Cement & Concrete Composites 46 (2014) 56–64

Contents lists available at ScienceDirect

Cement & Concrete Composites

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

Effect of chemical composition of slag on chloride penetration resistance

of concrete
Mike Otieno ⇑, Hans Beushausen, Mark Alexander
Concrete Materials and Structural Integrity Research Unit, Department of Civil Engineering, University of Cape Town, Private Bag X3, Rondebosch 7701, South Africa

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

Article history: Three ground granulated slags (FeMn arc-furnace (GGAS), Corex (GGCS) and blastfurnace (GGBS) slags) of
Received 5 June 2013 varying chemical composition, and from different sources were used to make concretes using two w/b
Received in revised form 4 November 2013 ratios (0.40 and 0.60) and three slag replacement levels (20%, 35% and 50%). The effect of chemical com-
Accepted 16 November 2013
position and replacement level of slags on the chloride penetration resistance of the concretes was
Available online 23 November 2013
assessed using the chloride conductivity test. The results showed that the chloride penetration resistance
of concrete increases with decreasing w/b ratio and increasing slag replacement level. In the GGAS con-
Keywords:
cretes, despite having relatively low SiO2 and high MgO content, its significantly high Mn2O3 and low
Slag
Chloride penetration
Al2O3 content was found to have a negative effect on the chloride penetration resistance of the concrete.
Chloride conductivity The significantly high chloride penetration resistance of GGCS concretes was partly attributed to both its
high CaO content and particle fineness. Only GGCS concretes showed a trend of increasing chloride pen-
etration resistance with increased particle fineness; GGBS and GGAS concretes did not show any trend
between particle fineness and chloride penetration resistance. The slag activity index was found to be
a better indicator of chloride penetration resistance in concrete than the slag hydraulic index.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction (approximately 310 m2/kg Blaine) [8–10]. Its primary constituents

The use of supplementary cementitious materials (SCMs) such
as slag, fly ash, silica fume and metakaolin to make blended cement
concretes instead of plain CEM I (PC) concretes is common because
of the positive attributes associated with their use, which generally
relate to improved economic, durability and sustainability proper-
ties. From a chloride-induced reinforcement corrosion perspective,
the use of blended cement concretes is usually preferred to plain
PC concretes mainly due to their improved resistance to chloride
penetrability [1], extended time-to-corrosion initiation [2,3], and
relatively low corrosion rates [4,5]. The improved chloride penetra-
tion resistance and decreased corrosion rate in blended cement
concretes has been observed in both cracked and uncracked con-
cretes [6,7]. This publication focuses on the chloride penetration
resistance of slag-blended cement concretes and examines the
influence of chemical composition of various granulated slags.
Slag is a non-metallic by-product of the iron manufacturing
process. It has a relative density of approximately 2.9, with its
bulk density varying in the range of 1200–1300 kg/m3. It generally
has a higher particle fineness (>350 m2/kg Blaine) than PC
⇑ Corresponding author. Address: Department of Civil Engineering, Private Bag
X3, Rondebosch 7701, Cape Town, South Africa. Tel.: +27 21 650 5180.
E-mail addresses: OTNMIK001@myuct.ac.za, mikexbenjamin@gmail.com (M.
Otieno).
are silica (SiO2), alumina (Al2O3) and quicklime (CaO). Lime is
added as a fluxing agent [9]. It typically replaces 20–50% of PC in
concrete but higher replacement levels have been used in the past
[1,9,11]. On the practical side, high levels of slag replacement may
result in excessive delays in setting times and slower rate of
strength development but by contrast, lower replacement levels
may not produce all of the technical benefits possible with slag
concrete, for example those relating to improved durability. The
commonly used replacement level in the South African construc-
tion industry is 50%.
The hydraulic activity of slag as a cementitious material is influ-
enced by (i) its chemical composition, (ii) the alkali concentration
(pH) of the reacting system, (iii) its glass content, (iv), its particle
fineness, and (v) the temperature during the early phases of the
hydration process [12,13]. Compounds that increase its hydraulic
activity include CaO, MgO and Al2O3 while SiO2 reduces its hydrau-
licity [14,15]. These compounds (CaO, MgO, Al2O3 and SiO2) are
used to assess the suitability of a slag as a partial cement replacing
material in concrete. This is done by calculating the slag hydraulic
index (HI) which can be used as a first indicator to assess the poten-
tial of a slag to form cementitious hydration products in an alkaline
medium such as concrete i.e. it can be regarded as the slag’s
cementitious potential [16]. A number of formulae have been
proposed in the literature to predict the HI of slag (see [17]) but
the most commonly used one is the ratio of the sum of oxides

