Calcium Aluminate Blended

Calcium Aluminate Blended
Cements Incorporating
Engineered Residues
Jennifer ASTOVEZA
Main supervisors:
Prof. dr. ir. Y. Pontikes
Prof. dr. ir. A. Lecomte
Chairman:
Prof. R. Puers
Reporters:
Prof. S.A.B. Lopez
Dr. ir. V. Baroghel-Bouny
Other members of the jury:

Prof. dr. ir. R. Trauchessec (co-supervisor)
Dr. ir. R. Soth (co-supervisor)
Dr. ir. R. Snellings
Prof. J. Elsen
Dissertation presented in partial

fulfilment of the requirements for the
degree of Doctor of Engineering
Science (PhD): Materials Engineering
June 2022
École Doctorale Chimie Mécanique Matériaux Physique
Thèse
« sciences des matériaux »
Ciments composés à base d'aluminate de calcium

incorporant des résidus traités valorisables
Par: Jennifer ASTOVEZA
Thèse soutenue publiquement le 15 juin 2022 à Louvain, Belgique

devant le jury suivant :
Directeurs de thèse:
Yiannis Pontikes Professeur des universités
André Lecomte Professeur des universités
Examinateurs:
Romain Trauchessec Maître de conférences (co-directeur)
Ratana Soth Docteur (co-directeur)
Jan Elsen Professeur des universités
Ruben Snellings Docteur
Rapporteures:
Susan Andrea Bernal Lopez Professeure des universités
Véronique Baroghel-Bouny Docteur HDR
Invité:
Robert Puers Professeur des universités (président du jury)
This manuscript is presented in partial fulfilment of the requirements for a dual
PhD from the Faculty of Engineering Science of KU Leuven, and the Institut Jean
Lamour of the Université de Lorraine.
This work was funded by the SOCRATES Project under the European Union
Framework Program for Research and Innovation Horizon 2020 Grant Agreement
No.721385 (EU MSCA-ETN SOCRATES).
© 2022 KU Leuven Faculty of Engineering Science

Uitgegeven in eigen beheer, Jennifer Astoveza, 1 Rue Le Chatelier, Vaulx-Milieu
38090 (France)
Alle rechten voorbehouden. Niets uit deze uitgave mag worden vermenigvuldigd
en/of openbaar gemaakt worden door middel van druk, fotokopie, microfilm,
elektronisch of op welke andere wijze ook zonder voorafgaandelijke schriftelijke
toestemming van de uitgever.
All rights reserved. No part of the publication may be reproduced in any form by
print, photoprint, microfilm, electronic or any other means without written
permission from the publisher.
ISBN 9789464447224
___________________________________________Acknowledgements
Acknowledgements
This work would not have been possible without the contribution of the
following people/organizations:
the team behind the SOCRATES Project for the conception and
funding of this research
Imerys, KU Leuven and Université de Lorraine for the warm
accommodation and support
my thesis advisers, Prof. Yiannis Pontikes, Prof. André Lecomte,
Prof. Romain Trauchessec, and Dr. Ratana Soth for their relentless
supervision, and for always believing in my abilities especially
whenever I am overwhelmed with self-doubt
The members of the jury: Prof. Robert Puers, Prof. Jan Elsen, Dr.
Véronique Baroghel-Bouny, Prof. Susan Andrea Bernal Lopez and
Dr. Ruben Snellings for their constructive review
Dr. Chris Parr and Dr. Christoph Wöhrmeyer of Imerys for offering
me the opportunity to join their team, while providing continuous
support for the completion of my thesis
I would also like to acknowledge several colleagues for their direct
contribution to the experimental works and data analyses, namely:
Dr. Jeroen Soete of KU Leuven for his assistance in tomography and
for performing the analysis on cylindrical volumes in section 3.2.2
Dr. Arne Peys and Christina Siakati of KU Leuven for fitting the
Mössbauer spectra of the raw materials (section 2.2.5) and the
hydrated binary blends (section 3.4.3) using Model 1
Sébastien Diliberto of Institut Jean-Lamour for performing the
Mössbauer spectroscopy and for fitting the spectra in section 3.4.3
using Model 2
Sylvie Migot-Choux and Mélanie Emo of Institut Jean-Lamour for
performing with us the TEM and STEM, respectively
Dr. Tobias Hertel of KU Leuven for his assistance in ATR-FTIR
Dr. Bruno Espinosa of Imerys for the encouragement and for sharing
his insights on the fate of Fe during hydration
my supportive friends and colleagues at Imerys, KU Leuven and
Université de Lorraine for the training and insightful exchanges
Lastly, I would like to express my gratitude to my friends, my second
parents (Michèle & Daniel, Maria & Guido, Zuri & Gonza) and my family
(Nordin, Moises, Everlyn, Glaiza and Maycee). Thank you for choosing to
be part of my life, and for embracing my best and worst.
ASTOVEZA 2022 i
_____________________________________________________ Abstract
Abstract
The pressure to decarbonize the cement industry in light of sustainability
goals has been motivating the search for new types of supplementary
cementitious materials (SCM) in the recent decades. Industrial residues,
which otherwise end up in landfills or find low added-value applications,
are of particular interest in near-zero-waste process schemes.
This thesis explored the potential for valorization of pre-treated residues
as SCM in calcium aluminate cement (CAC)-calcium sulfate hemihydrate
(HH) system. These residues included non-ferrous metallurgy (NFM)
slags, municipal solid waste incinerator (MSWI) bottom ashes, jarosite and
paper-fabrication residues. Using isothermal calorimetry and mechanical
strength test, a screening test was first established to compare their early-
age and the long-term reactivity. The highly-amorphous NFM slags
exhibited superior reactivity showing evidence of long-term (> 28 days)
contribution to the mechanical strength of mortars at 30 wt% replacement
level. However, the slow hydration kinetics in the first days of
hydration could be an issue for CAC systems, given that their applications
greatly rely on their rapid-hardening property. Moreover, the amorphous
nature of the NFM slags and the dominance of iron (Fe) in their
composition, set them apart from traditional SCM . As of this writing,
there are only limited publications discussing the hydration mechanism of
Fe in CAC-based binders.
In order to overcome the slow kinetics of the slag hydration, the influence
of three factors, including: (1) the proportion of sulfates by varying the
CAC/HH ratio; (2) the addition of lime-rich paper residue at 5 wt% as
activator; and (3) the incorporation of 30 wt% PC in ternary CAC-HH-PC
systems, were investigated. For all these formulations, the early hydration
reactions were characterized using isothermal calorimetry on paste
samples. The dimensional stability and mechanical strength were followed
using standard mortars up to 1 year curing period. Furthermore, a
methodology to estimate the slag hydration degree using a non-destructive
technique based on X-ray Computed Tomography (XCT) combined with
volume analysis was developed in this study. Several techniques were used
to follow the phase assemblage evolution, including: X-ray diffraction
(XRD), thermogravimetric analysis (TGA), attenuated total reflection-
Fourier transform infrared spectroscopy (ATR-FTIR). Lastly, transmission
electron microscopy (TEM), scanning transmission electron microscopy
(STEM) and Mössbauer spectroscopy were also performed in order to gain
a better understanding of the fate of Fe, the main component of the slag,
during hydration.
ASTOVEZA 2022 iii

