Professional Documents
Culture Documents
h i g h l i g h t s
We developed limestone cement with high packing density to maximize binder properties.
Limestone powders with three main particle diameters were examined.
Combinations of limestone powders with several different particle sizes were studied.
Increasing surface area and packing density improve blended cement performances.
Blended cement with a combination of several particle sizes performed the best.
a r t i c l e i n f o a b s t r a c t
Article history: Limestone cement with high packing density was developed to maximize binder properties in spite of
Received 3 April 2014 increased limestone contents. Limestone powders with three main particle diameters relative to the clin-
Received in revised form 30 June 2014 ker particles were used. Cements with a single-size limestone particle and with combinations of several
Accepted 4 August 2014
particle sizes were compared. It was concluded that the replacement of an active material with an inert
Available online 7 September 2014
additive can improve cement paste performances by increasing the surface area and the packing density
of the cement-based particles, mainly when limestone powders with a combination of several different
Keywords:
particle size distributions were used due to increased packing density.
Hydration
Particle size distribution
Ó 2014 Elsevier Ltd. All rights reserved.
Blended cement
CaCO3
Sustainability
Specific surface
1. Introduction metakaolin [1–3] and (2) materials which are not considered poz-
zolanic, generally having low reactivity with cement minerals. Of
Due to environmental and energy efficiency concerns, there is the latter type, limestone is one of the most attractive additives
growing interest in the development of a blended Portland cement because it is considered natural, available, and economical. Several
in which the amount of clinker is reduced and partially replaced studies have reported that cement blended with limestone had
with mineral additives. There are three principal motivators improved initial compressive strengths with lower setting times
behind these efforts: (1) ecological benefits, as a result of lower compared to the original cement, i.e., without added limestone
CO2 emissions to the atmosphere, (2) economic benefits, since [4–6]. The addition of fine inert limestone powder, whose surface
reduced clinker cement is cheaper to produce, and (3) scientific/ area was greater than that of the clinker, increased the hydration
technological benefits, based on improved cement and concrete rate at early age and the generated heat of hydration [5]. However,
performance. There are two main types of mineral additives com- the final strength of the blended cement after 28 days was less
monly used: (1) pozzolanic additives such as fly ash, slag, and than that of the original cement paste.
Cement properties have been explained by the effect of packing
density, which is defined as the ratio between the solid phase vol-
⇑ Corresponding author. Address: Structural Engineering Department, Ben-Guri- ume and the total volume of the system (Fig. 1). The inclusion in
on University of the Negev, P.O. Box 653, Beer-Sheva 84105, Israel. Tel.: +972 8 647
the system of particles with improved particle size gradation con-
9672; fax: +972 8 647 9670.
E-mail addresses: yaniv.knop@gmail.com (Y. Knop), alvapeled@bgu.ac.il (A. fers on it increased packing density and, thus, decreased porosity,
Peled), ronenc@nesher.co.il (R. Cohen). as was also described by the linear packing density model of grain
http://dx.doi.org/10.1016/j.conbuildmat.2014.08.004
0950-0618/Ó 2014 Elsevier Ltd. All rights reserved.
Y. Knop et al. / Construction and Building Materials 71 (2014) 26–34 27
CEM I 52.5 R was partially replaced with limestone powders (>99.8% CaCO3)
with varying particle size distributions. The chemical composition of the original
cement is presented in Table 1. Three different limestone powders representing
several particle diameters—smaller than, larger than, or similarly sized to the origi-
nal CEM I with a mean particle size of 17 lm—were tested. The limestone powders
with the smaller and similarly sized particles to the original cement were fractions
of the same source, while the limestone with larger size was from a different source.
However, the purity and the density of the limestone powders were similar. Two
different powder systems were prepared and investigated:
Table 1
Chemical composition of the original cement.
