Evaluation of Portland pozzolan blended cements containing diatomaceous earth

Abstract

Clinker replacement by supplementary cementitious materials (SCMs) is one strategy to reduce CO2 emissions of cement industry. Diatomaceous earth, a natural pozzolan, has been used as SCM. So, this study aimed to evaluate two Portland pozzolan blended cements with distinct content of diatomaceous earth (16DE and 49DE) and compare then with a high clinker content cement (REF). Cements were physically and chemically characterized; pastes and mortars were analyzed in terms of reactivity, water demand, and mechanical strength; finally, environmental indicators were estimated. Cements with diatomaceous earth demanded a similar volume of mix water for mortars without superplasticizer and less water than REF for mortars with superplasticizer. The chemical bound water at 91 days reduced 21% and 27% for 16DE and 49DE, respectively, in relation with REF. For binder intensity indicator, both cements had worst results compared to REF, but for carbon intensity indicator 49DE obtained better results comparing to REF.

Full text

75
INTRODUCTION
The increasing production of cement will extend at least
until 2050; studies show a demand forecast for the Brazilian
market between 120 and 140 million of tons of cement
and worldwide between 3.7 and 4.4 billion tons [1]. This
growing production will be responsible for a large volume of
CO2 emissions, worldwide this value is approximately 10%
of total anthropogenic CO2 emissions and can reach 30%
by 2050 [2]. Therefore, academy and industry have been
developing strategies to reduce this environmental impact
generated by the cement industry. One of these strategies is
the substitution of clinker by supplementary cementitious
materials (SCMs). Supplementary cementitious materials
(SCMs) are nely divided materials that contribute to
the properties of the nal product due to: hydraulic or
pozzolanic activity or those that are nominally chemically
inert but contribute with ‘ller effect’, as for instance
limestone ller (SCMs can be natural, processed natural
or articial) [3, 4]. Apart from chemical contribution,
supplementary cementitious materials can also affect the
mixing water when blended with Portland cement, and
these two aspects impact on cement efciency in terms of
mechanical strength and durability. Nowadays the most
used supplementary materials are blast furnace slag and y
ash, by-products of the steel industry and thermoelectric
plants, respectively, but the increase in cement production
was not accompanied by the supply of these materials. This
is observed with the stabilization of world average clinker
ratio since 2000 [2]. Therefore, other supplementary
materials are being studied and some are already applied
by the industry as is the case of diatomaceous earth.
Diatomite is a material of sedimentary and biogenic
origin, that is formed from the accumulation of algae shells
that were fossilized due to the silica deposit on its structure
[5]. Its origin is usually associated with clay, quartz and
iron oxide, composing than the name diatomaceous earth.
It is classied as a pozzolan: materials that are formed by
an amorphous and a crystalline phase and that in presence
of water and at ordinary temperature react with calcium
hydroxide precipitating hydrated phases with cementitious
properties (C-S-H, C-A-S-H, AFm) [6]. The natural
pozzolans need no treatment but grinding is needed to
present pozzolanic activity, as is the case of pyroclastic
rocks, for instance, Santorini earth, zeolites, volcanic ashes
and sedimentary rocks such as diatomaceous earth [3].
Results on Portland blended cements with diatomaceous
earth showed a low reactivity on the rst days of hydration
and better results up to 28 days [7]. The addition of
diatomaceous earth normally leads to an increase in
the mixing water mainly because of agglomeration of
diatomaceous earth particles, high internal porosity and
specic surface area [8].
The replacement of clinker by pozzolans or any other
supplementary cementitious material is always attributed
to having a linear correlation with CO2 emission reduction,
and this approach is valid when only the cement production
Cerâmica 65 (2019) Suppl.1 75-86 http://dx.doi.org/10.1590/0366-6913201965S12596
Evaluation of Portland pozzolan blended cements
containing diatomaceous earth
P. C. R. A. Abrão1,2*, F. A. Cardoso1,2, V. M. John1,2
1University of São Paulo, Polytechnic School, Department of Construction Engineering,
Av. Prof. Almeida Prado 83, 05508-070, S. Paulo, SP, Brazil
2National Institute on Advanced Eco-efcient Cement-based Technologies (INCT), S. Paulo, SP, Brazil
Abstract
Clinker replacement by supplementary cementitious materials (SCMs) is one strategy to reduce CO2 emissions of
cement industry. Diatomaceous earth, a natural pozzolan, has been used as SCM. So, this study aimed to evaluate
two Portland pozzolan blended cements with distinct content of diatomaceous earth (16DE and 49DE) and compare
then with a high clinker content cement (REF). Cements were physically and chemically characterized; pastes
and mortars were analyzed in terms of reactivity, water demand, and mechanical strength; nally, environmental
indicators were estimated. Cements with diatomaceous earth demanded a similar volume of mix water for mortars
without superplasticizer and less water than REF for mortars with superplasticizer. The chemical bound water at 91
days reduced 21% and 27% for 16DE and 49DE, respectively, in relation with REF. For binder intensity indicator,
both cements had worst results compared to REF, but for carbon intensity indicator 49DE obtained better results
comparing to REF.
Keywords: reactivity, water demand, environmental indicators, Portland pozzolan blended cements.
*pedro.abrao@lme.pcc.usp.br
https://orcid.org/0000-0002-3416-3196

