Cement & Concrete Composites 55 (2015) 374382

Contents lists available at ScienceDirect

Cement & Concrete Composites

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

Hydration of quaternary Portland cement blends containing

blast-furnace slag, siliceous y ash and limestone powder
Axel Schler a,, Barbara Lothenbach a, Frank Winnefeld a, Maciej Zajac b
a
Empa, Swiss Federal Laboratories for Materials Science and Technology, Laboratory for Concrete and Construction Chemistry, berlandstrasse 129, 8600 Dbendorf, Switzerland
b
HeidelbergCement Technology Center GmbH, Rohrbacher Strasse 95, 69181 Leimen, Germany

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

Article history: In this study the hydration of quaternary Portland cements containing blast-furnace slag, type V y ash
Received 5 December 2013 and limestone and the relationship between the types and contents of supplementary cementitious
Received in revised form 4 September 2014 materials and the hydrate assemblage were investigated at ages of up to 182 days using X-ray diffraction
Accepted 1 October 2014
and thermogravimetric analysis. In addition thermodynamic modeling was used to calculate the total
Available online 12 October 2014
volume of hydrates. Two blast-furnace slag contents of 20 and 30 wt.% were studied in blends containing
y ash and/or limestone at a cement replacement of 50 wt.%. In all cases the experiments showed the
Keywords:
presence of CSH, portlandite and ettringite. In samples without limestone, monosulfate was formed;
Quaternary Portland cement
Blast-furnace slag
in the presence of limestone monocarbonate was present instead. The addition of 5 wt.% of limestone
Fly ash resulted in a higher compressive strength after 28 days than observed for cements with lower or higher
Thermodynamic modeling limestone content. Overall the presence of y ash exerts little inuence on the hydrate assemblage. The
Limestone strength development reveals that amounts of up to 30 wt.% y ash can be used in quaternary cements
Hydrates without signicant loss in compressive strength.
2014 Elsevier Ltd. All rights reserved.

1. Introduction 3C4 ASH12 2CC H18 ! C6 AS3 H32 2C4 ACH11 : 1

The production of cement clinker causes 58% of the global man
made CO2 emissions [1,2]. The reduction of these is an important
aspect of cement research in recent years, especially as the demand
for cement increases [2]. This has led to numerous studies on
materials suitable to partially replace the clinker in blended
cements, so-called supplementary cementitious materials (SCMs).
Possible and well investigated candidates for cement substitution
include silica-rich materials like blast-furnace slag (BFS), y ash
(FA) and limestone (LS) [38]. Binary systems including type V
y ash or blast-furnace slag are well studied and both materials
are known to show synergistic effects with limestone in ternary
systems [911].
The addition of low amounts of CaCO3 to Portland cements
results in the formation of mono- and/or hemicarbonate instead
of monosulfate at longer reaction times [12,13]. The formation of
monocarbonate prevents the destabilization of ettringite to
monosulfate as monocarbonate C4 ACH11 and ettringite
(C6 AS3 H32 ) are more stable than monosulfate C4 ASH12 and calcite
CC [12,1416]:
Corresponding author. Tel.: +41 58 765 4917; fax: +41 58 765 6935.
E-mail address: axel.schoeler@empa.ch (A. Schler).
(Cement notation with A = Al2O3, C = CaO, C CO2 , H = H2O,
S = SiO2, S SO3 :)
The stabilization of ettringite and monocarbonate in the pres-
ence of CaCO3 increases the total volume of the solid phases
[8,17], which has been shown to result in higher compressive
strength in OPC with low additions of limestone. The amount of
ettringite and monocarbonate formed in the hydrated system
depends on the quantity of Al2O3 and SO3 in the system as
described in detail in [18]. However, not only the total amount of
Al2O3 in the system, but also the reaction degree of the aluminum
containing phases (C3A, C4AF, SCMs) is an important factor [17].
The amount and neness of calcite plays a minor role [13].
Similar effects have been observed for blends of OPC with y
ash or blast-furnace slags [9,10,19,20]. The presence of y ash,
which contains generally more Al2O3 than OPC, leads to the forma-
tion of more AFt and AFm phases, such that the effect can be more
pronounced. Investigations in OPC-LS and OPC-FA-LS cements have
shown that this volume increase correlates strongly with the
development of compressive strength and an increase in compres-
sive strength has been observed up to 10 wt.% limestone [9,1921].
The contribution of Si-rich y ashes to the formation of AFt and
AFm phases is limited due its low reaction degree. It has been
reported that after 1 year only 2530 wt.% of the y ash present

http://dx.doi.org/10.1016/j.cemconcomp.2014.10.001
0958-9465/ 2014 Elsevier Ltd. All rights reserved.
 A. Schler et al. / Cement & Concrete Composites 55 (2015) 374382 375

in Portland cement blends has reacted [2224]. Besides these Table 2

ndings, efforts to reduce the carbon footprint of cements are also Mineralogical composition of the used raw materials determined by XRD Rietveld
(wt.%).a
reected in the work of CEN/TC 51 WG 6 Ad-Hoc-Group CEM X.
This group is concerned with the investigation and the possible OPC BFS FA LS
standardization of ternary cement compositions in EN 197-1 C3S 62.9 Mullite 19.2
[2527]. However, only ternary cements containing clinker, a0 C2S 5.5 Free lime 0.2
limestone and either granulated blast-furnace slag, y ash or b C2S 3.2 Maghemite 0.9
C3A cubic 1.4 Quartz 0.2 10.2 0.5
pozzolans are considered. C3A othorh. 4.2 Calcite 0.4 0.4 99.1
While ternary composite cements are well investigated there is C4AF 12.7 Anhydrite 0.4
a lack of knowledge regarding quaternary composite cements. The Calcite 1.5 Akermanite 0.2
latter include three SCMs besides OPC with blast-furnace slag, Arcanite 0.7 Merwinite 0.4
Periclase 0.2 Gypsum 0.3
limestone and Si-rich y ash (type V according to EN 197-1) being
Portlanditeb 1.7 Illite 0.4
the most commonly used materials. In terms of phase composition Bassanite 3.2 Amorphous 98.5 68.8
quaternary systems have similar composition to already existing Gypsum 1.2
composite cements based on OPC, including blast-furnace slag or Syngenite 1.4
type V y ash [28]. Such systems might therefore be promising Quartz 0.2

cement compositions. Total 100.0 100.0 100.1 100.0

This study aims to investigate the interactions between Port- a
The Uncertainties are in the range of 4.16.5% for C3S, 2.85.5% for C2S, 0.92.5%
land cement, blast-furnace slag, type V y ash and limestone pow- for C3A, 1.32.4% for C4AF, 1.01.6% for gypsum and 1.53.8% for calcite as inves-
der in quaternary cements with high clinker replacement level of tigated by round robin tests [49].
b
50 wt.%. Thermodynamic modeling is applied to gain information Free lime in OPC = 1.65, detected according to [50].

