Construction and Building Materials 407 (2023) 133511

Contents lists available at ScienceDirect

Construction and Building Materials

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

Durability parameters of three low-carbon concretes (low clinker,

alkali-activated slag and supersulfated cement)
Lola Doussang a, Gabriel Samson a, *, Fabrice Deby a, Bruno Huet b, Emmanuel Guillon b,
Martin Cyr a
a
Laboratoire Matériaux et Durabilité des Constructions (LMDC), INSAT/UPS Génie Civil, 135 Avenue de Rangueil, 31077 Toulouse Cedex 04 France
b
Holcim Innovation Center, 95 rue du Montmurier, F-38070 Saint Quentin Fallavier, France

A R T I C L E I N F O A B S T R A C T

Keywords: The general transfer properties (water porosity, mercury intrusion porosimetry, water permeability, capillary
Low-carbon concrete absorption, gas permeability and resistivity) and chloride penetration resistance (natural chloride diffusion test
Concrete durability and rapid chloride permeability test) of three low-carbon concretes C25/30 (low clinker (LCK) concrete, alkali-
Chloride transport
activated slag with sodium carbonate (Na2CO3-AAS) concrete and supersulfated cement (SSC) concrete) were
Transfer properties
evaluated and compared, in order to identify the strengths and weaknesses of each technology. The results show
that the LCK concrete is characterized by “excellent” transfer properties (low water porosity, gas permeability,
water permeability and capillary absorption) but high viscosity in the fresh state and very high chloride
permeability. AAS and SSC concretes have similar transfer properties, comparable to classical concretes C25/30,
and are extremely resistant to chloride penetration. Finally, the use of traditional protocols for durability tests
applied to alternative low-carbon binders is questioned and metrology adaptations are proposed.

1. Introduction The LCK concrete reduces the proportion of Portland cement in the
binder by partially replacing it by limestone filler to maintain an
The need to reduce the carbon dioxide (CO2) emissions in the con­ acceptable rheology (binder made of 27% CEM I and 73% limestone
struction field by the development of more sustainable concretes has filler in this study). The use of limestone filler as an addition in concrete
been widely demonstrated [1,2]. was allowed for the first time in 1965 in Germany, and later, in 2000, by
The main strategy is to reduce the proportion of Portland cement in the European standard EN 197–1 [6] as Portland limestone cements
the binder, by partially replacing it by supplementary cementitious CEM II/A-L (6–20% limestone content) and CEM II/B-L (21–35% lime­
materials (SCMs), which are more environmentally friendly, or even to stone content) [7]. Information is available in the literature on Portland
use alternative cements without Portland cement [1,3,4]. The present concretes with limestone additions and their durability properties.
study focuses on three low-carbon concrete technologies of the same However, in most cases, levels of replacement by fillers ranging from
strength class, C25/30, among those available on the market: low 0 to 35% have been extensively studied because they correspond to
clinker (LCK) concrete, alkali-activated slag with sodium carbonate standard recommendations (limestone content up to 50% in some
(Na2CO3-AAS) concrete and supersulfated cement (SSC) concrete. More studies). Nevertheless, it remains lower than the 73% filler content of
specifically, attention is paid to their durability: their general transfer the binder in the LCK concrete of this study. The mechanical strengths of
properties and chloride penetration resistance. The transfer properties of Portland limestone concretes, modulus of elasticity, porosity, water
a concrete are of primary importance as they are largely responsible for absorption and durability properties (chloride migration and diffusion,
determining the time taken for external aggressive agents, such as carbonation kinetics, sulfate or freeze–thaw resistance, creep and
chlorides, to reach the reinforcement and potentially initiate corrosion. shrinkage) are generally characterized and well-documented for low
These two phenomena (transport of aggressive species and corrosion of filler contents [7–11]. What emerges is that, when the amount of filler in
reinforcement) can be further accelerated by crack propagation during the binder is increased, the concrete properties remain unchanged for
the service life of concrete [5]. 15–20% cement replacement, while the mechanical and durability

* Corresponding author.
E-mail address: [email protected] (G. Samson).

https://doi.org/10.1016/j.conbuildmat.2023.133511
Received 28 April 2023; Received in revised form 19 September 2023; Accepted 22 September 2023
Available online 4 October 2023
0950-0618/© 2023 Elsevier Ltd. All rights reserved.
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

performances decrease for higher replacement levels (greater than 20% concern a very different strength class, thereby making the comparison
fillers) [8–10]. difficult between the mixtures and the prediction of their durability. The
The AAS concrete in this study was made with slag as the alumino­ present study aims to fill this gap, by providing data on the general
silicate precursor, combined with a sodium carbonate alkaline activator. transfer properties (water porosity, mercury intrusion porosimetry,
The sodium carbonate was chosen as the activator for its wider avail­ water permeability, capillary absorption, gas permeability and re­
ability, and its more eco-friendly manufacturing process, lower cost and sistivity) and transport of chlorides (natural chloride diffusion test and
lower health risk when used (as opposed to the handling of highly rapid chloride permeability test) in the three low-carbon concretes
alkaline solutions for sodium silicate or sodium hydroxide in liquid considered (LCK, Na2CO3-AAS and SSC). These three concretes were
form) [12,13]. AAS have been extensively studied over the years selected because they are promising alternatives for reducing green­
[14–18]: and their hydration mechanisms, mechanical properties, gen­ house gas emissions and were designed to have equivalent rheology and
eral transfer properties and durability properties are well documented in strength (self-compacting concrete, C25/30). Thus, in addition to their
the literature [13,19–23]. However, it should be noted that, in the large durability characterization, a comparison of their properties is pro­
majority of cases, these studies are performed for activations based on posed, in order to identify the strengths and weaknesses of each
sodium silicate or hydroxide. In comparison, studies on sodium technology.
carbonate-based activation (like the one used in this study) are much
more limited, although this activator presents the lowest carbon foot­ 2. Materials
print [24]. Moreover, these studies are not often conducted at the con­
crete scale (rather paste or mortar) [13,20,25]. However, the literature 2.1. Raw materials
agrees that the activator strongly influences the properties of the AAS
[13,20,26,27] and data on the scale of the concrete are needed to The chemical composition of cements and slag used in LCK, AAS and
accurately predict the long-term durability of this Na2CO3-AAS. Finally, SSC concretes determined by X-ray fluorescence (XRF) is given in
the literature agrees that AAS (regardless of the activator) have me­ Table 1. The quicklime that is part of the composition of AAS is only CaO
chanical properties comparable to those of Portland cements, together (100%). Other raw materials were also used, such as limestone filler
with generally lower porosity than OPC matrices, with smaller pore CaCO3, sodium carbonate Na2CO3 and anhydrite CaSO4. The carbon
sizes, reducing their overall transfer properties and favouring greater footprints from different databases of these raw materials are summa­
durability [21,23,26]. Better resistance to chloride penetration and rized in Table 2.
sulphate attack has also been demonstrated [20–22,26]. On the other The binder of the LCK was made with 27% CEM I 52.5 N CE CP2 and
hand, a higher susceptibility to shrinkage and carbonation than in OPC 73% limestone filler Betocarb HP 300 Sassenage from Omya. A Blaine
was identified [20–22]. The capacity of these matrices to bind chlorides specific surface of 3900 cm2/g for the CEM I and 5490 cm2/g for the
remains poorly quantified and needs further insight [27]. Of the fifty or filler were measured. The semi-crushed alluvial mix of aggregates
so articles reviewed by Zhang et al. [19], only four studies on chloride ranged from 0 to 20 mm. Aggregate proportions were optimized to
transport were found and none on carbonation – on Na2CO3-AAS, for achieve high packing to provide higher mechanical properties, com­
example (the same trend was found in the review by Osio-Norgaard et al. bined with the low water/binder ratio of 0.25. Water reducing agents
[27] focusing on chloride). Therefore, this study aims to position the were also used to improve the workability in the fresh state.
durability performance of Na2CO3-AAS concretes, for which information The AAS binder was composed of 95% Ground Granulated Blast
remains limited compared to other activation systems. Furnace Slag (GGBS) from Ecocem Fos/Mer as a precursor, with 4%
Finally, the composition of the SSC binder is defined by the European sodium carbonate and 1% quicklime as activators. The latter provided a
standard EN 15743 [28]. It consists of a mix of at least 75% slag com­ high pH to initiate dissolution and hydration of the slag. A Blaine spe­
bined with 5–20% calcium sulphate and an alkaline activator (0–5% cific surface of 5500 cm2/g was measured for the GGBS. The semi-
clinker) to obtain a sufficient reactivity. The activation of slag by sulfates crushed alluvial mix of aggregates ranged from 0 to 16 mm. This con­
was described in 1908 by Kühl [14], but SSC products developed crete had a water/binder ratio of 0.43. A commercial additive was also
particularly in Europe after the Second World War (1940–1960), due to used as a water reducing agent.
a lack of Portland cement leading to the standardization of this binder Finally, the SSC binder was composed of 89% GGBS from Ecocem
[29]. The hydration mechanisms involved, hydration products, micro­ Fos/Mer as the precursor, combined with 8% anhydrite of Lorraine and
structure, mechanical properties and sulfate resistance of SSC have been 3% CEM III/B 42.5 N La Malle as alkaline activators to obtain sufficient
extensively studied in the literature, in the majority of cases on paste reactivity. Finally, a limestone filler Orgon BL200 was added to the
samples [30–35]. Data concerning their general transfer properties mixture during concrete formulation. The semi-crushed alluvial mix of
(porosity, density, capillary absorption, gas permeability, resistivity) aggregates ranged from 0 to 16 mm. This concrete had a water/binder
and their durability performance (shrinkage, freeze–thaw resistance, ratio of 0.50 and was combined with water reducing agents.
accelerated carbonation, chloride migration, diffusion and binding, The three low-carbon concretes studied were designed to be C25/30
sulfate attack, acid attack, alkali-aggregate reaction) are also available. and also self-compacting (consistency class S5). Their respective com­
In summary, from a durability point of view, authors agree that SSC positions and characteristics are detailed in Table 3. The compressive
concrete performs extremely well in water saturated and chemically strengths given in Table 3 were obtained on cylindrical samples.
aggressive environments (high resistance to sulfate attack, acid attack, The carbon footprint of each concrete (without reinforcement) can
alkali-aggregate reaction and chloride penetration) but is more suscep­ be calculated directly by adding up the CO2 footprints of all raw mate­
tible to carbonation. However, a limited number of studies have been rials (as presented in Table 2) that are part of its composition (as defined
carried out at the concrete scale [36–41]. in Table 3) only considering their proportion in the formulation. The
In summary, some data about the transfer properties and transport of results are presented in Table 3 and can be ranked as follows: 122, 104
chlorides in the three low-carbon concrete technologies studied are and 50 kg eq. CO2/m3 for LCK, AAS and SSC respectively. The result
available in the literature. However, it is important to note that the data associated with a standard concrete made with CEM I of similar strength
available for LCK consider a lower Portland cement substitution (0–50% (C25/30) is around 207 kg eq. CO2/m3. Therefore, with their signifi­
filler) than that considered in this study (73% filler), which is going to cantly reduced carbon footprint (a reduction of at least 40% with respect
influence the measured concrete properties. The same situation applies to CEM I concrete) for equivalent performances, the three concretes of
to AAS, where data are available but for a different activator than in this this study are considered as “low- carbon”.
study (sodium carbonate). Furthermore, the tests are not always carried
out at the concrete scale, which is the most representative of reality, or

2
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

Table 1
Chemical composition of the cements and slag used in LCK, AAS and SSC concretes (% in mass) determined by XRF.
Chemical composition (% in mass)

CaO SiO2 Al2O3 MgO Fe2O3 K2O Na2O SO3 TiO2

CEM I 64.7 20.4 3.9 0.8 5.0 0.6 0.1 2.8 0.2
CEM III/B 50.5 37.7 8.1 4.5 1.9 0.5 0.2 2.0 0.5
GGBS 42.9 37.7 10.3 6.5 0.3 0.4 0.2 1.6 0.7

