Cement and Concrete Research 64 (2014) 5462

Contents lists available at ScienceDirect

Cement and Concrete Research

journal homepage: http://ees.elsevier.com/CEMCON/default.asp

The inuence of different drying techniques on the water sorption

properties of cement-based materials
D. Snoeck a,b, L.F. Velasco c, A. Mignon a,b, S. Van Vlierberghe b, P. Dubruel b, P. Lodewyckx c, N. De Belie a,
a
b
c

Magnel Laboratory for Concrete Research, Department of Structural Engineering, Faculty of Engineering, Ghent University, Technologiepark Zwijnaarde 904, B-9052 Ghent, Belgium
Polymer Chemistry and Biomaterials Group, Department of Organic Chemistry, Faculty of Sciences, Ghent University, Krijgslaan 281, B-9000 Ghent, Belgium
Department of Chemistry, Royal Military Academy, Renaissancelaan 30, B-1000 Brussels, Belgium

a r t i c l e

i n f o

Article history:
Received 28 March 2014
Accepted 20 June 2014
Available online 11 July 2014
Keywords:
Drying (A)
CalciumSilicateHydrate (CSH) (B)
Microstructure (B)
Adsorption (water sorption) (C)
Carbonation (C)

a b s t r a c t
Dynamic water vapor sorption (DVS) may be used to characterize the pore structure of cementitious materials,
but the technique is difcult to interpret as the microstructure is very sensitive to drying and rehydration due
to humidity exposure. The removal of interlayer water or chemically bound water can cause microstructural
shrinkage. As all drying techniques more or less dehydrate CSH and ettringite, they cause a restructuration
of the C-S-H.
In the present paper, DVS measurements were performed to characterize the changes induced by different drying
techniques in the textural and sorption properties of the material, while thermogravimetric analysis was used to
elucidate carbonation.
The ideal drying technique, which can preserve the microstructure and can remove only the non-bound water,
does unfortunately not exist. All drying techniques separately affect the microstructure to some extent. However,
these changes are minimized when using vacuum-drying and the solvent-exchange-method with isopropanol as
drying techniques.
2014 Elsevier Ltd. All rights reserved.

1. Introduction
At some level, the behavior of every material is related to its microstructure. One of the most important properties of a cement paste is
thus the developing pore structure. Other cementitious material characteristics (e.g. strength and permeability) can be derived based on the
knowledge of the total pore volume, the pore size distribution, the specic surface area and the pore connectivity [1,2].
Besides the differentiation between pores, there also exist three
types of water related to the different pore sizes. First, there is the pore
water in the capillary pores, which are a few hundred nanometres in
size. Secondly, the interlayer water which is held by the gel pores by
capillary tension and strong hydrogen bonds to the CSH. The size of
the gel pores is in the order of nanometres. The gel is a threedimensional build-up of CSH layers with physically adsorbed water
molecules and interlayer water between the sheet structure [5]. Removal of this water by drying can cause considerable microstructural shrinkage. The third type of water is the chemically bound water, which is part
of the chemical structure of the hydrated phases, and can only be
removed by hydrate decomposition. Besides capillary and gel pores,
voids are also a classication of pores. In this paper, however, a
Corresponding author. Tel.: +32 92645522; fax: +32 92645845.
E-mail address: nele.debelie@UGent.be (N. De Belie).

http://dx.doi.org/10.1016/j.cemconres.2014.06.009
0008-8846/ 2014 Elsevier Ltd. All rights reserved.

classication was made according to micro b 2 nm, meso 2-50 nm and

macro N 50 nm pores (originally proposed by Dubinin [3] and subsequently ofcially adopted by the IUPAC [4]).
Moisture transport processes cannot be understood without the
knowledge of the moisture xation in the concrete pore system.
Water vapor sorption (i.e. the relation between moisture content and
relative humidity) is therefore a key parameter. But, the moisture sorption isotherm is a function of the age, the moisture history, the temperature, etc. So, it is of great importance to know the extent and history of
the drying process while performing a water vapor sorption test.
Also and comparatively with other sorption techniques, the use of
water adsorption has some advantages, among them are: i) it can be
performed at room temperature, ii) water has a small kinetic diameter
(i.e. 0.28 nm) which allows it to enter pores even smaller than the
ones accessible to CO2 and N2, and iii) for some kind of materials it is
sensitive to surface chemistry. Moreover, it is not necessary to submit
the sample to a degassing protocol prior to the measurement, thus
avoiding the partial removal of the ettringite phase.
Generally, all drying techniques more or less dehydrate CSH and
ettringite. A strong drying method may remove interlayer water from
CSH which re-enters when the RH increases again [6]. Drying causes
the restructuration of low-density CSH and a collapse below 40% RH
as well as the conversion of loosely-packed CSH to a denser CSH.
Drying enhances the chemical aging, meaning that the degree of

