Antecedente 6

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Cement and Concrete Research 144 (2021) 106446

Contents lists available at ScienceDirect

Cement and Concrete Research


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

Carbonation of model cement pastes: The mineralogical origin of


microstructural changes and shrinkage
E. Kangni-Foli a, b, c, S. Poyet c, *, P. Le Bescop c, T. Charpentier d, F. Bernachy-Barbé c,
A. Dauzères a, E. L’Hôpital a, J.-B. d’Espinose de Lacaillerie b
a
IRSN, Institute of Radiation Protection and Nuclear Safety, PSE-ENV/SEDRE/LETIS, BP 17, F-92262 Fontenay aux Roses, France
b
Soft Matter Science and Engineering, UMR CNRS 7615, ESPCI Paris, Université PSL, 10 rue Vauquelin, 75005 Paris, France
c
Université Paris-Saclay, CEA, Service d’Etude du Comportement des Radionucléides, 91191 Gif-sur-Yvette, France
d
NIMBE, CEA, CNRS, Université Paris-Saclay, CEA Saclay, F-91191 Gif-sur-Yvette cedex, France

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

Keywords: This study explored the reactive processes of atmospheric carbonation and the consequences with respect to
Cement cementitious materials. Two model pastes were used: hydrated C3S (including C-S-H and portlandite) and a paste
Carbonation prepared by hydrating a blend of C3S and nanosilica (including C-S-H only). The two pastes were carbonated
C-S-H
under accelerated conditions in the laboratory. The resulting mineralogical assemblage was examined using X-
Shrinkage
Microstructure
ray diffraction, thermogravimetric analysis and nuclear magnetic resonance. The microstructural changes were
studied by X-ray tomography and porosimetry, and their macroscopic impacts were evaluated through gas
diffusion and shrinkage measurements. The use of model pastes allowed for the evaluation of the change in solid
volume induced by the carbonation of C-S-H. C-S-H decalcification and subsequent silica chain polymerisation
were found responsible for carbonation shrinkage (and potentially cracking). Finally, the results highlight the
protective role of portlandite: portlandite helped in limiting C-S-H decalcification and then reducing carbonation
shrinkage and cracking.

1. Introduction mineralogical profiles that are needed for validating geochemical sim­
ulations [12].
Atmospheric carbonation of cementitious materials refers to the re­ Similarly, the link between the changes in chemistry, microstructure
action between atmospheric carbon dioxide (CO2) and the calcium ions and macroscopic properties has not been clearly described until now.
(Ca2+) that are present in the pore solution [1]. This reaction leads to the For instance, the change in total porosity induced by carbonation re­
precipitation of calcium carbonate [2,3] at the expense of Ca-bearing mains uncertain: it is trivial when considering portlandite carbonation,
hydrates; among them, calcium hydroxide (portlandite, CH) and cal­ but the case of C-S-H is more complex. This is mainly due to the fact that
cium silicate hydrates (C-S-H). The main consequence of carbonation is the C-S-H molar volume is still uncertain and probably variable:
a strong pH reduction of the pore solution that can initiate active different values ranging from 68 to 126 cm3/mol have been used so far
corrosion of steel reinforcement [4,5]. Atmospheric carbonation has for high C/S C-S-H (ca. 1.5–1.7) [13–25]. Furthermore, the molar vol­
long been studied, and its mechanisms are now well-known and ume of the end-product of C-S-H carbonation – silica gel – also remains
described. This has made it possible to propose models to describe the uncertain with values ranging from 12 to 30 cm3/mol [26–28].
carbonation of concretes [6–11]. Another point of interest is carbonation shrinkage [29–33] and the
Nonetheless, the available models and the subsequent descriptions of cracking that can result from it [34–37]. Carbonation-induced cracking
carbonation are far from being fully satisfactory and some issues remain. was shown to depend on, among other things, the mineralogical
Although portlandite and calcium carbonate can be easily quantified composition (mainly portlandite and C-S-H contents) of the considered
using thermogravimetric analysis (TGA), the other Ca-bearing hydrates material [35]: blended cement specimens (such as CEM III, CEM V and
(C-S-H and sulfo-aluminates for instance) cannot. It is then difficult to low-alkalinity blends) cracked upon carbonation, which led to a strong
obtain comprehensive and reliable experimental results, such as increase in their transport properties whereas, on the contrary, it

* Corresponding author at: CEA Saclay, DES/ISAS/DPC/SECR/LECBA, B158 PC25, F-91191 Gif-sur-Yvette cedex, France.
E-mail address: [email protected] (S. Poyet).

https://doi.org/10.1016/j.cemconres.2021.106446
Received 10 November 2020; Received in revised form 16 March 2021; Accepted 24 March 2021
Available online 9 April 2021
0008-8846/© 2021 Elsevier Ltd. All rights reserved.
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

decreased in Portland cement specimens because they remained almost Table 2


crack-free. Because of the cracking, carbonation blurs the link between Composition and main properties of the two model pastes [42].
microstructure and transport as it is currently accepted: the finer the C3S paste (C/S = C-S-H paste (C/S = Unit
pore-size distribution, the lower the transport properties [38–41]. 3.0) 1.4)
In view of those considerations, it is the objective of this study to use C/S of the C-S-H 1.7a 1.4b –
two model pastes [42], the mineralogical composition of which was Porosity (drying at 38% 52% –
designed to be able to study the impact of carbonation on C-S-H alone, 80 ◦ C)c
independently of portlandite, and to obtain quantitative data. The Porosity (drying at 41% 56% –
105 ◦ C)c
changes in mineralogy and microstructure were probed concurrently to Saturated density 1.89 1.76 g/cm3
allow for investigation of the consequences of carbonation in terms of Amount of C-S-Hd 5.6 7.0 mol/L of
mineralogical volume changes and macroscopic shrinkage. The results paste
also highlight the central role of portlandite: the presence of portlandite Amount of CHe 7.3 0.0 mol/L of
paste
did reduce the carbonation rate, but it also helped in mitigating the
Young’s modulus Ef 15 ± 1 10 ± 1 GPa
consequences of carbonation (shrinkage and cracking).
a
Assumed value but not verified.
b
Measured using SEM-EDX.
2. Materials c
Drying at 105 ◦ C was used to enable comparison with the literature, but
drying at 80 ◦ C was considered to be less aggressive and more realistic.
Two model pastes were used in this study: a tricalcium silicate (C3S) d
Using the approach of Olson & Jennings [13].
paste (C/S = 3.0), namely a matrix of C-S-H embedding portlandite e
Using TGA and the tangential method [52].
crystals and another one composed of C-S-H only with a target C/S ratio f
Using the impulse excitation technique (GrindoSonic®).
of 1.4. Note that the latter is more or less the highest value that can be
obtained without precipitating portlandite [43–49]. The former is higher w/b ratio required to maintain workability of the fresh paste. The
hereafter referred to as the C3S paste and the latter as the C-S-H paste. difference in water content resulted in a different microstructure for the
The C-S-H paste was prepared by mixing triclinic C3S with the stoi­ two pastes (see Section 4.2).
chiometric amount of nanosilica. The triclinic C3S was provided by
Mineral Research Processing (Meyzieu, France); and it had a specific
3. Methods
surface area of 4600 cm2/g. The nanosilica was provided by BASF
(Rheomac AS150) as a colloidal suspension in water (50% in weight);
3.1. Accelerated carbonation
the nanosilica had a median diameter equal to 100 nm. The main goal of
the design was to obtain pourable pastes. A water-to-binder ratio of 0.50
After curing and prior to carbonation, all the samples were intro­
was enough to ensure workability of the C3S paste (Table 1). However,
duced into a CO2-free glovebox and kept in a desiccator (equipped with
due to the high water demand of the nanosilica, the w/b ratio of the C-S-
fans to avoid moisture gradients), in which the RH was maintained at
H paste had to be significantly increased in order to maintain work­
55% at ambient temperature using a saturated solution of Mg(NO3)2
ability [50,51]. It was then decided to use a commercial superplasticizer
[53]. After 170 days of drying (when the mass variation became con­
(Masterglenium Sky 537 from BASF) to limit the increase in w/b
stant), the specimens were covered with adhesive aluminium foils to
resulting in the composition reported in Table 1. Further details on the
ensure unidirectional carbonation. They were then introduced into the
design, batching and main properties of the pastes are provided else­
carbonation chamber [54] where they were carbonated at 25 ◦ C ±
where [42].
0.2 ◦ C, 55% RH ± 1% and 3.0% CO2 ± 0.2%. This CO2 content was
For each paste, two kinds of specimens were produced using PTFE
selected because it has been proven that accelerated carbonation at 3%
cylindrical moulds (∅ = 14 mm, H = 90 mm; and ∅ = 30 mm, H = 120
CO2 gave results that were similar to those of natural carbonation in
mm). The latter were designed to measure length variations and, more
terms of mineralogy, microstructure and cracking [55]. Accelerated
specifically, the carbonation shrinkage (see Section 3.5). After casting,
carbonation was monitored for nearly one year (326 and 316 days for
all the samples were kept for one month in their sealed moulds before
the C3S and C-S-H pastes, respectively).
they were unmoulded and then kept for two additional months in
airtight containers immersed in a specific solution to prevent leaching
(deionised water and portlandite in excess for the C3S paste and 3.2. Mineralogy
deionised water with crushed C-S-H samples for the C-S-H paste).
Some of the main properties of the two pastes are listed in Table 2, X-ray diffraction (XRD) was used to identify the change in miner­
and more information can be found in our preceding work [42]. The C3S alogy induced by carbonation. The cylinders were cut orthogonally to
paste displayed portlandite and C-S-H contents comparable to those of a the carbonation direction and the diffractograms were acquired on the
fully hydrated OPC paste, despite a very small residual signal of anhy­ (polished) solid surface [56], using a PANalytical X’Pert diffractometer
drous tricalcium silicate detected by X-ray diffraction (XRD) [42]. The equipped with a X’celerator detector using Cu-Kα radiation (λ = 1.54 Å).
C-S-H paste showed a higher porosity and lower saturated density The diffractograms were collected at ambient temperature from 5◦ to
compared with those of the C3S paste. This is the direct result of the 60◦ in 2θ with a step size of 0.017◦ and a dwell time of 40 s resulting in a
total test time of about 20 min. The diffractograms were compared to the
Table 1 following ICDD reference files: portlandite (CH), 44–1481; vaterite, 01-
Composition of the model pastes for 1 L of paste. 74-1487; calcite, 05–0586; aragonite, 70–9854; and triclinic tricalcium
C3S paste (C/S = 3.0) C-S-H paste (C/S = 1.4) Unit
silicate (C3S), 31–0301. Rietveld analyses were run for quantification
purposes; mainly to quantify the relative amounts of the three CaCO3
C3S (g) 1218.8 791.7 g
polymorphs (cf. Fig. 6). In hydrated cement pastes, quantitative XRD
Nanosilica (g) 0.0 237.9 g
Water (g) 609.4 617.8 g analysis is known to be difficult due to the presence of poorly crystalline
w/b ratio 0.50 0.63a – C-S-H [57]. In the carbonated materials, the presence of amorphous
Superplasticiser 0.0 31.1b g silica gel and amorphous calcium carbonate (that cannot be easily
Batching duration 5 8 min accounted for) increased the difficulty and the corresponding uncer­
a
Includes the water introduced by the superplasticiser. tainty. Rietveld analyses were thus not expected to give an accurate
b
Represents 3% of the binder weight (C3S + nanosilica). quantification of the different phases in presence, rather a rough

