See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/367750748

Microstructure and Moisture Transport in Carbonated Cement-Based

Materials Incorporating Cellulose Nanofibrils

Article in SSRN Electronic Journal · January 2022

DOI: 10.2139/ssrn.4040765

CITATIONS READS

3 58

2 authors:

Zhidong Zhang Ueli Angst

ETH Zurich ETH Zurich
60 PUBLICATIONS 863 CITATIONS 158 PUBLICATIONS 4,787 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Zhidong Zhang on 13 April 2023.

The user has requested enhancement of the downloaded file.

 Cement and Concrete Research 162 (2022) 106990

Contents lists available at ScienceDirect

Cement and Concrete Research

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

Microstructure and moisture transport in carbonated cement-based

materials incorporating cellulose nanofibrils
Zhidong Zhang *, Ueli Angst
Institute for Building Materials, ETH Zurich, 8093 Zurich, Switzerland

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

Keywords: Carbonation of cement-based materials may lead to durability problems, so that it is necessary to find a way to
Carbonation reduce the effect of carbonation on concrete durability. This study investigated the effects of cellulose nanofibrils
Permeability (CNFs) on the durability of carbonated cement-based materials. Two dosages of CNFs were used to prepare
Water absorption
cement pastes and mortars. Before and after accelerated carbonation (4 % CO2 concentration and 57 % relative
Cement hydration
Durability
humidity), their microstructure, mineralogical composition, moisture retention capacity, drying kinetics, and
water absorption were measured. Results show that the contents of hydration products (CH, C-S-H, and ettrin­
gite) slightly decrease with the increase of CNFs dosage. After carbonation, the carbonation ratios of hydration
products decrease with CNFs dosage so that more hydration products remain in the carbonated materials. The
calcite content was found to increase with the dosage of CNFs which may help the transformation of other
metastable calcium carbonates to calcite. The coarsening effect of carbonation on pore structure was clearly
observed in measured pore size distribution for all materials, while our results show that this effect is weakened
by CNFs. Moisture transport is clearly accelerated by carbonation, but the acceleration rate is diminished with
the increasing dosage of CNFs, suggesting that CNFs are able to reduce the microstructural damage by
carbonation.

1. Introduction decrease of 26 % and 33 % in the total porosity, measured by the water

saturation method [7]. However, for HCP made with blended binders
During the service life of concrete structures, carbonation of concrete (Portland cement, fly ash, slag, metakaolin binders) and various water-
commonly happens. In this process, CO2 reacts with hydration products, to-binder ratios, carbonated in a chamber with 1 % CO2, the measured
mainly calcium hydroxide (CH) but also calcium silicate hydrate (C-S-H) total water porosity was 6 % to 12 % lower than that before carbonation
and ettringite, depending on the carbonation stage. With the decrease of [2]. Concretes made of composite cement with slag and limestone
CH content, pH in the carbonated concretes decreases as well, which showed a reduction of 5 % to 12 % of the water-accessible porosity after
may thus destroy the passive layer on the steel surface and initiate steel the complete carbonation with 100 % CO2 concentration [6]. A study
corrosion. Meanwhile, the reactions of CO2 with hydration products compared HCP made of CEM I, CEM V/A, and low-pH mix and reported
produce new phases, resulting in the precipitation of carbonation the porosity reduction was directly related to clinker substitution by
products (mainly calcium carbonate) in the pore network. The increase pozzolanic additions; that is, the higher content of portlandite, the
of the volume of new solids then reduces the total pore volume in the higher decrease of porosity [3]. The drop of porosity is commonly
material. Measured porosity of carbonated materials has shown the explained by pore clogging, implying that the precipitation of calcium
reduction of total porosity for the Ordinary Portland cement-based carbonate is able to block some pores and of course alter the pore size
(OPC) materials [1–8]. Based on the literature data, the porosity distribution (PSD).
reduction depends on many factors, such as aggregate content, clinker The alteration of the pore structure may depend on the location
content, and carbonation condition. For hardened Portland cement where the calcium carbonate precipitates. Carbonation is generally
pastes (HCP) with water-to-cement (w/c) ratios of 0.4 and 0.5 and performed at relative humidity (RH) between 55 % and 76 %. At equi­
carbonated in an environment with 80 % to 90 % CO2, it was reported a librium state, these RHs correspond to water-filled pores with diameters

* Corresponding author.
E-mail address: [email protected] (Z. Zhang).

https://doi.org/10.1016/j.cemconres.2022.106990
Received 13 February 2022; Received in revised form 7 September 2022; Accepted 20 September 2022
Available online 24 September 2022
0008-8846/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

between 3.5 and 7.4 nm according to the Kelvin equation. The precip­ showed a clear improvement of both the flexural modulus and strength,
itation of calcium carbonate can only happen in pores smaller than these because the nanofibrils have higher intrinsic strength and stiffness [24].
sizes, which thus reduces the volume of these small pores and causes the Therefore, an excessively strong bonding is expected to form between
main peak of the PSD to shift towards the large pore size, resulting in the the cellulose fibers and the matrix because of the presence of free hy­
increase of the volumetric ratio of large pores as commonly reported in droxyl groups, which can be chemically bound with hydration products
the literature [1,2,5,9,10]. With prolonged exposure to CO2, the coars­ [25]. Furthermore, CNFs have the potential to slow down water trans­
ening effect can be further enhanced as the increase of solid volume may port in cementitious materials. It was reported that fiber reinforcement
create pressure to small pores and finally cause microcracks [11]. reduces the permeability of concrete with 0.1, 0.3, and 0.5 vol% of
Meanwhile, the prolonged exposure leads to carbonation of C-S-H gels cellulose fibers [24].
which decompose to silica gel [9] so more large pores are formed. The use of (nano)celluloses may have certain benefits in improving
As a result of the changes of porosity and PSD, the moisture retention the durability of cement-based materials. However, it is still not clear
and transport properties of the carbonated cementitious materials how nanocellulose can affect carbonation of cementitious materials. In
become different from the non‑carbonated (NC) ones. Carbonation leads the present study, cement pastes and mortars with two dosages of NCFs
to a drop of water content over all the RH range because of the decrease were prepared and their microstructural and moisture transport prop­
of porosity [2,3,5,7]. Despite porosity clogging, carbonation induces an erties, as two important indicators of durability of porous materials,
increase in the oxygen diffusion coefficient for most of the tested with and without carbonation were investigated. Liquid water perme­
hardened cement pastes with different binders at all relative humidity ability was also inversely determined by a moisture transport model
(RH) levels [2]. Carbonation of OPC pastes has an effect on their oxygen based on the results of drying tests and water absorption tests.
diffusion coefficient by an increase of one order of magnitude, and a
more significant effect was found for the blended pastes [9]. The in­ 2. Experiments
crease of oxygen permeability for supercritically carbonated specimens
was reported in the literature [12]. The water permeability determined The aim of this study is to characterize the microstructure and
by the cup test showed a decrease in the permeability of OPC pastes measure moisture transport properties of carbonated and non‑carbo­
carbonated at 3 % CO2, while a significant increase of permeability was nated cement-based materials incorporating CNFs. There is no doubt
found for the blended cements carbonated at the same condition [3]. that the microstructure of a material is closely related to its durability.
Possibly, the occurrence of microcracks is conducive to transport of Recent studies showed that the moisture condition inside carbonated
liquids and gasses through carbonated cementitious materials. For concrete is the controlling factor to the corrosion of steel rebar [26,27].
blended cement paste specimens (with 75–90 % Blast Furnace Slag), Therefore, various measurement methods and techniques were chosen
extensive cracking occurred as a consequence of carbonation, which to obtain the necessary information of microstructure and moisture
significantly increases the oxygen permeability [13]. Nevertheless, it transport properties. Thermogravimetric analysis (TGA) was used to
was also found that water permeability decreased by a factor of six and understand the carbonation effects on chemical composition. Nitrogen
three after carbonation of OPC and blended cement paste with limestone adsorption (NAD) and scanning electron microscopy (SEM) can provide
powders, respectively [14]. Therefore, the influence of carbonation on quantitative and visual information of the microstructure. Water vapor
the gas and liquid water permeabilities still needs more research work to sorption isotherms (WVSIs) indicate the capacity of a material retaining
have a clear overview [5]. moisture at given relative humidities. Drying and water absorption
In recent decades, studies found that adding nanofibers into concrete measurements directly reflect how fast moisture can move inside ma­
can improve mechanical properties, such as compressive strength and terials and are used to determine moisture transport coefficients.
ductility [15]. Nanocelluloses, including cellulose nanofibrils (CNFs)
and cellulose nanocrystals (CNCs), have now become a good alternative 2.1. Materials
reinforcing agent for concrete due to its remarkable mechanical prop­
erties, as well as owing to their comparable low price, hydrophilic na­ The cellulose nanofibrils (CNFs), also known as nanocellulose fibers
ture, low health risk, environmental friendliness, and biodegradability (NCFs) and nanofibrillated cellulose (NFC), were supplied by the Cel­
[16]. CNFs show the internal curing capability [17], which thus is able lulose and Wood Materials Laboratory at EMPA, Switzerland. The CNFs
to enhance cement hydration as shown by isothermal calorimetry and suspension was prepared from Eucalyptus pulp by chopping, swelling in
thermogravimetric analysis [18]. Nevertheless, measured chemical water, and mechanical grinding [28]. The width of CNFs is from nm to
shrinkage curves revealed that the main hydration peak was delayed by μm and the length is generally a few hundred μm (see Fig. 1 for SEM
1–2 h for cement paste with 0.5 w/c and 0.75 wt% CNFs of cement mass, images of CNFs). The nitrogen specific surface area, determined by the
while the total chemical shrinkage showed no difference after one day BET model, of the supercritically dried CNFs, is around 200 m2/g [28].
[19]. The delay effect may be caused by absorption of CNFs on the Two CNFs dosages were chosen in this study, 0.3 wt% and 0.8 wt% of
surface of cement grains (thus slowing down cement dissolution) and cement mass, which are in the range of commonly studied CNFs dosages
cations (e.g., K+, Na+) bound with the hydroxyl group of CNFs which in the literature [29]. The CNFs slurry was diluted in water to reach the
can reduce the reaction activity of cement hydration. As a consequence designed CNFs contents and then mixed with a high-speed mixer for 3
of hydration alteration, CNFs were reported to have the ability to min. The mixture was stored in a bottle and stirred by a magnetic stirrer
improve the microstructure of cement hydrates. Data measured by until the preparation of cement-based materials [19]. Cement paste and
mercury intrusion porosimetry (MIP) showed that the mean pore size of mortar specimens were prepared with CEM II/B-LL cement with <75 %
cement paste decreases with the addition of CNFs [18]. By adding a Portland cement clinker and about 25 % limestone, from Vigier Ciment
certain amount of cellulose fibers (e.g., 1–1.5 wt%), the measured (see Table 1 for the chemical composition of cement). They both have a
autogenous shrinkage and cracking induced by drying shrinkage were water-to-cement ratio of 0.6. For mortar, the sand-to-cement ratio was 3
significantly reduced [20,21]. The yield stress measured by a nano­ with the maximum river sand size of 2 mm. The dry materials (cement or
rheometer for cement pasted with 0.3 vol% of CNCs is at the same level cement and sand) were mixed in a mixer with a low speed (approxi­
as the plain cement. With a further increase of CNCs content, the yield mately 140 RPM) for 30 s and then the liquid mixture with CNFs was
stress increased significantly and surpassed the plain cement paste [22]. slowly added while the mixer continued with the low speed for 60 s.
The cement mortar reinforced by CNFs shows a similar toughness to that After cleaning the mixing bowl to avoid dry materials settling at the
with fine sand, but significantly greater flexural modulus and flexural bottom, the mixing continued for 60 s at a high speed (about 580 RPM)
resistance than those prepared with conventional pulps [23]. In addi­ [30]. After mixing, materials were cast in plastic bottles (33 mm
tion, results of three-point bending tests for mortar with 3.3 wt% of CNFs diameter for mortar and 31 mm for paste), which were slowly rotated for

