Version of Record: [Link]

com/science/article/pii/S0950061820338551
Manuscript_2cc9a94a98551fcdbe5166ce5be2270a

Internal sulfate attack in mortars containing

contaminated fine recycled concrete aggregates

Colman, C.a,c,∗, Bulteel, D.a , Thiery, V.a , Rémond, S.b , Michel, F.c , Courard,
L.c
a Univ. Lille, IMT Lille Douai, ULR 4515 - LGCgE, Laboratoire de Génie Civil et
géo-Environnement, F-59000 Lille, France
b Univ Orléans, Univ Tours, INSA CVL, LaMé, EA 7494, France
c University of Liège, Urban and Environmental Engineering, GeMMe Building Materials,

4000 Liège, Belgium

Abstract

Internal sulfate attack can be caused by the gypsum residues present in fine
recycled aggregates (FRA). As opposed to the better known external sulfate
attack or Delayed Ettringite Formation (DEF), the sulfates in this context are
provided by a gypsum contamination of the aggregates. Mortars made with
contaminated FRA were subjected to different conditions, to assess which pa-
rameters had an influence on the sulfate attack reaction. Their mechanical
properties and microstructure are investigated. Results showed that gypsum
content, porosity, temperature and alkalinity influenced the consequences of
sulfate attack. However, the gypsum size distribution and cement type did not.
Keywords: Recycled aggregates, sulfate attack, microstructure, waste
management, secondary ettringite formation

1. Introduction

1.1. Gypsum in fine recycled concrete aggregates

One of the key points within the framework of a sustainable construction

sector is the recycling of its waste products, completing the life cycle of these
5 materials. Recycled concrete aggregates (RCA), obtained by the demolition

∗ Correspondingauthor
Email address: [Link]@[Link] (Colman, C.)

Preprint submitted to Construction and Building Materials August 10, 2020

© 2020 published by Elsevier. This manuscript is made available under the Elsevier user license
[Link]
 or deconstruction of older concrete structures, can be used in a new concrete
as a replacement for natural aggregates [1]. 1.7 tonnes of these RCA are pro-
duced per person per year in Europe, waiting to be valorized [2]. Gypsum
(CaSO4 .2H2 O) is used in the construction sector firstly as an addition to Port-
10 land cement, to regulate the setting time of concrete and prevent a flash set
[3]. Besides that, gypsum is the major constituent of plaster walls in build-
ings. RCA, as a consequence, will contain a certain amount of gypsum. Larger
concrete and gypsum particles can be separated from each other based on a
difference in color [4] or density [5], but these techniques are not applicable on
15 the smallest size fractions of RCA. In fine recycled aggregates (FRA), gypsum is
an important contaminant to be considered: the water soluble sulfates coming
from the gypsum particles strongly limit their valorization potential [6].

1.2. Sulfate attack: sources and mechanisms

Sulfate attack is a deteriorating process where sulfates react with water and
20 aluminate hydrates in a hardened cement paste to form secondary ettringite. It
is assumed that this mineral exerts a pressure on its surrounding cement paste
and causes a volumetric deformation [7]. Macroscopically, the concrete structure
will show swelling behavior and the formation of cracks. Ettringite is a normal
hydration product in the cement paste: its formation only becomes dangerous
25 when it occurs after setting, in a rigid cement matrix. Depending on the source
of the sulfates responsible for the reaction, a distinction can be made between
an external and an internal reaction. To experience external sulfate attack, the
structure is submerged in a sulfate rich environment such as soil or seawater.
Diffusion mechanics and microcrack propagation from the surface inwards are
30 determining factors here [8]. Internal sulfate attack happens when there is a
delayed release of sulfates from the hardened cement matrix. In this sense, an
internal source of sulfates eliminates the diffusion and microcrack necessity of
external sulfate attack, possibly accelerating the reaction. Delayed Ettringite
Formation (DEF), which occurs when high curing temperatures have destroyed
35 the sulfate hydrates that were initially formed [7], has been known for some

2
 time. The current situation, where FRA is contaminated with gypsum residues,
is a relatively new problem that has not been studied extensively.
The following aspects of the internal sulfate attack reaction were selected to
be elaborated in this study:

