Cement and Concrete Composites 85 (2018) 22e31

Contents lists available at ScienceDirect

Cement and Concrete Composites

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

Alkali-activated slag concrete: Fresh and hardened behaviour

F. Puertas a, *, B. Gonza

lez-Fonteboa b, I. Gonza

lez-Taboada b, M.M. Alonso a,
M. Torres-Carrasco , G. Rojo , F. Martínez-Abella b
a b

a
Eduardo Torroja Institute for Construction Sciences (IETcc-CSIC), C/Serrano Galvache 4, 28033 Madrid, Spain
b
Department of Construction Technology, University of A Corun ~ a, E.T.S.I. Caminos, Canales, Puertos. Campus Elvin
~ a s/n, 15071 La Corun
~ a, Spain

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

Article history: The behaviour of fresh and hardened alkali-activated slag (AAS) and OPC concretes was compared and
Received 16 March 2017 the effect of mixing time assessed. OPC and AAS concrete slump and rheological results proved to differ,
Received in revised form particularly when the slag was activated with waterglass (WG). The nature of the alkaline activator was
27 September 2017
the key determinant in AAS concrete rheology. Bingham models afforded a good fit to all the OPC and
Accepted 3 October 2017
Available online 4 October 2017
AAS concretes. In OPC and NaOH-activated AAS concretes, longer mixing had an adverse effect on
rheology while improving hardened performance only slightly. In WG-AAS concrete, longer mixing
times, improved mechanical properties and also rheological behaviour was enhanced, in which those
Keywords:
Rheology
conditions were required to break down the microstructure. Longer mixing raised thixotropy in OPC and
Concrete NaOH-activated AAS concretes, but lowered the value of this parameter in waterglass-activated slag
Alkaline activation concrete.
Blast furnace slag © 2017 Elsevier Ltd. All rights reserved.
Bingham model
Hardened behaviour

1. Introduction These AAS systems are also characterised by high mechanical

strength and durability [17e22]. Nonetheless AAS cement, mortar
For over 20 years, the study and development of alkaline acti- and concretes exhibit two shortcomings that hamper their stand-
vated materials (AAMs), mainly cements and concretes has been ardisation and more general use, namely high drying shrinkage
the subject of great interest by the scientific community [1e6]. That [2,23e25] and quick setting when activated with waterglass solu-
is largely due to the enormous promise of these products as sus- tions [26e28].
tainable alternatives to portland cement-based binders. AAMs are The reason for such quick setting lies in the initial and speedy
natural or artificial aluminosilicates activated by hydroxide-, sili- formation of a primary C-S-H gel, which hardens sodium silica-
cate- or alkaline carbonate-solutions. Their manufacture not only activated AAS pastes. Shi and Day [29] used isothermal conduc-
generates lower GHG (especially CO2) emissions, but is more tion calorimetry to study slag activation with different activators.
economically and environmentally sustainable, for the raw mate- They observed two contiguous signals in the pre-induction period
rials or precursors used are often valorised industrial by-products in sodium silicate-activated pastes. The first was attributed to slag
[7e10]. The manufacture of some alkaline activators, sodium sili- dissolution and the second to the formation of a primary C-S-H gel
cate or waterglass (WG) especially, also carries adverse environ- as a result of the presence of [Si04]4- ions in the solution. This latter
mental impacts, however [11]. The production of possible signal intensified as the WG silica modulus rose. Primary C-S-H
alternative activators by processing different types of waste (such formation was also observed in constant shear rate rheology tests
as glass or rice husk ash) would consequently constitute a further [28,30] on waterglass-activated AAS pastes. After an initial rise,
step toward sustainability [12e15] in new binder development. shear declined and flattened, attendant upon paste deflocculation
Alkali-activated slag (AAS) cements and concretes are among possibly attributable to primary C-S-H flocks gel breakdown.
the most prominent AAMs, for their manufacture emits less GHGs Lengthening mixing time from 3 min to 10 min was also shown to
than portland cement and consumes less energy and water [16]. retard initial setting by 40 min and final setting by over 4 h. At a
mixing time of 30 min, initial and final settings were also delayed
[26].
* Corresponding author.
More recently, the intensity and timing of this rheological signal
E-mail address: [email protected] (F. Puertas). at constant shear rate in WG-activated AAS pastes have been shown

https://doi.org/10.1016/j.cemconcomp.2017.10.003
0958-9465/© 2017 Elsevier Ltd. All rights reserved.
 F. Puertas et al. / Cement and Concrete Composites 85 (2018) 22e31 23