Abstract _____________________________________________________
High early strengths (>20 MPa) were obtained for all formulations at
30 wt% slag-binder replacement level. The increased proportion of sulfates
(CAC/HH ratio = 1.6) improved the mechanical strength due to the more
extensive ettringite formation. The addition of PC and paper residue also
improved the strengths at early ages. The slag addition improved the
dimensional stability of mortars by diluting the binder, consequently
mitigating the excessive expansion in the ternary systems. The XCT results
showed that about 50 % of the slag has reacted after more than 180 days
of curing. This was in agreement with the estimations obtained from
Mössbauer spectroscopy, based on the relative amounts of absorption areas
of the newly-formed Fe species during the hydration. While the proportion
dissolution rate
based on the XCT results, the addition of PC and paper residue appeared
to have accelerated it in the first 28 days. At longer term however, the
reaction degree remained relatively comparable to those obtained for the
binary CAC-HH systems. Using the ATR-FTIR spectra, a hypothetical
peak displacement of the Si-O asymmetric stretching band of the slag at
around 930 cm-1 was proposed, supporting this reactivity.
As to the fate of Fe, TEM showed evidence of uptake in hydrated phases
grouped into two compositional clusters when plotted on a Ca-Si-Al
ternary diagram. These included: (1) an Al-rich intermixed phase for the
binary systems; and (2) an amorphous phase mainly containing Ca, Si and
Al for the ternary systems. After longer curing periods (>180 days), STEM
revealed the formation of an unidentified nano-crystalline phase richer in
Fe with a distinct rod/needle-like morphology and a d-spacing of
approximately 7.2 Å. The evolution of the Mössbauer spectra could
suggest Fe oxidation from the 2+ to 3+ state during slag hydration. This
was accompanied by a shift from tetrahedral to octahedral coordination as
a new Fe-containing hydrated phase is formed. There was no substantial
evidence of Fe uptake in the more abundant hydrated phases, such as in
ettringite or monosulfoaluminates, contrary to the models obtained from
thermodynamic simulations and synthetic systems.
To conclude, the overall properties of the blends containing the industrial
residues support their potential for valorization as SCM in CAC-based
systems. However, it will be critical to continue this investigation to clearly
identify the Fe-bearing hydrated phase/s and to, ultimately, utilize the
understanding of the hydration mechanism in improving
performance.
iv ASTOVEZA 2022
________________________________________________ Samenvatting
Samenvatting
De druk om de cementindustrie koolstofarm te maken in het licht van
duurzaamheidsdoelstellingen heeft de afgelopen decennia de zoektocht naar
nieuwesupplementaire cementerende materialen. (SCM) gemotiveerd.
Industriële residuen, die anders op stortplaatsen terechtkomen of enkel
toepassingen kennen met een lage toegevoegde waarde, zijn van bijzonder
belang in bijna-nul-afvalprocesschema's.
Dit proefschrift onderzocht het potentieel voor valorisatie van behandelde
residuen als SCM in calciumaluminaatcement (CAC)-calciumsulfaat
hemihydraat (HH) systeem. Deze residuen omvatten non-ferrometallurgie
(NFM) slakken, bodemas van de gemeentelijke afvalverbrandingsoven
(AEC), jarosiet en papierfabricageresiduen. Met behulp van isotherme
calorimetrie en mechanische sterktetest werd eerst een screeningstest
uitgevoerdom hun reactiviteit op jonge en oude leeftijd te vergelijken. De zeer
amorfe NFM-slakken vertoonden een superieure reactiviteit, waaruit blijkt dat
ze op lange termijn (> 28 dagen) bijdragen aan de mechanische sterkte van
mortels bij een vervangingsniveau van 30 gew.%. De langzame
hydratatiekinetiek van de slakken in de eerste dagen van hydratatie kan echter
een probleem zijn voor CAC-systemen, aangezien hun toepassingen sterk
afhankelijk zijn van hun snelle uithardingseigenschappen. Bovendien
onderscheiden de amorfe aard van de NFM-slakken en de dominantie van ijzer
(Fe) in hun samenstelling ze van traditionele SCM's. Op het moment van
schrijven zijn er slechts beperkte publicaties die het hydratatiemechanisme
van Fe in op CAC gebaseerde bindmiddelen bespreken.
Om de langzame kinetiek van de hydratatie van de slak te overwinnen zijn drie
factoren onderzocht, waaronder: (1) het gehalte aan sulfaten door de
CAC/HH-verhouding te variëren; (2) de toevoeging van 5 gew.% kalkrijk
papierresidu als activator; en (3) de opname van 30 gew.% PC in ternaire
CAC-HH-PC-systemen. Voor alle formuleringen werden de vroege
hydratatiereacties gekarakteriseerd met behulp van isotherme calorimetrie op
pastamonsters. De maatvastheid en mechanische sterkte van standaardmortels
werd gemonitord over een uithardingsperiode van 1 jaar. Verder werd in deze
studie een methodologie ontwikkeld om de hydratatiegraad van de slak te
schatten met behulp van een niet-destructieve techniek op basis van X-stralen
computertomografie (XCT) in combinatie met volume-analyse. Verschillende
technieken werden gebruikt om de evolutie van de fase-assemblage te volgen,
waaronder: röntgendiffractie (XRD), thermogravimetrische analyse (TGA),
verzwakte totale reflectie-Fourier-transformatie infraroodspectroscopie
(ATR-FTIR). Ten slotte werden ook transmissie-elektronenmicroscopie
(TEM), scanning transmissie-elektronenmicroscopie (STEM) en Mössbauer-
spectroscopie uitgevoerd om een beter begrip te verwervenover het lot van Fe,
het hoofdbestanddeel van de slak, gedurende de hydratatie.
ASTOVEZA 2022 v
Samenvatting ________________________________________________
Voor alle formuleringen met een vervangingsniveau van 30 gew.% NFM slak
werd een hoge vroege sterkte bekomen. Het verhoogde gehalte aan sulfaten
(CAC/HH-verhouding = 1,6) verbeterde de mechanische sterkte door een
verhoogde ettringiet vorming. De toevoeging van PC en papierresten
verbeterde ook de sterkte op jonge leeftijd. De toevoeging van slakken
verbeterde de maatvastheid van mortels door het bindmiddel te verdunnen,
waardoor de overmatige uitzetting in de ternaire systemen werd verminderd.
De XCT-resultaten toonden aan dat ongeveer 50 gew.% van de slak reageerde
na meer dan 180 dagen.. Dit was in overeenstemming met de schattingen
verkregen met Mössbauer-spectroscopie, gebaseerd op de relatieve
hoeveelheden absorptiegebieden van de nieuw gevormde Fe-soorten tijdens
de hydratatie. Hoewel het gehalte aan sulfaten geen duidelijke invloed had op
de oplossnelheid van de slak op basis van de XCT-resultaten, leek de
toevoeging van PC en papierresten dit in de eerste 28 dagen te hebben
versneld. Op langere termijn bleef de reactiegraad echter relatief vergelijkbaar
met die verkregen voor de binaire CAC-HH-systemen. Met behulp van de
ATR-FTIR-spectra werd een hypothetische piekverplaatsing van de Si-O
asymmetrische rekband van de slak bij ongeveer 930 cm-1 waargenomen, wat
deze reactiviteit ondersteunt.
Wat het lot van Fe betreft, leverde TEM bewijs aan voor de opname in
gehydrateerde fasen gegroepeerd in twee samenstellingsclusters op een Ca-Si-
Al ternair diagram. Deze omvatten: (1) een Al-rijke vermengde fase voor de
binaire systemen; en (2) een amorfe fase die voornamelijk Ca, Si en Al bevat
voor de ternaire systemen. Na langere uithardingsperioden (>180 dagen),
onthulde STEM de vorming van een niet-geïdentificeerde nanokristallijne fase
rijker aan Fe met een duidelijke staaf/naaldachtige morfologie en een d-
afstand van ongeveer 7,2 Å. De evolutie van de Mössbauer-spectra zou
kunnen wijzen op de oxidatie status van Fe (van 2+ naar 3+) tijdens de
hydratatie van de slak. Dit ging gepaard met een verschuiving van
tetraëdrische naar octaëdrische coördinatie als een nieuwe Fe-bevattende
gehydrateerde fase werdgevormd. Er was geen substantieel bewijs van Fe-
opname in de meer overvloedige gehydrateerde fasen, zoals in ettringiet of
monosulfoaluminaten, in tegenstelling tot de modellen verkregen uit
thermodynamische simulaties en synthetische systemen.
Globaal gezien, de resultaten in dit werk ondersteunen het
valorisatiepotentieel van industiële residuen als SCM in op CAC gebaseerde
systemen. Het zal echter van cruciaal belang zijn om dit onderzoek voort te
zetten om de Fe-dragende gehydrateerde fase(s) duidelijk te identificeren en
om uiteindelijk het begrip van het hydratatiemechanisme te gebruiken bij het
verbeteren van de prestaties van de mengsels.
vi ASTOVEZA 2022
______________________________________________________ Résumé
Résumé
Au vu des objectifs de durabilité visant à décarboniser l'industrie du
ciment, les liants alternatifs ont
l'échelle mondiale au cours des dernières décennies. Ces liants sont basés
-produits industriels comme substituts aux ciments,
ou comme granulats ou encore comme composants de matériaux de
construction, à performances équivalentes.
Dans le cadre de cette thèse, des résidus traités valorisables sont évalués
comme addition cimentaire notamment dans des mélanges binaires
(ciment alumineux CAC - hémihydrate HH). Trois des sous-produits ont
fait l'objet d'une évaluation plus approfondie : i) un laitier riche en fer
provenant d'une usine pilote de plomb-zinc sous forme granulée (trempe)
ii) un résidu de jarosite ; iii) un résidu du recyclage du papier. Pour
identifier et sélectionner ces résidus, il a été réalisé un test de
conductimétrie pour mesurer la cinétique de dissolution; des mesures de
flux de chaleur pour apprécier les réactions au jeune âge; et une mesure de
résistance mécanique sur mortiers standards pour caractériser la réactivité
à
détaillée de la réactivité. Les mortiers standards incorporant de laitiers ont

présenté une réactivité supérieure montrant d'une contribution positive à
long terme (> 28 jours) à la résistance mécanique. Cependant, il existe
actuellement peu de publications sur le mécanisme d'hydratation du fer
dans les liants à base de CAC.
Afin d'augmenter la cinétique d'hydratation du laitier, l'influence de trois
facteurs principaux, dont : (1) l'ajout de résidu de papier riche en chaux
comme activateur ; (2) la proportion de sulfates; et (3) l'ajout de ciment
Portland (PC), ont été évalués. Plusieurs techniques ont été utilisées pour
caractériser les propriétés des mélanges entre 1 jour et 1 an : La résistance
mécanique et la stabilité dimensionnelle ont été suivies en utilisant des
éprouvettes de mortier. Le degré d'hydratation du laitier a été quantifié à
l'aide de la microscopie électronique à balayage (MEB), de la tomographie
par rayons X (XCT) et de la spectroscopie Mössbauer. L'assemblage des
phases en fonction de la proportion de CAC a été simulé à l'aide de la
modélisation thermodynamique et plusieurs techniques, notamment la
calorimétrie isotherme, la diffraction des rayons X (DRX), l'analyse
thermogravimétrique (ATG) et la spectroscopie infrarouge à transformée
de Fourier (IRTF) ont été utilisées pour suivre expérimentalement
l'évolution des assemblages. Le mécanisme d'hydratation du fer contenu
dans le laitier a été décrit à partir des résultats de la spectroscopie
ASTOVEZA 2022 vii