40 Alkalinity. Many authors show the important role of the alkalinity of the in-
terstitial solution, as it interferes with the equilibrium between the different
sulfate phases. A higher alkalinity favors the existence of monosulfate and the
absorption of sulfur on the C-S-H gel instead of the formation of ettringite [9],
so ettringite formation triggers as pH lowers. Nevertheless, a higher swelling
45 due to ettringite formation is often found in mixes with a higher alkalinity
[9, 10, 11, 12, 13]. While alkalinity speeds up hydration and increases early
compressive strength [14], it leads to lower mechanical performances in a sul-
fate presence [9, 10]. Besides a possible interaction with sulfate attack, alkalinity
is also a risk factor for the alkali-silica reaction and other durability issues.

50 Cement type. Using a sulfate-resisting cement allows the use of FRA with a
high sulfate content [15]. These types of cement contain less C3 A, one of the
reactants needed to form ettringite. Moreover, fewer gypsum is added to this
cement type as a setting retarder, compensating for the additional sulfate source
to which the mixture will be exposed. The SO3 /Al2 O3 ratio of a cement is an
55 important factor regarding its potential to form ettringite [16].

Porosity. The most commonly accepted theory about the cause of expansion
is the heterogeneous crystal pressure exerted by the growing ettringite crystals
[7, 3, 17]. In this sense, a lower porosity means more confinement and a higher
internal pressure. On the other hand, in the case of external sulfate attack, a
60 lower porosity would prevent the inwards diffusion of sulfates and thus limit the
swelling potential [18].

Sulfates. Before dissolution, sulfates can be associated with different cations. It

has been found that the sulfates originating from Na2 SO4 lead to more swelling
than those from CaSO4 [9], and that MgSO4 is even more damaging [19]. This

3
65 would suggest that a gypsum contamination is less damaging than other types
of sulfate attack that have been researched. However, these are external sources.
The rapid availability of an internal CaSO4 contamination could make this dif-
ference smaller.
Coarser gypsum particles are hypothesized to not feed early ettringite for-
70 mation, but react later in an already rigid cement matrix. To keep the risk on
sulfate attack at a reasonable level, the current water soluble sulfate limit in
coarse recycled aggregates is established at 0.2% by EN 206 [20], with no men-
tion of FRA. At these quantities, sulfates are considered the limiting reagent
in the ettringite formation reaction so any augmentation would hypothetically
75 lead to more swelling. The conclusions of recent durability studies indicate a
level of 0.3% should be made possible [21].

Thaumasite formation. Next to ettringite, sulfates can also contribute to the

formation of the expansive mineral thaumasite. While damage caused by thau-
masite is more severe than that caused by ettringite, thaumasite formation does
80 not occur as often [22]. Only at temperatures lower than 10 ◦ C and in the
presence of a carbonate source, is thaumasite favored over ettringite [23].

1.3. Objectives

In this study, contaminated FRA were used in mortars to research the dam-
aging effects of sulfates. Each parameter of interest was varied while others
85 were kept constant, to identify the factors that can worsen or mitigate the sul-
fate attack results. Knowing which parameters to manipulate in a mix design
with highly contaminated FRA will ultimately promote the use of these recy-
cled aggregates. The results of this study could also contribute to the ongoing
discussion about the sulfate attack reaction mechanism [24].