to depend on the Na2O concentration and the SiO2/Na2O modulus superplasticisers [51e53] and the composition and particle size
in the activating solution used. The higher the latter and conse- distribution of the aggregates [54e57].
quently the greater the silicon content in the solution, the earlier In contrast, rheological studies on alkali-activated concretes are
and more intense the signal [31]. scant to non-existent. The addition of ultrafine fly ash [58] or lime
Such rapid setting obviously affects rheology and therefore on- slurry [59] has been observed to improve AAS concrete workability
site paste and concrete casting. As the literature shows [28,30,31], (lower lost of slump). No information has yet been forthcoming on
rheological studies contribute to an understanding of the reasons AAS concrete rheological parameters, however. More specifically,
for such behaviour and the pursuit of possible solutions. Few no studies comparing NaOH- or sodium silicate-activated AAS to
studies have focused on researching sodium silicate-activated AAS OPC concrete rheology have been published to date. As noted,
system rheology, however. moreover, as longer mixing times may induce initial breakdown of
Palacios et al. [28,30] found that NaOH-activated AAS pastes fit a the primary C-S-H gel in WG-AAS systems, improving the work-
Bingham model, whereas WG-activated materials conform to the ability as has been proven in previous studies on AAS paste and
Herschel-Bulkley model. Studies comparing OPC and AAS rheology mortars [31,60]. The motivation of this study was to know the effect
based on minislump test-determined workability [31,32] showed of longer mixing time of AAS concrete rheology with different
that WG-activated pastes were less workable than NaOH-activated alkaline activators (NaOH and Wg) and to compare the results with
and OPC pastes. Kashani et al. [33], studying hydroxide- and alka- OPC concretes. Mixing time must be factored into experimental
line silicate-activated AAS paste rheology, found that the latter had designs to determine NaOH- and WG-activated AAS concrete
lower initial shear stress than the former due to the adsorption of workability and rheology.
silicate ions on the particles, which had a fluidising effect as a result
of the increase in the double electrical layer. When these results 2. Experimental procedure
were studied in conjunction with isothermal conduction calorim-
etry findings on degree of reaction, the times at which yield stress 2.1. Materials
rose were found to be correlated to the heat release pattern. Those
data were consistent with the results reported by Varga et al. [31], CEM I 52.5 R portland cement (OPC) and ground-granulated
where initial yield stress was observed to be lower in WG- than in blast furnace slag (S) were used as binder to prepare the con-
NaOH-activated AAS pastes. Unlike Varga et al., however, Kashani cretes studied. The chemical composition of these materials was
et al. [33] observed that yield stress was lower in the first 20 min in listed in Table 1 and determined by a PHILIPS PW-1004 X-Ray
WG- than in alkaline hydroxide-activated AAS pastes. Different Fluorescence (XRF) spectroscometer, together with the loss on
authors may on occasion report contradictory findings due to var- ignition (found further to European standard EN-196-2:2014). The
iations in the activators used and paste mixing history, as well as in ground-granulated blast-furnace slag used in this study had a
the specific rheological study conducted. Blaine specific surface of 346 m2/kg and a specific gravity of
Fluidity loss has been shown to be greater in sodium silicate- 2880 kg/m3. The cement used, the chemical composition for which
activated AAS mortars than in OPC and NaOH-activated AAS mor- is given in Table 1, had a Blaine specific surface of 420 m2/kg and a
tars [2,32]. Waterglass-activated AAS mortars have also been found density of 3100 kg/m3. The rolled sand fines had a specific gravity of
to exhibit lower workability and greater fluidity loss than alkali- 2600 kg/m3 and an absorption of 0.23%, both coarse siliceous ag-
activated fly ash mortars [34,35]. gregates had a specific gravity of 2640 kg/m3, with 0.36% absorp-
Sight should not be lost of the fact that the end product is tion in the 4e8 mm material and 0.40% absorption in the 8e12 mm
concrete, however. An understanding of its workability and gravel (Table 2). The particle size distribution of the coarse aggre-
rheology is consequently of cardinal importance [36,37]. Empirical gates is shown in Fig. 1.
findings such as yielded by the slump test, V-funnel or J-ring [38]
not only furnish very valuable information on concrete work- 2.2. Mixture proportions and mixing protocols
ability, but can be readily and economically conducted. As operator-
sensitive, single-point tests [39,40], they may nonetheless find two The blast furnace slag and cement concretes were designed to a
very different concretes to be in the same workability class. AAS total content of 357 kg/m3. The OPC concrete was mixed with water
concrete systems should therefore be studied with multi-point and 0.1% (cement weight) chemical admixture, whereas the alkali-
tests that furnish information on rheological parameters such as activated slag concretes were mixed with one of two solutions
yield stress, plastic viscosity and information about thixotropy (Table 3):
[39e42].
Portland cement concrete rheology has been exhaustively - Panreac 98% pure, 10 M NaOH (5% Na2O by slag mass)
studied and found to fit a Bingham model (Equation (1)): - Sodium silicate hydrate or waterglass (5% Na2O by slag mass,
SiO2/Na2O modulus of 1.2), prepared with Merck sodium silicate
t ¼ t0 þ m$g_ (1) (SiO2: 27 wt%; Na2O: 8 wt%; H2O: 65 wt%).

where g_ is shear rate (s1), t0 is yield stress (an indication of the A slump of 120 mm or larger was sought in all cases.
force needed to initiate flow) and m is plastic viscosity (Pa$s) or the Two mixing protocols were deployed to prepare both the OPCC
material's resistance to flow [40,43]. OPC thixotropy has also been and AASC systems (see Fig. 2). Protocol 1 corresponds to AAS
explored by many authors [44e46]. Thixotropy is the property of concrete dosage previously determine in RILEM TC 247-DTA (the
these systems whereby viscosity gradually declines with rising stability and good mechanical strenght behaviuor were the main
shear rate and the initial structure is recovered when shear is factors considered). In protocol 2 a longer mixing time of all con-
removed [40,47]. It should not be confounded with the irreversible crete components were considered according to previous rheo-
‘structural breakdown’ in cement pastes defined by Tattersall and logical studies [26,31,60].
Banfill [40]. In the first (protocol 1), the aggregates were blended for 2 min in
Any number of studies have shown that portland cement con- the mixer with 1.65% of the water. The aggregates were then
crete rheology, thixotropy and workability vary with the presence allowed to stand for 8 min to absorb the mixing water. The binder
of additions [48e50], the presence and composition of (portland cement or ground-granulated blast furnace slag) were
24 F. Puertas et al. / Cement and Concrete Composites 85 (2018) 22e31

Table 1
Chemical composition of blast furnace slag and 52.5R OPC (XRF analysis).

Material Component (oxide; wt%)

a
SiO2 Al2O3 Fe2O3 MnO MgO CaO Na2O K2O TiO2 P2O5 LoI

Slag 38.71 10.46 0.33 0.21 7.58 40.62 0.54 0.35 0.37 0.05 0.30
OPC 19.32 7.90 2.69 0.06 0.81 61.81 0.11 1.02 0.23 0.06 2.90
a
LoI ¼ Loss on ignition at 1000  C.

Table 2
Physical properties of natural aggregates.