Résumé ______________________________________________________
Mössbauer, de la microscopie électronique à transmission (MET) et de la

microscopie électronique à transmission à balayage (METB).
La caractérisation approfondie des propriétés physiques et chimiques des
ciments composés incorporant des résidus a démontré leur potentiel de
valorisation dans cette application. Des mortiers de haute résistance et
stables dimensionnellement ont été obtenus en incorp
laitier amorphe riche en fer dans des liants binaires CAC-HH et ternaires
CAC-HH-PC. La méthodologie basée sur la tomographie a permis de
-delà de 180 jours. Ce
résultat a été confirmé par spectroscopie Mössbauer, qui a montré
également que l'hydratation du laitier s'accompagnait de l'oxydation du fer
2+ à 3+, et de la formation de nouveaux hydrates contenant du fer. À l'aide
de la microscopie électronique (MET et METB), le fer a été détecté au
jeune âge dans des phases complexes formées de gels amorphes riches en
aluminium. Le fer a également été détecté à des proportions plus élevées à
plus long terme, au sein une phase nanocristalline non identifiée. Dans
les systèmes ternaires, cette phase était également associée à du calcium,
du silicium et une petite quantité de magnésium.
En ce qui concerne l'augmentation de la cinétique d'hydratation du laitier,
les résultats ont montré qu'il n'y avait pas de différences substantielles entre
les systèmes à faible et à haute teneur en sulfate étudiés. Au contraire,
l'ajout de ciment Portland et de résidus du recyclage du papier ont montré
une augmentation du degré d'hydratation, toutefois cantonnée au jeune âge
2+
(< 28 jours). Cet effet pourrait résulter
dans le système, ainsi que de l'augmentation de l'alcalinité favorisant la
dissolution du laitier.
industriels traités valorisables et son impact positif sur les propriétés des
ciments composés à base de CAC. Néanmoins, il reste encore des
questions importantes en suspens, tant pour expliquer les mécanismes
fondamentaux de l'hydratation que pour tester la durabilité de ces liants.
Ainsi, les travaux futurs sur l'ajout de laitier devraient se concentrer sur des
analyses complémentaires pour identifier la phase nanocristalline
contenant du fer.
viii ASTOVEZA 2022

____________________________________ Thesis Summary in French
Ciments composés à base d'aluminate de

calcium incorporant des résidus traités
valorisables
Jennifer ASTOVEZA
Introduction
Le projet européen MSCA-ETN SOCRATES a pour objectif de
promouvoir l'économie circulaire en créant des procédés de valorisation
des résidus industriels qui génèrent le moins possible de nouveaux déchets.
Comme le montre la Figure 1, la technologie SOCRATES vise tout
toxiques. La fraction minérale restante, qui représente plus de 80-90% du

volume total, est ensuite valorisée comme précurseur de géopolymères,
dernière voie de valorisation qui est étudiée dans le cadre de cette thèse.
Les additions prétraitées devant être ajoutées au ciment sont appelées
résidus traités valorisables ou encore « sous-produits SOCRATES ».
Figure 1 Schéma simplifié du projet SOCRATES axé sur les travaux de l'auteure:
valorisation des résidus comme additions cimentaires dans les ciments composés.
ASTOVEZA 2022 ix
Thesis Summary in French _____________________________________
-produits comme substituts aux ciments, ou

comme granulats ou encore comme composants de matériaux de
avantages économiques et environnementaux [1,2]. Si les résidus du projet

SOCRATES s'avéraient être des matériaux qui satisfont ces critères, alors
les retombées de ce projet de valorisation constitueraient une avancée
matériaux de construction visés [3]:

2 anthropique mondial est attribué à la fabrication du
ciment;
dans le monde;
-50 % de la demande mondiale de
béton est prévu
Au vu de ces chiffres, on comprend pourquoi les liants alternatifs ont fait
dernières décennies. Néanmoins, pour contrôler le comportement physique

et chimique des liants incorporant des additions, il reste de nombreux défis
de recherche à relever, non seulement selon le type de résidu utilisé mais
aussi selon le type de liant visé. Dans la littérature, la majorité des travaux
traditionnelles (i.e. laitiers de
hauts-fourneaux, cendres volantes, calcaire, etc.) dans des liants
majoritairement composés de ciment Portland [4,5].
Les résidus du projet SOCRATES ont des propriétés chimiques différentes
de celles des additions traditionnelles. En effet, ces dernières sont
principalement constitués de CaO, SiO2 et Al2O3, alors que les laitiers issus
des métaux non-
SOCRATES sont beaucoup plus riches en fer ou en sulfates,
respectivement. Les différences chimiques entre les résidus du projet
SOCRATES, les additions traditionnelles et les ciments Portland (PC) ou
Alumineux (CAC) sont illustrées dans le diagramme de composition de la
Figure 2.
Parmi les résidus du projet, les laitiers issus des métaux non-ferreux sont
ciment car leur nature hautement amorphe est généralement reconnue

comme un critère déterminant de la réactivité de ce type de produits [10].
De plus, ces laitiers ont subi un prétraitement à haute température de sorte
que les métaux lourds et les composants toxiques y ont été pratiquement
présent d'études traitant de la réactivité et du mécanisme d'hydratation des
x ASTOVEZA 2022
laitiers riches en oxyde de fer (FeO-Fe2O3) dans les systèmes à base de

ciment d'aluminate de calcium (CAC).
Figure 2 Compositions chimiques de certains résidus de SOCRATES (orange) ; additions

traditionnelles (gris) et ciments (vert). Les composants *MgO et *Fe2O3 sont
exclusivement pris en compte pour les résidus SOCRATES (orange). Les données ont été
rassemblées pour créer une version modifiée des diagrammes ternaires rapportés dans
[6 9].
CAC est un terme générique adopté pour les ciments à haute teneur en
2O3) avec l'aluminate
monocalcique (CA) comme composant principal [11]. Dans l'industrie du
bâtiment, les CAC sont généralement associés en proportions variables au
ciment Portland et au sulfate de calcium (liants ternaires). Ils sont utilisés
dans des applications spéciales telles que les enduits de réagréage ou
enduits de lissage, les liants pour matériaux réfractaires ou les applications
en environnements agre
analyse du cycle de vie appliquée aux chapes autonivelantes [12] a montré
CAC que pour les liants riches en ciment Portland, tout en maintenant le
même niveau de performances. À ce jour, il existe peu d'articles
les liants à base de CAC. Les travaux de recherche publiés se limitent aux
additions traditionnelles, principalement les laitiers de hauts-fourneaux et
les cendres volantes [13 16].
étudiant notamment, sur le plan fondamental et appliqué, les mécanismes

d'hydratation propres à ces systèmes. La connaissance approfondie de
ASTOVEZA 2022 xi
l'influence de divers paramètres (formules, minéralogie, etc.) sur les

performances de ces liants composés constitue une étape clé qui servira par
ES
Objectifs et Problématique
Deux objectifs principaux ont été définis dans le cadre du projet
SOCRATES :
1. Évaluer la réactivité des sous-produits SOCRATES dans les
systèmes à base de CAC ;
2. Proposer des techniques d'activation de ces sous-produits afin
d'améliorer la cinétique de leur réaction, et les performances
associées.
Guidée par ces objectifs, une méthodologie a été mise en place afin de
repérer le résidu ayant le potentiel de valorisation le plus élevé. L'évolution
des propriétés physiques et chimiques du résidu sélectionné au cours de
l'hydratation a ensuite été étudiée en détail.
En effet, l'hydratation des liants ternaires (ciment alumineux - ciment

Portland - hémihydrate HH) engendre des réactions complexes souvent
difficiles à dissocier car elles se produisent simultanément. De plus,
l'origine variée des résidus du projet SOCRATES peut engendrer des
compositions irrégulières et des chimies très différentes, de sorte qu'une
comparaison directe de leur réactivité n'est pas toujours applicable, même
à volume égal de substitution. Enfin, l'abondance du fer et le caractère
amorphe de certains résidus, notamment les laitiers issus de métaux non-
ferreux, nécessitent des techniques de caractérisation complémentaires aux
donné que cette étude implique l'utilisation de matières premières réelles

plutôt que des produits purs de synthèse, il faut aussi s'attendre à ce que
certaines propriétés ne puissent pas être directement reliées à une réaction
pouvoir analyser leur comportement dans des systèmes complexes, de

sorte que, le cas échéant, une perspective industrielle puisse être
directement proposée pour ces liants plus écologiques.
xii ASTOVEZA 2022