2
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

Fig. 1. SEM images at different magnification of dry CNFs, after mechanical fibrillation using an ultra-fine friction grinder [28].

particles for the NAD measurements, were heated to 105 ◦ C at 10 ◦ C/min

Table 1
from room temperature and held at 105 ◦ C for 30 min to remove all
Chemical composition of CEM II/B-LL cement (based on
evaporable water, and then temperature increased to 1000 ◦ C at a rate of
European Standard EN 197-1).
5 ◦ C/min. Experimental details can be found elsewhere [19].
Composition Weight ratio %

CaO 60.78 2.3. Nitrogen adsorption (NAD)

SiO2 17.82
Al2O3 3.29
Fe2O3 2.55
The NAD measurements were conducted using a Micromeritics
MgO 1.61 TriStar II 3020 apparatus. About 1 g sieved samples were exchanged
SO3 2.52 with isopropanol for about 12 h and then dried at 40 ◦ C in a tube purged
K2O 0.28 with nitrogen gas. Then, about 0.5 g samples were degassed at 5 μm Hg
Na2O 0.16
pressure at 40 ◦ C for 12 h. The measurement started from relative
LOI 1.4
pressure p/p0 = 0.01 and increased to p/p0 = 0.99 to measure the
adsorption curve and then decreased to p/p0 = 0.01 step wisely to
one night to avoid solids separation and then sealed for curing. obtain the desorption curve.
At the age of six months, plastic bottles were removed and cylinders
(mortars and pastes) were cut into 2 cm long short cylinders. These short 2.4. Scanning electron microscopy (SEM)
cylinders were preconditioned in a chamber with 57 % RH for about 2
weeks. Then, they were divided into two groups. One group was moved A few large crushed pieces were impregnated by a low-viscosity
into a carbonation chamber with 57 % RH and CO2 concentration of 4 %. epoxy resin in a vacuum chamber. After about 24 h, when the epoxy
Several specimens without CNFs (mortars and pastes) were broken to resin had hardened, the embedded specimen was subjected to grinding
check the carbonation process by spraying phenolphthalein on the split and polishing, following the protocol reported in [30]. After grinding
surface. When there was no color change on the surface, the carbonation and polishing, the specimen was imaged as soon as possible to avoid
process was stopped. Meanwhile, specimens in the other group were still damage during the storage. The specimens were coated with a thin layer
kept in the drying chamber with 57 % RH but without additional CO2 of carbon (<10 nm thick). A FEI Quanta 600 environmental SEM with a
supply. When carbonation procedure was completed, all cylindrical BSE detector was used to image the polished surface under the high
specimens, including carbonated and non‑carbonated, were vacuum vacuum mode.
saturated with water and then kept in a 97 % RH desiccator until mass
equilibrium. Cylinders were subjected to drying kinetics measurements 2.5. Water vapor sorption isotherms (WVSIs)
and then water absorption tests.
When preparing short cylindrical specimens, a number of 2 mm thick Water vapor sorption isotherms were measured by a Dynamic Vapor
paste disks were also cut from the cylindrical materials. They were Sorption analyzer (DVS Advantage ET85, Surface Measurement Systems
carbonated in the same chamber as the cylindrical specimens and Ltd.). The DVS only needs very small amount of samples (generally
phenolphthalein was used to monitor the carbonation depth. For both <100 mg) which takes a much shorter time to reach the equilibrium
carbonated and non‑carbonated paste materials, disks for each type state therefore, only paste samples were used in this study. The saturated
were crushed and sieved to select particles in range of 0.6–1.2 mm, paste disk was crushed and a small piece was put in the DVS analyzer.
which were then used for TGA and NAD measurements. This size is the The controlled RH in the DVS started at 97 % and then decreased
generally used in the literature [31–34] as it was already proved as the stepwise to 11 % to obtain the desorption curve. The piece was trans­
optimal size for the NAD measurements [35]. ferred to a tube that was constantly flushed with nitrogen gas for one
week, to achieve the dry state. It was then put back to the DVA analyzer
2.2. Thermogravimetric analysis (TGA) and RH started at 11 % and increased to 97 % to obtain the adsorption
curve. The equilibrium criterion at each RH level was defined when a
The TGA tests were performed with a PerkinElmer® instrument in an mass change (dm/dt) was <0.0005 %/min over 10 min or a maximal
environment purged by ultra-pure nitrogen gas. About 30 mg samples, time of 1000 min for each step was reached. More information about

3
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

DVS measurements can be found elsewhere [36].

Even though the stop criterion is strict, the measured mass curves
still show a very slow increase at the end of measurements; thus, the
“real” final mass is expected to be different from the measured mass. To
minimize the effect, a moisture transport model (i.e., Fick's diffusion
equation) can be used to fit the measured mass change curve and the
“real” final mass (at equilibrium) can be estimated [37]. However, the
geometry of crushed specimens in the DVS analyzer is generally irreg­
ular, so that a measured mass change curve cannot be simulated by the
traditional moisture transport models which require the known geom­
etry. This problem can be overcome by using some empirical equations,
such as the Weibull function or the dual-Weibull function, which do not
depend on the geometry of specimens [36]. In this study, the dual-
Weibull function was used to correct the measured WVSIs.

2.6. Drying kinetics

After preconditioning at 97 % RH, the lateral surfaces of the cylin­

drical specimens were sealed with adhesive aluminum sheets. Two ends
of the cylindrical specimen were uncovered to allow moisture exchange
with the ambient environment. These cylinders were stored in desicca­
tors with different RHs, namely, 44 %, 53 %, 75 %, and 85 %. The masses
of these specimens were measured periodically to obtain the drying ki­
netics (mass change vs. time).

2.7. Water absorption

Drying kinetics measurements stopped at 128 days and then cylin­

drical specimens were subjected to water absorption tests. One end of a
cylindrical specimen was in contact with liquid water and the other end
was loosely covered with a plastic film to minimize moisture exchange.
The mass of a specimen was periodically measured for about 250 h.
During the measurements, each time when the specimen was taken out,
water on the surface was wiped away by a damping cloth. After the
water absorption tests, specimens were dried in an oven of 65 ◦ C until
the mass equilibrium. With knowing the measured volume of a specimen
and saturated mass before drying kinetics measurements, we can
calculate the total porosity (accessible-to-water porosity).

2.8. Naming rules

Fig. 2. TG (a) and DTG (b) curves of different paste materials heated from 105
In the experiments, different materials with different treatment and ◦
C to 1000 ◦ C.
CNFs dosages were studied at different RHs. To distinguish them, the
abbreviations are summarized in Table 2 which are used in the following carbonated or NC materials, the final mass change slightly changes with
sections of this paper, in particular for figures. the dosage of CNFs. The decomposition of CNFs may contribute to this
tendency. According to the TGA test on the pure CNFs [19], CNFs
3. Results thermally decompose at around 360 ◦ C, which is shown as a small peak
at this temperature in Fig. 2b. In addition to this peak, a few large peaks
3.1. Mineralogical change after carbonation can be identified in Fig. 2b. The first peak is between 105 and 200 ◦ C,
due to the emission of water from the decomposition of C-S-H and a
Cement hydration and carbonation were investigated by the means small amount of ettringite [38]. It is clear that the carbonated materials
of TGA results, which were obtained on paste materials that did not have slightly smaller peaks than the non‑carbonated materials. This may
contain aggregates. Results in Fig. 2a show that carbonated specimens result from the low contents of C-S-H and/or ettringite that were
have much higher mass losses than non‑carbonated materials, which is partially carbonated during the process of carbonation. Upon carbon­
caused by the thermal decomposition of carbonate phases. No matter for ation, ettringite generally decomposes to gypsum, calcium carbonate,
alumina gel, and water [39]. The main product - gypsum loses its two
Table 2 water molecules at temperature below 100 ◦ C and becomes anhydrite II
Abbreviations. & III [40] so that this peak cannot be seen in the presented figure (which
Abb. Meaning Abb. Meaning Abb. Meaning only shows curves above 105 ◦ C). The thermal decomposition of
P Paste CNF3 0.3 wt% CNFs NC Non‑carbonated alumina gel happens at 300 ◦ C [41], while this peak is not obvious in
dosage Fig. 2b, indicating that the amount of decomposed ettringite is
M Mortar CNF8 0.8 wt% CNFs C Carbonated insignificant.
dosage The following peak is found at 440 ◦ C, corresponding to the emission
RHxx Drying at xx
of water from the decomposition of one main hydration product – CH.
RH