2
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

estimate. Here we made use of the C-S-H models proposed by Richard­


son [58] to describe the C-S-H contribution to the experimental dif­
fractogram. Practically speaking, all the sixteen the C-S-H models
proposed by Richardson (in addition to portlandite, calcite, aragonite
and vaterite) were introduced into the Rietveld fitting tool of the Mal­
vern Panalytical Highscore Plus software that automatically adjusted the
mass fraction of each component to fit the experimental diffractogram.
Thermogravimetric analysis (TGA) was performed using a Netzsch
STA 409 PC Luxx apparatus. Analyses were run using 120 mg of
powdered sample under a constant N2 flowrate (80 mL/min). The
weight losses were recorded from 25 ◦ C to 1150 ◦ C with a heating rate of
10 ◦ C/min [59]. The quantification of the amounts of portlandite and
calcium carbonate was performed using the tangential method [52],
considering the mass loss between 400 ◦ C and 600 ◦ C for portlandite and
between 600 ◦ C and 900 ◦ C for calcium carbonates.
A single pulse 29Si magic angle spinning nuclear magnetic resonance
29
( Si MAS NMR) experiment was performed on a Bruker Avance III 500
spectrometer operating at the Larmor frequency resonance of 99.3 MHz.
Fig. 1. Illustration of the evaluation of the carbonation front position using
The operational conditions were set to π/2 pulses of 3.5 μs recycle delays equation equation (1) (Lmin = 1000; Lmax = 5000; A = 1; z0 = 25 mm).
of 20 s, spinning in a 7 mm zirconia rotor at 5.5 kHz, and a minimum of
4000 scans for each spectrum. Tetramethylsilane was used as an external
phenolphthalein: samples were cut along their vertical axis and the fresh
standard (0 ppm) to report the chemical shifts. The 29Si NMR spectra
surfaces were sprayed with phenolphthalein. The carbonated parts (with
gave an insight into the C-S-H structure. The silicate environments are
pH value lower than 9.0) remained colourless, whereas non‑carbonated
specified as Qn with 0 ≤ n ≤ 4 where n stands for the number of silicates
parts (with higher pH values) turned purple.
connected to a tetrahedral silicate denoted by Q [60–62]. Similar to
Density and total porosity of the pastes were measured using the
tobermorite - the C-S-H natural crystalline analogous [58,63–65] – three
buoyancy method using disks (Ø = 30.0 mm, H = 8 mm) that were cut
main environments can be found in C-S-H characterised by n values
from the central part of larger cylinders (∅ = 30 mm, H = 120 mm).
between 0 and 2 [43,66]. The first corresponds to the silicate end chain
Prior to the tests, the samples were re-saturated under vacuum [69]. The
unit, namely Q1, with resonance of about − 80 ppm. The other two are
mass loss associated with an oven drying at 80 ◦ C and 105 ◦ C enabled the
Q2 sites, which include Q2b and Q2p environment, that are associated to
total (accessible) porosity of the samples to be estimated.
the bridging (superscript b) tetrahedral silicate with resonance around
Some samples were crushed in centimetric pieces, immersed in liquid
− 83 ppm and the pairing (superscript p) silicates around − 85 ppm,
nitrogen and then freeze-dried for 24 h before they were tested using
respectively. Those environments are the three units of the tobermorite
mercury intrusion porosimetry (MIP). Four samples were used for each
dreierketten structure [64,67]. Silicate in C-S-H forms chains with var­
analysis. A Micromeritics Autopore IV device was used at a maximal
iable lengths, high calcium content in C-S-H leading to a higher pro­
pressure of 414 MPa (corresponding to pores of 3 nm in diameter).
portion of low coordination silicates Q1 [58], i.e. C-S-Hs with shorter
silicate chains. The 29Si MAS NMR results were processed using the
latest version of an in-house software developed previously [68]. The 3.4. Gas diffusion
spectra were fitted using Gaussian-Lorentzian line shapes with chemical
shifts and parameters included in the Appendix. The gas diffusion coefficient of the carbonated pastes was measured
using the through-diffusion method [70,71]. The tests were conducted
using three disks (Ø = 30.0 mm, H = 8 mm) each time. The samples to be
3.3. Microstructure tested were placed between two chambers; the first filled with synthetic
air – 20% of oxygen (O2) and 80% of nitrogen (N2) – and the second one
X-ray micro-Computed Tomographic (X-μCT) projections were ac­ filled with nitrogen (N2) (Fig. 2). Note that the conditions in the two
quired using a Bruker SkyScan 1173 device equipped with a flat panel chambers were the same (ambient temperature at atmospheric pressure
detector (2240 × 2240 pixels) using the following tube operating con­ and 55% RH). Also note that, due to the presence of water vapour, the
ditions: 115–130 kV and 61–69 μA. A full 360◦ set of projections was gas concentration within the two chambers was slightly lower than
obtained with a rotational step of 0.3◦ , exposure time of 1100 ms, a 100%, but this was not accounted for in the assessment of the gas
frame averaging from 8 to 10, and images with a pixel size of 16.8 μm. A diffusion coefficients.
grey level profile along the vertical axis was plotted and averaged over Prior to the tests, the carbonated disks (Ø 30 × 8 mm) were first
“slices” of two voxels in thickness. The carbonation front was obtained saturated in a desiccator under water with a vacuum [69] and then
by applying a logistic function (E- 1) to the sigmoid part of the curve
obtained by plotting the grey level with respect to the position (the
RH controlled
slices). The carbonation front was delimited as follows: Its beginning
was defined as the position associated with 104% of the lower grey value
(Lmin) and its end by the position associated with 96% of the highest
value (Lmax). In other words, the carbonation front was located between
the position values associated with 0.96Lmax and 1.04Lmin (see Fig. 1). Upstream Downstream
(20% O2, 80% N2) (100% N2)
(Lmax − Lmin )
L(z) = Lmin + (1)
1 + exp[A(z0 − z) ]

where L(z) is the grey level for the position z in the sample; Lmin and Lmax
minimal and maximal grey levels in the profile respectively; A positive Cement paste
constant and z0 position of the sigmoid midpoint.
The position of the carbonation front was also characterised using Fig. 2. Principle of the gas diffusion test.

3
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

equilibrated at 55% RH and ambient temperature (20 ◦ C ± 2 ◦ C). They sensors (Solartron Metrology).
were then sealed in position using epoxy adhesive and O-rings to avoid
gas leakage during the diffusion test. The respective diffusion of oxygen 4. Results
and nitrogen through the disks was monitored using gas micro-
chromatography, and the resulting diffusion coefficients were evalu­ 4.1. Mineralogy
ated using Fick’s 2nd law [72].
In parallel to the direct measurements, the simple model proposed by Fig. 3 presents some of the diffractograms that were collected in the
Papadakis et al. [73,74] was used to assess the diffusion coefficient of non‑carbonated zone; in the carbonation front and in the carbonated
CO2 of the two carbonated pastes. This proved to be necessary in some zone of the two pastes. It is evident that the carbonation of the C-S-H
cases because of cracking. The model by Papadakis creates a link be­ paste (C/S = 1.4) led to the dissolution of the C-S-H and precipitation of
tween the carbonation depth xc and the square root of time t: calcium carbonate: mainly vaterite and to some extent calcite. This was
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ consistent with the conclusion of Šauman [75] and Auroy et al. [55],
√̅ 2DcCO2 [CO2 ]
xc (t) = k t with k = (2) which stated that the presence of vaterite was indicative of C-S-H
[Ca]carbo carbonation. In the carbonation front (40.8 mm from the surface
exposed to CO2), the coexistence of calcium carbonate and C-S-H (broad
where k is the carbonation rate; DCO2c is the diffusion coefficient value peak at 29◦ [44,76,77]) could be clearly observed. In the carbonated
for CO2 through the carbonated material; [CO2] is the concentration of zone (15.0 mm), the signal related to C-S-H could no longer be observed,
CO2 and [Ca]carbo is the concentration of Ca that can be involved in the and a very large diffraction band could be observed between 18◦ and
carbonation process. Consequently, the value of the CO2 diffusion co­ 27◦ , which is characteristic of amorphous silica gel (Fig. 4) [78].
efficient DCO2c can be estimated if the carbonation rate k is measured The carbonation of the C3S paste was somewhat different: por­
experimentally: tlandite was never completely dissolved (even in the carbonated zone)
(
[Ca]carbo
) and calcite was the main calcium carbonate that was detected. The C-S-
DcCO2 = k2 (3) H was also not believed to have completely dissolved (even though it
2[CO2 ]
was virtually impossible to detect them due to the presence of calcite)
More details are provided in the Results section. because of the absence of the large diffraction band discussed above.
This was confirmed using 29Si NMR (Fig. 7).
3.5. Shrinkage measurement Fig. 5 presents the mineralogical assemblage obtained using XRD.
Note that these results should only be interpreted in a qualitative
Two of the smaller cylinders (Ø = 14 mm and H = 80 mm) were first manner. In the carbonated zone of the C-S-H paste, the C-S-H signal was
re-saturated under vacuum [69] before they were introduced in a cli­ below the detection limit; the C-S-H were then assumed to be completely
matic chamber where they were subjected to drying at 25 ◦ C and 55% and evenly depleted (this was confirmed using 29Si NMR results, Fig. 7).
RH (note that the specimens were not protected against the CO2 present Conversely, portlandite and C-S-H remained in significant amounts in
in the atmosphere during that phase). After 195 days of drying, they the carbonated zone of the C3S paste. The latter also appeared not to be
were introduced into the accelerated carbonation chamber where they homogeneous: the first four millimetres were more carbonated than the
were subjected to 3% CO2 at 25 ◦ C and 55% RH for a further 70 days. rest of the carbonated zone in which the amount of calcium carbonate
The change in length was monitored throughout the test using LVDT was reduced and the amounts of remaining portlandite and C-S-H

v
v
v
v
C-S-H paste (C/S = 1.4)

v
Intensity (au)

v c v
c c v
Carbonated zone
C-S-H + c
vv v
v v
v c c v v Carbonation front
C-S-H
C-S-H
C-S-H
Non-carbonated zone

5 10 15 20 25 30 35 40 45 50 55 60
2 K -Cu (°)

p
c
C3S paste (C/S = 3.0)

p c p cc p
Intensity (au)

c p c p c Carbonated zone
p C-S-H p
c p
p c c c c p p Carbonation front
p
p C-S-H
p
p p p
Non-carbonated zone

2 K -Cu (°)

Fig. 3. Diffractograms of the pastes at the end of accelerated carbonation (316 and 326 days for the C-S-H and C3S pastes, respectively).