Table 2 Table 3
Carbon footprint (in kg eq. CO2/t) of the raw materials used in LCK, AAS and Concrete composition, mixture parameters and characteristics in fresh and
SSC concretes from different databases. hardened state of LCK, AAS and SSC together with CO2 footprint of each con­
CO2 footprint(1) (kg eq. CO2/t)
crete, calculated by adding the CO2 footprints of all raw materials forming part
of the mixture (Table 2), considering their proportions.
CEM I 765 (2)
Limestone filler 6 (3) unit LCK AAS SSC
GGBS 100 (2) Concrete CEM I kg/m3 136.8
Sodium carbonate 1350 (3) composition Limestone filler kg/m3 375.0 49.1
Quicklime 893 (3) GGBS kg/m3 481.7 329.3
Anhydrite 14 (3) Na2CO3 kg/m3 20.3
CEM III/B 273 (2) Quicklime kg/m3 5.1
Water reducing agent 1850 (4) Anhydrite kg/m3 29.6
Sand and aggregates 2.3 (5) CEM III/B kg/m3 11.1
(1) Water reducing kg/m3 6.02 11.15 5.1
Value of CO2 footprint known as A1 (CO2 associated with the material) in
agent
the Environmental and Health Data Sheet method. Neither the transport of 3
Sand (0–5 mm) kg/m 678.5 686.5 739.7
raw materials (A2) nor the concrete manufacturing stage (A3) is evaluated. Fine aggregates kg/m3 272.1 609.6 656.9
(2)
French values from SFIC (Syndicat Français de l’Industrie Cimentière) (5–10 mm)
available on [42]. Coarse kg/m3 827.1 229.6 247.4
(3)
From the Swiss database ECOInvent [43]. aggregates
(4) (>10 mm)
Average value used by contractors regardless of the products and their
concentration. Total water kg/m3 130.0 220.0 210.0
(5)
from the French database INIES [44]. Mixture parameters w/c ratio (1) 0.95 0.43 0.57
w/b ratio (2) 0.25 0.43 0.50
weff/b ratio 0.22 0.40 0.46
2.2. Mixing and curing Theoretical kg/m3 2425.5 2263.8 2278.3
specific gravity
Quantity of L/m3 318 407 360
For each concrete, the mixing procedure lasted around 5 min and
paste
followed the same steps. First, the dry aggregates were dry mixed alone CO2 footprint kg eq. 122 104 50
in a pan mixer for 1 min to achieve a uniform dispersion. Then, 1/3 of CO2/m3
the water was added and the mixing continued for 1 min more. After Concrete Inversed cone seconds 9.7 1.7 2.0
that, the pan mixer was stopped, the binder was added and mixing characteristics in flow rate
fresh and (rheology)
continued for 1 min. Finally, the remaining 2/3 of the water with water
hardened state Abrams cone mm 525 580 490
reducing agent was added with a further 2 min of mixing. Then, the spread
concrete was cast in cylindrical (110 mm diameter × 220 mm) moulds in Air content % 1.3 1.2 1.8
3 layers, each one being vibrated, covered by a plastic cover to avoid 28 days MPa 33.0 33.3 32.0
compressive
evaporation, and stored in a wet curing room with 95% relative hu­
strength
midity (RH). The specimens were demoulded after 72 h and cured in the 90 days MPa 34.4 37.5 40.0
same room for 28 days or other duration, depending on the testing date. compressive
strength
3. Test methods (1)
The w/c ratio does not consider the inert filler in the cement part.
(2)
The w/b ratio considers the inert filler in the binder, unlike the w/c ratio.
All the tests were performed on concrete samples after curing in a
wet curing room with 95% relative humidity. The values presented are 50 ◦ C chosen for the modified test to prevent damage to slag-based
an average of several measurements on different samples for repeat­ samples (e.g. ettringite in SSC). Also, for the modified procedure, the
ability. All the tests performed are summarized in Table 4. sample thickness was adapted from 50 mm to 17 mm to facilitate the
drying at 50 ◦ C, as recommended by Cyr [36]. More details on the choice
of these new parameters are given in the Results (4.1.1) and Discussion
3.1. General transfer properties
(5.1) sections.
3.1.1. Water porosity
3.1.2. Mercury intrusion porosity
Water porosity was measured according to NF P18-459 [45], on
MIP was carried out at 28 days of curing, on concrete samples from
concrete samples after 28 days, 90 days, 6 months and 1 year of curing.
cylindrical specimens 110 mm diameter × 220 mm crushed to select
This test was repeated as the “water porosity modified test” (except for
only pieces of mortar around 5 mm diameter. These pieces of “mortar”
the age of 90 days), so as to adapt it to the AAS and SSC concretes of this
from concrete samples were conditioned for 10 days in a vacuum
study. These adaptations were motivated by the fact that different au­
desiccator before testing. Tests were done with 2 g of sample and
thors had drawn attention to the drying temperature used for pre­
repeated 3 times for each mix composition to obtain representative
conditioning samples made with alternative binders to Portland cement
results.
[36,53,54]. The traditional water porosity test from NF P18-459 re­
quires a drying period at 105 ± 5 ◦ C, instead of the lower temperature of

3
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

Table 4
Summary of tests performed on LCK, AAS and SSC concretes to characterize their general transfer properties, pore solution and resistance to chloride penetration with
the associated standards, together with curing, sample size and preconditioning.
Test Standard Sample Sample shape and size Sample preconditioning and details Number of
curing samples for
average

GENERAL Water porosity NF P18-459 [45] 28 days, 6 Cylindrical 110 mm diameter × Drying step at 105 ◦ C 3
TRANSFER (ε) months, 1 220 mm, sawn to a thickness of 50
PROPERTIES year mm, after discarding the 1 cm-top
and bottom
Water porosity Based on NF P18-459 28 days, 6 110 mm diameter × 220 mm, Drying step at 50 ◦ C instead of 105 ◦ C, 9
modified (lower drying months, 1 sawn to a thickness of 17 mm to avoid destabilization of hydrates
temperature) year instead of 50 mm, after discarding such as ettringite or deterioration of
the 1 cm-top and bottom the microstructure
Mercury 28 days 110 mm diameter × 220 mm Vacuum desiccator for 10 days before 3
intrusion crushed to select only pieces of testing
porosimetry paste around 5 mm diameter
Water EN 12390–8 [46] 28 days 100 mm cubes 3
permeability
Capillary iso – 15,148 [47] 28 days 110 mm diameter × 220 mm, Side surfaces sealed with a waterproof 2
absorption (Aw) sawn to a thickness of 50 mm, Sicomin resin
after discarding the 1 cm-top and
bottom
Gas permeability XP P18-463 [48] 90 days 110 mm diameter × 220 mm sawn Side surfaces sealed with Devcon bi- 3
(k) to a thickness of 50 mm, after component epoxy resin
discarding the 1 cm-top and
bottom
Gas permeability Based on XP P18-463 90 days 110 mm diameter × 220 mm sawn Drying until mass stabilization at 3
modified (lower drying to a thickness of 50 mm, after 45 ◦ C, 65 ◦ C, 80 ◦ C and 105 ◦ C to
temperature) discarding the 1 cm-top and highlight the influence of drying
bottom temperature on the results
Resistivity (ρ) RILEM TC-154 EMC 28 days, 90 Direct measurement on concrete Measurements made directly after 3
recommendations [49] days 110 mm diameter × 220 mm removal from the wet curing room (no
ground flat on both ends prior to saturation under vacuum)
testing
PORE SOLUTION Pore solution Based on the protocol 38 days Cylindrical 33 mm diameter × 70 1
extraction described in [50] mm pastes samples
CHLORIDE Natural chloride EN 12390–11 [51] 28 days, 90 On half 110 mm diameter × 220 All surfaces covered with chloride ion 2
TRANSPORT diffusion test days mm diffusion proof Sicomin resin, except
(Dnss) the exposure surface
RCPT test, charge ASTM C1202 [52] 28 days, 90 110 mm diameter × 220 mm sawn 3
passed (Q) days to a thickness of 50 mm, after
discarding the 1 cm-top and
bottom

3.1.3. Water permeability and capillary absorption Measurements were taken immediately after the sample was removed
Water permeability and capillary absorption were assessed after 28 from the wet curing room, to avoid drying, which could influence the
days of curing, following EN 12390–8 [46] and ISO-15148 [47] stan­ values of resistivity obtained. The material was considered sufficiently
dards, respectively. Before capillary absorption, the specimens were saturated after the curing period and, for this reason, the step of satu­
stored under the test conditions until the mass of each specimen, ration under vacuum was not performed.
measured over 24 h, had stabilized to within ± 0.1% of its total mass.
The main parameters of these tests are summarized in Table 4.
3.2. Pore solution extraction
3.1.4. Gas permeability
Gas permeability was measured following XP P18-463 [48], on A pore solution extraction was performed on pastes aged for 38 days
samples having undergone 90 days of curing. A CEMBUREAU device in order to measure the pH. The ionic concentration of the pore solution
was used for the experiment. The permeability was then determined was also determined by inductively coupled plasma (ICP) analysis. Paste
from the flow measurement in the steady state. A modified procedure of samples were selected (rather than mortar or concrete) to extract a
this test was also proposed for AAS and SSC concretes, with lower drying maximum of pore solution by this method, in the absence of aggregates.
temperatures but keeping the sample thickness of 50 mm this time, for A Sintco 600 T press was used to perform the extraction as described in
the same reasons as for the water porosity test. For AAS and SSC sam­ greater detail in [50]. The pressure was applied with a speed of 3 kN/s
ples, the gas permeability measurements were performed with an up to 1000 kN, where it was maintained for 300 s to extract some mil­
adapted drying protocol, after mass stabilization at 45 ◦ C, 65 ◦ C, 80 ◦ C lilitres of pore solution. The pore solution was then filtered and imme­
and 105 ◦ C. This modified procedure was not applied for the LCK con­ diately analysed with ICP optical emission spectrometry to quantify Na,
crete, as it is a Portland cement-based binder, considered adapted to the K, Fe, Al, Mg, Ca, Si species.
standard preconditioning.

3.1.5. Resistivity 3.3. Chloride transport

Direct electrical resistivity measurements on non-reinforced concrete
samples 110 mm diameter × 220 mm that had been ground flat at both Chloride transport through the concrete was evaluated in both nat­
ends prior to testing, were taken at 28 and 90 days of curing. The test set- ural and accelerated conditions, by a natural chloride diffusion test and
up was based on RILEM TC-154 EMC [49] using a resistance meter. by the Rapid Chloride Permeability Test (RCPT), respectively.

4
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

3.3.1. Chloride diffusion test

Chloride transport in natural conditions was characterized by a
diffusion test performed according to EN 12390–11 [51], on concrete
after 28 and 90 days of curing. Five half-samples per mixture (two for
the test and three to follow the evolution of chloride penetration in
concrete by silver nitrate (AgNO3) pulverization) were immersed in 50 L
of sodium chloride (NaCl) solution at 30.93 g/L representing natural
seawater. The NaCl solution was renewed every 90 days. One half-
sample was conditioned in a sealed plastic bag after sawing so that the
initial chloride content in the concrete could be analysed. The reference
period for chloride exposure was 90 days, but it was necessary to extend
this immersion duration in some cases as chlorides had to penetrate
sufficiently far into the material to enable grinding of at least 8 parallel
layers of concrete, which were required to achieve the chloride profile.
Thus, for samples cured for 28 days, 91 days of immersion in 3% NaCl
solution were sufficient to obtain the chloride profile. For samples cured
for 90 days, 96 days of immersion were sufficient for LCK concrete, but
longer chloride exposure was required for AAS (454 days) and SSC (184
days) concretes. Once chlorides had penetrated the material sufficiently,
profile grinding was carried out, followed by chloride analysis. Total
chlorides were determined following standard NF EN 14629 [55], while
free chlorides were determined according to the French procedure rec­ Fig. 1. Water porosity, ε, for LCK (in green), AAS (in red) and SSC (in yellow)
ommended by GranDuBé [56]. The latter consists of placing 5 g of concretes after 28 days, 90 days, 180 days and 365 days of wet curing, with a
powder in a beaker with 200 mL of demineralized water and stirring for drying step of either 105 ◦ C and a sample thickness of 50 mm (coloured lines) or
3 min. The solution is then filtered, rinsed with only 10 mL of demin­ 50 ◦ C and 17 mm (black lines). (For interpretation of the references to colour in
eralized water and acidified with 2 mL of nitric acid. Bound chlorides are this figure legend, the reader is referred to the web version of this article.)
calculated as the difference between total chlorides and free chlorides.
Finally, the chloride profile obtained enabled the non-steady state Moreover, porosity of AAS and SSC may have been over-estimated
diffusion coefficient (Dnss) to be determined, and the chloride content at because the drying step at 105 ◦ C could have destabilized hydrates or
the exposed concrete surface (Cs) to be calculated by regression analysis. damaged the concrete microstructure [53], as detailed in part 5.1 of the
Discussion.
3.3.2. RCPT test For this reason, water porosity measurements were repeated with a
Chloride transport in accelerated conditions was assessed by the lower drying temperature of 50 ◦ C, a choice based on literature data
RCPT test (Rapid Chloride Permeability Test), following ASTM C1202 [36,53,59–62]. The results are presented and compared to those with a
[52]. The standard voltage of 60 V was applied and the test was run for 6 drying step of 105 ◦ C in Fig. 1. A decrease of porosity for AAS (16.2% to
h. This test gave the charge that passed through each concrete, which 16.1% to 15.8%) and SSC (16.4% to 15.9% to 14.2%) concretes can be
can be related to different levels of resistance to chloride ion penetra­ observed between 28 days, 180 days and 365 days. This result was ex­
bility. The main parameters of this test are summarized in Table 4. pected because the hydration of slag is slower than that of cement and is
not complete at 28 days. This trend was not found for results obtained
4. Results with the drying step of 105 ◦ C, which reinforced the hypothesis that this
temperature was too high for slag-based concretes and probably
The results are deliberately presented here in a rather succinct damaged the samples, leading to a misleading porosity assessment. The
manner. A more detailed analysis is proposed in the Discussion section. porosity of LCK concrete remained relatively stable between 28 days,
180 days and 365 days (10.6% to 9.6% to 9.8%), because the hydration
4.1. General transfer properties of cement was already almost completed at 28 days. No significant effect
of the drying temperature was observed on LCK water porosity, which
4.1.1. Water porosity was expected and suggests that the temperature of 50 ◦ C is sufficient to
Results of water porosity with a drying step at 105 ◦ C or 50 ◦ C are dry a sample with a reduced thickness of 17 mm.
presented in Fig. 1 for different periods of curing.
The porosity of concretes obtained with a drying step of 105 ◦ C 4.1.2. Mercury intrusion porosity
remained stable over the time for the three formulations studied Mercury intrusion porosity (MIP) was performed to complete the
(considering the error bars). This was also the case for slag-based con­ information provided by the water porosity test, such as the pore size
cretes (AAS and SSC), known to have a slower hydration than CEM I- distribution, as illustrated on Fig. 2 and Fig. 3.
concretes [57] and for which a decrease in porosity could have been Different pore size distributions could be identified between the
expected between 28 days and 90 days due to a higher hydration level. mortars extracted from the three concretes studied (Fig. 2 and Fig. 3).
This result can possibly be explained by a drying step of 105 ◦ C in the LCK concrete was the least porous concrete (10% at 28 days from water
experimental protocol, which is not suitable for slag-based concretes and porosity measurement), but had larger pore size measured on extracted
therefore gives unreliable results, as developed in the discussion in part mortar (average pore diameter: 23 nm) whose distribution was centred
5.1. These water porosity values are consistent with those found in the around 40 nm. AAS and SSC had a higher concrete water porosity
literature: 9% for Portland limestone concrete (50–65% limestone) [9], (around 16% at 28 days) with smaller pore size measured on extracted
16–17% for SSC [37] and 14% for Na2CO3-AAS concrete of higher mortars (average pore diameter: 13 nm for both). AAS presented a
strength class (65 MPa at 28 days instead of 30 MPa here) [58]. bimodal pore size distribution centred around 9 and 21 nm, while SSC
The porosity of LCK concrete was significantly lower than that of distribution was centred around 12 nm for the pore entry diameter. The
AAS and SSC. This seems logical considering the very low w/b ratio of pore size distribution, presented on Fig. 3, also clearly highlights that
LCK concrete (0.25 against 0.43 and 0.50, respectively) and the fact that AAS and SSC have a similar porous network. Moreover, Fig. 3 shows
an aggregate optimization was performed to achieve high packing. that, for the three mixtures, about 45% of the porosity is accessible

5
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

Fig. 2. (a) Pore-size distribution and (b) cumulative mercury volume from mercury intrusion for LCK (in green), AAS (in red) and SSC (in yellow) mortar samples
removed from concrete after 28 days of cure. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of
this article.)

