Construction and Building Materials 140 (2017) 374–384

Contents lists available at ScienceDirect

Construction and Building Materials

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

Microstructure and hardened state properties on pozzolan-containing

concrete
H. Paiva a,⇑, A.S. Silva b, A. Velosa c, P. Cachim c, V.M. Ferreira a,c
a
Civil Engineering Dept./CICECO, University of Aveiro, 3810-193 Aveiro, Portugal
b
LNEC – Materials Dept., National Laboratory for Civil Engineering, Av. Brasil, 101, 1700-066 Lisboa, Portugal
c
Civil Engineering Dept./RISCO, University of Aveiro, 3810-193 Aveiro, Portugal

h i g h l i g h t s

 SEM microstructure analysis as a tool to evaluate pozzolan’s effects in concrete.

 Effect of two chemically and physically different pozzolans in concrete behaviour.
 Comparison of concrete rheology controlled by water or a water reducing admixture.
 Impact of type of workability control over microstructure and hardened state features.

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

Article history: The homogeneous state of the concrete in the fresh state has consequences on their properties in the
Received 27 September 2016 hardened state. The use of pozzolanic concrete must be carefully evaluated, because the physical and
Received in revised form 21 February 2017 chemical differences of each pozzolan have consequences on concrete hardened state properties. In this
Accepted 22 February 2017
work, two physical and chemically different pozzolans were used to evaluate the effect on concrete
Available online 6 March 2017
microstructure and its basic hardened state properties. Scanning electron microscopy (SEM) was used
and can be an effective tool in the analysis of the microstructure allowing a better understanding of
Keywords:
the properties of concrete in hardened state. It allows showing that the differences in the two pozzolans
Concrete
Metakaolin
(metakaolin and diatomite) promote different microstructures as well as different final properties of poz-
Diatomite zolanic concrete. Moreover, it was also shown that the use of a water reducing agent, to adjust workabil-
Microstructure ity instead of water, promotes an improvement in both the microstructure and the hardened state
SEM/EDS features such as porosity and mechanical strength.
Porosity Ó 2017 Elsevier Ltd. All rights reserved.
Reactivity
Admixtures

1. Introduction
It is widely accepted the use of pozzolanic materials such as fly
ash and microsilica as partial replacement of Portland cement to
produce mortars and concretes [1]. Pozzolans are natural or artifi-
cial fine materials which can partially replace cement in mortars
and concretes. The main component of the pozzolan is amorphous
silica which, in the presence of water, reacts with calcium hydrox-
ide (Ca(OH)2), forming compounds with cementitious properties.
Some pozzolanic materials have also amorphous alumina in their
composition, which can be very reactive, producing aluminosili-
cates such as in cements. The effectiveness of a pozzolan depends
on its pozzolanic reactivity and this concept may include two fac-
⇑ Corresponding author.
E-mail address: hpaiva@ua.pt (H. Paiva).
tors, namely, the maximum amount of calcium hydroxide that can
react with the pozzolan and the speed at which the pozzolanic
reaction occurs [2]. However, beyond the pozzolanic reactivity,
these materials, being very fine, can also have a ‘‘filler” effect, pro-
moting a decrease in the total porosity of the system due to the fill-
ing of voids and capillary pores. The packing mechanism is greatly
enhanced by the addition of very fine particles leading to densifica-
tion. It was also demonstrated, using a simulation system, that the
finer particles tend to concentrate near the transition zone of the
cementitious matrix interface and the aggregates (IZT), leading to
a reduction in the porosity and increasing the internal cohesion,
since there is an increased action of van der Waals forces between
the particles due to the increased proximity between them [3].
Besides silica fume and fly ash, metakaolin (MK) has been also
studied as a pozzolanic material. The metakaolin is usually
obtained from the calcination of kaolinite-rich clays at tempera-