4
 CEM I 52.5 N Water Limestone aggregates (mm) Superplasticizer
0/4 2/7 7/14 14/20
350 175 216 658 436 612 0.4%

Table 1: Composition, in kg, of the original concrete

90 2. Materials and methods

2.1. Used materials

FRA were made in the laboratory by fabricating a concrete and subsequently

crushing it. The composition of this original concrete is given in Table 1, and was
designed to obtain a consistency class S3 and strength class C30/37. After 90
95 days of curing, this concrete was crushed by a jaw crusher and the resulting 0/4
mm fraction was used as FRA in all described tests. The use of this ’model’ FRA
gave exact control of the chemical composition of the materials and removed any
possible variability or contamination at the level of the aggregates by chlorides,
organics, etc. This FRA was then manually contaminated with gypsum to
100 obtain a ’clean’ material where only sulfates could contribute to a deteriorating
reaction.
Figure 1 and Table 2 summarize the properties of the resulting FRA. Water
absorption and particle density of the FRA were determined via the method de-
scribed by Zhao et al. [25]. Characterization techniques for natural aggregates,
105 described in EN 1097-6 [26], consistently underestimate the water absorption
of FRA because of the fineness and agglomeration issues between the particles.
The method - designed in response to this difficulty - by IFSTTAR [27] seems
to overestimate the water absorption of FRA but works well for particles in the
0.5/4 mm range. Thanks to an very good correlation between the hardened ce-
110 ment paste content or mass loss at 475 ◦ C and the water absorption, the water
absorption of the fines can then be extrapolated. Using the water absorption
of each size fraction (either measured for the coarser particles or calculated for
the fines) is more accurate than using either of the two mentioned experimen-
tal methods for the whole 0/4 mm bulk [28]. Even though no gypsum was

5
 100

80

60
Passed (%)

40

20

0
0.1 1 10
Particle size (mm)

Figure 1: Size distribution 0-4 mm of the used FRA

Water absorption Particle density SO4 2- content

9.78% 1.95 g/cm3 0.18%

Table 2: Characterization of the used FRA

115 addedto the model concrete, 0.18% of water soluble sulfates were measured via
ion chromatography, originating from the used cement.
The gypsum used to contaminate this FRA was a CaSO4 .2H2 O powder
(D50 13 µm) obtained from VWR Chemicals. The sulfates from this gypsum
contamination are added to the 0.18% of water soluble sulfates already in this
120 FRA, originating from cement particles. A CEM I 52.5 N cement from HOLCIM
was used as the default cement, in one test replaced by a High Sulfate Resisting
(HSR) CEM I from the same manufacturer. The chemical composition of these
cements is shown in Table 3.

6
 Chemical CEM I 52.5 N CEM I HSR
CaO 64.3 64.6
SiO2 18.3 21.4
Al2 O3 5.2 3.7
Fe2 O3 4.0 4.6
MgO 1.4 0.8
Na2 O 0.32 0.27
K2 O 0.43 0.40
SO3 3.5 2.5
Cl- 0.06 0.06
LOI 2.3 1.3
C3 A 6.6 2.4
C4 AF 12 14
C3 S 61.9 68.8
C2 S 11.2 9.4

Table 3: Chemical composition (mass%) of the used cement types

7
 2.2. Mortar fabrication

125 All mortars were prepared with the FRA described in section 2.1, and were
contaminated with 5% (by weight of the granular fraction) of gypsum. This
gyspum, together with the residual sulfates in the model FRA, brings the total
sulfate content of this mix to 3.08%. 7 days before mixing, the FRA was pre-
saturated with its absorbed water and 10% of the mixing water. The standard
130 procedure described in EN 196-1 [29] for mortar fabrication was followed, where
normalized sand was replaced volumetrically by FRA: using a particle density
of 2.6 g/cm3 for normalized sand and 1.95 g/cm3 for the FRA, the aggregate
envelope volume was kept constant. After a cure of 24 hours, the mortars were
kept in water at 21◦ C. These described compositions or conditions were then
135 varied accordingly, depending on the parameter that was tested.

2.3. Tested parameters

The following parameters were hypothesized to have an effect on the internal

sulfate attack reaction. The influence of each one was tested with an exaggerated
high (+1) and a low (-1) level, and in one case also an intermediate (0) level.
140 Table 4 summarizes these levels and how each of them was obtained by adapting
the standard mortar composition. Table 5 shows the compositions in more
detail. To isolate the response of only one parameter, each series of replicates
is kept in its own container, so as not to be influenced by the leaching water
of another [30]. While one factor is being researched, all other parameters are
145 kept as described in section 2.2.

Alkalinity. The lower level of this parameter is the normal alkalinity present in
the used cement. The higher level is double this amount, achieved by adding
NaOH to the mixing water.

Cement type. To test the influence of the available C3 A, a HSR cement was
150 used. A small difference in alkalinity between the HSR cement and the CEM I
was mitigated by adding NaOH to the mixing water, bringing the Na2 O eq. of
both cements on the same level.

8
 Gypsum grain size. The same fine gypsum powder used in the other samples,
was hardened and subsequently crushed to obtain particles in the 2/4 mm range.
155 These two size distributions - the powder or the coarser particles - are used to
contaminate the FRA.

Porosity. To research the influence of the available porosity, the water to cement
ratio (W/C) was varied.