4.1.3. Water permeability and capillary absorption

The water permeability and capillary absorption of the three con­
cretes are presented in Fig. 4.
The water permeability of slag-based concretes (AAS and SSC) had
the same order of magnitude (Fig. 4). This seems reasonable as they
already had similar porosity, and porosity can influence permeability.
Capillary absorption was slightly higher for AAS than for SSC. For LCK
concrete these values were about one third of those of slag-based con­
cretes. This can mainly be explained by LCK’s lower porosity, which
offsets the large pore entry diameter (section 4.1.2) and thus reduces the
water transport. Therefore, LCK concrete is less permeable to water than
AAS and SSC.

4.1.4. Gas permeability

The gas permeability of the three concretes previously dried at
different temperatures is presented in Fig. 5. Contrary to what was done

Fig. 3. Contribution of the different pore sizes to the total measured mercury
porosity (inspired from [59]) for LCK, AAS and SSC mortar samples removed
from concrete after 28 days of cure, complementary to the one shown in Fig. 3.

through pores having diameters between 10 and 50 nm (49%, 43% and

43% of its total porosity for LCK, AAS and SSC, respectively), corre­
sponding to medium capillary pores [63]. Then, there are some differ­
ences between LCK and slag-based binders for the smaller and the larger
pore size families. Slag-based binders have twice the porosity accessible
through pores of diameter between 2.5 and 10 nm, corresponding to
small capillaries of gel pores, (14%, 36% and 34% of the total porosity
for LCK, AAS and SSC, respectively), while LCK has more porosity
accessible by pores of diameter greater than 50 nm, corresponding to
large capillary pores (36%, 20% and 23% of its total porosity for LCK,
AAS and SSC, respectively) that favour the transport of water or other
aggressive species. These orders of magnitude correspond to the litera­
ture data for Na2CO3-AAS concrete [59,64] and SSC concrete [37] and
also confirm that slag-based binders have smaller pores than those made
with only Portland cement, as reported in the literature [21,26]. These
characteristics regarding the total porosity and the pore size distribution
specific to each concrete are going to influence the transport and Fig. 4. Water permeability (in blue) and capillary absorption (in black)
durability properties studied later. measured on LCK, AAS and SSC concretes after 28 days of curing. (For inter­
pretation of the references to colour in this figure legend, the reader is referred
to the web version of this article.)

6
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

Fig. 5. Gas permeability (×10-18 m2) measured on LCK, AAS and SSC concretes
after 90 days of curing, for different drying temperatures (105 ◦ C, 80 ◦ C,
60 ◦ C, 45 ◦ C).

in the water porosity test, the sample thickness was not reduced to adapt
to the drying step at lower temperatures (50 mm for all the drying Fig. 6. Electrical resistivity of LCK (in green), AAS (in red) and SSC (in yellow)
temperatures tested). For the LCK concrete, drying was only carried out concretes measured with a resistance meter at 28 days (hatched bars) and 90
at the temperature of 105 ◦ C as it is a Portland cement-based binder, days (full bars) of curing. (For interpretation of the references to colour in this
considered adapted to the standard preconditioning according to the figure legend, the reader is referred to the web version of this article.)
results obtained in section 4.1.1.
Large differences in gas permeability were identified according to increasing with the compressive strength [58,65–67]. Finally, the re­
the type of binder (Fig. 5). The gas permeability, after drying at 105 ◦ C, sistivity of a traditional OPC concrete is generally between 100 and 300
of the three concretes studied can be ranked from the highest to the Ω.m [65,67,68], which means that the replacement by fillers leads to a
lowest as follows: AAS (3520 × 10-18 m2) > SSC (1770 × 10-18 m2) ≫ decrease in the resistivity. To explain the differences between the three
LCK (39 × 10-18 m2). The gas permeability of slag-based concretes (AAS matrices, several factors are involved, such as the quantity of paste and
and SSC) was several orders of magnitude higher than for the LCK the pore structure of the concrete or the conductivity of the pore fluid
concrete. This is consistent with the literature findings [60] and the solution, as discussed in more detail in section 5.3.
previous results presented in section 4.1.1 for porosity, which is a Evolution of the resistivity between 28 days and 90 days
parameter strongly influencing the gas permeability.. However, these An increase of resistivity for LCK and SSC concretes was observed
values remain higher than those available in the literature for slag-based between 28 days and 90 days. In contrast, the resistivity of AAS
concretes. For example, for SSC of higher strength class (between 40 and remained globally constant, whereas an increase with the duration of
50 MPa at 28 days) the gas permeability measured in the literature is curing would have been expected, as previously observed [58,69,70]
between 323 [37] and 921 [36] × 10-18 m2 (against 1770 × 10− 18 m2 in due to the densification of the microstructure with the slag or cement
this study), which suggests a possible alteration of the material during hydration.
the test (idem for AAS). Moreover, for drying temperatures above 60 ◦ C,
AAS gas permeability was around twice that of SSC. 4.2. Ionic conductivity of the pore solution
Finally, the comparing gas permeability obtained for lower drying
temperatures with that using a drying step at 105 ◦ C, showed a strong To investigate the composition and the ionic concentration of the
effect of the drying temperature on the results. The same trend as for pore solutions, an ICP analysis was performed after 38 days of endoge­
water porosity was observed: the gas permeability measured was high nous cure. The cations Na+, K+, Ca2+, Al3+, Mg2+ and Fe2+ were
for high drying temperatures and decreased with lower drying temper­ quantified. H3O+ ions were neglected because of the high pH of the pore
atures. However, this time, the changes of drying temperature influ­ fluid solution. For the AAS binder, CO2– 3 ions were supplied by sodium
enced the results strongly (almost a factor 20 for AAS concrete between carbonate activation, but Bernal et al. [71] show that they are
105 ◦ C and 45 ◦ C), as discussed in greater detail in section 5.1. completely consumed after about 10 days. As the extraction of pore
solution was performed at 38 days of cure here, they are neglected. The
4.1.5. Resistivity amount of OH– ion was deduced as the value needed to maintain elec­
Results of resistivity for different curing periods (28 or 90 days) are troneutrality for LCK, and from the pH measurements for AAS and SSC
presented in Fig. 6. due to the presence of other anions, as explained below. It is important
Comparison among the three binders to note that the concentration of sulfide species S2- was not measured by
LCK has a lower resistivity than the slag-based concretes, the re­ the ICP analysis because it is difficult to analyse, as it evolves over time
sistivity of AAS being of the same order of magnitude but slightly higher and it oxidizes easily. However, the results available in the literature
(Fig. 6). The resistivity of SSC was much higher than that of the other show that its concentration in slag-based binders is not negligible [72].
two with a factor of more than 10 between AAS and SSC. To summarize, For example, on concretes after 3 months of curing, Ma et al. [65]
the ranking obtained was the following: ρ (SSC) ≫ ρ (AAS) > ρ (LCK). measured a concentration between 2000 and 6000 ppm of sulfide in the
The very high resistivity of the SSC concrete is known in the scientific pore solution of AAS. These results confirmed measurements by Grus­
literature and has been measured by other authors [41]. For AAS, kovnjak et al. [73] and Nedeljković et al. [74] of 260 and 573 mmol/L,
scattered values ranging from 100 to 1000 Ω.m can be found, the values respectively, of sulfide in AAS cured for 28 days against negligible

7
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

amounts for a traditional CEM I-sample. Sulfide ions S2- were therefore [51]. From the chloride profile obtained for each concrete (Fig. 8), the
neglected for LCK. To estimate the amount of sulfides in AAS and SSC, it respective diffusion coefficient Dnss can be calculated, as summarized in
was deduced as the amount needed to maintain the electroneutrality of Table 6. The test was performed on concrete after 28 days and 90 days of
the solution. The values obtained were estimations (not direct mea­ curing to evaluate the influence of the curing period on the chloride
surements) of the concentrations present in the pore solution of the three diffusion resistance of each concrete. Only the results after 90 days of
binders at 38 days of curing. These values are not representative of the curing are presented for the LCK concrete, due to technical problems for
concentrations at a young age and may still evolve over time. The results the test after 28 days of curing. One sample was analysed per condition.
are detailed in Table 5 in mg/L with the associated conversion in mmol/ The proposed 95% confidence interval was calculated considering
L, based on the atomic weight of each species, detailed in Table 9 in the possible variations on the input data. For this purpose, the mathematical
supplementary data. fitting required by the EN 12390–11 standard was performed with the
These data were then used to estimate the conductivity of the pore two measured Ci values (on the sample with 28 days of curing and the
solution by calculation according to Equation 1 [75]. one with 90 days of curing), and also considering an error of ± 10% on
∑n the measured chloride concentrations for each ground layer.
σ= i=1
λi [Xi ] (1) Fig. 8 shows marked differences in chloride diffusion resistance for
the three concretes studied. The type of binder plays an important role in
with σ the conductivity of the pore solution in S/m, λi the molar con­ this property, as slag-based concretes (AAS and SSC), which have similar
ductivity of species i in S.m2.mol− 1 (taken from the literature and pre­ chloride profiles at 28 and 90 days, are much more resistant to chloride
sented in Table 9 of the supplementary data) and [Xi] the concentration diffusion than those made with Portland cement (LCK). This trend has
of species i in the pore solution in mol/m3 (measured by ICP and also been observed by other authors [26,37,78–80].
available in Table 5). The LCK matrix was characterized by fast transport of chlorides in
The contribution of each ionic species to the conductivity of the pore the concrete, even though it presented a low water porosity of 10%
solution is shown in Fig. 7 and detailed in Table 10 of the supplementary (4.1.1). This chloride diffusion coefficient (50.78 × 10− 12 m2/s) is
data. Note that the molar conductivity of S2- was attributed to the HS- higher than the one found in the literature for OPC based-concretes,
ions, another possible ion formed from sulfur [76], as this characteristic which is around 23 × 10− 12 m2/s [81] as discussed in greater detail in
was not found for S2-. section 5.4. Fig. 8 also shows that the LCK concrete bound a small
First of all, it can be noted that the conductivity related to the fraction of the chlorides (on average 25% of bound chlorides among
quantity of ions in the pore solution is very variable for the three mix­ total chlorides), suggesting high chloride diffusion kinetics.
tures considered (by a factor of around 9 between SSC and AAS) as, the Unlike the LCK matrix, AAS and SSC concretes showed a strong
more ions a pore solution contains (Table 5), the more its conductivity resistance to chloride diffusion, even though the AAS and SSC concretes
increases. It can also be noted that OH– ions are the most conductive, had a higher porosity, around 16% (but with smaller pore sizes as shown
followed by alkalis (Na+, K+), as illustrated by their respective molar on Fig. 2). Similar orders of magnitude for diffusion coefficient (Dnss) are
conductivities, presented in Table 9 in the supplementary data. The found in the literature for SSC concretes of higher strength class, C45/55
conductivity of the SSC is the lowest (1 S/m) and is equitably distributed or C50/60 (0.07–0.10 × 10− 12 m2/s by Divet and Le Roy [37]) and AAS
among Ca2+, K+, Na+, S2- and OH– ions. This low conductivity agrees concretes of higher strength class (0.5 × 10− 12 m2/s by Zhang et al.
with the high resistivity measured on concrete in section 4.1.5. Then, the [82]), 3.73 ± 1.98 × 10− 12 m2/s by Osio-Norgaard et al. [27]. It was not
LCK has an intermediate conductivity (8 S/m) due mainly to hydroxide possible to accurately quantify free chlorides using the GranDuBé pro­
(74%) and potassium (21%) ions. Considering that the sample used for tocol to estimate the influence of chloride binding on the Dnss co­
the pore solution extraction was saturated (endogenous curing), the efficients, as the procedure was developed for Portland cement binders
pore solution conductivity of LCK is comparable to that obtained by and gives values out of the expected order of magnitude and not relevant
Olsson et al. [77] on a CEM I. Finally, the AAS binder had the highest for AAS and SSC concretes. More details are given in the Discussion part
conductivity (11 S/m), mainly due to sodium (47%) from the activator, 5.2.
which is sodium carbonate, and also to sulfide (29%) and hydroxide ions Finally, the chloride diffusion coefficient Dnss (Table 6) obtained on
(21%). This high conductivity agrees with the low resistivity measured concretes after 28 days of curing was higher than the one measured on
on concrete in the previous section. samples after 90 days of curing (lower chloride content and lower
penetration depth measured). There is a factor of 4 or 2 for AAS or SSC,
4.3. Chloride transportation in concrete respectively, between the two curing periods (results unusable for LCK
concrete after 28 days of curing), indicating a significant pore refine­
4.3.1. Chloride diffusion test ment between 28 and 90 days, consistently with the water porosity re­
The chloride diffusion resistance of each concrete was assessed from sults at 50 ◦ C presented in Fig. 1. This seems logical as slag-based
the chloride diffusion test in natural conditions using EN 12390–11

Table 5
Concentration of the main species present in the pore fluid solution of LCK, AAS and SSC pastes after 38 days of curing (in mg/L and conversion into mmol/L),
quantified by ICP analysis for Na+, K+, Ca2+, Al3+, Mg2+, Fe2+, calculated either from pH measurement or assessed to maintain the electroneutrality for OH– and S2-
ions.
Concentration of species in pore fluid solution (38 days’ curing)

ICP analysis Calculation Calculation

Na+ K+ Ca2+ Al3+ Mg2+ Fe2+ OH– S2- Total ionic concentration

LCK mg/L 1260 8745 157 0.961 0 0.183

mmol/L 55 224 4 0.036 0 0.003 286 (2) 0 569
AAS mg/L 23,370 1353 3 204 0 0.341
mmol/L 1017 35 0.087 8 0 0.006 112 (1) 481 (2) 1652
SSC mg/L 988 1209 637 0.227 0 0
(1) (2)
mmol/L 43 31 16 0.008 0 0 13 46 149
(1)
From pH measurements: [OH–] = 10pH-14 (with pH = 13.2, 13.1 and 12.1 for LCK, AAS and SSC, respectively).
(2)
From electroneutrality.