http://dx.doi.org/10.1016/j.conbuildmat.2017.02.120
0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.
 H. Paiva et al. / Construction and Building Materials 140 (2017) 374–384
tures ranging between 600 and 800 °C, depending on the charac-
teristics of the raw material. However, some authors [4] consider
that the optimum calcination temperature should be always deter-
mined in order to obtain maximum strength. Moses and Cabrera
[5] have studied the distribution of the pore diameter and the
degree of hydration of cement pastes with metakaolin content
(10, 15, 20 and 25%) for a water/cement ratio of 0,55 and observed
a decrease in porosity for 28 days. For longer ages, the total poros-
ity increases when compared with formulations without the poz-
zolan. Observing the works that have been done on the use of
metakaolin in cement-based materials, it appears that the use of
this pozzolanic material improves the behaviour of mortars and
concretes. This beneficial effect is mainly due to the high poz-
zolanic reactivity of metakaolin that, on one side, reacts rapidly
and extensively with calcium hydroxide resulting from the hydra-
tion process of cement and, on the other hand, accelerates the
hydration reaction of cement. The calcium silicate hydrate formed
is in the form of a gel that penetrates the pores, promoting the
decrease of the average pore size and increasing the number of
smaller pores. This effect is also observed in the interfacial transi-
tion zone (ITZ) between the binder and the aggregate, promoting
densification. The refinement of the pores and densification of
the interfacial transition zone can justify the improvement in
mechanical strength, the decreased water absorption by capillarity,
improved resistance to chemical attack and the increased durabil-
ity [6].
Diatomite (D) is a natural material from a sedimentary rock,
formed mainly by the deposition of microscopic organisms with
a crystalline and amorphous silica shell [7]. Diatomite was used
as a pozzolan, directly after milling or after a heat treatment at
about 1000 °C followed by grinding. Subjecting this material to
thermal treatment shows an increase in the content of amorphous
silica and grinding reduces it particle size. The hydraulic and poz-
zolanic activity of silica mineral additions depends on its crys-
talline or amorphous state as well as on its size and particle
shape [8]. The study of diatomite as partial replacement of cement
is still recent but some authors have shown that its use in the pro-
duction of mortar and concrete does not improve its mechanical
properties, due to its nature and low pozzolanic reactivity [7,9].
A disadvantage of the use of diatomite in the production of mortar
and concrete is the need for a very high amount of water which
promotes a simultaneous reduction in the mechanical strength
[10,11].
However, Ergun [12] has also studied the effect of diatomite on
the mechanical properties of the concrete where cement was
replaced by diatomite (5% to 10%), maintaining the water constant
and using a constant amount of a water reducing agent. The com-
pressive strength of the concrete increased with diatomite. Ergun
[12] considered that this behaviour was due to the combination
of this pozzolan (containing highly reactive amorphous silica) with
a water reducing admixture which allows the use of lower
amounts of mixing water.
Scanning electron microscopy (SEM) is recognized as a good
tool to evaluate the microstructure of materials, including mortars
and concretes. Some authors [13–15] have studied the interfacial
transition zone (IZT) between the cement matrix and aggregates
by SEM, for instance, Rossignolo [14] has used SEM to determine
the thickness of the IZT. Diamond [15] studied the effect of a water
reducing agent in the concrete microstructure and found that con-
cretes containing this admixture are more homogeneous and with-
out agglomerates that usually promote density variations in the
material. Oertel et al. [16] have also analysed particle and agglom-
erates size of silica fume with SEM and its results indicate that, if
particles agglomeration is prevented, a higher mechanical strength
will result in the concrete.
375
The aim of this work was to use the microstructure analysis by
SEM to better understand the hardened state characteristics of
concrete containing two physically and chemically different poz-
zolans, specifically, a metakaolin and a diatomite.
2. Materials and methods
In this study, the formulation for the reference concrete mixture
was obtained using the Faury method [17]. The standard or refer-
ence concrete (named B_0,6W) constitution involves Portland
cement (CEM type I 42.5R) as a binder, a siliceous natural sand
and two types of crushed limestone as coarser aggregates [18].
Metakaolin (MK) was used as a partial cement substitute mate-
rial in contents of 10, 20 and 30% (wt%). This material is a dehy-
droxylated aluminium silicate, with a general formula of
Al2O3.2SiO2. It is an amorphous non-crystallized material consti-
tuted of lamellar particles. This MK presents a pozzolanic index
(measured by the modified Chapelle test according to NF P18-
513:2010) of 1100 mg Ca(OH)2/g of metakaolin and a specific sur-
face area (BET) of 17 m2/g [19,20].
Diatomite (D) was used as a partial cement substitute in con-
tents of 3, 5, 8 and 10% (wt%). This diatomite, previously calcined
and grinded, presents a specific surface area (BET) of 9 m2/g. The
Chapelle test was not conducted in diatomite since its residence
time in contact with the calcium hydroxide (Ca(OH)2) is not
enough to obtain reliable results. Alternatively, a comparative
study adapted for these two pozzolans to evaluate their reactivity
was performed. This test compares the compressive strength of
lime based mortars with metakaolin or with diatomite, based on
the pozzolan reaction with hydrated lime. The choice of lime as a
binder was made to ensure the availability of calcium hydroxide
to react with pozzolans. This method consisted of producing a ref-
erence lime based mortar whose composition is 80% of sand used
in the study and 20% of hydrated lime (with 90% Ca(OH)2). In the
pozzolan-containing mortars, lime was replaced with 50% of diato-
mite or metakaolin.
Table 1 presents the results obtained for the pozzolanic charac-
terization of the employed pozzolans (metakaolin and diatomite).
The compressive strength (Rc) of the base lime mortar and the lime
mortars with metakaolin or diatomite at different curing times (at
20 ± 5 °C and 60 ± 5% RH) confirms that the metakaolin have higher
pozzolanic reactivity.
Yilmaz and Ediz [10] showed with SEM that diatomite is very
porous and does not show a very high surface area even after cal-
cination and grinding. It was considered important to determine
the diatomite water absorption. An expedite method (an adapta-
tion of the EN 1097-6 standard) was developed where a diatomite
sample is diluted in water and then passes through a filter that
retains particles larger than 8 lm. When all the free water has
passed through the filter, the material filled of water is scraped
with a spatula and weighed (Fig. 1). The difference between the
weight of wet material and dry material, relative to the dry mate-
rial weight, is the percentage of water absorption of diatomite The
experiment was repeated 10 times for two different amounts of
sample (1 and 5 g). The average value of water absorption pre-
sented by the diatomite was higher than 96%.
Table 1
Values of compressive strength (Rc) for the pozzolanic evaluation of metakaolin and
diatomite.
Composition Rc (MPa) Rc (MPa) Rc (MPa)
3 days 7 days 28 days
Base mortar 0,29 0,29 0,47
Mortar with MK 1,01 2,99 3,73
Mortar with D 0,41 0,62 1,08
376 H. Paiva et al. / Construction and Building Materials 140 (2017) 374–384

Fig. 1. Determination of diatomite water absorption.