Aggregate Maximum grain size (mm) Specific gravity (kg/m3) Water absorption (%)

Sand 0e4 mm 4 2600 0.23

Coarse (4e8 mm) 8 2640 0.40
Coarse (8e12 mm) 12 2640 0.36

The same stages were followed in protocol 2, although the times

varied. The aggregate, binder and liquid were mixed for 11 min
rather than 5 min. To maintain the same 21.5 min total time as in
protocol 1, that 6 min difference was adjusted as follows: the rest
time after blending liquid and aggregate was shortened by 3 min
(ending at 7 min), the aggregate-binder blending time by 1 min, the
standing time after mixing aggregate, binder and liquid by 1 min
and the final mixing time by 1 min (Fig. 2(b)).
Hereinafter, the concretes will be named with a suffix denoting
the mixing protocol used.

2.3. Testing program

The slump test described in European standard EN-12350-2 was

conducted on all the fresh concretes at each test age. The fresh
concrete was also analyzed in an ICAR portable rotational rheom-
eter fitted with a four-bladed vane able to ensure axial symmetry at
variable speeds (Fig. 3). Static (or at-rest) yield stress was deter-
Fig. 1. Particle size distribution of coarse aggregates in OPCC and AASC concretes.
mined by stress growth test, while Bingham parameters for dy-
namic yield stress and plastic viscosity were found by flow curve
test method. In this study, rheology tests were performed at 15 min,
Table 3 30 min, 45 min and 60 min after the liquid was added to the solid.
Concrete batching (per 1 m3). After the mixing process, part of the concrete was poured from
Material (kg) OPCC AASC N AASC WG the mixer into the rheometer container. At each testing age, the
slump test was carried out. Before beginning this test, the concrete
Cement 357.0 e e
Blast furnace slag e 357.0 357.0 was remixed 30 s. At the same time, the rheological tests were
Water 197.5 177.75 145.0 started from the concrete being at rest:
Water in solution e 19.75 52.5
Fine aggregate 720.5 711.3 563.0 I. The stress growth test started when the vane immersed in the
Coarse aggregate 4-8 540.3 533.5 548.0
Coarse aggregate 8-12 540.3 533.5 728.3
concrete. A constant speed of 0.025 rps was applied as torque
L/S 0.55 0.55 0.55 was monitored on the computer screen.
Admixture (wt%) 0.1 e e II. When peak torque was reached, the vane was removed and
NaOH and sodium silicate e 23.0 42.5 the concrete was remixed with a shovel.
SiO2/Na2O modulus e e 1.2
III. The vane was reinserted and the flow curve test started. The
pH solution 7.0 13.6 13.8
material rotated at a constant 0.5 rps for 20 s, after which torque
was measured at decreasing speeds.
IV. The vane was removed at the end of the test and the concrete
added and mixed with aggregates during 2 min 30 s. At this was re-shovelled and allowed to stand until the following test
moment, 90% solution (water or alkaline solution) was added and time.
the reference time started when the solution was in contact with
the binder. Two minutes later, the rest of the liquid was added, The concrete was moulded into 100 mm cubic specimens,
along with the admixtures in the OPCC systems. In this mixing demoulded after 24 h and cured in a chamber for 7 d or 28 d, when
protocol, the mixing time between the aggregates, binder and so- it was tested for compressive strength as per European standard EN
lution (water or alkaline activator) was 5 min. Finally, the mixture is 12390-3. The testing frame was programmed to gradually increase
allowed to stand for 2 min and again mixed for 2 additional minutes the load at a rate of 0.6 ± 0.2 MPa s1 to failure.
before to start the rheological tests. Entire test lasts 21 min 30 s (see Cylindrical 150 mm  200 mm specimens were also prepared
Fig. 2(a)). for the water permeability test, conducted on the 28 d samples to
 F. Puertas et al. / Cement and Concrete Composites 85 (2018) 22e31 25

Fig. 2. Mixing protocols for OPCC and AASC: (a) protocol 1; (b) protocol 2; (note differences in mixing and standing times).

in a special device fitted with a mechanical stirrer and a syringe to

add the liquid to the solids and record the reaction data from the
very earliest stages. The liquid/solid ratios used were the same as to
prepare the concretes (Table 3).

3. Results and discussion

The results obtained for the fresh and hardened concretes are
described below.

3.1. Fresh behaviour

The slump values for the OPCC and AAS concretes (AASC N and
AASC WG) in Fig. 4(a) showed that the AASC (and particularly the
waterglass-activated) concretes had a larger slump than the OPC
concretes.
Similar findings have been reported for mortars [60]. Mixing
Fig. 3. ICAR rheometer. time was also observed to affect slump, for longer times lowered
fluidity in all the mortars. As expected, the slump for OPCC and
AASC N declined at longer test times. In the AASC WG concretes, the
European standard EN 12390-8. These specimens were placed in a decline was so steep that it could not be measured at 30 min or
water tank connected to an air compressor fitted with a valve to 45 min (see Fig. 4(b). Mixing protocol 2 had a beneficial effect on
adjust the pressure, which was applied at 500 kPa (5 bar) for 72 h. these concretes, which remained fluid for longer periods.
They were then tested for splitting tensile strength as per European Different models such as the Herschel-Bulkley model or the
standard EN 12390-6 and the water penetration depth was modified Bingham model can be considered to describe the
measured on the split specimens. Density was found in the fresh behaviour of certain vibrated or self-consolidating concretes.
concrete as described in European standard EN 12350-6 and den- However, as the applied rheological model should be as simple as
sity of hardened concrete in the 1 d cured 100 mm concrete cubes possible, the Bingham model (Equation (1)) for yield stress and
as per EN 12390-7. plastic viscosity is recommended.
Total porosity and pore size distribution were found for the 7 d
and 28 d specimens on a Micromeritics Autopore IV 9500 mercury t ¼ t0 þ m$g_
intrusion porosimeter.
Before applying this model, the effect of thixotropy should be
Hydration was monitored in all systems with a Thermometric
verified by checking for torque equilibrium at each rotational ve-
TAM Air isothermal conduction calorimeter. The pastes were mixed
locity on the flow curve, whilst the mix should be examined
26 F. Puertas et al. / Cement and Concrete Composites 85 (2018) 22e31