76
is evaluated. Nowadays, most studies that assess CO2
emissions of cement industry are within this frontier of
study [9-11]. However, cement is produced to be used
for some purpose, either for concrete, mortar, precast
and others, and in these cases the clinker replacement by
SCMs can affect the eco-efciency of the nal product
[8]. Few researches assess the eco-efciency of cement in
its use; some of these works quantify the CO2 emissions
per m³ of concrete [12-15], but it is important to relate
these impacts to the performance of the material [13, 16].
The performance of the nal product (concrete, mortar,
precast, etc.) is tied with cement efciency. This efciency
is related to two parameters: rst is the binder ability to
chemically combine water, and second is the water required
to promote a specic rheological behavior suitable for each
type of technological application. Both parameters affect
the performance of the cementitious product in terms of
rheological behavior, mechanical strength, and durability.
Therefore, the aim of this study is to evaluate two Portland
pozzolan blended cements with distinct content of
diatomaceous earth (16DE and 49DE) and compare them
with a high clinker content cement (REF). Cement pastes
and mortars, prepared with and without superplasticizer
(dispersant), were analyzed in terms of reactivity, water
demand, and mechanical strength, then environmental
indicators were calculated based on experimental results
and literature data.
MATERIALS AND METHODS
A cement with a high content of clinker (REF)
and two Portland pozzolan blended cements from the
Brazilian market were analyzed as to their chemical and
mineralogical compositions, physical characteristics, water
and superplasticizer demand, reactivity and mechanical
properties. Types of cements: two commercial Portland
pozzolan blended cements from different plants were
analyzed with distinct percentages of diatomaceous earth
(16DE and 49DE) and as a reference a cement with 94%
of clinker+CaSO4 and 6% of limestone ller. Table I
shows detailed information about the cements, the codes
indicate: i) percentage of pozzolan on cement, determined
by Rietveld renement; and ii) the type of pozzolan,
diatomaceous earth (DE). Cement type is according to
Brazilian standards [17, 18], CP means Portland cement,
CPV is a high early age strength cement with high clinker
content. CPIIZ and CPIV are cement classes dened as
Portland pozzolan blended cements, the difference between
them is the amount of pozzolan allowed.
Characterization of anhydrous materials. Chemical
and mineralogical characterization: Table II presents the
results of chemical analysis obtained by X-ray uorescence
spectroscopy (Axios Advanced, PANalytical). Cement
49DE showed higher content of silica and lower content of
calcia comparing to REF; on the other hand, 16DE and REF
presented similar amount of calcium oxide, but 16DE had
a slightly higher content of silica from the diatomaceous
earth and also a higher content of magnesium oxide
probably from the limestone ller. Table III presents the
mineralogical composition of the materials, evaluated via
X-ray diffraction (XRD, X’Pert MPD, PANalytical). For
phases quantication a renement was performed by the
Rietveld method, using a software (HighScore Plus v.4.6a)
and Panalytical Inorganic Structure Database. Materials
with diatomaceous earth addition showed lower content of
clinker phases than REF, as expected for blended cements.
16DE and 49DE presented quartz, kaolinite and vitreous
phase from the amorphous silica of diatomite. Only 16DE
and REF had limestone ller on their compositions. Results
of limestone ller obtained from XRD and TG/DTG were
quite similar, showing that both techniques agreed, also the
statistical indicators Rwp (weighted-prole R-factor) and
GOF (goodness of t) were in the range dened as a good
renement [19].
P. C. R. A. Abrão et al. / Cerâmica 65 (2019) Suppl.1 75-86
Code Type of
cementaDescriptionaClinker+CaSO4
contenta (%)
Pozzolan
contenta (%)
Filler
contenta (%)
Type of
pozzolanb
REF CPV-ARI High early strength cement 95-100 - < 5 -
16DE CP II Z Portland pozzolan blended cement
with low pozzolan addition 76-94 6-14 < 10 Diatomaceous
earth
49DE CPIV Portland pozzolan blended cement
with high pozzolan addition 45-85 15-50 < 5 Diatomaceous
earth
a - according to Brazilian standards [17, 18]; b - according to the manufacturer.
Table I - General description of the investigated cements.
Oxide (%) CaO SiO2Al2O3Fe2O3SO3MgO Na2O K2OTiO2Loss on ignition
OPC REF 59.7 19.1 4.5 2.6 3.0 1.3 0.3 0.9 0.2 4.7
Diatomaceous
earth
16DE 57.1 21.9 4.9 4.4 2.2 4.0 0.2 0.2 0.3 4.2
49DE 44.1 35.2 4.2 3.4 2.6 1.8 0.4 0.7 0.5 4.6
Table II - Chemical composition of the used raw materials determined by X-ray uorescence spectroscopy.

77
Scanning electronic microscopy: materials were assessed
by scanning electronic microscopy (SEM, FEI, Quanta
600FEG) with energy dispersive spectroscopy (EDS,
Bruker, SSD Xash Quantax 400). The operating conditions
for obtaining SEM images were: high vacuum; high voltage
- 10 kV; and secondary electron detector. Software Esprit
was used to analyze EDS results. Fig. 1 presents the images
of 49DE cement, which had the addition of diatomaceous
earth. Fig. 1a shows a diatomite stick covered with kaolin
plates identied by EDS with silicon and aluminum:
the aluminum came from the kaolinite as diatomite had
only silicon in its composition (Fig. 1b); also, the stick
morphology is characteristic of a diatomite [20]; in both sides
clinker grains can be seen with some hydration spots. Fig.
1c presents a particle with multiple layers and pronounced
texture that appears to be a cluster of kaolin plates, identied
by EDS with silicon and aluminum (Fig. 1d). In both cases,
EDS showed platinum element, because the samples were
coated with a ne layer of platinum to provide electrical
conduction as well as to obtain high magnication images.
Fig. 1e presents a clinker particle covered by kaolin plates
and with some hydration spots, while Fig. 1f displays the
corrugated texture of the clinker and the kaolin plates lying
on its surface.
Physical characterization: the physical parameters
were obtained by the following methods: i) particle size
distribution was determined by laser diffraction (Helos/
KR, Sympatec) with the particles dispersed in deionized
water (Sucell, Sympatec); the powder was mixed with water
for 60 s at 1500 rpm with a rotational mixer, then part of
the suspension was inserted on the basin that already had
deionized water; the stirrer and pumping parameters were
set at 40% and 50%, respectively; immediately before the
test, ultrasound dispersion was employed for 90 s and, then,
the measurements were performed during 10 s three times in
a row; the same protocol was followed for the suspensions
with and without superplasticizer; admixture content
used was the one obtained as described ahead; ii) specic
surface area was determined by BET method (Belsorpmax,
Bel Japan) using nitrogen adsorption; and iii) true density
was measured by a helium pycnometer (Multipycnometer
Quantachrome MVP 5DC).
Fig. 2a shows the particle size distribution curves of
cements with and without superplasticizer. For materials
without superplasticizer (solid lines) cement REF presented
a higher volume of particles between 10 and 30 mm when
compared with Portland pozzolan blended cements; on
the other hand, cement 49DE presented higher volume of
particles between 0.2 and 6 mm comparing to 16DE and REF
indicating that the addition of diatomaceous earth produced
a cement with ner particles. With the incorporation of
Table III - Mineralogical composition of the used raw
materials determined by XRD with Rietveld renement.
Compound (%) OPC Diatomaceous
earth
REF 16DE 49DE
Clinker and sulfate phases
Alite 60.8 43.6 30.9
Belite 13.3 14.7 5.3
C3A 8.2 4.4 5.3
Ferrite 8.0 5.2 3.1
Periclase 0.0 4.8 1.9
Calcium sulfate phases 3.5 3.8 4.3
Clinker and sulfate content 93.8 76.5 50.9
Pozzolan phases
Quartz 0.0 1.6 7.9
Kaolinite 0.0 3.0 9.0
Vitreous phase 0.0 11.5 32.3
Pozzolan content 0.0 16.1 49.2
Limestone ller
Limestone llera4.9 7.4 0.0
Limestone llerb6.0 8.0 0.0
Statistical indicators of renement
Rwp 9.2 13.1 11.6
GOF 2.1 3.3 3.1
Figure 1: SEM micrographs of 49DE cement: a) diatomite stick
with corresponding EDS spectrum (b); c) kaolinite plates with
corresponding EDS spectrum (d); e) clinker grain with kaolin
plates and zoom of the delimited area (f).
P. C. R. A. Abrão et al. / Cerâmica 65 (2019) Suppl.1 75-86
a - determined via TG/DTG; b - determined via XRD.
1
1
10
10
cps/eV
8
8
6
6
4
4
0
0
2
2
2
2
3
3
O
O
Si
Si
Al
Al
Pt
Pt