Component CaO SiO2 Al2O3 Fe2O3 MgO TiO2 K2O Na2O P 2 O5 Mn2O3 SO3
% 65.07 18.96 4.5 2.46 1.16 0.36 0.33 0.21 0.32 0.30 2.86
The PSD and the surface area of the original cement and of lime- Fig. 2. Particle size distribution (PSD) of the tested powders (cement and
stone powders are given in Table 2. limestone).
Fig. 2 shows the particle size as a function of the cumulative
volume of the solid for the three tested limestone powders and
the cement. The mean particle size of each powder, based on the
PSD analysis, was determined by the volume of the particles. In
Fig. 2, the mean size of each limestone powder and of the original
cement are represented by the dashed lines. The three different
limestone sizes used had mean particle sizes of 53 lm, 25 lm,
and 3 lm, each of which is signified by CCX, such that X stands
for the particle size, e.g., CC3 is the limestone powder with mean
particle size 3 lm. Note that for the limestone powder with the
largest mean particle size of 53 lm, only a portion of its particles
were larger than (measuring from 45 lm to 100 lm) those of the
original cement. In the case of the limestone powder with the
smallest mean particle size (3 lm), most of its particles were smal-
ler than those of the original cement.
The ratio between the surface area of each blended powder
(limestone + cement) and of the original cement as a function of
the percent of original cement replaced with the limestone is
shown in Fig. 3a for the different limestone powders (e.g., CC3 lm,
CC25 lm, and CC53 lm). A scaling factor (AM) was introduced by
Kumar et al. [10] to describe the relative change in the solid surface
area induced by the filler addition in comparison to the surface
area provided by unit mass of cement. The influence of the lime-
stone replacement content on the solid surface area is shown in
Fig. 3b by using an area multiplier (AM, unitless). In both cases,
the surface area is strongly correlated with that of the respective
limestone powder used in the mixture for the three tested systems.
The finest limestone powder (CC3 lm) observed the greatest AM
factor, which exhibited the highest increase in the AM factor when
increasing its content in the blended cement.
Fig. 3. (a) Surface area of the tested powders (cement and limestone) and (b) the
correlation between the area multiplier to the limestone content in the blended
3.2. Water amounts for normal consistency cement and subsequent
cement.
packing density
Water was added to the blended cement powders until normal pare normal consistency cement and the packing density is given
consistency was obtained based on EN 196/3. Water demand is in Eq. (1) [22], taking into consideration the specific density and
affected by several factors such as particle shape, particle surface weight of the examined powder.
area, density of the material, and the packing density of the parti-
1
cles. The correlation between the amount of water required to pre- ;¼ ð1Þ
1 þ qP wp
with water. The presence of this layer indicates that C–S–H grows
not only around the clinker particle, as expected, but also around
the limestone particle. Similar observations were also reported
by Kumar et al. [10] that examined a variety of mixtures composed
of different cements. These observations demonstrate that the
limestone particles provide additional nucleation centers. Thus,
partial replacement of the cement with limestone containing par-
ticles with larger surface areas compared to those of the original
cement particles produced a blend with a greater number of nucle-
ation centers than in the original cement. Because the hydration
products also develop on the limestone particles and not only on
the cement particles, the subsequent hydration rate of such a blend
is high. In contrast, when the limestone in the cement–limestone
blend was composed of particles with smaller surface areas than
those of original cement particles, the number of nucleation cen-
ters—and subsequently, the hydration rate—decreased.
As discussed in Section 3.1, the increase in limestone content
led to an increase in the surface areas of the fine limestone blended
cement powders (Fig. 3). Therefore, for the cement blended with 3-
lm limestone particles (CC3 lm), i.e., the smallest investigated
particle with the highest surface area, faster hydration rates were
observed compared to the original cement for all replacement con-
tents, as expected (Fig. 7c). However, for this fine particle system,
no straightforward correlation was observed between the surface
areas of the different limestone content systems and the setting
process, as can clearly be seen in Figs. 7c and 8. Similar behaviors
of penetration depth vs. time were observed for the cement
blended with either 5% or 10% limestone, i.e., similar setting rates
despite the greater surface area of the 10% vs. the 5% limestone
blended cement (Fig. 3). This trend is clearly noticed in Fig. 8, pre-
senting similar time values of 166 and 168 min for 5% and 10% con-
tents, respectively, to reach a 20-mm penetration depth. A similar
trend was also observed when comparing the 20% with the 30%
blended cement systems (Figs. 7c and 8).