on the formed hydrates and on the volume of those as a function

of blast-furnace slag/y ash ratio. In addition thermogravimetric
analysis (TGA) and X-ray diffraction (XRD) is carried out to identify 100
the hydrates formed and their changes over time. cumulative volume [vol.%] OPC

80
2. Experimental
limestone
60
2.1. Materials and mix design
FA
The cement used in this study was an ordinary Portland cement 40
(CEM I 52.5 R) (OPC) with a calcite content of 1.5 wt.%. Set regula-
tors were bassanite and gypsum with shares of 3.2 and 1.2 wt.%. As 20 BFS
cement substitutes three SCMs have been studied: a blast-furnace
slag (BFS), a siliceous y ash (type V according to EN 197-1) (FA), 0
and a limestone powder with a calcite content of >99% (LS). The
chemical and mineralogical composition of these materials deter-
relative frequency [vol.%]

OPC BFS
8
mined by means of X-ray uorescence (XRF) and quantitative
X-ray diffraction are given in Tables 1 and 2. The blast-furnace slag, FA
6

Table 1
Chemical compositiona and physical properties of the used raw materials.
4

OPC BFS FA LS
2
SiO2 19.86 41.25 56.77 2.35 limestone
Al2O3 4.64 7.57 26.03 0.78
TiO2 0.26 0.3 1.37 0.04 0
MnO 0.1 0.71 0.03 0.01 0.1 1 10 100 1000
Fe2O3 3.52 0.38 5.07 0.28 particle diameter [ m]
CaO 63.67 36.43 3.19 53.18
MgO 1.11 9.67 1.34 0.33 Fig. 1. Particle size distribution of the raw materials measured by laser diffraction.
K2O 0.63 0.62 2.53 0.18 The relative error is 1 vol.%.
Na2O 0.14 0.37 0.31 n.d.
SO3 2.89 1.51 0.06 n.d.
P2O5 0.14 0.02 0.4 0.02
LOIb 2.13 0.62e 2.12 42.33 the y ash and the limestone were ground separately in a labora-
Total 99.09 98.83 99.22 99.50
tory ball mill, while the OPC was used as received. Fig. 1 shows
CO2c 0.7 0.0 1.16 42.2 the particle size distribution (PSD) of the raw materials measured
Ctotald n.d. n.d. 1.28 n.d. by laser diffraction with a Malvern Mastersizer. The overall PSD of
K2Owater soluble 0.43 0.01 0.04 n.d. the OPC, the BFS and the FA do not differ signicantly from each
Na2Owater soluble 0.06 <0.01 0.03 n.d.
other while the limestone shows a distinct higher fraction of parti-
Blaine surface (cm2/g) 5180 4200 3360 4650
Density (g/cm3) 3.13 2.91 2.22 2.72 cles with a diameter larger than 10 lm.
d50 (lm) 11 14 10 16 Blast-furnace slag and y ash both consist of amorphous and
a
crystalline compounds, which can be determined and quantied
Determined by XRF analysis (wt.%).
b
950 C (OPC, BFS), 1050 C (FA, LS).
by XRD. However, no information on the composition of amor-
c
Calculated from TGA. phous phases can be derived by XRD. Thus, the average composi-
d
C content determined by combustion analysis. TOC = 1.22 wt.%. tion of the amorphous content of the blast-furnace slag and the
e
Gain of ignition due to oxidation of S2. y ash was calculated by subtracting the readily soluble alkalis
376 A. Schler et al. / Cement & Concrete Composites 55 (2015) 374382

and the fraction of crystalline phases from the total XRF composi- Table 4
tion and normalizing the results to 100 wt.% (Table 3). Due to the Compositions of the investigated quaternary cements.

low crystalline content of the blast-furnace slag, the composition BFS-FA-L Composition (data in wt.%)
of the amorphous content differs only slightly from the total com- OPC BFS FA LS
position. The amorphous content of the y ash shows clear differ-
20-30-0 50 20 30 0
ences to the total composition, mainly in the SiO2 and Al2O3 20-25-5 50 20 25 5
content. This is due to the presence of mullite (19.2 wt.%) and 20-20-10 50 20 20 10
quartz (10.2 wt.%). 20-15-15 50 20 15 15
Table 4 summarizes the compositions of the investigated qua- 20-10-20 50 20 10 20

ternary cements. The cement substitution level is always 50 wt.%, 30-20-0 50 30 20 0

while the SO3 level of all systems is adjusted with anhydrite to 30-15-5 50 30 15 5
30-10-10 50 30 10 10
3 wt.%. Two BFS levels of 20 and 30 wt.% were investigated. At both
30-5-15 50 30 5 15
of these levels FA is gradually replaced by limestone powder in 30-0-20 50 30 0 20
steps of 5 wt.% up to a limestone content of 20 wt.%.