78
superplasticizer (dashed lines), the curves moved to the left
(lower particle sizes); for cement REF the curve maintained
a similar shape, but for 16DE and 49DE there was a visual
increase on volume of particles between 0.5 and 8 mm, less
pronounced for 16DE. Fig. 2b shows the difference between
particle size distribution of cements without and with
superplasticizer; for cement REF and 16DE the addition
of superplasticizer led to a decrease in volume of particles
between 20 and 100 mm and an increase in the range of 0.2
and 20 mm; this also happened for 49DE, but in this case there
was a higher increase in volume of particles between 0.7 and
8 mm when compared with REF and 16DE. Combining these
results with the images obtained by SEM, it was supposed
that when superplasticizer is added on cements containing
diatomaceous earth there was a dispersion of some clusters
of kaolin plates (Fig. 1c) and therefore a system with ner
particles, which explain why Portland pozzolan blended
cements with superplasticizer presented a higher increase on
volume of ne particles when compared to REF.
Table IV presents the results of physical properties of the
materials, differentiating as to the use (SP) or not (noSP) of
superplasticizer, when applicable. All cements had similar
particle true density, but there were clear differences on
specic surface area. The cements with diatomaceous earth
had a higher specic surface area and, consequently, higher
volumetric surface area and shape factor. These parameters
are important since they affect water demand, especially the
shape factor [21] that gives an idea of morphology; particles
with high specic surface area, angular shape or with high
internal porosity demand more water to achieve a specic
rheological behavior. The parameters related to particle
size distribution (D10, D50, and D90) were reduced by half
when superplasticizer was incorporated.
Analysis of paste. Superplasticizer saturation content: for
the determination of superplasticizer saturation ratio, pastes
with 0.3 water/cement were produced. The superplasticizer
used was a polycarboxylate (ADVA CAST 527), which
acted with electrosteric stabilization mechanism, had a
water/solid ratio of 0.6 and a specic mass of 1.075 g/cm³.
The mixture of the pastes occurred in three steps: i) adding
water in the powder in 5 s; ii) manual mixing with scoop for
50 s; and iii) dispersion in 10000 rpm rotation mixer for 1.5
min. Immediately after mixing, the pastes were subjected to
rotational rheometry (Rheometer MARS 60, Haake), using
plate-plate geometry with 35 mm in diameter and gap of
1 mm. The shear program used was the stepped ow test
in two cycles of acceleration and deceleration, varying the
shear rate from 0 to 50 s-1. The rst cycle was run only for
paste normalization (structural breakdown), while the second
cycle was considered for data collection. At this stage, for
each superplasticizer content yield stress was obtained in
shear rate of 0 s-1. The saturation content was dened as the
posterior point to stabilization of the yield stress.
Reactivity: of the cements was measured by quantifying
the chemically combined water and portlandite content at 7,
28 and 91 days of hydration. Pastes with water/cement ratio of
0.5 were mixed in the same manner as described above, then
they were placed in cylindrical molds of 2.5 cm in diameter
and 8 cm in height and stored at 23 °C and 90-100% relative
humidity. After 24 h, they were relocated to larger containers
(3 cm in diameter and 10 cm in height) lled with deionized
P. C. R. A. Abrão et al. / Cerâmica 65 (2019) Suppl.1 75-86
Parameter
OPC Diatomaceous earth
REF 16DE 49DE
noSp Sp noSp Sp noSp Sp
ρs
a (g/cm³) 3.08 3.11 3.03
D10 (mm) 2.3 1.6 1.8 1.2 1.9 1.4
D50 (mm) 14.4 9.0 12.5 5.9 12.5 6.2
D90 (mm) 39.0 23.4 37.2 16.0 39.4 19.1
SSALD
b (m²/g) 0.36 0.49 0.44 0.63 0.39 0.58
SSABET
c (m²/g) 1.64 2.41 4.30
VSABET
d (m²/cm³) 5.0 7.5 13.0
Shape factor, ξe4.6 3.3 5.5 3.8 11.0 7.4
a - particle true density by He pycnometry; b - specic surface area by laser diffraction
granulometry; c - specic surface area by BET method; d - volumetric surface area
obtained by the SSABET and ρs; e - shape factor calculated from SSABET and SSALD.
Table IV - Physical properties of the used raw materials.
Figure 2: Particle size distribution curves obtained by laser
diffraction of cements without (solid line) and with (dashed line)
superplasticizer (a), and difference on particle size distribution
curves of compositions without and with superplasticizer - positive
values mean that total volume of particles in that specic size range
is higher when comparing to compositions without dispersant and
vice versa (b).
10
4
6
0
2
-6
8
2
4
-4
-2
0
-80.1
Particle size (µm)
Volume (%)Volume (%)
1 10 100