Table 2
Cement (Cem), metakaolin (MK) and diatomite (D) chemical analysis.

(%) SiO2 Al2O3 Fe2O3 MnO MgO CaO Na2O K2O TiO2 P2O5 LoI*
Cem 22,23 5,70 3,50 0,09 1,13 63,30 n.d. 0,86 0,28 0,10 3,13
MK 52,17 44,50 0,45 nd nd 0,01 nd 0,15 1,42 0,12 1,42
D 81,71 0,56 0,31 0,01 0,14 7,29 nd 0,08 0,05 0,06 10,16
*
LoI – Loss of ignition. nd: not determined.

Table 3 Further characterization of raw materials has involved X-ray

Composition of studied concrete samples. diffraction (XRD) and X-ray Fluorescence (XRF) of cement, meta-
Composition Pozzolan Water/ HRWRA content kaolin and diatomite.
(%) binder as wt% of total solids Table 2 presents the metakaolin and diatomite chemical analy-
(%) sis obtained by XRF. The metakaolin presents, as major compo-
Mortars nents SiO2 and Al2O3, while the diatomite presents basically SiO2
Standard A_0,6W 0 0,6 0 and some CaO [18]. This chemical difference is one of the reasons
C_10MK_0,65W 10MK 0.65 0 why the metakaolin have higher pozzolanic reactivity since alu-
C_20MK_0,70W 20MK 0,70 0
C_30MK_0,76W 30MK 0,76 0
mina is very reactive.
C_3D_0,65W 3D 0,65 0 The X-ray diffraction (XRD) was also performed and, in case of
C_5D_0,67W 5D 0,67 0 cement, this material presents as major elements, tricalcium sili-
C_8D_0,70W 8D 0,70 0 cate (C3S) and dicalcium silicate (C2S). As minor phases tricalcium
C_10D_0,73W 10D 0,73 0
aluminate (C3A) and tetracalcium aluminoferrate (C4AF) were
C_10MK_0,60W_0,10WR 5MK 0,6 0,10
C_20MK_0,60W_0,15WR 20MK 0,6 0,15 detected. The other XRD peaks are related to anhydrous com-
C_30MK_0,60W_0,25WR 30MK 0,6 0,25 pounds and gypsum present in the composition of the Portland
C_3D_0,60W_0,06WR 3D 0,6 0,06 cement used (type I) [18]. The diffraction pattern of metakaolin
C_5D_0,60W_0,08WR 5D 0,6 0,08 is essentially amorphous but with the presence of some crystalline
C_8D_0,60W_0,10WR 8D 0,6 0,10
C_10D_0,60W_0,12WR 10D 0,6 0,12
silica, and a clay mineral, muscovite, which is an aluminosilicate.

Fig. 2. SEM/BEI of base concrete (S: sand, B: crushed limestone; C: cement, P: pore). Fig. 3. SEM/BEI of base concrete with demarcated area for EDS analysis.
 H. Paiva et al. / Construction and Building Materials 140 (2017) 374–384 377

Regarding XRD analysis of the diatomite, it was essentially amor- 3. Results and discussion
phous but with some crystalline peaks of quartz, calcite and arag-
onite [18]. Pozzolanic fine particles systems tend to agglomerate and this
The concrete samples in this study (listed in Table 3) were ref- may lead to changes in the fresh and hardened state properties
erenced as C_xP_yW_zWR, in which C means concrete, x is the of mortars and concretes [21]. There is an effect on the pozzolanic
pozzolan content in wt% of cement (MK or D), y is the water/binder concrete workability caused by a system thickening shown by the
ratio and z is the amount of water reducer admixture (HRWRA) in
wt% of total solids. The binder total content is considered as the
weight of the cement and pozzolan together. The water-reducing
admixture (WR) is based on a polycarboxylic acid, with a density
between 0,67 and 1,1 and solid contents between 28,5 and 31,5%
[18].
In order to assess the basic features of the resulting concretes
containing these pozzolanic additions, fresh and hardened state
characteristics were measured and compared with the reference
or base concrete.
The concrete workability was controlled by slump determined
by the Abrams cone test. The desired slump set up for this study
was kept inside the range of 8 to 10 cm, in order to produce con-
cretes with very similar workability for comparison purposes.
The use of an identical workability is important because differ-
ences in hardened state samples properties can only be due to
the components of each composition. Mechanical compressive
strength was measured in samples cured for 28 days at 20 ± 5 °C
and 95 ± 5% RH, according to EN 1015-11:1999. Base Concrete
The study of concrete microstructure was performed using a
scanning electron microscope (SEM) (JEOL JSM6400) and the
observation mode of the samples was through electron backscat-
tered imaging (BEI) at 15 keV. Supplementary information on their
chemical composition was obtained, when needed, with energy
dispersive spectroscopy (EDS) (Inca EDS X-Sight system, Oxford
Instruments) with the same electron beam voltage. Due to their
porosity, samples had to be impregnated with an epoxy resin
and then subjected to successive polishing to obtain a polished
but cohesive surface for microstructure observation and analysis.
Porosity measurements were also carried out on hardened sam-
ples by mercury intrusion porosimetry (MIP), which allows assess-
ing not only the total porosity but also the pore size distribution.
The samples preparation involved the separation of the mortar part
from the coarse aggregate in order that only the mortar part was
analysed.
Thermal analysis (differential and thermogravimetric analysis)
was also used to calculate the consumption ratio of Ca(OH)2 in
the concrete due to the metakaolin (MK) and diatomite (D) intro- 20% MK
duction. Small portions of various locations were removed from
the concrete sample, disaggregated together and passed through
a 63 lm sieve. Only the cementitious matrix was analysed. Fig. 5. SEM/BEI of base and 20% MK concretes. Base concrete above and 20% MK
Table 3 presents the compositions of studied concrete samples. below.