2.2. Thermodynamic modeling

Thermodynamic modeling was carried out to investigate the
changes in volume of the solid phases as well as in the hydrates
assemblage, both as a function of added blast-furnace slag and
y ash. The geochemical modeling program GEMS [29,30] was
used with thermodynamic data from the PSI-GEMS database
[31,32] supplemented by cement specic data [33,34]. CSH is
described by a solid solution between Ca-rich (C1.67S1.0H2.1) and
Si-rich CSH (C0.83S1.0H1.33) end-members as described in detail
in [34,35]. The use of a solid solution model allows the description
of varying Ca/Si ratios in the CSH. The activities of the different
species were calculated from the measured total concentrations at
the respective temperatures using GEMS. The activity coefcients
of aqueous species yi were computed with the built-in expanded
extended DebyeHckel equation in TruesdellJones form with
common parameter ai of 3.67 for KOH solutions:
p
Ay z2i I
log ci p by I;
1 By ai I
where zi denotes the charge of species i, I the effective molal ionic
strength, by is a semi-empirical parameter (b  0.123 for KOH at
25 C), and Ay and By are P, T-dependent coefcients. This activity
correction is thought to be applicable up to approx. 1 M ionic
strength [36]. Data on the density and the molar volume of phases
considered in the thermodynamic calculations are given in Table 5.
For the thermodynamic calculations different long-term
degrees of reaction were assumed to calculate the phase assem-
blage after longer hydration times. The used degrees of reaction
are 100 wt.% for OPC, 70 wt.% for the blast-furnace slag and
30 wt.% for the y ash. The values for the SCMs are based on inves-
tigations of blast-furnace slag and y ash blended Portland
cements after one year of hydration [22,24,37,38]. As the crystal-
line phases in both SCMs are either considered to be non-reactive
Table 3
Glass composition of blast-furnace slag and y ash (wt.%).
Values are estimated by subtracting the water soluble alkalis
and the fraction of crystalline phases from the total XRF
composition.
(e.g. quartz or mullite) or are only present in very low amounts,
only the amorphous content was assumed to react. The XRD data
indicate no signicant reaction of the quartz and mullite present
in y ash up to 91 days of hydration. This in turn means that the
before mentioned overall long term degrees of reaction have to
be based on the amorphous part of the blast-furnace slag and the
y ash which leads to elevated values of 71.1 wt.% and 43.6 wt.%,
respectively. For the calculation of CSH a solid solution of Ca-
rich CSH (C1.67S1.0H2.1) and Si-rich CSH (C0.83S1.0H1.33) with
varying C/S ratio has been implemented as described in detail in
[34,39]. The oxide composition as given in Table 3 was used as
input for the equilibrium calculations which give the mass, the vol-
ume and the composition of the modeled systems as output.
2.3. Methods of analysis
2.3.1. Investigation of cement pastes
To study the phase development, cement paste samples were
2 prepared from 200 g of dry material with water to binder ratio of
0.45 using a Twister Evolution vacuum mixer from Renfert. The
samples were mixed for 2 min at 300 rpm and stored at 20 C in
60 ml sealed plastic vessels for TGA and XRD studies.
For TGA, slices of approximately 3 mm where cut from the
hardened cement paste and gently crushed with a mortar and pes-
tle. To stop the hydration the resulting powder was immersed in
isopropanol for 15 min and afterwards ltrated with a Buchner
funnel before being ushed with diethylether in order to remove
all remaining isopropanol. Drying of the lter cake in a compart-
ment dryer at 40 C for 8 min assured the elimination of remaining
diethylether. The analysis was carried out directly afterwards on
approximately 50 mg of the resulting powder using a Mettler
Toledo TGA/SDTA851e by monitoring the weight while heating
up from 30 to 980 C at a heating rate of 20 C/min under nitrogen
atmosphere.
The corresponding changes in bound water and portlandite as
given in Table 7 are calculated as percentage of the dry sample
weight at 500 C according to
mtotal  mdry solids
BW  100 %; 3
mdry solids

BFS FA and
SiO2 42.7 62.2 mH2 O CH MCaOH2 mH2 O CH 74
Al2O3 7.9 18.5 CH   100 %   100 %;
mdry solids MH2 O mdry solids 18
Fe2O3 0.4 6.3
CaO 37.5 3.9 4
MgO 10.0 2.0
K2O 0.6 3.8 with mtotal = total mass of wet paste, mdry solids = mass of dry paste
Na2O 0.4 0.5 corresponding to sample weight at 500 C, mH2 O CH = weight loss due
TiO2 0.3 2.1 to dehydroxylation of portlandite obtained by the tangent method,
P2O5 0.6
M(Ca(OH)2) = molar mass of portlandite and M(H2O) = molar mass
Si/Al 5.4 3.4
of H2O. For XRD experiments, slices where immersed in isopropanol
 A. Schler et al. / Cement & Concrete Composites 55 (2015) 374382 377

Table 5 were placed in a PANalytical XPert Pro MPD diffractometer in a

Density and molar volume of hydrates used in the thermodynamic calculations [34]. h  2h conguration with an incident beam monochromator and
Mineral Formula Density Molar volume CuKa radiation (k = 1.54 ). The samples were scanned for
(g/cm3) (cm3/mol) 120 min between 5 and 75 with an XCelerator detector. For the
Monosulfate Ca4Al2(SO4)(OH)126H2O 2.01 309 evaluation of the gained data in terms of phase identication the
Hemicarbonate Ca4Al2(CO3)0.5(OH)135.5H2O 1.98 285 XPert High Score Plus program version 3.0e from PANalytical was
Monocarbonate Ca4Al2(CO3)(OH)125H2O 2.17 262 used.
Ettringite Ca6Fe2(SO4)3(OH)1226H2O 1.77 707
Portlandite Ca(OH)2 2.24 33
Si-rich CSH (CaO)0.8333(SiO2)(H2O)1.3333 2.50 59 2.3.2. Compressive strength of mortar samples
Ca-rich CSH (CaO)1.6667(SiO2)(H2O)2.1 2.67 78 Compressive strength of mortar prisms was determined after 2,
Hydrotalcite Mg4Al2(OH)143H2O 2.01 220 7 and 28 days according to EN 196-1.
FH FeO(OH) 2.61 34
Calcite CaCO3 2.71 37
3. Results and discussion