4
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

polymorph that was precipitated in the carbonated zone of the C3S


paste. Aragonite and vaterite were only detected close to the surface
exposed to CO2. There is no clear explanation for the absence of vaterite
in the depth of the carbonated zone of the C3S paste, despite the
carbonation of its C-S-H. Nonetheless, phenomena that might explain
Intensity (au)

the observed behaviour include: discrepancies in Ca-supersaturation


[79–81] in pH values [82] (since formation of calcite is reported at
Silica gel saturation index (SI) higher than that of vaterite), the presence of
foreign ions (sulfate, silica or magnesium) [83–85], or the effect of the
nucleation surface [86] as disturbance of the solvated surface, which can
decrease the energy growth barrier to the order of thermal energy.
In the carbonated zone of the C-S-H paste (Fig. 7-a), and between
0 and 40 mm from the exposed surface, no signal related to C-S-H was
detected using 29Si NMR. Rather, large resonances (Q3gel and Q4gel ca.
100 and 110 ppm, respectively) could be observed. This indicated that
the C-S-H was completely degraded/dissolved and that silica chains
polymerised to form an amorphous silica product (silica gel) [87]. This
confirmed the observations made using XRD (Figs. 3 and 5). In the
2 K -Cu (°) carbonation front (around 43.8–44.6 mm), there was decreased evi­
dence of silica gel but the C-S-H was obviously affected by carbonation,
Fig. 4. Evidence of the silica gel presence in the C-S-H paste carbonated zone because the relative intensities of the Q1 and Q2 environments were
(C/S = 1.4).
different from those of the non‑carbonated C-S-H (beyond 47.6 mm) and
typical of a more polymerised environment. This indicated that the C/S
increased. ratio of the C-S-H had decreased [88–91] and that Ca was taken from the
Rietveld refinements were conducted to estimate the relative pro­ C-S-H structure.
portions of the three CaCO3 polymorphs (vaterite, aragonite and calcite) Once again, the results from the C3S paste were somewhat different
in the carbonated zones of the two pastes (Fig. 6). The results confirmed from those of the C-S-H paste (Fig. 7-b). The C-S-H was almost
the observations of Fig. 3. Vaterite was the main polymorph that was completely degraded very close to the surface exposed to CO2 (0–2 mm),
precipitated in the carbonated zone of the C-S-H paste (around 90% by but a significant amount remained in the carbonated zone (between 2
weight). It must be noted though that some calcite (less than 10%) was and 19 mm), with a C/S ratio decrease that was evidenced by the
detected throughout the carbonated zone as well as aragonite close to decrease of the relative intensities of the Q1 and Q2 compared to the
the surface exposed to CO2. Conversely, calcite was, by far, the main non‑carbonated part.
Intensity (arbitrary unit)
(a) C-S-H paste (C/S = 1.4)

C-S-H
CaCO3

0 5 10 15 20 25 30 35 40 45 50 55
Depth (mm)
b
n

n
r

r
t

C-S-H
CaCO3
Intensity (arbitrary unit)
(b) C3S paste (C/S = 3.0)

CH

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Depth (mm)

Fig. 5. Mineralogical assemblage after accelerated carbonation (316 and 326 days for the C-S-H and C3S pastes respectively) obtained using XRD.

5
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

100%

Polymorphic abundance
80%

C-S-H paste (C/S = 1.4)


60% Vaterite
Calcite
40% Aragonite

20%

0%
0 5 10 15 20 25 30 35 40 45 50 5
Depth (mm)

100%
Polymorphic abundance

80% Aragonite
C3S paste (C/S = 3.0)

Vaterite
60% Calcite

40%

20%

0%

Depth (mm)

Fig. 6. CaCO3 polymorphic abundance (%wt) in the carbonated zone after accelerated carbonation (316 and 326 days for the C-S-H (top) and C3S pastes (bottom)
respectively) obtained using Rietveld refinement. The vertical bar represents the uncertainty of the quantification using Rietveld refinement (evaluated to 5%).

Q2p Q1
Q2p
Pristine Q0
46.6-47.4 mm Pristine
2b
Q 1Q 27.9-30.1 mm
43.8-44.6 mm

41.0-41.8 mm 20.4-21.4 mm

36.0-39.0 mm 16.6-18.2 mm

15.0-20.0 mm 6.8-7.6 mm
Q4gel Q4gel
Q3gel 0-3.6 mm Q3gel 0-2 mm

δ (ppm) 29 Si δ (ppm) 29 Si

29
Fig. 7. Si NMR spectra of the C-S-H and C3S pastes after accelerated carbonation (316 and 326 days for the C-S-H and C3S pastes, respectively).

The quantitative mineralogical profile of calcium carbonate (CC)


Q1 + Q2b + Q2p
that was obtained using TGA is presented in Fig. 8(a) for the C-S-H paste %Si(C − S − H) = (4)
∑4
(C/S = 1.4). For the purposes of comparison, the calcium carbonate Qn
profile is presented together with that of C-S-H that was assessed from n=0

the 29Si NMR results using:


where %Si(C − S − H) stands for the atomic fraction (at.%) of total
silicon present in the C-S-H.
The carbonation depth was about 40–45 mm and the carbonation
front was quite sharp (Fig. 8-a). This was confirmed by both TGA (for

6
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

Fig. 8. Quantitative mineralogical profiles obtained using TGA and NMR after accelerated carbonation (316 and 326 days for the C-S-H and C3S pastes, respectively).
The horizontal bar represents the thickness of the disk that was sampled and tested using TGA. The vertical bar represents the uncertainty of measurement (evaluated
to 5%). Note that the scale of the %Si(C-S-H) profile obtained using NMR was modified to ensure comparison with the %Ca(C-S-H) profile obtained using TGA.

calcium carbonate CC) and 29Si NMR (for C-S-H). The Ca content in ( )
CaCO3 in the carbonated zone appeared to be lower than 100% (by %Ca(C − S − H) = 100% − %Ca(CH) − %Ca CC (5)
about 10% on average). This did not mean that some C-S-H remained in
The results indicate that, in the carbonated zone, the C-S-H retained
the carbonated zone, as its absence was unambiguously proven using
29 almost 60% of its Ca. This was more or less consistent with the results of
Si NMR (Figs. 7 & 8-a). Rather, this could to be due to some discrep­ 29
Si NMR even though both of them strongly differed near the cylinder
ancy in the TGA quantification, because CaCO3 has been shown to
surface (Fig. 8-b). This result made it possible to evaluate the amount of
sometimes decompose at temperatures lower than 600 ◦ C [3,6,92,93],
Ca that was involved in the carbonation of each paste (Table 3). For the
or this could indicate that some Ca was present in the silica gel instead of
C3S paste, not all the Ca of CH or C-S-H was involved in carbonation: it
precipitating as CaCO3. In this latter case, considering that all the silica
corresponded only to 75% of CH and 40% of the C-S-H, whereas it
was present under the form of silica gel and that 10% of the Ca remained
corresponded to all the C-S-H for the C-S-H paste. Furthermore, despite a
in the silica gel (Fig. 8-a), it was easy to estimate the C/S ratio of the
strong difference in their initial Ca content, the carbonation of the two
silica gel: 10% of the C/S of the paste = 0.1 × 1.4 = 0.14.
pastes appeared to have yielded similar amounts of CaCO3 (9.3 versus
The situation was not so clear for the C3S paste (Fig. 8-b): the
9.8 mol/L).
carbonation front was not well defined and the carbonation depth was
lower than that of the C-S-H paste by two times: about 17–22 mm. As
already shown by XRD (Figs. 3 & 5), portlandite remained in the 4.2. Microstructure/cracking
carbonated zone, and its residual content was estimated to be 25% of the
initial content (Table 3). Comparing the amounts of Ca in portlandite % Carbonation led to a decrease in total porosity (from 10% to 20%)
( )
Ca(CH) and calcium carbonate %Ca CC obtained using TGA, it was and an increase in density (from 0.32 to 0.39 g/cm3) (Table 4) as has
been commonly observed [39,73,94–97]. It is noteworthy that the
possible to estimate the amount of Ca that remained in the C-S-H %Ca(C
change in dry density induced by carbonation (i.e. induced by the fix­
− S − H) as follows:
ation of CO2) was quite similar for the two pastes. This is surely
consistent with the fact that the amounts of carbonated Ca were similar
for the two pastes (9.8 and 9.3 mol/L of paste for the C-S-H and C3S
Table 3 pastes, respectively, see Table 3). The reduction in porosity was twice
Amount of calcium involved in the carbonation reaction.
higher for the C-S-H than for the C3S paste, but this was explained simply
Paste Amount of Ca carbonated Value (mol/L of paste) by considering the phases that were present in the carbonated pastes and
C3S paste 75 % of [CH] + 40 % of [C − S − 0.75 × 7.3 + 0.4 × 1.7 × 5.6 = the corresponding volumes (see Section 5.1).
H] 9.3 The pore size distribution was also significantly modified by
C-S-H 100 % of [C − S − H] 1.4 × 7.0 = 9.8 carbonation (Fig. 9). In both cases, the bimodal distribution was shifted
paste
to an almost unimodal distribution located near 10 nm [35,55], the

7
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

Table 4 carbonation and C-S-H decalcification. The second (II, between 10 and
Porosity and density of the carbonated pastes. 40 mm) was somewhat different. The cracks were thinner than those of
C-S-H paste (C/S C3S paste (C/S = Unit the first zone and almost parallel. In addition, strong heterogeneities in
= 1.4) 3.0) grey levels could be observed (Fig. 12). Here, it was assumed that cracks
Density (saturated) 1.98 2.11 g/ were generated by restrained carbonation shrinkage and that CO2 was
cm3 able to quickly diffuse into these open cracks, helping in carbonating the
Total porosity (drying at 80 ◦ C) 35% 28% – matrix close to the cracks. This was believed to have induced the
Total porosity (drying at 36% 29% – observed heterogeneities. It must be noted, though, that no significant
105 ◦ C)
Dry density (drying at 80 ◦ C) 1.63 1.83 g/
differences were obtained between zones I and II using XRD and NMR.
cm3 The third zone (III) corresponded to what appeared as the carbon­
Dry density (drying at 105 C) ◦
1.62 1.82 g/ ation front (between 40 and 45 mm) using TGA (Fig. 8-a) and XRD
cm3 (Fig. 5-a), where C-S-H were progressively decalcified. Similar to zone II,
Change in porosity (drying at − 20% − 10% –
zone III presented strong heterogeneities in grey level. This zone was
80 ◦ C)a
Change in dry density (drying +0.39 +0.32 g/ thought to be cracked (but with openings smaller than the XCT resolu­
at 80 ◦ C)a cm3 tion, assumed to be less than 1 μm) which led to the coexistence of
a carbonated and non‑carbonated paste within the same region. The last
See Table 2 for the porosity and density values of the non‑carbonated pastes.
zone (IV) corresponded to the non‑carbonated zone (almost uniform
dark grey level).
intensity of which seemed to depend on the degree of carbonation (close
Examination of the C3S paste showed three main zones (Fig. 13). The
to 100% for the C-S-H paste and limited for the C3S paste). In the
first (I, the first five millimetres from the surface) appeared in light grey
carbonated C-S-H paste, the amount of the C-S-H pores (3 nm in diam­
(denser than the rest) and was then expected to be the most carbonated
eter) was almost nil. This was consistent with the 29Si NMR results that
part of the cylinder. This corresponded quite well with the mineralogical
showed that the C-S-H was depleted after carbonation. This was not the
results that showed that the CaCO3 content was maximal. The grey level
case for the carbonated C3S paste for which 60% of the C-S-H remained
of the second zone (II, from 5 to 20 mm) was darker than that of zone I,
after carbonation (Table 3). Another point of interest was the change in
indicating that it was less dense and then less carbonated. This was
critical pore diameter (the diameter corresponding to the major peak
verified by the mineralogical results that showed that C-S-H and CH
encountered when moving from big to small pores) [98,99]: the pore
remained in significant amounts. The last zone (III) corresponded to the
structure after carbonation was coarsened for the C-S-H paste (the
non‑carbonated part (darker grey). The carbonation front appeared to
critical pore diameter was shifted from 20 nm to 150 nm by carbonation)
be tortuous due to the presence of few cracks that locally provided
whereas the critical pore diameter was reduced from 35 nm to 20 nm by
pathways for CO2 and increased the carbonation depth close to the
carbonation (for the C3S paste). On this basis, carbonation would then be
cracks.
expected to increase the transport properties (and then gas diffusion) of
the C-S-H paste and to reduce those of the C3S paste. Note that this
4.3. Carbonation rate/gas diffusion
divergent behaviour has already been observed for OPC and substituted
binders [34,35,100].
Fig. 14 shows the progress of carbonation depth with the square root
Fig. 10 presents two scans that were obtained using X-μCT after
of time. All the techniques used (X-μCT, phenolphthalein and XRD) were
carbonation. The carbonated zones are clearly visible in light grey. It can
in good agreement. As expected, the carbonation rate of the C-S-H paste
be observed that the carbonation front appeared tortuous in both cases
was higher than that of the C3S paste (2.1 mm/d0.5 versus 1.1 mm/d0.5).
and that the carbonation depth values were consistent with those ob­
Here, the reduced quantity of calcium in the C-S-H paste may have
tained using XRD (Fig. 5) and TGA (Fig. 8). The carbonated zone of the
played a role in increasing the carbonation rate [101–105].
C-S-H paste appeared to be severely cracked, especially for the first 10
The results of the gas diffusion tests also provided valuable infor­
mm. Conversely, the carbonated zone of the C3S paste only shows a few
mation. The value of the diffusion coefficient of the carbonated C3S
thinner cracks.
paste was 2.7 × 10− 8 m2/s. This value is in good agreement with results
Thorough examination of the C-S-H paste scan revealed the presence
usually observed in carbonated OPC pastes [39,100]. Because it was
of four different zones in the cylinder (Fig. 11). The first (I) corre­
severely cracked (Fig. 10), it was impossible to measure the gas diffusion
sponded to the first 10 mm of the carbonated zone. It was severely
coefficient of the C-S-H paste. Instead, the gas diffusion coefficient was
cracked (crazing) and the grey level was almost uniform due to complete
assessed using Papadakis’ model (Eq. (3)). In doing so, the experimental

0.70 0.40
Non-carbo 0.35 Non-carbo
0.60
Carbo Carbo
0.30
dV/dln(D) (mL/g)
dV/dln(D) (mL/g)

0.50
0.25
0.40
0.20
0.30
0.15
0.20
0.10
0.10 0.05
0.00 0.00
0.001 0.01 0.1 1 10 0.001 0.01 0.1 1 10
Pore entry diameter (µm) Pore entry diameter (µm)

Fig. 9. Pore-entry size distribution obtained using MIP after accelerated carbonation.