4
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

Clearly, the carbonated specimens have a much smaller CH peak

because most CH was carbonated. Peaks in the temperature range higher
than the CH peaks are from the decomposition of calcium carbonates,
which release CO2. The NC materials only have one peak at around 720
◦
C from the decomposition of limestone (calcite). However, the
carbonated materials show two peaks, at 550 and 740 ◦ C. The peak at
the lower temperature results from the decomposition of a metastable
calcium carbonate formed at the early stage of carbonation of calcium
hydroxides [42], which is believed to be vaterite. Depending on the
availability of moisture and temperature, vaterite may transform to
aragonite (with high temperature) and poorly crystallized calcite (when
sufficient moisture supply) [43,44]. The thermodynamic stability of the
polymorphs increases in the order vaterite → aragonite → calcite and
their solubility decreases in the same order [45]. Therefore, the peak at
the higher temperature is from the decomposition of calcite.
With the increase of CNFs dosage, the peak for vaterite becomes
smaller, while the calcite peak becomes higher. Clearly, the presence of
CNFs helps vaterite transform into calcite. Our hypothesis is that this is
probably caused by the presence of CNFs which can in principle reduce
the amount of Ca+ in pore solution.
Based on the above description, a TG curve can be divided into three
sections, the temperature range of the CH peak with two additional
temperature ranges, below the CH peak and above the CH peak, corre­
sponding to the decomposition of C-S-H & ettringite and calcium car­
bonates, respectively. Note that the temperature range of the CH peak
should be identified individually for each DTG curve. By analyzing the
mass losses in different temperature ranges, we can compare the con­
tents of different phases in different materials. Mass losses from the
decomposition of different phases are provided in Fig. 3. In these curves,
the mass ratio of each phase was corrected by removing the CNFs-related
mass loss at different temperature (from the thermal decomposition
curve of CNFs). The presented mass ratio is the ratio of measured mass
loss to the dried mass of material at 1000 ◦ C (already subtracted the
dried mass of CNFs). Note that mass losses were caused by releasing
different gases in the decomposition of different phases, such as water
for C-S-H & ettringite and CH, and CO2 for carbonates. Therefore, values
reported in Fig. 3 are not used to indicate the mass ratios in the actual
materials. Instead, curves only compare the effects of CNFs on different
phases.
Fig. 3a shows that, in the NC materials, both CH and C-S-H &
ettringite contents decrease with the increasing CNFs dosage, indicating
that the degree of cement hydration was reduced by CNFs. In the liter­
ature, the heat evolution measured with an isothermal calorimeter Fig. 3. TGA mass losses from decomposition of different phases: a) non‑car­
showed that CNFs delayed and increased the main hydration peak bonated and b) carbonated cement pastes.
compared with specimens without CNFs [17], while a recent study re­
ported that the measured chemical shrinkage curves for cement pastes
with CNFs do not show an increase in the main hydration peak [19]. αi,C =
mi,NC − mi,C
(1)
These results indicate that CNFs may slightly influence cement hydra­ mi,NC
tion at early ages (<3 days). For the long term, to a certain degree, CNFs
may still work as a retarder to prevent the formation of hydration where i = CH or C-S-H & ettringite and mi, NC and mi, C are measured
products. mass losses for the phase i in NC and carbonated specimens, respectively.
After carbonation, the contents of CH and C-S-H & ettringite decrease Results in Table 3 display that the carbonation ratios for all phases
as they are partially carbonated; meanwhile, the amount of carbonate decrease with the content of CNFs. After carbonation, only about 28.5 %
phases become very high, comparing Fig. 3b with Fig. 3a. The uncar­ CH remains in the neat cement paste, while about 40 % CH still can be
bonated hydration products are higher for the high dosage of CNFs and found in the paste with 0.8 wt% CNFs. For all types of pastes, C-S-H and/
the amount of carbonates in pastes with CNFs is lower than that in the or ettringite are partially carbonated. About 33 % C-S-H & ettringite in
neat cement paste. These results imply that CNFs are able to help hy­ the neat paste are carbonated, while this value is halved for the paste
dration products resist carbonation and preserve slightly more hydra­ with 0.8 wt% CNFs. These results indicate that all hydration products
tion products.
Assuming that the mass reduction of CH and C-S-H & ettringite after
Table 3
carbonation is only caused by the carbonation reactions, the carbonation
Carbonation ratios of CH and C-S-H & ettringite.
ratios of hydration phases αi, C can be determined by the mass difference
of this phase in NC and carbonated materials to its mass in the NC ma­ Materials CH carbonation C-S-H & ettringite carbonation

terial, which is written as PC 0.715 0.328

P CNF3 C 0.658 0.281
P CNF8 C 0.600 0.167

5
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

begin to be carbonated when exposed to CO2 but CH is apparently for the larger pores. The NC paste with 0.8 wt% CNFs always has lower
carbonated faster than other phases as it is generally agreed that CH is PSD curves than the other two NC pastes. The main peaks for the NC
more soluble and much easier to react with CO2 than C-S-H. Moreover, pastes appear at around 2 nm. For the carbonated pastes, the PSD curves
the presence of CNFs seems to inhibit carbonation of these pastes. are much lower than the NC materials. The insert in Fig. 5b shows that
the peaks of carbonated pastes shift to the right to 3–4 nm compared
with the NC ones, resulting from the coarsening effect of carbonation.
3.2. Microstructure The peak increases with the dosage of CNFs, indicating a more signifi­
cant reduction in the volume of small pores in the neat cement paste
The SEM images can provide visual information about the micro­ than other materials.
structure. Fig. 4 compares SEM images of NC and carbonated pastes with The BET surface area was calculated based on data from the
0.8 wt% CNF. The NC specimen on the left-hand side shows a very adsorption curve in the range 0.05 ≤ p/p0 ≤ 0.3. The NC pastes always
heterogeneous microstructure and diverse phases. The needle-like C-S-H have much higher BET surface areas than the carbonated pastes (see
and ettringite/CH with long rod shapes are clearly visible. The wide CNF Fig. 6), indicating that carbonation may induce damages to the micro­
fibers are also seen and few hydration products are on the surface of structure. From the SEM images in Fig. 4, it seems that the carbonates
CNFs. In terms of the carbonated specimen, the wide CNFs are clear, occupy the volume of some pores, which are smaller pores as shown by
while the other regions seem more homogenous than the NC materials. the decrease of micro-pore volume in measured PSDs in Fig. 5.
There no needles or rods of hydration products can be seen, which are It is generally accepted that the higher fraction of small pores leads to
believed to transform into carbonation products (as confirmed by EDX a higher BET surface area. Nevertheless, adding 0.3 wt% CNFs does not
results in Fig. 4b), showing in many bright small clusters. increase BET surface area for NC pastes even though Fig. 5b shows that
To have overall structure information of the cementitious materials, this paste has a higher volume of small pores. This may be because the
NAD data are useful. The nitrogen adsorption curves in Fig. 5a depict lower volume of large pores cancels out the increase of BET area due to
very different curve shapes between carbonated and NC cement pastes. the higher volume of small pores. Consequently, two pastes have very
The NC materials have much higher total absorption volume and more similar BET surface areas. With the further increase of CNFs content to
obvious hysteresis between adsorption and desorption curves. Amongst 0.8 wt%, the measured BET surface decreases by about 15 %. This may
the NC pastes, the neat paste and the 0.3 wt% CNFs paste have both be due to the fact that fewer hydration products in the material if the
higher total absorption volume and larger hysteresis than the 0.8 wt% content of CNFs is high as already shown in Fig. 3a that the amount of
CNFs paste. This implies that the microstructure of these materials is CH and C-S-H & ettringite decreases with the increase of CNFs content.
more heterogeneous than the paste with 0.8 wt% CNFs. In other words, After carbonation, BET surface area slightly increases with the CNFs
the differences in pore size between the large pores and the connected content, which is in agreement with the observation in Fig. 3b and
small pores are higher. On the contrary, the carbonated pastes have flat calculated carbonation ratios in Table 3 that less CH and C-S-H &
sorption curves and low hysteresis, indicating that pores become more ettringite are carbonated for the high content of CNFs. These results
uniform in terms of size and well connected so that the irregularity is indicate that the presence of CNFs is able to reduce the degree of
low. In addition, we see a jump from p/p0 = 0, which corresponds to the carbonation.
volume of the micro-pores. Clearly, the NC pastes have larger micro-
pore volume than the carbonated pastes.
The pore size distributions (PSD) in Fig. 5b were calculated by the 3.3. Water vapor sorption isotherms
Barrett-Joyner-Halenda (BJH) method by assuming cylindrical pores.
This method can provide reliable PSD for pore size between 2 and 50 nm WVSIs are commonly described in a function of the degree of satu­
for cementitious materials [46]. We clearly see that the NC pastes have ration, S, vs. RH as shown in Fig. 7, in which black lines are from NC
higher pore volume for the whole range of measured pore size. The NC pastes and red lines represent results for carbonated materials. Clearly,
paste with 0.3 wt% CNFs shows slightly a higher PSD curve than the NC the results vary with the CNFs dosage, measurement methods, RH level,
neat cement for pores smaller than 15 nm and the tendency is reversed and carbonation status. Considering the NC materials, the decrease rate

Fig. 4. SEM images of non‑carbonated (a) and carbonated (b) pastes with 0.8 wt% CNFs.

