Construction and Building Materials

Thermal reactivation of hydrated cement pastes: properties and impact on cement

hydration
--Manuscript Draft--

Manuscript Number: CONBUILDMAT-D-22-04410

Article Type: Research Paper

Keywords: Recycled cement paste; Thermal activation; Chemical reactivity; Supplementary

cementitious material; Blended cement; Hydration

Corresponding Author: Ruben Snellings, PhD

VITO
Mol, Belgium

First Author: Asghar Gholizadeh-Vayghan

Order of Authors: Asghar Gholizadeh-Vayghan

Guillermo Meza Hernandez

Felicite Kingne Kingne

Jun Gu

Nicole Dilissen

Michael El Kadi

Tine Tysmans

Jef Vleugels

Hubert Rahier

Ruben Snellings, PhD

Abstract: In this research, the physico‑chemical properties and cementitious performance of

thermally activated cement pastes (referred to as DCPs) are investigated. Hydrated
pastes prepared from Portland cement (CEM I) and slag blended cement (CEM III) are
subjected to four different thermal treatments: 350 °C for 2 h, 550 °C for 2 h, 550 °C for
24 h, and 750 °C for 2 h, and milled to cement fineness. Particle characteristics, phase
composition, thermogravimetric response, and chemical reactivity of the DCPs are
determined and explored in relation to the original composition and the thermal
treatment conditions. The influences of incorporating the above materials as
supplementary cementitious materials (SCMs) are also investigated through a
systematic hydration study. Direct correlations were observed between the
temperature and duration of the thermal treatment and the grindability and BET specific
surface area of the DCP, the degree of recrystallisation of C 2 S phases, and the
chemical reactivity. In contrast, the mechanical strength results for the blended
cements indicate that thermal treatment at 350 °C for 2 h provides a better
performance.

Suggested Reviewers: Miruna Florea

University of Technology Eindhoven: Technische Universiteit Eindhoven
m.v.a.florea@tue.nl
Expert in recycling of waste materials in cement and concrete. She has authored
several impactful papers on recycling of cement and concrete.

Francesco Di Maio
TU Delft: Technische Universiteit Delft
D.DiMaio@tudelft.nl
Expert in upcycling of C&DW, in particular focusing on concrete demolition waste.

Valdecir Quarcioni
IPT: Instituto de Pesquisas Tecnologicas
quarciva@ipt.br

Ricardo Serpell

Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
 EPFL-ENAC: Ecole Polytechnique Federale de Lausanne Faculte Environnement
Naturel Architectural et Construit
ricardo.serpell@epfl.ch
Expert who has published several papers on thermal recycling of cement.

Kathy Bru
BRGM: Bureau de Recherches Geologiques et Minieres
k.bru@brgm.fr
Recycling expert with several publications on concrete recycling using thermal
treatment

Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Cover Letter

May 21st, 2019

Subject: Research paper titled:

“The effects of carbonation conditions on the physical properties of recycled
concrete aggregates”
By: Asghar Gholizadeh-Vayghan and Ruben Snellings

Dear Editor of Construction and Building Materials;

Attached please find our research paper titled “The effects of carbonation conditions on the physical properties
of recycled concrete aggregates” that is submitted for peer review and potential publication in Construction
and Building Materials. In this paper, an experimental program has been undertaken with the
objective of investigating and improving the main drawbacks of recycled concrete aggregate
(namely, high water absorption and inferior surface wear resistance) by means of controlled
carbonation. While the subject of recycled concrete aggregate (RCA) carbonation is gaining more
and more attention and, one after the other, researchers across the globe certify its great merit, a
systematic research characterizing the effects of different carbonation parameters can barely be
found in the literature. The authors believe that the carbonation parameters and pre-treatments
investigated in this research are unprecedented. Thanks to the carbonation facilities available at
VITO, the effects of RCA moisture condition, carbonation environment’s relative humidity,
carbonation pressure, temperature and duration are investigated in wide ranges and significant
improvements in water absorption and wear resistance have been found. Lime bathing of RCA
with the aim of further enhancing the carbonation outcome is also investigated for the first time
with satisfying results. In addition, a closed-form analytical formula relating porosity of RCA to its
density and water absorption is also derived and the carbonation conditions leading to the highest
drop in RCA porosity are identified. The authors believe that this paper can serve as a good
resource for the researchers who intend to investigate the subject of carbonation as it sheds light
on the roles and interactions of different carbonation conditions and introduces the concept of
lime bathing for further improving the carbonation outcome.
We would like to thank you and the reviewers in advance for evaluating this manuscript. Please
refer any questions to the corresponding author:
Asghar Gholizadeh-Vayghan
Sustainable Materials Management
VITO (Flemish Institute for Technological Research)
2400 Mol, Belgium
(Phone) +32 1 433-5757
(Email) asghar.gholizadehvayghan@vito.be

With best regards,

Asghar Gholizadeh-Vayghan
Sustainable Materials Management
VITO (Flemish Institute for Technological Research)
Manuscript (For revision, remove the original version and upload Click here to view linked References
clean version of revised paper)