3.1. Thermodynamic modeling

Table 6
Compositions of AFm and AFt phases as used in the thermodynamic calculations.
Thermodynamic modeling was applied to investigate the inu-
Mineral Reaction ence of limestone on the hydrate assemblage (Fig. 2). To illustrate
Ettringite Ca6Al2(SO4)3(OH)1226H2O ? 6Ca2++2Al(OH) 4 + 3SO4
2
the effect of limestone a CO2-free OPC containing 30 wt.% y ash
+ 4OH + 26H2O and 20 wt.% blast-furnace slag was chosen as starting composition.
Monosulfate Ca6Al2(SO4)(OH)126H2O ? 4Ca2++2Al(OH) 4 + SO4
2
In the absence of CaCO3, CSH and monosulfate are calculated to
+ 4OH + 6H2O
Hemicarbonate Ca4Al2(CO3)0.5(OH)135.5H2O ? 4Ca2++2Al(OH) 4
be the main hydrates in a ternary system with 20 wt.% blast-fur-
+ 0.5CO2 
3 + 5OH + 5.5H2O nace slag.
Monocarbonate Ca4Al2(CO3)(OH)125H2O ? 4Ca2++2Al(OH)-4 + CO23 The partial replacement of y ash by CaCO3 is calculated to lead
+ 4OH + 5H2O
to the consumption of portlandite as the SiO2 present in the y ash
will react with Ca(OH)2 according to Eq. (5):
yCaOH2 SiO2 xH2 O ! yCaO  SiO2  y xH2 O: 5
Table 7
Amount of bound water, CSH and portlandite at 20 and 30 wt.% blast- The complete replacement of y ash by CaCO3 leads to the pres-
furnace slag and 30 and 20 wt.% FA, respectively. Calculated by thermody-
ence of more portlandite as no pozzolanic reaction occurs.
namic modeling; referred to weight of dry hydrates.
A replacement of y ash by CaCO3 leads to the formation of
BFS content (wt.%) 30 20 hemicarbonate, monocarbonate and more ettringite instead of
Bound water (wt.%) 26.2 25.0 monosulfate according to Eqs. (6) and (7) (additional information
CSH (wt.%) 66.7 62.1 on the stoichiometry of AFm amd AFt phases are presented in
CH (wt.%) 1.0 0.0 Table 6):

3C4 ASH12 CH CC 19H ! C6 AS3 H32 2C4 AC0:5 H12

for 20 min followed by drying at 40 C for 15 min. As immersion in 3  309 cm3 33 cm3 37 cm3 ! 707 cm3 2  285 cm3 ;
isopropanol is known to affect hydrate phases [40] this procedure DV 28 vol:%
was kept as short as possible to avoid loss of information. The slices 6

fly ash [wt%]

30 25 20 15 10 5 0
monosulfate
60 hemicarbonate

50 monocarbonate

ettringite
volume [cm 3/100g]

40
Ca-rich C-S-H portlandite 1.8
(jennite)
30 1.6
C/SC-S-H

1.4
Si-rich C-S-H
20 1.2
(tobermorite)
1.0
inert phases
10 calcite
FA - amorphous
BFS - amorphous
hydrotalcite FH3
0
0 5 10 15 20 25 30
CaCO3 [wt%]

Fig. 2. Hydrates as predicted by thermodynamic modeling as a function of CaCO3/y ash content at 20 wt.% blast-furnace slag. The dash-dot lines represent the compositions
of the systems under investigation. Volume expressed as cm3/100 g unhydrated binder.
378 A. Schler et al. / Cement & Concrete Composites 55 (2015) 374382

blast-furnace slag [wt%]

50 40 30 20 10 0

max. volume of solids

60 at 50 wt.% BFS

50 monocarbonate

volume [cm3/100g]
ettringite
40
portlandite 1.8

1.6
30

C/S C-S-H
1.4
Ca-rich C-S-H Si-rich C-S-H
(jennite) (tobermorite) 1.2
20
1.0
calcite inert phases
10
BFS - amorphous FA - amorphous
hydrotalcite FH3
0
0 10 20 30 40 50
fly ash [wt%]

Fig. 3. Changes in the total volume of hydrates as an effect of different additions of blast-furnace slag and y ash in a quaternary OPC containing 5 wt.% CaCO3. The dash-dot
lines represent the two blast-furnace slag levels that were investigated in the present study. Volume expressed as cm3/100 g unhydrated binder.

3C4 ASH12 2CC 18H ! C6 AS3 H32 2C4 ACH11 see also Eq:1
CH
3  309 cm3 2  37 cm3 ! 707 cm3 2  262 cm3 ; DV 23 vol:%
E AFmss
7
intensity [arbitrary units]

Numbers in parentheses shows the molar volume of solids and HC

the volume gain due to the reaction with respect to the reactants. HT
MS
E
In the presence of more CaCO3, monocarbonate is stable according MC Mu E
to Eq. (7). From the diffractograms given in Fig. 4 it can be seen that
both, hemicarbonate and monocarbonate are present at 182 days
FA-LS
of hydration. In experimental studies, initially the formation of
30-0
hemicarbonate is observed as the formation of monocarbonate is
kinetically slow [13]. In the long term a transformation of hemicar-
bonate to monocarbonate is observed. These effects result in a 25-5
maximum volume of stable hydrates at about 3 wt.% CaCO3 con-
tent, similar to the effects observed for ternary OPC-BFS-LS and 15-15
OPC-FA-LS systems [10,17]. Investigations of OPC-limestone and
OPC-FA-LS cements have shown that this volume maximum corre- 8 9 10 11 12 13 14 15 16 17 18 19 20
lates with a maximum of compressive strength [9,1921]. This
position [2CuK]
positive effect is enhanced by increasing Al2O3 content as more
ettringite can be stabilized by CaCO3. Thus, the partial substitution Fig. 4. XRD diffractograms of samples containing 20 wt.% blast-furnace slag and
of OPC by blast-furnace slag and y ash should increase the posi- 30 wt.% y ash or mixtures of y ash and limestone at 182 days of hydration. AFmss
tive effect of the presence of CaCO3. However, the contribution of solid solution of hemicarbonate and OH substituted monosulfate, CH
portlandite, E ettringite, HC hemicarbonate, HT hydrotalcite, MC monocr-
blast-furnace slag and y ash to the amount of Al2O3 that is avail- bonate, MS monosulfate, Mu mullite.
able to form monocarbonate and to stabilize ettringite is limited as
both materials exhibit slower reaction rates than OPC. Further-
more, the used blast-furnace slag has a low Al2O3 content. After (i) The total volume of hydrated binder and the amount of CS
one year, the reaction of approximately 2530 wt.% of y ash and H is slightly reduced due to the lower degree of reaction of
70 wt.% of blast-furnace slag has been reported [2224,37,38]. the y ash compared to the blast-furnace slag.
Considering the only partial reaction of y ash and blast-furnace (ii) Decrease of the volume of portlandite caused by the pozzo-
slag, blast-furnace slag contributes approx. 6 wt.% and y ash lanic reaction with the SiO2 from the y ash.
approx. 8 wt.% of reactive Al2O3. Thus, their presence increases (iii) The higher Al2O3 and the lower MgO content of the y ash
the amount of available Al2O3 in the long term. lead to an increase in monocarbonate and a decrease in
Thermodynamic modeling was also used to study the inuence hydrotalcite.
of different fractions of y ash and blast-furnace slag on the
hydrate assemblage. In a ternary system containing 50 wt.% of If a complete reaction of the binder would have been assumed a
blast-furnace slag and 5 wt.% of CaCO3 as shown on the left hand slightly higher total volume of hydrates would be obtained for the
side of Fig. 3, the main hydrates are calcium silicate hydrate (C y ash rich blends, mainly due to the formation of additional CS
SH), ettringite, monocarbonate, hydrotalcite and portlandite. Suc- H. Furthermore the complete depletion of portlandite and the for-
cessive replacement of blast-furnace slag by y ash exerts different mation of strtlingite at high y ash contents would be calculated
effects: as has been shown elsewhere [8].
 A. Schler et al. / Cement & Concrete Composites 55 (2015) 374382 379