6
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

Fig. 6. Nitrogen BET surface area for carbonated and non‑carbonated cement
pastes with different CNFs dosages. Error bars represent the standard de­
viations, which are clearly very small so NAD measurements have a very good
repeatability.

than the hydration products. For the adsorption curves, the tendency is
reversed, different from pastes with low and without CNFs.
Fig. 7 clearly displays that carbonation can markedly reduce the
moisture retention capacity. This effect decreases with the increasing
content of CNFs. The carbonation effect on moisture retention capacity
can be quantified by a water content ratio [42],
wC
Rw = 1 − (2)
wNC

where wC and wNC are measured water contents of carbonated and NC

materials at a given RH. Rw shows the relative change of the moisture
retention capacity due to carbonation. Results in Fig. 8 are calculated Rw
based on measured desorption isotherms. The general trend is that the
values of Rw at the moderate RHs are higher than those at low and high
RHs. The Rw is about 0.6 for the paste without CNFs, meaning that the
moisture retention capacity is reduced by about 60 % compared with the
Fig. 5. Measured nitrogen adsorption curves (a) and pore size distributions NC paste. With the increase of CNFs content, the curve shifts down,
calculated by BJH method (b). The enlarged PSDs for carbonated pastes are
indicating that the carbonation effect becomes weaker. The curve shapes
shown in the insert in figure (b).
of pastes without and with 0.3 wt% CNFs are very similar, while for the
paste with 0.8 wt% CNFs, the carbonation effect on moisture retention
of the desorption curve at extremely high RHs (e.g., from 100 % to 83 %) capacity is much lower and the Rw even become negative at low RHs.
becomes higher with the increase of CNFs dosage. It has been reported Therefore, we can say that the higher CNFs content is able to better
that CNFs have a strongly hydrophilic nature, so they can absorb much reduce the carbonation effect on the microstructure so that CNFs can
water when there is enough liquid water [17], while the absorbed water help cementitious materials retain moisture.
can be easily lost when the RH is lower than 100 %. However, with the
decrease of RH, WVSIs with celluloses sharply decrease, resulting in low
3.4. Drying kinetics
water content at the moderate RHs [47]. After carbonation, WVSIs for
the neat pastes and mortars, shift down, which agrees with the common
The measured drying kinetics for mortars and cement pastes with
observations in the literature [3], resulting from the pores coarsening
different dosages of CNFs are shown as symbols in Fig. 9 and Fig. 10,
effect of carbonation. In other words, the carbonated materials are less
respectively. It can be seen that the total amount of mass loss increases
sensitive to the change of RH than NC materials for the moderate RHs
with the drying RH, regardless of carbonated or non‑carbonated mate­
[7]. If a material has more large pores, it cannot hold more water at the
rials. The lower environmental RH is able to remove more water from
moderate and low RHs, meaning low moisture retention capacity. Pastes
the porous media. For a given RH, the measured mass loss is higher for
with 0.3 wt% CNFs show the similar results to the neat cement pastes,
cement pastes than mortars with the same content of CNFs because
but the differences between NC and carbonated materials are less
cement pastes contain higher HCP per unit of volume than mortars
obvious. The further increase of CNFs content completely changes the
which have a much lower ability to retain moisture than HCP.
relation between NC and carbonated materials as shown in Fig. 7c. For
After carbonation, the drying mass loss at a given RH is much lower
the desorption curves, the NC materials are still higher at high RHs,
than the NC specimens dried at the same RH. As aforementioned, the
while at low RHs, the carbonated materials are able to hold slightly more
moisture retention capacity of carbonated materials is lower at the
moisture than the NC ones, which is most likely caused by the fact that
moderate RH because of the increase of volume fraction of large pores.
the moisture retention capacity of CNFs is less sensitive to carbonation
During preconditioning, in which the saturated specimens were dried at

7
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

Fig. 8. Carbonation effect on moisture retention capacity of cement pastes with

different contents of CNFs.

97 % RH, more water has been removed from the carbonated materials.
Therefore, in the subsequent drying, less water remains in the materials
for dying.
With the change of the dosage of CNFs, the mass loss at different RH
conditions varies. In general, the moisture loss slightly decreases at the
CNFs content of 0.3 wt%, while it slightly increases at 0.8 wt% CNFs.
This trend is less pronounced for carbonated materials than the NC ones.
The reason behind this is that the PSDs are altered by the additional
CNFs, which further affects the moisture retention capacity of these
materials. As shown in Fig. 5, the NC materials with 0.3 wt% CNFs have
the highest measured pore volume (for pores smaller than 50 nm) and
the NC materials with 0.8 wt% CNFs have the lowest measured pore
volume. This would help to understand the drying mass loss curves in
Fig. 9 and Fig. 10. For the carbonated materials, the nitrogen adsorption
curves are very similar as shown in Fig. 5, so there is no significant
difference in the drying mass loss in Fig. 9 and Fig. 10.

3.5. Water absorption

After the drying kinetics measurements, water absorption tests were

performed on the same specimens. Water penetration into the porous
materials is driven by capillary pressure. The measured water absorption
data are shown in Fig. 11 and Fig. 12 for mortars and cement pastes,
respectively. In these figures, the x-axis is shown in the square root of
time, because traditionally the amount of cumulative absorbed water in
the first stage (in the short term) linearly increases with the square root
of time (t1/2) for most porous media [48,49]. The slope of the linear
region represents the ability of a material to absorb and transmit water
by capillarity. Therefore, the amount of cumulative absorbed water I can
be written as a linear relation with t1/2.

I = Sp t1/2 + A (3)

where Sp is the slope of the linear curve, called sorptivity, and A is a

parameter associated with the end effect (such as buoyancy, lateral in­
vasion) [50]. By fitting the linear region of water absorption curves, the
sorptivity of studied materials is determined and provided in Table 5 and
Table 6 for mortars and cement pastes, respectively.
Fig. 11 and Fig. 12 show that most measured water absorption curves
have a linear relation with t1/2, except for some cement pastes. The
Fig. 7. Water vapor sorption isotherms (S vs. RH) for materials without CNFs linear regions end at about 9 h for the fast absorption (higher Sp) and
(a), with 0.3 wt% CNFs (b), and with 0.8 wt% CNFs. about 64 h for the slow absorption (lower Sp). For mortars, the
carbonated ones and these NC mortars previously dried at low RHs (e.g.,
44 % and 53 %) have steeper slopes (or higher Sp) than the other

8
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

Fig. 9. Drying kinetics for mortars with different dosages of CNFs: a) without
CNFs, b) with 0.3 wt% CNFs, and c) with 0.8 wt% CNFs. Symbols are measured Fig. 10. Drying kinetics for cement pastes with different dosages of CNFs: a)
data and lines are simulated results. without CNFs, b) with 0.3 wt% CNFs, and c) with 0.8 wt% CNFs. Symbols are
measured data and lines are simulated results.

9
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

Fig. 11. Water absorption of mortars with different dosages of CNF. Symbols
are measured data and lines are simulated results. Fig. 12. Water absorption of cement pastes with different dosages of CNF.
Symbols are measured data and lines are simulated results.

10
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

materials. The sorptivity Sp is related to two main factors, the fraction of of a specimen and the further increase of the curves only results from the
capillary pores and the capillary pore size distribution. The former de­ secondary water absorption.
termines the final amount of absorbed water and the latter controls the The final amount of absorbed water depends on the total available
speed of absorption. If a porous material has more large capillary pores, capillary pores in the mortar specimen, which clearly decreases with the
it has a higher Sp. Specimens that were previously dried at low RHs have increase of previous drying RHs. According to the Kelvin equation, 44 %,
been removed more water (see Fig. 9) so that they have more large 53 %, 75 %, and 85 % RH can remove water from pores with diameters
capillary pore space available for water absorption. On the contrary, larger than 2.6, 3.3, 7.4, and 13.2 nm, respectively. The cumulative
specimens dried at high RHs (e.g., 75 % and 85 %) show slower water absorbed water at the end in Fig. 11 shows that the NC specimens with
absorption because they have less large capillary pore volume for water 0.3 wt% CNFs have more pores larger than 2.6 and 3.3 nm than mortars
penetration. After carbonation, the differences in water absorption without CNFs and with 0.8 wt% CNFs, agreeing with PSDs shown in
curves between different RHs are smaller than the NC ones. This is Fig. 5. However, the NC specimens with 0.3 wt% CNFs have fewer pores
because that the desorption WVSIs of carbonated materials are flatter larger than 7.4 and 13.2 nm than the other NC mortars. After carbon­
than the NC ones as shown in Fig. 7. Therefore, the water retention ation, the differences in different pore sizes become smaller than for the
capacity at the studied RHs is not significantly different. NC materials. The carbonated specimens with 0.3 wt% CNFs have a
After the first stage of water absorption, curves do not show a clear slightly higher amount of pores larger than 2.6 and 13.2 nm than the
transition region and rapidly enter another almost linear region, in other carbonated materials.
which water absorption does not change too much. The two linear stages Regarding water absorption results for cement pastes, the NC spec­
of water absorption for cementitious materials have been reported in the imens without CNFs show very gentle water absorption curves, while
literature and the second linear stage is considered as the secondary after carbonation, the water absorption curves sharply increase once
water absorption [51,52]. The slow increase of the second linear region they contact liquid water. Their water absorption are much faster and
is presumably caused by microstructural alteration [51,53,54], gas higher than other carbonated and NC specimens. Checking the specimen
dissipation [52], and water redistribution [55]. Different from most surfaces after water absorption experiments (see images in Fig. 13), we
studies in the literature, the specimens used in the present study are only found that cracks have been formed in some carbonated specimens,
2 cm high, so at the end of the first stage, water has reached the top side which are most likely caused by drying. These cracks are able to

P C RH44 P C RH75

P CNF3 NC RH44 P CNF3 C RH44

P CNF8 NC RH44 P CNF8 C RH53

Fig. 13. Images of surfaces of paste samples after water absorption measurements. The damaged specimens in the first row correspond to the sharply increasing
curves in Fig. 12a.

11
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

significantly accelerate water uptake. Such damages were not so obvious

for other specimens with CNFs, presumably because CNFs act as bridges
to increase the bonds between hydration products and thus avoid
cracking [20,21]. However, the carbonated cement pastes clearly show
faster water absorption than the NC ones, while this tendency gradually
disappears with the increasing dosage of CNFs. For cement pastes with
0.8 wt% CNFs, the effect of drying RH surpasses the effect of carbonation
as shown in Fig. 12c that water absorption of the NC paste dried at 44 %
RH is the fastest one.
The effects of CNFs, drying RH, and carbonation on water absorption
can be quantified by Sp reported in Table 5 and Table 6. Either for
mortars or cement pastes, if they are not carbonated, the calculated Sp
decreases with the drying RH, while after carbonation, this tendency is
not so clear anymore. Before carbonation, the average Sp increases with
the dosage of CNFs, meaning that the presence of CNFs leads to more
large capillary pores in the materials. However, after carbonation, the
average Sp decreases with the dosage of CNFs, which implies that CNFs
are able to reduce the effect of carbonation on the volume of capillary
pores. By calculating the ratio of average Sp before and after carbon­
ation, we have 2.4 for mortars without CNFs, 2.0 for mortars with 0.3 wt
% CNFs, and 1.7 for mortars with 0.8 wt% CNFs. These ratios are 18.7,
2.0 and, 1.3 for cement pastes with different dosages of CNFs. These data
further suggest that the impact of carbonation on the microstructure is
less significant for the materials with CNFs than the neat cement pastes
and mortars.