1
2
3

Thermal reactivation of hydrated cement

4
5 1
6
7
8
2 pastes: properties and impact on cement
9
10 3 hydration
11 4
12
13 5 Asghar Gholizadeh-Vayghan1, Guillermo Meza Hernandez2, Felicite Kingne Kingne2, Jun
14
15 6 Gu2, Nicole Dilissen2,4, Michael El Kadi3, Tine Tysmans3, Jef Vleugels4, Hubert Rahier2,
16
17 7 Ruben Snellings1,*
18
19 8 1 Sustainable Materials, Vlaamse Instelling voor Technologisch Onderzoek, 200 Boeretang, Mol, Belgium
20 9 2 Physical Chemistry and Polymer Science, Department MACH, Vrije Universiteit Brussel, 1050 Brussels, Belgium
21
22 10 3 Department of Mechanics of Materials and Constructions (MeMC), Vrije Universiteit Brussel, 1050 Brussels, Belgium
23 11 4 Department of Materials Engineering, KU Leuven, Kasteelpark Arenberg 44, 3001 Heverlee, Belgium
24
25 12 * Correspondence: ruben.snellings@vito.be; Tel.: +32 14 33 56 26
26 13
27
28
29
30
14 Abstract
31
32
15 In this research, the physico-chemical properties and cementitious performance of thermally
33
34 16 activated cement pastes (referred to as DCPs) are investigated. Hydrated pastes prepared from
35
36 17 Portland cement (CEM I) and slag blended cement (CEM III) are subjected to four different thermal
37
18 treatments: 350 °C for 2 h, 550 °C for 2 h, 550 °C for 24 h, and 750 °C for 2 h, and milled to
38
39 19 cement fineness. Particle characteristics, phase composition, thermogravimetric response, and
40
41 20 chemical reactivity of the DCPs are determined and explored in relation to the original composition
42
43
21 and the thermal treatment conditions. The influences of incorporating the above materials as
44 22 supplementary cementitious materials (SCMs) are also investigated through a systematic
45
46 23 hydration study. Direct correlations were observed between the temperature and duration of the
47
48 24 thermal treatment and the grindability and BET specific surface area of the DCP, the degree of
49 25 recrystallisation of C2S phases, and the chemical reactivity. In contrast, the mechanical strength
50
51 26 results for the blended cements indicate that thermal treatment at 350 °C for 2 h provides a better
52
53 27 performance.
54
55 28
56
57 29 Keywords: Recycled cement paste; Thermal activation; Chemical reactivity; Supplementary
58
59 30 cementitious material; Blended cement; Hydration
60
61
62
1
63
64
65
 1
2
3
4
5 31 1. Introduction
6
7
8 32 The production of Portland cement accounts for 5 – 8 % of the total anthropogenic CO2 emission
9 33 and consumes large amounts of natural resources [1,2]. On average, 842 kg of CO2 is generated
10
11 34 for producing one tonne of clinker [3]. The reduction in the use of clinker in new cement via
12
13 35 supplementary cementitious materials (SCMs) is known to be one of the most effective ways of
14 36 lowering the CO2 footprint of the cement and concrete industries. An interesting way for reducing
15
16 37 the clinker use is its partial replacement by recycled concrete fines (RCF) in new cement. RCF as
17
18 38 one of the main concrete recycling products is usually not recommended for use as fine aggregate
19
39 in new concrete and has thus far received relatively limited attention [4]. A few different venues
20
21 40 have been identified for the utilisation of RCF as a cementitious material. The use of RCF as a
22
23 41 raw feed for the production of clinker is one such approach. The main challenges of this approach
24
42 that limit high level incorporation are the high SiO2 content compared to Portland clinker raw meal
25
26 43 and the variations in the chemical composition of RCF. The compositional variability reflects the
27
28 44 wide ranges in the chemical composition of cements and concrete aggregates, and the varying
29
45 ability of concrete recycling processes to liberate aggregates from hardened cement paste [5]. For
30
31 46 instance, Gastaldi et al. [6] investigated the use of recycled concrete powder as a raw feed for
32
33 47 clinker production, and concluded that due to its composition, the RCF content has to be limited
34
35
48 to 20 – 40 %. The direct use of milled RCF as SCM has also been studied in the past [7,8,9]. This
36 49 approach appears to offer more robustness for implementation on industrial scale compared to
37
38 50 the former approach. As the cement in the RCF has largely reacted, the RCF mainly acts as a
39
40
51 filler material when directly used [10,11]. Thermal treatment is reported to partially reactivate the
41 52 hydrated cement paste contained in RCF giving some cementitious properties to this material
42
43 53 [12,13,14,15,16].
44
45 54 Thermal treatment leads first to the gradual dehydration of different hydrates in the cement paste
46
47 55 and, eventually, to the formation of new high-temperature phases with potential hydraulic activity.
48
49 56 The degree of dehydration of each phase and the new phases formed depend on the treatment
50 57 conditions such as temperature and time. The heating rate, oven/furnace type, particle size, and
51
52 58 cooling rate also affect the extent of dehydration and the phases formed [17,18]. Additionally, the
53
54 59 type and composition of the initial cement play a key role in the dehydration behaviour and the
55 60 phase composition of the output material [19]. A review by Carriço et al. [18] concluded that the
56
57 61 literature shows some ambiguity regarding the phases formed during thermal dehydration of
58
59 62 Portland cement. Formation of α’- β- and γ-C2S has been reported at temperatures ranging from
60 63 600 °C to 900 °C by different authors. Below 600 °C, a continuous dehydration of the interlayer
61
62
2
63
64
65
 1
2
3
4 64 water and gel water in C-S-H is reported between 50 °C and 600 °C [20,21]. At 750 °C, XRD
5
6 65 peaks corresponding to poorly-crystallised wollastonite (CS) and larnite (β-C2S) with low reactivity
7
8 66 have been reported by Wang et al. [12]. Serpell and Zunino [17] carried out a study on the
9 67 formation of different C2S polymorphs during dehydration of Portland cement in the 600 – 800 °C
10
11 68 and reported a decreasing α’- and γ-C2S content with increase in temperature at the expense of
12
13 69 an increasing β-C2S content.
14
15 70 It is important to determine the optimum dehydration temperature to achieve the highest degree
16
17 71 of reactivity and strength performance in the dehydrated cement. Recent studies have adopted
18 72 dehydration temperatures of 300 – 800 °C for the decomposition of C-S-H and dehydroxylation of
19
20 73 portlandite [22,23,24,25]. Wang et al. [12] found the highest compressive strength at a dehydration
21
22 74 temperature of ~450 °C of Portland cement. Angulo et al. [26] reported a comparable optimum
23 75 dehydration temperature (500 °C). On the other hand, Shui et al. [27] studied the effects of thermal
24
25 76 dehydration of hydrated cement paste on its cementitious characteristics. They found a direct
26
27 77 correlation between increasing the dehydration temperature and the water demand and rate of
28 78 setting of blended cements incorporating DCP. Improvement in the cementitious performance of
29
30 79 DCP up to a dehydration temperature of 800 °C was reported, followed by a decline above 800 °C
31
32 80 [27].
33
34 81 Florea [24] investigated the effects of using thermally activated RCF as SCM on the strength of
35
36 82 mortar mixtures. It was concluded that the very fine fraction of recycled concrete treated at 800 °C
37
83 can replace Portland cement up to 20 wt% without significant loss of strength. Up to 20 % increase
38
39 84 in the compressive strength was reported when replacing 10 wt% of blended cement with high
40
41 85 slag content by thermally dehydrated RCF. Wu et al. [28] investigated the effects of thermal
42
86 treatment on the properties and cementitious performance of waste concrete, mortar, and paste
43
44 87 powder. They observed a similar behaviour for all three materials concluding improvement in the
45
46 88 performance of thermally treated waste in comparison with untreated waste when the dehydration
47
48
89 temperature ranged from 600 °C to 1000 °C. An improvement in the 28-d strength activity index
49 90 of waste paste powder from 70 % to 90 % was reported upon application of a thermal dehydration
50
51 91 at 600 °C (when compared to the untreated powder). Inferior performance (compared to untreated
52
53
92 powder) was however observed when the dehydration temperature was as high as 1200 °C.
54
55 93 There seems to be no consensus regarding the proper dehydration conditions when it comes to
56
57 94 the strength performance and the phases formed during dehydration and rehydration. While the
58 95 literature is generally in agreement regarding the high early heat of hydration of thermally activated
59
60 96 cement, the impacts of the dehydration conditions on the extent of heat evolution, rehydration
61
62
3
63
64
65
 1
2
3
4 97 reactions, and how they influence cement hydration reactions are still controversial [18]. The
5
6 98 temperature dependency of the rehydration reactions and products and how they translate to
7
8 99 strength development also need to be more systematically investigated.
9
10 100 The objective of this paper is to obtain a deeper understanding of the influence of thermal
11
12 101 treatment on the properties and cementitious performance of hydrated cement paste (as a proxy
13 102 to recycled concrete fines). To this end, a systematic study is carried out on the effects of different
14
15 103 thermal treatments on the physical, chemical, and mineralogical characteristics of hydrated
16
17 104 cement paste (produced from CEM I and CEM III). The impacts of using such materials as SCM
18 105 on the hydration process and strength development of cement mixtures is also investigated
19
20 106 through a comprehensive hydration study, and the underlying root causes for the observed effects
21
22 107 are explored and interpreted. The cementitious performance of dehydrated cement paste (DCP)
23 108 produced under different conditions are also compared and the most suitable treatment conditions
24
25 109 are identified for both CEM I and CEM III cases.
26
27
28 110 2. Materials and methods
29
30
31 111 2.1. Materials selection and preparation
32
33
34 112 CEM I and CEM III (prepared by blending 300 kg of CEM I with 220 kg of ground granulated blast
35
113 furnace slag – GGBFS) were used for producing the initial hydrated cement pastes (later to be
36
37 114 dehydrated and studied). Table 1 shows the chemical, physical, and mineralogical properties of
38
39 115 the CEM I, GGBFS, and CEM III used in this research. A fine quartz filler was also used as a
40
116 reference inert material to be compared with the thermally activated dehydrated cement paste
41
42 117 (DCP) fines. Figure 1 shows the particle size distributions and the characteristic particle sizes of
43
44 118 the above materials. Particle size distribution was analysed using a Horiba laser diffraction particle
45
119 size analyser in isopropyl alcohol (IPA).
46
47
48 120 For the preparation of the dehydrated cement fines, cement pastes with a water-to-binder ratio of
49
50 121 0.5 were first prepared from both CEM I and CEM III. All pastes were cured in saturated lime water
51 122 for 28 days at 21 ± 2 °C. Next, the hydrated cement pastes were crushed below 20 mm and dried
52
53 123 at 40 °C for 48 h. The dried HCPs were then inserted into a muffle furnace and heated at a rate
54
55 124 of 10 °C/min up to maximum temperatures of 1) 350 °C for 2 h, 2) 550 °C for 2 h, 3) 550 °C for
56 125 24 h, and 4) 750 °C for 2 h. Upon completion of the thermal cycle, the materials were removed
57
58 126 from the furnace and allowed to cool down to ~170 °C (temperature monitored via an IR-pyrometer)
59
60 127 and subsequently transferred to a desiccator. A planetary ball mill with a tungsten carbide jar (500
61
62
4
63
64
65
 1
2
3
4 128 mL) and 13×2-mm tungsten carbide milling balls were used for milling the materials. 60 grams of
5
6 129 dried DCP (added with 5 wt% of isopropyl alcohol as the grinding aid) was milled at 350 rpm for
7
8 130 12 min (with 1-min inversion intervals) in each cycle. The milled portions were then dried at 60 °C
9 131 for 24 h, fully blended and characterised for their physicochemical properties, and used for
10
11 132 preparing new blended cement mixtures for further studies. Table 2 shows an overview of the
12
13 133 dehydrated cement pastes produced for this study along with their labels.
14
15
16
134 2.2. Characterisation methods
17
18 135 2.2.1. Materials characterisation
19
20 136 The particle size distributions of the milled DCPs were analysed using a Horiba laser diffraction
21 137 particle size analyser in IPA. Ultrasonication was applied to separate the particles in another IPA
22
23 138 solution prior to characterisation. The reported results are the average of three measurements.
24
25 139 The fineness of the DCP powders was measured via the Brunauer-Emmett-Teller (BET) nitrogen
26 140 gas sorption method using a Quantachrome Autosorb iQ Series instrument.
27
28
29 141 The DCPs were subjected to scanning electron microscopy in order to explore the effects of the
30 142 thermal treatment on the morphology of the DCP particles after milling. Secondary electron images
31
32 143 were acquired using a FEI FEG Nova NanoSEM 450 operated at an electron acceleration voltage
33
34 144 of 5 kV.
35
36 145 The bound water, portlandite, and carbonate contents of the DCPs were measured via
37
38 146 thermogravimetry in order to observe the effects of dehydration conditions. Samples were heated
39
40
147 at 10 °C/min from 30 °C up to 1000 °C under nitrogen atmosphere at a 50 ml/min N2 flow rate.
41 148 The bound water was determined by measuring the mass change between 50 °C and 550 °C, and
42
43 149 the portlandite and carbonate contents were estimated respectively as the mass loss in the 400 –
44
45
150 500 °C and 600 – 790 °C ranges using the tangential method [29].
46
47 151 The phase composition of the DCPs was determined via X-ray diffraction. XRD data were
48
49 152 collected using an Empyrean diffractometer (Panalytical) equipped with a CoKα tube operated at
50 153 40 kV and 45 mA. Diffraction scans were recorded from 5 to 110 °2θ, with step size 0.013 °2θ,
51