On the other hand, fine limestone particles can agglomerate as
presented in Fig. 9. This figure presents a SEM image of mortar
made of blended cement with the small-particle limestone
(CC3 lm). Agglomeration, a result of interparticle interaction, cre-
ated a massive, ‘‘particle’’ whose size was over 100 lm. The effect
of agglomeration is expected to be dependent on the replacement
content of the original cement with limestone having fine particles.
Fig. 7. Penetration depth vs. time behavior of original cement and blended cements Replacement of a small percentage of original cement with fine
with 5%, 10%, 20%, and 30% wt. limestone with particle size of: (a) CC53 lm (b) limestone particles is expected to cause, if any, only a local level
CC25 lm, and (c) CC3 lm.
Sand parcle
Fig. 8. The time to reach 20 mm penetration depth of blended cements with Limestone
limestone particle sizes of CC3 lm, CC25 lm, and CC53 lm limestone contents of agglomeraon
5%, 10%, 20%, and 30% by weight. 0% represents the original cement.
centers [5,6,12]. Inan et al. [23] observed a layer of calcium–sili- Fig. 9. SEM image of agglomeration of limestone particles near a sand particle of
cate–hydrates (C–S–H) around the surface of the limestone particle mortar with blended cement with the smallest limestone particles (CC3 lm) using
despite the inactive character of limestone, which does not react EDS analysis.
Y. Knop et al. / Construction and Building Materials 71 (2014) 26–34 31
effect because of the dispersion of the limestone in the original 53 lm + 2% 3 lm. Finally, the longest times up to setting and the
cement. However, by the increased content of limestone having slowest hydration process were obtained for the paste that incor-
fine particles, agglomeration is expected to be more dominant. porated the large limestone particles (CC53 lm). Note that in order
Thus, by increasing the content of fine particles above a certain to reach normal consistency, the amount of water was not the
percentage, the packing density decreased because of the agglom- same for all mixtures, i.e., the water/cement ratio was not the same
eration effect, and this in turn decreased the effective surface area. for all systems; it was higher for the single-particle-size systems,
It should be remembered that with these very fine powder sys- 0.290 and 0.285 for 3-lm and 53-lm particles, respectively and
tems, the amount of water needed to mix a paste of normal consis- 0.280 for the combined 3% 53-lm + 2% 3-lm system. For the origi-
tency is increased with the increase in limestone contents (relative nal cement, the water/cement ratio was 0.295. These ratios can
to that needed for the original cement) (Fig. 4), and the packing also influence the hydration rates discussed above.
density decreases (Fig. 5). The low packing density and the Analysis of the amounts of water required to mix the cement
increased number of ‘‘empty’’ spaces within the paste may slow pastes to normal consistency is presented in Fig. 11a. The figure
the setting times despite the larger surface area. shows that significantly less water was required for the cement
These results suggest that two principal mechanisms—surface system blended with the limestone combination of 3% 53-
area and packing density—should be considered when discussing lm + 2% 3-lm particles compared with the systems with single-
hydration rate behaviors. These two parallel mechanisms entail particle-size distributions (CC53 lm, CC3 lm). The reduced
opposing influences. Based on the results obtained here, small par- amount of water needed to reach normal consistency indicates
ticle systems have large surface areas that, although they increase an increase in the packing density of the blended cement paste
the hydration rate, they also exhibit lower packing density, which with the combined particle sizes. The calculated packing density
slows the setting of the paste. When limestone with larger parti- of the different limestone systems with the small and large parti-
cles was used, the particles’ surface area was relatively low, leading cles is presented in Fig. 11b, including that of the original cement.