8
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

Fig. 7. (a) Pore fluid conductivity of LCK, AAS and SSC and (b) contribution of each ionic species to this conductivity (OH–, S2-, Na+, K+, Ca2+, Al3+, Mg2+, Fe2+).

Fig. 8. Chloride profile (total chlorides) associated with LCK, AAS and SSC concretes after 28 or 90 days of curing and immersion in NaCl solution at 30.93 g/L.
Profiles of free and bound chlorides are also available for LCK concrete after 90 days of curing.

4.3.2. RCPT test

Table 6
Charge passed, Q, in the RCPT test and the associated penetration
Chloride diffusion coefficient Dnss of LCK, AAS and SSC concretes, after 28 and
levels of chloride ions for LCK, AAS and SSC concretes after 28 and 90
90 days of curing, after exposure in saline solution [NaCl] = 30.93 g/L. The
meanings of abbreviations are as follows: t is the exposure time in saline solu­
days of curing, are presented in Fig. 9.
tion, Ci is the initial chloride content, Cs is the chloride content at the concrete The charge passed measured by the RCPT test is directly dependent
surface, Dnss is the chloride diffusion coefficient, and R2 is the square of the on the material porosity and the connectivity of the porous network, as
linear correlation coefficient. Both Ci and Cs are % by mass of concrete. well as the presence of ions in the concrete pore solution (free chlorides).
Concrete Chloride Second Fick’s law parameters
The presence of ions makes the electric current transport easier, and thus
exposure corresponds to measurement of a high charge passed and so a high
permeability to chlorides [83]. Fig. 9 shows that charges passed for slag-
Curing Binder t Ci Cs (%) Dnss (x10- R2 (exclusion
(days) (%) 12 2
m /s) 1st point) based concretes are much lower than for LCK concrete. AAS and SSC
have a very low charges passed (427 and 125 Coulombs, respectively, at
28 AAS 91 0.012 0.15 ± 0.84 ± 0.954
days 0.03 0.10
90 days of curing), i.e. a low chloride permeability, while LCK has high
SSC 91 0.010 0.14 ± 1.06 ± 0.999 charge passed (6572 Coulombs at the same curing period) correspond­
0.01 0.09 ing to a high chloride permeability. These results are in accordance with
90 LCK 96 0.015 0.32 ± 50.78 ± 0.978 those of the natural chloride diffusion test presented in section 4.3.1 and
days 0.02 12.06
with literature data [41,84], and are discussed in more detail in section
AAS 454 0.009 0.42 ± 0.19 ± 0.997
0.04 0.01 5.3. The charge passed presented in Fig. 9 is also proportional to the
SSC 184 0.008 0.39 ± 0.58 ± 0.991 concrete resistivity presented in Fig. 6, as it is obtained by integrating
0.04 0.21 the current over time, which is itself a function of the resistance of the
material according to Ohm’s law. The proportionality between the
charge passed and the resistivity of the concrete is shown in Fig. 10,
concretes are not entirely hydrated at 28 days and therefore have a less
which illustrates that, when the resistivity of the material decreases, i.e.
dense porous network and a physical barrier that is less effective in
when its conductivity increases, the associated charge passed also in­
resisting chloride penetration [82].
creases because the current transport is favoured.

9
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

tests. As summarized by [59,64], the challenge is to use a drying tem­

perature that efficiently removes the water from capillary pores without
altering the material’ microstructure and chemistry. Therefore, it is
important to adapt this temperature to each binder to avoid erroneous
conclusions and misinterpretations concerning their durability (possible
over- or under-estimation of parameters). The literature shows that the
high drying temperatures generally used for Portland cements are not
suitable for slag-based binders.
For SSC, a high drying temperature leads to the destabilization of
ettringite, which is an important hydrate of this binder [4,85,86], the
proportion of which is estimated at about 30–40% of hydrates by ther­
modynamic modelling [30,87]. It has been shown that ettringite is un­
stable around 60 ◦ C and for higher temperatures [62]. This result was
confirmed by Cyr et al. by X-ray diffraction measurements on SSC paste
samples [36]. They showed that the SSC pastes conditioned at 20 ◦ C
(without drying) and 40 ◦ C contain ettringite and gypsum as the main
crystalline hydrates while, after drying at 105 ◦ C, these substances are
replaced by hemihydrate and anhydrite.
AAS binders do not contain ettringite as a hydration product but
some authors have, nevertheless, investigated their suitable drying
Fig. 9. Charge passed, Q, and the associated penetration levels of chloride ions temperatures [53,59–61]. For Na2CO3-AAS binders, the optimal drying
for LCK, AAS and SSC concretes after 28 days (hatched bars) and 90 days (full
temperature, according to Trincal et al. [64], is a temperature between
bars) of curing, from RCPT test.
80 ◦ C and 105 ◦ C, which provides optimal drying kinetics (not too long)
and preserves the mechanical strength and mineralogy (only an increase
of the pore size was highlighted). In contrast, it is better to avoid the
range of 40–60 ◦ C at low relative humidity, as a loss of mechanical
strength was detected (more significant at 60 ◦ C). According to the au­
thors, it is possibly attributable to carbonates and C-(A)-S-H reactions,
such as desaturation of intercrystallite pores, dissolution/precipitation,
carbonation or the fact that higher drying temperatures and kinetics can
limit capillary pressure.
The durability parameters that require a preconditioning step at high
temperatures in this study are the water porosity test from standard NF
P18-459 [45] (105 ◦ C required) and the gas permeability test from XP
P18-463 [48] (80 and 105 ◦ C required). To evaluate the influence of the
drying temperature on these parameters, the tests were performed at
45 ◦ C, 60 ◦ C, 80 ◦ C and 105 ◦ C on AAS and SSC with a constant sample
thickness of 50 mm, as presented in Fig. 11. For the LCK concrete, which
is a Portland cement-based binder, the standard preconditioning at
105 ◦ C was considered adapted (section 4.1.1), that is why no additional

Fig. 10. Relationship between the charge passed measured according to the
RCPT test and the resistivity of concrete after 28 or 90 days of curing, for LCK,
AAS and SSC.

5. Discussion

The tests carried out in this work to characterize the durability of the
concretes were defined for Portland cement-based binders and it is
important not to apply them for alternative binders without questioning
their suitability. Sections 5.1 and 5.2 focus on this metrology aspect in
order to discuss the relevance of the tests carried out with or without
adaptation, to characterize the three low-carbon concretes of the study.
The results obtained are then analysed in greater detail in section 5.3,
and compared to commonly used references (CEM I, CEM II/A, CEM III/
A) using a performance-based approach in section 5.4, before finally
being extrapolated in section 5.5 to conclude on the resistance of each
concrete to chloride ingress.

Fig. 11. Gas permeability measured for different drying temperatures as a

5.1. Metrology: Effect of drying temperature for preconditioning of slag- function of water porosity measured for the same drying temperatures, on LCK
based binders (in green), AAS (in red) and SSC (in yellow) samples with a constant thickness
of 50 mm. (For interpretation of the references to colour in this figure legend,
Drying is required as a preconditioning step for several durability the reader is referred to the web version of this article.)

10
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

tests at lower drying temperatures were performed. mechanism influencing the chloride transport through the concrete, it­
Firstly, it can be observed that the measured porosity increases with self conditioned by the material’s porous network and the porosity of the
the drying temperature used (+9% between 45 ◦ C and 105 ◦ C for AAS interfacial transition zone, which can evolve over time [89,90]. In
and SSC), as more water can be removed from the capillary porosity in classical Portland-based materials, chlorides are mainly chemically
addition to all physically sorbed water, and there may also be damage to bound by AFm and to a lesser extent, physically bound by C-S-H, which
the material as discussed above. For the drying temperature of 60 ◦ C, the slows sown their ingress. Finally, only the free chlorides in the pore
measured increase in porosity is higher for SSC than for AAS, which solution are likely to lead to corrosion of the reinforcement.
tends to confirm the literature findings about ettringite destabilization It is well known that slag-based binders (such as AAS and SSC) are
for SSC. Considering conditions of 20 ◦ C and 50% relative humidity (RH) resistant to chloride penetration due to their high binding capacities
in the laboratory where the tests were performed, the associated dew (observation of chloride binding by hydration products using XRD, TGA,
point is 9.3 ◦ C. Assuming a constant dew point of 9.3 ◦ C in the labora­ etc.), but few studies in the literature deal with the quantification of free
tory, a relative humidity of 12.5% can be deduced at 45 ◦ C, 6% RH at and bound chlorides in such matrices. For example, Osio-Norgaard et al.
60 ◦ C, 2.5% RH at 80 ◦ C and 1% RH at 105 ◦ C. It has been shown, that [27] proposed a review of 66 articles dealing with chloride transport in
ettringite becomes unstable around 5% RH and below [88], meaning AAS binders but, among them, only one investigates chloride binding
that it is stable with an acceptable reliability for the drying temperature (by NaSi-AAS). As presented in part 4.3.1, it was not possible to quantify
of 45 ◦ C of Fig. 11. Finally, the porosity of 8% measured for the drying free chlorides accurately in this study by using the French protocol [56]
temperature of 45 ◦ C for slag-based binders seems very low (even lower for AAS and SSC concretes, as the procedure was developed for Portland-
than that of LCK) and is certainly underestimated due to the low drying based materials and is quite sensitive to experimental conditions (stir­
temperature, preventing the evaporation of water in the smaller capil­ ring time, amount of water added, temperature, chlorides bound
lary pores, even if a constant mass is reached [64]. differently than for OPC…) [68]. Extremely high free chloride contents
Concerning the gas permeability, it increases strongly with the dry­ were found (equal to or greater than total chlorides), as presented in
ing temperature used, for the same reasons as explained above. This Table 11 in the supplementary data, whereas slag-based binders are
increase is rather linear for SSC, although the values measured at 45 and known for their higher ability to bind chlorides compared to Portland
105 ◦ C (274 to 1770 × 10-18 m2) differ by a factor of 6. For the AAS cement [22,26,39,82,91]. Some later investigations showed that it is
concrete, this increase is even higher (186 to 3520 × 10-18 m2) with a mentioned by the standard ASTM C1218 [92] for the determination of
factor of 19. At 60 ◦ C, the permeability value is already multiplied by 15 water soluble chloride in mortar and concrete, that “sulfides are known
compared to the one measured at 45 ◦ C, which is in agreement with to interfere with the determination of chloride content”, as the slag
damage of the material at this temperature. contains “concentrations of sulfide capable of such interference and of
Finally, it is important to keep in mind that, at 105 ◦ C, the values producing erroneously high test results”, which confirms the previous
were probably over-estimated due to damage of the concrete or an findings [92]. A treatment with hydrogen peroxide is proposed in the
alteration of its microstructure while, at 45 ◦ C the values are surely standard to eliminate such interference by oxidizing the sulfides. Actu­
under-estimated, preventing the evaporation of water in the smaller ally, this interference of sulfides in the determination of chloride content
capillary pores [36]. It can be expected that the real water porosity and is mostly documented in the literature in other fields, such as analytical
gas permeability of concrete are between these two extremes. chemistry, environment or water treatment [92–96]. The standard ISO
That is why the temperature of 50 ◦ C was finally adopted, combined 9297:1989 [96] in the field of water quality, explains that sulfides (S2-)
with a reduction of the sample thickness to 17 mm (not done for gas hinder the determination of chloride in water, forming insoluble silver
permeability but recommended), that is to say 1/3 of its initial thickness, compounds, while thiosulfate (S2O2- 3 ) ions form complexes with silver
to keep only one acceptable drying temperature for LCK, AAS and SSC in ions. A test that allows better quantification of free and bound chlorides
the modified procedures proposed previously. 50 ◦ C was chosen as a is the fixation isotherm test, where the concrete is crushed to a certain
compromise considering the destabilization of ettringite around 60 ◦ C particle size and the depletion of the NaCl solution in contact with it is
and above for SSC and the damage of the material observed from 60 ◦ C measured [97]. This test was performed by Ye et al. [91], and illustrates
for the gas permeability measurement of AAS. Moreover, at the chosen a higher chloride binding capacity of AAS (including Na2CO3-AAS)
temperature of 50 ◦ C, the associated relative humidity of 10%, consid­ compared to OPC, particularly at high NaCl concentrations. The bound
ering a constant dew point of 9.3 ◦ C, ensures the stability of ettringite chloride content in AAS pastes equilibrated at 1.0 M NaCl condition
[88]. No particular recommendations were necessary for the LCK as ranges from 13.7 to 19.7 mg/g, i.e. 70%–150% higher than that of pure
behaviour similar to that of a Portland cement is expected during drying. OPC pastes with a similar Liquid/Solid ratio [91]. However, this is a
Mechanical strength measurements on 4x4x16 cm3 mortar samples were time-consuming test, which is no longer widely used. Quantitative
made in parallel on AAS (Fig. 14 in supplementary data) and no decrease analysis by ionic chromatography is also possible and some rare de­
was observed after drying at 50 ◦ C, as previously reported [58,64]. terminations of free and bound chlorides for AAS and SSC binders can be
Finally, Fig. 11 shows that the drying temperature still has consid­ found in the literature. Nguyen et al. [39] succeeded in quantifying free
erable influence on the results obtained. The temperature of 50 ◦ C is and bound chlorides in SSC pastes and Ismail et al. [61] quantified free
proposed in this study for the thermal preconditioning of low-carbon and bound chlorides in NaSi-AAS pastes, following ASTM C1152 for
binders but it remains preferable, when possible, to favour direct tests acid-soluble chloride (total chloride) and ASTM C1218 for water-soluble
on a pathology without prior thermal preconditioning (RCPT, migration chloride (free chloride). Neither of these groups state whether they used
or diffusion tests to assess the performance of concrete against chloride hydrogen peroxide. Zhang et al. [82] quantified free chlorides in one
penetration for example) for greater reliability and representativity of NaSi + NaOH-AAS concrete, in accordance with the Chinese standard
the real conditions. It should thus be recommended to avoid using only JTJ270-98 [98], where 2 g of concrete powder is put into 50 mL distilled
general durability indicators such as porosity and permeability to water and shaken for 20 min. Twenty-four hours later, the suspension is
qualify the long-term behaviour of these binders and to adapt the con­ filtered, 20 mL is put into a flask with 2 drops of phenolphtalein and
ditioning to the concrete constituents. diluted H2SO4 is added until the solution becomes colourless. After­
wards, 10 drops of K2CrO4 solution are added and 0.02 mol/L AgNO3
5.2. Metrology: Determination of free and bound chlorides for slag-based solution is used to titrate until the solution becomes red. The volume of
binders AgNO3 solution consumed is recorded and used to calculate the free
chloride content. However, the free chloride values obtained were not
Chloride ions in concrete exist either in the form of free ions in the compared to the total chloride values to quantify the chloride binding
pore solution or bound to the hydration products. Diffusivity is the main capacity. Finally, some authors have proposed quantifying the chloride