100 3.2. Hydrates present

weight loss [wt.%]

FA-LS
95 30-0
90 25-5
15-15
Results of TGA and XRD investigations at 182 days of hydration
85 for selected samples containing 20 wt.% blast-furnace slag at dif-
80 ferent y ash and limestone quantities are depicted in Figs. 4 and
75 5. In absence of limestone, the main phases detected by XRD and
70 TGA are calcium silicate hydrate (CSH), ettringite, AFm phases
0.00 (mainly hemicarbonate and a mixed AFm containing sulfate, car-
bonate and hydroxide [42]) and portlandite. The presence of lime-
DTG [wt.%/K]

-0.04 C-S-H
HT
-0.08
MC/HC
-0.12
-0.16 E 15-15
0 100 200 300
temperature [C]
Fig. 5. TGA curves for samples containing 20 wt.% blast-furnace slag at 182 days of
hydration. Cc calcite, CH portlandite, CSH calcium silicate hydrate, E
ettringite, HC hemicarbonate, HT hydrotalcite, MC monocarbonate.
Thermodynamic modeling predicts no or very little portlandite,
as a homogeneous system with CSH of C/S  1.5 is predicted. In
real systems generally the presence of signicant quantities of
portlandite is still observed while CSH with a more variable C/
S from 1.2 to 1.6 is present [17,41]. Due to inhomogeneities and
a disconnected pore network portlandite can persist in some parts
of the microstructure while in other areas a low C/S is present.
SEM/EDX investigations in blended Portland cements showed a
very inhomogeneous microstructure [17,40]; in some areas port-
landite persisted while in others the C/S of the CSH equaled to
1.2. Such inhomogeneous distribution of the C/S cannot be cap-
tured by thermodynamic modeling.
In this study quaternary blends with two different blast-furnace
slag levels were investigated as indicated in Fig. 3. The calculated
total volume of solids as shown in Fig. 2 decreases only very
slightly by 1 vol.% from 30 wt.% to 20 wt.% blast-furnace slag con-
tent. The corresponding changes in bound water, CSH and port-
landite are calculated using GEMS as given in Table 7.
stone leads to the formation of hemi- and monocarbonate instead
CH
Cc
of the monosulfate phases [12]. Furthermore slightly more ettring-
FA-LS ite is detected as can be seen from the TGA data. XRD data recorded
30-0
25-5 at different sample ages (not presented here) show that during the
rst weeks only hemicarbonate is observed, while hemi- and
400 500 600 700 800 900 monocarbonate are both present at later ages. However, from a
thermodynamic point of view only monocarbonate is stable. This
initial kinetic hindrance of the monocarbonate formation has pre-
viously been reported, e.g. by [12,13]. The formation of mono- and
hemicarbonate prevents the monosulfate formation, indirectly sta-
bilizes ettringite (Fig. 2) and is the cause for the increased ettring-
ite content as can be seen from Fig. 5. The slight increase of
portlandite in the presence of 5 and 15 wt.% limestone is due to
the simultaneous reduction of the y ash content whereby less
portlandite is consumed to form CSH (see Fig. 6).
XRD experiments on systems containing 30 wt.% blast-furnace
slag do not show differences to the data presented in Fig. 4. As
shown in Table 8 the bound water content increases up to 91 days
in all systems. After 91 and 182 days the total bound water content
is comparable to calculated values (as given in Table 7). The sam-
ples with higher blast-furnace slag content have slightly more
bound water in agreement with the modeling. This observation
underpins the faster reaction of the blast-furnace slag. The portlan-
dite content does not vary signicantly between the two blast-fur-
nace slag levels and does not indicate a strong inuence of the y
ash used on the extent of the pozzolanic reaction.
The portlandite contents are slightly lower in the samples con-
taining more y ash which is in agreement with the calculated
trends in Figs. 2 and 3. The absolute portlandite quantities deter-
mined by TGA decrease slowly with time but are considerably

wt.% fly ash wt.% fly ash

30 25 20 15 10 20 15 10 5 0
28 28
(a) 20 wt.% BFS (b) 30 wt.% BFS
26 26
portlandite / bound water [g/100g binder]

portlandite / bound water [g/100g binder]

24 24

22 22

20 20
bound water

18 18 28 days
bound water 91 days
16 28 days 16 182 days
91 days
182 days portlandite
14 portlandite 14

12 12

10 10

8 8
0 5 10 15 20 0 5 10 15 20
wt.% limestone wt.% limestone

Fig. 6. Content of bound water (excluding portlandite) and portlandite obtained from TGA for both blast-furnace slag levels of (a) 20 wt.% and (b) 30 wt.% at 28, 91 and
182 days of hydration. The relative error due to preparation and measurement is between 5% and 10%.
380 A. Schler et al. / Cement & Concrete Composites 55 (2015) 374382

Table 8
Bound water (excluding portlandite) and portlandite as calculated from TGA at 1 to 182 days of hydration. Data is given in (wt.%).