Fig. 4. (a) Slump test findings for the concretes studied, where suffix denote the mixing protocol used; (b) 22.5 min slump for AASC WG prepared with mixing protocol 1.

visually for segregation both during and after the test. These and dynamic yield stress and lower plastic viscosity than the alkali-
measurements may describe apparently non-linear behaviour due activate materials. More specifically, the highest yield stress was
to thixotropy and segregation and should be eliminated [61]. observed in OPCC -2, although it shared very similar plastic vis-
In this study, this possible non-linear behaviour was checked cosity with OPCC -1. Dynamic yield stress barely rose in the first
and as it did not occur, in light of the foregoing, the Bingham model 45 min in the OPC concretes, and by only 20% after 60 min, whilst
was applied, the results of which are shown in Fig. 5 for all the plastic viscosity rose by 40% after 60 min.
concretes studied. The 15 min static and dynamic yield stress values were higher
It can be seen that the standard Bingham model accurately and viscosity was lower in the AASC N than in the AASC WG con-
represents all the systems studied, for as the coefficients of deter- cretes. The values of these parameters were lowest in protocol 1,
mination found for all the curves denoted good correlations. sodium hydroxide (N) mixes and somewhat higher in the sodium
Therefore the flow curves for the concretes studied could be silicate hydrate (WG) materials mixed to the same protocol. When
assumed to fit a linear model. mixed to protocol 1, the AASC N concretes exhibited 60 min dy-
The standard Bingham model accurately represented all the namic yield stress 52% higher than at the outset and 43% higher
systems studied, for as the coefficients of determination found for when mixed to protocol 2, whilst plastic viscosity grew by 46% in
all the curves denoted good correlations. protocol 1 and 44% in protocol 2.
The Reiner-Riwlin equations [62] for the Bingham model were These parameters rose so dramatically in protocol 1 AASC WG
applied to the experimental data to convert the torque-rotational concretes that the rheological tests could not be conducted after
velocity relationship into a shear stress-shear rate relationship, 15 min, or after 22.5 min with protocol 2, at which time the rheo-
expressed in fundamental units (Pa and Pa$s) (Equations (2) and logical parameters were observed to rise significantly.
(3)). This procedure transforms a relationship between torque Further to the aforementioned findings, longer mixing times
and rotational velocity into a relationship between shear stress and raised dynamic yield stress in both OPCC and AASC N. Whilst the
shear rate, expressed in fundamental units (Pa and Pa$s). mixing protocol did not appear to affect plastic viscosity in OPCC
! concretes, viscosity rose at later ages when the AASC N mixes were
1 1
prepared to protocol 2. Longer mixing times improved AASC WG
R21
 R22 rheology (dynamic yield stress and plastic viscosity).
t0 ¼  G (2) Some authors describe three mixing stages [63] (Fig. 7). The
4phln RR12 first, in which the solid is dispersed in the liquid, is characterised by
the transition from initially high inter-particle forces (calling for
! high mixer power) to a suspension with a significant decline in such
1
forces. The second or optimum stage exhibits asymptotic power
R21
 R12 demand. It delivers the maximum slump and encompasses the
2
mp ¼ H (3) optimal mixing time. Upon finalisation of this stage, inter-particle
4p2 h
where: and particle-mixer collisions may induce wear on the aggregate
and cement particles and abrasion in the early hydration products.
R1 is the vane radius (m) The result is the third or ‘over-mixing’ stage, i.e., the appearance of
R2 is the outer container radius (m) new reaction nuclei that raise rheological parameters and accel-
h is the vane height (m) erate the decline in workability.
G is the y-intercept on the flow curve (Nm) The rise in plastic viscosity and yield stress attendant upon
H is the slope of the flow curve (Nm$s) longer mixing times in OPCC and AASC N was indicative of a more
compact fresh microstructure. While occasioning over-mixing in
The findings on plastic viscosity and dynamic and static yield these materials, protocol 2 appeared to generate optimum stage
stress are shown in Fig. 6. conditions in the AASC WG concretes.
In the 15 min trials, the OPC concretes exhibited higher static As gels seem to be generated at very early ages in AASC WG,
 F. Puertas et al. / Cement and Concrete Composites 85 (2018) 22e31 27

Fig. 5. ICAR rheometer-determined flow curves for: (a) OPCC-1; (b) OPCC-2; (c) AASC N-1; (d) AASC N-2; (e) AASC WG-1 and (f) AASC WG-2.

longer mixing times would be required to break down the micro- activated concretes mixed to protocol 2 procedures, whereas those
structure and improve rheological behaviour in these materials. same procedures reduced the stress in materials activated with
When a 15 min at-rest period was used, static yield stress rose waterglass so effectively that rheological parameters could even be
by 26% between 15 min and 30 min in both OPC concretes (Fig. 6(c)) measured after 22.5 min.
and remained steady after 45 min (also with 15 min at-rest). After The calorimetric findings afforded an explanation for the
60 min (15 min at-rest), however, yield stress remained constant behaviour induced by protocol 2. Further to those findings (Fig. 8),
under mixing protocol 1 only, rising significantly under protocol 2. heat of hydration declined in the OPC and AAS N pastes at around
In the latter, the over-mixing effect induced by the longer mixing 20 min-60 min (induction period), i.e., approximately the duration
time might have accelerated hydration. These findings infer that of the rheology tests, an indication that no significant hydration-
the longer mixing protocol prompted significant and irreversible induced gel formation took place in this stage. In contrast, under
coagulation among cement particles (decline in workability). the same conditions the AAS WG pastes exhibited a first heat flow
Static yield stress rose over time (with a constant 15 min at-rest peak, attributed to the formation of a C-S-H gel. Gel formation
period) in the concretes activated with either solution. The would explain the rheological behaviour of the WG-activated
waterglass-activated samples mixed to protocol 1 could not be alkaline concretes, which would improve under protocol 2 due to
tested in the rheometer (for times of over 15 min) due to the high the partial destruction of such primary gels.
torque generated at the test speed. See Fig. 6(d) for a photograph of Slump values are plotted against dynamic yield stress per den-
this concrete 30 min after mixing. Static yield stress rose in NaOH- sity in these concretes in Fig. 9 [64,65]. The density values are listed
28 F. Puertas et al. / Cement and Concrete Composites 85 (2018) 22e31

Fig. 6. (a) Plastic viscosity; (b) dynamic yield stress; and (c) static yield stress for the concretes studied; (d) 30 min AASC WG-1.