to a slow hydration rate, but at the same time to high packing den- The figure clearly demonstrates that replacing the original cement
sity (Fig. 5).These results may explain the trends discussed above with limestone that is either single-particle-size or combined-par-
and the inconsistency between the hydration rates and surface ticle-size highly increased the packing density of the paste. Fur-
areas of the fine particle systems (Figs. 7c and 8). thermore, replacing the original cement with limestone
Combining large and small particles in one blended cement sys- containing a combination of several particle size distributions
tem may promote a large surface area and a greater number of increased the packing density of the paste compared to the cement
nucleation centers, but at the same time, it will also lead to higher blended with limestone that had only one particle size distribution
packing density (Fig. 1) that will speed the hydration process and (CC53 lm, CC3 lm), mainly for the combined 3% 53-lm + 2%
decrease the setting time of the paste. Blended cement systems 3 lm-particles. However, increasing the content of the fine lime-
comprising particle-size combinations are discussed below. stone powder from 2% to 4% and decreasing the content of the
large-particle limestone from 3% to 1% decreased the packing den-
3.3.2. Blended cement systems—combined-particle-size distributions sity, respectively.
Blended cement systems in which the original cement was par- Increases in packing density mean lower paste porosity, which
tially replaced with 5% limestone powder containing a combina- can lead to significantly decreased setting times of the blended
tion of several different particle size distributions were prepared cement with the combined particle sizes (Fig. 10). Faster time val-
and investigated. Limestone powders were prepared as follows:
(i) particle sizes of 53 lm (3%) + 3 lm (2%) combined; (ii) particle
sizes of 53 lm (1%) + 3 lm (4%) combined; (iii) 5% single particle
size of 53 lm; and (iv) 5% single particle size of 3 lm. The particle
size combinations clearly had a strong influence on the setting pro-
cess of cement paste (Fig. 10). Combined particle size systems
reduced the time required to reach penetration depths up to final
setting, i.e., increased the hydration rate compared to those of
the original cement paste. Thus, cement blended with the lime-
stone combination of 3% 53-lm + 2% 3-lm particles showed the
shortest time values up to setting among all the tested systems.
It can also be seen that altering the relative contents of the larger
and smaller particles to 1% 53 lm + 4% of 3 lm increased the time
values to reach final setting compared to the combination of 3%
Fig. 11. Properties of original cement and of cement pastes blended with single-
particle-size or two-particle-sizes with 5% limestone: (a) water requirements to
Fig. 10. Penetration depth vs. time behavior of original cement and of cement obtain normal consistency related to original cement and (b) calculated packing
pastes blended with single-particle-size or two-particle-sizes with 5% limestone. density.
32 Y. Knop et al. / Construction and Building Materials 71 (2014) 26–34
ues to reach penetration depths up to final setting were measured The reduced intensity of the peak, which indicates the quantity
despite the reduced surface area of the combined blended system of the two main clinker minerals C3S and C2S means that the
(3% 53 lm + 2% 3 lm). Surface area of 1.04% relative to the original blended cement with the finer limestone particles (CC3 lm) and
cement was measured for the blended cement with the combined greater surface area contains lower amounts of C3S and C2S, i.e.,
particle sizes, compared with that of the finest limestone blended greater hydration and a higher reaction rate of the blended cement
cement system (CC 3 lm) of 1.15%. This trend can be explained by with limestone having more surface area.