52 154 and a measurement time of 50 s per step. Rietveld analysis was performed using HighScore Plus
53
54 155 software (version 4.6a) and the external standard approach was used to determine the amorphous
55 156 phase content (rutile; Kronos 2300 TiO2 external standard calibrated against NIST SRM 676a α-
56
57 157 Al2O3). The starting crystal structure data were taken from Snellings [30], and the chemical
58
59 158 compositions of the dehydrated materials (calculated using Table 1 data) were used to estimate
60 159 the mass attenuation coefficients needed for the absorption correction.
61
62
5
63
64
65
 1
2
3
4 160 The chemical reactivity of the DCP powders was also determined using the R3 (Rapid, Relevant,
5
6 161 Reliable) reactivity test according to ASTM C1897. Synthetic pastes were produced incorporating
7
8 162 the slags and Ca(OH)2, H2O, with additions of CaCO3, K2SO4, and KOH to simulate the reaction
9 163 medium of a hydrating Portland cement [31]. The designated amounts of DCP, calcium hydroxide,
10
11 164 and calcite were weighed on weighing papers, homogenised, and stored at 40 °C. The dry mixture
12
13 165 of the main ingredients was then mixed with the potassium solution at 1600 ± 50 rpm for 2 min
14 166 using a high shear blender, and the time of start of mixing was recorded as the ‘time zero’. A 15.0
15
16 167 g freshly made paste specimen was prepared and the cumulative heat was measured and
17
18 168 recorded following ASTM C1897 – Method A procedure.
19
20 169 2.2.2. Hydration study
21
22 170 A hydration study was carried out on the DCPs in comparison with the selected inert filler as
23
24
171 follows. Cement pastes incorporating DCPs were tested for their phase evolution (via X-ray
25 172 diffractometry), heat evolution (via calorimetry), and thermogravimetric response (TGA). The
26
27 173 control mixture was produced with 100 % CEM I or CEM III with a water-to-binder ratio of 0.40.
28
29
174 The blended mixtures were produced by replacing 30 wt% of the cement with either DCP of the
30 175 same type. The mixtures are labelled based on the type of the cement used and the replacement
31
32 176 type. I-350/2 for instance refers to the paste made with CEM I and 30 wt% replacement with
33
34 177 DCI/350°C/2h. A mixture incorporating 30 wt% cement replacement with quartz powder (I-QP and
35 178 III-QP) as inert filler was also produced to account for the filler effect of the DCPs. All pastes were
36
37 179 prepared by first dry-mixing the powders for 60 min using a Turbula multidirectional mixer. The
38
39 180 dry mixture of the main ingredients was then mixed with water at 1600 ± 50 rpm for 2 min using a
40 181 high shear blender so that a homogeneous paste was achieved.
41
42
43 182 All prepared cement pastes were exposed to extensive XRD investigations as follows. The early-
44 183 age behaviour of all CEM I paste series were explored using in-situ XRD in 15-min time intervals
45
46 184 (started within 15 min after mixing) up to 24 h. The fresh pastes were cast into XRD sample holders
47
48 185 and covered with a semi-spherical X-ray transmissible polymer dome to avoid surface evaporation.
49 186 The closed sample holder was placed inside the XRD measurement position for continuous data
50
51 187 collection during early hydration. Diffraction scans were recorded from 5 to 55 °2θ under the
52
53 188 conditions mentioned above. This was done to monitor the early-age formation of hydrated phases
54 189 and allow the comparison of the effects of incorporating different DCPs in the cement pastes. The
55
56 190 longer term evolution of hydration phases was also studied via ex-situ XRD from 1 d to 91 d for
57
58 191 both CEM I and CEM III pastes (in the 5 to 90 °2θ range). To this end, the prepared cement pastes
59 192 were cast into 12-ml plastic moulds and stored at 21 ± 2 °C for 24 h. Next, the hardened pastes
60
61
62
6
63
64
65
 1
2
3
4 193 were demoulded and stored inside slightly larger plastic bottles under a thin film of water until the
5
6 194 testing age. At each testing age, the hardened pastes were taken out, a slice was cut, and the
7
8 195 surface of the freshly cut specimen was smoothened using a 400 grit sandpaper and then flushed
9 196 with water. The extra water film was removed from the surface and the specimen was directly
10
11 197 subjected to the test under the same conditions as mentioned above.
12
13 198 The thermogravimetric response of the hydrated cement pastes was measured in order to
14
15 199 determine and compare the bound water and portlandite contents of the control and blended
16
17 200 pastes over time. Representative hydration-stopped samples were heated at 10 °C/min from 30 °C
18 201 up to 1000 °C under nitrogen atmosphere at 50 ml/min flow rate. The bound water was determined
19
20 202 by measuring the mass change between 50 °C and 550 °C, and the portlandite content was
21
22 203 estimated as the mass loss in the 400 – 500 °C range using the tangential method.
23
24 204 A sample of each paste mixture was prepared to determine the heat of hydration using isothermal
25
26 205 calorimetry. The heat evolution was then monitored at 20 °C using a TAM Air calorimeter up to
27 206 7 days according to the Method B of ASTM C1702.
28
29
30 207 2.2.3. Mechanical performance
31
32 208 To assess the cementitious performance of the produced DCPs, standard mortar mixtures
33 209 conforming to EN 196-1 specifications were produced and tested at different ages for their
34
35 210 compressive strength. A water-to-binder ratio of 0.5 and a sand-to-binder ratio of 3.0 were used
36
37 211 for all mixtures. The binder in the control mixtures was composed solely of either CEM I or CEM III.
38 212 In the case of test (blended) mixtures, 30 wt% replacement with each DCP was made in order to
39
40 213 observe the influence of such replacements on the water demand and strength development of
41
42 214 the mortars. For each CEM type, DCPs originally produced from the same CEM type were used
43 215 as partial replacement. The mixture compositions are the same as those of the cement pastes.
44
45 216 For each composition, polycarboxylate-based high-range water reducing admixture (also referred
46
47 217 to as superplasticizer: SP) was added to the mixing water in order to maintain the mortar flow in
48 218 the 175 ± 15 mm range. The SP addition ranged between 0.04 and 0.22 wt.%, increasing with
49
50 219 DCP treatment temperature. This approach was taken to eliminate the effects of DCPs on the
51
52 220 consistence and compactability of the mixtures and their subsequent effects on the mechanical
53 221 properties of the hardened mortars. Mixtures with a flowability outside the designated ranges were
54
55 222 disposed and new mixtures with adjusted superplasticizer contents were produced. Once the
56
57 223 flowability in the desired range was obtained, the mixture was cast into three 40 × 40 × 160 mm 3
58 224 prismatic moulds, covered with plastic sheets and stored at 20 ± 2 °C for 24 h. Each mixture was
59
60
61
62
7
63
64
65
 1
2
3
4 225 produced four times (one batch for each testing age: 1, 7, 28, and 91 d). The hardened specimens
5
6 226 were removed and cured in a climate chamber at 20 ± 1 °C and RH > 95 % until the age of testing.
7
8
9 227 3. Results and discussion
10
11
12 228 3.1. Material characterisation
13
14
15 229 Figure 2 shows the particle size analysis results of the dehydrated cement paste powders along
16 230 with the summarised characteristic particle sizes. It could be observed that the milled materials
17
18 231 are in all accounts of similar size to the respective starting cement or finer (compare the
19
20 232 characteristic particles sizes shown in Figure 2 with respective values in Figure 1). Comparing the
21 233 particle size of DCPs produced under different treatment conditions, it could be concluded that an
22
23 234 increase in the dehydration temperature and time generally leads to a higher grindability (note the
24
25 235 order of the PSD curves in Figure 2 with respect to the colour codes). For both CEM I and CEM III,
26 236 the DCPs produced at 350 °C have the largest average particle size and D90 (see the blue curves),
27
28 237 whereas DC/550°C/24h and DC/750°C/2h show the smallest values (see the orange and red
29
30 238 curves, respectively).
31
32 239 The BET specific surface area values of the DCI and DCIII powders are shown in Figure 3. For
33
34 240 DCI, the BET surface monotonically increases with increase in the dehydration time and
35
241 temperature. Similar behaviour was also observed in the case of DCIII with the exception of
36
37 242 DCIII/750°C/2h. A considerable decline in the BET surface area of DCIII is observed when the
38
39 243 dehydration temperature exceeds 550 °C. It could be argued that as the dehydration temperature
40
244 increases, the C-S-H continues to dehydrate, shrinks and becomes more porous [32]. However,
41
42 245 once the dehydration temperature exceeds a certain level, a major part of C-S-H recrystallises
43
44 246 into C2S which entails a collapse in the porous structure by nucleation and growth of C2S crystals.
45
247 An important finding from the BET results is that the DCP (and consequently RCF) has a very high
46
47 248 specific surface area as a potential SCM. The BET surface areas observed in the DCPs range
48
49 249 from 15 m2/g to 23 m2/g, which is more than one order of magnitude higher than those of the
50
51
250 starting materials (i.e., CEM I and GGBFS: 1.76 and 1.35 m2/g, respectively). Such high porosity
52 251 could be considered a drawback for DCP/RCF as it increases the water demand of the binder. A
53
54 252 closer look at the microstructure and morphology of the DCPs produced under different thermal
55
56
253 conditions is provided in Figure 4 and Figure 5. It is observed that the DCP grains are irregularly
57 254 shaped agglomerates consisting of flake to globular sub-micron sized particles. The DCPs
58
59 255 produced at lower temperatures tend to a more flake-like texture while those dehydrated at higher
60
61
62
8
63
64
65
 1
2
3
4 256 temperatures are more globular. Some agglomerates showed clear internal microporosity (e.g. in
5
6 257 the case of DCIII/550°C/2h (see Figure 5d), however this type of grains represented only a minority
7
8 258 and was found for different treatment conditions. Therefore, this texture-type is likely inherited
9 259 from more porous zones in the original hydrated cement paste.
10
11
12 260 Figure 6a and b show the TGA responses of the DCI and DCIII materials, respectively. Using the
13 261 TGA curves, the amount of bound water, portlandite content, and carbonate content of DCI and
14
15 262 DCIII were calculated and plotted in Figure 7a and b, respectively. The first remarkable difference
16
17 263 is found in the bound water contents of DCPs prepared at 350 °C compared to those prepared at
18 264 higher temperatures. With increase in the dehydration temperature and time, the amounts of
19
20 265 bound water, and portlandite content decrease in all cases. It is interesting to note that the
21
22 266 dehydration never led to a complete absence of portlandite, as would be expected for treatment
23 267 temperatures of 550 and 750 °C. A shift to lower temperatures is observed for the dehydroxylation
24
25 268 event of portlandite for 550°C/24h and 750°C/2h materials. This indicates a change in portlandite
26
27 269 properties such as crystal size, and thus may indicate a different origin, for instance rapid hydration
28 270 of free lime is known to result in small submicrometer crystal sizes [33]. Free lime formed by
29
30 271 thermal decomposition of portlandite in cement pastes is known to be reactive and convert readily
31
32 272 upon brief exposure to ambient air to either calcium hydroxide or calcium carbonates [34]. Minor
33 273 mass losses were also observed at temperatures exceeding 600 °C, which indicates occasional
34
35 274 formation of carbonates as a result of ambient air exposure during the handling steps.
36
37
275 The phase composition of the DCPs was studied in detail via XRD. The summarised Rietveld
38
39 276 analysis results are presented in Figure 8 a) and b). It can be clearly observed in both figures that
40
41 277 the thermal treatment causes major drops in the amorphous content of the DCPs. This is to the
42
278 most part due to the formation of C2S phases, and at 750 °C/2h also C4AF and C3A. Various C2S
43
44 279 phases could be refined, β-C2S is present after all treatments and increases in content with
45
46 280 temperature. Below 550 °C, β-C2S is interpreted to be a residual clinker phase deriving from the
47
48
281 initial CEM I and CEM III, at 750 °C it appears to form significantly, despite being, as a pure phase,
49 282 thermodynamically unstable. The more stable γ -C2S is the dominant C2S phase at 550 °C, yet
50
51 283 becomes subordinate to β-C2S at 750 °C. No such phase was detected for DCP produced at
52
53
284 350 °C. While the α-C2S content appears to be in the error margin of ±2 wt% [35], it shows a
54 285 notable increase at 750 °C in DCI. Although α-C2S is a high-temperature phase, its formation
55
56 286 during HCP dehydration at low temperatures has been reported by other researchers as well
57
58 287 [17,19]. The formation of high-temperature C2S polymorphs may be unexpected. In case of α-C2S,
59 288 the phase content refinement could be affected by misfitting of other phases of low crystallinity.
60
61
62
9
63
64
65
 1
2
3
4 289 However, β-C2S formation was clearly distinguished and identified at temperatures well below its
5
6 290 established stability field. This apparent disparity may be explained by the observation that, in the
7
8 291 absence of an efficient reaction medium such as a solution or a melt phase, the thermally induced
9 292 solid-state conversions may be controlled by kinetic factors such as precursor or templating effects,
10
11 293 such as seen in the transformation of meta-clays from clay minerals [36]. Such effects have not
12
13 294 been explored extensively for C-S-H in actual cement pastes as yet. In addition, the presence of
14 295 impurities and defects in C-S-H may induce the formation of metastable phases and preserve
15
16 296 these during cooling [37]. The C3S identified in the DCPs is interpreted to be a remainder of the
17