0958-9465/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.cemconcomp.2013.11.003
 M. Otieno et al. / Cement & Concrete Composites 46 (2014) 56–64
known to increase slag hydraulicity (CaO, MgO and Al2O3) to SiO2
which decreases its hydraulicity [14,15]. However, these formulae
have been criticized for their inadequacy to sufficiently predict the
hydraulic activity of a slag since the hydration reactions that occur
are far more complex than indicated by these formulae [3]. Fur-
thermore, even though the type of PC used has also been shown
to have a significant effect on hydraulic activity of a slag [17], this
is not reflected in these formulae.
Even though the hydration product formed when slag is
blended with PC and water is essentially the same as the principal
product formed when PC hydrates, i.e., calcium silicate hydrate, C–
S–H [18], the properties of hydrated paste systems incorporating
slag such as strength, porosity and heat of hydration are influenced
by the properties of the slag such as chemical and mineralogical
composition, glass content, particle fineness, and the type of acti-
vation provided [8,19].
Partial replacement of PC with slag has been shown to lead to
significant reduction in chloride ion penetrability by densification
of concrete microstructure [20–22]. In its hydration process, it re-
acts not only with calcium hydroxide (Ca(OH)2) but also with water
to form calcium aluminates and C–S–H [23]. The latter product,
additional to that formed by hydration of PC, contributes to refine-
ment of pores and blocking of chloride diffusing paths [24]. An opti-
mum slag replacement level to maximize concrete resistance to
chloride penetration has so far not been established. The commonly
used replacement level of 50% is based solely on compressive
strength considerations [9], and some studies have reported in-
creased chloride resistance as the slag replacement level increases
from 40% to 65% by mass of total cementitious material [25].
The presence of slag in concrete also influences the chemical
chloride binding capacity of the hardened cement paste [26]. This
affects chloride transport in concrete through (a) partial blocking
of pores in concrete, which results from the formation of calcium
chloro-aluminates (Friedel’s salt), (b) demobilization of chloride
ions, or (c) both [24,27]. The main compounds in cement which
take part in chloride binding are tricalcium-aluminate (C3A) and
tetracalcium-aluminoferrite (C4AF) [28]. In the case of plain PC
concrete, its chloride binding capacity can be predicted by applying
the Bogue formulae to estimate the quantity of these compounds
[29]. In the case of blended cements, suitable chloride binding iso-
therms are used e.g. Langmuir and Freundlich isotherms [27,30].
However, the applicability of these isotherms is usually limited
to the chloride contents in the concrete and the SCM replacement
levels used in their derivation. Furthermore, they do not take into
account the chemical composition of the blended cement. Factors
that affect the chloride binding capacity of slag-blended cement
concretes include slag replacement level, w/b ratio, temperature,
pH, carbonation, chloride desorption, sulphate content and particle
fineness of the blended cement [30]. A summary of past literature
on selected factors (relevant to this publication) affecting the chlo-
ride binding capacity of slag blended concretes is presented here:
(i) Slag replacement level: Chloride binding capacity increases
with increasing slag replacement level [31]. This trend is
attributed to the dilution effect of sulphates by slag [32].
The net effect of SCM on the binding capacity of a blended
cement mainly depends on the resulting types and amounts
of calcium–aluminate–hydrate (C–A–H), C–S–H and cal-
cium–aluminate–silicate–hydrate (C–A–S–H) in the cement
paste [30]. These hydrates are influenced by the chemical
composition of the blended cement and the SCM replace-
ment level.
(ii) Water-to-binder (w/b) ratio: Even though previous research-
ers have reported conflicting results with respect to the
influence of w/b ratio on chloride binding capacity, with
some reporting significant increase [33,34] and others
57
reporting insignificant increase [35] with increase in w/b
ratio, the general consensus is that the chloride binding
capacity of a slag-blended cement concrete increases with
increasing w/b ratio [36]. This trend is attributed to the
higher porosity at higher w/b ratios which makes potential
binding sites more exposed and more accessible to chloride
ions [30].
2. Research significance
Even though the partial replacement of PC with slag can result
in significant benefits with respect to chloride penetration resis-
tance of concrete, differences in composition and reactivity of the
slag, depending on the source, in addition to variations in slag
replacement levels can affect its efficacy. An understanding of
the effects of both the chemical compositions and physical charac-
teristics of slag on the durability performance of concrete is there-
fore important, and can help in the development of more durable
concretes. This study is specifically important in the South African
context. As more types of slags are produced, it is necessary that
their durability performance is investigated so that the extent of
their strengths and weaknesses are known. A separate publication
by Beushausen et al. [19] investigated the early-age properties,
strength development and heat of hydration of concrete containing
three types of slags. This publication reports on the same slags, and
focuses on the influence of chemical composition and replacement
level of slags on the chloride penetration resistance of concrete.
3. Experimental details
Three types of slags produced in South Africa were used in this
research:
(i) Ground granulated blastfurnace slag (GGBS): This is the most
common type of slag used in concrete in South Africa. It is
produced when pig iron is manufactured in a blastfurnace.
In the manufacturing process, the iron oxide is reduced to
metallic iron using, as a fluxing agent, limestone or dolomite,
which combines with the silica and alumina constituents in
the ore to form a molten slag, which is then further treated
to produce the finely ground slag used in the manufacture of
concrete.
(ii) Ground granulated Corex slag (GGCS): This slag is produced in
the Corex process which is a more environmentally friendly
process to produce iron. In the Corex process, coke ovens and
a blastfurnace are replaced with a direct reduction shaft and
a melter-gasifier. This process yields a quenched slag, called
GGCS, as a by-product. Changes in the manufacturing pro-
cess inevitably result in subtle differences in chemical and
physical properties of slags produced by the Corex process
in comparison with the blastfurnace slags.
(iii) Ground granulated FeMn arc-furnace slag (GGAS): Recently,
trials in South Africa have shown that modified basic oxygen
furnace steel and arc-furnace steel slags can be granulated
successfully to comply with the requirements of granulated
blastfurnace slag for use in concrete [37]. One of these slags,
namely a ground granulated FeMn arc furnace slag was
included in this investigation. The ground granulated FeMn
slag is a ‘‘Processed Air-Cooled FeMn Arc Furnace Slag’’,
which was re-melted on a pilot plant scale (800 kg) with
the addition of dolomitic limestone to obtain the same com-
position as GGBS. At the same time, the MnO content was
reduced from ca. 20% to between 4% and 6%. The slag com-
plies with all of the requirements for GGBS in terms of SANS
1491-1 [37] and BS EN 15167-2 [38].
58 M. Otieno et al. / Cement & Concrete Composites 46 (2014) 56–64