79
water, sealed and stored at 23 °C until the age of interest.
Slices from the cylindrical specimens were removed with
2 mm of thickness and, then, underwent to three immersion
cycles in isopropanol, the rst two immersions of 1 h each
and the third immersion of 24 h. Finally, the isopropanol was
exchanged for diethyl ether, with the samples immersed for
16 h, and subsequently dried in chamber at 50 °C for 30 min.
After the hydration stoppage, samples were ground and then
subjected to thermogravimetric test (TGA, STA 409PC/PG,
Netzsch). The quantication of the chemically bound water
(Bw) and the portlandite content (CH) were determined with
the following equations [22]:
Bw=W40-W500
W0
(A)
CH=W400-W500
W500
M
M (B)
where M is the molar mass, Wx is the percentage of mass
loss at temperature x (°C) and LOI is loss on ignition.
Analysis of mortar. Mortars preparation: in the study of
water demand, 1:3 (cement:sand weight ratio) mortars were
produced with a water/cement ratio adjusted for each sample.
Standard sand [23] was employed and a mix composed of
four fractions of different nominal sizes, 0.15, 0.3, 0.6, 1.2
mm, each fraction corresponding to 25 wt%, was prepared.
The superplasticizer type and dosage were according to
described above; cements used were the ones presented in
Table I. This mortar composition was elaborated according
to the Brazilian standard [24] for determination of cement
strength classes which agreed with European standards [25].
The mortars were mixed according to EN 196-1 standard
[25]. Right after the mixing procedure, it was measured the
incorporated air according to NBR 13278 standard [26].
Water demand: ow table test was used to analyze
the water demand for a specic rheological behavior of
cements. After measuring the incorporated air, mortars were
placed in a conic mold, the excess material and the mold
were removed, and the table then underwent 30 falls in 30 s
from a height of 12.5 mm, method according to NBR 13276
standard [27]. After this procedure, nal mortar spread on
the table was recorded by measuring three diameters, which
composed an average, also pictures of mortars before and
after the ow table test were taken to visual analysis. A
different mortar batch was used for each water/cement ratio.
The tests for each composition were initiated from the water/
cement (w/c) ratio of 0.48, the value used by the Brazilian,
European and American standards for the classication of
cement compressive strength class [25, 28, 29]. The other
mortars were prepared with the w/c ratio dosed with the
objective of achieving an average spread of 240±10 mm,
in which the mortars presented adequate consistency for
molding. The spread value of 240 mm was adopted after
several preliminary tests were performed; at rst, the range
of 265±10 mm was used, but it was noticed phase separation
in some mortars, the range was then reduced.
Compressive strength: after the ow table test, for
evaluation of compressive strength, mortars were molded
in specimens of 40x40x160 mm³, according to EN 196-1
standard [25]. For each age of interest, three specimens were
molded and then placed on moist air room for the rst 24
h, then they were demolded and immersed in water until
testing. At the age of interest (7, 28 and 91 days), they were
tested for exural strength through the three-point loading
method, followed by compressive strength test on the half
parts of the prism according to EN 196-1 standard.
Porosity: was calculated according to Power’s model
[30]. However, in this case, the degree of hydration and
bound water in the age of interest were calculated based
on experimental data obtained as described in Reactivity
subsection. The apparent density was calculated by
measuring the dimensions and mass of the specimens for
each age and the theoretical density was calculated from the
density of raw materials and the mortar composition. The
total mortar porosity was calculated according to:
Totalporosity=+
Vcapillary porous
Vtotal
1-Dapparent
Dtheoretical
(C)
where Vtotal is total volume, Vcapillary porous is the volume of
capillary pores, Dapparent is the density of the test body,
and Dtheoretical is theoretical density. As 49DE had a large
proportion of pozzolan it was considered that all materials
were totally hydrated at 180 days. The combined water at
180 days (Bw180days) was estimated through extrapolation of
the experimental curve (7, 28 and 91 days).
Binder intensity, carbon intensity, and CO2 emissions:
two indicators were calculated in this work: binder intensity
(BI) and carbon intensity (CI). Both were presented in
[16]: the BI relates the amount of binder per m³ of concrete
to deliver a unit of a performance indicator; the CI relates
the amount of CO2 emitted to a unit of a performance
indicator. Different from the method applied in [16], in this
work, BI and CI were calculated for modeled concretes
with 28 and 91 days of hydration, since Portland pozzolan
blended cements tended to react more at long ages. In this
study for calculations of BI and CI, a concrete with 300
dm³ of paste was modeled; from this value, the volume of
cement according to the water/cement ratio of each system
was calculated. The compressive strength of concrete was
estimated by multiplying the compressive strength of
mortars by a factor of 0.8 [31]. It was considered a CO2
emission factor of 857 kgCO2/t clinker, Brazilian average
value [32], for CI and CO2 emission calculations. In both
cases, emissions relating to the process of production,
grinding, and transportation of supplementary cementitious
materials were not incorporated into the calculations, a
common procedure in the literature [33]. For calculation of
CI, the emissions concerning the extraction, beneciation,
and transportation of the aggregates, as well as production
and transportation of the concrete, were considered null
P. C. R. A. Abrão et al. / Cerâmica 65 (2019) Suppl.1 75-86