18 297 initial cements. The variations in the C3S content are within the error range of the analysis. Some
19 298 increase in the C3A and C4AF contents was observed at 750 °C, which is indicative of initial
20
21 299 recrystallisation at such temperature. Moreover, it was noticed that the AFm and AFt phases were
22
23 300 decomposed to the most part during thermal treatment at temperatures as low as 350 °C. No
24 301 traces of such phases were observed at higher temperatures, where the presence of anhydrite
25
26 302 indicates that such phases were at least partially decomposed and not merely dehydrated.
27
28 303 The changes in the phase compositions of the DCPs was observed to entail variations in their
29
30 304 chemical reactivity. Depending on the treatment conditions, DCI and DCIII showed 7-d R3
31
32 305 reactivities in the ranges of 125 – 210 J/g and 225 – 283 J/g, respectively (see Table 3). The R3
33 306 heat release and heat flow results of DCI and DCIII materials are shown in Figure 9a and b. Notice
34
35 307 the differences in the heat release values of DCPs produced under different conditions.
36
37 308 Dehydration at 350 °C yields the least reactive DCP in both cases. DCI produced at such
38 309 temperature shows 45 % lower reactivity compared to the DCPs obtained at higher temperatures,
39
40 310 and DCIII shows 25 % lower reactivity.
41
42
311 DCPs produced at 550 and 750 °C show more or less similar reactivities. An attempt was made
43
44 312 to find correlations between the phase composition of the DCPs and their R3 heat release so as
45
46 313 to determine which phases potentially contribute to the generation of heat. The analyses indicate
47
48
314 that while all high-temperature phases have positive effects on the R3 heat release, no clear
49 315 statistically significant correlation could be detected between such phases and the 1-, 3-, and 7-d
50
51 316 R3 heat results. Table 4 lists the linear correlation coefficients between the R3 heat results (along
52
53
317 with the corresponding p-values) and different mineral phases in DCI and DCIII. It is likely that the
54 318 dynamics and the evolving characteristics (composition, crystallinity, defect density, etc.) of all
55
56 319 phases contained in the DCPs are interfering with a straightforward correlation analysis. In
57
58 320 particular, the properties of the large amorphous fraction would be significantly different for the
59 321 different thermal treatments. Nevertheless, it is clearly observed that thermal treatment at 550 °C
60
61
62
10
63
64
65
 1
2
3
4 322 and above lead to an increase in the chemical reactivity of the DCPs. It should also be noted that
5
6 323 the heat release values recorded for DCIII are greater than those of DCI in all cases in spite of
7
8 324 DCIII having lower clinker, β-C2S, and portlandite contents. This is most likely due to the chemical
9 325 reactivity of the residual (i.e., unhydrated) GGBFS in DCIII.
10
11
12 326 3.2. Hydration study
13
14 327 The effect of 30 wt% replacement of the Portland cements (CEM I and CEM III) by their respective
15 328 DCPs on the hydration processes and the strength development was studied in a hydration study
16 329 comprising isothermal calorimetry, thermogravimetric and XRD analysis. This was done with
17 330 respect to a control cement of 100 wt% original CEMI and CEMIII (I-CTRL and III-CTRL,
18
331 respectively) and the non-reactive 30 wt% replacement by quartz powder (I-QP and III-QP,
19
332 respectively). The calorimetry results (cumulative heat and heat flow) of the CEM I pastes are
20
333 plotted in
21
22
334 Figure 10. It is observed that all mixtures containing DCP have a significantly higher heat flow in
23
24 335 the initial hydration period (< 1.5 h) and a shorter induction period compared to I-CTRL and I-QP.
25
26 336 Such increase in the instant heat release and the shortening of the induction period is more
27
337 significant in the case of DCI/550°C/24h and DCI/750°C/24h. Incorporation of all DCPs also leads
28
29 338 to an acceleration of the main hydration peak, as indicated by the steeper slope of the heat flow
30
31 339 curve during the acceleration stage and the earlier occurrence of the maximum heat flow.
32
33 340 With increase in the dehydration temperature and time, the heat flow peak occurs earlier, yet the
34
35 341 area under the first peak decreases. Simultaneously, the main hydration peak is observed to split
36
37 342 in two and a second, broad shoulder or peak is observed at later hours (peak at 12 – 14 h), the
38 343 second peak is clearly evident in I-750/2. Specifically in the case of this paste (i.e., I-750/2) and I-
39
40 344 550/24 to some extent, a significant deviation in the heat release pattern from that of the control
41
42 345 mixture is evident. Judging by the R3 results (Figure 9), it appears that the more reactive the DCP,
43 346 the larger the deviation is the heat flow pattern from that of the control paste. It can therefore be
44
45 347 argued that the rapid hydration of the DCP interferes with the hydration of the Portland clinker
46
47 348 phases of the CEM I. Since the main hydration peak period is generally associated with the
48 349 hydration of C3S [38], the very early-age C3S hydration is investigated later on using in-situ XRD.
49
50
51 350 The incorporation of DCPs in CEM III pastes leads to similar effects as described for CEM I. In
52 351 Figure 11, the heat flow and cumulative heat data are presented for the CEM III pastes. It can be
53
54 352 observed that the split of the main hydration peak into an accelerated early peak and a retarded
55
56 353 second peak is even more pronounced than in the CEM I pastes. In case of the III-550/24 and
57 354 III-750/2, the early peak has shifted to such an extent that it coincides with the initial dissolution
58
59 355 peak. The large peak observed in the case of III-QP is indicative of sulfate depletion (due to the
60
61
62
11
63
64
65
 1
2
3
4 356 dilution effect), which is because CEM III was prepared merely by combining CEM I and GGBFS
5
6 357 without addition of calcium sulfates.
7
8 358 Regarding the main pastes, similar heat evolution patterns as those of DCI-blended pastes could
9
10 359 be identified for the DCIII-blended pastes with a few remarks. The instant heat release observed
11
12 360 in the pastes incorporating DCIII is significantly larger than that of respective CEM I pastes, and
13 361 large heat flows were detected for CEM III pastes incorporating DCIII dehydrated at 550 °C and
14
15 362 above (see Figure 11). The shift of the acceleration peak to the left (compared to the control) is
16
17 363 also more severe when comparing each blended CEM III paste with the respective paste in the
18 364 CEM I series (
19
20
21 365 Figure 10). While III-350/2 and III-550/2 show an initial peak at early hours, III-550/2 also shows
22 366 a secondary peak at ~10 h which was not the case for I-550/2. III-350/2 also shows a minor
23
24 367 secondary peak at 15 – 17 h. No initial peak in the heat flow of III-550/24 and III-750/2 could be
25
26 368 identified while similar to III-550/2, a secondary peak around ~ 10 h could be observed for these
27 369 pastes. All pastes incorporating DCIII exhibit higher heat of hydration compared to the control at
28
29 370 all times while falling short from that of III-QP between 8 and 15 h. At later times, pastes
30
31 371 incorporating DCIII show higher heat compared to both III-CTRL and III-QP, and those
32 372 incorporating DCP produced at higher dehydration temperatures and times generate higher heats.
33
34
35 373 The modification of the heat release pattern in the first 48 h does not translate to lower cumulative
36
374 heat in the longer term. By 7 days, the 550/24 and 750/2 DCPs for both CEM I and CEM III show
37
38 375 the highest heat release, as normalised to the initial CEM I content (
39
40
376 Figure 10b). On the other side of the spectrum, I-350/2 and III-350/2 show the least deviation from
41
377 the heat evolution of I-CTRL and the lowest total heat release among all blended mixtures (and
42
378 lowest R3 heat release both in the short and long terms). A good correlation could be established
43
379 between the long-term heat release of the pastes incorporating each DCP and the R3 cumulative
44
45
380 heat release of the same DCPs. In
46
47
381 Figure 12, the isothermal calorimetry results of CEM I and CEM III pastes are plotted against the
48 382 R3 heat results of the same at different test durations. Notice the direct correlation between the
49
50 383 two.
51
52 384 Further investigations are carried out using thermogravimetric and XRD analysis to establish the
53
54 385 effect of the DCPs on the formation and properties of the hydration product assemblages. Figure
55
56 386 13 and Figure 14 show the TG and DTG plots of CEM I and CEM III pastes for different hydration
57 387 ages, respectively. The pastes incorporating DCPs show higher mass loss than the pastes with
58
59 388 quartz powder at all ages. This indicates a higher content of hydrates contributed by (re)hydration
60
61
62
12
63
64
65
 1
2
3
4 389 of the DCP. Compared to the control pastes (I-CTRL and III-CTRL), similar TG/DTG responses
5
6 390 have been recorded for the pastes incorporating DCP, indicating that the reaction of the DCPs
7
8 391 produces similar types and amounts of hydration products. Part of these products are initially
9 392 contributed by the DCPs. The bound water and portlandite contents of the blended pastes
10
11 393 produced with and without DCP are plotted in Figure 15 and Figure 16. The values are reported
12
13 394 with respect to the residual masses at 550 °C to cancel out the dilution effects due to the
14 395 incorporation of water. Figure 15a confirms that the blends with quartz powder (I-QP) contain
15
16 396 significantly less bound water than the I-CTRL and the DCP pastes, which is explained by the
17
18 397 dilution of the cement by the inert quartz powder. The control and the DCP blends show similar
19 398 levels of bound water.
20
21
22 399 Figure 16a shows the bound water variations of different CEM III pastes over time. In comparison
23 400 with CEM I pastes, it is clearly observed that the CEM III pastes have a slower rate of hydration
24
25 401 (slower rate of bound water generation). Similar as in the CEM I mixtures, III-QP produced
26
27 402 significantly less bound water compared to III-CTRL and the DCP blends at all ages. The bound
28 403 water levels in the DCP-blends were comparable to that in the III-CTRL paste, clearly indicating
29
30 404 the contribution of the rehydrated DCP to the overall hydrate content. Contrary to the increasing
31
32 405 trend of the portlandite content with age in the CEM I mixtures, the portlandite content in the
33 406 CEM III mixtures decreases over time in all cases. This is interpreted to be due to the presence of
34
35 407 slag which consumes the portlandite. The lower portlandite contents in the III-QP pastes compared
36
37 408 to III-CTRL can be accounted for by the 30 wt% dilution factor of the III-QP blend. In contrast, the
38 409 lower portlandite levels in the DCIII pastes could indicate a higher overall degree of reaction of the
39
40 410 slag, partially inherited from the DCIII materials, or, alternatively, a difference in the clinker
41
42 411 hydration degree.
43
44 412 The consumption of the clinker phases and the formation of the hydration products was studied in
45
46 413 more detail by quantitative XRD analysis. To identify the reaction mechanisms underlying the
47
48
414 deviant early-age heat release profiles of the DCP pastes, CEM I pastes were closely investigated
49 415 in terms of their early-age phase evolution via in-situ XRD. As an example, Figure 17a and b show
50
51 416 the evolution of phases in the I-350/2 paste in the 8 – 15° 2θ and 33 – 41° 2θ ranges, respectively.
52
53
417 Figure 17a shows the formation of ettringite and hemicarboaluminate in the early hours of
54 418 hydration. Figure 17b illustrates the consumption of the clinker phases (A and B; alite and belite)
55
56 419 over time giving rise to the formation of portlandite up to 24 h of hydration. Later age specimens
57
58 420 were investigated by ex-situ measurements on cut slices of hydrated paste. For illustration, the
59 421 diffractograms of the I-350/2 and III-350/2 pastes for different hydration ages are presented in
60
61
62
13
63
64
65
 1
2
3
4 422 Figure 18. The longer term hydration of the DCP-cements shows broad similarities with the
5
6 423 hydration of a reference Portland cement. Ettringite and portlandite form early on, while the AFm
7
8 424 phases (hemi-, and monocarboaluminate) form subsequently. In the CEM III blended cements,
9 425 hydrotalcite is formed additionally as a reaction product of the hydration of the slag. A comparison
10
11 426 of the hydrate assemblage of the various cements at the age of 28 days is made in Figure 19.
12
13 427 Clearly, the DCP-blended pastes show higher, more pronounced ettringite and AFm (hemi- and
14 428 monocarboaluminate) reflections than the control pastes.
15
16
17 429 The XRD data were processed by Rietveld analysis for quantification of phase abundancies in the
18 430 course of hydration. Figure 20 presents the quantification results for both in-situ and ex-situ XRD
19
20 431 measurements of the CEM I pastes. Inspection of the C3S degrees of reaction in Figure 20a) and
21
22 432 b) learns that at very early age C3S hydration initiates more early on in the DCP, yet is surpassed
23 433 within several hours by the control paste as the C3S hydration rate in the DCP pastes appears
24
25 434 slower after the first few hours. In line with the calorimetry results the rate of reaction of the C3S
26
27 435 in I-CTRL slows down strongly after 7 h of hydration, allowing the DCP pastes to gradually catch
28 436 up over the first few days of hydration, reaching equivalent levels by 2 days of hydration and
29
30 437 beyond. The higher the dehydration temperature and duration of the DCP, the more pronounced
31
32 438 the difference in the C3S reaction profile with the control.
33
34 439 Considering the total degree of hydration of clinker phases, Figure 20 c and d present generally
35
36 440 slower and lower degrees of reaction that persist until 28 days for DCPs dehydrated at