11
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

binding ability of AAS by alternative methods such as by experimental capacity on the one hand, and for high conductivity (thus low re­
work on model hydrates and simulated pore solution [97] or by sistivity) because there are fewer hydrates to trap ions, on the other
modelling [99]. The main conclusions are that AAS and SSC binders hand. Lower Portland cement content means a reduced capacity to bind
have strong ability to bind chlorides. For example, Nguyen et al. [39] chlorides, as shown in Fig. 8: the LCK concrete bound few chlorides,
found that, at 28 days, 87% of total chlorides measured were bound by which, combined with its very low resistivity presented in section 4.1.5
SSC hydrates in pastes cast with 3 %wt NaCl in the mix. This is much can partly explain the fast chloride ingress. This observation is in
higher than the proportion of bound chlorides measured for the LCK agreement with other studies for which the chloride diffusion coefficient
concrete in section 4.3.1 (between 15% and 40% with an average of measured on Portland limestone (30% limestone) concretes (13.6–23.5
25%, depending on the grinding depths). × 10− 12 m2/s) was considerably higher than on Portland cement
(7.0 – 8.0 × 10− 12 m2/s) [101–103]. The review by Elgalhud et al. [8]
5.3. Microstructure and chemistry analysis estimates that the chloride ingress of a Portland limestone concrete with
35% limestone can increase by about 65% compared to a concrete
This part highlights the characteristics of each of the low-carbon without limestone. This increase is expected to be even higher for 73%
concretes of similar mechanical performances, by coupling the anal­ limestone (as in the LCK of the study). Finally, and to a lesser extent, LCK
ysis of their microstructure and their chemistry (results obtained in has a majority of capillary pores (greater than10 nm), 85% according to
section 4 and summarized in Table 7). These new elements give a better Fig. 3, and this promotes the mobility of chloride ions [82]. The physical
understanding of the performance of LCK, AAS and SSC against the barrier resisting penetration of aggressive agents is high (reduced
penetration of aggressive agents. transfer properties due to a low w/b ratio), but is counterbalanced by a
Low clinker (LCK) low chemical barrier due to the low clinker content.
The results of section 4, summarized in Table 7, show that, in gen­ Slag-based binders (AAS and SSC)
eral, LCK concrete has good transfer properties: low porosity (water or The results of section 4, summarized in Table 7, show that AAS and
MIP), low water and gas permeability, and low capillary absorption. SSC concretes have several similar transfer properties: same porosity
This is mainly due to the low w/b ratio (0.25) of this mixture, combined (water and MIP), similar pore sizes, same water permeability, and high
with an optimized granular packing, which, on the one hand, balances capillary absorption and gas permeability in both cases. In general, these
the reduced cement content of this mixture (27% of the binder) and thus transfer properties are high compared to those of LCK (except for pore
maintains the desired mechanical resistances (C25/30) and, on the other size), due to a higher w/b ratio (about twice as high) and a higher paste
hand, minimizes the transfer properties (physical barrier). This is posi­ volume (318 L for LCK, compared to 407 L for AAS and 360 L for SSC). In
tive from a durability point of view. However, the drawback of this low detail, the lower paste volume of SSC concrete (360 L) compared to AAS
w/b ratio is a more viscous concrete that is more difficult to handle (407 L) is balanced by a higher w/b ratio (0.50 vs. 0.43 for AAS),
(impact on rheology despite an optimized admixture), with a reversed resulting in similar transfer properties. Moreover, the high w/b ratio of
cone flow rate of 9.7 s (Table 3). The other weak point of this matrix is its these mixtures provides higher flowability than that of LCK concrete
low resistance to chloride penetration and diffusion, which depends (flow rate 5 times as high) and thus easier workability.
mainly on the chemistry of its hydration products (influencing the Then, both matrices offer very good performance against chlorides:
chloride binding) and its pore structure (influencing the diffusive low chloride diffusion coefficient, Dnss, and low charge passed during
transport), as demonstrated mathematically by Ollivier et al. [100]. the RCPT test in both cases, but of different orders of magnitude. The
Despite its reduced porosity, a high chloride diffusion coefficient, Dnss, chloride diffusion coefficient (at 90 days of curing) of AAS concrete
high charge passed in the RCPT test and low resistivity were measured (0.19 ± 0.01 × 10− 12 m2/s) is lower than that of SSC (0.58 ± 0.21 ×
for LCK. This can be explained by its low clinker content (137 kg/m3 or 10− 12 m2/s), which suggests a difference in the diffusive transport and/
27% of the binder), which is responsible for a lower chloride-binding or chloride binding capacity between the two binders. A difference in

Table 7
Summary of test results obtained for LCK, AAS and SSC concretes to characterize their general transfer properties, pore solution and performance against chloride
penetration.
Test or property measured Unit Sample curing LCK AAS SSC

GENERAL TRANSFER PROPERTIES Water porosity (ε) (105 ◦ C) % 28 days 10.0 18.2 17.3
90 days 10.3 17.6 17.5
6 months 10.0 18.8 17.3
1 year 11.5 19.1 18.8
Water porosity modified (50 ◦ C) % 28 days 10.6 16.2 16.4
6 months 9.6 16.1 15.9
1 year 9.8 15.8 14.2
Mortar porosity from MIP % 28 days 19 28 28
Average pore diameter from MIP nm 28 days 23 13 13
Pore entry diameter of the largest pore size from MIP nm 28 days 40 9 and 21 12
Water permeability mm 28 days 5 16 15
Capillary absorption (Aw) kg/m2.h1/2 28 days 0.112 0.259 0.346
Gas permeability (k) (105 ◦ C) x10-18 m2 90 days 39 3520 1770
Gas permeability modified (80 ◦ C, 60 ◦ C, 45 ◦ C) x10-18 m2 90 days – 3510 1230
– 2710 761
– 186 274
Resistivity (ρ) Ω.m 28 days 48 112 1537
90 days 77 104 1955
PORE SOLUTION Ionic conductivity of the pore solution (σ) S/m 38 days 8 11 1
Concentration of species in pore fluid solution mmol/L 38 days 569 1652 149
CHLORIDE TRANSPORT Natural chloride diffusion test (Dnss) x10-12 m2/s 28 days * 0.84 ± 0.10 1.06 ± 0.09
90 days 50.78 ± 12.06 0.19 ± 0.01 0.58 ± 0.21
RCPT test (Q) Coulombs 28 days 4770 614 159
90 days 6572 427 125

*This result is not available, due to technical problems for the test after 28 days of curing.

12
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

diffusive transport is possible because of the higher capillary absorption, passed and resistivity) of the SSC can be explained by the very small
gas permeability and charge passed measured on AAS than on SSC (even amount of ions in its pore solution (149 mmol/L), leading to a very low
though the values are of the same order of magnitude compared to those conductivity (1 S/m) [105,106]. Then, although SSC has a high porosity
of the LCK), as well as a much lower resistivity, which would be in line (water and MIP), its low pore diameter leads to more tortuosity and
with stronger diffusion mechanisms for AAS. However, the fact that the involves the double layer effects that reduce the mobility of remaining
Dnss coefficient, which takes the chloride binding into account, is lower ions. Finally, SSC has a lower paste volume (360 L) than AAS (407 L),
for AAS than for SSC suggests a stronger chloride binding capacity of which also favours lower resistivity [69]. Concerning AAS, although its
AAS, compensating for the higher diffusion. The mechanisms of chloride porosity and pore size are similar to those of SSC, its higher ionic charge
binding for AAS and SSC depend on their hydration products. For can mainly be explained by a very high concentration of species in its
Na2CO3-AAS, 40% to 70% of the total bound chlorides are bound by pore solution (1652 mmol/L), due to the high sodium content (Fig. 7)
hydrotalcite and monocarbonates [91]. These are hydration products brought by activation with sodium carbonate, leading to a high con­
called Layered Double Hydroxides (LDH) and represent between 22 and ductivity (11 S/m). Finally, unlike LCK concrete, AAS and SSC have a
40% of the solid phases, according to thermodynamic modelling weaker physical barrier against aggressive agents, which is balanced by
[99,104]. The chloride binding mechanism is either anion exchange in a strong chemical barrier against chlorides.
their intermediate layer or surface absorption in the electrical double
layer [97]. C-A-S-H also contributes to chloride binding (surface
adsorption as the principal mechanism), but to a lesser extent than LDH 5.4. Comparison with references based on the performance approach
[97]. The chloride binding by SSC is governed principally by the for­
mation of Friedel’s salts from ettringite and by hydrotalcite [39], which In order to better position the performance of the three low-carbon
make up between 25% and 47% of the solid phases, according to ther­ concretes studied (LCK, AAS and SSC), a comparison was made with
modynamic modelling [30], following the same mechanisms as the one commonly used reference concretes (CEM I, CEM II/A, CEM III/A), using
described for AAS. the performance-based approach as illustrated in Fig. 12. General
Finally, a difference in ionic charge between the two mixtures can transfer properties, such as water porosity, resistivity, gas permeability,
also be identified, as the SSC binder has lower charge passed (125 C at and capillary absorption were compared between the three low-carbon
90 days) than AAS (427C) and a much higher resistivity (1955 Ω.m concretes of the study and three reference concretes, as were their nat­
versus 104 Ω.m for AAS at 90 days). The low ionic charges (charge ural chloride diffusion coefficient and their carbon footprint. The
reference concretes were chosen to have a strength class equal or close to

Fig. 12. Comparison of LCK, AAS and SSC general transfer properties, chloride transport and CO2 footprint with those of reference concretes (CEM I, CEM II/A and
CEM III/A) of similar strength class from the PERFDUB database. An upward arrow indicates that a high value is aimed for the considered property, in order to have a
maximum performance from a durability point of view (inversely with a downward arrow).