80
in this study. The amount of CO2 absorbed during the
carbonation was also neglected. All of these factors when
summed up did not signicantly change the results since
the production of clinker is the most contributing step for
the CO2 emissions of the concrete [12].
RESULTS AND DISCUSSION
Superplasticizer demand: the results of rotational
rheometry test is shown in Fig. 3 where the yield stress
is expressed according to the superplasticizer content in
percentage of cement mass. The results showed that for the
same volume of mixing water, the cements with diatomaceous
earth demanded a higher volume of superplasticizer to
achieve the same rheological behavior (Fig. 3), due to their
larger specic surface area and shape factor. Data in Table
V demonstrate that saturation content ranged from 0.6 to
1.4 wt%, showing the importance of correctly determining
the volume of additive. In addition to the saturation content,
analysis of the specic consumption was made, that is
additive consumption divided by the specic surface area as
measured by BET. The specic content of superplasticizer
was in the same range for REF and 49DE and a little
above for 16DE, which may be connected to the chemical
composition or because of particle adsorption [34].
Reactivity: Fig. 4 presents the results obtained by
hydration stoppage. Cement REF combined more water
and had a higher portlandite content in all ages, compared
to 16DE and 49DE, due to its higher percentage of reactive
phases. Both Portland pozzolan blended cements had a
similar evolution on bound water but the results were less
pronounced when compared with REF. The progressive
reduction of portlandite content is an indicative of pozzolanic
activity, in which pozzolan phases consume Ca(OH)2 to
produce hydration phases with cementitious properties.
16DE was the most reactive Portland pozzolan blended
cement on early ages, but over time the results of portlandite
content showed that pozzolan on this cement reacted less than
the diatomaceous earth on 49DE. Cement 16DE had 11%
of amorphous phase on its composition and consumed only
0.24 gCH/100 g cement; on the other hand, 49DE cement
had 32% of amorphous phase and consumed 1.44 gCH/100
g cement between 7 and 91 days of hydration; so 16DE had
three times less amorphous phase but consumed six times
less portlandite than 49DE. Concluding, diatomaceous earth
and limestone ller of 16DE actually acted in a mechanism
called ‘ller effect’ [35], providing sites for nucleation and
space for the hydration products of the clinker phases, thus
giving a greater reactivity to the material in the rst ages,
although stabilizing in the long term.
Fig. 5 shows the results of bound water relative to REF
P. C. R. A. Abrão et al. / Cerâmica 65 (2019) Suppl.1 75-86
Material w/s
Saturation
dosage
(wt%)
Specic
dispersant
(mg/m²)
τo
(Pa)
REF
0.3
0.6 3.4 0.3
16DE 1.1 4.6 0.9
49DE 1.4 3.3 0.2
Table V - Water/solid (w/s) ratio, saturation dispersant
content, specic dispersant content and yield stress (τ0) of
the studied systems.
Figure 3: Yield stress at 0 s-1 from the downwards shear rate ramp as
a function of the superplasticizer content for the analyzed systems.
0.1
Superplasticizer content (%)
Yield stress (Pa)
0.7 1.30.3
150
100
75
50
25
0
125
0.9 1.50.5 1.1
Figure 4: Content of bound water (a) and portlandite (b) obtained
from TGA for all systems at 7, 28 and 91 days of hydration.
24
20
16
12
22
18
14
10
0
Time (day)
Bound water (g/100 g cement)
604020 80 100
24
20
16
12
22
18
14
10
Bound water (g/100 g cement)
0
Time (day)
604020 80 100

81
versus clinker substitution. Cement 16DE on 7 days of
hydration stayed above the dilution line (dashed line), that
is, the reduction in the chemically bound water was lower
than the reduction in the clinker content when compared to
REF, indicating an enhance on reactivity on the rst days,
but at 91 days of hydration the relative bound water moved
to below the dilution line, ratifying that, in this case, the
diatomaceous earth on 16DE did not react much on long
term ages. The 49DE cement achieved better results, staying
above the dilution line at all ages. At 91 days of hydration
this cement had a 27% reduction in bound water, but with a
clinker substitution of 43%, compared to REF. Therefore, the
clinker substitution by diatomaceous earth on this case was
favorable to reactivity, because part of diatomaceous earth
also reacted by consuming the portlandite and producing
other hydration products with cementitious properties, thus
contributing with an increase in the chemical bound water.
Water demand: maintaining a xed water/cement ratio
of 0.48 in mass (1.45 to 1.49 in volume), as suggested by
the Brazilian, European and North American standards, the
mortars without superplasticizer showed dry aspect and
inappropriate consistency for molding as shown in Figs. 6a
and 6e. Images captured after the test indicated that mortars
did not spread, but instead underwent fracturing due to
the lack of cohesion between the cementitious matrix and
the aggregates (Figs. 6b and 6f). Determination of cement
compressive strength based on a xed water/cement ratio
is not the best option, because the result of the test is highly
dependent on the energy applied during the molding process.
Therefore, the minimum water demand must be based on
a rheological behavior adequate for easily molding the
mortar samples. In this study, a xed rheological behavior
was adopted: spreading on ow table equal to 240±10 mm;
for this value, mortars presented better cohesion without
segregating, which guaranteed a good molding and proper
spreading as indicated on Figs. 6c, 6d, 6g, and 6h.
Fig. 7a shows the results of ow table test for mortars
without (empty icons) and with (lled icons) superplasticizer.
Without superplasticizer, all cements demanded a similar
amount of water to reach 240±10 mm spread, but with
the addition of diatomaceous earth a slight reduction on
mix water was achieved, 2% for 49DE and 4% for 16DE
compared to REF. With the incorporation of superplasticizer,
the cements with diatomaceous earth required a smaller
volume of water to achieve the same spread; for 49DE it
was obtained a reduction of 8% and for 16DE 10% when
compared to REF. With incorporation of superplasticizer,
cements containing diatomaceous earth presented a higher
volume of ne particles compared to REF, results obtained
from particle size distribution analysis (Fig. 2a), indicating
a dispersion of kaolin clusters; these ne particles than
had an effect of lling voids among the others, increasing
the packing of the system, reducing the necessary water
for the same uidity, and explaining why cements with
P. C. R. A. Abrão et al. / Cerâmica 65 (2019) Suppl.1 75-86
Figure 6: Images before (left) and after (right) the ow table test
(dropping the table 30 times) of mortars made with different water/
cement (w/c) ratio: a,b) REF mortar with 0.48 w/c; c,d) REF mortar
with 0.53 w/c; e,f) 49DE mortar with 0.48 w/c; and g,h) 49DE
mortar with 0.53 w/c.
Figure 5: Relative bound water (REF) versus clinker substitution.
The empty icon corresponds to 7 days of hydration, pointed icon
28 days and lled icon 91 days. The black line corresponds to a
direct dilution.
1.0
0.9
0.8
0.7
0.6
Clinker substitution (%)
Relative bound water
10 403020 50 60