37
441 temperatures of 550 °C and above. It should be noted that the C2S, C3A and C4AF phases,
38
39 442 including their polymorphs, contained in the DCPs were integrated in the total amount of clinker
40
41 443 phases, as it was not possible to make a distinction between these and corresponding phases
42
444 present in the CEM I. Therefore, to some extent the lower degrees of reaction of the DCP
43
44 445 containing pastes can be allocated to a lower reactivity of part of the DCP “clinker” phases, in
45
46 446 particular the γ-C2S. The absence of such recrystallised clinker phases in I-350/2 can also explain
47
48
447 the higher clinker degrees of reaction in the corresponding pastes. The early age portlandite
49 448 content profiles in Figure 20 e) mainly reflect the C3S reaction curves when taking into account
50
51 449 that the DCP pastes initially contained portlandite. As shown in Figure 20 f), at later age total
52
53
450 portlandite levels in the DCP pastes appear to be slightly lower or equivalent to the control paste,
54 451 which is more or less in line with the trend in the TGA results.
55
56
57 452 The XRD analysis results in Figure 20 g) and h) indicate more rapid and more extensive formation
58 453 of ettringite in the cements blended with DCPs, in particular for DCPs treated at higher
59
60 454 temperatures. Also, the formation of AFm-hemicarboaluminate occurs significantly earlier in the
61
62
14
63
64
65
 1
2
3
4 455 blends with the higher temperature DCPs. This is a striking difference with the I-CTRL paste in
5
6 456 which hemicarboaluminate was not observed in the in-situ experiment and is only found from 7
7
8 457 days of hydration in the ex-situ measurements. Simultaneously, gypsum in the DCP cements is
9 458 very rapidly consumed between leaving no detectable traces in early hours (< 1 h), while persisting
10
11 459 until 4 – 5 h of hydration (beyond the main hydration peak) in the control paste. See Figure 21a
12
13 460 and b to compare the gypsum primary peak between the I-CTRL paste and I-350/2 in early hours.
14
15 461 Rapid gypsum depletion and early hemicarboaluminate formation are both indicative for
16
17 462 undersulfation of the DCP cements. Undersulfation is well-known to affect the hydration of C3S,
18 463 the nucleation and growth of C-S-H [39,40], and leads to slower strength development [41]. In
19
20 464 case sulfate depletion occurs before the main hydration peak, the C3S hydration peak becomes
21
22 465 lower and broader [42]. It is argued here that the addition of DCP as a high surface area, reactive
23 466 material perturbs the regular C3S hydration process by rapidly consuming sulfates from the system,
24
25 467 either by precipitation of calcium aluminate hydrates, as observed by XRD, or by sorption onto
26
27 468 C-S-H or DCP phases [43]. In addition, it can be argued that the DCP hydrates could act as
28 469 suitable nuclei for new hydrates, and in this way shorten the induction period between the initial
29
30 470 peak and the main C3S hydration peak significantly. This is clearly observed in the calorimetry
31
32 471 curves in Figure 10 and Figure 11, and to a lesser extent in the XRD results in Figure 20 by the
33 472 earlier initiation of C3S hydration in the DCP pastes. A similar effect is sorted by commercial
34
35 473 additives using homogeneously dispersed C-S-H nuclei to accelerate the hydration of C3S [44],
36
37 474 however without the occurrence of the subsequent retardation seen by this study. The in-situ XRD
38 475 results indicate that the retardation is reflected in simultaneous slow-down in C3S hydration,
39
40 476 however the mechanism explaining this retardation is still unclear. Possible causes are one or a
41
42 477 combination of the following i) modification of the pore solution composition, ii) changes in the
43
478 nucleation and growth rate of the C-S-H and other hydrates, and iii) differences in the resulting
44
45 479 microstructure of the hydrated cement. The latter could entail changes in the mechanical
46
47 480 performance of the cement which are presented and discussed in the subsequent section.
48
49
50 481 3.3. Mechanical performance
51
52 482 The mechanical strength results indicate that the modification of the hydration process by
53
54 483 introduction of the DCP phases can strongly affect the mechanical performance. Figure 22a and
55 484 b represent the absolute and relative strength development results of the CEM I mortars. It is
56
57 485 observed that I-550/24 and I-750/2 developed the lowest strength values among other mixtures
58
59 486 containing DCP. They also showed lower strength values compared to I-QP. I-350/2 and I-550/2,
60 487 on the other hand, showed higher strength results compared to I-QP almost at all ages. This
61
62
15
63
64
65
 1
2
3
4 488 indicates that the inverse influence of incorporating such DCPs on the strength of CEM I mixtures
5
6 489 is beyond their dilution effect and even indicative of some non-compatibility with CEM I in terms
7
8 490 of strength development. I-550/2 and I-350/2, on the other hand, develop a 28-d compressive
9 491 strength of nearly 70 % that of the control and continue to close the gap at later ages. Therefore,
10
11 492 they could be deemed as moderately reactive SCMs. Mixtures incorporating DCI/350°C/2h, in
12
13 493 particular, reach 85 – 90 % of the 90-d compressive strength of the control mixture which could
14 494 be considered a satisfactory result given the high replacement ratio (30 wt%).
15
16
17 495 A close look at the strength development plots of the CEM III mortar series (Figure 23) reveals
18 496 that the presence of slag causes some notable changes in the strength development. The gap
19
20 497 between the strength performance of the DCP-blended pastes compared to the control is much
21
22 498 smaller in such blends compared to the CEM I paste series. All DCP-blended mixtures have
23 499 performed better compared to the mixture III-QP incorporating the inert filler (quartz powder)
24
25 500 almost at all ages. III-550/24 was found to develop the highest strength values compared to all
26
27 501 other DCP-blended mixtures at all ages, and even better than the control mortar at early ages
28 502 (< 3 d). This mixture was also found to develop a 56-d strength exceeding the 28-d strength of the
29
30 503 control mixtures. While the rate of strength development is slower in the CEM III mixtures
31
32 504 compared to CEM I mixtures (compare Figure 22b and Figure 23b), the CEM III mixtures reach
33 505 higher strength values at later ages in comparison with their respective CEM I mixtures. Such
34
35 506 systematic behaviour suggests that recycling of concrete produced from CEM III is technically
36
37 507 more favourable when it comes to the long-term mechanical performance. A more detailed
38 508 comparison of the performance of the mixtures incorporating different DCIII materials reveals that
39
40 509 there is not a remarkable added value in increasing the dehydration temperature of DCIII beyond
41
42 510 350 °C. III-350/2 and III-550/24 develop similar strengths at 28 and 91 d. The advantage of the
43
511 latter appears to be limited only to the very early-age strength development (i.e., 1 d). Moreover,
44
45 512 it could be observed that an increase in the dehydration temperature of DCIII to 750 °C has some
46
47 513 detrimental effects on its strength performance. A similar trend was observed for the CEM I pastes,
48
514 allowing to conclude that a thermal dehydration at 350 °C for 2 h is sufficient for generating a
49
50 515 suitable DCP and exceeding temperatures, especially beyond 550 °C, could result in adverse
51
52 516 effects.
53
54 517 The phase composition and strength performance of DCPs are further analysed with respect to
55
56 518 one another to determine the significant correlations between the two. No positive correlation was
57
58 519 detected between the belite phases formed in DCI/DCIII and their strength performance (see
59 520 Figure 8, Figure 22, and Figure 23). DCI/750°C/2h, for instance, contained the highest β-C2S (see
60
61
62
16
63
64
65
 1
2
3
4 521 Figure 8) and showed the lowest strength. This does not mean that belites have a negative effect
5
6 522 on strength. Rather, the increase in the dehydration temperature results in recrystallisation of
7
8 523 reactive amorphous phases that would apparently better contribute to strength performance.
9 524 Statistical analyses suggest that the amorphous content in the DCI materials has a direct
10
11 525 correlation with the strength performance of this material at different ages (see Table 5). This was
12
13 526 not always the case for DCIII. Moreover, some positive correlation was also observed, especially
14 527 for DCIII, between the portlandite content and strength performance.
15
16
17 528 The bound water and portlandite contents of the mixtures containing DCP are examined against
18 529 their compressive strength at different ages and the results are shown in Table 6. It is observed
19
20 530 that there is a negative correlation between the compressive strength of the CEM I mixtures
21
22 531 containing DCP, and the amounts of bound water and portlandite produced in the same. This
23 532 observation coupled with the inverse correlation observed between compressive strength and
24
25 533 heat release of DCP-blended mixtures (Figure 24) suggests that the hydration reactions of such
26
27 534 mixtures are apparently not favourable in terms of microstructure and mechanical strength.
28 535 DCI/350°C/2h that generated the lowest R3 heat, and which had lowest bound water and
29
30 536 portlandite content, showed the highest compressive strength results compared to other blended
31
32 537 mixtures at all ages. However, the minor differences between the bound water and portlandite
33 538 content values may not necessarily justify the significant differences in the strength results. It
34
35 539 appears that the extent of deviation in the heat release pattern compared to that of the control
36
37 540 mixture might better explain the observed differences in mechanical performance.
38
39 541 Similar to the case of CEM I mixtures, the CEM III mixtures containing DCP showed an inverse
40
41 542 correlation between the compressive strength, and the bound water and portlandite content (see
42
543 Table 6). It appears that more hydration reactions sourced from the DCPs (evidenced by the heat
43
44 544 of hydration and bound water) does not translate to higher mechanical strength, which is probably
45
46 545 due to the type and microstructure of the phases formed during hydration. The contribution of
47
48
546 GGBFS, on the other hand, appears to have some positive effects on the strength. Judging by the
49 547 decreasing portlandite contents of CEM III pastes incorporating DCP, it could be argued that more
50
51 548 portlandite consumption usually translates to higher compressive strength (see III-350/2 and III-
52
53
549 550/24 in Figure 16 in comparison to other mixtures).
54
55 550
56
57
58
59
60
61
62
17
63
64
65
 1
2
3
4
5 551 4. Conclusions
6
7
8 552 Dehydrated cement (DC) powders (referred to as DCPs) prepared via thermal treatment and
9 553 comminution of CEMI and (synthetic) CEMIII hydrated pastes were investigated in this research
10
11 554 for their physico-chemical properties and cementitious performance as SCMs. A full hydration
12
13 555 study was carried out to explore the effects of partial replacement of fresh cement (both CEMI and
14 556 CEMIII) with such SCMs. The main findings of this research are summarised as follows:
15
16
17 557 - Increase in the dehydration temperature from 350 °C to 750 °C generally results in an increase
18
558 in the grindability of such materials as well as their surface area (with an exception observed in
19
20 559 the case of DCIII dehydrated at 750 °C).
21
22
23
560 - Recrystallisation of C-S-H at 550 °C and higher temperatures results in the formation of γ-C2S
24 561 and β-C2S. The content of potentially hydraulic phases increased with dehydration temperature at
25
26 562 the expense of the amorphous or nanocrystalline phase (likely C-S-H and other cement hydrates).
27
28 563 - According to the R3 reactivity results, DCPs prepared at higher dehydration temperatures and
29
30 564 times generate significantly higher initial rates of reaction compared to those prepared at lower
31
32 565 temperatures, and were also found to have higher reactivity in the long term (after 1 – 2 days).
33
34 566 - Positive correlations were found between the R3 heat results and the amount of potentially
35
36 567 hydraulic phases formed in the DCPs (for both DCI and DCIII) while most correlations were not
37 568 statistically significant. No direct correlation was found, however, between the R3 heat release
38
39 569 and the strength performance of blended cements comprising 30 wt% DCP.
40
41 570 - Blended cement pastes incorporating DCP generated similar or lower bound water and
42
43 571 portlandite content compared to the control paste (i.e., I-CTRL). The pastes incorporating DCPs
44
45 572 prepared at higher temperatures generated higher bound water compared to the other blended
46 573 pastes, but this did not translate to a better strength performance.
47
48
49 574 - Incorporation of DCP in the pastes results in a high initial heat generation in the first 1 – 2 h and
50
575 also expedites the acceleration period. However, the pastes incorporating DCPs produced at
51
52 576 higher temperatures (i.e., 550°C/24h and 750°C/2h) show a significant retardation of the main
53
54 577 hydration of C3S. This is in agreement with the lower C3S hydration rate (according to the Rietveld
55
578 analysis results of in-situ XRD) of such pastes at that time.
56
57
58 579 - DCP blended cements contain higher levels of crystalline ettringite and AFm hydration products
59
60 580 throughout the hydration of the cement. AFm phases are formed early on, and gypsum is depleted
61
62
18
63
64
65
 1
2
3
4 581 before the main C3S hydration event, indicating undersulfation and a disturbance of the regular
5
6 582 hydration process with potential consequences for hydrate nucleation and growth and
7
8 583 microstructure development.
9
10 584 - The best strength performance in the case of CEMI pastes was observed for DCI/350°C/2h