Fig. 7. Effect of mixing time [63]. Fig. 8. Calorimetric curves for OPC, AAS N and AAS WG pastes.

in Table 4. greater the difference between peak and equilibrium shear stress,
An analysis of the data in Fig. 9 showed that both the OPC and the higher the thixotropy [66,67]. As noted earlier, the test could
AAS concretes, irrespective of the mixing protocol, exhibited not be run on AAS WG1 irrespective of the rest time.
similar trends that were, moreover, consistent with the patterns The data in Fig. 10(b) show that longer mixing times generated
reported in the literature [65]. In other words, the relationship greater thixotropy in OPCC and AASC N. Those results, which are
between empirical (slump) and rheological (dynamic yield stress) consistent with the rheological results, can be attributed to the
findings were similar in NaOH- and WG-activated AAS concretes formation of fine particles induced by over-mixing [63,68]. In
and both were similar to the data for OPC concretes [64,65]. contrast, lengthening the mixing time in AASC WG favoured system
Concrete thixotropy was assessed as the difference between the deflocculation [31], thanks to which the test could be conducted;
peak shear stress and the shear stress at equilibrium in the Stress thixotropy was found to be much higher in this than in the OPC and
Growth Test at 30 min (15 min of resting time), except for the AASC AASC N systems. At shorter mixing times (protocol 1), which
WG 2 at 22.5 min (7.5 min of resting time) (see Fig. 10 (a)). The
 F. Puertas et al. / Cement and Concrete Composites 85 (2018) 22e31 29

Fig. 9. Slump vs dynamic yield stress/density.

hampered deflocculation, these measurements could not be taken

under the protocol conditions established for this study. The find-
ings suggest that longer and perhaps more energetic mixing may
have yielded less thixotropic AASC WG concretes.

3.2. Hardened behaviour

Fig. 11 shows the 7 d and 28 d compressive strength findings for

all the concretes, as well as the 28 d total porosity data. The highest
strength was observed for the slag concrete alkali-activated with
the commercial sodium silicate solution (waterglass), with a 28 d
value upward of 60 ± 4.5 MPa. Such performance was comparable
to earlier results [20] in which the use of a commercial sodium Fig. 10. (a) Test protocol; (b) peak stress at equilibrium in the concretes.
silicate hydrate (waterglass) induced high strength due to the sili-
con present in the solution. Lower 28 d strengths, on the order of
16 ± 0.4 MPa, were found when slag was activated with a 5% Na2O low in portland cement as expected on the grounds of its me-
solution of NaOH. Portland cement exhibited strength 25% lower chanical strength, however. The explanation lies in the micro-
than AASC WG at both ages studied. Mechanical strength appeared structural differences (due to the formation of different hydration
to be scantly affected by the mixing protocol used, although when products) in alkali-activated and portland cement concretes that
mixed to protocol 2 the materials exhibited slightly higher strength. determine the way the cementitious matrices bond to the aggre-
An increase in mixing time may have favoured reorganisation of the gates. In alkali-activated systems, aggregate and cementitious
system particles, yielding a somewhat more orderly and compact matrix bond well, affording these systems high mechanical stability
microstructure that would explain these findings. [20].
The strength findings were related to the total porosity (shown Another explanation for the difference in mechanical perfor-
as per cent of total sample volume), density and water permeability mance between alkali-activated and portland cement concretes lies
data shown in Table 4. in the difference in their respective hydration products: C-A-S-H
The NaOH-activated slag concretes had 16 %e20% total porosity, and C-S-H gels. The main product of blast furnace slag activation is a
whereas the lower total porosity (7 %e8%) in the sodium silicate- calcium aluminosilicate hydrate (C-A-S-H) gel, similar to the gel
activated materials was attributed to the presence of silicon in generated in ordinary portland cement hydration, but with a lower
the medium. These total porosity data were directly related to the Ca/Si ratio, which generally ranges from 0.9 to 1.2 [69e73]. When
compressive strength values observed. Total porosity was not as the activator used is a sodium silicate hydrate (waterglass), the

Table 4
28 d water permeability front, total porosity (%) and density in the concretes studied.

Concrete Water Permeability front (mm) Total Fresh density (kg/m3) Hardened density (kg/m3)
porosity (%)

OPCC -1 11.0 16.4 2342.7 2359.4

OPCC -2 6.8 12.4 2337.9 2362.2
AASC N-1 11.0 16.8 2325.3 2371.1
AASC N-2 9.5 19.3 2339.7 2366.5
AASC WG-1 0 8.0 2385.8 2406.6
AASC WG-2 0 7.6 2348.8 2390.1
30 F. Puertas et al. / Cement and Concrete Composites 85 (2018) 22e31