82
diatomaceous earth demanded less water than REF when
superplasticizer was added. In all cements the addition of
superplasticizer reduced the mix water: for cement REF a
decrease of 23% was achieved, for 16DE 28% and 49DE
26%. This water reduction inuenced the performance of
the material, such as mechanical properties, durability, and
environmental indicators.
The maximum paste thickness (MPT) [36] was
calculated for mortars with and without superplasticizers
(Fig. 7b). MPT represents the distance between the aggregate
particles and depends on the volumetric surface, packing
porosity and volumetric solid concentration of the coarse
particles in the mortar. A higher MPT reduces the friction
between the coarse particles and consequently increases
the uidity of the mortar. From Fig. 7b it is observed that
an increase on MPT led to an increase of spread on table,
as expected. To achieve the stipulated spread of 240±10
mm, mortars with different cements had a similar value of
MPT: for mortars without superplasticizer (empty icons)
the MPT varied from 3.6 to 3.7 mm and for mortars with
superplasticizer (lled icons) this variation stayed between
2.8 to 3 mm. These results demonstrated that the MPT can
be a good tool to predict the water demand for each cement
to achieve the specic spread on table. Since for MPT
calculation is only necessary the physical characterization
of the raw materials, the laboratory work can be reduced.
However, a bigger data to verify this is necessary; it is also
important to know that in this case it was used a standard
sand and same mortar composition and mixture procedure
to produce all mortars.
Compressive strength and porosity: Fig. 8 shows the
results of compressive strength obtained for 7, 28 and 91
days of hydration. It was observed that the cements REF
and 16DE had a greater evolution of strength in the rst 7
days than 49DE. The last, a cement with high addition of
diatomaceous earth, presented slower hydration kinetics due
to its lower content of clinker phases and showed a higher
increase on compressive strength between 28 and 91 days of
hydration because of the pozzolanic activity of diatomaceous
earth. These results agreed with the reactivity evaluation.
It was also clear the benet of using the superplasticizer
that reduced the mixing water, reducing the porosity and
consequently increasing the compressive strength. For
mortars produced with REF, 16DE and 49DE, at 7 days,
there was an increase of 46%, 50% and 67% on compressive
strength by adding superplasticizer and reducing the mixing
water, and at 91 days of hydration, this increase changed to
36%, 40%, and 53%, respectively.
The compressive strength of brittle materials, like
ceramics and cementitious materials, have an exponential
relation with porosity [37, 38]. Since the cements have a
different reactivity and water demand, they inuence on
mortar total porosity and consequently on compressive
strength. Fig. 9 shows three groups of results, each group
corresponding to a specic cement; inside each group it
can be observed: i) the effect of superplasticizer addition,
difference between empty and lled icons; and ii) the
inuence of hydration process, difference between icons
that goes from right (7 days) to left (91 days). Taking an
example, for cement 49DE the addition of superplasticizer
led to a decrease of 5.8% on total porosity of mortar at 7
days; this decrease on porosity was linked with reduction of
P. C. R. A. Abrão et al. / Cerâmica 65 (2019) Suppl.1 75-86
Figure 7: Spread on table versus water/cement volume ratio
obtained from the ow table test (a) and maximum paste thickness
– MPT (b) for mortars without (empty icons) and with (lled icons)
superplasticizer. The purple band corresponds to the optimum
ow (240±10 mm) dened in this study as the ‘xed rheological
parameter’.
320
300
300
340
380
280
240
200
160
260
260
220
220
180
180
140
2.0
1.1 1.51.3 1.71.2 1.61.4 1.8
MPT (µm)
Vwater/Vcement
Spread on table (mm) Spread on table (mm)
3.0 4.02.5 3.5 4.5 5.0
a)
b)
Figure 8: Compressive strength of the mortar samples without
(noSP) and with (SP) incorporation of superplasticizer at 7, 28 and
91 days of hydration, for the same spread on table (240±10 mm).
noSP
80
50
20
70
40
10
60
30
0
REF
Compressive strength (MPa)
16DE 49DE
noSP noSPSP SP SP

83
mix water caused by the dispersant addition. An example on
the inuence of hydration process can be observed for 49DE
(without superplasticizer): there was a reduction of 4.5% on
total porosity between mortars at 7 and 91 days of hydration;
over time materials can chemically combine more water
and thus forming a larger volume of hydrated products,
obtaining a lower porosity. The three groups of results,
each one corresponding to a type of cement, had a different
relation between total porosity and compressive strength.
This fact was related to the different formed microstructure
in the nal product, since it depends on the chemical and
mineralogical composition of cements, physical properties
of raw materials and mix water.
Fig. 10 shows the results of relative mechanical strength
of the mortars prepared with the blended cements in relation
to REF. Mortars with and without dispersant produced
with 16DE cement stayed below the dilution line, which
represents that the reduction in mechanical strength at all
ages was higher than the reduction of clinker content when
compared to REF. For 49DE, the replacement of clinker
by diatomaceous earth was more effective, taking into
consideration the mechanical strength. Mortar without
dispersant at 91 days had a reduction on compressive
strength of 36% and clinker substitution of 43% compared
to REF, and with the addition of superplasticizer, there was
a reduction on compressive strength of 27% for mortar at 91
days compared to REF.
Binder intensity, carbon intensity, and CO2 emissions:
Fig. 11a shows the results of binder intensity (BI) for
modeled concretes; this indicator relates the amount of
binder in an m³ of concrete necessary to provide 1 MPa of
compressive strength. The grey curves indicate constant
amounts of binder in weight per m³ of concrete; in a best
practice scenario concretes with 250 kg of binder per m³ are
at the ‘state of the art’. From Fig. 11a it can be concluded
that binder intensity depended on curing age and the
incorporation of superplasticizer, in other words, it depended
on reactivity and quantity of mixing water. Concretes with
Portland pozzolan blended cements presented higher binder
intensity compared with REF, for concrete with and without
superplasticizer. The cements with diatomaceous earth
combined less water in their hydration process than REF, so it
was necessary more binder to achieve 1 MPa of compressive
strength. Concretes without superplasticizer prepared with
49DE and 16DE cements at 28 days of hydration needed 6.12
and 1.92 kg more binder to obtain 1 MPa when compared
to REF; with the addition of superplasticizer these numbers
reduced to 3.62 and 1.79 kg, respectively; these values are
expressive when the analysis is scaled up to the volume of
concrete used nowadays.
Fig. 11b presents the estimated results of carbon intensity
(CI) for the studied concretes; the grey curves represent
constant amounts of CO2 emitted to produce one m³ of
concrete. It was also clear that the carbon intensity indicator
depended on clinker content, reactivity and mixing water;
materials that chemically combine more water and require
a smaller volume of mixing water produce concretes with
P. C. R. A. Abrão et al. / Cerâmica 65 (2019) Suppl.1 75-86
Figure 9: Compressive strength versus mortar porosity calculated
from Power’s model for the systems without (empty icon) and with
(lled icon) superplasticizer incorporation. There are three icons
for each composition: the rst icon (from right to left) corresponds
to 7 days of hydration, second to 28 days and third to 91 days.
4
90
80
40
70
30
60
20
50
10
Mortar total prorosity (%)
Compressive strength (MPa)
12 208 16 24
Figure 10: Relative compressive strength (fc) - REF versus clinker substitution fraction on mortars without (a) and with (b) superplasticizer.
The empty icons correspond to 7 days of hydration, pointed icons to 28 days and lled icons to 91 days. The black line corresponds to a
direct dilution.
1.0 1.0
0.9 0.9
0.8 0.8
0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
10 10
Clinker substitution (%) Clinker substitution (%)
Relative fc (MPa/MPa)
Relative fc (MPa/MPa)
30 30
50 50
20 20
40 40
60 60
a)