Matériaux et Méthodologie
Six sous-produits SOCRATES ont été étudiés : un laitier riche en fer
provenant d'une usine pilote de plomb-zinc, disponible sous forme
de fayalite (produit prétraité) ; une cendre d'incinération de déchets

municipaux ; un résidu de jarosite ; un résidu du recyclage du papier. Sur
d'entre eux ont fait l'objet d'une évaluation plus approfondie. Ces trois
résidus sont représentés Figure 3.
Figure 3 Les principaux sous-produits SOCRATES évalués dans cette étude, notamment :
le laitier riche en fer granulé (a) ; le résidu jarosite (b) ; et les résidus du recyclage du
papier (c)
Pour identifier et sélectionner ces résidus, il a été réalisé : i) un test de

conductimétrie pour mesurer la cinétique de dissolution ; ii) des mesures
de flux de chaleur pour apprécier les réactions au jeune âge ; iii) une
mesure de résistance mécanique sur mortiers standards pour caractériser la
réactivité à long terme. En utilisant la fluorescence X (FX) et l'analyse par
que le laitier granulé de fayalite est composé de 46 % de FeO. La fraction

amorphe représente 92 % du produit, les 8 % restant étant de la wüstite. Ce
réactivité. Les résidus de jarosite et ceux issus du recyclage du papier sont

principalement composés de sulfates (51 %) et de chaux (68 %),
respectivement. Ils ont été étudiés en tant que sources de sulfates ou de
chaux dans des études distinctes.
Les principales formules incorporant le laitier de fayalite sont présentées
dans le Tableau 1. Les formules A et B correspondent à un ajout du laitier
dans des mélanges binaires CAC-HH, tandis que pour les liants A-PC et
B-PC, l'ajout est réalisé dans des mélanges ternaires CAC-HH-PC.
L'utilisation de deux rapports CAC/HH - 4,5 (formules A) et 1,6 (formules
B) est destinée à étudier la réactivité dans des mélanges respectivement
à faible et à forte teneur en sulfate de calcium. De plus, pour chaque
ition (le
ASTOVEZA 2022 xiii

laitier) est remplacée par un volume égal de quartz broyé. Les pâtes de
ciment ont été préparées avec un rapport eau/liant (e/l) de 0,65 et stockées
sous enveloppes hermétiques à 20°C. Les mortiers standards ont un rapport
e/l de 0,5 et ils ont été conservés sous eau à 20°C.
Tableau 1 Composition de liants incorporant le laitier riche en fer granulé

somme
(en % massiques) laitier CAC HH PC CAC/HH
code liant
A 4.5CAC-HH 30 57 13 0 100 4.5
B 1.6CAC-HH 30 43 27 0 100 1.6
4.5CAC-HH-
A-PC 30PC 30 33 7 30 100 4.5
1.6CAC-HH-
B-PC 30PC 30 25 15 30 100 1.6
formulations de référence :
REFciment (référence ciment) pas d'ajout de sous-produits ni de quartz
(référence quartz) laitier remplacé par un volume égal de
REFquartz quartz
Plusieurs techniques ont été utilisées pour caractériser les propriétés des
mélanges (entre 1 jour et 1 an), et certaines ont été spécifiquement adaptées
aux propriétés particulières du laitier et à sa teneur élevée en fer (i.e.
densité élevée, masse atomique élevée, effet Mössbauer important) :
la résistance mécanique et la stabilité dimensionnelle ont été
suivies en utilisant des éprouvettes de mortier;
le degré d'hydratation du laitier a été quantifié à l'aide de la
microscopie électronique à balayage (MEB), de la tomographie
par rayons X (XCT) et de la spectroscopie Mössbauer;
l'assemblage des phases en fonction de la proportion de CAC a été
simulé à l'aide de la modélisation thermodynamique et plusieurs
techniques, notamment la calorimétrie isotherme, la diffraction des
rayons X (DRX), l'analyse thermogravimétrique (ATG) et la
spectroscopie infrarouge à transformée de Fourier (IRTF) ont été
utilisées pour suivre expérimentalement l'évolution des
assemblages;
le mécanisme d'hydratation du fer contenu dans le laitier a été
décrit à partir des résultats de la spectroscopie Mössbauer, de la
microscopie électronique à transmission (MET) et de la
microscopie électronique à transmission à balayage (METB).
xiv ASTOVEZA 2022

Résultats et Discussion
Pour un taux de remplacement massique de 30 % de laitier issu de métaux
non-ferreux, l'indice d'activité (IA) déduit de la résistance mécanique sur
étude a été faite avec un taux

d'incorporation du laitier de 30 % dans les liants binaires CAC-HH et
ternaires CAC-HH-PC.
Propriétés physiques
Les mortiers incorporant le laitier issu de métaux non-ferreux granulé et
1,3 % d'acide citrique (Figure 4) ont atteint des valeurs de résistance à la
jours. La faible quantité d'acide citrique, initialement ajouté pour retarder

la prise, a également contribué à améliorer les propriétés au jeune âge.
été obtenues dans les systèmes à haute teneur en sulfate, grâce à la

formation plus importante d'ettringite. Mais le caractère expansif de
otamment dans la formule de
référence du système ternaire (avec ciment Portland) où une expansion
excessive a finalement conduit à des macro-fissures (Figure 5) et à une
perte de résistance des mortiers. L'ajout du laitier dans les systèmes
binaires et ternaires a limité cette expansion. Cette addition améliore donc
la stabilité dimensionnelle du liant (Figure 5). Cet effet est probablement
gouverné par un effet de dilution au jeune âge (facteur physique), et par la
nature de phases hydratées formées à plus long terme (contribution
chimique du laitier).
Figure 4 Résistance à la compression des mortiers incorporant le laitier
ASTOVEZA 2022 xv
Figure 5 Photos des mortiers B et B-PC après 45 jours de conservation sous eau
Quantification du degré d'hydratation du laitier

Une méthodologie pour estimer le degré d'hydratation du laitier a été
développée par analyse des images issues de la tomographie à rayons X
(XCT). Les principales étapes déployées sont illustrées à la Figure 6. Cette
technique a pu être utilisée car les propriétés du laitier (quantité et densité
du fer) et de la matrice cimentaire sont très contrastées. Les résultats ont
été comparés à ceux obtenus par spectroscopie Mössbauer (Figure 7).
ra
L'estimation du degré d'hydratation du laitier à l'aide de la spectroscopie

Mössbauer et de la tomographie ont mis en évidence une cinétique plus
rapide de l'hydratation du laitier en présence de ciment Portland.
converge vers 50 %. Les valeurs généralement plus faibles obtenues à

partir de la tomographie sont probablement liées à deux facteurs : i)
l'hypothèse selon laquelle le laitier est complètement non réactif au cours
des 24 premières heures d'hydratation ; ii) le réglage de la résolution (voxel
à 2,15 µm) qui exclut la contribution des particules de laitier les plus fines.
xvi ASTOVEZA 2022

Figure 6 Principales étapes de l'estimation du degré d'hydratation du laitier à l'aide de

XCT, notamment : (1) reconstruction 2D des scans ; (2) segmentation de seuil pour
distinguer les particules du laitier ; (3) création d'images binaires 2D sur la base de
l'étape précédente ; (4) modélisation des volumes 3D à partir desquels le degré
d'hydratation a été estimé.
Figure 7 Degré d'hydratation du laitier estimé à l'aide de la XCT et la

spectroscopie Mössbauer
ASTOVEZA 2022 xvii

Évolution de l'assemblage des phases
contenant du laitier ont été prédites à l'aide d'une modélisation

thermodynamique (logiciel GEMS) et comparées à celles observées
expérimentalement à l'aide de la diffraction des rayons X, des analyses
thermogravimétriques et de la spectroscopie infrarouge. Lors de la
50 %. Un résumé des résultats est présenté dans le Tableau 2.
Tableau 2 Résumé des assemblages de phases obtenus à partir de la modélisation