It is important to note that, of the three types of slags men-
tioned here, only GGBS and GGAS are commonly used in South
Africa. GGAS has not been exploited as a replacement for PC in con-
crete but has the potential of being utilized in future. It is therefore
worth investigating.
Experiments were carried out to establish the effect of the fol-
lowing parameters on chloride penetration resistance of concrete:
(i) type of slag (GGBS, GGCS and GGAS), (ii) slag replacement level
(0%, 20%, 35% and 50% by mass total binder), and (iii) w/b ratio
(0.40 and 0.60). For a given w/b ratio, the binder, aggregate, and
water contents were kept constant for all the slag replacement lev-
els used. Plain Portland cement CEM I 42.5N (PC) was used in all
the cement blends. A summary of the concrete mix proportions
is given in Table 1. The oxide composition (determined by standard
X-ray diffraction (XRF) technique) and particle fineness of the slags
used are given in Table 2. The fineness was measured on both the
unblended and slag-blended cements used. The glass content was
measured in transmitted light with a petrographic microscope as
well as by X-ray diffractomery using silicon as an internal stan-
dard. Aggregates were 19 mm andesite coarse aggregate and natu-
ral quartz sand. No specific tests were performed on the aggregates
as they are standard materials used in local concretes. No concrete
admixtures were used because the fresh concrete properties such
as slump were being investigated, and have been reported in an
earlier publication [19]. The 28-day compressive strengths are gi-
ven in Table 3.
The rapid chloride conductivity test [39–41] was used to test
the efficacy of resistance of the concretes to chloride ion penetra-
tion. This test measures the ionic flux across a concrete disc spec-
imen (68 ± 2 mm diameter and 25 ± 2 mm thickness) due to a 10 V
potential difference (see Fig. 1). The 10 V potential difference was
obtained using a commercial power supply (CICÒ PS1830) with
output ranges of 0–30 V and 0–5 A for, respectively, voltage and
current. The voltage and current were measured using commercial
digital multimeters (BrymenÒ TBM805) with measurement ranges
of 10 A and 600 V. For each of the concrete mixes, two 100 mm
cubes were cast; two discs were obtained by coring from each
cube. Hence, for each concrete mix, four concrete disc specimens
used to carry out the rapid chloride conductivity test. Pre-condi-
tioning of the concrete disc specimens, which typically are cored
from 100 mm cubes, involves oven-drying at 50 ± 2 °C for
7 days ± 4 h followed by vacuum saturation (75 to 80 kPa) in
a 5 M NaCl solution for 1 h ± 15 min, followed by soaking in the
same solution for a further 18 ± 1 h to ensure a saturation equilib-
rium. The specimens are oven-dried at 50 ± 2 °C for 7 days to en-
sure uniform very low moisture contents within concrete pores.
The drying temperature of 50 ± 2 °C is preferred to 100 °C to limit
microstructural damage in the concrete [42]. The drying duration
is adopted based on extensive laboratory tests performed at the
University of Cape Town which have shown that mass loss in the
specimens is negligible after 7 days ± 4 h oven-drying. The oven-
drying helps prevent dilution of the saturating solution (5 M NaCl)
by pore water [43]. The chloride conductivity value is defined as
(Eq. (1)):
it
r¼ ð1Þ
VA
where r is the conductivity of the specimen (mS/cm), i is the cur-
rent (mA), V is the potential difference (V), t is the average thickness
of specimen (cm), and A the cross-sectional area of the specimen
(cm2). Chloride conductivity values of four specimens of the same
concrete are averaged to give one result. An increase in the chloride
conductivity value depicts a decrease in chloride penetration resis-
tance. Typical chloride conductivity index values range from 0.2 to
2.5 mS/cm with the lower limit applying to highly chloride resistant
concretes (e.g. 50/50 PC/GGBS) and the upper limit applying to pen-
etrable concretes (e.g. 100% PC) [44,45]. The test was performed on
the specimens after 28-day curing under water (23 ± 2 °C).
The specimens used in the chloride conductivity test were used
to determine the concrete porosity (n) as follows:
Ms  MD
n¼  100 ð2Þ
Atqs
where Ms is the vacuum 5 M NaCl-saturated mass of the specimen
determined using the procedure already mentioned, Md is oven-dry
mass of the specimen after 7 days ± 4 h oven-drying at 50 ± 2 °C, A
is the cross-sectional area of the specimen, d is the average thick-
ness of the specimen and qs is the density of the 5 M NaCl salt solu-
tion (1.19  103 g/mm3).
4. Results
4.1. Effect of w/b ratio and slag replacement level on chloride
penetrability
The 28-day chloride conductivity results for the different con-
cretes, w/b ratios and slag replacement levels are given in Figs. 2
and 3. The results show that for a given w/b ratio and type of slag
(GGBS, GGCS or GGAS), and regardless of the slag replacement le-
vel, the chloride conductivity values for the PC concretes were
higher (i.e. lower resistance to chloride penetration) than those
for the slag-blended concretes (Fig. 3). This trend is attributed to
the refined microstructure of, and chloride binding in, the blended
cement concretes compared to that of PC due to the formation of
additional C–S–H phases and higher alumina content. Microstruc-
ture and durability properties of cementitious materials are

Table 1
Summary of concrete mix proportions.

Material (kg/m3) w/b Ratio Binder composition

100% PC 50/50 PC/slag 65/35 PC/slag 80/20 PC/slag
Portland cement, PC (CEM I 42.5 N) 0.40 488 244 317 390
0.60 300 150 195 240
Slag (GGBS, GGCS, or GGAS) 0.40 – 244 171 98
0.60 150 105 60
Water 0.40 195 195 195 195
0.60 180 180 180 180
Sand (2 mm max.) 0.40 616 616 616 616
0.60 803 803 803 803
Andesite (19 mm max.)a 1104 1104 1104 1104
a
Applicable to both 0.40 and 0.60 w/b ratio concretes.
 M. Otieno et al. / Cement & Concrete Composites 46 (2014) 56–64 59

Table 2
Chemical composition and physical properties of the binders and various blends.

Binder Percentage composition (%) Finenessa (m2/kg)

CaO Al2O3 SiO2 MgO Fe2O3 Mn2O3 TiO2 P2O5 K2O Na2O LOI Glass
PC (CEM I 42.5N) 64.71 3.82 20.39 2.15 2.64 0.63 0.33 0.03 0.44 0.15 3.1 nab 355
GGCS 34.17 12.74 37.80 11.14 0.65 1.01 1.43 0.01 0.80 0.26 1.28 98 385
GGBS 37.23 15.23 33.67 11.32 0.90 0.07 0.72 0.01 0.60 0.18 1.95 99 410
GGAS 36.96 6.16 33.74 14.79 0.07 6.93 0.27 0.01 0.03 0.07 0.52 93 415

PC/GGBS
80/20 58.62 5.64 24.10 3.97 2.24 0.72 0.54 0.03 0.50 0.17 2.17 na 355
65/35 54.63 6.93 26.67 5.35 1.95 0.78 0.72 0.03 0.55 0.17 1.48 na 350
50/50 49.96 8.27 29.39 6.63 1.65 0.83 0.86 0.02 0.55 0.16 0.76 na 345

PC/GGCS
80/20 59.20 6.16 22.93 3.97 2.29 0.52 0.40 0.03 0.47 0.16 1.98 na 370
65/35 54.91 7.79 24.84 5.31 1.99 0.43 0.46 0.03 0.49 0.15 1.14 na 375
50/50 51.29 9.60 27.20 6.77 1.75 0.35 0.51 0.02 0.49 0.14 0.29 na 380

PC/GGAS
80/20 59.25 4.35 23.25 4.66 2.14 1.90 0.30 0.03 0.32 0.10 2.23 na 360
65/35 55.70 4.67 24.97 6.57 1.81 2.86 0.30 0.00 0.31 0.10 1.60 na 355
50/50 51.68 5.04 27.27 8.51 1.38 3.84 0.30 0.03 0.24 0.08 1.05 na 360
a
Blaine fineness.
b
na: data not available (this was not measured).