D. Snoeck et al. / Cement and Concrete Research 64 (2014) 5462

polymerization of the silicate chains has increased and the CSH has
changed irreversibly towards stiffer, stronger and denser ones [7]. Incomplete water removal likely leaves plugs or residual water in the
narrowest choke points, resulting in a complete pore blockage in some
areas of the paste [8,9]. During chemical aging, CSH tends to link
with each other, causing a compression of the hydrates. A strong drying
step accelerates this reaction because the CSH are brought closer
together [5].
The four main drying techniques described in literature to dry
cementitious materials are discussed hereafter.
1.1. Freeze-drying
In freeze-drying (F-drying), the ice sublimates, causing less capillary
stresses. Some investigations even showed a preservation of the pores
in the ne pore region (r b 5 nm) [10,11]. For that reason, this drying
technique is used frequently for MIP and TGA measurements [12] as it
also stops hydration. However, F-drying may also change the microstructure [13]. It causes limited signicant damage, due to thermomechanical stress in the inner CSH porosity [10,12,14]. So overall,
the literature is not conclusive on the effects of F-drying.
1.2. Oven-drying
Oven-drying is widely used because it is a fast drying technique.
Oven-drying, however, damages and alters the microstructure signicantly, even to a greater extent than freeze-drying [14]. For example,
oven-drying at 105 C removes unbound water but the capillary hydrostatic stresses due to surface tension of the receding water menisci generate a collapse of the ne pores [11,1416] as CSH is partly
dehydrated at 105 C [1,171]. Oven-drying also leads to the decomposition of monosulfoaluminate AFm and ettringite AFt phase at 60 C ([18]
as cited by [13]). It thus causes ettringite and CSH to lose a signicant
amount of non-evaporable water [14]. Drying at 50 C coarsens the pore
structure due to the polymerization of silicate anion chains and the development of a cohesive structure in CSH [19]. Generally, oven-drying
thus results in large capillary porosity due to capillary stresses, cement
hydrates (ettringite, AFm and CSH) desiccation, and micro-crack generation due to thermo-hydric stresses and the differential thermal expansion of the composition.
1.3. Vacuum-drying
Vacuum-drying is a slow technique and Zhang and Glasser ([18]
as cited by [13,14]) already observed that ettringite and monosulfoaluminate may degrade (structural and physical collapse of hydrates like AFm and AFt phases), causing damage to the pore structure
and increasing the pore volume ([18], as quoted by [12]).
1.4. Solvent-exchange-method
Solvent-exchange-methods replace the water by a solvent during
submersion of a sample in the solvent. The solvent is afterwards removed by ambient drying or vacuum-drying. The solvent-exchangemethods show the best results in preserving the microstructure as
they stress the microstructure less [10,12,20] and they are therefore
used to stop the hydration and to study the microstructure. However,
it remains possible that a solvent replacement creates chemical artifacts
in the specimens by reaction with hydration products or through strong
adsorption. Upon heating, methanol for example will react with CSH
to form CO2 and this is reected in a higher carbonate peak [2022]. It
binds with hydrating and mature cement paste. Some researchers stated that there might be a possible reaction of isopropanol with calcium
hydroxide [23], but in general, it is accepted that isopropanol does not
react with cement [20] as does methanol.

55

Which drying technique is the optimal is feed for discussion. Zhang

and Scherer [13] have summarized pros and cons of the various drying
techniques available as follows:

solvent-replacement N vacuum-drying N F-drying N

oven-drying
F-drying N oven-drying N solvent-replacement
oven-drying N F-drying N solvent-replacement

To preserve microstructure:
To preserve composition:
To save time:

In the present study, the effects of different drying techniques on the

microstructure were approached using a supplementary technique,
namely water vapor adsorption isotherms. In this way, knowledge is
gained on the effects of the drying techniques on the sorption properties, namely the formed microstructure of cementitious materials. In addition, thermogravimetric analysis was used to elucidate carbonation.
2. Materials and methods
2.1. Materials
A cement paste with a water-to-cement ratio of 0.5 was made and
cast. The cement used was CEM I 52.5 N (chemical and phase composition in Table 1) and the standard used for the mixing procedure was EN
196-1. After one day of storage in a relative humidity of 95 5% and a
temperature of 20 2 C, the samples were stored in water at 20
2 C for a minimum period of 6 months to ensure a stable formation
of the cementitious matrix. Prior to testing, the samples were ground
and sieved (5001000 m). To exclude the effect of contact with CO2
and the effects of carbonation, the samples were stored in water or in
the presence of soda lime during drying (Fig. 1b). The soda lime was replaced every day.
2.2. Methods
2.2.1. Drying techniques
Several drying techniques were studied. These include:
no-drying (20 2 C): specimens are stored in water;
freeze-drying: specimens are dried in liquid nitrogen and placed in a
freeze-dryer instantaneously for two weeks;
oven-drying at 40 C: for two weeks (ventilated oven);
oven-drying 105 C: for two weeks (non-ventilated oven);
air-drying in the presence of silica gel (20 2 C): for one month;
vacuum-drying (20 2 C): for two weeks at 0.1 bar;
the solvent-exchange-method in methanol and isopropanol followed by
vacuum-drying (20 2 C): stored in the solvent (solution:sample
volume ratio more than 100:1) for one week and subsequently
vacuum-dried for two weeks.
Table 1
Chemical and phase composition of CEM I 52.5 N. The
phase composition is calculated based on the formulas
of Bogue, corrected with the amount of SO3 for C3S.
CEM I 52.5 N
mass-%
CaO
SiO2
Al2O3
Fe2O3
SO3
MgO
K2O
Na2O
C3S
C2S
C3A
C4AF

Phase composition
mass-%

63.12
18.73
4.94
3.99
3.07
1.02
0.77
0.41
66.9
3.3
6.3
12.1

56

D. Snoeck et al. / Cement and Concrete Research 64 (2014) 5462

Temperature
controlled
chamber

Microbalance
module
Humidity regulated
sample module

H2O
dry gas

Flow
control
module

Vapour
module
N2
sample reference

(a)

(b)
Fig. 1. Schematic overview of the dynamic vapor sorption (DVS) methodology (a) and the used equipment, where samples stored in the presence of soda lime are put into the sample
container and the whole is put in the DVS equipment (b).

In addition, ground specimens were stored for six months in water

and six months in air to ensure complete carbonation for comparison
with a changed microstructure.