11
12 585 followed by DCI/550°C/2h, 550°C/24h, and 750°C/2h. As such, dehydration at temperatures
13 586 exceeding 350 °C is not recommended for this DC type.
14
15
16 587 - In the case of CEMIII pastes, while DCIII/550°C/24h showed the best early-age strength
17 588 performance, similar strength activity was identified for DCIII/350°C/2h at later ages (after 28 d).
18
19 589 Increasing the dehydration temperature to 550 °C for extended durations of time is therefore
20
21 590 deemed not justifiable unless a high early-age strength is desired.
22
23 591 - All DCIII materials and their respective blended mixtures were found to have a slower rate of
24
25 592 strength development compared to the corresponding DCI materials and pastes. However, they
26 593 showed higher strength results at later ages compared to DCI.
27
28
29 594
30
31
32 595 5. Acknowledgements
33
34
35 596 This study was conducted under the BnB “Beton naar Hoogwaardig Beton” project funded by the
36
37 597 Interreg V Vlaanderen-Nederland programme.
38
39
40 598 6. References
41
42
43
44
45 [1] I.Z. Bribián, A.V. Capilla, A.A. Usón, Life cycle assessment of building materials: Comparative analysis
46 of energy and environmental impacts and evaluation of the eco-efficiency improvement potential. Building
47 and environment 46(5) (2011) 1133-1140.
48 [2] E. Worrell, L. Price, N. Martin, C. Hendriks, L.O. Meida, Carbon dioxide emissions from the global cement
49 industry. Annual review of energy and the environment, 26(1) (2001) 303-329.
50 [3] WBCSD and CSI. Cement Industry Energy and CO 2 Performance. Getting the Numbers Right (GNR).
51 12/30/2016 - 8/24/2021]; Available from: https://docs.wbcsd.org/2016/12/GNR.pdf (accessed 8 May 2022).
52 [ 4 ] RILEM Recommendation, Specifications for concrete with recycled aggregates. Materials and
53 Structures 27(9) (1994) 557-559.
54 [5] J. Schoon, K. De Buysser, Fines extracted from recycled concrete as alternative raw material for Portland
55 clinker production. In 1st Concrete Innovation Conference (CIC) (2014).
56 [6] D. Gastaldi, F. Canonico, L. Capelli, L. Buzzi, E. Boccaleri, S. Irico, An investigation on the recycling of
57
hydrated cement from concrete demolition waste. Cement and Concrete Composites 61 (2015) 29-35.
58
[7] M. Lidmila, P. Tesárek, T. Plachy, Z. Rácová, P. Padevět, V. Nežerka, O. Zobal, Utilization of recycled
59
fine-ground concrete from railway sleepers for production of cement-based binder. In: Applied Mechanics
60
and Materials, Trans Tech Publications Ltd., Vol. 486, 2014, pp. 323-326.
61
62
19
63
64
65
 1
2
3
4
5
6
[8] K. Šeps, I. Broukalová, Properties of cement based composite with fine ground recycled concrete.
7
In Advanced Materials Research, Trans Tech Publications Ltd, Vol. 1000, 2014, pp. 110-113.
8
[9] Q. Liu, T. Tong, S. Liu, D. Yang, Q. Yu, Investigation of using hybrid recycled powder from demolished
9
10 concrete solids and clay bricks as a pozzolanic supplement for cement. Construction and Building
11 Materials 73 (2014) 754-763.
12 [ 10 ] J. Topič, Z. Prošek, Properties and microstructure of cement paste including recycled concrete
13 powder. Acta Polytechnica 57 (2017) 49-57.
14 [11] L. Oksri-Nelfia, P.Y. Mahieux, O. Amiri, P. Turcry, J. Lux, Reuse of recycled crushed concrete fines as
15 mineral addition in cementitious materials. Materials and Structures 49, 2016, 3239-3251.
16 [12] J. Wang, M. Mu, Y. Liu, Recycled cement. Construction and Building Materials 190 (2018) 1124-1132.
17 [13] M.C.R. Farage, J. Sercombe, C. Galle, Rehydration and microstructure of cement paste after heating
18 at temperatures up to 300 C. Cement and Concrete Research 33 (2003) 1047-1056.
19 [14] C. Alonso, L. Fernandez, Dehydration and rehydration processes of cement paste exposed to high
20 temperature environments. Journal of materials science 39 (2004) 3015-3024.
21 [15] F. Splittgerber, A. Mueller, Inversion of the cement hydration as a new method for identification and/or
22 recycling. In International congress on the chemistry of cement, Durban, South Africa, 2003 , pp. 1282-1290.
23 [16] R. Serpell, M. Lopez, Reactivated cementitious materials from hydrated cement paste wastes. Cement
24 and Concrete Composites 39 (2013) 104-114.
25 [17] R. Serpell, F. Zunino, Recycling of hydrated cement pastes by synthesis of α′ H-C2S. Cement and
26 Concrete Research 100 (2017) 398-412.
27 [18] A. Carrico, J.A. Bogas, M. Guedes, Thermoactivated cementitious materials–A review. Construction
28 and Building Materials, 250 (2020) 118873.
29 [19] M. Vyšvařil, P. Bayer, M. Chromá, P. Rovnaníková, Physico-mechanical and microstructural properties
30 of rehydrated blended cement pastes. Construction and Building Materials 54 (2014) 413-420.
31 [20] G. Wang, C. Zhang, B. Zhang, Q. Li, Z. Shui,. Study on the high‐temperature behavior and rehydration
32 characteristics of hardened cement paste. Fire and Materials 39 (2015) 741-750.
33 [21] Q. Zhang, G. Ye, E. Koenders, Investigation of the structure of heated Portland cement paste by using
34 various techniques. Construction and Building Materials 38 (2013) 1040-1050.
35 [22] R. Yu, Z. Shui, J. Dong, Using dehydrated cement paste as new type of cement additive. ACI Materials
36 Journal 110 (2013) 395.
37 [23] R. Yu, Z. Shui, Efficient reuse of the recycled construction waste cementitious materials. Journal of
38 cleaner production 78 (2014) 202-207.
39 [24] M.V.A. Florea, Z. Ning, H.J.H. Brouwers, Activation of liberated concrete fines and their application in
40 mortars. Construction and Building Materials 50 (2014) 1-12.
41 [25] Z. Shui, D. Xuan, H. Wan, B. Cao, Rehydration reactivity of recycled mortar from concrete waste
42 experienced to thermal treatment. Construction and Building materials 22 (2008) 1723-1729.
43 [26] S.C. Angulo, M.S. Guilge, V.A. Quarcioni, R. Baldusco, M.A. Cincotto, Rehydration of cement fines: A
44 TG/calorimetry study. Institute for Technological Research: São Paulo, Brazil, 2013.
45 [27] Z. Shui, D. Xuan, W. Chen, R. Yu, R. Zhang, Cementitious characteristics of hydrated cement paste
46 subjected to various dehydration temperatures. Construction and building materials 23 (2009) 531-537.
47 [28] H. Wu, J. Xu, D. Yang, Z. Ma, 2018. Utilizing thermal activation treatment to improve the properties of
48 waste cementitious powder and its newmade cementitious materials. Journal of Cleaner Production,
49 129074.
50 [29] B. Lothenbach, P. Durdzinski, K. De Weerdt, Thermogravimetric analysis. In: K.L. Scrivener, R.
51 Snellings; B. Lothenbach (Eds.), A practical guide to microstructural analysis of cementitious
52 materials, CRC Press, La Boca, 2016, pp. 177-211.
53 [30] R. Snellings, K.L. Scrivener, Rapid screening tests for supplementary cementitious materials: past and
54 future. Materials and Structures 49 (2016) 3265-3279.
55 [31] American Standard for Testing and Materials, 2020. Measuring the Reactivity of Supplementary
56 Cementitious Materials by Isothermal Calorimetry and Bound Water Measurements (ASTM C1897 − 20).
57 Retrieved from https://www.astm.org/c1897-20.html.
58 [32] M.J. DeJong, F.J. Ulm, The nanogranular behavior of CSH at elevated temperatures (up to 700
59 C). Cement and Concrete Research, 37 (2007) 1-12.
60 [33] C. Rodriguez‐Navarro, E. Hansen, W.S. Ginell, Calcium hydroxide crystal evolution upon aging of lime
61 putty. Journal of the American Ceramic Society 81 (1998) 3032-3034.
62
20
63
64
65
 1
2
3
4
5
6
[34] R. Snellings, M. De Schepper, K. De Buysser, I. Van Driessche, N. De Belie, Clinkering reactions during
7
firing of recyclable concrete. Journal of the American Ceramic Society 95 (2012) 1741-1749.
8
[35] R. Snellings, X-ray powder diffraction applied to cement. In: K.L. Scrivener, R. Snellings; B. Lothenbach
9
10 (Eds.), A practical guide to microstructural analysis of cementitious materials, CRC Press, La Boca, 2016,
11 pp. 107-176.
12 [36] J.F. Lambert, W.S. Millman, J.J. Fripiat, Revisiting kaolinite dehydroxylation: A silicon-29 and aluminum-
13 27 MAS NMR study. Journal of the American Chemical Society 111 (1989) 3517-3522.
14 [37] H.F. Taylor, Cement chemistry, Thomas Telford, London, 1997.
15 [38] J.W. Bullard, H.M. Jennings, R.A. Livingston, A. Nonat, G.W. Scherer, J.S. Schweitzer, ... & J.J. Thomas,
16 Mechanisms of cement hydration. Cement and concrete research 41 (2011) 1208-1223.
17 [ 39 ] B. Mota, T. Matschei, K.L. Scrivener, The influence of sodium salts and gypsum on alite
18 hydration. Cement and Concrete Research, 75 (2015) 53-65.
19 [40] E.M. Bérodier, A.C. Muller, K.L. Scrivener, Effect of sulfate on CSH at early age. Cement and Concrete
20 Research, 138 (2020) 106248.
21 [41] L. Wadsö, F. Winnefeld, K. Riding, P. Sandberg, Calorimetry. In: K.L. Scrivener, R. Snellings; B.
22 Lothenbach (Eds.), A practical guide to microstructural analysis of cementitious materials, CRC Press, La
23 Boca, 2016, pp. 37-106.
24 [42] M.D. Niemuth, L. Barcelo, J. Weiss, Effect of fly ash on optimum sulfate levels measured using heat
25 and strength at early ages. Advances in Civil Engineering Materials 1 (2012) 1-18.
26 [ 43 ] A. Quennoz, K.L. Scrivener, Interactions between alite and C3A-gypsum hydrations in model
27 cements. Cement and Concrete Research 44 (2013) 46-54.
28 [44] E. John, T. Matschei, D. Stephan, Nucleation seeding with calcium silicate hydrate–A review. Cement
29 and Concrete Research 113 (2018) 74-85.
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
21
63
64
65
 1
2
3
4
5 List of Tables
6
7
8 Table 1. The chemical composition, physical properties, and mineral composition of the binder
9 materials
10
11 Table 2. Overview of the studied dehydrated cement paste samples
12
13 Table 3. The R3 cumulative heat results for the dehydrated cement paste powders
14
15 Table 4. The correlation coefficients (and corresponding p-values) between the amount of different
16 phases and the 7-d R3 cumulative heat release values
17
18 Table 5. The correlation coefficients (and corresponding p-values) between the amount of different
19 phases and the (relative) strength performance at different ages
20
Table 6. The correlation coefficients (and corresponding p-values) between the compressive
21
strength and TGA results (normalised to CEM I mass) at different ages (only for mixtures
22
containing DCP)
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
22
63
64
65
 1
2
3
4
5 List of Figures
6
7
8 Figure 1. The particle size distribution of the raw materials
9
Figure 2. The particle size distribution of the dehydrated cement paste powders treated under
10
different conditions
11
12 Figure 3. BET surface areas of the dehydrated cement paste powders treated under different
13
conditions
14
15 Figure 4. SEM images of the DCI dehydrated cement paste powders at 5,000x (left) and 50,000x
16 (right) magnifications. (a): 350°C/2h, (b): 550°C/2h, (c): 550°C/24h, (d): 750°C/2h
17
18 Figure 5. SEM images of the DCIII dehydrated cement paste powders at 5,000x (left) and 50,000x
19 (right) magnifications. (a): 350°C/2h, (b): 550°C/2h, (c): 550°C/24h, (d): 750°C/2h
20
21 Figure 6. TGA and DTG plots of the dehydrated cement paste powders. (a): DCI, (b): DCIII
22
23 Figure 7. Bound water, portlandite, and carbonates contents of the dehydrated cement paste
24 powders. (a): DCI, (b): DCIII
25
26 Figure 8. Rietveld analysis results of the dehydrated cement paste powders. (a): DCI, (b): DCIII
27
28 Figure 9. The R3 cumulative heat release and specific heat flow plots of dehydrated cement paste
29 powders in the first 24 h (left) and 7 days (right). (a): DCI, (b): DCIII
30
31 Figure 10. The isothermal calorimetry results of CEM I pastes (a) up to 24 h, and (b): up to 7 d
32
33 Figure 11. The isothermal calorimetry results of CEM III pastes (a) up to 24 h, and (b): up to 7 d
34
35 Figure 12. The isothermal calorimetry versus R3 heat release results for CEM I and CEM III pastes
36 at different ages (the embedded table shows the correlation coefficients between calorimetry and
37 R3 heat release and corresponding p-values at different test durations)
38
Figure 13. The TGA and DTG curves of the CEM I pastes. (a) 3 d, (b) 7 d, (c) 28 d, (d) 91 d
39
40 Figure 14. The TGA and DTG curves of the CEM III pastes. (a) 3 d, (b) 7 d, (c) 28 d, (d) 91 d
41
42 Figure 15. The bound water (a) and portlandite content (b) values of the CEM I pastes (values
43 normalised to the binder content)
44
45 Figure 16. The bound water (a) and portlandite content (b) values of the CEM III pastes (values
46 normalised to the binder content)
47
48 Figure 17. The in-situ XRD pattern of I-350/2 paste over time. (a): 8 – 15°, (b): 33 – 41°; A: alite, B:
49 belite, E: ettringite, F: ferrite, H: hemicarboaluminates, P: portlandite
50
51 Figure 18. The ex-situ XRD pattern of (a) I-350/2 paste, and (b): III-350/2 pastes over time; A: alite,
52 AF: AFm phases, B: belite, C: C3A; E: ettringite, F: ferrite, H: hemicarboaluminates, M:
53 monocarboaluminates; P: portlandite, Q: quartz, T: hydrotalcite
54
55 Figure 19. The ex-situ XRD patterns of (a) CEM I series, and (b): CEM III series at 28 d; A: alite, AF:
56 AFm phases, B: belite, C: C3A; E: ettringite, F: ferrite, H: hemicarboaluminates, M:
57 monocarboaluminates; P: portlandite, Q: quartz, T: hydrotalcite
58
59
60
61
62
23
63
64
65
 1
2
3
4 Figure 20. The evolution of primary phases in the CEM I pastes over time (left: first 24 h; in-situ,
5 right: up to 91 d; ex-situ). (a and) C3S degree-of-reaction, (c and d) clinker degree-of-reaction, (e
6 and f) portlandite content, (g and h) ettringite
7
8 Figure 21. The evolution of gypsum (primary peak) in the in-situ XRD diffractograms. (a) I-CTRL,
9 (b) I-350/2
10
11 Figure 22. The compressive strength results of the CEM I pastes. (a) absolute compressive
12 strength results, (b) relative strength ratios with respect to the 28-d strength of the control mixture
13
14 Figure 23. The compressive strength results of the CEM III pastes. (a) absolute compressive
15 strength results, (b) relative strength ratios with respect to the strength of the control mixtures at
16 the same age
17
18 Figure 24. The 7-d compressive strength of blended CEM I and CEM III pastes versus the 7-d
19 calorimetry heat release
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
24
63
64
65
All tables Click here to access/download;Table;Tables_all.docx