13
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

that of the concretes in this study, which is C25/30. Thus, only the time required for chloride to reach a rebar at a certain concrete cover.
references C20/25, C25/30 and C30/37 were considered, in order to The idea is to use the solution of Fick’s second law for a semi-infinite
remain comparable to the concretes studied and because they had suf­ case with a constant boundary condition Cs and initial condition Ci
ficiently complete databases (5 references for CEM I, 4 references for (Equation 2), using the natural chloride diffusion coefficient, Dnss, ob­
CEM II/A and 1 reference for CEM III/A) to be representative of the tained experimentally on concrete cured for 90 days and immersed in
characteristics of the various reference concretes. The database of the reconstituted seawater (presented in section 4.3.1).
national project PERFDUB [81] was used to make the comparisons [ ( )]
x
because this project allowed the evaluation of many characteristics C(x, t) = Ci + (Cs − Ci ) 1 − erf √ ̅̅̅̅̅̅̅̅̅
̅ (2)
2
Dnss t
related to the durability of concretes considering various binders. It was
also interesting because its data were published and involved cross
where C(x, t) (% by mass of concrete) is the chloride content measured at
testing between several laboratories [107]. In total, the PERFDUB
average depth × (m) and exposure time t (seconds); Cs is the calculated
database is composed of 42 concretes. For the calculation of carbon
chloride content at the exposed surface (% by mass of concrete); Ci is the
footprints, 765, 676 and 437 kg eq. CO2/m3, respectively, were taken for
initial chloride content (% by mass of concrete); Dnss is the non-steady
the CEM I, CEM II/A and CEM III/A according to the French values from
state chloride diffusion coefficient (m2.s− 1).
SFIC (Syndicat Français de l’Industrie Cimentière) available on [42].
It has been shown that “chloride-transport tests (…) can be used as
Concerning general transfer properties (bar charts in yellow in
inputs to models to provide reasonably reliable predictions of perfor­
Fig. 12).
mance under certain limitations” [109], as confirmed by [110], and that
Due to its low w/b ratio, LCK has low porosity and capillary ab­
the estimation of the time needed for chlorides to reach a rebar at a
sorption compared to the references, while AAS and SSC have porosity
defined depth from Fick’s second law provides good matches to reality
located in the high average of the references and slightly lower capillary
[111]. The use of the non-steady state diffusion coefficient Dnss allows
absorption (keeping in mind that, for a drying temperature of 105 ◦ C,
the diffusion of chlorides through the concrete, as well as the concrete’s
the porosities are surely over-estimated for slag-based binders, as dis­
chloride binding capacity, to be considered in an indirect way (without
cussed in section 5.1). LCK and AAS have resistivity of the same order of
the use of chloride fixation isotherms for each matrix). A first hypothesis
magnitude as that of the references (similar to CEM I for LCK and to CEM
taken here is that the model considers only the diffusion mechanism,
II/A for AAS), while SSC has very high resistivity, mostly due to its low
while other chloride transport processes may take place [26]. It is also
concentration of ions in the pore fluid solution. The same observations
assumed that Dnss is constant over time, even though it is shown in the
can be made at 90 days of curing, as detailed in Fig. 15 in the supple­
literature that it is temperature dependent [109] and time dependent
mentary data. Gas permeability is low for LCK concrete due to its low w/
[70,112], until it reaches a constant value after 5 to 10 years of service
b ratio, while it is extremely high (several orders of magnitude higher)
[113]. Therefore, this hypothesis may lead to a slight overestimation of
for AAS and SSC compared to the references, because of damage of the
the kinetics obtained. To approach a stabilized Dnss value more closely,
sample microstructure during the preconditioning in temperature.
the model is used for the 90 days of cure. The parameters Ci, Cs and Dnss
Except for this biased result, the three low-carbon concretes studied
are those presented in Table 6. From this point, for a given cover x, it is
have similar or better transfer properties than the reference concretes of
possible to calculate, using a solver, the diffusion time required for
similar strength, which makes them competitive and attractive in
chlorides to reach a critical concentration, Ccrit, (meaning that C(x,t) is
addition to their reduced carbon footprints, which are slightly lower
taken equal to Ccrit) at the reinforcement for each concrete, as illustrated
than that of CEM III/A for LCK and AAS and even 2.5 times lower for
in Fig. 13. Ccrit is the critical chloride concentration threshold that alters
SSC.
the rebar passive film stability and leads to corrosion initiation. A wide
Concerning chloride transport (bar chart in blue on Fig. 12).
range of values have been suggested for this parameter in the literature
The resistance to chloride penetration of the three concretes studied
[114–118]. The value of 0.4% is often chosen for Ccrit in the case of
and the three references proposed varied according to the binder
aerated conditions, while 1% is recommended in immersed conditions
considered. In general, the more slag the binder contained, the more
as the oxygen is little renewed and more chlorides are needed to initiate
resistant the concrete was to chloride penetration. Thus, CEM III/A
corrosion [117,119]. Therefore, Ccrit is fixed at 1%/cement mass in this
concrete was more resistant to chloride penetration than CEM I and CEM
study, as the diffusion coefficient Dnss is measured in seawater
II/A binders. AAS and SSC concretes were even less permeable to
chlorides than the references because they contained a larger amount of
slag. In addition to different diffusive transport, these mixtures are
known to strongly bind chlorides to their hydration products (even
though this could not be quantified experimentally as explained in
section 5.2). This slows down the penetration of chlorides through the
material [39,97]. In contrast, LCK concrete is very permeable to chlo­
rides despite its low porosity because it has a reduced clinker content
(27% of its binder). Less CEM I means a lower capacity to bind chlorides
by the formation of Friedel salts. This result has also been reported in the
literature by other authors [108]. Therefore, despite its good transfer
properties, it is advisable to avoid the use of this formulation of LCK
concrete in a marine environment, contrary to AAS and SSC concretes,
which are extremely efficient in chloride-rich areas.

5.5. Use of a simplified model to characterize resistance to chloride ion

penetration

The objective of this section is to use a “simplified model” as an

interpretation tool, to characterize the resistance to chloride penetration
of the three concretes studied, while remaining aware that more elab­ Fig. 13. Time required to reach the critical chloride threshold (1%/cement
orate and realistic models exist. The aim is to use the experimental data mass) at the reinforcement for each concrete, considering different covers from
obtained previously in order to obtain orders of magnitude and trends of 10 to 60 mm.

14
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

(immersed conditions). It corresponds to a threshold Ccrit of 0.211, 0.224 low-carbon concretes were designed to have equivalent rheology and
and 0.184 % by mass of concrete for the LCK, AAS and SSC mixtures mechanical resistances (self-compacting concrete, C25/30). A compar­
respectively. ison of their durability performances between each other and with
Fig. 13 illustrates the very different chloride penetration resistances commonly used reference concretes is presented. The following con­
of the three concretes considered in this study, which were formulated to clusions can be drawn:
have the same strength class C25/30. These are approximate values and
cannot be taken as certain. However, they allow the three low-carbon • The tests performed to characterize the durability of concretes made
concretes to be compared. For LCK concrete, the critical chloride with alternative binders require some adaptations. For the water
threshold (1%/cement mass) at the reinforcement is reached much porosity and gas permeability, a drying temperature of 50 ◦ C instead
faster than for the slag-based formulations, despite a low porosity. This of 105 ◦ C, combined with a reduction of the sample thickness, should
is due to its low clinker content, which results in a low chloride binding be recommended for the preconditioning of slag-based samples. For
capacity (section 4.3.1) and a low resistivity (4.1.5). In 5.2 years, the the determination of free and bound chlorides in slag-based binders,
threshold of 1% chlorides/cement mass is reached at the reinforcement ionic chromatography or water soluble chloride analysis combined
for a 30 mm cover. Finally, it seems that the LCK concrete does not with hydrogen peroxide should be favoured over redox electrode.
guarantee a 50-year service life, even for a thick cover (60 mm). In • The LCK concrete has better transfer properties than reference con­
contrast, slag-based formulations are more resistant to chloride pene­ cretes of similar strength (CEM I, CEM II/A, CEM III/A) due to its low
tration, as they are known to have a high chloride binding capacity. AAS w/b ratio, but this low ratio also results in a higher viscosity. The
concrete appears to perform best despite the similar porosities and much AAS and SSC concretes studied have transfer properties similar to
higher resistivity of SSC concrete (5.2). This difference can be partly those of reference concretes, which makes them and the LCK
explained by a different diffusion kinetics and a different chloride competitive and attractive, in addition to their reduced carbon
binding capacity due to the presence of different hydrates between the footprint.
two. For example, for a 30 mm cover, the critical chloride threshold is • Due to its reduced clinker content, the LCK concrete is very perme­
expected to be reached after 5.2 years, 181 years and 736 years for LCK, able to chlorides, even when compared to traditional concretes. Its
SSC and AAS, respectively. Increasing the cover by 5 mm significantly use in the presence of chloride should be avoided as it does not fulfil
increases the time taken for chlorides to reach the reinforcement: 2 years the criteria of exposure classes XS1 and XD1.
more for LCK, 65 years for SSC and 266 years for AAS by moving from 30 • The slag-based concretes AAS and SSC are extremely resistant to
to 35 mm. chloride penetration and are even better than the references, mainly
The French classification PERFDUB led to the creation of the stan­ because of their high ability to bind chlorides. They are perfectly
dard FD P18-480 [120] that details thresholds to be respected based on suited to use in a marine environment, even for the most unfav­
different durability parameters, depending on the exposure class. There ourable exposure classes, XS3m (tidal or splash zone) and XD3tf
are two exposure classes related to chloride exposure: XS (risk of (very frequent salting).
corrosion induced by the chlorides present in the sea water) and XD (risk
of corrosion induced by chlorides having a non-marine origin, de-icing This article has focused on chloride penetration resistance, but these
salts for example). To be classified XS or XD, a concrete has to reach a same low carbon concretes will be evaluated with respect to carbonation
performance threshold defined from a chloride migration test performed resistance in a forthcoming paper.
after 90 days of curing (Drcm_90d) and modulated by an aging factor α,
which is a parameter describing the decrease of the apparent diffusion CRediT authorship contribution statement
coefficient of chloride ions in concrete with time. It is calculated by
considering the composition of the binder (proportions of slag, fly ash or Lola Doussang: Writing – review & editing, Writing – original draft,
silica fume). For LCK, the associated aging factor α (calculated according Visualization, Methodology, Investigation, Formal analysis, Conceptu­
to FD P18-480 [120] is 0.30, as against 0.45 for AAS and SSC. No alization. Gabriel Samson: Writing – review & editing, Writing –
chloride migration test was performed in this study, but the Perfdub original draft, Validation, Supervision, Project administration, Meth­
report [107] shows a proportionality between migration coefficient odology, Investigation, Formal analysis, Conceptualization. Fabrice
(Drcm) and chloride diffusion coefficient (Dnss measured in this study). Deby: Writing – review & editing, Writing – original draft, Validation,
The comparison is therefore made directly with Dnss measured after 90 Supervision, Project administration, Methodology, Investigation,
days of curing, presented in section 4.3.1. Although the Drcm_90d Formal analysis, Conceptualization. Bruno Huet: Writing – review &
thresholds presented in Table 8 have not been defined for slag-based editing, Validation, Supervision, Resources, Project administration,
binders, AAS and SSC concretes could be classified as XS3m (tidal or Investigation, Formal analysis. Emmanuel Guillon: Writing – review &
splash zone) and XD3tf (very frequent salting), while LCK concrete does editing, Validation, Supervision, Resources, Project administration,
not achieve XS1 (exposed to air carrying sea salt but not in direct contact Investigation, Formal analysis. Martin Cyr: Writing – review & editing,
with sea water) and XD1 (moderate humidity) classes, as reported in Validation, Supervision, Project administration, Methodology, Investi­
Table 8. gation, Funding acquisition, Formal analysis, Conceptualization.

6. Conclusion Declaration of Competing Interest

The general transfer properties and chloride penetration resistance The authors declare that they have no known competing financial
of LCK, Na2CO3-AAS and SSC concretes were investigated. These three interests or personal relationships that could have appeared to influence
the work reported in this paper.
Table 8
Data availability
XS and XD PERFDUB classification of LCK, AAS and SSC concretes, with the
associated thresholds (chloride migration coefficient Drcm) from to FD P18-480
that must not be exceeded. Data will be made available on request.
LCK AAS SSC
Appendix A. Supplementary data
− 12 2 12
XS classification – XS3m (3 × 10 m /s) XS3m (3 × 10− m2/s)
12
XD classification XD3tf (9 × 10− m2/s) XD3tf (9 × 10− 12
m2/s)
Supplementary data to this article can be found online at https://doi.
–

15
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

org/10.1016/j.conbuildmat.2023.133511. [30] A. Gruskovnjak, B. Lothenbach, F. Winnefeld, R. Figi, S.-C. Ko, M. Adler,