Le
monosulfoaluminate est abrégé en Ms.
modélisation thermodynamique résultats expérimentaux

composants: principal mineur principal mineur
ettringite,
ettringite, gibbsite,
A Ms gibbsite,
hydrogrenat strätlingite
strätlingite
hydrogrenat, ettringite, Ms,
B ettringite
gibbsite gibbsite strätlingite
strätlingite, hydrogrenat, strätlingite,
A-PC C-S-H
Ms C-S-H Ms
strätlingite, strätlingite,
Ms,
B-PC ettringite, ettringite C-S-H,
C-S-H
hydrogrenat Ms
Les résultats expérimentaux sont généralement en accord avec les
modélisations, notamment sur la nature des phases formées. Les
principales différences portent sur les proportions respectives. Elles ont été
attribuées à la cinétique d'hydratation, non prise en compte lors de la
expérimentalement observée au jeune âge dans les systèmes à haute teneur

en sulfates, tandis que le laitier et le ciment portland présentent une
cinétique d
L'ettringite formée étant relativement stable dans le temps, sa présence
modifie la proportion des ions disponibles pour l'hydratation par rapport
au système initial. C'est également la raison principale proposée pour
expliquer pourquoi l'hydrogrenat et le monosulfoaluminate contenant du
fer, prédit dans les modèles, n'ont pas été détectés expérimentalement, du
moins en quantités significatives. Le faible degré de cristallinité de ces
phases peut également expliquer la difficulté de quantifier la proportion
réelle des divers hydrates.
Par rapport aux formules de référence, l'évolution au cours du temps de
l'assemblage des phases déterminée expérimentalement montre que
l'influence du laitier se manifeste principalement sur les silicates (i.e.
xviii ASTOVEZA 2022

strätlingite, gehlénite, C-S-H) où des changements majeurs ont été

observés. Le composant cristallisé du laitier, la wüstite, est également
apparu non réactif, démontrant que seule la fraction vitrifiée du laitier se
dissolvait de manière significative tout au long de la période de
durcissement d'un an. De plus, la réactivité du laitier a pu être démontrée
à partir des spectres IRTF sur la base du déplacement du pic de la bande
asymétrique Si-O du laitier à environ 930 cm-1. Cependant, la complexité
des systèmes étudiés et la difficulté de caractériser les phases amorphes en
phase
hydratée contenant du fer.
Mécanisme de l'hydratation du fer

Le fer initialement contenu dans le laitier a été détecté à l'aide de la
microscopie électronique (MET et METB) principalement dans des phases
de gel amorphe (Figure 8). Ces gels sont plus riches en Al dans les
systèmes binaires, tandis que les proportions de Ca et de Si sont plus
élevées dans les systèmes ternaires (Figure 9). Cette différence est liée à
la contribution du ciment Portland dans ces derniers. Les images MET de
certaines des phases observées dans les pâtes contenant du laitier sont
présentées à la Figure 8. Dans ces images, la morphologie distincte de
chaque phase est bien visible.
Figure 8 Images MET de certaines phases observées dans les pâtes contenant du laitier,
notamment : gels contenant du fer (a-b) avec leur diffractogramme (c) ; ettringite (d) ;
strätlingite (e) et wüstite (f)
À plus longs termes, les résultats obtenus par microscopie électronique ont
montré que la composition des gels tend à se déplacer, dans le diagramme
ASTOVEZA 2022 xix

période de durcissement plus longue (> 180 jours), les cartographies

obtenues par METB ont révélé la formation d'une phase nanocristalline
riche en fer dans le système binaire. Cette phase non identifiée présentait
une morphologie en forme de tige/aiguille (Figure 9) avec un espacement
inter couches atomiques d'environ 7,2 Å, estimé à partir de son diagramme
de diffraction électronique. Sur la base de ces caractéristiques et de la
composition chimique (EDS), les phases pouvant correspondre seraient
l'hydroxyde de fer-aluminium et l'hydroxyde doubles lamellaires (LDL)
contenant du fer. Cette dernière structure pourrait expliquer davantage la
quantité relativement importante de Mg détectée dans cette phase.
révéler avec certitude l'identité de cette phase hydratée riche en fer. Des
techniques ou des méthodologies complémentaires sont nécessaires pour
transforme à termes en hydrogrenat ou en monosulfoaluminate contenant

du fer, comme suggéré par les modèles thermodynamiques, doit également
être envisagée.
Figure 9 Diagramme ternaire de composition atomique (At.%) de gels d'hydrates avec 1-

12% de fer obtenu à partir d'analyses MET et METB à dispersion d'énergie (EDS). Des
points de référence pour la gibbsite, la strätlingite, la katoite, l'hydrogrenat siliceux et le
C-A-S-H ont été ajoutés à partir des données de la littérature [17]
xx ASTOVEZA 2022
Figure 10 Images MET (a-b) et diffractogramme (c) mettant en évidence les

caractéristiques de la phase nanocristalline non identifiée contenant du fer détectée après
1 an d'hydratation dans le système binaire
D'autre part, l'évolution des spectres Mössbauer lors de l'hydratation a mis

en évidence la transformation de l'environnement atomique du fer,
principal composant du laitier. Deux modèles indépendants de
décomposition des spectres ont suggéré que l'hydratation du laitier
changement de coordination tétraédrique à octaédrique avec des atomes

d'oxygène. Les paramètres Mössbauer utilisés pour les nouvelles phases
formées sont similaires à ceux rapportés pour les gels de ferrihydrite et les
monosulfoaluminates contenant du fer. L'un des modèles a également
indiqué un composant supplémentaire de Fe (II) ayant une coordination
Les autres sous-produits SOCRATES

Le potentiel de valorisation du résidu de jarosite en tant que source de
sulfate et du résidu du recyclage du papier en tant qu'activateur minéral
dans les systèmes binaires CAC-HH ont également été explorés.
Les résultats ont montré que lorsque le résidu de jarosite remplace la source
de sulfate de calcium (jusqu'à 50 % en masse), la résistance à la
compression est maintenue à 26 MPa après 1 jour et 60 MPa après 28 jours
d'hydratation. Ces performances sont obtenues malgré la faible quantité de
soufre soluble dans le résidu de jarosite mesuré à partir du test de
conductivité. Cependant, compte tenu de la nature de ce résidu, des
analyses rigoureuses de toxicité et de lixiviabilité restent indispensables
pour identifier les opérations de prétraitement nécessaires.
D'autre part, l'ajout de 5 % de résidus provenant du recyclage du papier a
été ajouté pour induire la dissolution du laitier. Cette addition, riche en
chaux, a été accompagnée de 1,3 % d'acide citrique pour le contrôle de la
prise. Ce résidu a significativement augmenté la résistance mécanique du
mélange (différence de > 10 MPa) alors que seule une augmentation
minimale (2-5 %) du degré d'hydratation du laitier a été estimée à l'aide de
ASTOVEZA 2022 xxi

la tomographie. L'ajout du résidu du recyclage du papier a semblé avoir un

impact sur le pH et la composition de la solution interstitielle ce qui a
favorisé la formation de phases AFm.
Conclusion et perspectives
La caractérisation approfondie des propriétés physiques et chimiques des
projet européen SOCRATES a démontré le potentiel de valorisation de ces

produits dans ce type de liant. Des mortiers de haute résistance et stables
amorphe riche en fer dans des liants binaires CAC-HH et ternaires CAC-
HH-PC. La méthodologie développée (tomographie, etc.) a permis de
mon -delà de 180 jours. Ce
résultat a été confirmé par spectroscopie Mössbauer, qui a montré
également que l'hydratation du laitier s'accompagnait de l'oxydation du fer
2+ à 3+, et de la formation de nouveaux hydrates contenant du fer. À l'aide
de la microscopie électronique (MET et METB), le fer a été détecté au
jeune âge dans des phases complexes formées de gels amorphes riches en
aluminium. Le fer a également été détecté à des proportions plus élevées à
plus long terme, au sein une phase nanocristalline non identifiée. Dans les
systèmes ternaires, cette phase était également associée à du calcium, du
silicium et une petite quantité de magnésium.
L'effet du ciment portland, des sulfates et des résidus du recyclage du
papier sur la réactivité du laitier ont également été explorés dans cette
étude. Les résultats ont montré qu'il n'y avait pas de différences
substantielles dans le degré d'hydratation du laitier entre les systèmes à
faible et à haute teneur en sulfate étudiés. Au contraire, l'ajout de ciment
portland et de résidus du recyclage du papier ont montré une augmentation
du degré d'hydratation, toutefois cantonnée au jeune âge (< 28 jours). Cet
2+
dans le système,
ainsi que de l'augmentation de l'alcalinité favorisant la dissolution du
laitier.
Pour conclure, il reste encore des questions importantes en suspens, tant
pour expliquer les mécanismes fondamentaux de l'hydratation que pour
tester la durabilité de ces liants. Ainsi, les travaux futurs devraient se
concentrer sur des analyses complémentaires pour identifier la phase
nanocristalline contenant du fer. Elles pourraient utiliser des techniques de
caractérisation supplémentaires, ou bien reproduire les systèmes de
manière synthétique (phases pures) pour limiter notamment le problème
méthodologie par tomographie et les modèles utilisés pour la spectroscopie
xxii ASTOVEZA 2022