a combination of both specific surface area and interparticle spac- Therefore, partial replacement of the original cement with lime-
ing effects [10,24]. The combined 3% 53-lm + 2% 3-lm particle stone whose surface area was greater than that of the original
system had relatively low surface area compared to the system cement promoted reaction of a greater volume content of clinker
with 5% limestone with 3-lm particles, but reduced interparticle components with water and their subsequent transformation to
spacing, leading to lower setting times. calcium silicate hydrates. This led to an increased volume of the
The effect of surface area on hydration rate of the blended hydration products and a decrease in setting time (Fig. 7). As
cement pastes with 5% limestone was estimated by measuring explained in Section 3.3.1, the higher hydration rate was caused
the cumulative content of their four main mineral components: by the increase of the nucleation centers in the blended cement
C3S, C2S, C3A, and C4AF using Rietveld quantitative phase analysis with the finer limestone filler. In summary, the partial replacement
of X-ray diffraction 2.5 h after casting. Calculations of the cumula- of the original cement with limestone comprising a combination of
tive content of the blended cement minerals were related to that of several particle size distributions increases the surface area and the
the original cement. The analyzed cumulative contents of the four packing density of the fresh paste—mutual effects that can signifi-
mineral components were: 109% and 94%, for the single 53-lm and cantly reduce setting times (initial and final).
single 3-lm limestone blended cements, respectively, related to
the original cement. This indicates greater hydration for the fine 3.4. Compressive strength
particle blended cement (CC3 lm), i.e., less clinker and faster
hydration rate at age 2.5 h. The effect of the limestone surface area 3.4.1. Blended cement systems—one-particle-size distribution
on the hydration rate was also observed at later ages and with The compressive strengths on days 1 and 28 after casting the
greater limestone content by analyzing the XRD patterns of mortar with the cement systems blended with 5% wt. limestone
blended cements with CC3-lm and CC25-lm particle sizes and (particle size 53 lm, 25 lm, or 3 lm) were examined (Fig. 13).
20% limestone at the age of 1 d (Fig. 12). Fig. 12a presents the full The measured compressive strength of the original cement at 1 d
XRD curves of the two blended cements obtained during testing was 18.3 MPa and at 28 d, it was 65.0 MPa. The coefficient variance
with a magnified portion of the XRD pattern. This magnified zone was less than 1% for all tested systems. The ratio between the com-
is highlighted in Fig. 12b, which presents the intensity peaks pressive strength of the blended cement system to the strength of
related to the two main clinker minerals C3S and C2S of the cement the original cement system (without added limestone) is pre-
powder as reported by Yongqi et al. [25]. A clear decrease in the sented. At 1 d, a good correlation is clearly observed between the
intensity of this peak can be seen for the finest particle (CC3 lm) blended cement’s compressive strength and the surface area of
blended cement compared to the blended cement with CC25-lm the blend. The highest compressive strength was observed for
limestone. The blended cement with the CC25-lm limestone has the blended limestone system with the greatest surface area
less surface area of 1.371 m2/gr compared with 6.222 m2/gr for (CC3 lm), while the system with the least amount of surface area
the fine 3CC-lm limestone blended cement (Fig. 3a and Table 2). had the lowest strength (CC53 lm). In addition, the particle system
with the greater surface area was 4% stronger than the original
cement system, despite the reduction in the active component
(i.e., clinker) in the mortar. Such improved compressive strength
at 1 day of age can be attributed to the increased number of nucle-
ation centers created when limestone is added to cement, as dis-
cussed in Section 3.3.1.
The porosity of the different systems with and without lime-
stone replacement was measured by image analysis using TESCAN
VEGA3 Control software at 1 d. 15 SEM images of each one-
particle-size system: original cement, CC3 lm, CC25 lm, and
Fig. 12. XRD patterns of blended cement with limestone having CC3 lm particles Fig. 13. Compressive strengths (days 1 and 28 after casting) of mortars in which the
and CC25 lm, after 1 day of casting: (a) full scale pattern and (b) magnified region cement was replaced with 5% limestone powder with mean particle size of either
of C3S and C2S peaks. CC3 lm, CC25 lm, or CC53 lm.
Y. Knop et al. / Construction and Building Materials 71 (2014) 26–34 33