Sulfate content. A sulfate amount, one order of magnitude smaller than in the
160 other mortars, was used to demonstrate the importance of this parameter.

Thaumasite formation. These mortars were kept in water at 5◦ C to promote

thaumasite formation over ettringite. For thaumasite formation, a source of
carbonates is necessary besides C3 A, sulfates and water. 20% by mass of CEM
I was therefore replaced with a limestone filler. Here again, NaOH was added
165 so the alkalinity of the mix resembled the others.

9
 Factor Level
Alkalinity (A) -1 0.61% Na2 O eq.
+1 1.2% Na2 O eq.
C3 A content (C) -1 2.4% (CEM I HSR)
+1 6.6% (CEM I)
Gypsum grain size (G) -1 Powder (D50 13 µm)
+1 2/4 mm distribution
Porosity (P) -1 W/C 0.35
0 W/C 0.5
+1 W/C 0.65
Sulfate content (S) -1 0.47% of water soluble sulfates
+1 3.08% of water soluble sulfates
Temperature (T) -1 5 ◦ C and a carbonate addition
+1 21 ◦ C

Table 4: Summary of how the mortar composition was changed to obtain the levels of the
different parameters

10
 Name Cement FRA Absorbed Mixing Gypsum Extra
water water
A-1, C+1, G-1, P0, S+1, T+1 1.35 0.96 0.094 0.675 0.05
A+1 1.35 0.96 0.094 0.675 0.05 10.30 g NaOH
C-1 1.35 0.96 0.094 0.675 0.05 1.2 g NaOH
G+1 1.35 0.96 0.094 0.675 0.05
P-1 1.35 0.96 0.094 0.473 0.05
11

P+1 1.35 0.96 0.094 0.878 0.05

S-1 1.35 1.005 0.098 0.675 0.005
T-1 1.08 0.96 0.094 0.675 0.05 270 g limestone filler,
2.1 g NaOH

Table 5: The compositions, in kg, of the different mortar samples. The changes between a mix and its reference composition is placed in bold. For
the C-1 samples, the difference is in the type of cement, for the G+1 samples the size of the gypsum particles. Since the gypsum contamination is
expressed as a mass% of the aggregate part, a lower gypsum content in S-1 means more FRA and thus also more absorbed water.
 2.4. Monitoring of the reaction

To follow the development of the internal sulfate attack reaction, the mortar
specimens were subjected to different tests. On a macroscopic level, the mass,
length and ultrasonic wavespeed were recorded weekly to observe features of
170 sulfate attack such as swelling and possible internal cracking. At 7, 28, 90
and 180 days the mortars were characterized mechanically for their compressive
strength [29] and porosity by Mercury Intrusion Porosimetry. Every described
test was done for 3 replicate mortars. A microstructural analysis for the samples
at 90 days was carried out to examine the pore structure and ettringite deposits,
175 to complement the swelling results and provide grounds for their interpretation.
The samples were prepared according to routine procedures of embedding and
polishing [31] using a 2020 resin from Huntsman and MD System from Struers
with water-free diamond pastes and lubricants. Images in scanning electron
microscopy were obtained on a Hitachi S-4300/SE-N and coupled with EDS
180 analyses.

3. Results and discussion

In Figure 2, the length change of the mortars is shown. Standard devia-

tions are not shown on these figures to improve their readability, but are taken
into account when performing an unpaired t-test to check whether the swelling
185 results differ from each other statistically.
According to Table 6, C-1, G+1 and P+1 are similar to the standard mix
where all parameters have their normal value. Four other samples deviate from
this trend: P-1 and T-1 had a larger expansion, A+1 and S-1 had a lower expan-
sion. Next to their length, the mechanical properties that were monitored are
190 presented in Figure 3. The compressive strength of all samples kept steadily in-
creasing over time but did not seem to have any correlation to the corresponding
swelling amounts. The samples that showed a high or low swelling did not have
a low or high compressive strength, respectively. There was a large variation of
the measured strengths between 25 and 40 MPa, the parameters that did not

12
 Name Length Stdev Sample Difference
(180 days) (180 days) size for p>0.05
A-1, C+1, G-1, P0, 0.0675 0.007 3
S+1, T+1
A+1 0.017 0.002 3 yes
C-1 0.081 0.005 3 no
G+1 0.060 0.002 3 no
P-1 0.113 0.004 3 yes
P+1 0.063 0.004 3 no
S-1 0.037 0.010 3 yes
T-1 0.137 0.004 3 yes

Table 6: Unpaired t-test to evaluate if there is a significant difference between the samples
and their reference mix, shows that 3 parameters did not change the swelling results (C-1,
P+1 and G+1), and 4 did (T-1, P-1, S-1, A+1).