2.2.2. Water vapor sorption test

The dynamic vapor sorption apparatus used was from Surface Measurement Systems, London, UK (Fig. 1). The general principle of a sorption measurement is shown in Fig. 1a. The most important part is the
humidity-controlled sample chamber and the microbalance module.
By measuring the mass change as a function of time with changing
relative humidity (RH) of the sample versus an empty reference, sorption isotherms can be calculated. The humidity is regulated by mixing
humid and dry nitrogen gases. Fig. 1b shows the storage of samples in
the presence of soda lime and the material studied next to the DVS
equipment.
The temperature was set at 20 C, and the mass criterion to proceed
to the next RH step was dm/dt b 0.002 wt.%/min. The RH levels at which
samples (510 mg) were subsequently equilibrated included 9890
80706050403020105210% RH. Wet samples underwent
two desorptionadsorption cycles. Dried cement pastes were rst

conditioned at 0% RH inside the DVS equipment followed by two

adsorptiondesorption cycles.
The effect of the samples size was studied by performing DVS experiments with different granulometries. These were 250500, 5001000,
10002000 and 20005000 m. The larger the particle size, the longer
the time required for equilibration of the masses while increasing or
decreasing the RH. If the sample is too big, the time to reach equilibrium
is much higher and impracticable. Only the rst two size classes proved
to be acceptable. Practically, however, if a sample is too small, carbonation may occur and handling is less straightforward. In literature, the
most frequent sample sizes are: 5001000 [6], 6001180 [15], 800
1000 [24] and 6001200 m [11]. Therefore, a sample size of 500
1000 m was used in the present work.
The effect of the equilibration time was studied by imposing longer
times to reach the equilibrium. Since negligible differences were found
at longer equilibrium times, it was conrmed that the selected criterion
(dm/dt b 0.002 wt.%/min) ensured the reliability of the experiments.
2.2.3. Thermogravimetric analysis
The TGA instrument used in this investigation is a TGA Q50 Thermogravimetric Analyzer. A sample was hereby gradually heated at a rate of

D. Snoeck et al. / Cement and Concrete Research 64 (2014) 5462

10 C min1 till 900 C under a controlled N2 atmosphere. The mass loss

due to degradation of the material was recorded and used to check
whether the samples showed carbonation or not. The samples used
were the same as used for the dynamic vapor sorption tests.
The amount of physically bound water, chemically-bound water and
portlandite could be quantied using thermogravimetric analysis. The
amount of physically bound water was calculated as the difference
between the weight percent (wt.%) at 105 C minus the starting value.
The chemically-bound water will mainly be driven off at temperatures
above 105 C. It was determined by the mass loss between 105 C and
900 C minus the loss due to dehydration of the portlandite and the
loss due to decarbonation.
When portlandite CH decomposes to CaO and H2O (dehydroxylation
of Ca(OH)2), a mass loss is recorded around 420 C. Taking into account
the molecular weight of portlandite (MWCH) and water (MWH2O), the
weight loss in percent (WLCH) can be converted to the amount of CH
according to the rst part of Eq. (1). However, due to the decomposition
of calcite in CaO and CO2 (decarbonation of CaCO3) around 680 C and
the fact that calcite originates on the one hand from the original components (WLoriginal CO2) and on the other hand from the chemical reaction
between CH and CO2, an additional term is added to calculate the total
CH content irrespective of carbonation (with MWCO2 the molecular
weight of carbon dioxide). All amounts were expressed relative to the
nal mass at 900 C.

CH% WLCH % 

 MW
MW CH

CH
WLCaCO3 %WLoriginalCO2 %
MW H2 O
MW CO2
1

This CH value may be slightly overestimated as part of the CaCO3 also

originates from the carbonation of CSH [25]. This effect is negligible
for Portland cements and is more pronounced when one uses supplementary cementitious materials.
2.2.4. Used models for microstructural calculations
The type of branch of the water sorption isotherms taken into
account to calculate the microstructural properties is a point for discussion. The adsorption branch is not entirely correct as the menisci formation is delayed. The desorption branch is not usable due to delayed
evaporation or pore emptying [11]. As a consequence, the obtained
pore size distributions will not exactly represent the real pore
structures, but some general conclusions can be derived. In this paper,
and taking into account the shape of the measured isotherms (see
Section 3.2), the most frequently used adsorption curve was studied.
Although the understanding and modeling of water isotherms in
porous materials have been extensively researched, the underlying
mechanism of water adsorption is not yet fully elucidated to date and
remains the object of discussion [26]. Therefore, the interpretation of
the experimental water sorption isotherms is not always straightforward. Notwithstanding, some classical models (described below) can
be applied in order to calculate the textural parameters of the materials
from their water isotherms. The obtained results may differ from reality,
but they are still relevant and useful for comparative studies.
The BJH (Barrett, Joyner and Halenda) method [27] allows the
determination of the pore size distribution in the mesopore range
(250 nm) from experimental isotherms using the Kelvin model
of pore lling. In the Kelvin model, both capillary and adsorbed
water phases exist in cylindrical pores and iterative step-by-step
calculation leads to the pore size distribution [24].
The micropore volumes (b 2 nm) have been calculated by applying the DubininRadushkevich equation [28]. It is used to describe
the adsorption in microporous materials, mostly for activated
carbons and zeolites. The equation is based on the assumptions of a
change in potential energy between the gas and adsorbed phases,

57

and the characteristic energy of the solid [29]. The adsorption potential hereby corresponds to the change in vapor pressure (Polanyi
model) [30,31], leading to the DubininRadushkevich equation
(Eq. (2)):