3.6. Porosity and density

The saturated mass of a specimen was measured prior to the drying

kinetic measurement and the dried mass at 65 ◦ C was measured after
water absorption measurement. With knowing the volume of the spec­
imen, assuming no change of the volume, the saturated density ρsat, the
dry density ρdry, and total porosity ϕ can be determined. Results for
mortars and cement pastes are compared in Fig. 14. As generally known,
mortars have higher ρsat and ρdry and lower ϕ than cement pastes
because mortars contain sand. For the neat cement paste without
carbonation, its total porosity is two times higher than the neat mortar,
well agreeing with the calculation result based on the mix design.
Fig. 14. Porosity and density of cylindrical specimens: (a) cement pastes and
However, with additional CNFs in the cement pastes and mortars, this (b) mortars.
relation is not valid anymore.
For cement pastes, with the increase of CNFs dosage, ρsat and ρdry
Presumably, evaporable water in dried specimens is completely
linearly decrease and ϕ increases. A clear reason is that CNFs have a
removed, so that the increase of ρdry after carbonation is purely from
lower density than cement, but this effect is much smaller than the
CO2 that is captured by reaction with hydration products. Hence, the
observed differences in Fig. 14. For instance, the calculated differences
amount of CO2 absorbed by materials can be determined by calculating
between neat cement pastes and those with 0.8 wt% CNFs according to
the differences in ρdry before and after carbonation and multiplied by a
the mix design are 0.006 g/l for ρsat and 0.008 for ϕ, while Fig. 14 shows
factor of 44/26 as the carbonation process generates water that is
differences of 137 g/l and 0.064, respectively. Another possible reason is
removed by drying when measuring ρdry. The calculated results are
that CNFs delay hydration [17,19], which is in favor of the formation of
provided in Table 4 (mean values of four specimens). For cement pastes,
more high-density C-S-H as confirmed by the results of nanoindentation
adding 0.8 wt% CNFs is able to slightly decrease the amount of CO2. In
[25]. As a consequence, there are more empty spaces between hydration
general, mortars capture less CO2 than cement pastes. Results show that
products. This is partially confirmed by the average Sp (see Table 6)
the amount of CO2 varies with the CNFs dosage. Specifically, it clearly
which increases with the dosage of CNFs for the NC cement pastes. After
decreases for mortars, while for cement pastes, the amount of absorbed
carbonation, the ρsat and ρdry curves shift up and the ϕ curve moves
CO2 increases for 0.3 wt% CNFs but sharply decreases for 0.8 wt% CNFs.
down because of the increase of precipitated calcium carbonate which
Therefore, the additional CNFs show an ability to reduce carbonation
occupies some pores.
reaction in this study, in particular for mortars. This result agrees with
For mortars, ρsat and ρdry are almost constant for different dosages of
the calculated carbonation ratios in Table 3.
CNFs, while ϕ slightly increases. It is clear that the compaction of hy­
dration products in mortars is higher than these in cement pastes. The
4. Discussion
formed more high-density C-S-H due to the presence of CNFs should
contribute to this difference. After carbonation, however, ρsat, ρdry, and
4.1. Inverse analysis on water permeability
ϕ all slightly decrease from neat cement to 0.3 wt% CNFs and no further
change for the higher CNFs dosage. In addition to the fact that the
The above experimental results indicate that the moisture transport
precipitation of calcium carbonate leads to the increase of specimen
speed is markedly increased by carbonation, by the dosage of CNFs, and
mass and reduction of total pore volume, it may be possible that the
by cracks induced by drying. To quantify their effects on transport co­
density of calcium carbonate formed in neat cement mortar is higher
efficients, the liquid permeability kl can be determined based on
than that in mortars with CNFs.

12
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

measured drying kinetics and water absorption curves. When kl is the main adsorption curve with the increasing RH.
only unknown in a moisture transport model, it can be back calculated The simulated drying mass loss and water absorption curves (solids
by fitting the measured drying kinetics or water absorption curves. This for NC materials and dashes for carbonated ones) are compared with
approach is known as the inverse analysis method for the determination measured data in Fig. 9 and Fig. 10 for drying kinetics and Fig. 11 and
of liquid water permeability [56–58]. Studied reported that inverse Fig. 12 for water absorption. The inversely determined intrinsic
analysis can provide reasonable results compared with other experi­ permeability Kl by using measured drying kinetics and water absorption
mental measurements, such as the Beam Bending method [50]. To data is provided in Table 5 and Table 6 for mortars and cement pastes,
perform the inverse analysis, a reliable moisture transport model is respectively. Figures show that most simulated drying mass loss curves
required. Previous studies used a two-phase model which showed a good can match well the measured data with only adjusting Kl, except for a
ability for simulating unsaturated moisture in different cement-based few cement paste specimens. For water absorption, we only tried to fit
materials [56,57,59,60]. The unsaturated moisture transport in a rigid the first linear region as this region reflects water transport in capillary
porous material is formulated by the Richards' equation, written as pores. The slow moisture absorption in the later region, caused by the
complex microstructure of cement-based materials [51–54,64,65], is not
∂S
= ∇[Da (S)∇S ] (4) included in the present moisture transport model.
∂t
For NC mortars, Kl determined from drying kinetics is in the order of
where Da(S) is an apparent diffusivity, which is a function of S. 10− 21 m2, in the same range with data measured by various experi­
It was reported that a two-phase model is necessary to simulate mental methods in the literature [57,66,67]. Regardless of the carbon­
moisture transport in the whole range of RH [50]. The apparent diffu­ ation conditions, Kl determined from water absorption data is about
sivity Da(S) has contributions from liquid water and water vapor, 5–500 times (50 on average) higher than that from drying kinetics. It has
been reported that the sorptivity method provides permeability about
Da (S) = Dl (S) + Dv (S) (5) two orders of magnitudes higher than these determined by the Beam
Bending method and inverse analysis [50]. The main reason is that
with
specimens used for water absorption must be pre-dried in an environ­
Kl dPc ment with low RHs to remove a certain amount of water. As well known,
Dl (S) = − krl (6)
ϕηl dS any drying methods can induce damage to the microstructure [32],
generally creating more capillary pores. Nevertheless, the sorptivity
( )2
Mv Pvs RH dPc method does not consider the drying damage, so the determined Kl is
Dv (S) = − Dv0 ϕxD (1 − S)xD +2 (7) higher.
ρl RT ϕ dS
The drying effect on the pore structure of mortars is well shown in Kl
where ηl is the dynamic viscosity of liquid water, Mv is the molar mass of data in Table 5. No matter which methods are used to calculate Kl, the
water molecule, ρl is liquid water density, xD is a resistant parameter to clear tendency is that Kl values decrease with the increase of drying RH.
consider the microstructural effect on vapor diffusion (taking as 2.74 for It clearly demonstrates that the degree of drying damage decreases with
cement-based materials [61]), and Pvs is the saturated vapor pressure in the increasing RH. So, it is reasonable to say that the inversely deter­
air. The term dP mined Kl at high RH is much closer to the “real” Kl due to the least
dS represents the moisture retention capacity, which
c

should be calculated from a WVSI. A measured WVSI is fitted by the van damage to the microstructure [50]; that is drying at 85 % RH in this
Genuchten equation [62]. The relative permeability krl is expressed as a study.
function of S by using the van Genuchten - Mualem model [62,63]. After carbonation, Kl values in Table 5 are higher than NC specimens
These equations can be found elsewhere [56,60]. at the same experimental conditions. Only looking at average Kl values
Permeability in this model is expressed as kl = krlKl, in which the or drying at 85 % RH (less drying damage to the microstructure), we see
intrinsic permeability Kl is the only unknown; therefore, Kl can be that Kl slightly increases for NC specimens with the CNFs dosage which
inversely determined by fitting a measured mass change curve of a has the same trend to the total porosity as shown in Fig. 14, but after
specimen subjecting to drying or a water absorption curve. For simu­ carbonation, Kl decreases with CNFs dosage. By calculating the ratio of
lating drying kinetics, a desorption WVSI is used to calculate dP average Kl based on drying kinetics before and after carbonation, we
dS because
c

water content in a specimen decreases from high RH. After drying when
performing water absorption measurements, however, specimens start
at different initial S, which corresponds to 44 %, 53 %, 75 %, or 85 %
RH. From these points to let specimens absorb water, the S - RH relation
does not follow either the main desorption or main adsorption curves.
Instead, because of the existing sorption hysteresis, the S - RH relation
follows a scanning curve of WVSIs between two main sorption curves.
Therefore, a sorption hysteresis model is necessary to correctly predict
the path of the scanning curve. By comparing available sorption hys­
teresis models in the literature, previous studies have shown that the
revised Mualem II model provides the best results well matching
experimental data [60]. An example of scanning curves calculated by
the revised Mualem II model with different starting points is illustrated
in Fig. 15, which shows that the scanning curves gradually approach the

Table 4
The amount of CO2 captured by different materials.
Materials Cement paste (g/g of cement) Mortar (g/g of cement)

Without CNFs 0.372 0.281

0.3 wt% CNFs 0.387 0.199
0.8 wt% CNFs 0.359 0.162
Fig. 15. The scanning curves calculated by the revised Mualem II model.