8
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

Fig. 10. Cracking pattern observed using X-μCT after accelerated carbonation (316 and 326 days for the C-S-H and C3S pastes, respectively).

carbonation rate k was used as input data. Firstly, and for validation paste (Fig. 10). Because the amount of pores of entry diameter 150 nm
purposes, the model by Papadakis was also used to evaluate the diffusion remained limited in the carbonated C-S-H paste, cracking was believed
coefficient of the carbonated C3S paste. Thus, the quantity of carbo­ to be the major cause for the high gas diffusion coefficient in line with
natable calcium [Ca]carbo was estimated using the TGA results (Fig. 8): the conclusions of Auroy et al. [35] and Dutzer et al. [100].
/
[Ca]carbo = 75%of [CH] + 1.7 × 40%of [C − S − H] = 9.3 mol L of paste
4.4. Carbonation shrinkage
(6)
The diffusion coefficient of the carbonated C3S paste was evaluated Fig. 15 presents the length variation of the smallest cylinders (Ø14
as 5.2 × 10− 8 m2/s. This was twice the measured value (2.7 × 10− 8 m2/ mm) that were submitted for drying and then accelerated carbonation.
s), but the order of magnitude of the diffusion coefficient estimated The drying shrinkage of the C3S paste quickly reached a stable value
using Papadakis’ model was correct. For the C-S-H paste, all the C-S-H (0.3% = 3 mm/m) that was consistent with the results of Mindess et al.
were assumed to be available for carbonation, which was supported by [106], whereas the C-S-H paste kept slowly shrinking and was not sta­
the NMR results (Fig. 7). Consequently, E- 3 simply became: bilised even at 195 days. This could be due to (1) the refined pore
( ) structure of the C-S-H paste (Fig. 9) that induced very low water
DcCO2 =
[Ca]carbo 2
k = 1.4k2
[C − S − H]
(7) transport properties and (2) to the atmospheric carbonation (0.04%
2[CO2 ] [CO2 ] CO2) of the C-S-H paste during drying that could have induced some
carbonation shrinkage during the preliminary drying phase. After 195
where days of drying, the specimens were subjected to 3% CO2, which imme­
[Ca]carbo = 1.4 × [C − S − H]= 9.8 mol/L of paste diately generated additional shrinkage. Assuming that there was no
[CO2] = 1.25 × 10− 3 mol/L (3% CO2). interaction between the drying and carbonation, and that drying ended
The value of the diffusion coefficient of CO2 in the carbonated C-S-H at 195 days, the carbonation shrinkage was simply estimated by sub­
paste (C/S = 1.4) was evaluated to be 20.0 × 10− 8 m2/s; one order of tracting the contribution of drying: εc = εtot − εd. The shrinkage of the C-
magnitude greater than the value measured in the carbonated C3S paste. S-H paste appeared to be much higher than that of the C3S paste
This strong difference was explained by two concomitant phenomena: (Table 5).
(1) the critical pore diameter of the carbonated C-S-H paste was higher
than that of the carbonated C3S paste (150 nm versus 20 nm, see Fig. 9)
and (2) the significant cracking induced by carbonation in the C-S-H

9
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

Fig. 11. View of the four observed zones using X-μCT in the C-S-H paste after carbonation (316 days). For comparison, the corresponding mineralogical profile of
Fig. 8(a) is reproduced to scale.

was examined first because it was the simplest system. The non‑carbo­
nated paste was considered to be a porous medium made of only C-S-H
(with C/S = 1.4, cf. Fig. 16). The C-S-H carbonation reaction was
assumed to be the following:

C1.4 − S − H1.4 + 1.4C→1.4CC(Vaterite) + S − Ht + (1.4 − t)H (8)


As observed using XRD, the CaCO3 resulting from the carbonation of
the C-S-H was assumed to be vaterite (Fig. 6). Without further infor­
mation, the H/S ratio of the C-S-H was assumed to be 1.4 so that the C-S-
H specific gravity was close to 2.604 [25], and the end product of the
decalcification of the C-S-H was assumed to be a hydrated silica-gel
(S–H, Fig. 7).
Table 6 and Table 7 detail the volume and mass calculations for the
different phases present in the non‑carbonated and fully carbonated C-S-
H paste (C/S = 1.4).
If we first consider the non‑carbonated C-S-H paste (C/S = 1.4), the
C-S-H molar volume can be easily calculated (please note that here we
used the porosity obtained by drying at 80 ◦ C, Table 2):
/
1 − ∅nc 1 − 0.35
Vm (C − S − H) = = ≈ 69 cm3 mol (9)
[C − S − H] 7
Fig. 12. Close-up of zone II of the C-S-H paste and view of the cracks and
scatter in grey level. The obtained value is close to that measured by Allen et al. [24] (72
cm3/mol) and further used by Jennings [25] (72 cm3/mol for C/S = 1.7)
5. Discussion: linking mineralogy with structural changes and the value computed using molecular dynamics by Pellenq et al. [23]
(70.8 cm3/mol for C/S = 1.65).
5.1. Volume change After complete carbonation of the C-S-H paste (C/S = 1.4), the same
kind of calculation was performed to estimate the properties of the silica
All the above-mentioned results were used to evaluate the change in gel (S–H), namely molar volume Vm(S − H) and molar mass Mm(S − H):
solid volume and weight induced by carbonation. The C/S = 1.4 paste

10
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

Fig. 13. View of the three zones that were observed using X-μCT in the C3S paste after carbonation (326 days). The presence of some cracks was highlighted using
white dashed lines. For comparison, the corresponding mineralogical profile of Fig. 8(b) is reproduced to scale.

/
(1 − ∅c ) − 1.4[C − S − H]Vm (Vat)
Vm (S − H) = ≈ 40 cm3 mol (10)
1.4[C − S − H]
/
C/S = 1.4 (XCT) C/S = 3.0 (phenol) Mm (S − H) =
ρcd − 1.4[C − S − H]Mm (Vat)
≈ 93 g mol (11)
1.4[C − S − H]
C/S = 3.0 (XRD) C/S = 3.0 (XCT) The case of the C3S paste (C/S = 3.0) was more complicated than the
C-S-H paste (C/S = 1.4), because the non‑carbonated paste included
45 portlandite (CH) in addition to C-S-H (Fig. 17). Moreover, carbonation
was incomplete and then the remaining hydrates (CH and C-S-H)
Carbonation depth (mm)

40
coexisted with the end-products of carbonation (silica gel and calcium
35 carbonate) in the carbonated paste (Fig. 17). It must be noted that we
have implicitly assumed that a C-S-H with C/S lower than 1.7 could be
30 described by an ad-hoc combination of silica gel and C-S-H of C/S = 1.7.
25 Alternatively, it would have been necessary to describe the link between
C/S and molar volume just like in the work of Wu & Ye [28], but this
20 would have been much more difficult and uncertain.
Following the mineralogical observations (XRD and 29Si NMR, see
15 Section 4.1), the carbonation of CH and C-S-H was assumed to precipi­
10 tate calcite (neither aragonite nor vaterite) and silica gel (S–H); note
that the C-S-H was considered to be in the form C1.7-S-H1.8 [25]:
5 {
CH + C→CC(Calcite) + H
0 (12)
C1.7 − S − H1.8 + 1.7C→1.7CC(Calcite) + S − Ht + (1.8 − t)H
0 4 8 12 16 20 Table 8 and Table 9 detail the volume and mass calculations of all the
Square root of time (d1/2) different phases present in the non‑carbonated and (partially) carbon­
ated C3S paste (C/S = 3.0).
Fig. 14. Carbonation rates of the two pastes obtained using phenolphthalein, As for the C-S-H paste (C/S = 1.4), the results of the non‑carbonated
XRD and XCT. C3S paste were used to evaluate the molar volume of the C-S-H, ac­
cording to:

11
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

2.0%

Linear shrinkage (%)


1.6%

1.2%

0.8%

0.4%

0.0%
0 50 100 150 200 250 300
Days

Fig. 15. Variation in length of the Ø14 mm cylinders exposed to drying (25 ◦ C 55% RH) and then accelerated carbonation (25 ◦ C, 55% RH and 3% CO2).

Table 5 Table 7
Results of the carbonation shrinkage tests. Volume and mass of the different phases in the fully carbonated C-S-H paste (C/
S = 1.4).
Paste Total shrinkage Drying Carbonation
εtot shrinkage εd shrinkage εc Phase Volume Mass
C-S-H paste (C/S 1.4% (14 mm/ 0.7% (7 mm/m) 0.7% (7 mm/m) Voids Øc 0
= 1.4) m) Vaterite 1.4[C-S-H]Vm(vat) 1.4[C-S-H]Mm(vat)
C3S paste (C/S = 0.5% (5 mm/m) 0.3% (3 mm/m) 0.2% (2 mm/m) S-H (1-Øc)-1.4[C-S-H]Vm(vat) ρcd-1.4[C-S-H]Mm(vat)
3.0) Total 1 ρcd

Øc porosity of the carbonated paste.


ρcd dry density of the carbonated paste.
[C-S-H] concentration of C-S-H in the non‑carbonated paste.
Vm(vat) molar volume of vaterite (38 cm3/mol)
Mm(vat) molar mass of vaterite (100.1 g/mol).

Table 6
Volume and mass of the different phases of the non‑carbonated C-S-H paste
(C/S = 1.4).
Phase Volume Mass

Voids Ønc 0
C-S-H (1-Ønc) ρnc
d
Total 1 ρnc
d

Fig. 16. Assumed effect of carbonation on the C-S-H paste; the C-S-H are fully Ønc porosity of the non‑carbonated paste.
carbonated and replaced by silica gel and calcium carbonate (vaterite). ρnc
d dry density of the non‑carbonated paste.