Table 3 in this investigation. However, the higher concrete porosity in

Mean 28-day compressive strengths. the slag blended concretes made with 0.60 w/b ratio at a given slag
replacement level (see Fig. 4) should be expected to result in in-
w/b ratio Slag replacement level (%) Compressive strength (MPa)
creased chloride binding because the higher porosity of such con-
PC GGBS GGCS GGAS
cretes makes potential binding sites exposed and more accessible
0.40 0 68.0 – – – to chloride ions [30]. It is important to note, at this point, that
20 – 63.0 71.0 68.0
for both the 0.40 and 0.60 w/b ratios, the GGAS blended concretes
35 – 60.5 75.0 65.0
50 – 55.5 72.0 62.5 performed poorest with respect to chloride penetration resistance
i.e. they had the highest chloride conductivity values (see Figs. 2
0.60 0 40.0 – – –
20 – 41.5 46.0 42.0 and 3). This will be discussed in detail in the next section.
35 – 39.0 48.5 40.0 The results also show that for a given binder type (slag type and
50 – 34.0 49.0 36.5 slag replacement level), the chloride conductivity increases (i.e. the
chloride penetration resistance increases) with increasing w/b ra-
tio (Fig. 5). This trend is attributed to the relatively open pore
structure of the concretes made with 0.60 w/b ratio compared to
DC power source
those made with 0.40 w/b ratio. The porosity results for the various
A concretes show that concretes made with 0.60 w/b ratio had signif-
icantly higher porosities than those made with 0.40 w/b ratio
(Fig. 4), and that porosity decreased with increasing slag replace-
V ment level (Fig. 6(a) and (b)). The increase in porosity with increase
in w/b ratio is well documented in the literature [29]. For a given
5 M NaCl solution
w/b ratio and type of slag, the chloride conductivity decreased with
increasing slag replacement levels from 20% to 35% to 50%
(Fig. 7(a)–(c)). This trend can, again, be attributable to the increas-
ing degree of microstructure densification and chloride binding
Carbon anode
Steel cathode

Plastic
with increasing slag replacement level. It is important to mention
tubes
that even though the trends observed in this section have been, to
some extent, attributed to chloride binding, this is based on the
assumption that the phenomenon actually takes place. Further
Concrete experimental work is required to ascertain this.
specimen
Rubber
LEFT CELL collar RIGHT CELL
4.2. Effect of oxide composition on chloride penetrability
Fig. 1. Schematic of the chloride conductivity test [40].
From Table 2, it can be seen that for a given type of slag, the oxi-
des Al2O3, MgO and SiO2 increase with increasing slag replacement
directly related, as has been reported by several researchers level, while that of CaO (as expected) decreases with increasing
[46,47]. While it is acknowledged that some degree of chloride slag replacement level. These oxides were initially focused on be-
binding takes place when chloride ions come into contact with cause they are known to increase or decrease slag’s chemical reac-
the aluminate hydration products during the specimen pre-condi- tivity as a cementitious material as follows: (i) increased SiO2
tioning stage of the chloride conductivity test (i.e. pre-saturation content has been reported to decrease slag hydraulicity [14,15],
with 5 M NaCl), the degree to which it occurs was not quantified (ii) high Al2O3 content of slag results in a lower degree of hydration
60 M. Otieno et al. / Cement & Concrete Composites 46 (2014) 56–64

1.6 0.40 w/b ratio

Chloride conductivity (mS/cm)

Key
1.4 Max. = Average + t = standard deviation
t = 2.353 (for sample size, n = 4,
1.2 at 95% confidence interval,)
Average
1.0 0.99 1.00
0.8 Min. = Average – t
0.69
0.6 0.66

0.4 0.38 0.41

0.30 0.35
0.27
0.2 0.17
0.0
PC/GGBS PC/GGCS PC/GGAS PC/GGBS PC/GGCS PC/GGAS PC/GGBS PC/GGCS PC/GGAS
PC 50/50 PC/slag 65/35 PC/slag 80/20 PC/slag

Fig. 2. 28-day chloride conductivity results for PC and slag-blended concretes – 0.40 w/b ratio.

1.8 0.60 w/b ratio

Chloride conductivity (mS/cm)

1.6
1.48
1.4
1.31
1.2 1.22

1.0 0.98 1.01

0.8 0.80
0.6 0.65 0.66 0.65

0.4 0.42

0.2
0.0
PC/GGBS PC/GGCS PC/GGAS PC/GGBS PC/GGCS PC/GGAS PC/GGBS PC/GGCS PC/GGAS
PC 50/50 PC/slag 65/35 PC/slag 80/20 PC/slag

Fig. 3. 28-day chloride conductivity results for PC and slag-blended concretes – 0.60 w/b ratio.

12 45° line 1.6

GGCS-20 GGCS-35 GGCS-20 GGCS-35 45° line

11 GGCS-50 GGBS-20 1.4 GGCS-50 GGBS-20

GGBS-35 GGBS-50
chloride conductivity (mS/cm)

GGBS-35 GGBS-50 1.2

10
GGAS-20 GGAS-35 GGAS-20 GGAS-35
0.40 w/b ratio

1.0
porosity (%)