13
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

Table 5
Inversely determined intrinsic permeability and calculated sorptivity based on measured data on mortars.
Treatment CNFs dosage RH (%) Kl, drying (m2) Sp (g⋅s-1/2⋅m− 2) Kl, water absorption (m2)

Non‑carbonation (NC) Without CNFs 44 3.9E-21 679.6 2.1E-18

53 4.1E-21 261.1 2.8E-19
75 2.6E-21 113.0 6.0E-20
85 1.5E-21 100.1 5.0E-20
Average 3.0E-21 288.4 6.2E-19

0.3 wt% 44 6.2E-21 559.8 7.0E-19

53 7.9E-21 367.5 3.9E-19
75 2.1E-21 257.4 1.5E-19
85 1.6E-21 127.8 5.5E-20
Average 4.5E-21 323.9 3.2E-19

0.8 wt% 44 9.1E-21 616.1 9.4E-19

53 1.2E-20 383.8 4.2E-19
75 9.4E-21 191.9 9.0E-20
85 4.1E-21 166.7 7.0E-20
Average 8.7E-21 339.6 3.8E-19

Carbonation (C) Without CNF 44 5.5E-19 816.6 8.6E-18

53 5.7E-19 549.7 3.7E-18
75 4.2E-19 521.2 3.6E-18
85 2.5E-19 893.5 2.9E-18
Average 4.5E-19 695.3 4.7E-18

0.3 wt% 44 1.4E-19 544.0 2.7E-18

53 1.8E-19 747.6 4.6E-18
75 9.5E-20 595.6 3.5E-18
85 4.4E-20 433.0 1.8E-18
Average 1.1E-19 580.0 3.2E-18

0.8 wt% 44 7.6E-20 563.7 8.6E-19

53 6.2E-20 647.0 1.4E-18
75 2.3E-20 561.7 9.8E-19
85 1.3E-20 518.9 9.1E-19
Average 4.4E-20 572.8 1.0E-18

have 147.9 for mortars without CNFs, 25.5 for mortars with 0.3 wt% diffusion process of dissolved ions, the availability of water and space
CNFs, and 5.0 for mortars with 0.8 wt% CNFs. The ratios become 7.6, [70], cementitious materials with high w/c are expected to have a high
9.8, and 2.7 for water absorption results. This clearly proves that the long-term degree of hydration [71]. Nevertheless, when CNFs are pre­
presence of CNFs is able to reduce the enlarging effect of carbonation on sent, these fibers act as physical barriers and can absorb some ions; thus,
moisture transport. ion diffusion is slowed down. This effect may be more significant for the
For cement pastes, drying kinetics or water absorption data of a few higher w/c cementitious materials. Therefore, less hydration products
specimens cannot be simulated by the moisture transport (indicated by were observed in cementitious materials with CNFs.
“*” in Table 6). The surface inspection shows that these specimens were In terms of the microstructure, PSDs calculated from the nitrogen
damaged during drying as microcracks are clearly seen on some of them. adsorption curves show fewer small pores in cement pastes with CNFs
This may indicate that the use of CNF is able to reduce drying cracking as than the neat cement paste (see Fig. 5b). Together with the fact of the
microcracks were not observed for pastes with high CNFs content. Be­ increasing total porosity with the dosage of CNFs (see Fig. 14), we can
sides these specimens, the other determined Kl values show very similar infer that the presence of CNFs results in more large capillary pores. This
tendencies to the results of mortars. For instance, Kl decreases with the is confirmed by the calculated sorptivity and inversely determined
increase of drying RH. The ratios of average Kl based on drying kinetics permeability for NC materials, as both of them increase with the CNFs
before and after carbonation are 36.8 for pastes with 0.3 wt% CNFs and content (see Table 5 and Table 6). These results do not show promising
16.0 for pastes with 0.8 wt% CNFs. The ratios become 10.9, and 1.8 for benefits of CNFs in improving the durability of NC cement-based ma­
water absorption results. Therefore, the reduction of the microstructural terials. Nevertheless, the use of CNFs is able to reduce shrinkage and
damage induced by carbonation is also observed for cement pastes with cracking, which was reported in the literature [20,21] and also
CNFs. confirmed in this study (see Fig. 13).
However, after carbonation, the above-mentioned tendencies are
4.2. Effects of CNFs on durability of cement-based materials completely reversed. TGA data show that the amount of remaining hy­
dration products in carbonated pastes increases with the dosage of CNFs
As for non‑carbonated cement-based materials, TGA data reveal that (see Fig. 3b) and the carbonation ratios for all hydration products
the formation of hydration products is hindered by the presence of CNFs decrease with the CNFs dosage (see Table 3). The nitrogen adsorption
(see Fig. 3a). Theoretically, nanomaterials have the steric stabilization results show that BET surface area slightly increases with the CNFs
ability to create a uniform distribution of cement particles in the fresh content and the pore coarsening effect of carbonation is weakened by
cement slurry and also enhance nucleation of the hydrated products CNFs (see Fig. 5 and Fig. 6). The measured water vapor sorption iso­
because of the large surface area of CNFs; consequently, the early degree therms further suggest that the carbonation effect on the microstructure
of cement hydration is improved [68]. However, the delay of the main is reduced by CNFs. After carbonation, both the calculated sorptivity and
hydration peak was reported in the previous studies [17,19,69]. The inversely determined permeability decrease with the CNFs content (see
cause to this delay is still not clear yet, but the higher w/c, the more Table 5 and Table 6). The comparison of dry densities before and after
significant delay was found [19]. Considering that the deceleration carbonation implies that the amount of CO2 reacting with cement-based
period of cement hydration is controlled by factors including the materials decreases with the dosage of CNFs (see Table 4). All these

14
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

Table 6
Inversely determined permeability and calculated sorptivity based on measured data on cement pastes.
Treatment CNF dosage RH (%) Kl, drying (m2) Sp (g⋅s-1/2⋅m− 2) Kl, water absorption (m2)

Non‑carbonation (NC) Without CNF 44 7.0E-21* 358.9 7.0E-19*

53 8.0E-21* 317.8 1.2E-18*
75 7.2E-21 119.0 6.0E-20
Average 7.4E-21 265.2 6.5E-19

0.3 wt% 44 3.9E-21* 1467.9 3.8E-18

53 5.2E-21* 1356.9 2.6E-18
75 3.2E-21 307.5 2.1E-19
Average 4.1E-21 1044.1 2.2E-18

0.8 wt% 44 5.2E-21 1799.9 4.5E-18

53 5.0E-21 1412.0 2.9E-18
75 2.0E-21 417.6 2.1E-19
Average 4.1E-21 1209.8 2.5E-18

Carbonation (C) Without CNF 44 1.8E-19 5302.6 3.8E-17*

75 2.5E-19 4628.1 3.8E-17*
Average 2.2E-19 4965.4 3.8E-17

0.3 wt% 44 1.2E-19 2219.2 2.1E-17

53 1.8E-19 1496.3 1.3E-17
75 1.5E-19 2457.2 3.8E-17
Average 1.5E-19 2057.6 2.4E-17

0.8 wt% 44 6.2E-20 1796.8 5.0E-18

53 9.7E-20 1671.8 5.4E-18
75 3.6E-20 1340.6 3.4E-18
Average 6.5E-20 1603.1 4.6E-18

Note: Values with “*” indicate that experimental results cannot be simulated by the moisture transport model because either the measured transport is too fast or the
measured data are less stable.

results clearly show the benefits of CNFs in reducing the effects of CNFs. After carbonation, both water sorptivity and water perme­
carbonation on the durability of the studied cementitious materials. As ability became higher than before, but they decrease with the CNFs
the transport coefficients are lower in carbonated materials with CNFs, dosage. CNFs are able to reduce carbonation effects (coarsening) on
the penetration of moisture and oxygen, two essential factors for steel pore structure, in particular creating fewer large capillary pores than
corrosion, are expected to be much slower than that in carbonated materials without CNFs.
materials without CNFs. Therefore, we expect that the risk of steel
corrosion is lower after carbonation if CNFs are used in cementitious In addition to these benefits, other conclusions drawn in this study
materials. are:

5. Conclusions and outlook − The content of vaterite decreases with the increasing dosage of CNFs
after the accelerated carbonation. Nanofibers seem to help the
Cellulose nanofibrils (CNFs) were used to prepare cement pastes and transformation of vaterite to a more stable polymorph of calcium
mortars and their microstructure, mineralogical composition, moisture carbonates - calcite.
retention capacity, drying kinetics, and water absorption were investi­ − The TGA results show that the content of hydration products (CH, C-
gated in this study. Results indicate that the use of CNFs can reduce S-H, and ettringite) slightly decreases with the increase of CNFs
drying cracks but may not have other promising benefits to non‑car­ dosage. Therefore, the retardation effect (via absorption on cement
bonated materials as CNFs can retard cement hydration, coarsen the grain surface or chemically bound with cations) not only can delay
microstructure, increase moisture transport and porosity. However, cement hydration at early ages but also last longer than expected.
after carbonation, the durability benefits of CNFs are clear: − The inversely determined permeability decreases with the increasing
drying RH. The most like reason is that low RHs can induce more
− Mineralogy. The presence of CNFs can help the material resist serious drying damage to the microstructure so permeability in­
carbonation. The carbonation ratios of hydration products are lower creases accordingly, while high RHs have lower effects on the
for materials containing CNFs so that more hydration products microstructure so permeability is lower.
remain in the carbonated materials. In the carbonation regime of this − The reliability of experimental data for cement pastes is slightly
study, there are about 72 % CH being carbonated if no CNFs were lower than mortars, because, during the drying tests, the drying
added, but this value drops to 60 % with 0.8 wt% CNFs. The amount cracks can be clearly seen on cement pastes, which thus changes the
of carbonated C-S-H and ettringite is halved comparing materials microstructure of cement pastes.
with 0.8 wt% CNFs with no CNFs. The amount of CO2 captured by
the fully materials decreases with the increasing CNFs dosage. The present study only focused on paste and mortar made from one
− Microstructure. The measured PSDs for noncarbonated pates indi­ type of cement which were carbonated at six months. To represent the
cate that the higher fractions of small pores are found in pastes with real situations of concrete structure, concrete with more types of ce­
CNFs than the neat cement paste. After carbonation, BET surface area ments and long-term experiments will be carried out to study the effects
as well as the amount of small pores (<10 nm) slightly increases with of CNFs and other nanomaterials on the durability of cementitious
the CNFs content. WVSIs also confirm that the higher CNFs content is materials.
able to better weaken the carbonation effect on the microstructure.
− Moisture transport. For the noncarbonated materials, water sorp­
tivity and liquid permeability slightly increase with the dosage of

15
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

CRediT authorship contribution statement Build. Mater. 82 (2015) 376–390, https://doi.org/10.1016/J.