84
lower carbon intensity indices. Concretes with 49DE without
superplasticizer at 28 and 91 days presented reductions of
4% and 20% on carbon intensity when compared to REF;
on the other hand, concretes with 16DE had an increase of
7% and 5%, respectively. By incorporating superplasticizer,
concretes with 49DE at 28 and 91 days showed reductions
of 18% and 27% on carbon intensity indicator when
compared to REF. However, 16DE concrete had increases
of 9% and 5%, respectively. Even though 49DE had a lower
reactivity than REF, the best results on carbon intensity
indicator was linked with the high replacement of clinker
(53%), which was the material with most impact on CO2
emissions of concrete; in this case the reduction on clinker
content was accompanied by a reduction on carbon intensity,
but for 16DE this did not happen. Concretes made with
16DE with or without superplasticizer led to an increase of
carbon intensity when compared to REF, and the reduction
on clinker content was not compensated by the pozzolan,
resulting in higher carbon intensity.
Fig. 12 shows the carbon intensity of modeled concretes
versus the CO2 emitted on the production process of each
cement. Each cement had an only value of CO2 emission factor
that depends on clinker content, but had many values of carbon
intensity, that depend on clinker content, reactivity (curing
age) and water demand (with or without superplasticizer).
Concretes with 49DE, 16DE and REF at 28 days of hydration
without superplasticizer had a carbon intensity indicator of
6.16, 6.88 and 6.40 kgCO2.m-³.MPa-1, respectively, but a CO2
emission factor of 399, 622 and 774 kgCO2/t cement; in this
scenario analyzing the carbon intensity, 16DE would be the
cement with higher environmental impact, but analyzing
the CO2 emission factor REF would be by far the material
with higher environmental impact. Thus, there was no
direct relation between both indicators, showing that the
reduction of CO2 emissions in the production of cement is
not always associated with a reduction of the impacts of
cement in its use. It is also important to mention that the
correct choice of cement based on BI and CI depends on the
concrete compressive strength class required. For instance,
applications that require concrete with high compressive
strength (above 60 MPa), it would be advised to use cement
REF with superplasticizer, that presents lower BI and CI
than the Portland pozzolan blended cements. However, if
a concrete with 40 MPa is needed, the best choice would
be cement 49DE with superplasticizer; so, the correct user’s
choice is also important for reduction of environmental
impact of cement and concrete industry.
CONCLUSIONS
The SEM images showed that kaolin plates were spread
on the surface of 49DE and 16DE grains, increasing their
specic surface area. It was also observed a cluster of kaolin
plates. Cements with diatomaceous earth required more
superplasticizer than the cement with a high content of
clinker (REF) to obtain the same rheological behavior, due
to their higher specic surface area. 16DE cement presented
Figure 11: Estimated binder (a) and carbon (b) intensity indicator
of concretes containing a xed paste volume of 300 dm³ without
(empty icons) and with (lled icons) superplasticizer. There are
two icons for each composition, the rst icon, from right to left,
corresponds to 91 days of hydration and the second to 28 days. The
grey lines represent concretes with 250, 500 and 1000 kg of binder/
CO2 per m³ of concrete.
16
12
8
8
4
4
14
10
6
6
2
2
0
0
1
3
5
7
10
Compressive strength (MPa)
Cl300 (CO2kg.m-3.MPa-1)Bl300 (kg.m-3.MPa-1)
40300 6020 50 70 80
Figure 12: Carbon intensity indicator versus CO2 emissions. Carbon
intensity was calculated for concretes without (empty icons) and
with (lled icons - sp) superplasticizer. There are two icons for
each composition, the rst icon (from bottom to top) corresponds
to 91 days of hydration and the second to 28 days.
8
4
6
2
0
1
3
5
7
CO2 emissions (kgCO2/t cement)
Cl300 (CO2kg.m-3.MPa-1)
300 500400 700 800600
P. C. R. A. Abrão et al. / Cerâmica 65 (2019) Suppl.1 75-86

