–108. activation. Minerals Engineering 14(11): 1359–1377.
Dubina E, Black L, Sieber R and Plank J (2010) Interaction of Turner LK and Collins FG (2013) Carbon dioxide equivalent (CO2-
water vapour with anhydrous cement minerals. Advances in e) emissions: a comparison between geopolymer and OPC
Applied Ceramics 109(5): 260–268. cement concrete. Construction and Building Materials
Fernández-Jiménez A, Palomo JG and Puertas F (1999) Alkali- 43(June): 125–130.
activated slag mortars. Mechanical strength behavior. Cement Van Deventer JSJ, Provis JL and Duxson P (2012) Technical and
and Concrete Research 29(8): 1313–1321. commercial progress in the adoption of geopolymer cement.
Glukhovsky VD (1959) Soil-silicates. Gosstroiizdat, Kiev, USSR. Minerals Engineering 29(March): 89–104.
Glukhovsky VD, Krivenko PV, Rostovskaya GS, Timkovich VJ and Wang SD and Scrivener KL (1995) Hydration products of alkali
Pankratov VL (1983) Binder. US Patent 4410365, Oct. activated slag cement. Cement and Concrete Research 25(3):
Hubler MH, Thomas JJ and Jennings HM (2011) Influence of 561–571.
nucleation seeding on the hydration kinetics and compressive Wang SD, Scrivener KL and Pratt PL (1994) Factors affecting the

9
 strength of alkali-activated slag. Cement and Concrete Xu H, Provis JL, Van Deventer JSJ and Krivenko PV (2008)
Research 24(6): 1033–1043. Characterization of aged slag concretes. ACI Materials
Wang SD, Pu XC, Scrivener KL and Pratt PL (1995) Alkali- Journal 105(2): 131–139.
activated slag cement and concrete: a review of properties Yang KH, Song JK, Ashour AF and Lee ET (2008) Properties of
and problems. Advances in Cement Research 7(27): 93–102, cementless mortars activated by sodium silicate. Construction
http://dx.doi.org/10.1680/adcr.1995.7.27.93. and Building Materials 22(9): 1981–1989.
Whittaker M, Dubina E, Al-Mutawa F et al. (2013) The effect of Živica V (2007) Effects of type and dosage of alkaline activator
prehydration on the engineering properties of CEM I Portland and temperature on the properties of alklali-activated slag
cement. Advances in Cement Research 25(1): 12–20, http:// mixtures. Construction and Building Materials 21(7): 1463–
dx.doi/10.1680/adcr.12.00030. 1469.

10