195 influence the swelling amount did influence the compressive strength. There
was less variation between the samples in terms of porosity, only the sample
with a limited W/C ratio had a distinct lower porosity. The macroscopic dif-
ferences between the parameters are interpreted together with microstructural
observations.

13
 A-1, C+1, G-1, P0, S+1, T+1
0.14
A+1
C-1
0.12 G+1
P-1
P+1
0.1 S-1
Length change (%)

T-1
0.08

0.06

0.04

0.02

0
0 20 40 60 80 100 120 140 160 180
Age (days)

Figure 2: 6 month swelling behavior of the mortar samples in function of the tested parameters

14
 45
A-1, C+1, G-1, P0, S+1, T+1
A+1
40 C-1
G+1
P-1
Compressive strength (MPa)

P+1
35 S-1
T-1

30

25

20

15
0 20 40 60 80 100 120 140 160 180
Age (days)

A-1, C+1, G-1, P0, S+1, T+1

50 A+1
C-1
G+1
P-1
40
P+1
S-1
T-1
Porosity (%)

30

20

10

0
0 20 40 60 80 100 120 140 160 180
Age (days)

Figure 3: Compressive strength and porosity of the samples at 7, 28, 90 and 180 days

15
200 3.1. General case

The four curves on Figure 2 that stay together between the 0.06% and 0.08%
marks, show that increasing the porosity, using larger gypsum grains, or lim-
iting the available C3 A did not influence the amount of swelling. SEM results
confirmed that the morphology of these samples was very similar. Figure 4 is a
205 typical image found with recycled materials: the recycled aggregate is a cluster
of natural aggregates in the original cement paste. The new cement paste, sur-
rounding the recycled aggregates, had a notably higher air content, which has
been observed before by Bouarroudj et al. [32] for the same material. This air
content can be explained by the surface roughness of the recycled aggregates
210 [33], which captures more air into the mixture than a round (natural) aggregate.
Next to that, the difference in surface free energy between the new cement paste
and the recycled aggregates could also have played a role in this elevated air
content [34]. Figure 5 illustrates how ettringite deposits in these samples were
mainly found in pores or air bubbles, which sometimes also resulted in cracks
215 in the surrounding paste.

16
Figure 4: Contrast between a recycled aggregate and the new cement paste with a high air
content

Figure 5: Ettringite was found in pores and air bubbles, which exerted a pressure and cracked
the surrounding cement paste

17
 3.2. Alkalinity

Sample A+1 with an increased alkalinity did not show any evidence of cracks.
Air bubbles were often filled with ettringite, but not in a way that caused
damage. Instead, a high percentage of them showed the presence of a piece of
220 calcite or portlandite in their center, with ettringite crystals growing outwards
of this center and not inwards from the cement paste. This is demonstrated
in Figure 6. Calcite has been described as a nucleation center for very fast
ettringite growth at high alkalinity [35, 36]. The kinetics of this reaction explain
the lack of available sulfates to cause damage in a cured cement matrix. A higher
225 alkalinity has also been observed by Juenger et al. [14] to increase the initial
rate of hydration and cause a higher early compressive strength, which has been
confirmed in Figure 3. However, the swelling and compressive strength results
obtained in these experiments contradicted the findings of numerous authors
described in section 1.2. This could be due to the chosen experimental setup:
230 the high alkali values in this study were obtained by adding NaOH, which is
immediately available. Alkali’s coming from the adherent cement paste of FRA
would take more time to leach into the interstitial solution. The rapid initial
hydration in this sample prevented the higher swelling that is normally provoked
by the alkalinity of FRA.