Fig. 4. EDS spectrum of the base concrete cement past formulation.

378 H. Paiva et al. / Construction and Building Materials 140 (2017) 374–384
Fig. 6. Large MK agglomerates indicated by square marks.
Fig. 7. SEM/BEI of concrete with 30% MK, where metakaolin agglomerates are
visible.
increase in plastic viscosity and yield stress when there is an
increase in pozzolan content [21]. Although pozzolans promote
loss of workability, this study was conducted using all formula-
Fig. 8. EDS spectrum of metakaolin agglomerate.
tions with the same workability (slump). To tune workability and
keep it in close values, adjustments were made either through
the water content or the use of a water reducing agent (WRA).
The scanning electron microscope coupled with an energy dis-
persive system (SEM/EDS) can quickly provide information about
microstructural features such as morphology but also on chemical
composition of the components in the sample. All samples anal-
ysed by SEM/EDS were subjected to 28 days of curing in a con-
trolled environment. Fig. 2 shows a backscattered electron image
(BSE) obtained at SEM of the reference concrete sample. In this
SEM mode the images obtained are sensitive to the atomic mass
of the nuclei they scatter from. As a result, heavier elements appear
brighter than lighter elements in a SEM/BEI.
In Fig. 2, B1 is the coarse aggregate (crushed limestone), B2 is a
smaller crushed limestone, S is the sand, C the cement paste and P
is porosity. At this magnification level porosity, resulting from par-
ticle packing, appears black with a rounded shape. It was found
that there are zones where the bonding of the cement with the
aggregate matrix is good but, nevertheless, after 28 days, a large
amount of not hydrated cement grains (white particles in the
matrix) are still observed.
In Fig. 3, with a higher magnification, it is possible to observe a
darker halo around the cement particles, corresponding to cement
hydrate. It also shows a squared area where EDS analysis was per-
formed, allowing determination of the elemental composition of
the cement paste, which can be seen in Fig. 4 that shows that the
major elements are calcium and silicium. Elements such as oxygen,
carbon and alumina have also representative quantities. The pres-
ence of oxygen and carbon is due to the fact that the material pre-
sents silica (SiO2) and calcium carbonate (CaCO3). The minor
elements are potassium, iron, magnesium and titanium.
Microstructures of concrete samples with metakaolin observed
by SEM are presented in Figs. 5–9. Fig. 5 shows SEM/BEI images of
the base concrete with a 20% MK concrete, at the same magnifica-
tion and workability adjustment with water.
In comparison, after 28 days of curing, the base concrete has a
higher cement content that is still not hydrated. This fact can be
observed by the amount and size of whiter particles in the matrix
(non-hydrated cement grains) in both samples. However, the base
concrete sample shows an aggregate matrix interface much denser
than the concrete sample with 20% MK.
In Fig. 6 it is possible to observe metakaolin agglomerates of
considerable dimensions (signalled by squares) in the concrete
sample with 20% of MK.
Fig. 7 shows one of these metakaolin agglomerates in the sam-
ple of concrete with 30% MK. One of those agglomerates, about
 H. Paiva et al. / Construction and Building Materials 140 (2017) 374–384 379