ln W ln W 0

RT
 E0

2

 ln

p
p

!
2

where W0 is the micropore volume, W is the volume that has been

lled at a relative pressure of p/p0, E0 is the characteristic energy of
adsorption for a reference vapor, is the afnity coefcient, R is
the gas constant and T is the temperature (293 K). The DR equation
was applied at low relative humidities (110% RH).
The physical adsorption of vapor molecules on a solid surface was
explained by the BET theory (BrunauerEmmettTeller) [32,33]. This
model is an extension of the Langmuir theory [34] to multilayer adsorption. The BET equation considers that the rate of adsorption onto surface
sites is equal to the rate of evaporation from occupied sites, but it also
assumes that the energy of adsorption for the second and subsequent
layers is equal to that of liquefaction, and the total number of molecules
adsorbed is the sum of all the adsorbed layers. Although this model has
some drawbacks, it has been widely used ever since. In this regard, one
of the limitations of the BET method is that it can only be applied in
porous samples if these possess an open porosity [35], which is the
case for cementitious materials. With the BET theory, the specic
surface area accessible to water molecules (SBET) can be calculated within the low RH range of the adsorption isotherms. In this study, the
adsorption isotherm values at 10, 20 and 30% RH were used.
From the isotherms, the CSH gel amount can be calculated as the
ratio of the water amount adsorbed by the hardened cementitious
material to the water amount adsorbed by the CSH, following the
method described in [24]. It is calculated as the ratio of water adsorbed
at 20% RH (instead of 22.8% RH in [24]) to the water amount adsorbed
by the CSH at a certain RH. The latter value was determined in [7,
24] and is 0.219 g/cm3 at 22.8% RH and 0.26 at 20% RH. In this study,
both values are used to obtain an interval for the CSH content.
2.2.5. Statistical analysis
All standard deviations shown are deviations on individual results. All
sorption tests were conducted twice. A statistical analysis was performed
using the program SPSS in order to compare the obtained results. Multiple averages were compared using an analysis of variance (ANOVA) test
with a signicance level of 5%. The homogeneity of the variances was controlled with a Levene's test. The post-hoc test for data with homogenous
variances was a StudentNewmanKeuls test and if no homogenous variances were obtained, a Dunnett's T3 test was used.
3. Results and discussion
3.1. Thermogravimetric analysis
The obtained derivatives of the TGA curves are shown in Fig. 2.
There is a high mass loss between 50 and 500 C due to the release
of water from cement binding phases, like CSH. Generally, four
peaks are visible, corresponding to ettringite (AFt) at approximately 60 C, monosulfoaluminate (AFm) at 120 C, portlandite (dehydroxylation of Ca(OH) 2 ) at 420 C, and calcite (decarbonation of
CaCO3) at 680 C.
TGA results from carbonated cement pastes (dashed line in Fig. 2)
correspond to the proles found in [36]. The TGA curve of the samples
stored in water only shows a minor form of carbonation (saturated in
Fig. 2a). Oven-drying in a ventilated oven (40 C) and in air with silica
gel caused signicant carbonation, even though the soda lime was replaced regularly every day. Oven-drying at 105 C resulted in almost
no carbonation, as the air was not ventilated and soda lime was able

58

D. Snoeck et al. / Cement and Concrete Research 64 (2014) 5462

(a)

(a)

d(weight) [dm/dt]

CH through Portlandite (wt%)

5
10

CH through Calcite (wt%)

15
20

0.3
Saturated

Saturated
105C
Silica
Gel 3% RH
Air-drying

0.2

Carbonated
40C

Carbonated
F-drying
105C

0.1
40C

Temperature [C]

Air-drying

0
0

(b)

200

400

600

800

Vacuum dried
SEM Isopropanol

d(weight) [dm/dt]
SEM Methanol

0.3

Saturated
F-drying

0.2

Carbonated
Vacuum dried

(b)

Physically bound water (wt%)

0

Chemically bound water (wt%)

10

Saturated

0.1

Carbonated

Temperature [C]
0

F-drying
105C

200

400

600

800
40C

(c)

d(weight) [dm/dt]

Air-drying

0.3

Vacuum dried
SEM Isopropanol

Saturated
SEM Isopropanol

0.2

Carbonated
SEM Methanol

SEM Methanol

Fig. 3. Thermogravimetric analysis results [wt.%] of the portlandite content (a) and the
physically bound water and chemically bound water (b) of all specimens studied.

0.1
Temperature [C]
0
0

200

400

600

800

Fig. 2. Thermogravimetric analysis (TGA) showing partial carbonation of air- and ovendried specimens (a), no carbonation of F- and vacuum-dried specimens (b) and no
carbonation of specimens dried with solvent-exchange (c). The curves show the derivative of the TGA weight curve as a function of the temperature.

to extract most of the carbon dioxide from the air. Vacuum-drying did
not induce any signicant carbonation and F-drying caused partial carbonation (Fig. 2b). Specimens dried with the solvent-exchange-method
followed by vacuum-drying (Fig. 2c) also showed almost no carbonation (isopropanol) and partial carbonation (methanol). As methanol,
upon heating, will react with CSH to form CO2, this is reected in a
higher carbonate peak [22]. This heating was caused by performing
TGA analysis and would typically not happen at room temperature.
Oven-drying at 105 C removed part of the non-evaporable water,
dehydrated CSH, monosulfoaluminate (AFm) and ettringite (AFt)
phases as the peaks till 200 C in Fig. 2a lie lower compared to all
other drying techniques.
These ndings are also reected in the calculation of the total
amount of CH through portlandite and CH through the carbonation
peak (Fig. 3a). The total amount of portlandite is approximately the
same for all mixtures. Fig. 3a indicates that the carbonation increases
with the following drying techniques: F-drying, oven-drying at 105 C,
the solvent-exchange-method with methanol, oven-drying at 40 C in
a ventilated oven, drying in the presence of silica gel and the totally carbonated specimen. The other drying techniques did not show signicant
carbonation compared to the non-dried sample. The explanation of the
higher amount of carbonation reected by using methanol is given
above.
The amount of physically bound water left (Fig. 3b) reects the
strength of the drying technique. It is clear that the oven-drying technique at 105 C causes a signicant reduction in physically bound

water compared to other drying techniques as it is a strong drying technique. The lower value of drying in the presence of silica gel is due to the
change of the microstructure by carbonation, thus leading to a different
adsorption in the interior. Moreover, drying in the presence of silica gel
occurs near 3% RH, and the physically bound water in samples dried in
the presence of silica gel is thus still present in a higher extent compared
to samples dried with the other drying techniques. Only the F-drying,
vacuum-drying and the solvent-exchange-methods show the same
amount of physically bound water. The amount of physically bound
water in saturated specimens could not be determined as there was
still a lot of free water present during testing due to the continuous storage under water.
The amounts of chemically-bound water present do not seem to
differ, except for the oven drying at 105 C and the solvent-exchangemethod with methanol. During drying at high temperatures, part of
the chemically-bound water is removed. The lower value, when using
methanol, can be explained by the fact that methanol is anticipated to
react with the original constituents [22].