18
 Figure 6: A piece of calcite promoted the formation of ettringite at high alkalinity at an early
age

235 3.3. Porosity

While increasing the W/C ratio did not have an effect on the reaction, lower-
ing it raised the swelling amount. Figure 3 shows a distinctly lower porosity for
sample P-1, but not an equally large difference between P0 and P+1. In Figure
7, these findings are confirmed by showing a similar morphology for P0 and
240 P+1, but a significantly denser matrix for P-1. No superplasticizer was added
to counteract a loss in workability due to the lower W/C. The microstructure
of P-1 also showed the presence of more unreacted cement particles. The P-
1 samples displayed many cracks throughout which were filled with ettringite
crystals, as illustrated in Figure 8. A high swelling level would seemingly be a
245 good indicator for internal damage.

19
Figure 7: Comparison of the cement paste density between P-1, P0, and P+1

20
 Figure 8: Cracks filled with ettringite in a sample with lower porosity

3.4. Sulfates

The S-1 samples contained only 0.28% of water soluble sulfates, which is
still well above the maximum allowable limit in recycled aggregates [20]. Still,
no significant damage occured, as shown by its swelling behavior in Figure 2
250 and microstructure in Figure 9. The long term compressive strength of these
samples, shown in Figure 3, was lower than those with higher sulfate contami-
nations, but only because it did not increase after the first month of aging.

21
 Figure 9: Lowering the sulfate content resulted in an undamaged sample

3.5. Temperature

The samples kept at lower temperatures showed an important amount of

255 swelling compared to the other mixes, but the reaction seems slower and did
not reach a stabilization point yet after 6 months. Compressive strength, shown
in Figure 3, was much lower compared to the other samples, which is due to the
limited amount of cement in this mix. Figure 10 shows the internal damage in
these samples, while EDS confirmed that pores, air bubbles or cracks are filled
260 with an ettringite/thaumasite mixture.

22
 Figure 10: The T-1 samples were damaged by a mixture of ettringite and thaumasite

4. Conclusion

The water soluble sulfates in FRA are responsible for a deterioration when
they react with C3 A and water in a new cementitious mix. These sulfates could
originate from gypsum residues at the demolition site, but also from cement
265 particles in otherwise ’uncontaminated’ FRA. The severity of this deteriorating
reaction is determined by other factors: some can limit the swelling potential,
others enhance it. Six parameters were chosen to research and their influence
on the sulfate attack reaction has been identified. The use of a model FRA
manually contaminated with gypsum made it possible to isolate the responses
270 of only one parameter at a time, without interference of other variabilities or
contaminations.
Limiting the C3 A content, using coarse gypsum particles, or augmenting the
W/C ratio did not significantly change the swelling results. The formation of
thaumasite or limiting the available porosity lead to a larger expansion. On
275 the other hand, increasing the alkalinity of the mix or using a lower sulfate
content seemed to limit the swelling results. This indicates that the maximum
sulfate content established in EN206 may be too strict, which is in line with
the findings of the PN RecyBéton [21] and their subsequent proposal to set this
limit at 0.3%. Unrelated to their influence on the swelling amount, the variation

23
280 of these parameters also changed the compressive strength of the mortars.
In general, the use of FRA in mortars led to a high air content, which was
responsible for a lower compressive strength when compared to the compressive
strength of a standard mortar with natural aggregates. This air content was
explained by the surface roughness of recycled aggregates.
285 These results provide industrials with helpful information for mix designs
with high sulfate contents. This could in turn help promote the use of contam-
inated FRA, which are up to now not valorized.
Future work should focus on upscaling these tests to concrete, and/or re-
search a possible interaction between the parameters.

290 Acknowledgements

This research was conducted in the framework of the Interreg project VAL-
DEM ‘Solutions intégrées de valorisation des flux ”matériaux” issus de la démolition
des bâtiments: approche transfrontalière vers une économie circulaire’ (Conven-
tion n◦ 1.1.57 of Interreg France–Wallonie-Vlaanderen 2014-2020).

295 References

[1] N. Oikonomou, Recycled concrete aggregates, Cement & Concrete Com-

posites 27 (2005) 315–318. doi:10.1016/[Link].2004.02.020.

[2] S. Delvoie, Z. Zhao, F. Michel, L. Courard, Market analysis of recycled

sands and aggregates in northwest europe: drivers and barriers, in: IOP
300 Conference series: Earth and Environmental Science, 2019, pp. 1–8. doi:
10.1088/1755-1315/225/1/012055.