Mössbauer pourraient être encore

leur interprétation.
il faudrait encore optimiser les proportions de CAC, HH, PC et laitier,

voire aussi augmenter la réactivité du laitier au jeune âge (broyage,
composition, trempe) et enfin évaluer la durabilité de ce nouveau liant
alumineux.
Enfin, optimiser les proportions de CAC, HH, PC et laitier ; augmenter
davantage la réactivité du laitier surtout au jeune âge ; et l'évaluation de la
durabilité, sont des étapes essentielles pour augmenter le potentiel de
commercialisation de ce liant.
Bibliographie
[1] United Nations Environement Program (UNEP), Buildings and
climate change: Summary for decision-makers, 2009,
www.unep.fr/scp/sun.
[2] K.L. Scrivener, V.M. John, E.M. Gartner, Eco-efficient cements:
Potential economically viable solutions for a low-CO2 cement-based
materials industry, Cement and Concrete Research 114 (2018) 2 26.
[3] P.J.M. Monteiro, S.A. Miller, A. Horvath, Towards sustainable
concrete, Nature materials 16 (2017) 698 699.
[4] A. Vollpracht, B. Lothenbach, R. Snellings, J. Haufe, The pore
solution of blended cements: a review, Mater Struct 49 (2016) 3341
3367.
[5] S. Samad, A. Shah, M.C. Limbachiya, Strength development
characteristics of concrete produced with blended cement using ground
granulated blast furnace slag (GGBS) under various curing conditions,
1213.
[6] SOCRATES EU MSCA-ETN, Project proposal: Innovative training
networks (ITN) call: H2020-MSCA-ITN-2016, Part B, 2016, https://etn-
socrates.eu/.
[7] J.E. Rossen, Composition and morphology of C-A-S-H in pastes of
alite and cement blended with supplementary cementitious materials,
Lausanne, EPFL, 2014.
[8] R. Snellings, G. Mertens, J. Elsen, Supplementary Cementitious
Materials, Reviews in Mineralogy and Geochemistry 74 (2012) 211 278.
ASTOVEZA 2022 xxiii

[9] D. Marchon, S. Kawashima, H. Bessaies-Bey, S. Mantellato, S. Ng,

Hydration and rheology control of concrete for digital fabrication:
Potential admixtures and cement chemistry, Cement and Concrete
Research 112 (2018) 96 110.
[10] W. Chen, H.J.H. Brouwers, The reaction of slag in cement, theory
and computer modelling: Materials Science, 2006.
[11] M.F. Zawrah, A.B. Shehata, E.A. Kishar, R.N. Yamani, Synthesis,
hydration and sintering of calcium aluminate nanopowder for advanced
applications, Comptes Rendus Chimie 14 (2011) 611 618.
[12] E. Henry-Lanier, B. Espinosa, C. Eychenne-Baron, Environmental
footprint of calcium aluminate cement (CAC), in: C.H. Fentiman, R.J.
Mangabhai, K. Scrivener (Eds.), Calcium aluminates: Proceedings of the
international conference 2014, Palais des Papes, Avignon, France, 18-21
May 2014 / edited by Charles Fentiman, Raman Mangabhai and Karen
Scrivener, IHS, Bracknell, Berkshire, 2014, pp. 3 16.
[13] A.M. Dunster, F. Moulinier, K.C. Quillin, G.J. Osborne, Durability
of concrete made with calcium aluminate cement and ground granulated
blastfurnace slag in sulfate and marine environments, in: C. Fentiman, R.
Mangabhai, K. Scrivener (Eds.), Calcium Aluminate Cements
Proceedings of the centenary conference 2008, 2008.
[14] S. Kucharczyk, J. Deja, M. Zajac, Effect of Slag Reactivity
Influenced by Alumina Content on Hydration of Composite Cements,
ACT 14 (2016) 535 547.
[15] L. Fernández-Carrasco, E. Vázquez, Reactions of fly ash with
calcium aluminate cement and calcium sulphate, Fuel 88 (2009) 1533
1538.
[16] A.J. Majumdar, R.N. Edmonds, B. Singh, Hydration of Secar 71
aluminous cement in presence of granulated blast furnace slag, Cement
and Concrete Research 20 (1990) 7 14.
[17] I. Garcia-Lodeiro, A. Palomo, A. Fernández-Jiménez, D.E.
Macphee, Compatibility studies between N-A-S-H and C-A-S-H gels.
Study in the ternary diagram Na2O CaO Al2O3 SiO2 H2O, Cement
xxiv ASTOVEZA 2022

________________________________________________ Table of Contents
Table of Contents
Acknowledgements .............................................................................. i
Abstract .............................................................................................. iii
Samenvatting ....................................................................................... v
Résumé .............................................................................................. vii
List of Figures ................................................................................ xxxi
List of Tables ................................................................................ xxxix
Glossary............................................................................................. xli
Introduction ....................................................................................... 1
Chapter 1: State-of-the-art ............................................................... 9
1.1. The SOCRATES Project............................................................ 13
1.2. Calcium aluminate blended cements ......................................... 17
1.2.1. Pure CAC systems .............................................................. 18
1.2.2. Binary CAC-C$ systems ..................................................... 21
1.2.3. Ternary CAC-PC-C$ systems ............................................. 24
1.3. Supplementary Cementitious Materials ..................................... 27
1.3.1. Traditional SCM .................................................................. 27
1.3.2. Non-conventional SCM ....................................................... 30
1.4. The role of Fe in inorganic polymers and cement binders ......... 33
1.4.1. Fe speciation in slag-containing inorganic polymers ........ 33
1.4.2. Fe speciation in cementitious binders ................................ 34
1.4.3. Other techniques to characterize and quantify the reactivity
of slags ........................................................................................... 37
Chapter 2: Materials, methods and residue selection .................. 43
2.1. Materials and formulation .......................................................... 45
2.1.1. The Socrates residues ......................................................... 45
2.1.2. Cements .............................................................................. 48
2.1.3. Other blend components ..................................................... 48
2.1.4. Dry mix formulations .......................................................... 48
2.2. Methods ..................................................................................... 50
2.2.1. Basic characterization of raw materials ............................ 50
xxvi ASTOVEZA 2022
Table of Contents ______________________________________________
2.2.2. Physical properties ............................................................. 53

2.2.3. Thermal and thermo gravimetric analyses ......................... 56
2.2.4. Tomography ........................................................................ 56
2.2.5. Spectroscopy ....................................................................... 58
2.2.6. Microscopy ......................................................................... 64
2.2.7. Other techniques ................................................................. 68
2.3. Selection of industrial residue .................................................... 69
2.3.1. Electric conductivity measurements ................................... 69
2.3.2. Heat Release ....................................................................... 70
2.3.3. Setting time and mechanical strength ................................. 71
2.4. Evaluating jarosite residue as a sulfate source ........................... 73
Chapter 3: Fe-rich slag addition in binary CAC-HH binder ...... 81
3.1. Physical properties ..................................................................... 84
3.1.1. Low calcium sulfate system ................................................ 84
3.1.2. High calcium sulfate system ............................................... 87
3.2. Quantifying the degree of slag hydration ................................... 89
3.2.1. SEM-BSE ............................................................................ 89
3.2.2. Tomography ........................................................................ 92
3.3. Phase assemblage evolution over time....................................... 96
3.3.1. Thermodynamic modelling ................................................. 96
3.3.2. Early hydration reactions ................................................... 99
3.3.3. Long term hydration ......................................................... 100
3.4. The fate of Fe during hydration ............................................... 107
3.4.1. TEM .................................................................................. 108
3.4.2. STEM ................................................................................ 112
3.4.3. Mössbauer spectroscopy................................................... 120
3.5. Activation through paper residue addition ............................... 129
Chapter 4: Fe-rich slag addition in ternary CAC-HH-PC ........ 137
4.1. Physical properties ................................................................... 140
4.1.1. Low calcium sulfate system .............................................. 141
4.1.2. High calcium sulfate system ............................................. 143
4.2. Quantifying the degree of slag hydration using XCT .............. 147
ASTOVEZA 2022 xxvii
________________________________________________ Table of Contents
4.3. Phase assemblage evolution over time .................................... 151

4.3.1. Thermodynamic modelling ............................................... 151
4.3.2. Early hydration reactions ................................................. 155
4.3.3. Long term hydration ......................................................... 157
4.4. The fate of Fe during hydration ............................................... 165
4.4.1. TEM .................................................................................. 166
4.4.2. STEM ................................................................................ 170
4.4.3. Mössbauer spectroscopy .................................................. 175
Conclusions and Perspectives ....................................................... 181
Appendices ..................................................................................... 189
A. Life cycle assessment and exergy analysis .............................. 189
B. Mineralogy of the raw materials using XRD ........................... 203
C. Particle size distribution of raw materials ................................ 207
E. Image processing operations for XCT ..................................... 210
F. Mössbauer Spectroscopy Parameters ....................................... 211
G. Estimation of slag hydration degree using SEM-BSE ............. 213
H. Estimation of the theoretical slag volume in paste .................. 215
I. Supplementary data for thermodynamic modelling ................. 216
J. Supplementary data for Mössbauer spectroscopy.................... 221
K. Hydration of pure PC system ................................................... 229
List of Publications ........................................................................ 231
References ...................................................................................... 233
xxviii ASTOVEZA 2022

_______________________________________________ List of Figures
List of Figures
Figure 1 Simplified flow sheet of the SOCRATES Project .................... 1
Figure 2 Chemical compatibility of some of the SOCRATES residues .. 3
Figure 3 Some applications of CAC-based binders: ............................... 5
Figure 1.1 The scope of the SOCRATES project ................................. 13