0.40 w/b ratio

9 GGAS-50 PC GGAS-50 PC
0.8
8
0.6
7
0.4
6
0.2

5
5 6 7 8 9 10 11 12 0.0

0.60 w/b ratio 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
Porosity (%) 0.60 w/b ratio
chloride conductivity (mS/cm)
Fig. 4. Effect of w/b ratio on concrete porosity (Note: 50, 35, 20 refer to percentage
slag replacement level). Fig. 5. Effect of w/b ratio on chloride penetrability (Note: 50, 35, 20 refer to
percentage slag replacement level).
at early ages (up to 28 days) a coarse pore structure (pore size:
0.05–5 lm) [15,48], and (iii) increased MgO content results in a de- hydraulicity, it is neither possible from these results to quantify
crease in porosity and the formation of more hydrotalcite, C(–A)– the extent to which each oxide affected the refinement of the con-
S–H, which has a lower density and thus a higher degree of space crete microstructure nor attribute the increase in concrete chloride
filling than C–S–H [15,48]. penetration resistance with increase in slag replacement level to a
An analysis of the oxide composition of the different slags given specific oxide.
in Table 2 show that even though for a given type of slag, there As already mentioned in the previous section, for both the 0.40
was, as expected, an increase in Al2O3, MgO and SiO2 contents with and 0.60 w/b ratios, the GGAS blended concretes performed poor-
increasing slag replacement level, the percentage increases were est with respect to chloride penetration resistance i.e. they had the
different, with MgO showing higher percentage increases (up to highest chloride conductivity values (see Figs. 2 and 3). This poorer
41%) than both SiO2 (up to 10%) and Al2O3 (up to 26%), see Table 4. performance of GGAS is in contrast to what would be expected
It is also important to note that of these oxides (CaO, MgO, Al2O3 considering its MgO and Al2O3 contents at a given replacement le-
and SiO2), MgO and Al2O3 showed, for a given type of slag, the vel. For a given slag replacement level, GGAS had the highest MgO
highest coefficients of variation for each of these oxides for a given and lowest Al2O3 content. These oxides, as already mentioned, are
slag replacement level (see Table 5). However, even though these expected to, respectively, increase and decrease slag hydraulic
four oxides have, in past studies, been observed to affect slag activity and hence the concrete’s durability performance. However,
 M. Otieno et al. / Cement & Concrete Composites 46 (2014) 56–64 61

(a) 12 0.40 w/b ratio 80/20 PC/GGCS (b) 12 0.60 w/b ratio
80/20 PC/GGBS
11 11
80/20 PC/GGAS
10 10

Porosity (%)
9 9

8 8

7 7 80/20 PC/GGCS
80/20 PC/GGBS
6 6
80/20 PC/GGAS
5 5
15 20 25 30 35 40 45 50 55 15 20 25 30 35 40 45 50 55
Slag replacement level (%) Slag replacement level (%)

Fig. 6. Effect of slag replacement level on concrete porosity (Note: 0.40 and 0.60 w/b ratio).

Chloride conductivity, mS/cm

Chloride conductivity, mS/cm

Chloride conductivity, mS/cm

1.4
(a) (b) (c)
1.6 1.4
(20% slag replacement)

(35% slag replacement)

(0% slag replacement)

1.4 1.2 1.2

1.2 1.0 45° line 1.0
45° line 45° line
1.0 0.8 0.8
0.8
0.6 0.6
0.6
0.4 0.4
0.4 GGCS-0.4 GGCS-0.6 GGCS-0.4 GGCS-0.6 GGCS-0.4 GGCS-0.6
0.2 GGBS-0.4 GGBS-0.6 0.2 GGBS-0.4 GGBS-0.6 0.2 GGBS-0.4 GGBS-0.6
GGAS-0.4 GGAS-0.6 GGAS-0.4 GGAS-0.6 GGAS-0.4 GGAS-0.6
0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Chloride conductivity, mS/cm Chloride conductivity, mS/cm Chloride conductivity, mS/cm
(20% slag replacement) (35% slag replacement) (50% slag replacement)

Fig. 7. Effect of slag replacement level on chloride penetrability (Note: 0.4, 0.6 refer to w/b ratio).

blast-furnace slags having MnO content (5.4–21%), they found that

Table 4
Percentage increase in selected oxides with increase in slag replacement level. the higher the MnO content, the lower the early-age (up to 28-day)
strength and the higher the long-term (90-day) strength.
Blended Increase in slag replacement level Percentage increase
In short, high levels of manganese seem to inhibit the early-age
cement from (%) in oxide content
reactivity of the slag. Therefore, taking into account the oxide com-
SiO2 MgO Al2O3
position of the GGAS blended cements and that the chloride con-
PC/GGBS 20–35 10.7 34.8 22.9 ductivity test was carried out on 28-day old concrete specimens,
35–50 10.2 23.9 19.3 it can be concluded that the high Mn2O3 had a negative effect on
PC/GGCS 20–35 8.3 33.8 26.5 slag reactivity and hence on its potential to refine the concrete
35–50 9.5 27.5 23.2 pores. However, considering the results by Pera et al. just pre-
PC/GGAS 20–35 7.4 41.0 7.4 sented here, it can be expected that the chloride penetration resis-
35–50 9.2 29.5 7.9
tance of the GGAS concretes may have improved if the chloride
conductivity test was performed at a later age, say 56 days.
Nevertheless, this does not mean that the GGAS concretes
Table 5 would have performed better than the other slag-blended con-
Coefficient of variation of selected oxides for the slag-blended cements. cretes at later ages. This is because, considering the compressive
Slag replacement level (%) Coefficient of variation (%) strength developments of the three slags at different replacement
levels (20%, 35% and 50%) and w/b ratios (0.40 and 0.60) reported
CaO MgO Al2O3 SiO2
by Beushausen et al., [19] it is clear that even at an age of 56 days,
80/20 PC/slag 0.6 9.5 17.3 2.6
the GGCS, and not GGAS, had the highest strength (see [19]). The
65/35 PC/slag 1.0 12.5 24.9 4.0
50/50 PC/slag 1.8 14.3 30.7 4.5 results by Pera et al. may, therefore, only be applicable to slags
with similar oxide composition except for MnO content. In a case
where the other oxides also vary, the general trend of lower
from a chloride binding point of view, the poor performance of early-age strength and the higher the long-term strength with in-
GGAS concretes can partly be attributed to the low Al2O3 content creased MnO content may not be applicable.
which contributes to the aluminate phases that participate in chlo- In summary, the effect of Al2O3, MgO, SiO2 or Mn2O3 on the
ride binding. There was no significant difference in percentage hydraulic reactivity of slag-blended cements depends not only on
composition of SiO2 in the three slags at a given replacement level. the quantity of each oxide but also on the effect brought about
From Table 2, the only other oxide which showed high variation by the other oxides.
in the slag blended cements at a given replacement level was
Mn2O3. At a given replacement level, the Mn2O3 content in GGAS 4.3. Effect of glass content on chloride penetrability
was highest compared to the other types of slags, and increased
with increase in replacement level (up to 3.7, 6.7 and 11 times The glass content is also an important factor that influences the
higher in GGAS than in GGBS and GGCS at, respectively, 20%, 35% hydraulic activity of ground slag [8]. The glass content of all three
and 50% replacement levels). In a study by Pera et al. [49] using slag types investigated was above 90%, as shown in Table 2. The
62 M. Otieno et al. / Cement & Concrete Composites 46 (2014) 56–64