Click here to view linked References

Table 1. The chemical composition, physical properties, and mineral composition of the binder
materials

Content (m%)
Quantity CEM I GGBFS CEM III
CaO 60.7 37.5 51.8
SiO2 18.5 34.5 24.3
Al2O3 4.89 14.4 8.11
MgO 1.63 7.84 4.19
Fe2O3 2.88 0.40 1.85
TiO2 0.31 1.29 0.55
MnO 0.05 0.31 0.18
SO3 2.53 1.23 2.00
K2O 0.84 0.74 0.74
Na2O 0.12 0.47 0.22
Na2Oeq 0.67 0.96 0.71
LOI (%) 1.47 0.97 1.95
C3S M3 59.5 – 33.8
β-C2S 17.1 – 9.1
C3A cubic 5.2 – 2.8
C4AF 10.0 – 6.1
Portlandite 1.3 – 0.1
Gypsum 1.6 – 0.8
Anhydrite 0.9 – 0.2
Basanite 3.1 – 1.6
Quartz 0.6 – 0.17
Periclase 0.7 – 0.61
Amorphous 0.0 100 44.7
SSA (m2/g)* 1.76 1.35 1.42
* BET specific surface area
 Table 2. Overview of the studied dehydrated cement paste samples

Starting material Long name Dehydration temperature (°C) Dwelling time (h)
CEM I DCI/350°C/2h 350 2
DCI/550°C/2h 550 2
DCI/550°C/24h 550 24
DCI/750°C/2h 750 2
CEM III DCIII/350°C/2h 350 2
DCIII/550°C/2h 550 2
DCIII/550°C/24h 550 24
DCIII/750°C/2h 750 2
Table 3. The R3 cumulative heat results for the dehydrated cement paste powders

DCP label 1 day 3 day 7 day

J/g SCM J/g SCM J/g SCM
DCI/350°C/2h 95 115 125
DCI/550°C/2h 175 196 204
DCI/550°C/24h 137 177 189
DCI/750°C/2h 143 192 210
DCIII/350°C/2h 138 193 225
DCIII/550°C/2h 185 240 271
DCIII/550°C/24h 180 243 283
DCIII/750°C/2h 160 237 281
Table 4. The correlation coefficients (and corresponding p-values) between the amount of different phases and the 7-d R3 cumulative
heat release values

DC type R3 C3S M3 α-C2S β-C2S γ-C2S C3A cubic C4AF Portlandite Anhydrite Amorphous
(duration)
0.42 0.22 0.16 0.54 0.19 0.15 -0.66 0.97 -0.33
1d
(0.58) (0.78) (0.84) (0.46) (0.81) (0.85) (0.34) (0.03) (0.67)
0.52 0.51 0.5 0.78 0.48 0.29 -0.85 0.99 -0.65
DCI 3d
(0.48) (0.49) (0.5) (0.22) (0.52) (0.71) (0.15) (0.01) (0.35)
0.56 0.6 0.59 0.82 0.57 0.35 -0.89 0.96 -0.73
7d
(0.44) (0.4) (0.41) (0.18) (0.43) (0.65) (0.11) (0.04) (0.27)
0.11 -0.18 -0.11 0.71 -0.32 -0.1 0.22 0.81 -0.2
1d
(0.89) (0.82) (0.89) (0.29) (0.68) (0.9) (0.78) (0.19) (0.8)
0.37 0.24 0.34 0.95 0.11 0.29 -0.18 0.98 -0.61
DCIII 3d
(0.63) (0.76) (0.66) (0.05) (0.89) (0.71) (0.82) (0.02) (0.39)
0.42 0.38 0.5 0.99 0.26 0.41 -0.31 1 -0.73
7d
(0.58) (0.62) (0.5) (0.01) (0.74) (0.59) (0.69) (0) (0.27)
Table 5. The correlation coefficients (and corresponding p-values) between the amount of different
phases and the (relative) strength performance at different ages

DC type Relative strength β-C2S γ-C2S Portlandite Amorphous

(age)
-0.91 -0.94 0.82 0.95
3d
(0.09) (0.06) (0.18) (0.05)
-0.87 -0.81 0.53 0.84
7d
(0.13) (0.19) (0.47) (0.16)
DCI
-0.81 -0.93 0.59 0.84
28 d
(0.19) (0.07) (0.41) (0.16)
-0.8 -0.98 0.69 0.85
91 d
(0.2) (0.02) (0.31) (0.15)
-0.58 -0.29 0.81 0.6
3d
(0.42) (0.71) (0.19) (0.4)
-0.99 -0.54 0.98 0.93
7d
(0.01) (0.46) (0.02) (0.07)
DCIII
-0.86 -0.52 0.96 0.86
28 d
(0.14) (0.48) (0.04) (0.14)
-0.21 -0.36 0.43 0.37
91 d
(0.79) (0.64) (0.57) (0.63)
Table 6. The correlation coefficients (and corresponding p-values) between the compressive
strength and TGA results (normalised to CEM I mass) at different ages (only for mixtures
containing DCP)

Cement type Age (d) Bound water Portlandite content

1 NA NA
3 -0.69 (0.31) -0.91 (0.09)
CEM I series 7 -0.66 (0.34) -0.69 (0.31)
28 -0.85 (0.15) -0.92 (0.08)
91 -0.95 (0.05) -0.80 (0.20)
1 NA NA
3 -0.11 (0.89) -0.65 (0.35)
CEM III series 7 -0.57 (0.43) -0.82 (0.18)
28 -0.95 (0.05) -0.75 (0.25)
91 0.40 (0.60) -0.45 (0.55)
All Figures Click here to access/download;Figure;Figures_all.docx

Click here to view linked References

100
CEM I
CEM III
90
GGBFS
Quartz powder
80

70
Undersize (%)

60

50

40

30
Material Mean (µm) D10 (µm) D50 (µm) D90 (µm)

20 CEM I 28.2 0.9 8.8 93.3

CEM III 11.4 1.1 9.6 23.0

10 GGBFS 13.7 2.6 12.0 25.6

Quartz powder 9.42 2.15 7.45 19.4

0
0.1 1 10 100 1000
Size (µm)
Figure 1. The particle size distribution of the raw materials
 100
DCI/350°C/2h
90 DCI/550°C/2h
DCI/550°C/24h
80 DCI/750°C/2h

70 DCIII/350°C/2h
DCIII/550°C/2h
Undersize (%)

60 DCIII/550°C/24h
DCIII/750°C/2h
50
Material Mean (µm) D10 (µm) D50 (µm) D90 (µm)
40 DCI/350°C/2h 25.6 1.16 7.21 80.0

DCI/550°C/2h 14.1 1.33 6.83 34.4

30 DCI/550°C/24h 9.85 1.18 4.39 22.9

DCI/750°C/2h 12.7 1.00 4.46 32.6

20 DCIII/350°C/2h 19.0 1.36 8.29 49.6

DCIII/550°C/2h 10.8 1.47 7.25 22.5

10 DCIII/550°C/24h 12.5 1.51 6.39 25.7

DCIII/750°C/2h 11.6 1.19 5.16 25.3

0
0.1 1 10 100 1000
Size (µm)
Figure 2. The particle size distribution of the dehydrated cement paste powders treated under
different conditions
 25
DCI
DCIII
20
BET surface area (m2/g)