BFS-FA-LS Time (d) BW CH BFS-FA-LS Time (d) BW CH

20-10-20 1 10.4 9.3 30-0-20 1 10.4 9.4
2 14.9 11.6 2 15.0 11.7
7 18.0 11.6 7 18.1 12.6
28 18.1 12.0 28 21.6 12.8
91 22.1 12.2 91 24.7 13.5
182 20.9 12.0 182 24.5 13.2
20-15-15 1 10.8 9.1 30-5-15 1 10.5 9.1
2 14.4 10.7 2 14.7 11.1
7 17.3 11.0 7 18.5 12.6
28 18.2 12.0 28 21.8 12.6
91 22.2 11.6 91 22.8 12.8
182 21.6 11.5 182 23.7 12.5
20-20-10 1 10.6 8.7 30-10-10 1 10.7 9.2
2 14.5 11.1 2 14.5 11.3
7 18.0 11.6 7 18.4 12.1
28 18.7 11.6 28 22.0 12.2
91 22.9 11.2 91 25.0 12.4
182 22.0 11.7 182 25.0 12.0
20-25-5 1 10.5 8.9 30-15-5 1 10.5 9.3
2 14.3 10.9 2 14.7 11.0
7 18.2 11.5 7 19.3 12.3
28 19.0 11.4 28 21.8 11.9
91 22.8 11.0 91 25.1 12.2
182 21.7 10.6 182 23.1 11.5
20-30-0 1 10.2 8.6 30-20-0 1 10.5 8.6
2 14.8 10.8 2 14.6 10.9
7 18.0 11.4 7 18.1 11.6
28 18.6 11.5 28 20.8 11.8
91 23.7 11.9 91 21.6 11.7
182 21.8 11.0 182 23.4 11.2

60
60 (100-X)% OPC
2d 7d 28 d 50% OPC + (50-X)% BFS
50% OPC + (50-X)% FA
55 (100
20% slag 30% slag 58 -X )% 50% OPC + 20% BFS + (30-X)% FA
OPC 50% OPC + 30% BFS + (20-X)% FA
compressive strength [MPa]

50
volume [cm /100g]