3.2. Water vapor sorption isotherms

Taking into account that water is likely to impair the mechanical
properties and durability of concrete structures, water vapor adsorption
measurements constitute an essential tool when characterizing the
properties of cement-bound materials.
The obtained water vapor sorption curves are shown in Fig. 4 and are
in correspondence with the ones found by Baroghel-Bouny [24]. The
repetition in testing falls between the boundaries and accuracy of the
test.
The shape of the isotherms gives us valuable information regarding the pore structure of the samples. The water molecules are rstly
adsorbed at the narrow micropores and secondly the sorption

D. Snoeck et al. / Cement and Concrete Research 64 (2014) 5462

(a)
20

Mass water
content [%]

Saturated
Carbonated
105C
40C
Air-drying

15

10

Relave humidity [%]

0
0

20

40

60

80

100

(b)
20

Mass water
content [%]

Saturated
Carbonated

15

F-drying
Vacuum dried

10

Relave humidity [%]

0

20

40

60

80

100

(c)
20

Mass water
content [%]

Saturated
Carbonated

15

59

bottle pores: during the desorption process small pores will constrict
the openings to larger pores such that the adsorbed water in the larger pores is not released until the relative pressure corresponds to
that of the smaller pore radius. An alternative approach suggests
that during the adsorption or/and desorption process the water is
in a metastable state leading to the assembly of water molecules
into different molecular structures during adsorption and desorption
[38]. It is clearly observed that the desorption branch of the obtained
isotherms presents a steep decrease at RH b 40%, this being attributable to the existence of ink-bottles pores in these materials and the
occurrence of cavitation. In this regard, it has to be explained that if
the diameter of the pore entrance is smaller than a certain critical
width (estimated to be ca. 5 nm for nitrogen at 77.4 K), the mechanism of desorption from the pore body involves the spontaneous nucleation and growth of gas bubbles in the metastable condensed uid
(cavitation). In this case, the body empties while the pore neck remains lled. When this occurs, the pressure of desorption depends
on the adsorbate and temperature and it is not correlated with the
size of connecting pores. Hence, pore size distributions calculated
from the desorption branch of the hysteresis loop are articial and
they do not reect the real pore sizes [39].
A deeper study of the isotherms has been carried out by applying
different models and the results are presented in the next subsections.

(a)
200

dV/d(logD)
[mm/g]

Saturated
Carbonated
105C
40C
Air-drying

SEM Isopropanol
SEM Methanol

150

10

100
5

50
0

Relave humidity [%]

0

20

40

60

80

100

Fig. 4. Inuence of the drying on the obtained sorption curves: air- and oven-drying (a), Fand vacuum-drying (b), and solvent-exchange followed by vacuum-drying (c). The curves
show the mass water content as a function of the relative humidity.

Pore diameter [m]

0
0.001

0.01

(b)
200

dV/d(logD)
[mm/g]
Saturated
Carbonated
F-drying
Vacuum dried

150

mechanism will gradually proceed through the lling of the pores of

increasing size. Thus, the low water uptakes in the rst part of the
isotherm and the steep rises of the slope of the adsorption curve at
relative pressures above 50% indicate that these samples are mainly
comprised of mesopores. There also are not that much bigger pores
present as there is no huge water uptake at a high relative humidity
(N 96%). This can be due to the crushing of the samples.
However, in the case of the carbonated specimen and the samples
dried in the presence of silica gel and oven-dried at 105 C, the slope
of the isotherms is almost linear until high relative pressures. This result,
coupled with the fact that their total water uptake is lower than for the
rest of the specimens, indicates that the drying technique has provoked
a partial collapse of the porous structure. Also, with increasing the
degree of carbonation, the porosity of the specimens is reduced. In this
regard it has been previously stated that carbonation reduced the porosity among the capillary mesopores in the outer CSH (10100 nm)
[37]. It is also worth it to mention that the specimens dried by
solvent-exchange, especially in the presence of methanol, lead to practically identical isotherms than in the case of the saturated sample, thus
indicating that this drying technique preserves the microstructure and
sorption properties of the material.
In all the cases, the isotherms exhibit a very pronounced hysteresis
loop. This phenomenon is still under investigation and nowadays
there are two feasible theories. The rst one is the existence of ink-

0.1

100

50
Pore diameter [m]
0
0.001

0.01

0.1

(c)
200

dV/d(logD)
[mm/g]
Saturated
Carbonated
SEM Isopropanol
SEM Methanol

150

100

50
Pore diameter [m]
0
0.001

0.01

0.1

Fig. 5. Inuence of the drying on the calculated microstructure with the BJH method: airand oven-drying (a), F- and vacuum-drying (b), and solvent-exchange followed by
vacuum-drying (c). The curves show the pore size distribution as a function of the pore
diameter, with V the volume of the pores and D the diameter.

60

D. Snoeck et al. / Cement and Concrete Research 64 (2014) 5462

Vmicro
0

50

Saturated

Vmeso [mm/g]
150

100

34

SBET [m/g]

200

250

23

F-drying
105C

95
142

40C

Air-drying

141

28

Vacuum dried

115

SEM Isopropanol

Vacuum

30

Isopropanol

33

Methanol

34

130

SEM Methanol

121 mg-1
81 mg-1
122 mg-1
93 mg-1
110 mg-1
122 mg-1
138 mg-1

151

Fig. 8. Obtained specic surface area SBET for all studied drying techniques.

156

Fig. 6. Amount (mm3g1) of micropores calculated with the DR-method and amount of
mesopores calculated with the BJH-method. (IUPAC pore classication: micropores
(b2 nm), mesopores (250 nm) and macropores (N50 nm)).