 Longer mixing times had an adverse effect on dynamic and

static yield stress and prompted significant and irreversible
coagulation among cement particles (decline in workability) in
OPC and AAS N concretes. In contrast, longer mixing times were
favourable in AAS WG concrete, in which they would be
required to break down the microstructure to enhance rheo-
logical behaviour. These results proved that for AAS Wg concrete
preparation is recommended to longer mixing time. That would
have effect on future standardization of these eco-efficient
building materials.
 Thixotropy rose in OPC and AAS N concretes with mixing time,
whereas in AASC WG the opposite was observed. The findings
reported hereunder showed that in waterglass-activated AAS
concretes, longer or more energetic mixing would yield less
thixotropic materials.
 The mixing protocol also appeared to affect concrete rheology
(fresh behaviour) more than its strength (hardened behaviour).
In AASC WG, both rheological and mechanical behaviour
Fig. 11. 7 and 28 d compressive strength and 28 d total porosity. improved at longer mixing times. Mixing protocol 2 had an
adverse effect on OPCC and AASC N rheology, while improving
the performance of the hardened materials only slightly.
silicon present in the activating solution, and consequently in the
system, induces a decline in the Ca/Si ratio to around 0.8. That in
Acknowledgments
turn prompts the formation of longer mean chain lengths and
structures comparable to tobermorites [72], yielding an end prod-
This study was funded by the Spanish Ministry of the Economy
uct that exhibits high mechanical strength and the low porosity
under Projects BIA2013-47876-C2-1-P and BIA2014-58063-R.
required to ensure durability.
Further to the water penetration data (Table 4), permeability
front depths were similar in the OPCC and AASC N systems, and References
slightly shallower with protocol 2 than with protocol 1, a finding
ndez-Pereira, M. Izquierdo, Ontributions to the study
[1] Y. Luna-Galiano, C. Ferna
consistent with the strength data. Longer mixing times yielded of porosity in fly ash-based geopolymers. Relationship between degree of
higher strength and durability. Water failed to penetrate the AASC reaction, porosity and compressive strength, Mater. Construcc 66 (324) (2016)
e098.
WG systems, an observation likewise consistent with the me-
[2] F. Pacheco-Torgal, J.A. Labrincha, C. Leonelli, A. Palomo, P. Chindaprasirt,
chanical behaviour observed. In these systems, the silicon sourced Handbook of Alkali-activated Cements, Mortars and Concretes, Woodhead
from the activating solution induced the formation of longer gels, Publishing series in civil and structural engineering, 2015.
which would account for the low porosity of the final products. [3] C. Shi, P. Krivenko, D. Roy, Alkali-activated Cements and Concretes, Taylor and
Francis, London and New York, 2006.
Fresh and hardened concrete density data are shown in Table 4. [4] M.C.G. Juenger, F. Winnefeld, J.L. Provis, J.H. Ideker, Advances in alternative
The AASC WG concretes were the densest and most compact in cementitious binders, Cem. Concr. Res. 41 (2011) 1232e1243.
both states. These concretes are known to exhibit a very compact [5] J.L. Provis, A. Palomo, C. Shi, Advances in understanding alkali-activated ma-
terials, Cem. Concr. Res. 78 (2015) 110e125.
microstructure [72]. The longer mixing time (protocol 2) appeared [6] B. Singh, G. Ishwarya, M. Gupta, S.K. Bhattacharyya, Geopolymer concrete: a
to have a beneficial effect on these concretes. In OPCC and AASC N, review of some recent developments, Constr. Build. Mater 85 (2015) 78e90.
which exhibited similar and lower densities, the mixing procedure [7] C. Shi, A. Jime
nez-Ferna
ndez, A. Palomo, New cements for the 21st century:
the pursuit of an alternative to Portland cement, Cem. Concr. Res. 41 (2011)
appeared to have less of an impact than in (especially the fresh) 750e763.
AASC WG concretes. [8] J.S.J. Van Deventer, J.L. Provis, P. Duxson, Technical and commercial progress
The mixing protocol appeared to affect concrete rheology more in the adoption of geopolymer cement, Min. Eng. 29 (2012) 89e104.
[9] J.L. Provis, Geopolymers and other alkali activated materials: why, how, and
than its strength. In AASC WG, both rheological and mechanical
what? Mater. Struct. 47 (2014) 11e25.
behaviour improved at longer mixing times. Mixing protocol 2 had [10] A.F. Abdalqader, F. Jin, A. Al-Tabbaa, Development of greener alkali-activated
an adverse effect on OPCC and AASC N rheology, while improving cement: utilisation of sodium carbonate for activating slag and fly ash mix-
the performance of the hardened materials only slightly. tures, J. Clean. Prod. 113 (2016) 66e75.
[11] A.S. Brykov, V.I. Korneev, Production and usage of powdered alkali metal
silicate hydrates, Metallurgist 52 (2009) 648e652.
[12] F. Puertas, M. Torres-Carrasco, Use of glass waste as an activator in the
preparation of alkali-activated slag. Mechanical strength and paste charac-
4. Conclusions terisation, Cem. Concr. Res. 57 (2014) 95e104.
[13] J.M. Mejía, R. Mejía de Gutie
rrez, F. Puertas, Rice husk ash as a source of silica
in alkali-activated fly ash and granulated blast furnace slag systems, Mater.
The main conclusions obtained from this study are: Construcc 63 (311) (2013) 361e375.
[14] M. Torres-Carrasco, F. Puertas, Waste glass in the geopolymer preparation.
 The slump and rheological behaviour of OPCC and AASC Mechanical and microstructural characterisation, J. Clean. Prod. 90 (2015)
397e408.
differed, particularly where AASC was activated with waterglass. [15] M.A. Villaquir
an-Caicedo, R. Mejía de Gutie
rrez, N.C. Gallego, A novel Mk-
It has been demonstrated that the nature of the alkaline acti- based geopolymer composite activated with rice husk ash and KOH: perfor-
vator is the key determinant in AAS concrete rheology. For first mance at high temperature, Mater. Construcc 67 (326) (2017) e117.
[16] M. Jiang, X. Chen, F. Rajabipour, A.M. Asce, C.T. Hendrickson, D.M. Asce,
time has been proved that the relationship between the
Comparative life cycle assessment of conventional, glass powder, and alkali-
empirical and rheological parameters in OPCC and AASCC fol- activated slag concrete and mortar, J. Infrastruct. Syst. 20 (2014).
lowed a similar pattern, which was likewise similar to the trend [17] A. Fern
andez-Jime
nez, J.G. Palomo, F. Puertas, Alkali-activated slag mortars
identified in the literature for OPC systems. Mechanical strength behaviour, Cem. Concr. Compos 29 (1999) 1313e1321.
[18] S. Aydın, B. Baradan, Effect of activator type and content on properties of
 Bingham models afforded a good fit to all the OPCC and AASC alkali-activated slag mortars, Compos. Part B Eng. 57 (2014) 166e172.
studied. [19] F. Puertas, R. Gutie
rrez, A. Fern
andez-Jime
nez, S. Delvasto, J. Maldonado,
 F. Puertas et al. / Cement and Concrete Composites 85 (2018) 22e31 31