[3] M. Collepardi, A state-of-the-art review on delayed ettringite attack on

concrete, Cement & Concrete Composites 25 (2003) 401–407. doi:10.
1016/S0958-9465(02)00080-X.

305 [4] H. Asakura, Y. Watanabe, Y. Ono, M. Yamada, Y. Inoue, A. M. Al-

faro, Characteristics of fine processed construction and demolition waste

24
 in japan and method to obtain fines having low gypsum component
and wood contents, Waste Management & Research 28 (2010) 634–646.
doi:10.1177/0734242X09339951.

310 [5] W. Ambrós, C. Sampaio, B. Cazacliu, G. Miltzarek, L. R. Miranda, Sepa-

ration in air jigs of mixed construction and demolition waste, in: Advances
in recycling and management of construction and demolition waste, 2017,
pp. 25–28.

[6] F. Agrela, M. S. de Juan, J. Ayuso, V. Geraldes, J. Jiménez, Limiting

315 properties in the characterisation of mixed recycled aggregates for use in
the manufacture of concrete, Construction and Building Materials 25 (2011)
3950–3955. doi:10.1016/[Link].2011.04.027.

[7] S. Diamond, Delayed ettringite formation - processes and problems, Cement

& Concrete Composites 18 (1996) 205–215.

320 [8] C. Yu, W. Sun, K. Scrivener, Mechanism of expansion of mortars immersed

in sodium sulfate solutions, Cement and Concrete Research 43 (2013) 105–
111. doi:10.1016/[Link].2012.10.001.

[9] G. Escadeillas, J. Aubert, M. Segerer, W. Prince, Some factors affecting

delayed ettringite formation in heat-cured mortars, Cement and Concrete
325 Research 37 (2007) 1445–1452. doi:10.1016/[Link].2007.07.004.

[10] A. Pavoine, X. Brunetaud, L. Divet, The impact of cement parameters on

delayed ettringite formation, Cement & Concrete Composites 34 (2012)
521–528. doi:10.1016/[Link].2011.11.012.

[11] M. Halaweh, Effect of alkalis and sulfates on portland cement systems,

330 Ph.D. thesis, University of South Florida (2006).

[12] N. Leklou, Contribution à la connaissance de la réaction sulfatique interne,

Ph.D. thesis, Université Paul Sabatier III (2008).

25
 [13] V. Nguyen, N. Leklou, J. Aubert, P. Mounanga, The effect of natural
pozzolan on delayed ettringite formation of the heat-cured mortars, Con-
335 struction and Building Materials 48 (2013) 479–484. doi:10.1016/j.
conbuildmat.2013.07.016.

[14] M. Juenger, H. Jennings, Effects of high alkalinity on cement pastes, ACI

Materials Journal 98a (2001) 251–255.

[15] F. Agrela, M. Cabrera, A. Galvı́n, A. Barbudo, A. Ramirez, Influence of the

340 sulphate content of recycled aggregates on the properties of cement-treated
granular materials using sulphate-resistant portland cement, Construction
and Building Materials 68 (2014) 127–134. doi:10.1016/[Link].
2014.06.045.

[16] O. Batic, C. Milanesia, P. Maizab, S. Marfilb, Secondary ettringite for-

345 mation in concrete subjected to different curing conditions, Cement and
Concrete Research 30 (2000) 1407–1412. doi:10.1016/0003-4916(63)
90068-X.

[17] D. Breysse, Deterioration processes in reinforced concrete: an overview,

in: C. Maierhofer, H. Reinhardt, G. Dobmann (Eds.), Non-Destructive
350 Evaluation of Reinforced Concrete Structures, Woodhead Publishing, 2010,
Ch. 3, pp. 28–56.

[18] M. Zhang, J. Chen, Y. Lv, D. Wang, J. Ye, Study on the expansion of

concrete under attack of sulfate and sulfate–chloride ions, Construction
and Building Materials 39 (2013) 26–32. doi:10.1016/[Link].
355 2012.05.003.

[19] J. De Souza, M. Medeiros, J. Hoppe Filho, Evaluation of external sulfate

attack (na2so4 and mgso4): Portland cement mortars containing fillers,
Revista IBRACON de Estruturas e Materiais 13 (2020) 644–655. doi:
10.1590/s1983-41952020000300013.