45
56 50% O
PC + (5
0-X)%
3

40 BFS
50% OPC +
30% BFS +
(20-X)% FA
35
54
50% OPC +
30 20% BFS + (30-X)%
FA
25
52 50% OPC +
(50-X)% FA
20

15
10
30-0 25-5 20-10 15-15 10-20 20-0 15-5 10-10 5-15 0-20
fly ash-limestone [wt%]
Fig. 7. Compressive strength at 2, 7 and 28 days of hydration for systems
containing 20 wt.% and 30 wt.% blast-furnace slag.
higher than the modeled ones even after 182 days. This could be
related to a lower degree of y ash reaction than assumed or also
due to a inhomogeneous system with lower Ca/Si ratio of CSH,
as has been observed by [17,19].
3.3. Development of compressive strength
The development of compressive strength up to 28 days is
shown in Fig. 7. At 2 and 7 days all strength values are located in
the same range independent from the composition of the systems.
The highest values are reached with 5 wt.% additional limestone
while both, higher limestone content or limestone absence lead
to lower compressive strength. Thus, the presence of low
50
0 5 10 15 20
CaCO3 [wt.%]
Fig. 8. The effect of limestone in cement blends containing blast-furnace slag and/
or y ash calculated by thermodynamic modeling. Assumed degrees of reaction are
100% for OPC, 70% for blast-furnace slag and 30% for y ash. Volume expressed as
cm3/100 g unhydrated binder. The X represents the weight fraction of added CaCO3.
quantities of limestone increases compressive strength as observed
in OPC or OPC-FA blends [11,43]. CaCO3 is able to take part actively
in the hydration reaction up to 23 wt.% as shown in the thermo-
dynamic calculations presented in Fig. 2. This chemical reaction of
limestone has also been reported by [12,44]. At higher shares the
remaining calcite will only act passively as ller. The ller effect
itself is characterized by supplying higher specic surface available
for nucleation and a higher effective water/cement ratio [4446].
The amount of calcite that is able to react depends on the available
Al2O3 and thereby on the present SCMs and their composition.
The calculated effect of the blast-furnace slag and y ash
contents as a function of CaCO3 on the total volume of solids is
 A. Schler et al. / Cement & Concrete Composites 55 (2015) 374382
depicted in Fig. 8. All systems blended with blast-furnace slag and/
or y ash show a slightly lower volume of solids compared to OPC.
Cement substitution by blast-furnace slag leads to slightly higher
calculated volume of solids than substitution by y ash due to
the limited reactivity of the latter. When both materials are
incorporated the volume of solids increases with increasing
blast-furnace slag content. Blast-furnace slag can form signicantly
more CSH than y ash due to its faster reaction. The Al2O3
available for reaction from the y ash leads to more monocarbon-
ate but this cannot compensate for the lower CSH content. The
steeper decrease of the blast-furnace slag system that is visible
from about 2 wt.% CaCO3 originates from a decrease in blast-
furnace slag content and thereby a decrease in hydrotalcite as well.
The formation of hydrotalcite requires both Al2O3 and MgO. The
rst is provided by the y ash and the latter by blast-furnace slag.
Al2O3 is known to be preferably bound in hydrotalcite [47,48]. As
this phase remains constant in the y ash containing systems a less
steep decrease is visible that mainly results from a decrease in
CSH with decreasing blast-furnace slag content. A slightly higher
calculated total volume of solids with higher blast furnace slag
content goes along very well with the slightly higher compressive
strength observed for corresponding systems (Fig. 7).
4. Conclusions
The present study shows the effect of the binder composition in
quaternary OPC systems with 50 wt.% cement substitution by
blast-furnace slag, siliceous y ash and limestone.
(i) The relative fraction of blast-furnace slag, siliceous y ash or
limestone in OPC-blends with 50 wt.% substitution exerts no
signicant inuence on strength development at early ages
up to 7 days. At later ages, a CaCO3 content of 25 wt.% leads
to the stabilization of monocarbonate and ettringite, a max-
imum volume of hydrates (Figs. 2 and 8) and thus to a higher
compressive strength as shown in Fig. 7.
(ii) The ideal amount of additional limestone depends on the
composition and amounts of incorporated y ash and/or
blast-furnace slag as well as the total SO3. Further on the
amount of CaCO3 that is already incorporated in OPC has
to be taken into account.
(iii) The types of hydrates present in ternary or quaternary
blends with 50 wt.% of OPC are similar to OPC. However,
there are some differences in the volume of hydrates and
less portlandite is present in blended OPC while more C
SH and AFm phases can be observed.
(iv) The presence of both, y ash and blast-furnace slag, has little
effect on the hydrate assemblage. The replacement of some
blast-furnace slag by y ash leads to a slight decrease in
the amount of portlandite and CSH and to the formation
of more AFm phases (monocarbonate and hemicarbonate).
Portlandite is consumed in the pozzolanic reaction with
the y ash to form CSH. However, the limited assumed
degree of reaction as well as the low amorphous content of
the y ash itself leads to lower calculated CSH. The low
reactivity also causes the slightly reduced total volume of
solids and compressive strength at high y ash content.
Yet, these effects are small and consequently amounts of
up to 30 wt.% y ash can be used in quaternary blends with-
out signicant loss of compressive strength.
Acknowledgements
This study was nancially and analytically supported by Heidel-
bergCement, Leimen, Germany. The authors wish to acknowledge
381
H.-M. Ludwig and F. Bellmann (Bauhaus University Weimar) for
helpful and inspiring discussions and Luigi Brunetti and Boris
Ingold (Empa, Laboratory for Concrete and Construction Chemis-
try) for technical support.
References
[1] Damtoft JS, Lukasik J, Herfort D, Sorrentino D, Gartner EM. Sustainable
development and climate change initiatives. Cem Concr Res
2008;38(2):11527.
[2] Schneider M, Romer M, Tschudin M, Bolio H. Sustainable cement production at
present and future. Cem Concr Res 2011;41(7):64250.
[3] Tennis PD, Thomas MDA, Weiss WJ. State-of-the-art-report on use of limestone
in cement at levels of up to 15%. Skokie: Portland Cement Association; 2011.
[4] European construction in service of society ecoserve network, Cluster 2:
Production and Application of Blended Cements; 2005.
[5] Tsivilis S, Batis G, Chaniotakis E, Grigoriadis G, Theodossis D. Properties and
behavior of limestone cement concrete and mortar. Cem Concr Res
2000;30(10):167983.
[6] Wang X-Y, Lee H-S. Modelling the hydration of concrete incorporating y ash
or slag. Cem Concr Res 2010;40(7):98496.
[7] Vuk R, Tinta V, Grabrovek R, Kauci V. The effects of limestone addition, clinker
type and neness on properties of Portland cement. Cem Concr Res
2001;31(1):1359.
[8] Lothenbach B, Scrivener KL, Hooton RD. Supplementary cementitious
materials. Cem Concr Res 2011;41(12):124456.
[9] Carrasco MF, Menendez G, Bonavetti V, Irassar EF. Strength optimization of
tailor-made cement with limestone ller and blast-furnace slag. Cem Concr
Res 2005;35(7):132431.
[10] Menendez G, Bonavetti V, Irassar EF. Strength development of ternary blended