Alkaline cement mortars. Chemical resistance to sulfate and seawater attack, on the stability and rheology of self-consolidating concrete containing min-
Mater. Construcc. 52 (267) (2002) 55e71. eral admixtures, Constr. Build. Mater 53 (2014) 102e109.
[20] M. Torres-Carrasco, M. Tognonvi, A. Tagnit-Hamou, F. Puertas, Durability of [49] O. Boukendakdji, S. Kenai, E.H. Kadri, F. Rouis, Effect of slag on the rheology of
alkali-activated slag concretes prepared using waste glass as alternative fresh self-compacted concrete, Constr. Build. Mater 23 (2009) 2593e2598.
activator, ACI 112 (2015) 791e800. [50] A.A.A. Hassan, M. Lachemi, K.M.A. Hossain, Effect of metakaolin on the
[21] M. Chi, Effects of dosage of alkali-activated solution and curing conditions on rheology of self-consolidating concrete, in: Des. Prod. Place. Self Consol. Concr.
the properties and durability of alkali-activated slag concrete, Constr. Build. RILEM State Art Reports, 2010, pp. 103e112.
Mater 35 (2012) 240e245. [51] A. Lange, T. Hirata, J. Plank, Influence of the HLB value of polycarboxylate
[22] A.M. Rashad, A comprehensive overview about the influence of different ad- superplasticizers on the flow behavior of mortar and concrete, Cem. Concr.
ditives on the properties of alkali-activated slag - a guide for Civil Engineer, Res. 60 (2014) 45e50.
Constr. Build. Mater 47 (2013) 29e55. [52] A. Mardani-Aghabaglou, M. Tuyan, G. Yilmaz, O. € Ario€ z, K. Ramyar, Effect of
[23] C. Duran Atiş, C. Bilim, O.€ Çelik, O. Karahan, Influence of activator on the different types of superplasticizer on fresh, rheological and strength proper-
strength and drying shrinkage of alkali-activated slag mortar, Constr. Build. ties of self-consolidating concrete, Constr. Build. Mater 47 (2013) 1020e1025.
Mater 23 (2009) 548e555. [53] R. Flatt, I. Schober, Superplasticizers and the rheology of concreteR. Flatt, I.
[24] M. Palacios, F. Puertas, Effect of shrinkage-reducing admixtures on the Schober//, in: Underst. Rheol. Concr, 2012, pp. 144e208.
properties of alkali-activated slag mortars and pastes, Cem. Concr. Res. 37 [54]
nez, E.F. Ledesma, Rheological
F. Cartuxo, J. De Brito, L. Evangelista, J.R. Jime
(2007) 691e702. behaviour of concrete made with fine recycled concrete aggregates - influence
[25] H. Ye, A. Radllinska, Shrinkage mechanisms of alkali-activated slag, Cem. of the superplasticizer, Constr. Build. Mater 89 (2015) 36e47.
Concr. Res. 88 (2016) 126e135. [55] L. Evangelista, J. De Brito, Concrete with fine recycled aggregates: a review,
[26] M. Palacios, F. Puertas, Effectivenss of mixing time on hardened properties of Eur. J. Environ. Civ. Eng. 18 (2013) 129e172.
waterglass-activated slag pastes and mortars, ACI Mater. 108 (2011) 73e78 [56] D. Carro-Lo
pez, B. Gonza
lez-Fonteboa, J. De Brito, F. Martínez-Abella,
n . I. Gonza
lez-Taboada, P. Silva, Study of the rheology of self-compacting con-
[27] A. Fernandez-Jimenez, F. Puertas, Setting of alkali-activated slag cement. In- crete with fine recycled concrete aggregates, Constr. Build. Mater 96 (2015)
fluence of activator nature, Adv. Cem. Res. 13 (2001) 115e121. 491e501.
[28] M. Palacios, P. Banfill, F. Puertas, Rheology and setting of alkali-activated slag [57] R. Cepuritis, S. Jacobsen, B. Pedersen, E. Mørtsell, Crushed sand in concrete -
pastes and mortars: effect of organic admixture, ACI Mater. J. 105 (2008) effect of particle shape in different fractions and filler properties on rheology,
140e148. Cem. Concr. Compos 71 (2016) 26e41.
[29] C. Shi, R.L. Day, A calorimetric study of early hydration of alkali-slag cements, [58] F. Collins, J.G. Sanjayan, Effects of ultra-fine materials on workability and
Cem. Concr. Res. 25 (1995) 1333e1346. strength of concrete containing alkali-activated slag as the binder, Cem. Concr.
[30] M. Palacios, Empleo de aditivos org
anicos en la mejora de las propiedades de Res. 29 (1999) 459e462.
cementos y morteros de escoria activada alcalinamente, PhD Tesis, UAM [59] F.G. Collins, J.G. Sanjayan, Workability and mechanical properties of alkali
Madrid, Spain, 2006. activated slag concrete, Cem. Concr. Res. 29 (1999) 455e458.
[31] F. Puertas, C. Varga, M.M. Alonso, Rheology of alkali-activated slag pastes. [60] M.M. Alonso, S. Gismera, M.T. Blanco, M. Lanzo
n, F. Puertas, in: Alkali-acti-
Effect of the nature and concentration of the activating solution, Cem. Concr. vated Mortars: Workability and Rheological Behaviour, vol. 145, 2017,
Compos 53 (2014) 279e288. pp. 576e587.
[32] M. Palacios, F. Puertas, Effect of superplastisizer and shrinkage reducing ad- [61] O.H. Wallevik, D. Feys, J.E. Wallevik, K. Khayat, Avoiding inaccurate in-