U. Mäder, Hydration mechanisms of super sulphated slag cement, Cem. Concr.
Res. 38 (2008) 983–992, https://doi.org/10.1016/j.cemconres.2008.03.004.
References [31] Noor-ul-Amin, An overview on comparative study of alternatives for Ordinary
Portland cement, J. Basic Appl. Chem. 4 (2014) 15–22.
[1] G. Habert, S.A. Miller, V.M. John, J.L. Provis, A. Favier, A. Horvath, K. [32] C. Angulski da Luz, R.D. Hooton, Influence of curing temperature on the process
L. Scrivener, Environmental impacts and decarbonization strategies in the cement of hydration of supersulfated cements at early age, Cem. Concr. Res. 77 (2015)
and concrete industries, Nat Rev Earth Environ. 1 (2020) 559–573, https://doi. 69–75, https://doi.org/10.1016/j.cemconres.2015.07.002.
org/10.1038/s43017-020-0093-3. [33] H. El-Didamony, A.A. Amer, E.M. Ewais, K. Shokry, Physico-chemical and
[2] K.L. Scrivener, V.M. John, E.M. Gartner, Eco-efficient cements: Potential mechanical properties of supersulphated cement pastes, Bull. Facul. Sci., Zagazig
economically viable solutions for a low-CO2 cement-based materials industry, Univ. 2016 (2016) 54–67, https://doi.org/10.21608/bfszu.2016.31065.
Cem. Concr. Res. 114 (2018) 2–26, https://doi.org/10.1016/j. [34] T. Matschei, F. Bellmann, J. Stark, Hydration behaviour of sulphate-activated slag
cemconres.2018.03.015. cements, Adv. Cem. Res. 17 (2005) 167–178, https://doi.org/10.1680/
[3] E. Gartner, T. Sui, Alternative cement clinkers, Cem. Concr. Res. 114 (2018) adcr.2005.17.4.167.
27–39, https://doi.org/10.1016/j.cemconres.2017.02.002. [35] M. Singh, M. Garg, Behaviour of low heat sulphate resistant cement in aggressive
[4] M.C.G. Juenger, F. Winnefeld, J.L. Provis, J.H. Ideker, Advances in alternative media, Indian J. Eng. Mater. Sci. (2003) 4.
cementitious binders, Cem. Concr. Res. 41 (2011) 1232–1243, https://doi.org/ [36] M. Cyr, L. André, M. Ruau, N. Musikas, L. Frouin, Durability of supersulfated
10.1016/j.cemconres.2010.11.012. cement with improved early strength. 15th International Congress on the
[5] S. Chen, C. Duffield, S. Miramini, B. Nasim Khan Raja, L. Zhang, Life-cycle Chemistry of Cement, Prague, Czech Republic, 2019.
modelling of concrete cracking and reinforcement corrosion in concrete bridges: [37] L. Divet, R. Le Roy, Étude de la durabilité vis-à-vis de la corrosion des armatures
A case study, Eng. Struct. 237 (2021), 112143, https://doi.org/10.1016/j. des bétons formulés avec des ciments à forte teneur en laitier de haut fourneau,
engstruct.2021.112143. BLPC. (2013).
[6] AFNOR, NF EN 197-1 Ciment - Partie 1 : composition, spécifications et critères de [38] S. Ioannou, A. Badr, K. Kostova, K. Paine, T. Ibell, Utilization of Fabric Formwork
conformité des ciments courants, 2012. for Improving the Durability of Concrete from Supersulfated Cement, KEM. 711
[7] A.A. Elgalhud, R.K. Dhir, G.S. Ghataora, Carbonation resistance of concrete: (2016) 615–621, https://doi.org/10.4028/www.scientific.net/KEM.711.615.
limestone addition effect, Mag. Concr. Res. 69 (2017) 84–106, https://doi.org/ [39] K.S. Nguyen, A.T. Nguyen-Phung, H.T. Le, T.T. Ho, T.H. Nguyen-Ngoc, S.P. Yap,
10.1680/jmacr.16.00371. N. Chijiwa, N. Otsuki, Chloride binding ability and anti-corrosion properties of
[8] A.A. Elgalhud, R.K. Dhir, G. Ghataora, Chloride ingress in concrete: limestone supersulfated cement in seawater/sand mixing concrete, in: H.-.-H. Tran-Nguyen,
addition effects, Mag. Concr. Res. 70 (2018) 292–313, https://doi.org/10.1680/ H. Wong, F. Ragueneau, C. Ha-Minh (Eds.), Proceedings of the 4th congrès
jmacr.17.00177. international de géotechnique - ouvrages -structures, Ouvrages
[9] S. Palm, T. Proske, M. Rezvani, S. Hainer, C. Müller, C.-A. Graubner, Cements -StructuresSpringer Singapore, Singapore, 2018, pp. 367–376, 10.1007/978-981-
with a high limestone content – Mechanical properties, durability and ecological 10-6713-6_36.
characteristics of the concrete, Constr. Build. Mater. 119 (2016) 308–318, [40] D. Novak, H. Sommer, A new low-heat sulfate resistant binder for mass concrete,
https://doi.org/10.1016/j.conbuildmat.2016.05.009. HPC and SCC, Proceedings of the International Conference Held at the University
[10] R.K. Dhir, M.C. Limbachiya, M.J. McCarthy, A. Chaipanich, Evaluation of of Dundee, Scotland. 5 (2002) 213–222. https://doi.org/10.1680/scc.31777.
Portland limestone cements for use in concrete construction, Mater. Struct. 40 [41] R. Masoudi, Examining compositions, hydration mechanisms and properties of
(2007) 459–473, https://doi.org/10.1617/s11527-006-9143-7. supersulfated cement for use in concrete, University of Toronto, Thesis, 2018.
[11] ] D. Hooton, M. Nokken, M. Thomas, Portland-Limestone Cement: State-of-the- [42] ATILH, Ciments et LHR : analyse et inventaire du cycle de vie (ACV/ICV),
Art Report and Gap Analysis For CSA A 3000, SN3053. (2007). déclaration environnementale produit (DEP), Infociments. (2022). https://www.
[12] S. Bernal, Advances in near-neutral salts activation of blast furnace slags, RILEM infociments.fr/ciments/ciments-declaration-environnementale-inventaire-
Techn. Lett. 1 (2016) 39, https://doi.org/10.21809/rilemtechlett.2016.8. analyse-du-cycle-de-vie.
[13] P. Awoyera, A. Adesina, A critical review on application of alkali activated slag as [43] ecoinvent, ecoinvent Database, Ecoinvent. (2022). https://ecoinvent.org/the-
a sustainable composite binder, Case Stud. Constr. Mater. 11 (2019) e00268. ecoinvent-database/ (accessed November 23, 2022).
[14] H. Kühl, Verfahren zur Herstellung von Zement aus Hochofenschlacke, German [44] INIES, INIES: Les données environnementales et sanitaires de référence pour le
Patent No. 237777, 1908. bâtiment, (2017). https://www.base-inies.fr/iniesV4/dist/consultation.html.
[15] S.-D. Wang, K.L. Scrivener, P.L. Pratt, Factors affecting the strength of alkali- [45] AFNOR, NF P18-459 : Essai de porosité et de masse volumique, 2010.
activated slag, Cem. Concr. Res. 24 (1994) 1033–1043, https://doi.org/10.1016/ [46] AFNOR, EN 12390-8: Profondeur de pénétration d’eau sous pression, 2009.
0008-8846(94)90026-4. [47] AFNOR, ISO-15148 : Détermination du coefficient d’absorption d’eau par
[16] E. Douglas, J. Brandstetr, A preliminary study on the alkali activation of ground immersion partielle, 2003.
granulated blast-furnace slag, Cem. Concr. Res. 20 (1990) 746–756, https://doi. [48] AFNOR, XP P18-463 : Essai de perméabilité aux gaz sur béton durci, 2011.
org/10.1016/0008-8846(90)90008-L. [49] R. Polder, C. Andrade, B. Elsener, Ø. Vennesland, J. Gulikers, R. Weidert,
[17] S.-D. Wang, Alkaline activation of slag, University of London, Thesis, 1995. M. Raupach, RILEM TC 154-EMC recommendations: Test methods for on site
[18] D.M. Roy, Alkali-activated cements Opportunities and challenges, Cem. Concr. measurement of resistivity of concrete, Mater. Struct. 33 (2000) 603–611.
Res. 29 (1999) 249–254, https://doi.org/10.1016/S0008-8846(98)00093-3. [50] M. Cyr, P. Rivard, F. Labrecque, A. Daidié, High-pressure device for fluid
[19] X. Zhang, K. Long, W. Liu, L. Li, W.-J. Long, Carbonation and Chloride Ions’ extraction from porous materials: application to cement-based materials, J. Am.
Penetration of Alkali-Activated Materials: A Review, Molecules 25 (2020) 5074, Ceram. Soc. 91 (2008) 2653–2658, https://doi.org/10.1111/j.1551-
https://doi.org/10.3390/molecules25215074. 2916.2008.02525.x.
[20] A. Wang, Y. Zheng, Z. Zhang, K. Liu, Y. Li, L. Shi, D. Sun, The durability of alkali- [51] AFNOR, EN 12390-11 : Détermination de la résistance du béton à la pénétration
activated materials in comparison with ordinary portland cements and concretes: des chlorures, diffusion unidirectionnelle, 2015.
a review, Engineering 6 (2020) 695–706, https://doi.org/10.1016/j. [52] ASTM C1202–12, Test Method for Electrical Indication of Concretes Ability to
eng.2019.08.019. Resist Chloride Ion Penetration, ASTM Internat. (2012), https://doi.org/
[21] O.A. Mohamed, A review of durability and strength characteristics of alkali- 10.1520/C1202-12.
activated slag concrete, Materials. 12 (2019) 1198, https://doi.org/10.3390/ [53] Z. Zhang, Y. Zhu, H. Zhu, Y. Zhang, J.L. Provis, H. Wang, Effect of drying
ma12081198. procedures on pore structure and phase evolution of alkali-activated cements,
[22] G. Yang, J. Zhao, Y. Wang, Durability properties of sustainable alkali-activated Cem. Concr. Compos. 96 (2019) 194–203, https://doi.org/10.1016/j.
cementitious materials as marine engineering material: A review, Mater. Today cemconcomp.2018.12.003.
Sustainab. 17 (2022), 100099, https://doi.org/10.1016/j.mtsust.2021.100099. [54] M. Kovtun, Effect of preconditioning on durability indices of alkali-activated
[23] X. Sun, Y. Zhao, J. Qiu, J. Xing, Review: alkali-activated blast furnace slag for eco- concretes, SP 326. (2018) 19.1-19.6.
friendly binders, J. Mater. Sci. 57 (2022) 1599–1622, https://doi.org/10.1007/ [55] AFNOR, NF EN 14629 : Mesurage du taux de chlorure d’un béton durci, 2007.
s10853-021-06682-8. [56] H. Hornain, GranDuBé : grandeurs associées à la durabilité des bétons, Presses
[24] J. Provis, Alkali activated materials: state-of-the-art report, RILEM TC 224-AAM, Des Ponts. (2007).
Springer, New York, 2013. [57] A. Darquennes, B. Espion, S. Staquet, How to assess the hydration of slag cement
[25] X. Ke, Improved Durability and Sustainability of Alkali-Activated Slag Cements, concretes? Constr. Build. Mater. 40 (2013) 1012–1020, https://doi.org/10.1016/
University of Sheffield, Thesis, 2017. http://etheses.whiterose.ac.uk/17557/ j.conbuildmat.2012.09.087.
(accessed March 20, 2020). [58] P. Azar, G. Samson, F. Deby, H. Lahalle, V. Benavent, V. Trincal, M. Cyr, Chloride-
[26] J. Zhang, Y. Ma, J. Hu, H. Wang, Z. Zhang, Review on chloride transport in alkali- induced corrosion in ordinary and alkali-activated concrete. 11th ACI/RILEM
activated materials: Role of precursors, activators and admixtures, Constr. Build. International Conference on Cementitious Materials and Alternative Binders for
Mater. 328 (2022), 127081, https://doi.org/10.1016/j. Sustainable Concrete, 2021.
conbuildmat.2022.127081. [59] V. Trincal, V. Benavent, H. Lahalle, G. Samson, C. Patapy, Y. Jainin, M. Cyr,
[27] J. Osio-Norgaard, J.P. Gevaudan, W.V. Srubar, A review of chloride transport in Preconditioning of Specimens - Drying Influence on Alkali-Activated and
alkali-activated cement paste, mortar, and concrete, Constr. Build. Mater. 186 Geopolymer Mortar, in. XV International Conference on Durability of Building
(2018) 191–206, https://doi.org/10.1016/j.conbuildmat.2018.07.119. Materials and Components. eBook of Proceedings, 2020.
[28] AFNOR, NF EN 15743 : Ciment sursulfaté - Composition, spécifications et critères [60] K. Yang, C. Yang, B. Magee, S. Nanukuttan, J. Ye, Establishment of a
de conformité, 2010. preconditioning regime for air permeability and sorptivity of alkali-activated slag
[29] M. Papadakis, M. Venuat, J. Vandamme, Industrie de la chaux, du ciment et du concrete, Cem. Concr. Compos. 73 (2016) 19–28, https://doi.org/10.1016/j.
plâtre, Dunod, Paris, 1970. cemconcomp.2016.06.019.