50 lm long, is identified as ‘‘MK” inside the squared zone. EDS This is explained by the fact that the cement hydration and the
analysis in this agglomerate (Fig. 8) showed that major elements pozzolanic reactions are carried out within the capillary pores pro-
were aluminium and silicium, confirming the metakaolin elemen- moting the matrix densification. If this reaction is lower, such as
tal composition, determined previously by XRF. with diatomite, the porosity will be higher. Table 5 shows that
Fig. 9 shows the X-ray maps of the aluminium, silicium and cal- metakaolin presents higher calcium hydroxide consumption when
cium elements obtained by SEM/EDS, performed within the compared with the one presented by the diatomite. Thus, it is nor-
squared marked zone of Fig. 7. There are zones of high concentra-
tion of aluminium corresponding to metakaolin agglomerates. The
remaining aluminium is widespread on the cementitious matrix
and derives from the cement.
The silicium aggregates are easy to detect and it can be con-
cluded that they are related to sand (quartz). Regarding metakaolin
agglomerates, only the larger ones are possible to be observed
clearly. Finally, in the calcium map in the upper region, there is a
high concentration of calcium that should correspond to the coarse
aggregate (limestone) and, in the lower and central areas, higher
concentrations appear due to cement grain particles. The remain-
ing calcium is spread by the cementitious matrix, possibly in the
form of calcium hydroxide (Ca(OH)2) or associated with the hydra-
tion products of the cement.
Fig. 10 shows the EDS spectrum corresponding to this marked
region. As, expected the major elements are silicium and calcium,
which come mainly from sand and crushed limestone but also
from the cement. The presence of aluminium comes from the
cement and metakaolin. The considerable amount of carbon and
oxygen together with silica and calcium is related to alumina
(Al2O3) and calcium carbonate (CaCO3).
When pozzolanic materials are used, several studies
[15,16,18,22] confirm the need for a larger amount of water in
order to obtain the same consistency and improved compression
strength at least up to 10% diatomite content .
The microstructure of concrete with introduction of diatomite is
shown in Figs. 11 and 12. They show that the concrete sample with
8% diatomite presents higher porosity than the sample with 20% of
MK. The particles marked in the Fig. 11 top photo are MK agglom-
erates. Fig. 11 shows clearly the higher porosity of cementitious
matrix in concrete with 8% of diatomite, due to the increase of
water content that promotes a higher porosity and probably due
to the irregular shape of diatomite particles that leads to worst
packing and does not contribute to densification. Fig. 12 also shows
diatomite particles (example inside the circle) that behave as por-
ous coarse aggregates leading to a less-connected matrix. As the
interfacial zone between the matrix and aggregate particles are
connected by long crystals (Portlandite) perpendicular to the sur-
face of aggregates [2,23], the diatomite-containing concrete matrix
presents itself as a web of long particles, promoting a very porous
matrix (Fig. 12). The EDS analysis performed on diatomite particle
confirms the results of XRF, in which the major elements are sili-
cium (81,7%) and calcium (7,3%). The carbon and oxygen appears
as components related to carbonate and oxides, and also to the
resin used in the samples impregnation.
The differences in the microstructure porosity observed by SEM
confirmed the mercury intrusion porosimetry tests (Table 4).
Fig. 13 shows that concrete with 8% of diatomite presents the
higher porosity for the dimensions of the capillary pores (0,1 to
10 lm). The composition with 20% of metakaolin presents a beha-
viour near the base concrete; however, it shows a significate
increase of pores smaller than 0,01 lm, probably due to the
increase in water content and the air inside the metakaolin
agglomerates.
The high porosity in the diatomite-containing concrete can be
due to two factors. The first factor is due to the increase in water
content, promoted by the higher porosity of diatomite particles,
which implies that the concrete mortar is also very porous and,
the second factor, is related to the lowest pozzolanic reactivity of Fig. 9. Aluminium, silica and calcium X-ray maps within the demarcated squared
diatomite. zone in Fig. 6.
380 H. Paiva et al. / Construction and Building Materials 140 (2017) 374–384

Fig. 10. EDS spectrum of the demarcated area in Fig. 6.

Fig. 12. SEM/BEI of 8% diatomite concrete sample. The circle identifies a large
diatomite particle.

Table 4
Total porosity and mean pore diameter in the different studied compositions.
(Workability adjustment with water).

Compositions Average pore diameter (lm) Porosity (%)

C_0,6W 0,09 15,00
C_10MK_0,65W 0,02 13,32
C_20MK_0,70W 0,02 15,64
C_30MK_0,75W 0,03 18,22
C_5D_0,67W 0,10 23,92
C_8D_0,70W 0,07 19,24

Fig. 14 also shows the variation of compressive strength of con-

Fig. 11. SEM/BEI of 20% MK and 8% D concrete samples. Some MK agglomerates are
marked in top photo.
mal that the concrete with metakaolin shows lower porosity than
the concrete with diatomite, due to those two factors.
Table 6 shows the results for slump and mechanical strength. As
said before, the slump for the different formulations ranges from 8
to 10 cm (Abrams cone test) and, in this way, all compositions have
a similar workability.
cretes with metakaolin and diatomite. There is a decrease in the
compressive strength with increasing pozzolan content, which is
more pronounced in the diatomite-containing samples.
It becomes clear here that the physical and chemical features of
the pozzolans affect differently the final properties of concrete and
its microstructure. The addition of metakaolin, containing not only
amorphous silica but also alumina, has a higher pozzolanic reactiv-
ity than the high-porosity diatomite, generating a denser
microstructure and a concrete with a higher mechanical strength.
In some studies on the effect of pozzolan in the characteristics
of mortars and concretes, plasticizers or superplasticizers were
used to control the amount of mixing water and thus the workabil-
ity of systems [24]. Wild and Khatib [25] also have considered that
 H. Paiva et al. / Construction and Building Materials 140 (2017) 374–384 381

The use of water reducing agent (WR) instead of water to con-

trol the concrete workability promotes the partial dispersion of the
agglomerates and disperses the particles of cement and metakao-
lin. The dispersion of the metakaolin agglomerates is still not com-
plete because the amount of water reducing agent used was only
the necessary to obtain the desired workability.
Fig. 15 shows SEM/BEI image of a sample of base concrete and
other containing 20% metakaolin. The workability was adjusted
to 8–10 cm slump range, through the use of a water reducing
agent. Compared to the sample of concrete also with 20% metakao-
lin and workability control trough the water content (in Fig. 5),
there is a clear decrease in the number and size of pores. The con-
Diameter (µm) nection between the cement matrix and aggregates also seems to
be stronger.
Fig. 13. Pore size distribution by MIP for base concrete (▲), 20% MK concrete (d) Fig. 16 shows metakaolin agglomerates (marked as circles in
and 8%D concrete (j). photo). Compared with the concrete with 20% of metakaolin and
workability adjustment with water (Fig. 5), the agglomerates of
metakaolin in concrete with water reducing agent have much
Table 5
Free Ca(OH)2 amount obtained by thermal analy- smaller dimensions. The higher magnification used to observe this
sis in the different studied compositions (Worka- sample shows the high density obtained with the use of metakao-
bility adjustment with water). lin simultaneously with the water reducing agent.
Formulation Ca(OH)2 content Fig. 17 shows that concrete with 30% of metakaolin and water
(% by mass of cement) reducing agent presents a good connection in the interface
C_0,6W 15,1 matrix-aggregate and, although the increase of the content of
C_20MK_0,70W 5,4 metakaolin introduce more air in the system, it is possible to
C_30MK_0,75W 3,7 obtain a material with a high density.
C_5D_0,67W 13,4
C_8D_0,70W 12,2
C_10D_0,73W 11,1

Table 6
Analysis of concrete compressive strength and slump (Workability adjustment with
water).

Compositions Compressive strength (MPa) Slump (cm)

Base mortar C_0,60W 38,1 ± 0,9 8,50
C_10%MK_0,65W 38,4 ± 1,0 8,50
C_20%MK_0,70W 34,3 ± 0,9 9,50
C_30%MK_0,75W 30,3 ± 0,6 9,50
C_3D_0,65W 32,7 ± 0,9 9,00
C_5%D_0,67W 31,2 ± 0,5 9,00
C_8%D_0,70W 26,6 ± 1,0 9,00
C_10%D_0,73W 26,3 ± 0,9 9,00

Base concrete

Fig. 14. Compressive strength (Rc) values for compositions with different MK and D
contents. The workability was kept constant by changing water content.

20% MK
metakaolin is different from other pozzolanic material, not only
due to its high pozzolanic reactivity due to high reaction rates with
high levels of portlandite, but also by its ability to accelerate the Fig. 15. SEM/BEI of base concrete and with 20% of metakaolin and water reducing
cement hydration. agent.
382 H. Paiva et al. / Construction and Building Materials 140 (2017) 374–384

The concrete with 8% diatomite and superplasticizer (Fig. 18)

Fig. 16. SEM/BEI of 20% MK concrete showing metakaolin agglomerates (Worka-
bility adjusted with WR). MK agglomerates are marked with circles.
shows fewer pores and apparently an acceptable area of interface
between the matrix and the aggregate. In Fig. 19 it is possible to
observe clearly the darker halos of hydrated cement around the
cement particles. Comparing with samples where the workability
control was done changing the amount of water, it is possible to
confirm a much denser and better connection between the aggre-
gates and the cementitious matrix (Figs. 18 and 19). It is also pos-
sible to verify that higher amounts of diatomite promote a denser
matrix, since the reaction with the calcium hydroxide is more
extensive. Concrete with 5% of diatomite presents larger pores
and the interface between matrix and aggregates is weaker
(Fig. 19).
The porosimetry (MIP) results (Fig. 20) demonstrate the benefi-
cial effect of the use of water-reducing agent. The concrete with
diatomite presents a very low amount of capillary pores while
maintaining an increase in the number of pores smaller than
0,1 lm. The average pore diameter is still around 0,02 lm but
the total porosity shows a marked decrease (Table 7). The concrete
with diatomite shows a great decrease in the number of capillary

Fig. 18. SEM/BEI of concrete with 8% of diatomite and workability adjusted with
Fig. 17. SEM/BEI of concrete with 30% of metakaolin (Workability adjusted with WR agent.
WR).

One reason for the diatomite to present lower pozzolanic reac-

tivity than metakaolin is the difference in surface area, determined
by BET. The metakaolin surface area of is 17 m2/g while the surface
area of the diatomite is 8 m2/g, making it less reactive. Wen-Tien
Tsai et al. [26] showed that the diatomite having a surface area
of 4 m2/g (BET) can increase its area of about 10 m2/g by a chemical
washing process with hydrofluoric acid. In this study, diatomite
sent by the supplier was used as such and no processing to
improve its surface area was performed.
The deagglomeration of diatomite is improved with the pres-
ence of the water reducing agent, which has also a beneficial effect
on the pozzolanic reaction of diatomite. Probably, this increase of
the pozzolanic reaction of diatomite is due to the fact that the
water reducing agent also contributes to eliminate most of the
cement agglomerates, making the system more reactive, accelerat-
ing the hydration reactions and thus increasing the content of
available calcium hydroxide to the reaction with diatomite. Besides
this factor, the presence of the water reducing agent, destroying
the agglomerates, also promotes the increase of the mixtures Fig. 19. SEM/BEI of concrete with 5% of diatomite and workability adjusted with
homogeneity. WR agent.
 H. Paiva et al. / Construction and Building Materials 140 (2017) 374–384 383

55

50

45

Rc (MPa)
40

35 RC_D
Rc_MK
30

25

20
0 5 10 15 20 25 30 35
Pozzolan (%)

Fig. 21. Compressive strength (Rc) values for concretes with different MK and D
contents. The workability was adjusted by the water reducing agent (WR).
Fig. 20. Pore size distribution obtained by MIP for base concrete (▲), 20% MK
concrete (d) and 8% D concrete (j).

taining different contents of diatomite and metakaolin. Comparing

pores, approaching the base concrete pore distribution curve. with the results obtained for the workability control with water
Although the composition with 5% of diatomite still has an average content (in Table 5), one can observe a large increase in the con-
diameter and porosity values higher than those presented by the sumption of Ca(OH)2 with increase in the content of pozzolan.
base concrete, increasing the content of diatomite (8%) leads to a Although this increase was more significant for MK compositions,
decrease in these two variables below the level of the base con- the diatomite compositions also exhibit a marked increase in cal-
crete (C_0,6 W), as shown in Table 7. cium hydroxide consumption. These results demonstrate again
Regarding the consumption of calcium hydroxide by the differ- that the introduction of WR admixture increases the reactivity of
ent pozzolans, Table 8 shows the results obtained for concrete con- both systems. As these reactions occur inside the capillary pores,
there is a densification of the systems already observed in the
porosimetry tests. This densification has consequences on concrete
Table 7 compressive strength results (in Table 9 and Fig. 21).
Total porosity and mean pore diameter in the different studied compositions obtained Table 9 presents the results for slump and mechanical strength
by MIP (Workability adjustment with WR).
for compositions with WRA. Slump ranges from 8 to 10 cm in order
Compositions Average pore diameter (lm) Porosity (%) to have similar workability compositions.
C_0,6W 0,09 15,00 Fig. 21 shows the evolution of compressive strength values for
C_10MK_0,6W_0,10WR 0,02 10,02 concretes with different MK and D contents. It is possible to
C_20MK_0,6W_0,15WR 0,02 10,08 observe an increase in compressive strength with increasing MK
C_30MK_0,6W_0,20WR 0,02 11,93
content. Compositions with diatomite maintain the compressive
C_5D_0,6W_0,08WR 0,10 21,57
C_8D_0,6W_0,10WR 0,07 13,31 strength values near the values showed by the standard concrete
composition. This behaviour is observed in compositions with dia-
tomite due to their intrinsic porosity. While there is an increase in
Ca(OH)2 consumption and a decrease in capillary porosity there is
Table 8
also an increase in the number of pores smaller than 0,1 lm.
Free Ca(OH)2 amount obtained by thermal analysis in the
different studied compositions (Workability adjustment
with WR).
4. Conclusions
Composition Ca(OH)2 content
(% by mass of cement)
It is possible to observe in this study the effect of two physical
C_0.6W 15,1 and chemically different pozzolans on the concrete microstructure
C_20MK_0,6W_0,15WR 2,9
and basic hardened state properties.
C_30MK_0,6W_0,20WR 1,6
C_5D_0,6W_0,08WR 12,4 The pozzolanic reactivity is a factor which has an impact on the
C_8D_0,6W_0,10WR 7,4 microstructure and hardened state properties of the concrete,
C_10D_0,6W_0,12WR 5,4 reflected in the calcium hydroxide consumption, porosity develop-
ment and final compressive strength. The physical characteristics
of each pozzolan, such as the tendency to agglomerate or its intrin-
Table 9 sic porosity, exemplified by the metakaolin and diatomite respec-
Concrete compressive strength and slump for compositions with WRA. tively, clearly have affected the final concrete properties. Using
SEM/EDS, it was possible to identify the metakaolin aggregates
Compositions Compressive strength (MPa) Slump (cm)
and the diatomite particles. The EDS analysis allowed mapping
Base mortar C_0,60W 38,10 ± 0,6 8,50
out relevant elements such as aluminium, silicium and calcium.
C_10%MK_0,60W_0,10WRA 42,30 ± 0,8 8,50
C_20%MK_0,60W_0,15WRA 44,60 ± 0,3 9,50 Two different approaches were used to control workability of
C_30%MK_0,60W_0,20WRA 46,10 ± 0,6 9,50 concrete with both pozzolans, either by using water content or
C_3D_0,60W__0,05WRA 37,65 ± 0,8 9,00 using a water reducing (WR) admixture.
C_5%D_0,60W_0,08WRA 37,00 ± 0,7 9,00 When water was used to control the workability, it was clearly
C_8%D_0,60W_0,10WRA 37,38 ± 0,8 9,00
visible the difference between the two pozzolan based concrete
C_10%D_0,60W_0,12WRA 37,70 ± 0,5 9,00
microstructure. Diatomite-containing concrete presents very high
384 H. Paiva et al. / Construction and Building Materials 140 (2017) 374–384
porosity in the matrix and also in its interface with the aggregate.
Concrete with metakaolin presents fine pores and mostly in the
aggregate-matrix interface. Base concrete is the composition with
the lower porosity (15%), followed by the metakaolin (18% for
the 30%MK concrete) and then the diatomite concrete with the
highest one (19% for only 10%D). These differences are reflected
in the final properties. Compressive strength decreases with the
increase in pozzolan content. Using 30% MK, compressive strength
decrease from 38 MPa to 30 MPa, compared to the base reference
concrete value. This is because, although this pozzolan has a high
pozzolanic reactivity, its fine particles and agglomeration needs
more water to keep the workability. When diatomite is used, the
concrete shows a higher decrease in compressive strength than
the metakaolin one, from 38 MPa to 26 MPa (for 10%D). This effect
is due to its lower pozzolanic reactivity and to the higher diatomite
porosity, which needs a higher water content to maintain constant
workability.
However, when a water reducing agent was used to control the