significantly lower glass content of GGAS (93%), compared to GGBS blended cement mortar cubes (at a replacement level of 50%) to
(98%) and GGCS (99%) suggests a lower reactivity and may have the compressive strength of the plain cement, reference mortar
also contributed to its low chloride resistance (high chloride con- cubes at a designated age. In their publication using the same con-
ductivity values) observed; the reasonably similar reactivity of crete mix proportions used in this paper, Beushausen et al. [19] re-
GGBS and GGAS concretes (see [19]) shows that other factors such ported that the SAI values for the three slags increased with time
as oxide composition and fineness (especially for GGCS concrete – but that of GGCS mortar was significantly higher than those for
this is discussed later) influenced slag reactivity. Even though the the other slags at all ages (2, 7, 14 and 28 days). These results also
extent to which the glass content contributed to the slag reactivity show that the SAI is a better indicator of slag performance than HI.
cannot be quantified, research shows that it is not as important for It can be concluded that HI is a good initial (theoretical) indicator
its reactivity as its chemical composition, provided there is ade- of a slag’s hydraulic performance but SAI is a better indicator of its
quate glass content (generally > 90%) to enable the slag to show actual performance i.e. reactivity. In the same publication, they re-
satisfactory properties as a hydraulic binder [50]. In summary, ported that the 28 and 56-day compressive strengths (on the same
the effect of glass content on the chloride resistance of the different concretes) increased with decreasing w/b ratio (0.60–0.40), and
concretes cannot be conclusively quantified based on the limited with type of slag in the order GGBS ? GGAS ? GGCS. However,
results. Further tests are required to verify this trend. contrary to the chloride conductivity values where the GGCS val-
ues were superior (i.e. lower) than those for the other concretes
only at the 50% replacement level, the GGCS strengths were higher
4.4. Effect of slag reactivity on chloride penetrability
than those for GGBS and GGAS at all replacement levels (see Fig. 8).
These results also support the superior performance of the
The slag hydraulic indices (HI = (CaO + MgO + Al2O3)/SiO2) for
GGCS concretes, similar to those obtained by Mackechnie et al.
the three slags GGCS, GGAS and GGBS are, respectively, 1.9, 1.7
[9]. The lack of matching trends between the HI of the different
and 1.5. As indicated earlier, the HI is interpreted to give an indica-
slags and the increase in chloride penetration resistance in the
tion of a slag’s potential to react and form cementitious products,
slag-blended concretes at 20% and 35% partial replacement of PC
in this case C–S–H (additional to that of PC) which leads to refine-
(at a given w/b ratio) clearly suggests that a single replacement le-
ment of the concrete pore structure. Therefore, based on the HI of
vel cannot be universally applied to all types of slags. Each type of
the three slags, it was expected that, at a given w/b ratio and
slag may require a unique optimum replacement level in order to
replacement level, the chloride penetration resistance of the con-
maximize on the durability performance of a concrete. Considering
cretes would increase (or the chloride conductivity values would
that the poor performance of GGAS concretes was earlier attrib-
decrease) in the order GGBS ? GGAS ? GGCS. However, none of
uted to the high Mn2O3 content in GGAS, in the case of GGCS, its
the concretes followed this trend. The trends obtained for a given
superior performance can also be partly attributed to its very low
w/b ratio varied from one slag replacement level to another except
Mn2O3 content of 0.07% compared to 1.01 and 6.93 of, respectively,
for the 50/50 PC/slag replacement level where, for both the 0.40
GGBS and GGAS.
and 0.60 w/b ratios, a consistent trend of increasing chloride con-
ductivity value in the order GGCS ? GGBS ? GGAS was obtained;
however, this still did not follow the trend for the HI. These results
confirm the doubts that have been cast by previous researchers 4.5. Effect of binder fineness
regarding the reliability of HI to predict the potential performance
of a slag, chloride penetration resistance in this case. As already At the various slag replacement levels (20%, 35% and 50%), the
mentioned, the performance of a slag-blended concrete depends different slag blended cements had different particle finenesses
on several factors, inter alia the cementitious system studied ranging from 345 to 380 m2/kg (see Table 2). However, considering
[15]. The concept of HI can therefore not be extended to predict a given slag replacement level for the different slags, very small
the chloride penetration resistance of slags especially considering coefficients of variation (for particle fineness) of 2.1%, 3.7% and
the chemical and mineralogical influences they have on chloride 4.9% for, respectively, 20, 35 and 50 slag replacement levels were
penetration. observed. In this investigation, except for the GGCS blended con-
The superior performance of the GGCS concrete at the 50% slag cretes (Fig. 9), no general correlation could be established between
replacement level for both 0.40 and 0.60 w/b ratios conforms to the binder fineness (due to type of slag and replacement level) and
trend of the slag activity indices (SAI) obtained for these slags by chloride resistance for all the slag blended concretes. It is impor-
Beushausen et al. in an earlier publication [19]. ASTM C 989 [51] tant to mention that the influence of slag fineness was not a focus
defines the SAI as the percentage ratio of the strength of slag- of this investigation.

1.4 Chloride conductivity 80

28-day compressive strength (MPa)
Chloride conductivity (mS/cm)

Compressive strength 70
1.2
60
1.0
50
0.8
40
0.6
30
0.4
20
0.2 10

0.0 0
GGCS

GGBS

GGCS

GGBS

GGCS

GGBS

GGCS

GGBS

GGCS

GGBS

GGCS

GGBS
GGAS

GGAS

GGAS

GGAS

GGAS

GGAS

80/20 65/35 50/50 80/20 65/35 50/50 PC/Slag

0.40 w/b ratio 0.60 w/b ratio

Fig. 8. Trends for 28-day compressive strength and chloride conductivity for different slag replacement levels.
 M. Otieno et al. / Cement & Concrete Composites 46 (2014) 56–64 63

1.2 (f) Considering the oxides that are known to affect slag hydrau-
Chloride conductivity index (mS/cm)

80/20 PC/GGCS
licity (CaO, MgO, Al2O3 and SiO2), MgO and Al2O3 showed,
1.0 0.60 w/b ratio
for a given type of slag, the highest coefficients of variation
65/35 PC/GGCS at a given slag replacement level. However, it was neither
0.8
possible from these results to quantify the extent to which
0.40 w/b ratio
0.6 50/50 PC/GGCS
each oxide affected the refinement of the concrete micro-
structure nor attribute the increase in concrete chloride pen-
0.4 etration resistance with increase in slag replacement level to
a specific oxide.
0.2

0.0
365 370 375 380 385 Acknowledgements
Blaine fineness (m2 /kg)
The authors acknowledge the support of the University of Cape
Fig. 9. Effect of binder fineness on chloride penetrability of GGCS blended concrete. Town, the erstwhile Cement and Concrete Institute (C&CI), Sika Pty
Ltd. (SA), Ash Resources Pty Ltd, Concrete Manufacturer’s Associa-
5. Summary and conclusions tion, PPC Ltd and AfriSam South Africa.