mixtures on alkali-activated slag pastes and mortars, Cem. Concr. Res. 35 terpretations of rheological measurements for cement-based materials, Cem.
(2005) 1358e1367. Concr. Res. 78 (2015) 100e109.
[33] A. Kashani, J.L. Provis, G.G. Qiao, J.S.J. Van Deventer, The interrelationship [62] D. Feys, J.E. Wallevik, A. Yahia, K.H. Khayat, O.H. Wallevik, Extension of the
between surface chemistry and rheology in alkali activated slag paste, Constr. ReinereRiwlin equation to determine modified Bingham parameters
Build. Mater 65 (2014) 583e591. measured in coaxial cylinders rheometers, Mat. Struct. 46 (2013) 289e311.
[34] K.-H. Yang, J.-K. Song, A.F. Ashour, E.-T. Lee, Properties of cementless mortars [63] D. Lowke, P. Schiessl, Effect of mixing energy on fresh properties of SCC
activated by sodium silicate, Constr. Build. Mater 22 (2008) 1981e1989. SCC2005, in: proc 4th Int RILEM Symp Self-Compacting Concr Chicago, 2005,
[35] C. Varga, Cementos activados alcalinamente: Comportamiento reolo
gico y p. 2.
durable en medio
acido, PhD Tesis, UNED Madrid, Spain, 2015. [64] N. Roussel, C. Stefani, R. Leroy, From mini-cone test to Abrams cone test:
[36] K. Kovler, N. Roussel, Properties of fresh and hardened concrete, Cem. Concr. measurement of cement-based materials yield stress using slump tests, Cem.
Res. 41 (2011) 775e792. Concr. Res. 35 (5) (2005) 817e822.
[37] S. Amziane, C.F. Ferraris, E.P. Koehler, Measurement of workability of fresh [65] N. Roussel, Correlation between yield stress and slump: comparison between
concrete using a mixing truck sofiane amziane, chiara F. Ferraris, eric P. numerical simulations and concrete rheometers results, Mater. Struct. 39
Koehler, J. Res. Natl. Inst. Stand. Technol. 110 (2005) 55e66. (2006) 501e509.
[38] E.P. Koehler, D.W. Fowler, Summary of Concrete Workability Test Methods, [66] J. Assaad, Formwork Pressure of Self Consolodating Concrete - Influence of
2003, p. 92. Thixotropy, PhD Dissertation, University of Sherbrooke, Canada, 2004.
[39] O.H. Wallevik, J.E. Wallevik, Rheology as a tool in concrete science: the use of [67] I. Gonz
alez Taboada, Self Compacting Recycled Concrete: Basic Mechanical
rheographs and workability boxes, Cem. Concr. Res. 41 (2011) 1279e1288. Properties, Rheology, Robutness and Thixotropy, PhD Dissertation, University
[40] G.H. Tattersall, P. Banfill, The Rheology of Fresh Concrete, Pitman Books of A Corun ~ a, Spain, 2017.
Limited, Great Britain, 1983. [68] M.K. Rahman, M.H. Baluch, M.A. Malik, Thixotropic behavior of self com-
[41] M. Greim, Equipos y sistemas para el control de calidad de hormigones, pacting concrete with different mineral admixtures, Constrcc. Buid. Mat. 50
morteros y
aridos, Cem. Y Hormigo
n 99 (2009) 34e38. (2014) 710e717.
[42] P. Estell‘e, C. Lanos, High torque vane rheometer for concrete: principle and [69] S.D. Wang, X.C. Pu, K.L. Scrivener, P.L. Pratt, Alkali-activated slag cement and
validation from rheological measurements, Appl. Rheol. 22 (2012) 12881. concrete: a review of properties and problems, Adv. Cem. Res. 7 (1995)
[43] F. Mahmoodzadeh, S.E. Chidiac, Rheological models for predicting plastic 93e102.
viscosity and yield stress of fresh concrete, Cem. Concr. Res. 49 (2013) 1e9. [70] S. Wang, K.L. Scrivener, Hydration products of alkali activated slag cement,
[44] N. Roussel, A thixotropy model for fresh fluid concretes: theory, validation Cem. Concr. Res. 25 (1995) 561e571.
and applications, Cem. Concr. Res. 36 (2006) 1797e1806. [71] A. Fern
andez-Jime
nez, F. Puertas, Structure of calcium silicate hydrates
[45] N. Roussel, G. Ovarlez, S. Garrault, C. Brumaud, The origins of thixotropy of formed in alkaline-activated Slag: influence of the type of alkaline activator,
fresh cement pastes, Cem. Concr. Res. 42 (2012) 148e157. J. Am. Ceram. Soc. 86 (2003) 1389e1394.
[46] Y. Tan, G. Cao, H. Zhang, J. Wang, R. Deng, X. Xiao, B. Wu, Study on the [72] F. Puertas, M. Palacios, H. Manzano, J.S. Dolado, a. Rico, J. Rodríguez, A model
thixotropy of the fresh concrete using DEM, Procedia Eng. 102 (2015) for the C-A-S-H gel formed in alkali-activated slag cements, J. Eur. Ceram. Soc.
1944e1950. 31 (2011) 2043e2056.
[47] J.E. Wallevik, Thixotropic investigation on cement paste: experimental and [73] R.J. Myers, S. a Bernal, R. San Nicolas, J.L. Provis, Generalized structural
numerical approach, J. Nonnewt. Fluid Mech. 132 (2005) 86e99. description of calcium-sodium aluminosilicate hydrate gels: the cross-linked
[48] P. Ghoddousi, A.A. Shirzadi Javid, J. Sobhani, Effects of particle packing density substituted tobermorite model, Langmuir 29 (2013) 5294e5306.