85
a reduction on volume of chemically combined water at 91
days, higher than the reduction on its quantity of clinker, but
49DE proved to be more efcient in this regard, combining
proportionately a higher volume of water compared to REF,
indicating a contribution of the diatomaceous earth in the
reactivity of cement. In mortars without incorporation of
superplasticizer, all cements required a similar water volume
to obtain a spread on table of 240±10 mm, but with the
incorporation of superplasticizer, Portland pozzolan blended
cements required a smaller volume of water to achieve the
same rheological behavior. These two parameters, reactivity
and water demand, directly inuenced the compressive
strength and porosity of the nal product; for 16DE the
reduction on compressive strength was higher than the
reduction on its clinker content; on the other hand, 49DE
obtained an increase in the compressive strength relative to
REF for the ages of 28 and 91 days. The porosity of the
mortars was inuenced by the addition of superplasticizer
and hydration time. The binder intensity was greater for
cements with diatomaceous earth, an indicator that relates
the binder efciency and the use of materials. For carbon
intensity, only 49DE cement presented a reduction of this
indicator when compared with REF, showing that the
reduction in CO2 emissions of cement production (reduction
of the clinker content) did not have a direct relationship with
a reduction on the impacts of cement in its use. Therefore,
for the strategy of substituting clinker with supplementary
cementitious materials to be effective in reducing the
environmental impacts of the cement industry, it is not
possible to evaluate only the CO2 emission factor (clinker
factor), but the analysis should be more comprehensive
involving the reactivity and water demand of these cements.
ACKNOWLEDGMENTS
This research was supported by CNPq - Conselho
Nacional de Desenvolvimento Cientíco e Tecnológico
- Brazil (Process 485340/2013-5) and FAPESP (Process
14/50948-3 INCT/2014); the authors also thank the support
of InterCement. The information presented in this study are
those of the authors and do not necessary reect the opinion
of CNPQ, FAPESP or InterCement.
REFERENCES
[1] CEMBUREAU, “Activity report 2015”, Brussels (2016).
[2] K.L. Scrivener, V.M. John, E.M. Gartner, “Eco-efcient
cements: potential, economically viable solutions for a low-
CO2, cement-based materials industry”, UNEP (2016).
[3] R. Snellings, G. Mertens, J. Elsen, Rev. Mineral.
Geochem. 74, 1 (2012) 211.
[4] R.C. Mielenz, Concr. Int. 5, 8 (1983) 34.
[5] R.O.Y. Breese, in “Industrial minerals and rocks”, 6th ed.,
1, Soc. Min. Metall. Explor., Littleton (1994).
[6] F.M. Lea, The chemistry of cement and concrete, 3rd ed.,
Edward Arnold, London (1970).
[7] F. Massazza, Cem. Concr. Compos. 15, 4 (1993) 185.
[8] M.C.G. Juenger, R. Siddique, Cem. Concr. Res. 78
(2015) 71.
[9] E. Benhelal, G. Zahedi, E. Shamsaei, A. Bahadori, J.
Clean. Prod. 51 (2013) 142.
[10] K. Gäbel, A.-M. Tillman, J. Clean. Prod. 13, 13 (2005)
1246.
[11] E.M.R. Fairbairn, B.B. Americano, G.C. Cordeiro,
T.P. Paula, R.D. Toledo Filho, M.M. Silvoso, J. Environ.
Manage. 91, 9 (2010) 1864.
[12] D.J.M. Flower, J.G. Sanjayan, Int. J. Life Cycle Assess.
12, 5 (2007) 282.
[13] P. Van den Heede, N. De Belie, Cem. Concr. Compos.
34, 4 (2012) 431.
[14] M. De Schepper, P. Van den Heede, I. Van Driessche, N.
De Belie, Materials 7, 8 (2014) 6010.
[15] J. Turk, Z. Cotič, A. Mladenovič, A. Šajna, Waste
Manag. 45 (2015) 194.
[16] B.L. Damineli, F.M. Kemeid, P.S. Aguiar, V.M. John,
Cem. Concr. Compos. 32, 8 (2010) 555.
[17] Ass. Bras. Norm. Técn., NBR 5736, “Cimento Portland
pozolânico”, Rio Janeiro (1991).
[18] Ass. Bras. Norm. Técn., NBR 11578, “Cimento Portland
composto”, Rio Janeiro (1991).
[19] L.A. Gobbo, “Aplicação da difração de raios-X e
método de Rietveld no estudo de cimento Portland”, Tese
Dr., Un. S. Paulo (2009).
[20] G.P. Souza, M. Filgueira, R. Rosenthal, J.N.F. Holanda,
Cerâmica 49 (2003) 40.
[21] M. Hunger, H.J.H. Brouwers, Cem. Concr. Compos. 31,
1 (2009) 39.
[22] A. Schöler, B. Lothenbach, F. Winnefeld, M. Zajac,
Cem. Concr. Compos. 55 (2015) 374.
[23] Ass. Bras. Norm. Técn., NBR 7214, “Areia normal para
ensaio de cimento - especicação” (2015).
[24] Ass. Bras. Norm. Técn., NBR 7215, “Cimento Portland:
determinação da resistência à compressão” (1996).
[25] BS EN 196-1, “Methods of testing cement -
determination of strength” (2005).
[26] Ass. Bras. Norm. Técn., NBR 13278, “Argamassa
para assentamento de paredes e revestimento de paredes e
tetos - determinação da densidade de massa e do teor de ar
incorporado” (1995).
[27] Ass. Bras. Norm. Técn., NBR 13276 “Argamassa para
assentamento e revestimento de paredes e tetos - preparo da
mistura e determinação do índice de consistência” (2005).
[28] Ass. Bras. Norm. Técn., NBR 7215 “Cimento Portland
- determinação da resistência à compressão” (1996).
[29] ASTM, C 109/C 109M-02, “Method for compressive
strength of hydraulic cement mortars” (2002).
[30] T.C. Powers, T.L. Brownyard, J. Am. Concr. Inst. 43, 9
(1947) 101.
[31] P. Livesey, Constr. Build. Mater. 5, 3 (1991) 147.
[32] WBCSD, “Getting the numbers right, project emissions
report 2014” (2016).
[33] G. Habert, N. Roussel, Cem. Concr. Compos. 31, 6
(2009) 397.
[34] M. Adjoudj, K. Ezziane, E.H. Kadri, T.-T. Ngo, A. Kaci,
P. C. R. A. Abrão et al. / Cerâmica 65 (2019) Suppl.1 75-86

86
Constr. Build. Mater. 70 (2014) 549.
[35] B. Lothenbach, K. Scrivener, R.D. Hooton, Cem.
Concr. Res. 41, 12 (2011) 1244.
[36] I. de Oliveira, A.R. Studart, R.G. Pileggi, V.C.
Pandolfelli, Dispersão e empacotamento de partículas:
princípios e aplicações em processamento cerâmico,
Fazendo Arte Editor. (2000) 119.
[37] K. Kendall, A.J. Howard, J.D. Birchall, Phil. Trans. R.
Soc. Lond. A 310, 1511 (1983) 139.
[38] E.B.C. Costa, F.A. Cardoso, V.M. John, Constr. Build.
Mater. 156 (2017) 1114.
(Rec. 30/08/2018, Ac. 20/11/2018)
P. C. R. A. Abrão et al. / Cerâmica 65 (2019) Suppl.1 75-86