(1 − ∅nc ) − [CH]0 Vm (CH) (1 − 0.38) − 7.3 × 0.033


Vm (C − S − H) = =
[C − S − H] 5.6
/
≈ 68 cm3 mol The molar volume of silica gel was similar to that obtained for the C-
S-H paste (37 versus 40 cm3/mol). The molar mass was however
(13)
somewhat different (71 versus 93 g/mol). It should be noted that the
The obtained value (68 cm3/mol) was very close to that of the C-S-H values obtained using the results from the C3S paste were much more
paste (69 cm3/mol). In a second step, the results of the (partially) uncertain than those of the C-S-H paste (C/S = 1.4), mainly because of
carbonated C3S paste were used to evaluate the properties of the silica the remaining CH and C-S-H, which made it difficult to accurately es­
gel. timate the volume and mass of the silica gel.

/
(1 − ∅c ) − [CH]r Vm (CH) − [Cal]Vm (Cal) − [C − S − H]r Vm (C − S − H)
Vm (S − H) = ≈ 37 cm3 mol (14)
[C − S − H]0 − [C − S − H]r
/
ρcd − [CH]r Mm (CH) − [Cal]Mm (Cal) − [C − S − H]r Mm (C − S − H)
Mm (S − H) = ≈ 71 g mol
[C − S − H]0 − [C − S − H]r

12
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

Fig. 17. Effect of the carbonation on the C3S paste, the carbonation products and partially carbonated hydrated coexist.

(Table 10), resulting in the reduction of total porosity. Indeed, the


Table 8
change in volume depends on the calcium carbonate polymorph that is
Volume and mass of the different phases in the non‑carbonated C3S paste (C/S =
precipitated and on the C/S ratio of the C-S-H. However, the increase in
3.0).
solid volume induced by the carbonation of C-S-H is expected to lie
Phase Volume Mass between 19 and 35 cm3/mol. This means that the carbonation of high C/
Voids Ønc 0 S C-S-H (between 1.4 and 1.7) cannot be expected to generate an in­
CH [CH]0Vm(CH) [CH]0Mm(CH) crease in porosity as was observed by Wu and Ye [28] and Shah et al.
C-S-H (1-Ønc)-[CH]0Vm(CH) ρnc
d -[CH]0Mm(CH)
[109].
Total 1 ρnc
d
( ) ( )
[CH]0 concentration of portlandite in the non‑carbonated paste. C
∆Vc (C − S − H) = Vm (S − H) + Vm CC − Vm (C − S − H) (16)
Vm(CH) molar volume of portlandite (33 cm3/mol). S
Mm(CH) molar mass of portlandite (74.1 g/mol).
The change in solid volume induced by the carbonation of C-S-H is
one order of magnitude greater than that of portlandite (which ranges
Table 9 between 1 and 5 cm3/mol). C-S-H is then likely to contribute more to
Volume and mass of the different phases in the carbonated C3S paste (C/S = 3.0). porosity reduction than portlandite. This was confirmed in the case of
the C3S paste for which the change in solid volume induced by the
Phase Volume Mass
carbonation of both CH and C-S-H was evaluated (calcite was assumed
Voids Øc 0 to be the only calcium carbonate polymorph precipitated):
CH [CH]rVm(CH) [CH]rMm(CH)
C-S-H [C-S-H]rVm(C-S-H) [C-S-H]rMm(C-S-H) ∆VC− = 40%[C − S − H]0 ∆Vc (C − S − H) = 0.4 × 5.6 × 30.5
S− H
Calcite [Cal]Vm(Cal) [Cal]Mm(Cal) /
S-H (1-Øc)-Vm(CH)-Vm (Cal)-Vm (C-S-H) ρcd-Mm (CH)-Mm (Cal)-Mm (C-S-H) ≈ +68 cm3 L of paste
Total 1 ρcd

[CH]r concentration of the remaining portlandite in the carbonated paste. ∆VCH = 75%[CH]0 (Vm (Cal) − Vm (CH) ) = 0.75 × 7.3 × 2
/
[C-S-H]r concentration of the remaining C-S-H in the carbonated paste. ≈ + 11 cm3 L of paste (17)
[Cal] concentration of calcite in the carbonated paste
Vm(Cal) molar volume of calcite (35 cm3/mol) The change in porosity in the C3S paste was mostly due to C-S-H,
Mm(Cal) molar mass of calcite (100.1 g/mol).

Using the values computed above, it is easy to evaluate the change in 0.90
volume induced by the decalcification of the C-S-H, ∆Vd(C − S − H):
/
Chen et al. (2006)
∆Vd (C − S − H) = Vm (S − H) − Vm (C − S − H) ≈ 40 − 69 = − 29 cm3 mol 0.75
C3S paste
(15)
C-S-H paste
Shrinkage (%)

Because one mole of S–H is less voluminous than one mole of C-S-H, 0.60
the decalcification of the C-S-H leads to a significant reduction in the
solid volume (about 29 cm3/mol). This is consistent with the conclusions 0.45
of Chen et al. [107] and with the results of Morandeau et al. [97,108]
who found that ∆Vd(C − S − H) was between 20 and 40 cm3/mol.
Nonetheless, the precipitation of calcium carbonate also plays a signif­ 0.30
icant role in changes in volume during carbonation of C-S-H (E- 12). It is
evident that the reduction of solid volume generated by the decalcifi­
0.15
cation of C-S-H is compensated by the precipitation of calcium carbonate

Table 10 0.00
Change in solid volume (cm3/mol) induced by the carbonation of C-S-H, ac­ 0.0 0.5 1.0 1.5 2.0 2.5 3.0
cording to Eq. (16).
∆Vc(C − S − H) Aragonite Calcite Vaterite Unit (C/S) ratio of the hydrates
3
C/S = 1.4 cm /mol
Fig. 18. Comparison of the carbonation shrinkage results obtained in this study
+18.6 +20.0 +24.4
C/S = 1.7 +28.8 +30.5 +35.4 cm3/mol
with the NH4NO3 degradation data of Chen et al. [107].

13
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

even though the proportion of CH that was carbonated was greater than describe a common trend: the shrinkage increases with the decalcifica­
that of C-S-H. This was consistent with the conclusions of other studies tion level of the C-S-H. This unambiguously confirms that the shrinkage
[97,108,110]. These data might prove useful in evaluating the impact of generated by carbonation is due to the decalcification of C-S-H and
carbonation on transport properties [111–113]. It should be noted, polymerisation of the silica chains as it is the case for decalcification
however, that this is only relevant in the presence of portlandite. Indeed, [107]. At a first glance, it may be surprising that carbonation and
in the absence of portlandite, cracking is expected to prevail and make decalcification generate similar shrinkage because they are both quite
porosity reduction negligible in terms of transport [35,100] (cf. Section different. Carbonation induces CaCO3 precipitation, densification of the
5.3). matrix and an increase in mechanical properties [115–118], whereas
decalcification is responsible for Ca leaching, porosity increase and a
significant decrease in mechanical properties [119–123]. What really
5.2. Carbonation shrinkage actually matters is that carbonation and degradation both induce sig­
nificant decalcification and shrinkage of the C-S-H matrix that embeds
The carbonation shrinkage of the C-S-H paste (which was devoid of all the other minerals. It is the shrinkage of the matrix that induces the
portlandite) was more pronounced intense than that of the C3S paste shrinkage of the cementitious material.
including portlandite (Fig. 18 & Table 5). This strongly suggested that The following 2nd order polynomial was proposed to fit the data:
the dissolution of portlandite was not responsible for carbonation
shrinkage, which contradicts Powers’ hypothesis [114]. Rather, εc = 0.264∆2 (C/S) + 0.513∆(C/S) (19)
carbonation shrinkage was related to C-S-H alteration as it was inferred This simple approach allows for estimation of the shrinkage gener­
by Sereda et al. [30,31]. With that in mind, the carbonation shrinkage ated by the decalcification of high C/S C-S-H (between 1.4 and 1.7).
values (Table 5) were compared to those obtained by Chen et al. [107] Please note that this function is not expected to be valid for lower C/S
who studied the shrinkage of C3S samples exposed to accelerated values. More results would be greatly appreciated to verify this.
degradation (using NH4NO3). Chen et al. [107] concluded that
shrinkage was related to the decalcification of C-S-H and subsequent
polymerisation of silica chains. It is obvious from Fig. 18 that the 5.3. The role of portandite
carbonation shrinkage values obtained here are consistent with those
obtained by Chen et al. [107]. The results presented here stress the influence of the mineralogical
However, because of the presence of portlandite, the C/S ratio of the composition on atmospheric carbonation. The C-S-H paste appeared to
C3S paste could not be compared directly to that of the C-S-H paste (free be completely carbonated (no remaining C-S-H after carbonation), and
of portlandite). This shortcoming was easily overcome by computing the the porosity was strongly reduced due to CaCO3 precipitation, but the C-
change in C-S-H C/S ratio. Here we made the choice to characterise the S-H decalcification and the silica chain polymerisation generated sig­
decalcification level of the C-S-H using: nificant shrinkage and cracking that resulted in the increase in diffusion
(C/S)(t) − (C/S)0 coefficient. Conversely, the C3S paste carbonation gave quite different
∆(C/S) = (18) results: carbonation was not complete (portlandite and C-S-H remained
(C/S)0
in significant amounts in the carbonated zone), the porosity was reduced
where (C/S)(t) is the C/S ratio of the C-S-H at time t and (C/S)0 the C/S but less so than for the C-S-H paste. Furthermore, only a few cracks could
ratio of the unaltered C-S-H. The expression ∆(C/S) = 0% means that the be observed in the carbonated zone and the diffusion coefficient was
C-S-H are unaltered, whereas ∆(C/S) = 100% means that the C-S-H are reduced by carbonation.
completely degraded. The values of ∆(C/S) could be easily computed for The only reason why the C3S paste suffered less from carbonation
the tests of Chen et al. [107] because portlandite was completely dis­ compared to the C-S-H paste was due to the presence of portlandite. It
solved when the C/S ratio of the solid (CH + C-S-H) reached 1.4 (see must be recalled that the reaction between Ca2+ ions and (dissolved)
Section 4.1 of [107]). The shrinkage values were then plotted against the CO2 in the pore solution tends to decrease Ca2+ concentration, and thus
decalcification level ∆(C/S) in Fig. 19. Despite some inevitable portlandite dissolves to maintain equilibrium. In the C-S-H paste in
discrepancy, the results from Chen et al. and those from this study which portlandite was absent, Ca was withdrawn from the C-S-H,
leading to their complete decalcification. This generated significant
decalcification shrinkage and cracking and the gas diffusion coefficient
was increased, despite the strong reduction in porosity. In the C3S paste,
portlandite contributed to protecting the C-S-H from decalcification and
reducing the shrinkage induced by carbonation. As a consequence,
portlandite did provide some kind of mechanical protection to the C3S
paste because carbonation-induced cracking remained limited and the
reduction in porosity led to a reduction in the gas diffusion coefficient.
By protecting the C-S-H, portlandite also contributed to protecting
the C3S paste from carbonation. Because carbonation is (partly) driven
by CO2 diffusion through the carbonated layer (see Eq. (2)), the reduc­
tion in diffusion coefficient induced by carbonation enabled low values
(2.7–5.2 × 10− 8 m2/s) to be obtained that helped limit the carbonation
rate of the C3S paste. Conversely, the strong increase in diffusion coef­
ficient of the C-S-H paste (which was induced by carbonation shrinkage
and cracking) yielded a higher value of the diffusion coefficient (≈20.0
× 10− 8 m2/s) and finally a higher carbonation rate. It should be noted
that despite a strong difference in their initial Ca content, the carbon­
ation of the two pastes appeared to yield similar amounts of CaCO3 (9.8
and 9.3 mol/L of paste for the C-S-H and C3S pastes respectively, see
Table 3). Thus, the difference in carbonation rates was expected to be
Fig. 19. Carbonation shrinkage versus decalcification level ∆(C/S) of high C/S mainly due to the difference in diffusion coefficient. This was consistent
C-S-H. with the conclusions of Dutzer et al. [100].