26
360 [20] E. C. for Standardization, EN 206:2014 ’Concrete: Specification, perfor-
mance, production and conformity: Annex E.3 on Recycled Aggregates’
(2014).

[21] P. Rougeau, L. Schmitt, J. Nai-Nhu, A. Djerbi, M. Saillio, E. Ghorbel, J.-

M. Mechling, D. Bulteel, M. Cyr, A. Lecomte, N. Leklou, R. Trauchessec,
365 I. Moulin, T. Lenormand, O. Amiri, Propriétés liées à la durabilité, in:
F. D. Larrard, H. Colina (Eds.), Le béton recyclé, Iffstar, 2018.

[22] E. Irassar, Sulfate attack on cementitious materials containing limestone

filler — a review, Cement and Concrete Research 39 (2009) 241–254. doi:
10.1016/[Link].2008.11.007.

370 [23] M. Rahman, M. Bassuoni, Thaumasite sulfate attack on concrete: Mecha-

nisms, influential factors and mitigation, Construction and Building Mate-
rials 73 (2014) 652–662. doi:10.1016/[Link].2014.09.034.

[24] A. Neville, The confused world of sulfate attack on concrete, Cement

and Concrete Research 34 (2004) 1275–1296. doi:10.1016/[Link].
375 2004.04.004.

[25] Z. Zhao, S. Rémond, D. Damidot, W. Xu, Influence of hardened cement

paste content on the water absorption of fine recycled concrete aggregates,
Journal of Sustainable Cement-Based Materials 2 (2013) 186–203. doi:
10.1080/21650373.2013.812942.

380 [26] E. C. for Standardization, EN 1097-6 ’Tests for mechanical and physical
properties of aggregates. Part 6: Determination of particle density and
water absorption’ (2001).

[27] IFSTTAR, Test method no. 78:2011. Tests on aggregates for concrete: mea-
surement of total water absorption by a crushed sand (2011).

385 [28] T. Le, S. Rémond, G. Le Saout, E. Garcia-Diaz, Fresh behavior of mortar

based on recycled sand: Influence of moisture condition, Construction and

27
 Building Materials 106 (2016) 35–42. doi:10.1016/[Link].2015.
12.071.

[29] E. C. for Standardization, EN 196-1 ’Methods of testing cement. Part 1:

390 Determination of strength’ (2016).

[30] H. Taylor, C. Famy, K. Scrivener, Delayed ettringite formation, Cement

and Concrete Research 31 (2001) 683–693.

[31] K. Kjellsen, A. Monsoy, K. Isachsen, R. Detwiler, Preparation of flat-

polished specimens for sem-backscattered electron imaging and x-ray mi-
395 croanalysis - importance of epoxy impregnation, Cement and Concrete Re-
search 33 (2003) 611–616. doi:10.1016/S0008-8846(02)01029-3.

[32] M. Bouarroudj, S. Remond, F. Michel, Z. Zhao, D. Bulteel, L. Courard, Use

of a reference limestone fine aggregate to study the fresh and hard behavior
of mortar made with recycled fine aggregate, Materials and Structures 52
400 (2019) 1–14. doi:10.1617/s11527-019-1325-1.

[33] I. Mehdipour, K. Khayat, Understanding the role of particle packing char-

acteristics in rheophysical properties of cementitious suspensions: A lit-
erature review, Construction and Building Materials 161 (2018) 340–353.
doi:10.1016/[Link].2017.11.147.

405 [34] L. Courard, F. Michel, M. Martin, The evaluation of the surface free energy
of liquids and solids in concrete technology, Construction and Building
Materials 25 (2011) 260–266. doi:10.1016/[Link].2010.06.030.

[35] R. Komatsu, N. Mizukoshi, K. Makida, K. Tsukamoto, In-situ observation

of ettringite crystals, Journal of Crystal Growth 311 (2009) 1005–1008.
410 doi:10.1016/[Link].2008.09.124.

[36] C. Hargis, A. Telesca, P. Monteiro, Calcium sulfoaluminate (ye’elimite)

hydration in the presence of gypsum, calcite, and vaterite, Cement and
Concrete Research 65 (2014) 15–20. doi:10.1016/[Link].2014.
07.004.

28