This paper investigated the influence of chemical composition References

of different granulated slags, slag replacement level, w/b ratio
and glass content on concrete resistance to chloride penetration. [1] Song HW, Saraswathy V. Studies on the corrosion resistance of reinforced steel
A summary of the results obtained are as follows: in concrete with ground granulated blast-furnace slag – an overview. J Hazard
Mater 2006;138(2):226–33.
[2] Mackechnie JR. Predictions of reinforced concrete durability in the marine
(a) At a given slag replacement level, regardless of the type of environment. PhD Thesis, Department of civil engineering, University of Cape
slag used, chloride conductivity and porosity increased with Town, 1996.
[3] Mangat PS, El-Khatib JM, Molloy BT. Microstructure, chloride diffusion and
increasing w/b ratio (from 0.40 to 0.60), implying that the reinforcement corrosion in blended cement paste and concrete. Cem Concr
chloride penetrability also increased. Compos 1994;16(2):73–81.
(b) At a given w/b ratio, regardless of the type of slag used, chlo- [4] Otieno MB, Alexander MG, Beushausen H-D. Corrosion in cracked and
uncracked concrete – influence of crack width, concrete quality and crack
ride conductivity and porosity decreased with increasing
re-opening. Mag Concr Res 2010;62(6):393–404.
slag replacement level (from 20% to 35% to 50%), giving [5] Scott AN, Alexander MG. The influence of binder type, cracking and cover on
improved chloride penetration resistance. corrosion rates of steel in chloride-contaminated concrete. Mag Concr Res
(c) Regardless of the type of slag used, the slag hydraulic index 2007;59(7):495–505.
[6] Aldea CM, Shah SP, Karr A. Pemeability of cracked concrete. Mater Struct
(HI) was not a reliable performance indicator of the slag- 1999;32(5):370–6.
blended concretes with respect to chloride penetration [7] Mohamed B, Sakai K, Banthia N, Yoshida H. Prediction of chloride ions ingress
resistance. By contrast, the slag activity index (SAI), in uncracked and cracked concrete. ACI Mater J 2003;100(1):38–48.
[8] Pal SC, Mukherjee A, Pathak SR. Investigation of hydraulic activity of ground
appeared to be a more reliable indicator. At the 50% slag granulated blast furnace slag in concrete. Cem Concr Res 2003;33(9):1481–6.
replacement level, for both the 0.40 and 0.60 w/b ratios, a [9] Mackechnie JR, Alexander MG, Jaufeerally H. Structural and durability
consistent trend of increasing chloride conductivity (reduc- properties of concrete made with Corex slag. Research monograph No. 6.
Department of civil engineering, University of Cape Town, 2003
ing chloride resistance) in the order GGCS ? GGBS ? GGAS [10] Grieve G. Cementitious materials. Fulton’s Concrete Technology. Published by
was obtained but, this did not follow the trend for the HI. Erstwhile Cement and Concrete Institute, Midrand, South Africa, 2009. p. 1–16,
(d) For both the 0.40 and 0.60 w/b ratios, the GGAS blended con- ISBN 978-0-9584779-1-8. [Chapter 1].
[11] Beushausen H, Alexander M, Ballim Y. Early-age properties, strength
cretes performed poorest with respect to chloride penetra- development and heat of hydration of concrete containing various South
tion resistance. The poor performance of GGCS blended African slags at different replacement ratios. Constr Build Mater
concretes contrary to expectations was attributed to its 2012;29:533–40.
[12] ACI-233R-03. Slag cement in concrete and mortar, ACI 223R-03, American
low Al2O3 and significantly high Mn2O3 content compared
Concrete Institute. Farmington Hills, MI: American Concrete Institute,
to the other types of slags. Even though no literature exists American Concrete Institute. 2003
on the effect of Mn2O3 on slag hydraulicity, these results [13] Escalante JI, Gomez LY, Johal KK, Mendoza G, Mancha H, Mendez J. Reactivity
suggest that it may have had a negative effect on slag reac- of blastfurnace slag in Portland cement blends hydrated under different
conditions. Cem Concr Res 2001;31(10):1403–9.
tivity (e.g. suppressing the effect of MgO) and hence on its [14] Frearson JPH. Sulphate resistance of combination of Portland cement and blast
potential to refine the concrete pores. In fact, an index incor- furnace slag. ACI Special Publ 1986;91:1495–524.
porating the influence of the content of Mn2O3 would pro- [15] Haha MB, Lothenbach B, Le Saout G, Winnefeld F. Influence of slag chemistry
on the hydration of alkali-activated blastfurnace slag – Part II: Effect of Al2O3.
vide a better estimate of the chloride resistance of a slag Cem Concr Res 2012;42(1):74–83.
concrete that the HI. However, a detailed study would be [16] Bougara A, Lynsdale C, Milestone NB. Reactivity and performance of
required to achieve such as index. blastfurnace slags of differing origin. Cem Concr Compos 2010;32(4):319–24.
[17] Mantel DG. Investigation into the hydraulic activity of five granulated blast
(e) The superior performance of the GGCS concrete at 50% slag furnace slags with eight different portland cements. ACI Mater J
replacement level for both 0.40 and 0.60 w/b ratios con- 1994;91(5):471–7.
formed to the trend of the slag activity indices (SAI) and [18] Kramer W. Blastfurnace slags and slag cements. In: Proceedings of the 4th
international symposium on the chemistry of cement. Washington DC, USA,
compressive strength development for these slags obtained vol. 2, 1962, p. 957–973, Cited in Bougara A, Lynsdale C, Milestone NB.
in an earlier publication [19]. However, contrary to the chlo- Reactivity and performance of blastfurnace slags of differing origin. Cement
ride conductivity values where the GGCS values were higher and Concrete Composites 2010; 32(4): 319–24.
[19] Beushausen H, Alexander M, Ballim Y. Early-age properties, strength