cement with limestone ller and blast-furnace slag. Cem Concr Compos
2003;25(1):617.
[11] De Weerdt K, Kjellsen KO, Sellevold E, Justnes H. Synergy between y ash and
limestone powder in ternary cements. Cem Concr Compos 2011;33(1):308.
[12] Lothenbach B, Le Saout G, Gallucci E, Scrivener KL. Inuence of limestone on
the hydration of Portland cements. Cem Concr Res 2008;38(6):84860.
[13] Zajac M, Rossberg A, Le Saot G, Lothenbach B. Inuence of limestone and
anhydrite on the hydration of Portland cements. Cem Concr Compos
2014;46:99108.
[14] Matschei T, Lothenbach B, Glasser FP. The AFm phase in Portland cement. Cem
Concr Res 2007;37:11830.
[15] Damidot D, Glasser FP. Thermodynamic investigation of the CaOAl2O3
CaSO4H2O system at 50 C and 85 C. Cem Concr Res 1992;22(6):117991.
[16] Damidot D, Stronach S, Kindness A, Atkins M, Glasser FP. Thermodynamic
investigation of the CaOAl2O3CaCO3H2O closed system at 25 C and
inuence of Na2O. Cem Concr Res 1994;24:56372.
[17] De Weerdt K, Ben Haha M, Le Saot G, Kjellsen KO, Justnes H, Lothenbach B.
Hydration mechanisms of ternary Portland cements containing limestone
powder and y ash. Cem Concr Res 2011;41(3):27991.
[18] Matschei T, Lothenbach B, Glasser FP. The role of calcium carbonate in cement
hydration. Cem Concr Res 2007;37:5518.
[19] De Weerdt K, Justnes H, Lothenbach B, Ben Haha M. The effect of limestone
powder additions on strength and microstructure of y ash blended cements.
In: 13th international conference on the chemistry of cement. Madrid; 2011.
[20] Moesgaard M, Herfort D, Steenberg M, Frank Kirkegaard L, Yue YZ. Physical
performances of blended cements containing calcium alumosilicate glass
powder and limestone. Cem Concr Res 2011;41(3):35964.
[21] Damidot D, Lothenbach B, Herfort D, Glasser FP. Thermodynamics and cement
science. Cem Concr Res 2011;41(7):67995.
[22] Ben Haha M, De Weerdt K, Lothenbach B. Quantication of the degree of
reaction of y ash. Cem Concr Res 2010;40(11):16209.
[23] De Weerdt K, Ben Haha M, Le Saot G, Kjellsen KO, Justnes H, Lothenbach B.
The effect of temperature on the hydration of composite cements containing
limestone powder and y ash. Mater Struct 2012;45(7):110114.
[24] Deschner F, Mnch B, Winnefeld F, Lothenbach B. Quantication of y ash in
hydrated, blended Portland cement pastes by backscattered electron imaging.
J Microsc 2013;251(2):188204.
[25] CEN/TC 51 Cement and building limes, N 1146 Presentation of WG 6,
Denition and terminology of cement, Progress Report, Athens, March 2012.
[26] Espion B, Lebon B, Pierre C, Germain O, Hellebois A. Characterisation of new
ternary cements with reduced clinker content. In: Sakai K, editor. First
international conference on concrete sustainability. Tokyo: Japan Concrete
Institute; 2013. p. 14552.
[27] Gteberwachung und Qualittssicherung von Zement. VDZ-Ttigkeitsbericht
20092012. <http://www.vdz-online.de/leadmin/gruppen/vdz/3Literatur
Recherche/TaeB09-12/TaeB_09-12_gesamt.pdf>.
[28] Iiguez Snchez CA, Gmez Zamorano LY, Lothenbach B. Microstructure and
mechanical properties of composite cements with ultra ne silica waste
additions. In: 13th International Conference on the Chemistry of Cement.
Madrid 2011. 7p.
[29] Wagner T, Kulik D, Hingerl FF, Dmytrievana SV. Gem-selektor geochemical
modelling package: TSolMod library and data interface for multicomponent
phase models. Can Mineral 2012;50(5):117395.
382 A. Schler et al. / Cement & Concrete Composites 55 (2015) 374382
[30] Kulik DA, Wagner T, Dmytrieva SV, Kosakowski G, Hingerl FF, Chudnenko KV,
et al. GEM-Selektor geochemical modeling package: revised algorithm and
GEMS3K numerical kernel for coupled simulation codes. Comput Geosci
2013;17(1):124.
[31] Thoenen T, Kulik D. Nagra/PSI chemical thermodynamic database 01/01 for
GEMS-selektor (V.2-PSI) geochemical modelling code. PSI, Villingen; 2003.
[32] Hummel W, Berner U, Curti E, Pearson FJ, Thoenen T. Nagra Technical Report
NTB 02-16. Wettingen, Switzerland; 2002.
[33] Matschei T, Lothenbach B, Glasser FP. Thermodynamic properties of Portland
cement hydrates in the system CaOAl2O3SiO2CaSO4CaCO3H2O. Cem
Concr Res 2007;37(10):1379410.
[34] Lothenbach B, Matschei T, Mschner G, Glasser FP. Thermodynamic modelling
of the effect of temperature on the hydration and porosity of Portland cement.
Cem Concr Res 2008;38(1):118.
[35] Kulik DA, Kersten M. Aqueous solubility diagrams for cementitious waste
stabilization systems: II, End-member stoichiometries of ideal calcium silicate
hydrate solid solutions. J Am Ceram Soc 2001;84:301726.
[36] Helgeson HC, Kirkham DH, Flowers GC. Theoretical prediction of the
thermodynamic behaviour of aqueous electrolytes at high pressures and
temperatures: IV. Calculation of activity coefcients, osmotic coefcients, and
apparent molal and standard and relative partial molal properties to 600 C
and 5 Kb. Am J Sci 1981;281:1249516.
[37] Kocaba V, Gallucci E, Scrivener KL. Methods for determination of degree of
reaction of slag in blended cement pastes. Cem Concr Res 2012;42(3):51125.
[38] Lothenbach B, Le Saout G, Ben Haha M, Figi R, Wieland E. Hydration of a low-
alkali CEM III/BSiO2 cement (LAC). Cem Concr Res 2012;42(2):41023.
[39] Kulik DA. Improving the structural consistency of CSH solid solution
thermodynamic models. Cem Concr Res 2011;41:47795.
[40] Kocaba V. Development and evaluation of methods to follow microstructural
development of cementitious systems including slags. Lausanne: cole
Polytechnique Fdrale de Lausanne; 2009.
[41] Deschner F, Winnefeld F, Lothenbach B, Seufert S, Schwesig P, Dittrich S, et al.
Hydration of Portland cement with high replacement by siliceous y ash. Cem
Concr Res 2012;42(10):1389400.
[42] Pllmann H. Syntheses, properties and solid solutions of ternary lamellar
calcium aluminate hydroxi salts (AFm-phases) containing SO2 4 , CO3
2
and
OH. N Jb Miner Abh 2006;182(2):17381.
[43] Tsivilis S, Chaniotakis E, Kakali G, Batis G. An analysis of the properties of
Portland limestone cements and concrete. Cem Concr Compos 2002;24(3
4):3718.
[44] Bonavetti V, Rahhal VF, Irassar EF. Studies on the carboaluminate formation in
limestone ller blended cements. Cem Concr Res 2001;31(6):8539.
[45] Gutteridge WA, Dalziel JA. Filler cement the effect of the secondary
component on the hydration of Portland cement Part 1: ne nonhydraulic
ller. Cem Concr Res 1990;20:77882.
[46] Gutteridge WA, Dalziel JA. Filler cement: the effect of the secondary
component on the hydration of Portland cement: Part 2: ne hydraulic
binders. Cem Concr Res 1990;20:85361.
[47] Ben Haha M, Lothenbach B, Le Saout G, Winnefeld F. Inuence of slag
chemistry on the hydration of alkali-activated blast-furnace slag Part I: effect
of MgO. Cem Concr Res 2011;41(9):95563.
[48] Ben Haha M, Lothenbach B, Le Saot G, Winnefeld F. Inuence of slag
chemistry on the hydration of alkali-activated blast-furnace slag Part II:
effect of Al2O3. Cem Concr Res 2012;41(1):7483.
[49] Len-Reina L, De la Torre AG, Porras-Vzquez JM, Cruz M, Ordonez LM, Alcob
X, et al. Round robin on Rietveld quantitative phase analysis of Portland
cements. J Appl Cryst 2009;42:90616.
[50] Franke B. Bestimmung von Calciumoxyd und Calciumhydroxyd neben
wasserfreiem und wasserhaltigem Calciumsilikat. Z Anorg Allg Chem
1941;247(12):1804.