Figure 1.2 Submerged plasma-driven slag fuming process ................... 15
Figure 1.3 Two manufacturing processes for CAC production ............. 17
Figure 1.4 SEM images of the hexagonal CAH10 and cubic C3AH6 .... 20
Figure 1.5 Heat flow curves of formulations with different CAC
proportions ............................................................................................. 22
Figure 1.6 Phase assemblage evolution of CA-C$ (anhydrite) paste .... 23
Figure 1.7 SEM micrograph of synthesized monosulfoaluminate and.. 24
Figure 1.8 Phase assemblage evolution of: (a) a PC-rich ternary system;
and .......................................................................................................... 25
Figure 1.9 Heat flow curves of a pure PC system and . ........................ 29
Figure 1.10 (a) Particle size distribution of several SCM .................... 30
Figure 1.11 The structure of a synthetic NFM slag ............................... 34
Figure 1.12 Evolution of the Mössbauer spectrum of a pure PC hydrated
................................................................................................................ 36
Figure 1.13 ATR-FTIR spectra of three synthetic Fe-rich slags ........... 38
Figure 1.14 Figures collated from the quantification of hydration degree
................................................................................................................ 39
Figure 2.1 Picture of the granulated Fe-rich slag as received................ 45

Figure 2.2 Picture of the slowly-cooled Fe-rich slag sample as received
................................................................................................................ 46
Figure 2.3 Picture of the fumed fayalite slag sample as received ......... 46
Figure 2.4 Picture of the MSWI bottom ash sample as received........... 47
Figure 2.5 Picture of the jarosite residue sample as received ................ 47
Figure 2.6 Picture of the paper residue sample as received ................... 47
ASTOVEZA 2022 xxxi

List of Figures ________________________________________________
Figure 2.7 Standard mortar mounted in an extensometer ...................... 54

Figure 2.8 Walter+Bai shrinkage test set up (a) and the dimensions of the
triangular prism sample after the in-situ measurements ......................... 55
Figure 2.9 Picture of the XCT unit ....................................................... 57
Figure 2.10 Deconvoluted 57Fe Mössbauer spectra of the slag.............. 59
Figure 2.11 Deconvoluted 57Fe Mössbauer spectra of the CAC ............ 61
Figure 2.12 Deconvoluted 57Fe Mössbauer spectra of the PC ............... 61
Figure 2.13 ATR-FTIR of the raw (granulated FS) slag ....................... 62
Figure 2.14 ATR-FTIR of Secar 51®.................................................... 63
Figure 2.15 ATR-FTIR of the PC .......................................................... 63
Figure 2.16 ATR-FTIR of the calcium sulfate hemihydrate (HH) ........ 63
Figure 2.17 (a) Micrograph of a slag-containing hydrated mortar sample
................................................................................................................ 64
Figure 2.18 Micrograph of a slag-containing hydrated mortar sample
observed under an optical microscope in (a) polarized light and (b)
reflected light. ......................................................................................... 65
Figure 2.19 TEM images of pure synthetic C-S-H ............................... 66
Figure 2.20 STEM-EDS Fe-maps of C-S-H foils ................................. 67
Figure 2.21 Crystal structure of ettringite projected on to the (0001) plane.
................................................................................................................ 68
Figure 2.22 Schematic diagram of the conductimetry set-up ................ 69
Figure 2.23 Conductivity measurements on raw materials ................... 70
Figure 2.24 (a) Heat flow and (b) cumulative heat release of paste samples
for the screening test ............................................................................... 71
Figure 2.25 Initial and final setting time of mixtures for the screening test
................................................................................................................ 71
Figure 2.26 (a) Compressive strength evolution and (b) strength activity
index (SAI) of standard mortars for the screening test ........................... 72
Figure 2.27 Compressive strength evolution of blends with jarosite residue
addition ................................................................................................... 73
Figure 2.28 XRD patterns of hydrated pastes with different substitution
level of jarosite residue after 28 days of curing. ..................................... 74
Figure 2.29 Conductivity of the jarosite residue versus HH.................. 75
xxxii ASTOVEZA 2022

_______________________________________________ List of Figures
Figure 2.30 SEM-SE images of (a) no jarosite and (b) 100JAR paste
sample..................................................................................................... 75
Figure 2.31 Comparison of the ion concentrations after the dissolution test
................................................................................................................ 76
Figure 3.1 Initial and final setting time of formulation A and its references
................................................................................................................ 84
Figure 3.2 Compressive strength of formulation A (primary axis) and its
SAI evolution (secondary axis) .............................................................. 85
Figure 3.3 Flexural strength of formulation A (primary axis) and its SAI
evolution (secondary axis)...................................................................... 85
Figure 3.4 Length variation of formulation A and its reference ............ 86
Figure 3.5 Compressive strength of formulation B (primary axis) and its
SAI evolution (secondary axis) .............................................................. 88
Figure 3.6 Flexural strength of formulation B (primary axis) and its SAI
evolution (secondary axis)...................................................................... 88
Figure 3.7 Length variation of formulation B and its references .......... 88
Figure 3.8 (a) SEM-BSE micrograph of formulation A mortar after 90
days ........................................................................................................ 90
Figure 3.9 (a) SEM-BSE micrograph of formulation A reference cement
................................................................................................................ 91
Figure 3.10 SEM-BSE micrograph of formulation A pastes ................. 91
Figure 3.11 Illustration of the steps involved in the volume analysis of
XCT ........................................................................................................ 93
Figure 3.12 Estimated slag volume and hydration degree using XCT .. 94
Figure 3.13 Illustration of XCT analysis using cylindrical volumes ..... 95
Figure 3.14 Estimated (a) slag total volume and (b) slag hydration degree
................................................................................................................ 96
Figure 3.15 Phase assemblage of the reference (no slag) binary binder
calculated from GEMS thermodynamic modelling software ................. 98
Figure 3.16 Phase assemblage of the binary binder incorporating 30% slag
with varying CAC and HH proportions ................................................. 98
Figure 3.17 Heat flow measured ex-situ using isothermal calorimetry . 99
Figure 3.18 Cumulative heat release measured using isothermal
calorimetry ........................................................................................... 100
ASTOVEZA 2022 xxxiii

List of Figures ________________________________________________
Figure 3.19 XRD patterns of formulation A (solid black lines) .......... 101
Figure 3.20 XRD patterns of formulation B (solid blue lines) ............ 103
Figure 3.21 Evolution of the DTG curves of (a) formulation A and (b)
formulation B........................................................................................ 104
Figure 3.22 Total bound water content of formulations A and B ....... 104
Figure 3.23 ATR-FTIR spectra of formulation A (solid black lines). . 105
Figure 3.24 ATR-FTIR spectra of formulation B (solid blue lines). ... 106
Figure 3.25 TEM images and the atomic composition of areas in the (A)
.............................................................................................................. 108
Figure 3.26 TEM images and the atomic composition of areas in the (A)
.............................................................................................................. 109
Figure 3.27 TEM images, diffraction patterns and the atomic composition
of Fe-containing gels in the (A) 4.5CAC-HH paste sample ................. 110
Figure 3.28 TEM images before and after EDS analysis of Fe-containing
gels in the (A) 4.5CAC-HH paste sample ............................................ 110
Figure 3.29 TEM images and the atomic composition of crystalline phases
in the (A) 4.5CAC-HH paste sample .................................................... 110
Figure 3.30 TEM images and the atomic composition of marked areas in
the (B) 1.6CAC-HH paste sample ........................................................ 111
Figure 3.31 STEM elemental mapping, diffraction patterns ............... 113
Figure 3.32 STEM elemental mapping, diffraction patterns ............... 114
Figure 3.33 Ternary diagram of Fe-containing hydrate gels with 1-12% Fe
.............................................................................................................. 115
Figure 3.34 STEM image focused on an Fe-rich region...................... 116
Figure 3.35 STEM image focused on an Fe-rich region (a); along with its
corresponding annular dark field image (b); Fe map from EDS analysis (c);
and the diffraction pattern .................................................................... 116
Figure 3.36 STEM images focused on the needle-like Fe-rich phase . 117
Figure 3.37 TEM images of needle-like Fe hydroxide phases ............ 118
Figure 3.38 TEM images of (a) Al(OH)3 with its inset selected-area
electron diffraction ............................................................................... 119
Figure 3.39 57Fe Mössbauer spectrum evolution of formulation A ..... 121
Figure 3.40 57Fe Mössbauer spectrum of formulation A and its
components ........................................................................................... 123
xxxiv ASTOVEZA 2022