14
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

6. Conclusion [3] W.F. Cole, B. Kroone, Carbon dioxide in hydrated Portland cement, J. Am. Concr.
Inst. 56 (1960) 1275–1296.
[4] L. Bertolini, B. Elsener, P. Pedeferri, R. Polder, Corrosion of Steel in Concrete -
The use of two different model pastes (a C3S paste and a C-S-H paste Prevention, Diagnosis, Repair, Wiley, 2004.
with C/S = 1.4 devoid of portlandite) made it possible to investigate the [5] R. Neves, F.A. Branco, J. de Brito, A method for the use of accelerated
mechanisms and consequences of atmospheric carbonation. Firstly, and carbonation tests in durability design, Constr. Build. Mater. 36 (2012) 585–591,
https://doi.org/10.1016/j.conbuildmat.2012.06.028.
owing to the simplicity of the pastes’ mineralogical assemblage, the [6] M. Thiery, G. Villain, P. Dangla, G. Platret, Investigation of the carbonation front
molar volumes of C-S-H and silica gel (the end-product of C-S-H decal­ shape on cementitious materials: effects of the chemical kinetics, Cem. Concr.
cification) could be estimated. This allowed for evaluation of the change Res. 37 (2007) 1047–1058, https://doi.org/10.1016/j.cemconres.2007.04.002.
[7] B. Bary, A. Sellier, Coupled moisture—carbon dioxide–calcium transfer model for
in solid volume induced by the decalcification and the carbonation of C- carbonation of concrete, Cem. Concr. Res. 34 (2004) 1859–1872, https://doi.org/
S-H. A comparison between the change in volume induced by the 10.1016/j.cemconres.2004.01.025.
carbonation of portlandite indicates that C-S-H is more than likely to be [8] A.V. Saetta, B.A. Schrefler, R.V. Vitaliani, 2-D model for carbonation and
moisture/heat flow in porous materials, Cem. Concr. Res. 25 (1995) 1703–1712,
the main contributor in change in porosity. https://doi.org/10.1016/0008-8846(95)00166-2.
Secondly, the shrinkage induced by carbonation was measured and [9] A.V. Saetta, R.V. Vitaliani, Experimental investigation and numerical modeling of
compared with the NH4NO3 leaching results of Chen et al. [107]. The carbonation process in reinforced concrete structures part I: theoretical
formulation, Cem. Concr. Res. 34 (2004) 571–579, https://doi.org/10.1016/j.
comparison was satisfactory, which confirmed that carbonation cemconres.2003.09.009.
shrinkage was associated with the decalcification of C-S-H and poly­ [10] M. Te Liang, W. Qu, C.H. Liang, S.M. Lin, Mathematical modeling and
merisation of the silica chains. applications for concrete carbonation, J. Mar. Sci. Technol. 11 (2003) 20–33.
[11] A. Silva, R. Neves, J. De Brito, Statistical modelling of carbonation in reinforced
Thirdly, all the results highlight the protective role of portlandite.
concrete, Cem. Concr. Compos. 50 (2014) 73–81, https://doi.org/10.1016/j.
From a chemical point of view, portlandite has long been known to act as cemconcomp.2013.12.001.
a Ca-buffer that dissolves in the first instance to maintain the Ca- [12] N. Seigneur, É. Kangni-Foli, V. Lagneau, A. Dauzères, S. Poyet, P. Le Bescop,
concentration in the pore solution. This helps in limiting the carbon­ E. L’Hôpital, J.B. d’Espinose de Lacaillerie, Understanding the effects of the
atmospheric carbonation of cementitious materials using reactive transport
ation rate, protecting the C-S-H from decalcification and then limiting modelling, Cem. Concr. Res. 130 (2020) 105966, https://doi.org/10.1016/j.
subsequent cracking. In conclusion, the presence of portlandite does cemconres.2019.105966.
indeed provide indirect protection from cracking induced by [13] R.A. Olson, H.M. Jennings, Estimation of C-S-H content in a blended cement paste
using water adsorption, Cem. Concr. Res. 31 (2001) 351–356, https://doi.org/
carbonation. 10.1016/S0008-8846(01)00454-9.
[14] D.P. Bentz, Guide to Using CEMHYD3D: A Three-dimensional Cement Hydration
Data availability and Microstructure Development Modelling Package, 2005.
[15] L. Barcelo, Chemical shrinkage, in: Early Age Crack. Cem. Syst. - RILEM Rep. 025,
RILEM, 2003, pp. 21–25.
The data that supports the results presented in this article is not [16] D. Herfort, B. Lothenbach, Ternary phase diagrams applied to hydrated cement,
meant to be broadly disseminated but it can be provided by the authors in: K. Scrivener, R. Snellings, B. Lothenbach (Eds.), A Pract. Guid. to Microstruct.
Anal. Cem. Mater, CRC Press, Boca Raton (USA), 2016, pp. 485–502.
upon reasonable request. [17] Z.P. Bažant, S. Rahimi-Aghdam, Century-long durability of concrete structures:
Expansiveness of hydration and chemo-mechanics of autogenous shrinkage and
CRediT authorship contribution statement swelling, in: G. Meschke, B. Pichler, J.G. Rots (Eds.), Comput. Model. Concr.
Struct. - Proc. Conf. EURO-C 2018, Taylor & Francis, Bad Hofgastein (Austria),
2018, pp. 15–23, https://doi.org/10.1201/9781315182964-2.
E. KANGNI-FOLI: Investigation, Resources, Visualization, Writing- [18] P.D. Tennis, H.M. Jennings, A model for two types of calcium silicate hydrate in
original draft. the microstructure of Portland cement pastes, Cem. Concr. Res. 30 (2000)
S. POYET: Conceptualization, Methodology, Validation, Data cura­ 855–863.
[19] S. Bishnoi, K.L. Scrivener, μic: A new platform for modelling the hydration of
tion, Supervision, Writing-review & editing. cements, Cem. Concr. Res. 39 (2009) 266–274, https://doi.org/10.1016/j.
P. LE BESCOP: Methodology, Validation, Data curation. cemconres.2008.12.002.
T. CHARPENTIER: Formal analysis, Validation, Data Curation, [20] P. Blanc, X. Bourbon, A. Lassin, E.C. Gaucher, Chemical model for cement-based
materials: temperature dependence of thermodynamic functions for
Writing-review & editing. nanocrystalline and crystalline C-S-H phases, Cem. Concr. Res. 40 (2010)
F. BERNACHY-BARBE: Formal analysis, Methodology, Writing- 851–866, https://doi.org/10.1016/j.cemconres.2009.12.004.
review & editing. [21] B. Lothenbach, T. Matschei, G. Möschner, F.P. Glasser, Thermodynamic
modelling of the effect of temperature on the hydration and porosity of Portland
A. DAUZERES: Conceptualization, Funding acquisition, Supervision, cement, Cem. Concr. Res. 38 (2008) 1–18, https://doi.org/10.1016/j.
Writing-review & editing. cemconres.2007.08.017.
E. L’HÔPITAL: Formal analysis, Methodology, Writing-review & [22] B. Lothenbach, D.A. Kulik, T. Matschei, M. Balonis, L. Baquerizo, B. Dilnesa, G.
D. Miron, R.J. Myers, Cemdata18: a chemical thermodynamic database for
editing. hydrated Portland cements and alkali-activated materials, Cem. Concr. Res. 115
J.-B. D’ESPINOSE DE LACAILLERIE: Supervision, Formal analysis, (2019) 472–506, https://doi.org/10.1016/j.cemconres.2018.04.018.
Writing-review & editing. [23] R.J.-M. Pellenq, A. Kushima, R. Shahsavari, K.J. Van Vliet, M.J. Buehler, S. Yip,
F.-J. Ulm, A realistic molecular model of cement hydrates, Proc. Natl. Acad. Sci.
U. S. A. 106 (2009) 16102–16107, https://doi.org/10.1073/pnas.0902180106.
Declaration of competing interest [24] A.J. Allen, J.J. Thomas, H.M. Jennings, Composition and density of nanoscale
calcium–silicate–hydrate in cement, Nat. Mater. 6 (2007) 311–316, https://doi.
org/10.1038/nmat1871.
No potential conflict of interest was reported by the authors.
[25] H.M. Jennings, Refinements to colloid model of C-S-H in cement: CM-II, Cem.
Concr. Res. 38 (2008) 275–289, https://doi.org/10.1016/j.
Acknowledgments cemconres.2007.10.006.
[26] D.A. Kulik, Aqueous solubility diagrams for cementitious waste stabilization
systems: II, end-member stoichiometries of ideal calcium silicate hydrate solid
The authors gratefully acknowledge financial support from CEA solutions, J. Am. Ceram. Soc. 84 (2001) 3017–3026.
(grant DEN 4425) and IRSN (grant DAFCJ/SJ/LS 20301). The authors [27] D.A. Kulik, Improving the structural consistency of C-S-H solid solution
would like to thank Céline Cau Dit Coumes (CEA Marcoule, France) and thermodynamic models, Cem. Concr. Res. 41 (2011) 477–495, https://doi.org/
10.1016/j.cemconres.2011.01.012.
Barbara Lothenbach (EMPA, Switzerland) for helpful comments. [28] B. Wu, G. Ye, Development of porosity of cement paste blended with
supplementary cementitious materials after carbonation, Constr. Build. Mater.
References 145 (2017) 52–61, https://doi.org/10.1016/j.conbuildmat.2017.03.176.
[29] G.J. Verbeck, Carbonation of hydrated Portland cement, ASTM Spec. Tech. Publ.
205 (1958) 17–36.
[1] H.F.W. Taylor, Cement Chemistry, 2nd ed., Academic Press, London (UK), 1990
[30] K. Kamimura, P.J. Sereda, E.G. Swenson, Changes in weight and dimensions in
https://doi.org/10.1016/S0958-9465(98)00023-7.
the drying and carbonation of Portland cement mortars, Mag. Concr. Res. 17
[2] W.F. Cole, B. Kroone, Carbonate minerals in hydrated Portland cement, Nature.
(1965) 5–14.
184 (1959) BA57.