15

10

0
350°C/2h 550°C/2h 550°C/24h 750°C/2h
Treatment conditions
Figure 3. BET surface areas of the dehydrated cement paste powders treated under different
conditions
(a)

(b)

(c)

(d)

Figure 4. SEM images of the DCI dehydrated cement paste powders at 5,000x (left) and 50,000x
(right) magnifications. (a): 350°C/2h, (b): 550°C/2h, (c): 550°C/24h, (d): 750°C/2h
(a)

(b)

(c)

(d)

Figure 5. SEM images of the DCIII dehydrated cement paste powders at 5,000x (left) and 50,000x
(right) magnifications. (a): 350°C/2h, (b): 550°C/2h, (c): 550°C/24h, (d): 750°C/2h
(a) 0.10

0.00
100
Mass loss (% original mass)

-0.10

-0.20
95

DTG (%/min)
-0.30

-0.40
90
-0.50
DCI/350°C/2h
-0.60
DCI/550°C/2h
85 -0.70
DCI/550°C/24h
-0.80
DCI/750°C/2h
80 -0.90
50 150 250 350 450 550 650 750 850 950
Temperature (°C)
(b) 0.10

100 0.00
Mass loss (% original mass)

-0.10

DTG (%/min)
95
-0.20

90 -0.30
DCIII/350°C/2h
-0.40
DCIII/550°C/2h
85
DCIII/550°C/24h -0.50
DCIII/750°C/2h
80 -0.60
50 150 250 350 450 550 650 750 850 950
Temperature (°C)
Figure 6. TGA and DTG plots of the dehydrated cement paste powders. (a): DCI, (b): DCIII
(a)
18 17.03
Percentage (dry base) Bound water Portlandite Carbonates
16
14
12 11.11
10 9.09
7.87
8
6 5.36 5.02 5.29
3.64
4
1.78 1.29
2 0.66 0.43
0
DCI/350°C/2h DCI/550°C/2h DCI/550°C/24h DCI/750°C/2h
Sample label
(b)
14 12.91
Percentage (dry base)

Bound water Portlandite Carbonates

12
10
8
6.27 6.21
6
4.32
3.61
4 2.86
2.04
2 0.99 1.31 0.91
0.51 0.14
0
DCIII/350°C/2h DCIII/550°C/2h DCIII/550°C/24h DCIII/750°C/2h
Sample label
Figure 7. Bound water, portlandite, and carbonates contents of the dehydrated cement paste
powders. (a): DCI, (b): DCIII
 (a)
Amorphous
Hemicarbo
Calcite
Anhydrite
Component

Portlandite
C₄AF
C₃A cubic
γ-C₂S
DCI/350°C/2h
β-C₂S
DCI/550°C/2h
α-C₂S DCI/550°C/24h
C₃S M3 DCI/750°C/2h

0 10 20 30 40 50 60 70 80
Amount (m%)
(b)
Amorphous
Hemicarbo
Calcite
Anhydrite
Component

Portlandite
C₄AF
C₃A cubic
γ-C₂S
DCIII/350°C/2h
β-C₂S
DCIII/550°C/2h
α-C₂S DCIII/550°C/24h
C₃S M3 DCIII/750°C/2h

0 10 20 30 40 50 60 70 80 90
Amount (m%)
Figure 8. Rietveld analysis results of the dehydrated cement paste powders. (a): DCI, (b): DCIII
(a)

(b)

Figure 9. The R3 cumulative heat release and specific heat flow plots of dehydrated cement paste
powders in the first 24 h (left) and 7 days (right). (a): DCI, (b): DCIII
(a) 350 8

300 7

Heat flow (mW/g CEM I)

6
250
Heat (J/g CEM I)

I-CTRL 5
200 I-QP
I-350/2
I-550/2 4
150 I-550/24
I-750/2 3
100
2

50 1

0 0
0.5 4.5 8.5 12.5 16.5 20.5 24.5
Time (h)
(b) 600 8

7
500

Heat flow (mW/g CEM I)

6
Heat (J/g CEM I)

400
I-CTRL 5
I-QP
I-350/2
300 I-550/2 4
I-550/24
I-750/2 3
200
2
100
1

0 0
0 1 2 3 4 5 6 7
Time (days)
Figure 10. The isothermal calorimetry results of CEM I pastes (a) up to 24 h, and (b): up to 7 d
(a) 450 10

400 9

Heat flow (mW/g CEM III)

350 8
Heat (J/g CEM III)

7
300
III-CTRL
III-QP 6
250 III-350/2
III-550/2 5
200 III-550/24
III-750/2 4
150
3
100 2
50 1
0 0
0.5 4.5 8.5 12.5 16.5 20.5 24.5
Time (h)
(b) 900 10

800 9

Heat flow (mW/g CEM III)

700 8
Heat (J/g CEM III)

7
600
6
500
5
400
4
300 III-CTRL
III-QP 3
III-350/2
200 III-550/2 2
III-550/24
100 III-750/2 1
0 0
0 1 2 3 4 5 6 7
Time (days)
Figure 11. The isothermal calorimetry results of CEM III pastes (a) up to 24 h, and (b): up to 7 d
 850
Mixture type Age (days) CEMI - 1d
Calorimetry heat release (J/g CEMI)
1 3 7
0.49 0.9 0.92 CEMI - 3d
CEMI / DCI
(0.51) (0.1) (0.08)
750 0.23 0.82 0.96
CEMIII / DCIII
(0.77) (0.18) (0.04)
CEMI - 7d
CEMIII - 1d
650
CEMIII - 3d
CEMIII - 7d
550
Linear (CEMI - 1d)
Linear (CEMI - 3d)
450
Linear (CEMI - 7d)
Linear (CEMIII - 1d)
350
Linear (CEMIII - 3d)
Linear (CEMIII - 7d)
250
50 100 150 200 250 300
R3 heat release (J/g DCP)
Figure 12. The isothermal calorimetry versus R3 heat release results for CEM I and CEM III pastes
at different ages (the embedded table shows the correlation coefficients between calorimetry and
R3 heat release and corresponding p-values at different test durations)
(a) (b)

(c) (d)

Figure 13. The TGA and DTG curves of the CEM I pastes. (a) 3 d, (b) 7 d, (c) 28 d, (d) 91 d
(a) (b)

(c) (d)

Figure 14. The TGA and DTG curves of the CEM III pastes. (a) 3 d, (b) 7 d, (c) 28 d, (d) 91 d
(a)
35
3d 7d 28 d 91 d
Bound water (wt% at 550 °C)

30

25

20

15

10

0
I-CTRL I-QP I-350/2 I-550/2 I-550/24 I-750/2
Mixture label
(b)
25
3d 7d 28 d 91 d
Portlandite wt% at 550 °C)

20

15

10

0
I-CTRL I-QP I-350/2 I-550/2 I-550/24 I-750/2
Mixture label
Figure 15. The bound water (a) and portlandite content (b) values of the CEM I pastes (values
normalised to the CEM I content in the pastesbinder content)
(a)
35
3d 7d 28 d 91 d
Bound water wt% at 550 °C) 30

25

20

15

10

0
III-CTRL III-QP III-350/2 III-550/2 III-550/24 III-750/2
Mixture label
(b)
12
3d 7d 28 d 91 d
Portlandite (wt% at 550 °C)

10

0
III-CTRL III-QP III-350/2 III-550/2 III-550/24 III-750/2
Mixture label
Figure 16. The bound water (a) and portlandite content (b) values of the CEM III pastes (values
normalised to the binder content)
(a)

(b)

Figure 17. The in-situ XRD pattern of I-350/2 paste over time. (a): 8 – 15°, (b): 33 – 41°; A: alite, B:
belite, E: ettringite, F: ferrite, H: hemicarboaluminates, P: portlandite
(a)

(b)

Figure 18. The ex-situ XRD pattern of (a) I-350/2 paste, and (b): III-350/2 pastes over time; A: alite,
AF: AFm phases, B: belite, C: C3A; E: ettringite, F: ferrite, H: hemicarboaluminates, M:
monocarboaluminates; P: portlandite, Q: quartz, T: hydrotalcite
(a)

(b)

Figure 19. The ex-situ XRD patterns of (a) CEM I series, and (b): CEM III series at 28 d; A: alite, AF:
AFm phases, B: belite, C: C3A; E: ettringite, F: ferrite, H: hemicarboaluminates, M:
monocarboaluminates; P: portlandite, Q: quartz, T: hydrotalcite
(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 20. The evolution of primary phases in the CEM I pastes over time (left: first 24 h; in-situ,
right: up to 91 d; ex-situ). (a and b) C3S degree-of-reaction, (c and d) clinker degree-of-reaction, (e
and f) portlandite content, (g and h) ettringite
(a) (b)

Figure 21. The evolution of gypsum (primary peak) in the in-situ XRD diffractograms. (a) I-CTRL,
(b) I-350/2
(a) (b)

Figure 22. The compressive strength results of the CEM I mortars. (a) absolute compressive
strength results, (b) relative strength ratios with respect to the strength of the control mixtures at
the same age
(a) (b)

Figure 23. The compressive strength results of the CEM III mortars. (a) absolute compressive
strength results, (b) relative strength ratios with respect to the strength of the control mixtures at
the same age
 40
I-350/2 I-550/2
35
7-d Compressive strength (MPa)

30
I-550/24 III-550/2
III-350/2
III-550/24
25
I-750/2

20
III-750/2
15

10 CEM I series
CEM III series
5 Linear (CEM I series)
Linear (CEM III series)
0
450 500 550 600 650 700 750 800

7-d calorimetry heat release (J/g CEM I)

Figure 24. The 7-d compressive strength of blended CEM I and CEM III pastes versus the 7-d
calorimetry heat release
Highlights Click here to view linked References

Highlights

 Thermal treatment of hydrated cement paste at 550 °C and higher resulted in formation of β-
and γ-C2S polymorphs.
 The higher the treatment temperature, the higher the R3 test reactivity as SCM.
 Thermally treated hydrated cement strongly increased initial heat release and shortened
induction periods.
 Cement replacement by very reactive thermally activated cement paste can disturb the
cement hydration process, leading to early set and low strength performance.
 Best strength performance observed for treatment at 350 °C for 2 hours, higher contribution
to strength by thermally treated slag cement than neat Portland cement.
Credit Author Statement Click here to view linked References

Author contributions
Asghar Gholizadeh-Vayghan: Conceptualization, Investigation, Writing – Original draft preparation,
Guillermo Meza Hernandez: Investigation, Felicite Kingne Kingne: Investigation, Jun Gu:
Investigation, Resources Nicole Dilissen: Investigation, Writing – Review and Editing, Michael El Kadi:
Writing – Review and Editing, Tine Tysmans: Supervision, Jef Vleugels: Writing – Review and Editing,
Hubert Rahier: Supervision, Project administration, Funding acquisition Ruben Snellings:
Conceptualization, Supervision, Investigation, Writing - Review and Editing
Declaration of Interest Statement Click here to view linked References

Declaration of interests

☒The authors declare that they have no known competing financial interests or personal relationships
that could have appeared to influence the work reported in this paper.

☐The authors declare the following financial interests/personal relationships which may be considered
as potential competing interests: