You are on page 1of 13

Construction and Building Materials 191 (2018) 853–865

Contents lists available at ScienceDirect

Construction and Building Materials


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

Porous arrangement and transport properties of geopolymers


Quang Hung Nguyen a,b, Sylvie Lorente a, Anne Duhart-Barone b, Hervé Lamotte b
a
LMDC, INSA/UPS Génie Civil, 135 Avenue de Rangueil, 31077 Toulouse cedex 04, France
b
DEN/DSN/SEEC/LECD, Bât. 326, CEA Cadarache, 13108 Saint Paul Lez Durance, France

h i g h l i g h t s

 A new protocol and interpretation of the thermoporometry were proposed.


 Geopolymers have a multi-scale porosity.
 The effective diffusion coefficient is one order of magnitude smaller than the bulk coefficient diffusion.
 Geopolymers behavior as ink-bottled materials with mesopores connected to macropores in the core of the material.

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

Article history: This paper documents the results of a study on the relationship of the pore structure and the transport
Received 30 March 2018 properties of potassium- and sodium-based geopolymers. The pore structure was investigated by water
Received in revised form 28 September porosity, mercury intrusion porosimetry and thermoporometry. The geopolymers transport properties
2018
were determined using a set of experiments, including natural diffusion of tritiated water and two elec-
Accepted 4 October 2018
trical methods (direct current and Electrochemical Impedance Spectrometry). The geopolymers
microstructure corresponds to materials with a pore structure made of poor access dimensions around
7 nm (potassium-based geopolymers), and 20 nm (sodium-based geopolymers) connected to larger pores
Keywords:
Porous network
not detected by MIP but revealed by thermoporometry. As a result, the transport properties appear to be
Geopolymer higher than expected.
Tritium diffusion Ó 2018 Elsevier Ltd. All rights reserved.
Direct current
EIS
MIP
Thermoporometry

1. Introduction cement production) geopolymer shifted to be used in the construc-


tion sector [4]. Geopolymers show advantages in immobilizing
The term ‘‘geopolymer” was firstly introduced by Davidovits [1], low- and intermediate-level nuclear waste [5,6]. Moreover, they
who developed a binder obtained from the polycondensation of also possess a compressive strength similar to that of cement-
metakaolin in an alkali medium at the temperature of 100– based materials, which explains the primary application of
150 °C. As the raw material is mainly of geological origin (clay, kao- geopolymers as an alternative material to ordinary Portland
lin), Davidovits termed geopolymer as a mineral polymer obtained cement [7].
by chemical processes of geosynthesis [2]. According to Provis and Several properties of geopolymers have been investigated in the
Deventer [3], geopolymers are framework structures produced by literature: chemical durability (acid attack, sulfate attack, alkali sil-
the condensation of tetrahedral aluminosilicate units, with alkali ica reaction, carbonation. . ..) [7–11], fire-resistance [12,13] and
metal ions balancing the charge associated with tetrahedral Al. electrical properties [7,14–17]. The durability of geopolymers, as
The alumino-silicate source, which has a low content of calcium, well as porous materials in general, depends on the ability to with-
can be fly-ash class F (FA), natural pozzolan or metakaolin (MK) stand the intrusion of aggressive agents. The transport properties
as mentioned in Fig. 1. of a porous materials are influenced by the pore structure charac-
In the beginning, geopolymer was developed as a heat-resistant teristics [18,19]. The formation factor, defined as the ratio between
material after some catastrophic fires in Europe but later, thanks to the electrical conductivity of the bulk solution and the one of the
its eco-friendly properties (low CO2 emission in comparison to the material saturated by the same solution [20,21] can be related to

https://doi.org/10.1016/j.conbuildmat.2018.10.028
0950-0618/Ó 2018 Elsevier Ltd. All rights reserved.
854 Q.H. Nguyen et al. / Construction and Building Materials 191 (2018) 853–865

and results on the diffusion coefficient (effective diffusion coeffi-


cient – De or apparent diffusion coefficient – Da) of geopolymer
materials: (See Table 1).
As seen, the diffusion coefficient of geopolymer materials varies
in a large range, it is usually around 1011 m2/s essentially in the
case of sodium-based geopolymers.
Recent studies on transport properties at early age of
metakaolin-based geopolymer and cement paste indicate that the
geopolymer has a lower permeability than that of the cement paste
[36,37], despite a similar open porosity. However, other studies on
the permeability at the later ages of fly ash-based geopolymer
showed that the permeability of the latter is higher than that of
cement paste [38].
Here, we study the link between the porous arrangement of
sodium and potassium geopolymers and their transport properties,
at long term. Firstly, the porous structure of the porous samples
was characterized by water open porosity, MIP and thermoporom-
Fig. 1. CaO-Al2O3-SiO2 ternary diagram of cementitious materials [68]. etry. Later, the transport properties were determined by a direct
current test, an electrochemical impedance spectrometry test
(EIS) and a natural diffusion of tritiated water test.
the material pore structure characteristic by using an expanded
version of Archie’s law [22]:
2. Materials and characterization methods
ro s2
Ff ¼ ¼ ð1Þ 2.1. Geopolymers
rmat po T o
where Ff is the formation factor, ro is the electrical conductivity of GeoK and GeoNa were synthetized; the preparation of the activating solution
and the mixing process are detailed in [27]. The activating solution consisted of
the bulk solution, rmat is the electrical conductivity of the saturated an amorphous precipitated silica dissolved in alkali solution (soda and potash)
material, s is the tortuosity, To is the constrictivity and po, the open and stirred for 24 h. The silica powder was Tixosil 331Ò by Solvay, soda and potash
porosity. was supplied by VWR ProlaboÒ. Note that no sign of gelation was observed in the
The pore distribution of porous materials is often characterized activating solution. The median particle size of the silica powder was between
3 lm and 4 lm, the amount of amorphous silica (SiO2) was 95% and the sulfate
by mercury intrusion porosimetry (MIP) or nitrogen absorption content was less than 4 wt% by relative weight of silica. Metakaolin (PieriÒ Premix
porosimetry (NAP). According to the literature, the pore size distri- MK obtained from Grace Construction) was used as the aluminosilicate source. Its
bution of geopolymers highlights a bell curve centered in a meso- chemical composition (in % mass) is given in Table 2.
porous region [23–26]. For example, the potassium-based The composition of the GeoK and GeoNa pastes is detailed in Table 3. Samples
were in cylindrical form, with a diameter of 50 mm and a length of 70 mm for direct
geopolymer (GeoK) had a peak situated between the range of 8–
current and natural diffusion test, a diameter of 28 mm and a length of 50 mm for
30 nm [26] and between 7 and 10 nm [24,25,27]. Concerning EIS test. All specimens were endogenously cured (hermetically sealed) for
sodium geopolymer (GeoNa), the peak was shifted toward higher 28 months in the room temperature: T = (20 ± 3) °C.
values, up to 30 nm [24,27]. In addition, while studying the nitro-
gen adsorption of potassium and sodium geopolymer, Steins et al. 2.2. Water porosity
[24] and Benavent et al. [28] claimed that these materials have a
type IV isotherms behavior, which is attributed to the ‘‘ink bottle” Water porosity was measured at the end of the curing period. The protocol used
was an adaptation of the AFPC-AFREM recommendations [39], as the saturation
effect. The potassium geopolymer exhibited a higher specific area
time was 10 days instead of 24 h. The open porosity po of the samples was calcu-
than sodium geopolymer. Several methods of measurement like lated by:
NAP, Small-angle neutron scattering (SANS) and Small-angle X-
M sat  M dry
ray scattering (SAXS) showed that the specific area of potassium po ¼ q ð2Þ
M sat  M h water
geopolymer was about 3 times higher than that of sodium geopoly-
mer [24,28]. where Msat is the mass of the saturated sample in the air, Mdry is the dry mass at
105 °C, Mh is the hydrostatic mass and qwater is the water density. The dry bulk den-
To date, publications on the diffusivity of geopolymer are few in sity qa and wet bulk density qs were given by:
number, regardless of the kind of geopolymer (metakaolin
M dry
geopolymer, Fly ash geopolymer, geopolymer mortar, geopolymer qa ¼ q ð3Þ
M sat  M h water
concrete. . .). The following table summarize the relevant methods

Table 1
Summary of the determination of the diffusion coefficient of geopolymer materials.

Author Materials Results Additional information


Lloyd et al. [29] FA geopolymer paste, sodium activator De = 0.67–7.3 1012 m2/s Porosity 30%
Jämstorp et al. [30] MK geopolymer paste, sodium activator De = 1.6–120 1012 m2/s NA
Asprogerakas et al. [31] FA geopolymer mortar, sodium activator Da = 3.7–7.1 1011 m2/s NA
Ismail et al. [32] FA geopolymer mortar, sodium activator Da > 8 1011 m2/s Porosity = 22%
Yang et al. [14] FA geopolymer concrete, sodium activator Da = 1.543 1011 m2/s Porosity 33% (MIP test), 28 days
Chindaprasirt et al. [33] FA geopolymer concrete, sodium activator Da = 2.9–6.2 1010 m2/s Porosity = NA
3 years exposure in marine environment
Law et al. [7] FA geopolymer concrete, sodium activator Da = 3.1–3.7 1011 m2/s NA
Ganesan et al. [15] FA geopolymer concrete, sodium activator Da = 1.24 1011 m2/s NA
Gunasekara et al. [34] FA geopolymer concrete, sodium activator Da = 7.9–9.9 1011 m2/s 1 year curing, Porosity = NA
Kupwade-Patil et al. [35] FA geopolymer concrete, sodium activator Da = 1.45–18.7 1012 m2/s 1 year exposed in salt water
Q.H. Nguyen et al. / Construction and Building Materials 191 (2018) 853–865 855

Table 2
Chemical composition Properties PieriÒ Premix MK.

Chemical component CaO SiO2 Al2O3 Fe2O3 SO3 Na2O K2O Loss of ignition (LOI) Specific gravity g/cm3
Average value (% mass) 0.1 54.4 38.4 1.27 – less than 0.2 0.62 – 2.23

Table 3
Geopolymer composition (final molar ratio).

Molar ratio Activation solution Geopolymer paste


SiO2/M2O* H2O/M2O* SiO2/H2O M2O*/Al2O3 SiO2/Al2O3 H2O/M2O*
GeoK 1.4 12 0.117 1 3.8 12
GeoNa 1 12 0.083 1 3.4 12
*
M = K or Na.

M sat Thermoporometry experiments on geopolymers were performed with a Calvet-


qsat ¼ q ð4Þ
M sat  M h water type low-temperature calorimeter (BT 2.15) from Setaram Instrumentation (Fig. 2),
using the scanning mode (constant gradient of temperature). The cooling device
uses a supply system with ethanol instead of liquid nitrogen. Due to the limitation
2.3. Mercury intrusion porosimetry (MIP) of the latter, the apparatus can detect the presence of pore size above 4 nm. GeoK (3
samples) and GeoNa (2 samples) were analyzed. The samples were crushed into
The mercury intrusion porosimetry (MIP) was performed by an Autopore III smaller pieces of about 2 g before being vacuum saturated in water. The water vol-
9420, commercialized by Micromeritics. The maximal hydraulic pressure of the ume is much greater than that of the samples and the saturation period was long
apparatus was 414 MPa, the contact angle was 130° and the surface tension was enough to ensure that water would replace the original pore solution in the sample.
0.485 N/m. The samples were oven-dried at 80 °C, until the variation of the mass Moreover, we will see in the following section, that the high value of the effective
of the sample at two successive weighing was less than 0.1%. The equilibration rate diffusion coefficient (De  71010 m2/s) of geopolymer samples can explain a
for low pressure and high pressure is 0.03 lL/g. An alternative drying method, quasi-substitution of the interstitial pore solution by water. In addition, an evacu-
based on a lyophilization procedure was also performed but not retained, as dis- ation/renewal of the water in the desiccator to accelerate the substitution process
cussed in the following sections. For each geopolymer, 3 samples of about 6 g were was also performed. No nucleation agent was added to the saturated sample.
tested. The pore size distribution shown here is the average of 3 tests. As MIP is a Two thermal cycles made of a total of four consecutive sequences of cooling/
classical technique, we do not enter into details here. heating were applied to the samples, as shown in Fig. 3.
In the 1st thermal cycle, the 1st cooling sequence started at ambient tempera-
2.4. Thermoporometry ture (20 °C) then went down to 50 °C. It was followed by a 1st heating sequence to
0.5 °C. During the 2nd thermal cycle, the temperature was lowered again to
During the cooling process of a porous material saturated with a pure liquid 50 °C (2nd cooling sequence) and finally the temperature was increased to the
(water for example), the crystallization temperature (T) of the liquid phase confined ambient temperature (2nd heating sequence). The cooling rate was 0.07 °C/min
in the pore is lower than the normal temperature of the triphasic equilibrium (To). and the heating rate was 0.10 °C/min. The total time for such an experiment was
The relation between the freezing temperature depression DT = T  To and the pore about 60 h. At the end of the test, the samples were removed from the calorimeter
radius was derived by Kubelka [40]. Based on thermodynamic considerations which and oven-dried at 80 °C to determine the oven-dried mass (movd) and the quantity
include Laplace equation and Gibbs-Duhem equation, fully developed later by Brun of water held inside the porous material (mw). The mass of ice that froze during the
et al. [41,42], the equation proposed by Kubelka has the following form: 1st cooling sequence and the one melted in the 2nd heating sequence are supposed
to be close to the total amount of water present in the porous sample (mw). 80 °C
DT 2000rls
¼ ð5Þ was chosen rather than 105 °C in order to preserve the non-freezable layer of water,
To r q l Dh
which is not detectable by TMP.
where ql is the density of the liquid (kg/m3), rls is the liquid–solid surface tension (N/ The objective of the 1st heating sequence was to prevent the supercooling effect
m), r is pore radius assumed spherical (m), and Dh is the specific enthalpy of crystal- occurring during the 1st cooling sequence [47]. Supercooling accounts for the fact
lization (J/g). that water remains in liquid state when the temperature is below 0 °C in the pores,
The standard enthalpy of fusion of water in a bulk medium is DhStd = 334.1 [43], as crystallization is a process based on nucleation and crystal growth. While heating
but in the case of a confined medium, this value depends on the temperature. We up to 0.5 °C, the ice in the pores below 130 nm of diameter melts, and water in the
summarize in the following table the empirical formulae from the literature esti- pores of diameter above 130 nm remains as ice, according to Brun’s second equa-
mating the enthalpy change, with the general form Dh = a(DT)2 + bDT + c: tion (Eq. (7)). Therefore in the 2nd cooling sequence, water in the pore range of
If the liquid is water, Brun et al. [42] proposed a semi-empirical equation to (4–130) nm refreezes and the heat flux is the contribution of this pore range. The
determine the radius of the pore. This equation accounts for the thickness of the mass ratio of ice frozen between the 2nd cooling sequence and the 1st cooling
nonfreezable layer water, which they estimated at 0.8 nm. sequence, Pf, and the mass ratio of ice melted between the 1st heating sequence
During the cooling process, considering spherical pore shapes: and the 2nd heating sequence, Pm, were calculated as follows:

64:67 mc2
r¼ þ 0:57 ð6Þ Pf ¼ ice
 100 ð8Þ
DT mc1
ice

and the heating process, considering cylindrical pore shapes: and:


32:33 mh1
r¼ þ 0:68 ð7Þ Pm ¼ ice
 100 ð9Þ
DT mh2
ice

Table 4
Literature value of enthalpy of crystallization/fusion.

Notation Enthalpy (J/g) a b c


Crystallization Fusion
Fagerlund [44] DhFag x 0 2.19 333.7
Brun et al. [42] DhBru x 5.56102 7.43 332
x 0.155 11.39 332
Ishikiriyama et al. [45] DhIsh x 7.83103 2.119 334.1
Matala [46] DhMat x 1.25102 4.83 334
856 Q.H. Nguyen et al. / Construction and Building Materials 191 (2018) 853–865

Fig. 2. Cross-section of the BT 2.15 low-temperature calorimeter (by courtesy of Setaram Instrumentation).

Both values are expected to be in the same range. 4. Calculation of the increment of volume of ice frozen/melted: DVi = Dmi t,
The thermogram (heat flux as a function of the temperature) provides informa- where t is the specific volume of ice, t = 1.000132(1 + 9.1 105T + 1.035 10
5 2
tion on the porous system, such as the pore size distribution and the porosity. To T ) [42].
this end, a baseline must be obtained before any further step. The baseline of a 5. Calculation of the pore size with Eqs. (6) and (7), for negative temperature val-
calorimeter is the recording of its signal as a function of time in the absence of ues only.
the phenomenon intended to be measured [48]. We chose 2 methods of baseline
construction from Hemminger’s paper [49]: the straight line method and the degree Next, we calculated the total amount of ice frozen/melted in the 4 sequences by
P
of conversion method [50]. The degree of conversion method is recommended but summing the increment Dmi: mice ¼ i Dmi . Among the 4 values obtained, those
only can be applied to the 1st cooling sequence and 2nd heating sequence. Hence, from the 1st cooling sequence and 2nd heating sequence were compared to the
the straight line method was used in the 1st heating sequence and 2nd cooling mass of water held inside the material (mw), to verify the accuracy of the calcula-
sequence. tion. The porosity can be estimated by:
Considering an increment of time Dt = ti+1  ti, which corresponds to a variation
X Dmi q
of temperature DT = Ti+1  Ti, we followed the calculation procedure in accordance po ¼ sat
ð10Þ
with the method proposed by Brun et al. [42,51]: i
M sat qwater

1. Construction of the baseline and determination of the heat flux Q. The contribution of the mesoporous region, which is situated between 2 nm and
R tiþ1 50 nm, in each characterization method was determined by the following equation:
2. Determination of the heat released/absorbed by the sample DEi ¼ ti
dQ by
using a numerical integration, based on the trapezoidal method. pmeso
Bmeso ¼  100 ð11Þ
3. Determination of the incremental amount of ice frozen/melted, Dmi = DEi/Dhi, po
where Dhi is the enthalpy of crystallization/fusion. To the authors’ knowledge,
only Brun [42] gave the formulations of enthalpy during crystallization and where pmeso is the porosity of the mesopore region and po is the open porosity mea-
fusion, cf. Table 4. We considered that the other formulations in Table 4 were sured in each method.
applicable for both the crystallization and fusion processes. The temperature Finally, the pore size distribution of geopolymer samples is plotted, from the data
was calculated at (Ti + Ti+1)/2. of the 1st cooling sequence and 2nd cooling sequence. Note that the supercooling
Q.H. Nguyen et al. / Construction and Building Materials 191 (2018) 853–865 857

Fig. 4. Sketch of the migration cell and tested sample [54].

The resulting current was monitored during the experiment. The


details of the experimental test are given in [54] and shown in
Fig. 3. Sequence for the thermoporometry. Fig. 4.
The duration of the test was short enough to avoid any signifi-
cant changes of the electrolytic solution in both compartments and
effect occurs in the 1st cooling sequence. Moreover, data extracted from 2nd cooling any electro-osmotic flow between the 2 compartments [58]. The
sequence were used to determine the pore size distribution between 4 and 130 nm.
waterproofing of the system (PVC-resin-geopolymer) was carefully
checked. A blank test was previously conducted using only the
3. Transport properties epoxy resin and showed no current through the system. We calcu-
lated the electrical conductivity of the saturated material (rmat)
Diffusion and direct current tests were communally developed from Ohm’s law:
in our laboratories. We used natural diffusion test to calculate I L
the effective diffusion coefficient of geopolymer pastes, at the same rmat ¼ ð12Þ
DU A
time, electrokinetic tests (direct current + EIS) were performed to
where DU is the voltage applied to the sample (V), L is its thickness
determine the formation factor of geopolymer samples. The forma-
(m), A is the sample cross-section (m2), and I is the measured cur-
tion factor was not used to calculate the effective diffusion coeffi-
rent (A).
cient. We have recently reported that the analogy between the
The bulk conductivity was calculated from [59]:
formation factor and the diffusivity is not applicable in the meso-
pore range [52]. Note that diffusion and direct current tests were F2 X 2
used so far for mortar and concrete samples [53,54] but an adapta- ro ¼ z D cbulk
i i o;i i
ð13Þ
RT
tion of the test was necessary to avoid damaging the geopolymer
where F is the Faraday constant, R is the gas constant (J/K mol), T is
pastes. A 70 mm diameter cylinder was extracted concentrically
from a PVC cylinder (diameter 120 mm, height 100 mm). The the temperature (K), z is the charge number and cbulk is the concen-
geopolymer sample was then placed into the resulting hole and tration of the ionic species in the bulk solution (mol/m3).
secured to the PVC cylinder thanks to resin epoxy. The whole sys-
tem was then sawed into thin slices of 8 mm thick by a precision 3.2. Electrical impedance spectroscopy (EIS)
diamond wire (Fig. 4). The top and the bottom of the cylinders
were not used in order to ensure the homogeneity of the sample EIS tests were made following the protocol developed by Mer-
properties. The composition of the artificial solution used to satu- cado et al. [53]: a sinusoidal voltage of small amplitude (200 mV)
rate the samples was based on the internal pore solution of the was applied between 2 electrodes and the frequency varied in
geopolymers which was extracted following the protocol described the range of 40 Hz to 110 106 Hz. The saturated samples were
in [53]. The pH of the pore solution was measured immediately placed between copper electrodes specifically designed for the
after the pore solution extraction test by a standard pH meter EIS experiments as shown in Fig. 5.
(HACH 31) and gave a value of 13 for GeoK and GeoNa. A simplified The sample diameter was 28 mm and the sample thickness was
saturation solution, close to the pH value of the pore solution was 19 mm for GeoK, 18.5 mm for GeoNa. The amplitude, the phase
chosen (100 mol/m3 of NaOH for GeoNa samples, 100 mol/m3 of angle and the complex impedance were registered by an HP Agi-
KOH for GeoK samples). All the samples were saturated following lent 4294A.
the AFPC-AFREM protocol [39] and the saturation was extended The impedance Z(x) of the saturated sample was estimated by
to 2 weeks. the impedance of an electrical equivalent model:
Z ðxÞ ¼ Z real ðxÞ þ iZ im ðxÞ ð14Þ
3.1. Direct current test This latter is represented in a complex plane plot (Nyquist plot),
where the real part is plotted on the X-axis and the imaginary part
The direct current test is based on the application of a constant is plotted on the ordinate axis. The value of the resistance (Rmat) of
electric field through the sample [53,55–58]. Each sample was the saturated material is correlated to the minimum value of the
placed between two compartments named anodic and cathodic imaginary part of the Nyquist plot [53]. This model is valid only
depending on the electrical current direction. Both compartments when the material is totally saturated. When the material is par-
of equal volume (1.7 L) were filled with the same solution as the tially saturated, a complex model was developed in [60]. The satu-
samples saturation solution to avoid any concentration gradient. rated material electrical conductivity in this test is then calculated
A constant and uniform electric field of 400 V/m was applied as follows:
between the two parallel faces of the sample, using two electrodes
L
made of a stainless steel grid placed between the sample faces. The rmat ¼ ð15Þ
value of the electric field is typically used in the literature [53,58]. ARmat
858 Q.H. Nguyen et al. / Construction and Building Materials 191 (2018) 853–865

where KD is the distribution ratio (cm3/g) and qs is the density of the


solid phase of the material (g/cm3) [62]. More details on the exper-
imental set-up can be found in [61].

4. Results and discussion

We present in the following table the characteristic properties


for both geopolymers, obtained from water porosity and MIP tests
(see Table 5):
The open porosity of the geopolymers is higher than cement-
based materials that generally is found in a range of (23–35)% for
blended cement pastes [63]. This open porosity is in the same
range as values in the literature; for example in the case of GeoNa,
50–61% [64], 49–52% [27]. The open porosity for GeoK was 48% in
Fig. 5. Configuration of the sample for EIS measurements [53]. [27]. The specific area of GeoK is about 2.5 times higher than that
of GeoNa, in agreement with the results in [24,28]. The pore size
distribution curves obtained by MIP are given in Fig. 7.
The formation factor was calculated as above (Eq. (1)), while ro As seen, the geopolymers have a monomodal distribution with
was measured directly by means of a conductimeter. one peak centered on 7 nm for GeoK and 20 nm for GeoNa. These
results matched well the literature review, mentioned in the previ-
3.3. Diffusion test ous section. A complementary MIP test using a freeze drying pro-
cedure was also conducted on GeoK sample. We obtained a
After performing the direct current test, the tested sample was similar pore size distribution curve, with one peak situated in the
placed in a diffusion cell. The sample was sandwiched between vicinity of 7 nm. These values correspond to the range of meso-
two compartments containing the same saturated solution as the pores, according to the IUPAC classification [65]. The presence of
material sample (Fig. 6). macropores was not noticed, as reported in previous studies on
The system was tested previously to guarantee no leakage geopolymers [23,25,27].
through the epoxy resin. Tritiated water (HTO), with specific activ- The pore size distribution and the porosity were also examined
ity Co = 3.71010 Bq/m3, was added to the upstream compartment by thermoporometry in order to get more information on the
[61]. Due to the concentration gradient, HTO diffuses toward the microstructure of the materials. The thermogram of GeoK and
downstream compartment by molecular diffusion following Fick’s GeoNa are shown in Figs. 8 and 9.
first law of diffusion. The activity of tritiated water in both com- A steep increase in heat flux was observed during the first cool-
partments was regularly measured using a liquid scintillator coun- ing sequence (Figs. 8a and 9a). It was in the vicinity of 10 °C for
ter. The HTO upstream concentration was maintained constant by the GeoK samples, and 7 °C for the GeoNa materials. Such peak
adjusting it after each scintillation measurement, whereas the HTO is due to supercooling, which contains information on the pore size
downstream concentration was maintained close to null in order range. No other peak was found when the temperature continued
to have a constant concentration gradient between the 2 faces of to decrease. The second cooling sequence exhibits 2 peaks, much
the sample. smaller than the one observed in the first sequence (Figs. 8b and
The effective diffusion coefficient was calculated in steady state
by measuring the slope of the curve plotted [62] as:
Table 5
L aL2 Characteristic properties of the two geopolymers.
nðt Þ ¼ De t  ð16Þ
AC o 6 Material Open porosity (%) Dry bulk density Specific area
where n(t) is the cumulative quantity of HTO (Bq) in the down- (g/cm3) (m2/g)
Water MIP Water porosity MIP
stream compartment at time t, and a is defined analytically by: porosity
a ¼ po þ ð1  po Þqs K D ð17Þ GeoK 56.8 50.8 1.09 113.5
GeoNa 55.5 42.0 1.12 44.6

Fig. 6. Representative sketch of a diffusion cell [61]. Fig. 7. Pore size-distribution of geopolymers by MIP.
Q.H. Nguyen et al. / Construction and Building Materials 191 (2018) 853–865 859

Fig. 8. Thermogram of GeoK: (a) 1st thermal cycle (b): 2nd thermal cycle (c): 2 thermal cycles together.

Fig. 9. Thermogram of GeoNa: (a) 1st thermal cycle (b): 2nd thermal cycle (c): 2 thermal cycles together.

9b). The additional peak was detected at a temperature higher than details of Figs. 10 and 11, with the specific enthalpies given in
the one during the first cooling sequence, in the vicinity of 3 °C to Table 4 were presented in the Appendix 3 and 4 (Figs. 15 and 16).
2 °C. The impact of supercooling can be easily observed in both cases
Using the methodology described in Section 2.4, the pore size (Figs. 10a and 11a). Remember that the first thermal cycle (from
distribution of the geopolymers could be obtained. Results from ambient temperature to 50 °C, then up to 0.5 °C, cf. Fig. 3)
the 1st and 2nd cooling sequences are given in Figs. 10 and 11 was implemented with the objective of decreasing the supercool-
for GeoK and GeoNa respectively. Note that these figures were ing effect. The pore size distribution plotted from the 2nd cooling
obtained by using only Brun’s enthalpy, and that they are the sequence shows that the intensity of the peak was decreased
results from several samples of geopolymers in each case. Full (Figs. 10b and 11b). That peak corresponds to 15 nm for GeoK,
860 Q.H. Nguyen et al. / Construction and Building Materials 191 (2018) 853–865

paring pore size values is not relevant here as MIP is based on the
assumption of cylindrical pores, while the cooling sequences in
thermoporometry consider spherical pores (Eq. (6)). Important is
the occurrence of a pore family of larger dimension which could
not be detected by MIP. At present the explanation we can propose
is based on the ink-bottle effect, a well-known and long recognized
bias of the MIP technique. The measured pressure is the mercury
pressure necessary to access the pore entrance. What happens
behind the pore neck is unknown. As the pressure needed to push
mercury into the pore network is inversely proportional to the pore
size, pores with larger dimensions than the pore entrance cannot
be detected.
We present in Table 6 the calculations of the mass of water that
froze or melted during the 1st and 2nd cooling sequences of the
thermal process for the GeoK. The average values were obtained
using the standard specific enthalpies of crystallization/fusion,
DhStd, and the ones provided in Table 4. Full details of the calcula-
tions of 4 sequences for each sample are given in the Table 10
(Appendix 1). The values were obtained using the standard specific
enthalpies of crystallization/fusion, DhStd, and the ones provided in
Table 4. Given also in the second column of the table is the total
mass of water per mass of material, measured after the oven-
drying at 80 °C, as mentioned earlier. The specific latent heat pro-
vided by Brun et al. [42] is the one giving the closest results to mw.
Taking the example of the first sample in Table 10, the calculations
using DhBru, show that the sample contains 0.41 g of water per
Fig. 10. Pore size distribution of GeoK from thermoporometry (Brun’s enthalpy): (a) mass of material during the 1st cooling sequence (from 20 °C to
from the 1st cooling sequence (b): from the 2nd cooling sequence. 50 °C). Then, the 1st heating sequence (from 50 °C to 0.5 °C)
allowed to melt the ice in the pore range of (4 –130) nm. The mass
of ice melted in this sequence is mh1 ice ¼ 0:25 g. The remaining mass,
mc1
ice  m h1
ice ¼ 0:15 g per mass is the mass in the largest pores
(> 130 nm). Next, during the 2nd cooling sequence from 0.5 °C
to 50 °C, liquid water in the range of (4 –130) nm freezes again
and mc2 ice ¼ 0:31 g per mass of material. This value is close enough
to mh1ice to be considered acceptable. Finally, the entire volume of
water melts during the 4th sequence which allows to the reach
the ambient temperature of 20 °C. The difference between Pf and
Pm is minimized for the Brun’s enthalpy formulation. The porosity
acquired from the specific enthalpy DhBru is of 46.5% for the
potassium-based geopolymer, which is lower than the open poros-
ity obtained from water porosity test (56.8%).
In the case of the sodium-based geopolymer, the crystalliza-
tion/fusion specific enthalpy given by Brun et al [42] provides
the closest result to the mass of water mw. The average results of
the 1st and 2nd cooling sequences are given in Table 7, the details
of the calculations are also given in the Table 11 (Appendix 2).
Results from Thermoporometry (TMP) are highly different from
that of MIP, particularly in the mesoporous region. Using Eq. (11),
we present in the following table the proportion of mesopores
obtained in each methods (see Table 8).
MIP shows that most of the pores of geopolymer samples
(97%) are centered in the mesoporous region, whereas TMP high-
Fig. 11. Pore size distribution of GeoNa from thermoporometry (Brun’s enthalpy): lights a lower proportion of mesopores, particularly in the case of
(a) from the 1st cooling sequence (b): from the 2nd cooling sequence. GeoNa (11%). TMP also quantifies a large amount of the macrop-
ores, questioning the type of pore arrangement of all families
inside the geopolymers.
and 20 nm for GeoNa. The presence of a second peak can be Now we will expose the results from the transport tests, in
noticed. It is in the vicinity of 35 nm in the case of GeoK and order to correlate to the previous characterization results. Fig. 12
70 nm for GeoNa. Therefore, unlike the MIP results which indicate presents the voltage applied between the 2 parallel faces of the
only one pore mode (7 nm for GeoK), Fig. 10b shows that the samples, together with the electric current as a function of time.
potassium-based geopolymer has 2 pore sizes in the mesopores Here, we show the case of GeoNa, yet the other geopolymer exhib-
domain. Similar behavior is noticed in the case of sodium-based ited similar behavior. The reproducibility was checked by testing 3
geopolymers with again 2 pore families instead of 1 detected by samples of each geopolymer; repeatability was also insured by
MIP (20 nm for GeoNa), Fig. 11b. In addition to the mesopore range, testing each sample 3 times. The drop in voltage (and therefore
the presence of macropores (> 50 nm) is clearly highlighted. Com- in direct current) measured at the very start of the test is due to
Q.H. Nguyen et al. / Construction and Building Materials 191 (2018) 853–865 861

Table 6
Mass of frozen/melted water per mass of material, and porosity. Case of GeoK.

mw/movd (g/g) Crystallization/fusion specific enthalpy 1st cooling sequence 2nd heating sequence Pf Pm
(20 °C ? 50 °C) (50 °C ? 20 °C)
(J/g) mc1
ice (g/g)
Porosity (%) mh2
ice (g/g)
Porosity (%) (%) (%)

0.45 ± 0.01 DhStd 0.30 ± 0.005 35.1 ± 0.4 0.31 ± 0.005 36.1 ± 0.4 79.8 ± 0.8 62.6 ± 5.0
DhFag 0.33 ± 0.008 38.2 ± 0.5 0.32 ± 0.005 37.3 ± 0.4 78.9 ± 0.8 63.4 ± 4.9
DhBru 0.41 ± 0.008 47.3 ± 0.6 0.35 ± 0.005 40.7 ± 0.4 77.1 ± 0.8 65.5 ± 4.8
DhIsi 0.33 ± 0.008 38.2 ± 0.5 0.32 ± 0.005 37.2 ± 0.5 79.0 ± 0.8 63.4 ± 4.9
DMat 0.37 ± 0.008 42.5 ± 0.5 0.33 ± 0.005 38.8 ± 0.5 77.9 ± 0.8 64.4 ± 4.8

Table 7
Mass of frozen/melted water per mass of material, and porosity. Case of GeoNa.

mw/movd (g/g) Crystallization/fusion specific enthalpy 1st cooling sequence 2nd heating sequence Pf Pm
(20 °C ? 50 °C) (50 °C ? 20 °C)
(J/g) mc1
ice (g/g)
Porosity (%) mh2
ice (g/g)
Porosity (%) (%) (%)

0.37 ± 0.005 DhStd 0.24 ± 0.010 30.0 ± 1.1 0.26 ± 0.01 31.5 ± 0.9 47.5 ± 2.0 33.5 ± 3.5
DhFag 0.26 ± 0.005 31.5 ± 1.1 0.26 ± 0.01 31.6 ± 0.9 46.1 ± 1.6 33.7 ± 3.5
DhBru 0.29 ± 0.010 35.9 ± 1.4 0.26 ± 0.01 31.9 ± 0.9 42.8 ± 1.1 34.3 ± 3.5
DhIsi 0.26 ± 0.005 31.5 ± 1.2 0.26 ± 0.01 31.6 ± 0.9 46.1 ± 1.6 33.6 ± 3.5
DMat 0.28 ± 0.005 33.5 ± 1.3 0.26 ± 0.01 31.7 ± 1.0 44.4 ± 1.3 33.9 ± 3.5

Table 8
Proportion of the mesopore region (percentage) in each characterization method.

MIP TMP
GeoK 97.8 57.3
GeoNa 97.1 11.2

Fig. 13. Nyquist plot for the GeoNa sample.

rent tests. Presented in Fig. 13 is the example of one test for GeoNa.
Here, Rmat = 383 X, corresponding to 0.82 MHz. Eq. (15) allows to
calculate the material conductivity and Eq. (1) gives the formation
factor. The electrical conductivity of the bulk solution of NaOH
Fig. 12. Voltage and current in the direct current test of GeoNa. 0.1 M or KOH 0.1 M was respectively 22.3 mS/cm and 24.3 mS/
cm. These results match well to the ones estimated by Eq. (13).
As shown in Table 9, the material conductivities are similar what-
the time needed to obtain the targeted voltage. The material ever the electrical method, direct current or EIS. The electrical con-
conductivity and the corresponding formation factor were then ductivity of the saturated material rmat is one order of magnitude
calculated by means of Eqs. (12) and (1) (cf. Table 9). smaller than the bulk conductivity.
Electrical Impedance Spectroscopy tests were also performed In comparison with results on cement-based materials using
on the GeoK and GeoNa geopolymers. Reproducibility and repeata- the previous electrical methods, the formation factor obtained for
bility were again checked in the same manner as for the direct cur- a saturated blended cement paste (type V cement) with an EIS test
(Ff  300) [66] was found one order of magnitude higher than the
geopolymers tested in this work (Table 9). In Mercado et al. [66],
Table 9 the blended cement paste had an open porosity of 37% and MIP
Electrical conductivities and formation factors of the geopolymers. tests showed that most of the pore sizes were below 30 nm. The
Material Direct current EIS open porosity of geopolymers is about 1.5 times higher than the
rmat (mS/cm) Ff rmat (mS/cm) Ff
one of cement paste (55% vs 37%). The formation factor of cement
paste is about 300, compared to 20 for GeoNa, as to say 15 times
GeoK 2.15 ± 0.40 12 ± 2 1.88 ± 0.14 14 ± 1
GeoNa 1.12 ± 0.02 19 ± 1 1.04 ± 0.09 22 ± 2
higher. The difference in open porosity cannot explain such a
discrepancy.
862 Q.H. Nguyen et al. / Construction and Building Materials 191 (2018) 853–865

mono-modal materials but rather an example of ink-bottled


model when fully cured.

5. Conclusion

In this work, we characterized sodium- and potassium-based


geopolymers through their microstructure and transport proper-
ties. The main conclusions can be highlighted as follows:

– Mercury Intrusion Porometry tests indicate that the pore size is


centered around 20 nm for sodium-based geopolymers, and
around 7 nm for potassium-based geopolymers. Thermoporom-
etry results indicate that a second pore family exists: it is cen-
tered around 70 nm for sodium-based geopolymers, and
Fig. 14. Results of a diffusion test, case of GeoNa. around 35 nm for potassium-based geopolymers. Larger pores
(above 130 nm) were also detected but not quantified.
– The pore structure determined by MIP does not allow alone to
understand the results provided by transport properties exper-
Fig. 14 represents the cumulative quantity of molecular HTO in iments (direct current and diffusion) as those properties are
the downstream compartment over time. When the steady-state is much higher than expected. Indeed the tritium diffusion coeffi-
reached, the slope of this curve is constant, the effective diffusion cient is within the range 1010 m2/s, only one order of magni-
coefficient of the material can then be calculated. Again, the test tude smaller than its value in infinitely diluted solution.
was repeated to check the validity of the results. A simple – Thermoporometry combined with MIP provides a better view of
modelling using the data of the characterization tests from GeoNa not only the pore size distribution but also the pore arrange-
sample (De,HTO = 7 1010 m2/s, po = 0.555, L = 8.1 mm, Diame- ments. Geopolymers seem to behave as ink-bottled materials
ter = 50.1 mm, Co = 3.99 1010 Bq/m3) was ran. The steady-state in with pore necks dominated by mesopores connected to macro-
the theoretical case is reached within 8 h, as shown in Fig. 14. Note pores in the core of the material. The transport properties
how fast steady state is obtained. Such a fast transfer challenged appear clearly controlled by this pore family.
the applicability of Eq. (16) which assumes constant boundary con-
ditions in concentration on the two faces of the sample. The natural
Conflict of interest
diffusion tests were stopped after a few days, and the correspond-
ing effective diffusion coefficient has to be considered in an order
The authors declared that there is no conflict of interest.
of magnitude sense, with respect to the bulk diffusion coefficient
of tritiated water (2.44 109 m2/s [67]). In the case of GeoNa, we
obtained De,HTO  7 1010 m2/s. The GeoK was not tested. Acknowledgments
In summary, the measurements, whether based on an electri-
cal driving force – direct current and EIS, or a concentration dif- The authors would like to acknowledge the French National
ference – natural diffusion, indicate that the obtained transport Radioactive Waste Management Agency, Andra for supporting this
properties are much higher than expected. Such results corre- work and contributing to the funding of this research and the
spond to a pore network that cannot be made only of mesopores French Atomic Energy and Alternative Energies Commission at
as predicted by Mercury Intrusion Porometry experiments. The Marcoule (DEED/SEAD/LCBC) for the help they have brought to
second family of pores detected by thermoporometry may explain our research.
the high values of the effective diffusion coefficients obtained.
The existence of large pores in numbers is the reason why the Appendix 1. Full details of the sequences of Table 6
formation factor is much smaller than the one of a blended
cement paste. The conclusion may be that geopolymers are not See Table 10.

Table 10

mw/movd Crystallization/fusion 1st cooling sequence 1st heating sequence 2nd cooling sequence 2nd heating sequence Pf Pm
(g/g) specific enthalpy (20 °C ? 50 °C) (50 °C ? 0.5 °C) (0.5 °C ? 50 °C) (50 °C ? 20 °C)
(J/g) mc1
ice (g/g)
porosity (%) mh1
ice (g/g)
porosity (%) mc2
ice (g/g)
porosity (%) mh2
ice (g/g)
porosity (%) (%) (%)

1st sample 0.44 DhStd 0.30 34.6 0.21 24.9 0.23 27.2 0.31 35.7 78.7 69.6
DhFag 0.32 37.6 0.23 25.9 0.26 29.3 0.32 36.9 77.9 70.3
DhBru 0.40 46.5 0.25 29 0.31 35.4 0.35 40.3 76.1 72.1
DhIsi 0.32 37.6 0.23 25.9 0.26 29.3 0.32 36.9 77.9 70.2
DMat 0.36 41.8 0.23 27.3 0.28 32.2 0.33 38.4 76.9 71.1
2nd sample 0.45 DhStd 0.31 35.6 0.19 21.4 0.25 28.7 0.32 36.7 80.5 58.3
DhFag 0.34 38.7 0.19 22.4 0.27 30.8 0.33 37.9 79.7 59.1
DhBru 0.41 47.5 0.22 25.3 0.32 37.1 0.36 41.3 78.1 61.1
DhIsi 0.34 38.7 0.19 22.4 0.27 30.8 0.33 37.9 79.9 59.1
DhMat 0.37 42.8 0.21 23.7 0.29 33.8 0.34 39.5 78.8 60.1
Q.H. Nguyen et al. / Construction and Building Materials 191 (2018) 853–865 863

Table 10 (continued)

mw/movd Crystallization/fusion 1st cooling sequence 1st heating sequence 2nd cooling sequence 2nd heating sequence Pf Pm
(g/g) specific enthalpy (20 °C ? 50 °C) (50 °C ? 0.5 °C) (0.5 °C ? 50 °C) (50 °C ? 20 °C)
(J/g) mc1
ice (g/g)
porosity (%) mh1
ice (g/g)
porosity (%) mc2
ice (g/g)
porosity (%) mh2
ice (g/g)
porosity (%) (%) (%)

3rd sample 0.46 DhStd 0.30 35.2 0.19 21.4 0.25 28.3 0.31 35.8 80.3 59.8
DhFag 0.33 38.4 0.20 22.5 0.26 30.5 0.32 37.0 79.2 60.8
DhBru 0.42 47.9 0.22 25.5 0.32 36.8 0.35 40.4 77.0 63.3
DhIsi 0.33 38.4 0.20 22.4 0.26 30.4 0.32 36.9 79.2 60.8
DhMat 0.38 42.9 0.21 23.9 0.29 33.5 0.33 38.5 78.0 62.0

Appendix 2. Full details of the sequences of Table 7

See Table 11.

Table 11

mw/movd Crystallization/fusion 1st cooling sequence 1st heating sequence 2nd cooling sequence 2nd heating sequence Pf Pm
(g/g) specific enthalpy (20 °C ? 50 °C) (50 °C ? 0.5 °C) (0.5 °C ? 50 °C) (50 °C ? 20 °C)
(J/g) mc1
ice (g/g)
porosity (%) mh1
ice (g/g)
porosity (%) mc2
ice (g/g)
porosity (%) mh2
ice (g/g)
porosity (%) (%) (%)

1st sample 0.37 DhStd 0.25 30.7 0.10 11.7 0.12 15.0 0.27 32.4 48.9 36.0
DhFag 0.26 32.3 0.10 11.7 0.13 15.3 0.27 32.5 47.2 36.2
DhBru 0.30 36.9 0.10 12.0 0.13 16.0 0.27 32.8 43.5 36.7
DhIsi 0.26 32.3 0.10 11.7 0.12 15.2 0.27 32.5 47.2 36.1
DMat 0.28 34.4 0.10 11.8 0.13 15.6 0.27 32.6 45.3 36.3
2nd sample 0.36 DhStd 0.23 29.2 0.08 9.5 0.11 13.5 0.25 30.6 46.1 31.0
DhFag 0.25 30.7 0.08 9.5 0.11 13.8 0.25 30.6 44.9 31.2
DhBru 0.28 34.9 0.08 9.8 0.12 14.7 0.25 30.9 42.0 31.8
DhIsi 0.25 30.6 0.08 9.5 0.11 13.8 0.25 30.6 44.9 31.1
DhMat 0.27 32.6 0.08 9.6 0.11 14.2 0.25 30.7 43.4 31.4

Appendix 3. Fig. 15 (Full details of Fig. 10)


Appendix 4. Fig. 16 (Full details of Fig. 11).
See Fig. 15.
See Fig. 16.

Fig. 16. Pore size distribution of GeoNa from thermoporometry, calculated from all
Fig. 15. Pore size distribution of GeoK from thermoporometry, calculated from all enthalpy values (a) from the 1st cooling sequence (b): from the 2nd cooling
enthalpy values: (a) from the 1st cooling sequence (b): from the 2nd cooling sequence. sequence.
864 Q.H. Nguyen et al. / Construction and Building Materials 191 (2018) 853–865

References [31] A. Asprogerakas, A. Koutelia, G. Kakali, S. Tsivilis, Durability of fly ash


geopolymer mortars in corrosive environments, compared to that of cement
mortars, Adv. Sci. Technol. 92 (2014) 84.
[1] J. Davidovits, Synthesis of new high temperature geo-polymers for reinforced
[32] I. Ismail, S.A. Bernal, J.L. Provis, R. San Nicolas, D.G. Brice, A.R. Kilcullen, S.
plastics/composites, SPE PACTEC. 79 (1979) 151–154.
Hamdan, J.S. van Deventer, Influence of fly ash on the water and chloride
[2] J. Davidovits, The need to create a new technical language for the transfer of
permeability of alkali-activated slag mortars and concretes, Constr. Build.
basic scientific information, Proc. Symp. Held Luxemb. 10 12 June 1981,
Mater. 48 (2013) 1187–1201.
COMMISSION OF THE EUROPEAN COMMUNITIES, Luxembourg, 1981.
[33] P. Chindaprasirt, W. Chalee, Effect of sodium hydroxide concentration on
[3] D. Feng, J.L. Provis, J.S. Deventer, Thermal activation of albite for the synthesis
chloride penetration and steel corrosion of fly ash-based geopolymer concrete
of one-part mix geopolymers, J. Am. Ceram. Soc. 95 (2012) 565–572.
under marine site, Constr. Build. Mater. 63 (2014) 303–310.
[4] J. Davidovits, Geopolymer Chemistry and Applications, Geopolymer Institute,
[34] C. Gunasekara, D.W. Law, S. Setunge, Long term permeation properties of
2008.
different fly ash geopolymer concretes, Constr. Build. Mater. 124 (2016) 352–
[5] D.S. Perera, E.R. Vance, Z. Aly, J. Davis, C.L. Nicholson, Immobilization of Cs and
362.
Sr in geopolymers with Si/Al molar ratio of 2, in: Environ. Issues Waste Manag.
[35] Kupwade-Patil Kunal, N. Allouche Erez, Examination of chloride-induced
Technol. Ceram. Nucl. Ind. XI Proc. 107th Annu. Meet. Am. Ceram. Soc. Baltim.
corrosion in reinforced geopolymer concretes, J. Mater. Civ. Eng. 25 (2013)
Md. USA 2005 Ceram. Trans., Wiley-American Ceramic Society, 2005, p. 91.
1465–1476.
[6] A. Rooses, P. Steins, A. Dannoux-Papin, D. Lambertin, A. Poulesquen, F. Frizon,
[36] Z. Zhang, X. Yao, H. Zhu, Potential application of geopolymers as protection
Encapsulation of Mg–Zr alloy in metakaolin-based geopolymer, Appl. Clay Sci.
coatings for marine concrete: I. Basic properties, Appl. Clay Sci. 49 (2010)
73 (2013) 86–92.
1–6.
[7] D.W. Law, A.A. Adam, T.K. Molyneaux, I. Patnaikuni, A. Wardhono, Long term
[37] Z. Zhang, X. Yao, H. Zhu, Potential application of geopolymers as protection
durability properties of class F fly ash geopolymer concrete, Mater. Struct. 48
coatings for marine concrete: II. Microstructure and anticorrosion mechanism,
(2015) 721–731.
Appl. Clay Sci. 49 (2010) 7–12.
[8] T. Bakharev, Durability of geopolymer materials in sodium and magnesium
[38] Y. Ma, J. Hu, G. Ye, The pore structure and permeability of alkali activated fly
sulfate solutions, Cem. Concr. Res. 35 (2005) 1233–1246.
ash, Fuel 104 (2013) 771–780.
[9] T. Bakharev, Resistance of geopolymer materials to acid attack, Cem. Concr.
[39] AFPC-AFREM, Détermination de la masse volumique apparente et de la
Res. 35 (2005) 658–670.
porosité accessible à l’eau, in: Méthodes Recommandées Pour Mes. Gd. Assoc.
[10] A. Fernández-Jiménez, I. Garcia-Lodeiro, A. Palomo, Durability of alkali-
À Durabilité, INSA Toulouse, Toulouse, 1997.
activated fly ash cementitious materials, J. Mater. Sci. 42 (2007) 3055–3065.
[40] P. Kubelka, Über den Schmelzpunkt in sehr engen Capillaren, Berichte
[11] S.A. Bernal, J.L. Provis, D.G. Brice, A. Kilcullen, P. Duxson, J.S. van Deventer,
Bunsenges. Für Phys. Chem. 38 (1932) 611–614.
Accelerated carbonation testing of alkali-activated binders significantly
[41] M. Brun, A. Lallemand, J.F. Quinson, C. Eyraud, Changement d’état liquide –
underestimates service life: the role of pore solution chemistry, Cem. Concr.
solide dans les milieux poreux-II. Étude théorique de la solidification d’un
Res. 42 (2012) 1317–1326.
condensat capillaire, J. Chim. Phys. 70 (1973) 979–989.
[12] R. Zhao, J.G. Sanjayan, Geopolymer and Portland cement concretes in
[42] M. Brun, A. Lallemand, J.-F. Quinson, C. Eyraud, A new method for the
simulated fire, Mag. Concr. Res. 63 (2011) 163–173.
simultaneous determination of the size and shape of pores: the
[13] A. Fernández-Jiménez, A. Palomo, J.Y. Pastor, A. Martin, New cementitious
thermoporometry, Thermochim Acta 21 (1977) 59–88.
materials based on alkali-activated fly ash: performance at high temperatures,
[43] W.M. Haynes, CRC Handbook of Chemistry and Physics, 95th ed., Taylor &
J. Am. Ceram. Soc. 91 (2008) 3308–3314.
Francis, 2014.
[14] T. Yang, X. Yao, Z. Zhang, Quantification of chloride diffusion in fly ash–slag-
[44] G. Fagerlund, Determination of pore-size distribution from freezing-point
based geopolymers by X-ray fluorescence (XRF), Constr. Build. Mater. 69
depression, Mater. Struct. 6 (1973) 215–225.
(2014) 109–115.
[45] K. Ishikiriyama, M. Todoki, Evaluation of water in silica pores using differential
[15] N. Ganesan, R. Abraham, S.D. Raj, Durability characteristics of steel fibre
scanning calorimetry, Thermochim Acta 256 (1995) 213–226.
reinforced geopolymer concrete, Constr. Build. Mater. 93 (2015) 471–476.
[46] S. Matala, Effects of Carbonation on the Pore Structure of Granulated Blast
[16] X.-M. Cui, G.-J. Zheng, Y.-C. Han, F. Su, J. Zhou, A study on electrical
Furnace Slag Concrete, Helsinki University of Technology, 1995.
conductivity of chemosynthetic Al2O3–2SiO2 geoploymer materials, J. Power
[47] Z. Sun, G.W. Scherer, Pore size and shape in mortar by thermoporometry, Cem.
Sources 184 (2008) 652–656.
Concr. Res. 40 (2010) 740–751.
[17] S. Hanjitsuwan, P. Chindaprasirt, K. Pimraksa, Electrical conductivity and
[48] F.R.J. Rouquerol, P. Llewellyn, R. Denoyel, Calorimétrie: principes, appareils et
dielectric property of fly ash geopolymer pastes, Int. J. Miner. Metall. Mater. 18
utilisation, Éditions Tech. Ing. TIB384DUO (2012).
(2011) 94–99.
[49] W. Hemminger, S. Sarge, The baseline construction and its influence on the
[18] F.A. Dullien, Porous Media: Fluid Transport and Pore Structure, second ed.,
measurement of heat with differential scanning calorimeters, J. Therm. Anal.
Academic press, 1992.
Calorim. 37 (1991) 1455–1477.
[19] A. Delagrave, J. Marchand, M. Pigeon, Influence of microstructure on the
[50] U. Bandara, A systematic solution to the problem of sample background
tritiated water diffusivity of mortars, Adv. Cem. Based Mater. 7 (1998) 60–65.
correction in DSC curves, J. Therm. Anal. Calorim. 31 (1986) 1063–1071.
[20] K.A. Snyder, C. Ferraris, N.S. Martys, E.J. Garboczi, Using impedance
[51] M. Brun, A. Lallemand, J.F. Quinson, B. Martinie, J. Rossignol, B. Rasneur, C.
spectroscopy to assess the viability of the rapid chloride test for determining
Eyraud, Changement d’état liquide solide dans les milieux poreux-III.
concrete conductivity, J. Res. Natl. Inst. Stand. Technol. 105 (2000) 497.
Application à une méthode calorimétrique de détermination du spectre des
[21] K.A. Snyder, The relationship between the formation factor and the diffusion
rayons, de pores, J. Chim. Phys. 70 (1973) 990–996.
coefficient of porous materials saturated with concentrated electrolytes:
[52] Q.H. Nguyen, S. Lorente, A. Duhard-Barone, Modeling the ionic transport in
theoretical and experimental considerations, Concr. Sci. Eng. 3 (2001) 216–
mesoporous materials, Constr. Build. Mater. Revis. (2018).
224.
[53] H. Mercado, S. Lorente, X. Bourbon, Chloride diffusion coefficient: a
[22] P.J. Tumidajski, A.S. Schumacher, S. Perron, P. Gu, J.J. Beaudoin, On the
comparison between impedance spectroscopy and electrokinetic tests, Cem.
relationship between porosity and electrical resistivity in cementitious
Concr. Compos. 34 (2012) 68–75.
systems, Cem. Concr. Res. 26 (1996) 539–544.
[54] T. Wattez, A. Duhart-Barone, S. Lorente, On, the use of a migration technique to
[23] P. Rovnaník, Effect of curing temperature on the development of hard
measure the diffusion coefficient of cement-based materials, Seventh Int. Conf.
structure of metakaolin-based geopolymer, Constr. Build. Mater. 24 (2010)
Concr. Sev. Cond., Nanjing (2013).
1176–1183.
[55] K.D. Stanish, R.D. Hooton, M.D.A. Thomas, Testing the chloride penetration
[24] P. Steins, A. Poulesquen, F. Frizon, O. Diat, J. Jestin, J. Causse, D. Lambertin, S.
resistance of concrete: a literature review, FHWA Contract DTFH 61 (1997) 19–
Rossignol, Effect of aging and alkali activator on the porous structure of a
22.
geopolymer, J. Appl. Crystallogr. 47 (2014) 316–324.
[56] O. Amiri, A. Aıt-Mokhtar, P. Dumargue, G. Touchard, Electrochemical
[25] D.L. Kong, J.G. Sanjayan, K. Sagoe-Crentsil, Comparative performance of
modelling of chlorides migration in cement based materials. Part II:
geopolymers made with metakaolin and fly ash after exposure to elevated
experimental study—calculation of chlorides flux, Electrochim. Acta 46
temperatures, Cem. Concr. Res. 37 (2007) 1583–1589.
(2001) 3589–3597.
[26] D. Medpelli, J.-M. Seo, D.-K. Seo, Geopolymer with hierarchically
[57] E. Samson, J. Marchand, K.A. Snyder, Calculation of ionic diffusion coefficients
meso-/macroporous structures from reactive emulsion templating, J. Am.
on the basis of migration test results, Mater. Struct. 36 (2003) 156–165.
Ceram. Soc. 97 (2014) 70–73.
[58] F. Frizon, S. Lorente, C. Auzuech, Nuclear decontamination of cementitious
[27] C. Boher, I. Martin, S. Lorente, F. Frizon, Experimental investigation of gas
materials by electrokinetics: an experimental study, Cem. Concr. Res. 35
diffusion through monomodal materials. Application to geopolymers and
(2005) 2018–2025.
VycorÒ glasses, Microporous Mesoporous Mater. 184 (2014) 28–36.
[59] P. Atkins, J. de Paula, Physical Chemistry, ninth ed., W. H. Freeman, 2009.
[28] V. Benavent, F. Frizon, A. Poulesquen, Effect of composition and aging on the
[60] H. Mercado-Mendoza, S. Lorente, X. Bourbon, The diffusion coefficient of ionic
porous structure of metakaolin-based geopolymers, J. Appl. Crystallogr. 49
species through unsaturated materials, Transp. Porous Media 96 (2013) 469–
(2016) 2116–2128.
481.
[29] R.R. Lloyd, J.L. Provis, J.S. Van Deventer, Pore solution composition and alkali
[61] T. Wattez, A. Duhart-Barone, S. Lorente, Modeling of nuclear species
diffusion in inorganic polymer cement, Cem. Concr. Res. 40 (2010) 1386–1392.
diffusion through cement-based materials, Transp. Porous Media 98 (2013)
[30] E. Jämstorp, M. Strømme, G. Frenning, Modeling structure–function
699–712.
relationships for diffusive drug transport in inert porous geopolymer
[62] P. Locoge, M. Massat, J.P. Ollivier, C. Richet, Ion diffusion in microcracked
matrices, J. Pharm. Sci. 100 (2011) 4338–4348.
concrete, Cem. Concr. Res. 22 (1992) 431–438.
Q.H. Nguyen et al. / Construction and Building Materials 191 (2018) 853–865 865

[63] R.L. Day, B.K. Marsh, Measurement of porosity in blended cement pastes, Cem. [66] H. Mercado-Mendoza, S. Lorente, X. Bourbon, Ionic aqueous diffusion through
Concr. Res. 18 (1988) 63–73. unsaturated cementitious materials–a comparative study, Constr. Build.
[64] M.-S. Ko, H.-Y. Chen, S.-J. Lyu, T.-T. Wang, T.-H. Ueng, Permeation Mater. 51 (2014) 1–8.
characteristics and impact factors of geopolymers made of kaolin, Constr. [67] J.H. Wang, C.V. Robinson, I.S. Edelman, Self-diffusion and structure of liquid
Build. Mater. 93 (2015) 301–308. water. III. Measurement of the self-diffusion of liquid water with H2, H3 and
[65] J. Rouquerol, D. Avnir, C.W. Fairbridge, D.H. Everett, J.M. Haynes, N. Pernicone, O18 as Tracers1, J. Am. Chem. Soc. 75 (1953) 466–470.
J.D.F. Ramsay, K.S.W. Sing, K.K. Unger, Recommendations for the [68] K.L. Scrivener, A. Nonat, Hydration of cementitious materials, present and
characterization of porous solids (Technical Report), Pure Appl. Chem. 66 future, Cem. Concr. Res. 41 (2011) 651–665.
(1994) 1739–1758.

You might also like