than those for the other concretes only at the 50% replace-
development and heat of hydration of concrete containing various South
ment level, the GGCS strengths were higher than those for African slags at different replacement ratios. Constr Build Mater
GGBS and GGAS at all replacement levels. The superior per- 2012;29(4):533–40.
formance of GGCS was partly attributed to its very low [20] Yeau KY, Kim EK. An experimental study on corrosion resistance of concrete
with ground granulate blastfurnace slag. Cem Concr Res 2005;35:1391–9.
Mn2O3 content and particle fineness, in addition to its high [21] Rose JH. Effect of cementitious blastfurnace slag on chloride permeability of
CaO content. concrete. ACI Special Publ 1987;102:107–26.
64 M. Otieno et al. / Cement & Concrete Composites 46 (2014) 56–64
[22] Malhotra VM. Properties of fresh and hardened concrete incorporating ground
granulated blast furnace slag. In: Malhotra VM, editor. Supplementary
cementing materials for concrete. Canada: Minister of Supply and Services;
1987. p. 291–336.
[23] Jaul WC, Tsay DS. A study of the basic engineering properties of slag cement
concrete and its resistance to seawater corrosion. Cem Concr Res
1998;28(10):1363–71.
[24] Leng F, Feng N, Lu X. An experimental study on the properties of resistance to
diffusion of chloride ions of fly ash and blast furnace slag concrete. Cem Concr
Res 2000;30(6):989–92.
[25] ACI-233R-95. Ground granulated blastfurnace slag as a cementitious
constituent in concrete, ACI 223R-95, American Concrete Institute.
Farmington Hills, MI: American Concrete Institute, American Concrete
Institute. 1995.
[26] Arya C, Xu Y. Effect of cement type on chloride binding and corrosion of steel in
concrete. Cem Concr Res 1995;25(4):893–902.
[27] Glass G, Hassanein N, Buenfeld N. Neural network modelling of chloride
binding. Mag Concr Res 1997;49(181):323–35.
[28] Boddy A, Bentz E, Thomas MDA, Hooton RD. An overview and sensitivity study
of a multi-mechanistic chloride transport model. Cem Concr Res
1999;29(6):827–37.
[29] Mehta PK, Monteiro PJM. Concrete – microstructure, properties and
materials. New York: McGraw Hill; 2006.
[30] Zibara H. Binding of external chlorides by cement pastes. PhD Thesis, Graduate
department of civil engineering, University of Toronto. 2001.
[31] Dhir RK, El-Mohr MAK, Dyer TD. Chloride binding in GGBS concrete. Cem
Concr Res 1996;26(12):1767–73.
[32] Xu Y. The influence of sulphates on chloride binding and pore solution
chemistry. Cem Concr Res 1997;27(12):1841–50.
[33] Arya C, Buenfeld NR, Newman JB. Factors influencing chloride binding in
concrete. Cem Concr Res 1990;20(2):291–300.
[34] Nagataki S, Otsuki N, Wee TH, Nakashita K. Condensation of chloride ion in
hardened cement matrix materials and on embedded steel bars. ACI Mater J
1993;90(4):323–33.
[35] Tritthart J. Chloride binding in cement: II – The influence of hydroxide
concentration in the pore solution of hardened cement paste on chloride
binding. Cem Concr Res 1989;19(5):683–91.
[36] Tang L, Nilsson LO. Chloride binding capacity and binding isotherms of OPC
pastes and mortars. Cem Concr Res 1993;23(2):36–42.
[37] SANS-1491-1. Portland cement extenders Part 1: Ground granulated
blastfurnace slag, 2 ed. South African Standard Code of practice, 2005.
[38] BS-EN-15167-2. Ground granulated blast furnace slag for use in concrete,
mortar and grout. Conformity evaluation: British-Adopted European Standard,
ISBN: 0580489337, 2006. p. 28.
[39] Streicher PE. The development of a rapid chloride test for concrete and its use
in engineering practice. PhD Thesis, Department of Civil Engineering,
University of Cape Town. 1997.
[40] Streicher PE, Alexander MG. A chloride conduction test for concrete. Cem
Concr Res 1995;25(6):1284–94.
[41] Beushausen HD, Alexander MG. The South African durability index tests in an
international comparison. J S Afr Inst Civ Eng (Saice) 2008;50(1):25–31.
[42] Pigeon MJ, Garnier F, Pleau R, Aitcin P-C. Influence of drying on the chloride ion
permeability of HPC. Concr Int 1993;15(2):65–9.
[43] Streicher PE, Alexander MG. A chloride conduction test for concrete. Cem
Concr Res 1995;25(6):1284–94.
[44] Alexander MG, Ballim Y, Stanish K. A Framework for use of durability indexes
in performance-based design and specifications for reinforced concrete
structures. Mater Struct 2008;41(5):921–36.
[45] Alexander MG. Durability indexes and their use in concrete engineering. In:
Kovler K, Marchand J, Mindess S, Weiss J. (Eds.), Proceedings of the
International RILEM Symposium on Concrete Science and Engineering – A
Tribute to Arnon Bentur, Evanston, Illinois, USA. March 21–24, 2004. p. 9–22.
[46] Ortega M, Sánchez I, Climent MA. Durability related transport properties of
OPC and slag cement mortars hardened under different environmental
conditions. Constr Build Mater 2012;27(1):176–83.
[47] Sánchez I, López MP, Ortega JM, Climent MA. Impedance spectroscopy: an
efficient tool to determine the non-steady-state chloride diffusion coefficient
in building materials. Mater Corros 2011;62:139–45.
[48] Haha MB, Lothenbach B, Le Saout G, Winnefeld F. Influence of slag chemistry
on the hydration of alkali-activated blast-furnace slag – Part I: Effect of MgO.
Cem Concr Res 2011;41(9):955–63.
[49] Pera J, Ambroise J, Chebannet M. Properties of blast-furnace slags containing
high amounts of manganese. Cem Concr Res 1999;29(2):171–7.
[50] Gruskovnjak A, Lothenbach B, Winnefeld F, Figi R, Ko S-C, Adler M. Hydration
mechanisms of super sulphated slag cement. Cem Concr Res 2008;38:983–92.
[51] ASTM-C-989. Standard specification for slag cement for use in concrete and
mortars. ASTM International, 100 Barr Harbour Drive, West Conshohocken, PA
19428–2959, USA. 2012.