CONBUILDMAT.2015.02.093.
[15] B. Birgisson, A.K. Mukhopadhyay, G. Geary, M. Khan, K. Sobolev, Nanotechnology
Zhidong Zhang: Conceptualization, Data curation, Formal analysis, in Concrete Materials: A Synopsis, 2012, https://doi.org/10.17226/22672.
Investigation, Methodology, Software, Validation, Visualization, Fund­ [16] D. Klemm, F. Kramer, S. Moritz, T. Lindström, M. Ankerfors, D. Gray, A. Dorris,
ing acquisition. Ueli Angst: Funding acquisition, Methodology, Nanocelluloses: a new family of nature-based materials, Angew. Chem. Int. Ed. 50
(2011) 5438–5466, https://doi.org/10.1002/anie.201001273.
Conceptualization, Writing - review & editing. [17] O. Onuaguluchi, D.K. Panesar, M. Sain, Properties of nanofibre reinforced cement
composites, Constr. Build. Mater. 63 (2014) 119–124, https://doi.org/10.1016/j.
conbuildmat.2014.04.072.
[18] L. Jiao, H. Zhu, Y. Wang, H. Dai, L. Chen, M. Su, Natural cellulose nanofibers as
Declaration of competing interest
sustainable enhancers in construction cement, PLoS One 11 (2017), e0168422,
https://doi.org/10.1371/journal.pone.0168422.
The authors declare that they have no known competing financial [19] Z. Zhang, G.W. Scherer, Measuring chemical shrinkage of ordinary Portland
interests or personal relationships that could have appeared to influence cement pastes with high water-to-cement ratios by adding cellulose nanofibrils,
Cem. Concr. Compos. 111 (2020), 103625, https://doi.org/10.1016/j.
the work reported in this paper. cemconcomp.2020.103625.
[20] S. Kawashima, S.P. Shah, Early-age autogenous and drying shrinkage behavior of
Data availability cellulose fiber-reinforced cementitious materials, Cem. Concr. Compos. 33 (2011)
201–208, https://doi.org/10.1016/j.cemconcomp.2010.10.018.
[21] A. Mezencevova, V. Garas, H. Nanko, K.E. Kurtis, Influence of thermomechanical
No data was used for the research described in the article. pulp fiber compositions on internal curing of cementitious materials, J. Mater. Civ.
Eng. 24 (2012) 970–975, https://doi.org/10.1061/(asce)mt.1943-5533.0000446.
[22] Y. Cao, P. Zavaterri, J. Youngblood, R. Moon, J. Weiss, The influence of cellulose
Acknowledgements nanocrystal additions on the performance of cement paste, Cem. Concr. Compos.
56 (2015) 73–83, https://doi.org/10.1016/j.cemconcomp.2014.11.008.
[23] J. Claramunt, M. Ardanuy, L.J. Fernandez-Carrasco, Wet/dry cycling durability of
This study received financial support from the Swiss National Sci­
cement mortar composites reinforced with micro- and nanoscale cellulose pulps,
ence Foundation (SNSF, project no. PP00P2-194812), the European Bioresources 10 (2015) 3045–3055, https://doi.org/10.15376/biores.10.2.3045-
Research Council (ERC) under the European Union's Horizon 2020 3055.
research and innovation program (grant agreement no. 848794), and [24] M. Ardanuy, J. Claramunt, F. Pares, E. Aracri, T. Vidal, Nanofibrillated cellulose as
reinforcement for high performance cement mortar composites, in: ECCM 2012 -
ETH Career Seed Awards (project no. 1-007733). ZZ would also thank Compos. Venice, Proc. 15th Eur. Conf. Compos. Mater 7(3), 2012, pp. 3883–3894.
the Cellulose and Wood Materials Laboratory at EMPA for providing https://upcommons.upc.edu/handle/2117/16392. (Accessed 1 October 2021).
cellulose nanofibrils. [25] J. Flores, M. Kamali, A. Ghahremaninezhad, An investigation into the properties
and microstructure of cement mixtures modified with cellulose nanocrystal,
Materials (Basel) 10 (2017), https://doi.org/10.3390/ma10050498.
References [26] M. Stefanoni, U.M. Angst, B. Elsener, Electrochemistry and capillary condensation
theory reveal the mechanism of corrosion in dense porous media, Sci. Rep. 8
(2018) 1–10, https://doi.org/10.1038/s41598-018-25794-x.
[1] V. Shah, K. Scrivener, B. Bhattacharjee, S. Bishnoi, Changes in microstructure
[27] Z. Zhang, P. Trtik, F. Ren, T. Schmid, C. Dreimol, U. Angst, Dynamic effect of water
characteristics of cement paste on carbonation, Cem. Concr. Res. 109 (2018)
penetration into carbonated mortar on steel corrosion: a neutron imaging,
184–197, https://doi.org/10.1016/j.cemconres.2018.04.016.
electrochemical and modeling study, Cement (2022), 100043, https://doi.org/
[2] M. Boumaaza, P. Turcry, B. Huet, A. Aït-mokhtar, Influence of carbonation on the
10.1016/J.CEMENT.2022.100043. Submitted.
microstructure and the gas diffusivity of hardened cement pastes, Constr. Build.
[28] C. Antonini, T. Wu, T. Zimmermann, A. Kherbeche, M.J. Thoraval, G. Nyström,
Mater. 253 (2020), 119227, https://doi.org/10.1016/j.conbuildmat.2020.119227.
T. Geiger, Ultra-porous nanocellulose foams: a facile and scalable fabrication
[3] M. Auroy, S. Poyet, P. Le Bescop, J.M. Torrenti, T. Charpentier, M. Moskura,
approach, Nanomaterials 9 (2019) 1142, https://doi.org/10.3390/nano9081142.
X. Bourbon, Impact of carbonation on unsaturated water transport properties of
[29] A. Balea, E. Fuente, A. Blanco, C. Negro, Nanocelluloses: natural-based materials
cement-based materials, Cem. Concr. Res. 74 (2015) 44–58, https://doi.org/
for fiber- reinforced cement composites. A critical review, Polymers (Basel) 11
10.1016/j.cemconres.2015.04.002.
(2019), https://doi.org/10.3390/polym11030518.
[4] H. Cui, W. Tang, W. Liu, Z. Dong, F. Xing, Experimental study on effects of CO2
[30] Z. Zhang, M. Shakoorioskooie, M. Griffa, P. Lura, U. Angst, A laboratory
concentrations on concrete carbonation and diffusion mechanisms, Constr. Build.
investigation of cutting damage to the steel-concrete interface, Cem. Concr. Res.
Mater. 93 (2015) 522–527, https://doi.org/10.1016/j.conbuildmat.2015.06.007.
138 (2020), https://doi.org/10.1016/j.cemconres.2020.106229.
[5] G. Villain, M. Thiery, Impact of carbonation on microstructure and transport
[31] A. Korpa, R. Trettin, The influence of different drying methods on cement pastemi-
properties of concrete, in: 10DBMC Int. Conférence Durab. Build. Mater.
crostructures as reflected by gas adsorption: comparison between freeze-drying, D-
Components, 2005, p. 2005.
drying, P-drying and oven-drying methods, Cem. Concr. Res. 36 (2006) 634–649,
[6] C.A.R. da Silva, R.J.P. Reis, F.S. Lameiras, W.L. Vasconcelos, Carbonation-related
https://doi.org/10.1016/j.cemconres.2005.11.021.
microstructural changesin long-term durability concrete, Mater. Res. 5 (2002)
[32] Z. Zhang, G.W. Scherer, Evaluation of drying methods by nitrogen adsorption,
287–293, https://doi.org/10.1590/s1516-14392002000300012.
Cem. Concr. Res. 120 (2019) 13–26, https://doi.org/10.1016/j.
[7] Y.F. Houst, The role of moisture in the carbonation of cementitious materials, Int.
cemconres.2019.02.016.
J. Restor. Build. Monum. 2 (1996) 49–66. http://infoscience.epfl.ch/record/29446
[33] M.C. Garci Juenger, H.M. Jennings, The use of nitrogen adsorption to assess the
/files/I_J_Rest_Buildg_96.pdf.
microstructure of cement paste, Cem. Concr. Res. 31 (2001) 883–892, https://doi.
[8] U. Angst, F. Moro, M. Geiker, S. Kessler, H. Beushausen, C. Andrade, J. Lahdensivu,
org/10.1016/S0008-8846(01)00493-8.
A. Köliö, K.I. Imamoto, S. von Greve-Dierfeld, M. Serdar, Corrosion of steel in
[34] R.L. Rarick, J.J. Thomas, B.J. Christensen, H.M. Jennings, Deterioration of the
carbonated concrete: mechanisms, practical experience, and research priorities – a
nitrogen BET surface area of dried cement paste with storage time, Adv. Cem.
critical review by RILEM TC 281-CCC, RILEM Tech. Lett. 5 (2020) 85–100, https://
Based Mater. 3 (1996) 72–75, https://doi.org/10.1016/S1065-7355(96)90073-5.
doi.org/10.21809/rilemtechlett.2020.127.
[35] Z. Zhang, G.W. Scherer, Physical and chemical effects of isopropanol exchange in
[9] V.T. Ngala, C.L. Page, Effects of carbonation on pore structure and diffusional
cement-based materials, Cem. Concr. Res. 145 (2021), 106461, https://doi.org/
properties of hydrated cement pastes, Cem. Concr. Res. 27 (1997) 995–1007,
10.1016/J.CEMCONRES.2021.106461.
https://doi.org/10.1016/S0008-8846(97)00102-6.
[36] Z. Zhang, M. Thiery, V. Baroghel-Bouny, An equation of drying kinetics for
[10] S.T. Pham, W. Prince, Effects of carbonation on the microstructure of cement
cementitious materials, Dry. Technol. 36 (2018) 1446–1459, https://doi.org/
materials: influence of measuring methods and of types of cement, Int. J. Concr.
10.1080/07373937.2017.1408644.
Struct. Mater. 8 (2014) 327–333, https://doi.org/10.1007/s40069-014-0079-y.
[37] S. Tada, K. Watanabe, Dynamic determination of sorption isotherm of cement
[11] E.G. Swenson, P.J. Sereda, Mechanism of the carbonatation shrinkage of lime and
based materials, Cem. Concr. Res. 35 (2005) 2271–2277, https://doi.org/10.1016/
hydrated cement, J. Appl. Chem. 18 (1968) 111–117, https://doi.org/10.1002/
j.cemconres.2005.01.002.
jctb.5010180404.
[38] C. Hall, P. Barnes, A.D. Billimore, A.C. Jupe, X. Turrillas, Thermal decomposition of
[12] L. Yuanhua, Z. Dajiang, Z. Dezhi, Y. Yuanguang, S. Taihe, D. Kuanhai,
ettringite Ca6 [Al(OH)6]2(SO4)3⋅26H 2O, J. Chem. Soc. - Faraday Trans. 92
R. Chengqiang, Z. Deping, W. Feng, Experimental studies on corrosion of cement in
(1996) 2125–2129, https://doi.org/10.1039/FT9969202125.
CO2 injection wells under supercritical conditions, Corros. Sci. 74 (2013) 13–21,
[39] T. Nishikawa, K. Suzuki, S. Ito, K. Sato, T. Takebe, Decomposition of synthesized
https://doi.org/10.1016/J.CORSCI.2013.03.018.
ettringite by carbonation, Cem. Concr. Res. 22 (1992) 6–14, https://doi.org/
[13] P.H.R. Borges, J.O. Costa, N.B. Milestone, C.J. Lynsdale, R.E. Streatfield,
10.1016/0008-8846(92)90130-N.
Carbonation of CH and C-S-H in composite cement pastes containing high amounts
[40] F. Paulik, J. Paulik, M. Arnold, Thermal decomposition of gypsum, Thermochim.
of BFS, Cem. Concr. Res. 40 (2010) 284–292, https://doi.org/10.1016/j.
Acta 200 (1992) 195–204, https://doi.org/10.1016/0040-6031(92)85115-C.
cemconres.2009.10.020.
[41] M.S.R. Sarker, M.Z. Alam, M.R. Qadir, M.A. Gafur, M. Moniruzzaman, Extraction
[14] Q.T. Phung, N. Maes, D. Jacques, E. Bruneel, I. Van Driessche, G. Ye, G. De
and characterization of alumina nanopowders from aluminum dross by acid
Schutter, Effect of limestone fillers on microstructure and permeability due to
carbonation of cement pastes under controlled CO2 pressure conditions, Constr.