3.3. Amount and distribution of pores calculated with the BJH and DR
methods
As it has been previously explained, the mesopore size distributions
have been calculated by the BJH method applied to the adsorption
branch and the obtained proles are shown in Fig. 5. The results conrm
that the porous network of the materials is mainly formed of narrow
mesopores (dp ~ 56 nm), where the curves present a maximum. As
expected, all proles are co-alike except for the ones corresponding to
the specimens air-dried in the presence of silica gel, oven-dried at
105 C and the carbonated one, where the obtained pore size distribution is atter. This result evidence that these two drying techniques
are not adequate since the porous structure of the original sample has
been modied.
The volumes of micro and mesopores are presented in Fig. 6. As it is
clearly seen, the micropore volumes (calculated by the DR formulism)
are much lower in comparison with the volume of mesopores. In fact,
and for all the samples, the volume of micropores only accounts for
approximately the 20% of the total pore volume.
The CSH gel amounts are shown in Fig. 7. It is again clear that the
prominent changes are recorded in the same mixtures as above. The
lower values reect the densication of the CSH due to the loss of
physically bound water (Fig. 3b) which causes the CSH to change
from loose-packed CSH to denser ones. These results perfectly
match with the ones obtained by water adsorption. Thus, the three samples with the lowest water uptake are also the ones that present the
lowest CSH gel amount. Considering that the CSH is comprised of
gel pores and that it is the main phase responsible for the development
of the porosity in cementitious materials, when using invasive drying
techniques, not only the chemical composition of the material but also
its porous microstructure is affected.

C-S-H gel amount [mm/g]

0

77 mg-1

40C

34

Air-drying

105C

99

150
131 mg-1

F-drying

31
21

100

Saturated

165

Carbonated

Carbonated

50

50

100

150

200

250

Saturated
Carbonated
F-drying
105C
40C
Air-drying
Vacuum dried
SEM Isopropanol
SEM Methanol

Fig. 7. CSH amount for all samples after specic drying regimes, with deviations on the
boundaries.

3.4. Specic surface area (SBET)

The SBET-values are shown in Fig. 8. Typically, the SBET-values range
from 77 to 138 m2/g. These values are in good correspondence with
Baroghel-Bouny and other researchers [2,24,35], who report on the
values ranging from 80 to 143 m2/g.
A higher water removal does not result in a larger specic surface
area and more accessible porosity [11], which is also visible if one compares the value found for an oven-dried (105 C) specimen to all other
values for the specic surface area. Several authors observed that the
minimum surface area was obtained for oven-dried samples at high
temperatures [11,15]. When applying oven-drying at 105 C, the
amount of removed water is higher (see also Fig. 3b) but the surface
area is smaller. This can be interpreted as a possible pore collapse or
pore alteration. As more water is removed, ner pores are accessed,
but some gel pores cannot be accessed due to pore collapse. As a result,
not the original pore structure is studied as the microstructure is
changed.
The lower values for carbonated specimens and drying in air can be
explained by the degree of carbonation. Hence, and in accordance with
previous results [40], an inverse correlation is obtained between the
surface area and the amount of calcium carbonate present in the sample.
In addition, methanol-calcium ion products have a high surface area
which can affect the isotherm determination [23]. The latter explains
the highest value of SBET for the solvent-exchange-method with
methanol.
Based on the obtained results, the effect of the different drying techniques on the microstructural properties of the specimens is summarized as follows. Oven-drying causes capillary hydrostatic stresses due
to the surface tension of the receding water menisci and removed part
of the non-evaporable water. This results from the dehydration of C
SH and causes the collapse of the monosulfoaluminate (AFm) and
ettringite (AFt) phase at 60 C [18]. In fact, the thermogravimetric prole of the specimen oven-dried at 105 C shows, together with the carbonated sample, the lowest amount of these two phases (Fig. 2).
Generally, oven-drying results in cement hydrates (ettringite, AFm
and CSH) desiccation, and micro-crack generation due to thermohydric stresses and the differential thermal expansion of the composition. Capillary stresses due to receding water menisci may also induce
a modication of the textural properties of the sample. The pore collapse will cause a decrease of the amount of the smallest pores present
as capillary pressure increases with decreasing pore size. The effect of
cracking and microstructure alteration increases with increasing drying
temperature (40 C compared to 105 C). It removes the bound water
from ettringite and does not preserve the microstructure as the amount
of pores when dried at 40 C is higher compared to drying at 105 C.
Oven-drying is thus an unsuitable drying technique to preserve the
fragile microstructure of cement-based materials. Air-drying in the
presence of silica gel has to be avoided since it causes the carbonation
of the material and, therefore, the modication of its microstructural

D. Snoeck et al. / Cement and Concrete Research 64 (2014) 5462

20

4. Conclusions

Adsorbed lm
thickness []
t-curve boundaries
50..200
t-curve boundary commonly
used for cement pastes

15

10..14.5

Series7
Experimentally found t-curve
10

Relave pressure p/p0 [ -]

0
0

0.2

61

0.4

0.6

0.8

Fig. 9. Obtained t-curve compared to the curves used by Hagymassy Jr.

properties. Vacuum-drying may lead to the structural and physical

collapse of hydrates including AFm and AFt phases [18]. However, and
considering our results, it seems an acceptable drying technique that
preserves the amount of physically bound water and avoids carbonation. Finally, the solvent-exchange methods followed by vacuumdrying causes the smallest change in microstructural properties, thus
being the most suitable ones for the drying of cement-based materials.
Among both of them, isopropanol exchange should be selected when
submitting the sample at high temperatures, since methanol reacts
with CSH.