15
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

[31] E.G. Swenson, P.J. Sereda, Mechanism of the carbonation shrinkage of lime and [56] G. Le Saoût, T. Füllmann, V. Kocaba, K.L. Scrivener, Quantitative study of
hydrated cement, J. Appl. Chem. 18 (1968) 111–117. cementitous materials by x-ray diffraction - Rietveld analysis using an external
[32] Y.F. Houst, Carbonation shrinkage of hydrated cement paste, in: V.M. Malhotra standard, in: J.J. Beaudoin, J.M. Makar, L. Raki (Eds.), Proc. 12th Int. Congr.
(Ed.), 4th CANMET/ACI Int. Conf. Durab. Concr, ACI, Sydney (Australia), 1997, Chem. Cem, Cement Association of Canada, Montréal (Canada), 2007, 12p.
pp. 481–491. [57] R. Snellings, X-ray powder diffraction applied to cement, in: A Pract. Guid. to
[33] Y. Suda, J. Tomiyama, T. Saito, T. Saeki, Effect of relative humidity on Microstruct. Anal. Cem. Mater, 2016, pp. 107–176.
carbonation shrinkage and hydration product of cement paste (in japanese), [58] I.G. Richardson, Model structures for C-(A)-S-H(I), Acta Crystallogr. Sect. B:
Cement Sci. Concr. Technol. 73 (2019) 71–78, https://doi.org/10.14250/ Struct. Sci. Cryst. Eng. Mater. 70 (2014) 903–923, https://doi.org/10.1107/
cement.73.71. S2052520614021982.
[34] P.H.R. Borges, J.O. Costa, N.B. Milestone, C.J. Lynsdale, R.E. Streatfield, [59] G. Villain, M. Thiery, G. Platret, Measurement methods of carbonation profiles in
Carbonation of CH and C-S-H in composite cement pastes containing high concrete: thermogravimetry, chemical analysis and gammadensimetry, Cem.
amounts of BFS, Cem. Concr. Res. 40 (2010) 284–292, https://doi.org/10.1016/j. Concr. Res. 37 (2007) 1182–1192, https://doi.org/10.1016/j.
cemconres.2009.10.020. cemconres.2007.04.015.
[35] M. Auroy, S. Poyet, P. Le Bescop, J.M. Torrenti, T. Charpentier, M. Moskura, [60] E. Lippmaa, M. Magi, A. Samoson, G. Engelhardt, A.-R. Grimmer, Structural
X. Bourbon, Impact of carbonation on unsaturated water transport properties of studies of silicates by solid-state, J. Am. Chem. Soc. Am. 102 (1980) 4889–4893.
cement-based materials, Cem. Concr. Res. 74 (2015) 44–58, https://doi.org/ [61] J.R. Kirkpatrick, Nuclear magnetic resonance spectroscopy, in: V.
10.1016/j.cemconres.2015.04.002. S. Ramachandran, J.J. Beaudouin (Eds.), Handb. Anal. Tech. Concr, Noyes
[36] J. Han, W. Sun, G. Pan, W. Caihui, Monitoring the evolution of accelerated Publications, Park Ridge NJ (USA), 2001, pp. 205–230, https://doi.org/10.1007/
carbonation of hardened cement pastes by X-ray computed tomography, J. Mater. 978-1-62703-176-9_26.
Civ. Eng. 25 (2013) 347–354. [62] J. Skibsted, High-resolution solid-state nuclear magnetic resonance spectroscopy
[37] K. Wan, Q. Xu, Y. Wang, G. Pan, 3D spatial distribution of the calcium carbonate of Portland cement-based systems, in: K.L. Scrivener, R. Snellings, B. Lothenbach
caused by carbonation of cement paste, Cem. Concr. Compos. 45 (2014) 255–263, (Eds.), A Pract. Guid. to Microstruct. Anal. Cem. Mater, Taylor & Francis, Boca
https://doi.org/10.1016/j.cemconcomp.2013.10.011. Raton (USA), 2016, pp. 214–286.
[38] P. Halamickova, R.J. Detwiler, Water permeability and chloride ion diffusion in [63] S. Merlino, E. Bonaccorsi, T. Armbruster, Tobermorites: their real structure and
Portland cement mortars: relationship to sand content and critical pore diameter, order-disorder (OD) character, Am. Mineral. 84 (1999) 1613–1621, https://doi.
Cem. Concr. Res. 25 (1995) 790–802, https://doi.org/10.1016/0008-8846(95) org/10.2138/am-1999-1015.
00069-O. [64] E. Bonaccorsi, S. Merlino, A.R. Kampf, The crystal structure of tobermorite 14 Å
[39] Y.F. Houst, F.H. Wittmann, Influence of porosity and water content on the (plombierite), a C-S-H phase, J. Am. Ceram. Soc. 88 (2005) 505–512, https://doi.
diffusivity of CO2 and O2 through hydrated cement paste, Cem. Concr. Res. 24 org/10.1111/j.1551-2916.2005.00116.x.
(1994) 1165–1176, https://doi.org/10.1016/0008-8846(94)90040-X. [65] I.G. Richardson, The calcium silicate hydrates, Cem. Concr. Res. 38 (2008)
[40] F. Frizon, C. Gallé, Diffusive and convective transport of inert gas through cement 137–158, https://doi.org/10.1016/j.cemconres.2007.11.005.
pastes: influence of microstructure and water, J. Porous Media. 12 (2009) [66] F. Brunet, P. Bertani, T. Charpentier, A. Nonat, J. Virlet, Application of 29 Si
221–237, https://doi.org/10.1615/JPorMedia.v12.i3.30. homonuclear and 1 H-29 Si heteronuclear NMR correlation to structural studies
[41] M. Boumaaza, P. Turcry, B. Huet, A. Aït-mokhtar, Influence of carbonation on the of calcium silicate hydrates, J. Phys. Chem. B 108 (2004) 15494–15502, https://
microstructure and the gas diffusivity of hardened cement pastes, Constr. Build. doi.org/10.1021/jp031174g.
Mater. 253 (2020) 119227, https://doi.org/10.1016/j. [67] C. Biagioni, S. Merlino, E. Bonaccorsi, The tobermorite supergroup: a new
conbuildmat.2020.119227. nomenclature, Mineral. Mag. 79 (2015) 485–495, https://doi.org/10.1180/
[42] É. Kangni-Foli, S. Poyet, P. Le Bescop, T. Charpentier, F. Bernarchy-Barbé, minmag.2015.079.2.22.
A. Dauzères, E. L’Hôpital, M. Neji, J.B. d’Espinose de Lacaillerie, Model synthetic [68] F. Brunet, T. Charpentier, C.N. Chao, H. Peycelon, A. Nonat, Characterization by
pastes for low pH cements, Cem. Concr. Res. 136 (2020) 106168. solid-state NMR and selective dissolution techniques of anhydrous and hydrated
[43] I. Klur, B. Pollet, J. Virlet, A. Nonat, C-S-H structure evolution with calcium CEM V cement pastes, Cem. Concr. Res. 40 (2010) 208–219, https://doi.org/
content by multinuclear NMR, in: Nucl. Magn. Reson. Spectrosc. Cem. Mater, 10.1016/j.cemconres.2009.10.005.
Springer Berlin Heidelberg, Berlin, Heidelberg, 1998, pp. 119–141, https://doi. [69] H. Hornain, G. Arliguie, GranDuBé - Grandeurs associées à la durabilité des
org/10.1007/978-3-642-80432-8_8. bétons (in French), Presses des Ponts et Chaussées, 2007.
[44] K. Garbev, G. Beuchle, M. Bornefeld, L. Black, P. Stemmermann, Cell dimensions [70] J. Sercombe, R. Vidal, C. Gallé, F. Adenot, Experimental study of gas diffusion in
and composition of nanocrystalline calcium silicate hydrate solid solutions. Part cement paste, Cem. Concr. Res. 37 (2007) 579–588, https://doi.org/10.1016/j.
1: synchrotron-based x-ray diffraction, J. Am. Ceram. Soc. 91 (2008) 3005–3014, cemconres.2006.12.003.
https://doi.org/10.1111/j.1551-2916.2008.02484.x. [71] C. Boher, F. Frizon, S. Lorente, F. Bart, Influence of the pore network on hydrogen
[45] G. Renaudin, J. Russias, F. Leroux, F. Frizon, C. Cau-dit-Coumes, Structural diffusion through blended cement pastes, Cem. Concr. Compos. 37 (2013) 30–36,
characterization of C-S-H and C-A-S-H samples-part I: long-range order https://doi.org/10.1016/j.cemconcomp.2012.12.009.
investigated by Rietveld analyses, J. Solid State Chem. 182 (2009) 3312–3319, [72] J. Crank, The Mathematics of Diffusion, 2nd ed., Clarendon Press, Oxford, 1975
https://doi.org/10.1016/j.jssc.2009.09.026. https://doi.org/10.1016/0306-4549(77)90072-X.
[46] Y. Suda, T. Saeki, T. Saito, Relation between chemical composition and physical [73] V.G. Papadakis, C.G. Vayenas, M.N. Fardis, Fundamental modeling and
properties of C-S-H generated from cementitious materials, J. Adv. Concr. experimental investigation of concrete carbonation, ACI Mater. J. 88 (1991)
Technol. 13 (2015) 275–290, https://doi.org/10.3151/jact.13.275. 363–373.
[47] E. Tajuelo Rodriguez, K. Garbev, D. Merz, L. Black, I.G. Richardson, Thermal [74] V.G. Papadakis, M.N. Fardis, C.G. Vayenas, Effect of composition, environmental
stability of C-S-H phases and applicability of Richardson and Groves’ and factors and cement-lime mortar coating on concrete carbonation, Mater. Struct.
Richardson C-(A)-S-H(I) models to synthetic C-S-H, Cem. Concr. Res. 93 (2017) 25 (1992) 293–304.
45–56, https://doi.org/10.1016/j.cemconres.2016.12.005. [75] Z. Šauman, Carbonization of porous concrete and its main binding components,
[48] W. Kunther, S. Ferreiro, J. Skibsted, Influence of the Ca/Si ratio on the Cem. Concr. Res. 1 (1971) 645–662.
compressive strength of cementitious calcium-silicate-hydrate binders, J. Mater. [76] K. Mohan, H.F.W. Taylor, Analytical electron microscopy of cement pastes: IV,
Chem. A 5 (2017) 17401–17412, https://doi.org/10.1039/c7ta06104h. β-dicalcium silicate pastes, J. Am. Ceram. Soc. 64 (1981) 717–719, https://doi.
[49] C. Roosz, P. Vieillard, P. Blanc, S. Gaboreau, H. Gailhanou, D. Braithwaite, org/10.1111/j.1151-2916.1981.tb15893.x.
V. Montouillout, R. Denoyel, P. Henocq, B. Madé, Thermodynamic properties of [77] A. Nonat, The structure and stoichiometry of C-S-H, Cem. Concr. Res. 34 (2004)
C-S-H, C-A-S-H and M-S-H phases: results from direct measurements and 1521–1528, https://doi.org/10.1016/j.cemconres.2004.04.035.
predictive modelling, Appl. Geochem. 92 (2018) 140–156, https://doi.org/ [78] L. Black, C. Breen, J. Yarwood, K. Garbev, P. Stemmermann, B. Gasharova,
10.1016/j.apgeochem.2018.03.004. Structural features of C-S-H(I) and its carbonation in air-a Raman spectroscopic
[50] G. Quercia, G. Hüsken, H.J.H. Brouwers, Water demand of amorphous nano silica study. Part II: carbonated phases, J. Am. Ceram. Soc. 90 (2007) 908–917, https://
and its impact on the workability of cement paste, Cem. Concr. Res. 42 (2012) doi.org/10.1111/j.1551-2916.2006.01429.x.
344–357, https://doi.org/10.1016/j.cemconres.2011.10.008. [79] T. Ogino, T. Suzuki, K. Sawada, The formation and transformation mechanism of
[51] S. Abd.El.Aleem, M. Heikal, W.M. Morsi, Hydration characteristic, thermal calcium carbonate in water, Geochim. Cosmochim. Acta 51 (1987) 2757–2767.
expansion and microstructure of cement containing nano-silica, Constr. Build. [80] J. Kawano, N. Shimobayashi, A. Miyake, M. Kitamura, Precipitation diagram of
Mater. 59 (2014) 151–160, https://doi.org/10.1016/j.conbuildmat.2014.02.039. calcium carbonate polymorphs: its construction and significance, J. Phys.
[52] B. Lothenbach, P. Durdzinski, K. De Weerdt, Thermogravimetric analysis, in: Condens. Matter 21 (2009) 425102, https://doi.org/10.1088/0953-8984/21/42/
K. Scrivener, R. Snellings, B. Lothenbach (Eds.), A Pract. Guid. to Microstruct. 425102.
Anal. Cem. Mater, CRC Press, Boca Raton (USA), 2016, pp. 177–211. [81] M. Kitamura, Strategy for control of crystallization of polymorphs,
[53] L. Greenspan, Humidity fixed points of binary saturated aqueous solutions, J. Res. CrystEngComm. 11 (2009) 949–964, https://doi.org/10.1039/b809332f.
Natl. Bur. Stand. Sect. A Phys. Chem. 81A (1977) 89–96, https://doi.org/ [82] Y.S. Han, G. Hadiko, M. Fuji, M. Takahashi, Crystallization and transformation of
10.6028/jres.081A.011. vaterite at controlled pH, J. Cryst. Growth 289 (2006) 269–274, https://doi.org/
[54] E. Drouet, S. Poyet, P. Le Bescop, J.-M. Torrenti, X. Bourbon, Carbonation of 10.1016/j.jcrysgro.2005.11.011.
hardened cement pastes: influence of temperature, Cem. Concr. Res. 115 (2019) [83] L. Fernández-Díaz, Á. Fernández-González, M. Prieto, The role of sulfate groups in
445–459, https://doi.org/10.1016/J.CEMCONRES.2018.09.019. controlling CaCO3 polymorphism, Geochim. Cosmochim. Acta 74 (2010)
[55] M. Auroy, S. Poyet, P. Le Bescop, J.-M. Torrenti, T. Charpentier, M. Moskura, 6064–6076, https://doi.org/10.1016/j.gca.2010.08.010.
X. Bourbon, Comparison between natural and accelerated carbonation (3% CO2): [84] M. Kellermeier, F. Glaab, R. Klein, E. Melero-García, W. Kunz, J.M. García-Ruiz,
impact on mineralogy, microstructure, water retention and cracking, Cem. Concr. The effect of silica on polymorphic precipitation of calcium carbonate: an on-line
Res. 109 (2018) 64–80, https://doi.org/10.1016/j.cemconres.2018.04.012.