workability the results were different. The use of this admixture
promotes the particles agglomerates destruction and also the dis-
persion of the cement and pozzolans particles. The dispersion
increases the pozzolans reactivity, because dispersed and finer par-
ticles are more reactive.
In this case, the use of 30% MK increases compressive strength
of the base reference concrete from 38 MPa to 46 MPa and when
diatomite was used, the concrete even kept its compressive
strength around 38 MPa (for 10%D). Porosity was in this case lower
compared with the previous situation. Now the metakaolin and
diatomite concretes present 12% and 13%, respectively, compared
to the 15% of the base concrete. The particles deagglomeration
action of the water reducing agent promoted an increase in reactiv-
ity of the systems and, accordingly, an improvement in the con-
crete hardened state properties.
The use of polished samples in microstructural analysis by SEM
is an effective method to evaluate these features. This method
allowed observing the differences in the microstructure of a poz-
zolanic concrete when using water content or a water reducing
agent to adjust the workability and better understand the resulting
hardened state properties of a pozzolanic concrete.
References
[1] F.P. Glasser, J. Marchand, E. Samson, Durability of concrete – degradation
phenomena involving detrimental reactions, Cem. Concr. Res. 38 (2008) 226–
246.
[2] P.C. Hewlett, Lea’s Chemistry of Cement and Concrete, Butterworth-
Heinemann, UK, Oxford, 2001.
[3] P. Stroeven, M. Stroeven, Reconstructions by SPACE of the interfacial transition
zone, Cem. Concr. Compos. 23 (2001) 189–200.
[4] J. Ambroise, M. Murat, J. Pera, Hydration reaction and hardening of calcined
clays and related minerals V: Extension of the research and general
conclusions, Cem. Concr. Res. 15 (1985) 261–268.
[5] F. Moisés, J. Cabrera, Pore size distribution and degree of hydration of
metakaolin- cement pastes, Cem. Concr. Res. 30 (2000) 561–569.
[6] C.S. Spoon, S.C. Kou, L. Lam, Compressive strength, chloride diffusivity and pore
structure of high performance metakaolin and silica fume concrete, Constr.
Build. Mat. 20 (2006) 858–865.
[7] N. Degirmenci, A. Yilmaz, Use of diatomite as partial replacement for Portland
cement in cement mortars, Constr. Build. Mat. 23 (2009) 284–288.
[8] V. Rahhal, R. Talero, Calorimetry of Portland cement with silica fume, diatomite
and quartz additions, Constr. Build. Mat. 23 (2009) 3367–3374.
[9] A.C. Aydin, R. Gul, Influence of volcanic originated natural materials as additive
on the setting time and some mechanical properties of concrete, Constr. Build.
Mat. 21 (2007) 1277–1281.
[10] B. Yilmaz, N. Ediz, The use of raw and calcined diatomite in cement production,
Cem. Constr. Compos. 30 (2008) 202–211.
[11] M.G. Stamatakis, D. Fragoulis, G. Csirik, I. Bedelean, S. Pedersen, The influence
of biogenic-silica-rich rocks on the properties of blended cements, Cem. Concr.
Compos. 25 (2003) 177–184.
[12] A. Ergün, Effects of the usage of diatomite and waste marble powder as partial
replacement of cement on the mechanical properties of concrete, Constr.
Build. Mat. 25 (2011) 806–812.
[13] S. Diamond, J. Huang, The IZT in concrete – a different view based on image
analysis and SEM observations, Cem. Concr. Compos. 23 (2001) 179–188.
[14] J.A. Rosignolo, Interfacial interactions in concretes with silica fume and SBR
latex, Constr. Build. Mat. 23 (2009) 817–821.
[15] S. Diamond, The patch microstructure in concrete: The effect of
superplasticizer, Cem. Concr. Res. 36 (2006) 776–779.
[16] T. Oertel, F. Hutter, R. Tänzer, U. Helbig, G. Sextl, Primary particle size and
agglomerate size effects of amorphous silica in ultra-high performance
concrete, Cem. Concr. Compos. 3 (2013) 761–767.
[17] J. Faury, Le Beton. Influence de ses constituents inerts. Regles á adopter pour
sameilheure composition. Sa confection et son transport sur les chantier, 3rd
ed., Dunod, Paris, 1958.
[18] H. Paiva, A. Velosa, P. Cachim, V.M. Ferreira, Correlation between mortar and
concrete behavior using rheological analysis, J. Build. Eng. (2015) 177–188.
[19] E. Ferraz, S. Andrejkovičová, W. Hajjaji, A.L. Velosa, Silva A. Santos, F. Rocha,
Pozzolanic activity of metakaolins by the French standard of the modified
Chapelle test: a direct methodology, Acta Geodyn. Geomater. 12 (2015) 289–
298.
[20] NF P18–513:2010, Metakaolin. Pozzolanic addition for concrete. Definitions,
specifications and conformity criteria. Association Française de Normalisation,
La Plaine Saint-Denis, (in French).
[21] H. Paiva, A. Velosa, P. Cachim, V.M. Ferreira, Effect of metakaolin dispersion on
fresh and hardened state properties of concrete, Cem. Concr. Res. 4 (2012)
607–612.
[22] D. Kastis, G. Kakali, S. Tsivilís, M.G. Stamatakis, Properties and hydration of
blended cements with calcareous diatomite, Cem. Concr. Res. 36 (2006) 1821–
1826.
[23] G. Prokopski, J. Halbiniak, Interfacial transition zone in cementitious materials,
Cem. Concr. Res. 30 (2000) 579–583.
[24] G.R. Sensale, Strength development with rice-husk ash, Cem. Concr. Compos.
28 (2006) 158–160.
[25] S. Wild, J.M. Khatib, A. Jones, Relative strength, pozzolanic activity and cement
hydration in superplasticised metakaolin concrete, Cem. Concr. Res. 26 (1996)
1537.
[26] Wen.-Tien. Tsai, Chi.-Wei. Lai, Kuo.-Jong. Hsien, Characterization and
absorption of diatomaceous earth modified by hydrofluoric acid etching,
Coll. Interface Sci. 29 (2006) 749–754.