3.5. Comparison with the t-curve of Hagymassy Jr. et al. [41]

If the percentage of water mass obtained in the sorption isotherms is
divided by the SBET-value, the average thickness of the water lm
adsorbed on the solid surface is found (bold black curve of non-dried
specimens in Fig. 9). This curve can be compared to the ones suggested
by Hagymassy Jr. et al. [41] (solid gray and blue curves). The t-curve is
the statistical thickness of the lm adsorbed on nonporous adsorbents
as a function of the relative pressure [41].
All curves obtained are between the boundaries of the t-curves from
Hagymassy Jr. et al. [41]. They found that below the statistical thickness
of a monomolecular water layer (3 ) or thus a RH of approximately
23%, the strongly bound water is removed from the CSH. Moreover,
a linear relationship exists between 12 and 63% RH. The equilibrium
water amount adsorbed is proportional to the CSH amount. From
63% RH onwards, a signicant condensation takes place which implies
the overall lling of pores [24].
For cementitious materials, the blue curve is used as proposed by
Hagymassy Jr. et al. [41]. It can be seen that this curve is slightly higher
in values than the one found by recalculating it from the sorption isotherms. Up to 63% RH, this curve should be shifted downwards.

The conclusions are summarized in Table 2. The ideal drying technique, which can preserve the microstructure and can remove only
the non-bound water, does unfortunately not exist. All drying techniques affect the microstructure in their own way. Water sorption has
given additional experimental evidence that the best techniques to
dry the cementitious samples are vacuum-drying and the solventexchange-method with isopropanol.
Freeze-drying changed the microstructure due to thermo-mechanical
stress in the inner CSH. Oven-drying removed part of the nonevaporable water, dehydrated CSH, monosulfoaluminate (AFm) and
ettringite (AFt) phases, and caused thermo-hydric stresses due to differential expansion. Capillary stresses due to receding water menisci may
also induce a modication of the textural properties of the sample.
Oven-drying and air-drying proved to cause carbonation, which has its
impact on the microstructure. Also, oven-drying at 105 C and air drying
in the presence of silica gel led to the removal of part of the C-SH phase,
thus modifying the structural and textural properties of the sample.
Methanol reacted with CSH (only upon heating) but isopropanol
seemed to be inert with cementitious compounds.
Oven-drying is thus an unsuitable drying technique to preserve the
fragile microstructure of cement-based materials.
The porosity is reduced the most with increasing degree of carbonation. Carbonation reduced the porosity among the outer CSH (10
100 nm).
A higher water removal does not result in a larger specic surface
area and more accessible porosity. In oven-drying, the amount of
removed water is higher but the surface area is smaller. This can be
interpreted as a possible pore collapse or pore alteration.
All found isotherms are between the boundaries of the t-curves from
Hagymassy Jr. et al. [41]. The curve which should be used in the calculations is somewhat lower compared to the proposed one for cementitious materials as other materials were studied in Hagymassy Jr. et al.
[41].

Acknowledgments
As a Research Assistant of the Research Foundation-Flanders (FWOVlaanderen), D. Snoeck wants to thank the foundation for the nancial
support (11D7413N).
The authors also want to thank the Research Foundation-Flanders
for funding the project entitled Effect of Tunable Hydrogels on Concrete
Microstructure, Moisture Properties, Sealing and Self-Healing of Cracks
(3G019012).

References
[1] H.F.W. Taylor, Cement Chemistry, Thomas Telford Publishing, London, 1990.
[2] T.C. Powers, T.L. Brownyard, Studies of the Physical Properties of Hardened Portland
Cement Paste, Portland Cement Association, Research Laboratories, Cornell, 1948.
[3] M.M. Dubinin, The potential theory of adsorption of gases and vapors for adsorbents
with energetically nonuniform surfaces, Chem. Rev. 60 (1960) 235241.

Table 2
Conclusion of the effects of the drying techniques and carbonation.

Saturated
Carbonated
F-drying
105 C
40 C
Air-drying
Vacuum dried
Isopropanol
Methanol