16
Z. Zhang and U. Angst Cement and Concrete Research 162 (2022) 106990

dissolution process, Int. J. Miner. Metall. Mater. 22 (2015) 429–436, https://doi. [57] Z. Zhang, M. Thiery, V. Baroghel-Bouny, Investigation of moisture transport
org/10.1007/S12613-015-1090-2. properties of cementitious materials, Cem. Concr. Res. 89 (2016) 257–268, https://
[42] M. Auroy, S. Poyet, P. Le Bescop, J.M. Torrenti, T. Charpentier, M. Moskura, doi.org/10.1016/j.cemconres.2016.08.013.
X. Bourbon, Comparison between natural and accelerated carbonation (3% CO2): [58] O. Coussy, Mechanics of Porous Continua, John Wiley and Sons Ltd, Chichester,
impact on mineralogy, microstructure, water retention and cracking, Cem. Concr. United Kingdom, 1995.
Res. 109 (2018) 64–80, https://doi.org/10.1016/j.cemconres.2018.04.012. [59] V. Baroghel-Bouny, Water vapour sorption experiments on hardened cementitious
[43] Y. Sheng Han, G. Hadiko, M. Fuji, M. Takahashi, Crystallization and transformation materials. Part I: essential tool for analysis of hygral behaviour and its relation to
of vaterite at controlled pH, J. Cryst. Growth 289 (2006) 269–274, https://doi.org/ pore structure, Cem. Concr. Res. 37 (2007) 414–437, https://doi.org/10.1016/j.
10.1016/J.JCRYSGRO.2005.11.011. cemconres.2006.11.019.
[44] G. Wolf, E. Königsberger, H.G. Schmidt, L.C. Königsberger, H. Gamsjäger, [60] Z. Zhang, M. Thiery, V. Baroghel-Bouny, Numerical modelling of moisture transfers
Thermodynamic aspects of the vaterite-calcite phase transition, J. Therm. Anal. with hysteresis within cementitious materials: verification and investigation of the
Calorim. 602 (60) (2000) 463–472, https://doi.org/10.1023/A:1010114131577, effects of repeated wetting-drying boundary conditions, Cem. Concr. Res. 68
2000. (2015) 10–23, https://doi.org/10.1016/j.cemconres.2014.10.012.
[45] A.V. Radha, T.Z. Forbes, C.E. Killian, P.U.P.A. Gilbert, A. Navrotsky, [61] M. Thiery, P. Belin, V. Bouny, M. Nguyen, Modelling of isothermal drying process
Transformation and crystallization energetics of synthetic and biogenic amorphous in cementitious materials: analysis of the moisture transfer and proposal of
calcium carbonate, Proc. Natl. Acad. Sci. U. S. A. 107 (2010) 16438–16443, simplified approaches, in: 3rd Int. Conf. Coupled T-H-M-C, Process. Geo-Systems,
https://doi.org/10.1073/pnas.1009959107. Polytech Lille, France, 2008, pp. 571–581. http://hal.inria.fr/hal-00387501/.
[46] V. Baroghel-Bouny, Water vapour sorption experiments on hardened cementitious [62] M.T. van Genuchten, A closed-form equation for predicting the hydraulic
materials. Part II: essential tool for assessment of transport properties and for conductivity of unsaturated Soils1, Soil Sci. Soc. Am. J. 44 (1980) 892, https://doi.
durability prediction, Cem. Concr. Res. 37 (2007) 438–454, https://doi.org/ org/10.2136/sssaj1980.03615995004400050002x.
10.1016/j.cemconres.2006.11.017. [63] Y. Mualem, A new model for predicting the hydraulic conductivity of unsaturated
[47] M. Koch, S. Tenbohlen, T. Stirl, Diagnostic application of moisture equilibrium for porous media, Water Resour. Res. 12 (1976) 513–522, https://doi.org/10.1029/
power transformers, IEEE Trans. Power Deliv. 25 (2010) 2574–2581, https://doi. WR012i003p00513.
org/10.1109/TPWRD.2010.2048343. [64] C. Zhou, X. Zhang, Z. Wang, Z. Yang, Water sensitivity of cement-based materials,
[48] J.R. Philip, The theory of infiltration: 4. Sorptivity and algebraic infiltration J. Am. Ceram. Soc. 104 (2021) 4279–4296, https://doi.org/10.1111/JACE.17918.
equations, Soil Sci. 84 (1957) 257–264, https://doi.org/10.1097/00010694- [65] C. Zhou, F. Ren, Z. Wang, W. Chen, W. Wang, Why permeability to water is
195709000-00010. anomalously lower than that to many other fluids for cement-based material? Cem.
[49] C. Hall, Water sorptivity of mortars and concretes: a review, Mag. Concr. Res. 41 Concr. Res. 100 (2017) 373–384, https://doi.org/10.1016/j.
(1989) 51–61, https://doi.org/10.1680/macr.1989.41.147.51. cemconres.2017.08.002.
[50] Z. Zhang, G.W. Scherer, Determination of water permeability for a moisture [66] A.S. El-Dieb, R.D. Hooton, Water-permeability measurement of high performance
transport model with minimized batch effect, Constr. Build. Mater. 191 (2018) concrete using a high-pressure triaxial cell, Cem. Concr. Res. 25 (1995)
193–205, https://doi.org/10.1016/j.conbuildmat.2018.09.194. 1199–1208, https://doi.org/10.1016/0008-8846(95)00112-P.
[51] F. Ren, C. Zhou, Q. Zeng, Z. Zhang, U. Angst, W. Wang, Quantifying the anomalous [67] Z.C. Grasley, G.W. Scherer, D.A. Lange, J.J. Valenza, Dynamic pressurization
water absorption behavior of cement mortar in view of its physical sensitivity to method for measuring permeability and modulus: II. Cementitious materials,
water, Cem. Concr. Res. 143 (2021), 106395, https://doi.org/10.1016/j. Mater. Struct. Constr. 40 (2007) 711–721, https://doi.org/10.1617/s11527-006-
cemconres.2021.106395. 9184-y.
[52] C. Hall, A. Hamilton, Beyond the sorptivity: definition, measurement, and [68] R. Mejdoub, H. Hammi, J.J. Suñol, M. Khitouni, S. Boufi, A. M‘nif, Nanofibrillated
properties of the secondary sorptivity, J. Mater. Civ. Eng. 30 (2018) 1–7, https:// cellulose as nanoreinforcement in Portland cement: thermal, mechanical and
doi.org/10.1061/(ASCE)MT.1943-5533.0002226. microstructural properties, J. Compos. Mater. 51 (2017) 2491–2503, https://doi.
[53] C. Hall, Capillary imbibition in cement-based materials with time-dependent org/10.1177/0021998316672090.
permeability, Cem. Concr. Res. 124 (2019), 105835, https://doi.org/10.1016/j. [69] H. Dai, L. Jiao, Y. Zhu, C. Pi, Nanometer cellulose fiber reinforced cement-based
cemconres.2019.105835. material, CN105174768A. https://patents.google.com/patent/CN1051
[54] Z. Zhang, U. Angst, Modeling anomalous moisture transport in cement-based 74768A/en, 2015. (Accessed 22 June 2022).
materials with kinetic permeability, Int. J. Mol. Sci. 21 (2020) 837, https://doi. [70] J.W. Bullard, H.M. Jennings, R.A. Livingston, A. Nonat, G.W. Scherer, J.
org/10.3390/ijms21030837. S. Schweitzer, K.L. Scrivener, J.J. Thomas, Mechanisms of cement hydration, Cem.
[55] Z. Zhang, U. Angst, A dual-permeability approach to study anomalous moisture Concr. Res. 41 (2011) 1208–1223.
transport properties of cement-based materials, Transp. Porous Media 135 (2020) [71] J. Zhang, E.A. Weissinger, S. Peethamparan, G.W. Scherer, Early hydration and
59–78, https://doi.org/10.1007/s11242-020-01469-y. setting of oil well cement, Cem. Concr. Res. 40 (2010) 1023–1033, https://doi.org/
[56] M. Mainguy, O. Coussy, V. Baroghel-Bouny, Role of air pressure in drying of 10.1016/j.cemconres.2010.03.014.
weakly premeable materials, J. Eng. Mech. 127 (2001).

17

View publication stats