16
E. Kangni-Foli et al. Cement and Concrete Research 144 (2021) 106446

energy-dispersive X-ray diffraction (EDXRD) study, Nanoscale. 5 (2013) 7054, Cem. Concr. Res. 89 (2016) 297–309, https://doi.org/10.1016/j.
https://doi.org/10.1039/c3nr00301a. cemconres.2016.08.019.
[85] Z. Zou, W.J.E.M. Habraken, G. Matveeva, A.C.S. Jensen, L. Bertinetti, M.A. Hood, [104] A. Leemann, P. Nygaard, J. Kaufmann, R. Loser, Relation between carbonation
C.-Y. Sun, P.U.P.A. Gilbert, I. Polishchuk, B. Pokroy, J. Mahamid, Y. Politi, resistance, mix design and exposure of mortar and concrete, Cem. Concr. Compos.
S. Weiner, P. Werner, S. Bette, R. Dinnebier, U. Kolb, E. Zolotoyabko, P. Fratzl, 62 (2015) 33–43, https://doi.org/10.1016/j.cemconcomp.2015.04.020.
A hydrated crystalline calcium carbonate phase: Calcium carbonate hemihydrate, [105] V. Shah, S. Bishnoi, Carbonation resistance of cements containing supplementary
Science (80-. ) 363 (2019) 396–400. cementitious materials and its relation to various parameters of concrete, Constr.
[86] H. Deng, S. Wang, X. Wang, C. Du, X. Shen, Y. Wang, F. Cui, Two competitive Build. Mater. 178 (2018) 219–232, https://doi.org/10.1016/j.
nucleation mechanisms of calcium carbonate biomineralization in response to conbuildmat.2018.05.162.
surface functionality in low calcium ion concentration solution, Regen. Biomater. [106] S. Mindess, J.F. Young, F.V. Lawrence, Creep and drying shrinkage of calcium
2 (2015) 187–195, https://doi.org/10.1093/rb/rbv010. silicate pastes I. Specimen preparation and mechanical properties, Cem. Concr.
[87] I.G. Richardson, G.W. Groves, Models for the composition and structure of Res. 8 (1978) 591–600.
calcium silicate hydrate (CSH) gel in hardened tricalcium silicate pastes, Cem. [107] J.J. Chen, J.J. Thomas, H.M. Jennings, Decalcification shrinkage of cement paste,
Concr. Res. 22 (1992) 1001–1010, https://doi.org/10.1016/0008-8846(92) Cem. Concr. Res. 36 (2006) 801–809, https://doi.org/10.1016/j.
90030-Y. cemconres.2005.11.003.
[88] X. Cong, R. James Kirkpatrick, 17O and 29Si MAS NMR study of β-C2S hydration [108] A. Morandeau, M. Thiéry, P. Dangla, Impact of accelerated carbonation on OPC
and the structure of calcium-silicate hydrates, Cem. Concr. Res. 23 (1993) cement paste blended with fly ash, Cem. Concr. Res. 67 (2015) 226–236, https://
1065–1077, https://doi.org/10.1016/0008-8846(93)90166-7. doi.org/10.1016/j.cemconres.2014.10.003.
[89] X. Cong, R.J. Kirkpatrick, 29Si MAS NMR study of the structure of calcium silicate [109] V. Shah, K. Scrivener, B. Bhattacharjee, S. Bishnoi, Changes in microstructure
hydrate, Adv. Cem. Based Mater. 3 (1996) 144–156, https://doi.org/10.1016/ characteristics of cement paste on carbonation, Cem. Concr. Res. 109 (2018)
S1065-7355(96)90046-2. 184–197, https://doi.org/10.1016/j.cemconres.2018.04.016.
[90] Y. He, L. Lu, L.J. Struble, J.L. Rapp, P. Mondal, S. Hu, Effect of calcium–silicon [110] S.T. Pham, W. Prince, The carbonation of calcium-silicate-hydrate C-S-H in
ratio on microstructure and nanostructure of calcium silicate hydrate synthesized cement mortar studied using thermal analysis and gas pycnometer: determination
by reaction of fumed silica and calcium oxide at room temperature, Mater. Struct. of the quantity of calcium carbonate produced and the increase in molar volume,
47 (2013) 311–322, https://doi.org/10.1617/s11527-013-0062-0. Adv. Mater. Res. 931–932 (2014) 411–415.
[91] M. Castellote, L. Fernandez, C. Andrade, C. Alonso, Chemical changes and phase [111] R.J. Millington, J.P. Quirk, Permeability of porous solids, Trans. Faraday Soc. 57
analysis of OPC pastes carbonated at different CO2 concentrations, Mater. Struct. (1961) 1200–1207, https://doi.org/10.1039/TF9615701200.
42 (2009) 515–525, https://doi.org/10.1617/s11527-008-9399-1. [112] M. Xie, K.U. Mayer, F. Claret, P.A. Diederik, J. Carl, C. Chiaberge, J. Simunek,
[92] S. Goto, K. Suenaga, T. Kado, Calcium silicate carbonation products, J. Am. Implementation and evaluation of permeability-porosity and tortuosity-porosity
Ceram. Soc. 78 (1995) 2867–2872. relationships linked to mineral dissolution-precipitation, Comput. Geosci. 19
[93] G. Villain, G. Platret, Two experimental methods to determine carbonation (2014) 655–671, https://doi.org/10.1007/s10596-014-9458-3.
profiles in concrete, ACI Mater. J. 103 (2006) 265–271. [113] J. Hommel, H. Class, E. Coltman, Porosity – permeability relations for evolving
[94] S.E. Pihlajavaara, Some results of the effect of carbonation on the porosity and pore space: a review with a focus on (bio-)geochemically altered porous media,
pore size distribution of cement paste, Mater. Struct. 1 (1968) 521–527, https:// Transp. Porous Media 124 (2018) 589–629, https://doi.org/10.1007/s11242-
doi.org/10.1007/BF02473640. 018-1086-2.
[95] V.T. Ngala, C.L. Page, Effects of carbonation on pore structure and diffusional [114] T.C. Powers, A hypothesis on carbonation shrinkage, J. PCA Res. Dev. Lab. 4
properties of hydrated cement pastes, Cem. Concr. Res. 27 (1997) 995–1007. (1962) 40–50.
[96] P.A. Claisse, H. El-Sayad, I.G. Shaaban, Permeability and pore volume of [115] J. Xiao, J. Li, B. Zhu, Z. Fan, Experimental study on strength and ductility of
carbonated concrete, ACI Mater. J. 96 (1999) 378–381. carbonated concrete elements, Constr. Build. Mater. 16 (2002) 187–192, https://
[97] A. Morandeau, M. Thiéry, P. Dangla, Investigation of the carbonation mechanism doi.org/10.1016/S0950-0618(01)00034-4.
of CH and C-S-H in terms of kinetics, microstructure changes and moisture [116] J.J. Chang, W. Yeih, R. Huang, J.M. Chi, Mechanical properties of carbonated
properties, Cem. Concr. Res. 56 (2014) 153–170, https://doi.org/10.1016/j. concrete, J. Chin. Inst. Eng. 26 (2003) 513–522.
cemconres.2013.11.015. [117] J. Jerga, Physico-mechanical properties of carbonated concrete, Constr. Build.
[98] A.J. Katz, A.H. Thompson, Quantitative prediction of permeability in porous rock, Mater. 18 (2004) 645–652, https://doi.org/10.1016/j.conbuildmat.2004.04.029.
Phys. Rev. B 34 (1986) 8179–8181. [118] C.F. Chang, J.W. Chen, Strength and elastic modulus of carbonated concrete, ACI
[99] A.S. El-Dieb, R.D. Hooton, Evaluation of the Katz-Thompson model for estimating Mater. J. 102 (2005) 315–321, https://doi.org/10.14359/14710.
the water permeability of cement-based materials from mercury intrusion [119] C. Carde, R. François, Effect of the leaching of calcium hydroxide from cement
porosimetry data, Cem. Concr. Res. 24 (1994) 443–455. paste on mechanical and physical properties, Cem. Concr. Res. 27 (1997)
[100] V. Dutzer, W. Dridi, S. Poyet, P. Le Bescop, X. Bourbon, The link between gas 539–550.
diffusion and carbonation in hardened cement pastes, Cem. Concr. Res. 123 [120] C. Le Bellego, B. Gérard, G. Pijaudier-Cabot, Chemo-mechanical effects in mortar
(2019) 105795, https://doi.org/10.1016/j.cemconres.2019.105795. beams subjected to water hydrolysis, J. Eng. Mech. 126 (2000) 266–272.
[101] P.H.R.R. Borges, N.B. Milestone, J.O. Costa, C.J. Lynsdale, T.H. Panzera, A. [121] U. Schneider, S.W. Chen, Deterioration of high-performance concrete subjected to
L. Christophoro, Carbonation durability of blended cement pastes used for waste attack by the combination of ammonium nitrate solution and flexure stress, Cem.
encapsulation, Mater. Struct. 45 (2012) 663–678, https://doi.org/10.1617/ Concr. Res. 35 (2005) 1705–1713, https://doi.org/10.1016/j.
s11527-011-9788-8. cemconres.2004.11.011.
[102] A. Leemann, F. Moro, Carbonation of concrete: the role of CO2 concentration, [122] V.H. Nguyen, H. Colina, J.M. Torrenti, C. Boulay, B. Nedjar, Chemo-mechanical
relative humidity and CO2 buffer capacity, Mater. Struct. 50 (2017) 30, https:// coupling behaviour of leached concrete. Part I: experimental results, Nucl. Eng.
doi.org/10.1617/s11527-016-0917-2. Des. 237 (2007) 2083–2089, https://doi.org/10.1016/j.nucengdes.2007.02.013.
[103] J.L. Branch, D.S. Kosson, A.C. Garrabrants, P.J. He, The impact of carbonation on [123] I. Yurtdas, S.Y. Xie, N. Burlion, J.F. Shao, J. Saint-Marc, A. Garnier, Influence of
the microstructure and solubility of major constituents in microconcrete materials chemical degradation on mechanical behavior of a petroleum cement paste, Cem.
with varying alkalinities due to fly ash replacement of ordinary Portland cement, Concr. Res. 41 (2011) 412–421, https://doi.org/10.1016/j.
cemconres.2011.01.008.

17

You might also like