No carbonation

No reduction of free water

No change in microstructure

Overall conclusion

++

+
+

++
++

++

++

+
++
++
++

++

+
++
++

++

++
++
+

62

D. Snoeck et al. / Cement and Concrete Research 64 (2014) 5462

[4] IUPAC, Manual of symbols and terminology, Appendix 2, Pt. 1, colloid and surface
chemistry, Pure Appl. Chem. 31 (1972).
[5] R.M. Espinosa, L. Franke, Inuence of the age and drying process on pore structure
and sorption isotherms of hardened cement paste, Cem. Concr. Res. 36 (2006)
19691984.
[6] N. De Belie, J. Kratky, S. Van Vlierberghe, Inuence of pozzolans and slag on the
microstructure of partially carbonated cement paste by means of water vapour
and nitrogen sorption experiments and BET calculations, Cem. Concr. Res. 40
(2010) 17231733.
[7] R.A. Olson, H.M. Jennings, Estimation of CSH content in a blended cement paste
using water adsorption, Cem. Concr. Res. 31 (2001) 351356.
[8] S. Diamond, A discussion of the paper Effect of drying on cement-based materials
pore structure as identied by mercury porosimetry a comparative study between oven-, vacuum- and freeze-drying by C. Gall, Cem. Concr. Res. 33 (2003)
169170.
[9] C. Gall, Reply to the discussion by S. Diamond of the paper Effect of drying on
cement-based materials pore structure as identied by mercury intrusion
porosimetry: a comparative study between oven-, vacuum- and freeze-drying,
Cem. Concr. Res. 33 (2003) 171172.
[10] L. Konecny, S.J. Naqvi, The effect of different drying techniques on the pore size distribution of blended cement mortars, Cem. Concr. Res. 23 (1993) 12231228.
[11] A. Korpa, R. Trettin, The inuence of different drying methods on cement paste microstructures as reected by gas adsorption: comparison between freeze-drying (Fdrying), D-drying, P-drying and oven-drying methods, Cem. Concr. Res. 36 (2006)
634649.
[12] N.C. Collier, J.H. Sharp, N.B. Milestone, J. Hill, I.H. Godfrey, The inuence of water removal techniques on the composition and microstructure of hardened cement
pastes, Cem. Concr. Res. 38 (2008) 737744.
[13] J. Zhang, G.W. Scherer, Comparison of methods for arresting hydration of cement,
Cem. Concr. Res. 41 (2011) 10241036.
[14] C. Gall, Effect of drying on cement-based materials pore structure as identied by
mercury intrusion porosimetry a comparative study between oven-, vacuumand freeze-drying, Cem. Concr. Res. 31 (2001) 14671477.
[15] M.C. Garci Juenger, H.M. Jennings, The use of nitrogen adsorption to assess the
microstructure of cement paste, Cem. Concr. Res. 31 (2001) 883892.
[16] J.J. Beaudoin, B.T. Tamtsia, Effect of drying methods on microstructural changes in
hardened cement paste: an A.C. impedance spectroscopy evaluation, J. Adv. Concr.
Technol. 2 (2004) 113120.
[17] M. Moukwa, P.C. Atcin, The effect of drying on cement pastes pore structure as
determined by mercury porosimetry, Cem. Concr. Res. 18 (1988) 745752.
[18] L. Zhang, F.P. Glasser, Critical examination of drying damage to cement pastes, Adv.
Cem. Res. 12 (2000) 7988.
[19] Y. Aono, F. Matsushita, S. Shibata, Y. Hama, Nano-structural changes of CSH in
hardened cement paste during drying at 50 C, J. Adv. Concr. Technol. 5 (2007)
313323.
[20] R.F. Feldman, J.J. Beaudoin, Pretreatment of hardened hydrated cement pastes for
mercury intrusion measurements, Cem. Concr. Res. 21 (1991) 297308.
[21] J.J. Beaudoin, P. Gu, J. Marchand, B.T. Tamtsia, R.E. Myers, Z. Liu, Solvent replacement
studies of hydrated portland cement systems: the role of calcium hydroxide, Adv.
Cem. Based Mater. 8 (1998) 5665.

[22] R.L. Day, Reactions between methanol and portland cement paste, Cem. Concr. Res.
11 (1981) 341349.
[23] J.J. Beaudoin, Validity of using methanol for studying the microstructure of cement
paste, Mater. Struct. 20 (1987) 2731.
[24] V. Baroghel-Bouny, Water vapour sorption experiments on hardened cementitious
materials part I: essential tool for analysis of hygral behaviour and its relation
to pore structure, Cem. Concr. Res. 37 (2007) 414437.
[25] P.H.R. Borges, J.O. Costa, N.B. Milestone, C.J. Lynsdale, R.E. Streateld, Carbonation of
CH and CSH in composite cement pastes containing high amounts of BFS, Cem.
Concr. Res. 40 (2010) 284292.
[26] S. Furmaniak, P.A. Gauden, A.P. Terzyk, G. Rychlicki, Water adsorption on carbons.
Critical review of the most popular analytical approaches, Adv. Colloid Interface
Sci. 137 (2008) 82143.
[27] E.P. Barrett, L.G. Joyner, P.P. Halenda, The determination of pore volume and area
distributions in porous substances. I computations from nitrogen isotherms, J.
Am. Chem. Soc. 73 (1951) 373380.
[28] M.M. Dubinin, Physical adsorption of gases and vapors in micropores, Prog. Surf.
Membr. Sci. 9 (1975) 170.
[29] C. Nguyen, D.D. Do, The DubininRadushkevich equation and the underlying microscopic adsorption description, Carbon 39 (2001) 13271336.
[30] A. Gil, P. Grange, Application of the DubininRadushkevich and DubininAstakhov
equations in the characterization of microporous solids, Colloids Surf. A Physicochem.
Eng. Aspects 113 (1996) 3950.
[31] G.O. Wood, Afnity coefcients of the Polanyi/Dubinin adsorption isotherm equations: a review with compilations and correlations, Carbon 39 (2001) 343356.
[32] S.J. Gregg, K.S.W. Sing, Adsorption, Surface Area and Porosity, Academic Press Inc.,
LondonNew-York, 1982.
[33] S. Brunauer, P.H. Emmett, E. Teller, Adsorption of gases in multimolecular layers, J.
Am. Chem. Soc. 60 (1938) 309319.
[34] I. Langmuir, The constitution and fundamental properties of solids and liquids. Part I.
Solids, J. Am. Chem. Soc. 38 (1916) 22212295.
[35] I. Odler, The BET-specic surface area of hydrated Portland cement and related materials, Cem. Concr. Res. 33 (2003) 20492056.
[36] G. Villain, M. Thiry, G. Platret, Measurement methods of carbonation proles in
concrete: thermogravimetry, chemical analysis and gammadensimetry, Cem.
Concr. Res. 37 (2007) 11821192.
[37] M. Thiry, P. Faure, A. Morandeau, G. Platret, J.-F. Bouteloup, P. Dangla, V. BaroghelBouny, Effect of carbonation on the microstructure and moisture properties of
cement-based materials, 12DBMC 12th International Conference on Durability
of Building Materials and Components, Porto, 2011, p. 8.
[38] T. Ohba, K. Kaneko, Cluster-associated lling of water molecules in slit-shaped graphitic nanopores, Mol. Phys. 105 (2007) 139145.
[39] M. Thommes, Physical adsorption characterization of nanoporous materials, Chem.
Ing. Tech. 82 (2010) 10591071.
[40] J.J. Thomas, J. Hsieh, H.M. Jennings, Effect of carbonation on the nitrogen BET surface
area of hardened portland cement paste, Adv. Cem. Based Mater. 3 (1996) 7680.
[41] J. Hagymassy Jr., S. Brunauer, R.S. Mikhail, Pore structure analysis by water vapor
adsorption I. t-Curves for water vapor, J. Colloid Interface Sci. 29 (1969) 485491.