16
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

[61] I. Ismail, S.A. Bernal, J.L. Provis, S. Hamdan, J.S.J. van Deventer, Drying-induced review and new approaches, Adv. Cem. Res. 23 (2011) 265–275, https://doi.org/
changes in the structure of alkali-activated pastes, J. Mater. Sci. 48 (2013) 10.1680/adcr.2011.23.6.265.
3566–3577, https://doi.org/10.1007/s10853-013-7152-9. [88] L.G. Baquerizo, T. Matschei, K.L. Scrivener, Impact of water activity on the
[62] K. Ndiaye, M. Cyr, S. Ginestet, Durability and stability of an ettringite-based stability of ettringite, Cem. Concr. Res. 79 (2016) 31–44, https://doi.org/
material for thermal energy storage at low temperature, Cem. Concr. Res. 99 10.1016/j.cemconres.2015.07.008.
(2017) 106–115, https://doi.org/10.1016/j.cemconres.2017.05.001. [89] C. Andrade, R. d’Andrea, N. Rebolledo, Chloride ion penetration in concrete: The
[63] S. Mindess, J.F. Young, D. Darwin, Concrete, Prentice Hall, USA, 1981. reaction factor in the electrical resistivity model, Cem. Concr. Compos. 47 (2014)
[64] V. Trincal, V. Benavent, H. Lahalle, B. Balsamo, G. Samson, C. Patapy, Y. Jainin, 41–46, https://doi.org/10.1016/j.cemconcomp.2013.09.022.
M. Cyr, Effect of drying temperature on the properties of alkali-activated binders - [90] D. Sun, Z. Cao, C. Huang, K. Wu, G. De Schutter, L. Zhang, Degradation of
Recommendations for sample preconditioning, Cem. Concr. Res. 151 (2022), concrete in marine environment under coupled chloride and sulfate attack: A
106617, https://doi.org/10.1016/j.cemconres.2021.106617. numerical and experimental study, Case Stud. Constr. Mater. 17 (2022) e01218.
[65] Q. Ma, S.V. Nanukuttan, P.A.M. Basheer, Y. Bai, C. Yang, Chloride transport and [91] H. Ye, L. Huang, Z. Chen, Influence of activator composition on the chloride
the resulting corrosion of steel bars in alkali activated slag concretes, Mater. binding capacity of alkali-activated slag, Cem. Concr. Compos. 104 (2019),
Struct. 49 (2016) 3663–3677, https://doi.org/10.1617/s11527-015-0747-7. 103368, https://doi.org/10.1016/j.cemconcomp.2019.103368.
[66] A. Noushini, Durability of Geopolymer Concrete in Marine Environment, The [92] ASTM International, ASTM C1218 / C1218M–17, Standard Test Method for
University of New South Wales, Thesis, 2018. Water-Soluble Chloride in Mortar and Concrete, West Conshohocken, PA, 2017.
[67] Q. Yu, X. Wan, T. Zhao, T. Wang, X. Han, Z. Sun, Investigation on Resistance of [93] T. Altunbulduk, H.M. zu Köcker, W. Frenzel, Studies on the elimination of sulfide
Chloride Penetration of Alkali Activated Slag Concrete and Related Electrical Test interference in the potentiometric determination of chloride using ion selective
Methods, Materials Reports. 36 (2022). https://doi.org/10.11896/cldb. electrodes in a flow injection system, Fresenius J. Anal. Chem. 351 (1995)
20120067. 593–598, https://doi.org/10.1007/BF00323331.
[68] C. Chalhoub, Study of the initiation and propagation phases of chloride induced [94] F. Pargar, D. Koleva, Klaas Van Breugel, Determination of chloride content in
corrosion in reinforced concrete structures, Université Toulouse 3 Paul Sabatier, cementitious materials: from fundamental aspects to application of Ag/AgCl
Thesis (2020). chloride sensors, Sensors 17 (2017) 2482, https://doi.org/10.3390/s17112482.
[69] P. Azarsa, R. Gupta, Electrical Resistivity of Concrete for Durability Evaluation: A [95] D. Wu, Y. Hu, Y. Liu, R. Zhang, Review of Chloride Ion Detection Technology in
Review, Adv. Mater. Sci. Eng. 2017 (2017) 1–30, https://doi.org/10.1155/2017/ Water, Appl. Sci. 11 (2021) 11137, https://doi.org/10.3390/app112311137.
8453095. [96] Technical comittee ISO/TC 147/SC 2, ISO 9297:1989, Qualité de l’eau — Dosage
[70] C. Andrade, M. Castellote, Chloride aging factor of concrete measured by means des chlorures — Titrage au nitrate d’argent avec du chromate comme indicateur
of resistivity, in: International Conference on Durability of Building Materials and (Méthode de Mohr), 1989.
Components Porto -Portugal, 2011, p. 8. [97] X. Ke, S.A. Bernal, O.H. Hussein, J.L. Provis, Chloride binding and mobility in
[71] S.A. Bernal, J.L. Provis, R.J. Myers, R. San Nicolas, J.S.J. van Deventer, Role of sodium carbonate-activated slag pastes and mortars, Mater. Struct. 50 (2017)
carbonates in the chemical evolution of sodium carbonate-activated slag binders, 252, https://doi.org/10.1617/s11527-017-1121-8.
Mater. Struct. 48 (2015) 517–529, https://doi.org/10.1617/s11527-014-0412-6. [98] Tianjin Harbour Engineering Institute, JTJ 270-1998: Testing code of concrete for
[72] S. Mundra, G. Samson, G. Masi, R. Achenbach, D.M. Bastidas, S.A. Bernal, M. port and waterwog engineering, 1998.
C. Bignozzi, M. Criado, M. Cyr, N. Gartner, S. von Greve-Dierfeld, A. Legat, [99] S. Mundra, D.P. Prentice, S.A. Bernal, J.L. Provis, Modelling chloride transport in
A. Nikoonasab, J.L. Provis, M. Raupach, G.J.G. Gluth, Application of alkali-activated slags, Cem. Concr. Res. 130 (2020), 106011, https://doi.org/
electrochemical methods for studying steel corrosion in alkali-activated 10.1016/j.cemconres.2020.106011.
materials, maco.202313743, Mater. Corros. (2023), https://doi.org/10.1002/ [100] J.-P. Ollivier, M. Carcassès, J.-P. Bigas, O. Truc, Diffusion des chlorures dans le
maco.202313743. béton saturé, Revue Française de Génie Civil. 6 (2002) 227–250, https://doi.org/
[73] A. Gruskovnjak, B. Lothenbach, L. Holzer, R. Figi, F. Winnefeld, Hydration of 10.1080/12795119.2002.9692363.
alkali-activated slag: comparison with ordinary Portland cement, Adv. Cem. Res. [101] L. Bertolini, F. Lollini, E. Redaelli, Influence of concrete composition on
18 (2006) 119–128, https://doi.org/10.1680/adcr.2006.18.3.119. parameters related to the durability of reinforced concrete structures., in:
[74] M. Nedeljković, B. Ghiassi, S. van der Laan, Z. Li, G. Ye, Effect of curing Guimarães, Portugal, 2007.
conditions on the pore solution and carbonation resistance of alkali-activated fly [102] F. Lollini, E. Redaelli, L. Bertolini, Effects of portland cement replacement with
ash and slag pastes, Cem. Concr. Res. 116 (2019) 146–158, https://doi.org/ limestone on the properties of hardened concrete, Cem. Concr. Compos. 46
10.1016/j.cemconres.2018.11.011. (2014) 32–40, https://doi.org/10.1016/j.cemconcomp.2013.10.016.
[75] R. Pethig, Dielectric properties of body tissues, Clin. Phys. Physiol. Meas. 8 (1987) [103] F. Lollini, E. Redaelli, L. Bertolini, A study on the applicability of the efficiency
5–12, https://doi.org/10.1088/0143-0815/8/4A/002. factor of supplementary cementitious materials to durability properties, Constr.
[76] S. Mundra, J.L. Provis, Mechanisms of passivation and chloride-induced corrosion Build. Mater. 120 (2016) 284–292, https://doi.org/10.1016/j.
of mild steel in sulfide-containing alkaline solutions, J. Mater. Sci. 56 (2021) conbuildmat.2016.05.031.
14783–14802, https://doi.org/10.1007/s10853-021-06237-x. [104] X. Ke, S.A. Bernal, J.L. Provis, B. Lothenbach, Thermodynamic modelling of phase
[77] N. Olsson, B. Lothenbach, V. Baroghel-Bouny, L.-O. Nilsson, Unsaturated ion evolution in alkali-activated slag cements exposed to carbon dioxide, Cem. Concr.
diffusion in cementitious materials – The effect of slag and silica fume, Cem. Res. 136 (2020), 106158, https://doi.org/10.1016/j.cemconres.2020.106158.
Concr. Res. 108 (2018) 31–37, https://doi.org/10.1016/j. [105] S. Dean, M. Bassuoni, M. Nehdi, T. Greenough, Enhancing the Reliability of
cemconres.2018.03.007. Evaluating Chloride Ingress in Concrete Using the ASTM C 1202 Rapid Chloride
[78] K.Y. Yeau, E.K. Kim, An experimental study on corrosion resistance of concrete Penetrability Test, J. ASTM Int. 3 (2006) 13403, https://doi.org/10.1520/
with ground granulate blast-furnace slag, Cem. Concr. Res. 35 (2005) 1391–1399, JAI13403.
https://doi.org/10.1016/j.cemconres.2004.11.010. [106] Y. Bu, J. Weiss, The influence of alkali content on the electrical resistivity and
[79] D.M. Roy, W. Jiang, M.R. Silsbee, Chloride diffusion in ordinary, blended, and transport properties of cementitious materials, Cem. Concr. Compos. 51 (2014)
alkali-activated cement pastes and its relation to other properties, Cem. Concr. 49–58, https://doi.org/10.1016/j.cemconcomp.2014.02.008.
Res. 30 (2000) 1879–1884, https://doi.org/10.1016/S0008-8846(00)00406-3. [107] Ph. Turcry, F. Cassagnabère, F. Jacquemot, M. Guéguen Minerbe, National Project
[80] İ. Türkmen, M. Gavgalı, R. Gül, Influence of mineral admixtures on the PERFDUB: General report of working group GT1, 2021.
mechanical properties and corrosion of steel embedded in high strength concrete, [108] D.J.S. Carrinho, Resistance to chloride-ion penetration of different concrete
Mater. Lett. 57 (2003) 2037–2043, https://doi.org/10.1016/S0167-577X(02) mixtures subjected to long-term immersion tests, Mater. Sci. 11 (2018).
01136-9. [109] M. Alexander, M. Thomas, Service life prediction and performance testing —
[81] IREX, Projet National PERFDUB, (2022). https://www.perfdub.fr/. Current developments and practical applications, Cem. Concr. Res. 78 (2015)
[82] J. Zhang, Y. Ma, J. Zheng, J. Hu, J. Fu, Z. Zhang, H. Wang, Chloride diffusion in 155–164, https://doi.org/10.1016/j.cemconres.2015.05.013.
alkali-activated fly ash/slag concretes: Role of slag content, water/binder ratio, [110] E. Koenders, K. Imamoto, A. Soive, eds., Benchmarking Chloride Ingress Models
alkali content and sand-aggregate ratio, Constr. Build. Mater. 261 (2020), on Real-life Case Studies—Marine Submerged and Road Sprayed Concrete
119940, https://doi.org/10.1016/j.conbuildmat.2020.119940. Structures: State-of-the-Art Report of the RILEM TC 270-CIM, Springer
[83] O. Kayali, M. Khan, M.S. Ahmed, The role of hydrotalcite in chloride binding and International Publishing, Cham, 2022. https://doi.org/10.1007/978-3-030-
corrosion protection in concretes with ground granulated blast furnace slag, Cem. 96422-1.
Concr. Compos. 34 (2012) 936–945, https://doi.org/10.1016/j. [111] J.R. Lliso-Ferrando, I. Gasch, A. Martínez-Ibernón, M. Valcuende, Effect of
cemconcomp.2012.04.009. macrocell currents on rebar corrosion in reinforced concrete structures exposed to
[84] A. Noushini, A. Castel, Performance-based criteria to assess the suitability of a marine environment, Ocean Eng. 257 (2022), 111680, https://doi.org/
geopolymer concrete in marine environments using modified ASTM C1202 and 10.1016/j.oceaneng.2022.111680.
ASTM C1556 methods, Mater. Struct. 51 (2018) 146, https://doi.org/10.1617/ [112] Q. Tran, P. Ghosh, P. Lehner, P. Konečný, Determination of Time Dependent
s11527-018-1267-z. Diffusion Coefficient Aging Factor of HPC Mixtures, KEM. 832 (2020) 11–20,
[85] M. Moranville-Regourd, S. Kamali-Bernard, Cements Made From Blastfurnace https://doi.org/10.4028/www.scientific.net/KEM.832.11.
Slag, in: Lea’s Chemistry of Cement and Concrete, Elsevier, 2019: pp. 469–507. [113] C. Andrade, M. Castellote, R. d’Andrea, Measurement of ageing effect on chloride
https://doi.org/10.1016/B978-0-08-100773-0.00010-1. diffusion coefficients in cementitious matrices, J. Nucl. Mater. 412 (2011)
[86] A. Hegazy, A. Khalil, E. El-Alfi, M. El-Shahat, Durability of supersulphated cement 209–216, https://doi.org/10.1016/j.jnucmat.2010.12.236.
pastes activated with Portland cement in magnesium chloride solution, Egypt. J. [114] U. Angst, B. Elsener, C.K. Larsen, Ø. Vennesland, Critical chloride content in
Chem. 62 (2019) 1145–1155. https://doi.org/10.21608/ejchem.2019.6563.1 reinforced concrete — A review, Cem. Concr. Res. 39 (2009) 1122–1138, https://
579. doi.org/10.1016/j.cemconres.2009.08.006.
[87] A. Gruskovnjak, B. Lothenbach, F. Winnefeld, B. Münch, R. Figi, S.-C. Ko, [115] U. Angst, Chloride induced reinforcement corrosion in concrete: Concept of
M. Adler, U. Mäder, Quantification of hydration phases in supersulfated cements: critical chloride content – methods and mechanisms, Thesis, Norges teknisk-

17
L. Doussang et al. Construction and Building Materials 407 (2023) 133511

naturvitenskapelige universitet, Fakultet for ingeniørvitenskap og teknologi, [118] Y. Gao, Y. Zheng, J. Zhang, J. Wang, X. Zhou, Y. Zhang, Randomness of critical
accessed February 21, 2023, Institutt for konstruksjonsteknikk (2011), https://n chloride concentration of reinforcement corrosion in reinforced concrete flexural
tnuopen.ntnu.no/ntnu-xmlui/handle/11250/236720. members in a tidal environment, Ocean Eng. 172 (2019) 330–341, https://doi.
[116] U.M. Angst, Predicting the time to corrosion initiation in reinforced concrete org/10.1016/j.oceaneng.2018.11.038.
structures exposed to chlorides, Cem. Concr. Res. 115 (2019) 559–567, https:// [119] F. Deby. Approche probabiliste de la durabilité des bétons en environnement
doi.org/10.1016/j.cemconres.2018.08.007. marin, Thesis, University of Toulouse, 2008.
[117] U.M. Angst, O.B. Isgor, C.M. Hansson, A. Sagüés, M.R. Geiker, Beyond the [120] AFNOR, FD P 18-480, 2022.
chloride threshold concept for predicting corrosion of steel in concrete, Appl.
Phys. Rev. 9 (2022), 011321, https://doi.org/10.1063/5.0076320.

18