_______________________________________________ List of Figures
Figure 3.41 Absorption areas (AA) calculated from the two modelling
approaches ............................................................................................ 127
Figure 3.42 Slag hydration degree in formulation A paste.................. 128
Figure 3.43 (a) Heat flow and (b) cumulative heat release of pastes
containing 23 wt% paper residue (in blue) and 30 wt% slag (in black).129
Figure 3.44 Initial and final setting time of formulations A-TC; reference
no TC; and formulation A .................................................................... 130
Figure 3.45 Compressive strength (primary axis) and SAI evolution
(secondary axis) of formulation A-TC (blue), its reference formulation
(gray) and formulation A (black). ........................................................ 131
Figure 3.46 Flexural strength (primary axis) and SAI evolution (secondary
axis) of formulation A-TC (blue), its reference formulation (gray) and
formulation A (black). .......................................................................... 131
Figure 3.47 Length variation of formulations A-TC (blue) and A (black)
.............................................................................................................. 132
Figure 3.48 The slag hydration degree (primary axis) and slag total volume
of formulations A-TC and A estimated from XCT + volume analysis 132
Figure 3.49 (a) XRD patterns; (b) Mössbauer spectra; (c) DTG curves; and
(d) total bound water evolution of formulations A-TC ........................ 133
Figure 3.50 TEM images and the atomic composition of marked areas in
the (A-TC) 4.5CAC-HH-5TC paste sample ......................................... 134
Figure 4.1 Initial and final setting time of formulation A-PC, B-PC .. 140
Figure 4.2 Compressive strength of formulation A-PC (primary axis)141
Figure 4.3 Flexural strength of formulation A-PC (primary axis)....... 141
Figure 4.4 Length variation of formulation A-PC and its references .. 142
Figure 4.5 Compressive strength of formulation B-PC (primary axis) 144
Figure 4.6 Flexural strength of formulation B-PC (primary axis) ....... 144
Figure 4.7 Length variation of formulation B-PC and its references .. 146
Figure 4.8 Pictures of A-PC and B-PC mortars with their references . 147
Figure 4.9 Illustration of the steps involved in the volume analysis of XCT
.............................................................................................................. 148
Figure 4.10 Estimated slag volume and hydration degree using XCT 149
Figure 4.11 Phase assemblage of the reference ternary CAC-HH-PC
binder.................................................................................................... 152
ASTOVEZA 2022 xxxv

List of Figures ________________________________________________
Figure 4.12 Phase assemblage of the ternary binder ........................... 152

Figure 4.13 Composition by volume of the reference ternary binder .. 154
Figure 4.14 Composition by volume of the ternary binder.................. 154
Figure 4.15 Heat flow measured ex-situ using calorimetry for A-PC . 155
Figure 4.16 Cumulative heat release using isothermal calorimetry..... 156
Figure 4.17 XRD patterns of formulation A-PC (solid black lines) .... 158
Figure 4.18 XRD patterns of formulation B-PC (solid blue lines). ..... 159
Figure 4.19 DTG curves of (a) formulation A-PC and (b) formulation B-
PC ......................................................................................................... 160
Figure 4.20 ATR-FTIR spectra of formulation A-PC (solid black lines)
.............................................................................................................. 160
Figure 4.21 ATR-FTIR spectra of formulation B-PC (solid blue lines)
.............................................................................................................. 161
Figure 4.22 Superimposed ATR-FTIR spectra of the raw slag (red) to
those of the B-PC.................................................................................. 163
Figure 4.23 -FTIR spectrum superimposed to that of the
B-PC ..................................................................................................... 163
Figure 4.24 Total bound water content of formulations A-PC, B-PC . 165
Figure 4.25 TEM images and the atomic composition of areas in the (A-
PC) ........................................................................................................ 166
Figure 4.26 TEM images and the atomic composition of areas in the (A-
PC) ........................................................................................................ 166
Figure 4.27 TEM images, diffraction pattern and the atomic composition
of Fe-containing gels in the (B-PC) 1.6CAC-HH-30PC paste sample . 168
Figure 4.28 TEM images of marked areas in the (B-PC) .................... 169
Figure 4.29 STEM elemental mapping of a gel phase in the (B-PC) .. 171
Figure 4.30 STEM elemental mapping for a selected area in the (B-PC)
.............................................................................................................. 172
Figure 4.31 STEM images and their corresponding Fe and S maps.... 173
Figure 4.32 Ternary diagram of Fe-containing hydrated gels ............. 174
Figure 4.33 57Fe Mössbauer spectrum evolution of formulation B-PC176
Figure 4.34 57Fe Mössbauer spectrum of formulation B-PC at 28 days and
its deconvoluted components................................................................ 177
Figure 4.35 Evolution of B-PC absorption areas (AA) calculated . .... 177
xxxvi ASTOVEZA 2022

_______________________________________________ List of Figures
Figure 4.36 Slag hydration degree of formulation B-PC (primary axis)

estimated based on the changes in the total absorption areas (secondary
axis) of the unreacted slag .................................................................... 178
Figure 5.1 Summary of the compressive strength results .................... 182

Figure 5.2 Pictures of the B-PC and B (high-sulfate) mortars after 45 days
of curing under water ........................................................................... 182
Figure 5.3 Consolidated slag hydration degree estimated using XCT and
Mössbauer spectroscopy....................................................................... 183
Figure 5.4 Consolidated ternary diagram of Fe-containing hydrate gels
.............................................................................................................. 185
ASTOVEZA 2022 xxxvii

_________________________________________________ List of Tables
List of Tables
Table 1.1 Chemical composition of common by-products .................... 31
Table 2.1 Binder composition for the screening test ............................ 49

Table 2.2 Binder composition for the case study on jarosite ................. 49
Table 2.3 Composition of binders incorporating the granulated Fe-rich
slag ......................................................................................................... 50
Table 2.4 Chemical analysis of raw materials obtained using XRF ...... 51
Table 2.5 Minor components of some of the raw materials .................. 52
Table 2.6 Density, particle size distribution and specific surface area of
the raw materials .................................................................................... 52
Table 2.7 57Fe Mössbauer hyperfine parameters of the fitting proposed for
the slag.................................................................................................... 59
Table 3.1 Estimated slag area and hydration degree using SEM-BSE .. 92
Table 3.2 FTIR band assignments based on the distinct spectral peaks107
Table 3.3 57Fe Mössbauer hyperfine parameters used for Model 1. .... 124
Table 3.4 57Fe Mössbauer hyperfine parameters used for Model 2. .... 124
Table 4.1 FTIR band assignments based on the distinct spectral peaks161
Table 5.1 Composition of binders incorporating the Fe-rich slag ....... 181
Table 5.2 Consolidated phase assemblage based on thermodynamic
modelling.............................................................................................. 184
ASTOVEZA 2022 xxxix

_____________________________________________________ Glossary
Glossary
Cement shorthand notations for oxides:
H: H2O C: CaO A: Al2O3 S: SiO2 F: Fe2O3 $: SO3 c: CO2 T: TiO2
Abbreviations of main anhydrous and hydrated phases:
CA CaAl2O4 Monocalcium aluminate

C3S Ca3SiO5 Alite
C2S Ca2SiO4 Belite
Tetracalcium aluminoferrite
C4AF Ca4Al2Fe2O10 (ferrite)
C2AS Ca2Al2SiO7 Gehlenite (Geh)
Cc CaCO3 Calcite
C$H0.5 CaSO4 2O Calcium sulfate hemihydrate
C$H2 CaSO4 2O Gypsum (Gyp)
C6A$3H32 Ca6Al2(SO4)3(OH)12 26(H2O) Ettringite (Ett)
C4A$H12 Ca4Al2(SO4)(OH)12 6(H2O) Monosulfoaluminate (Ms)
C2ASH8 Ca2Al2SiO2(OH)10 (H2O) Strätlingite (Str)
C3AH6 Ca3Al2(OH)12 Hydrogarnet (Hdg)
AH3 2 Al(OH)3 Gibbsite (Gib)
FeO Wüstite (Wue)
Materials:
Calcium aluminate cement, Ciment Fondu® or Secar
CAC 51®
PC Ordinary Portand cement, Milke® Premium
HH -hemihydrate, Prestia Selecta®
C$ Calcium sulfate
(quartz) filler Fine sand, Millisil E400®
NFM slags Non-ferrous metallurgy slags
slag Granulated iron-rich slag (Chapters 3and 4)
paper residue Residue from paper recycling, Top-Crete®
SCM Supplementary cementitious materials
binder Cement (CAC, PC) + HH + SCM or filler
ASTOVEZA 2022 xli

Glossary _____________________________________________________
IP Inorganic polymer
Techniques:
PSD Particle size distribution
SSA Specific surface area
XRF X-ray fluorescence
XRD X-ray diffraction
TGA Thermogravimetric analysis
ATR- Attenuated total reflection-
FTIR Fourier transform infrared spectroscopy
XCT X-ray computed tomography
SEM Scanning electron microscopy
(Scanning) Transmission electron
(S)TEM microscopy
MS Mössbauer spectroscopy
XANES X-ray absorption near edge structure
NMR Nuclear magnetic resonace spectroscopy
XAS X-ray absorption spectroscopy
Other abbreviations:
wt% weight percent
at% atomic percent
vol% volume percent
w/b water to binder ratio
SAI strength activity index
VOI volume of interest in XCT
BSE backscaterred electron in SEM or TEM
EDS energy-dispersive (detector) in SEM or TEM
primary axis located on the vertical left
secondary axis located on the vertical right
xlii ASTOVEZA 2022

_________________________________________________ Introduction
Introduction
Circular economy is one of the fundamental building blocks of the
2021 European Green Deal [1]. This concept revolves around regenerative
resource cycles with a focus on minimization of waste through process
optimization, materials substitution and recycling. As a consequence,
research and development of sustainable processes and products is
neutrality targets by 2050.
Figure 2 Simplified flow sheet of the SOCRATES Project

with emphasis on the work of the author: residual matrix valorization as supplementary
cementitious materials (SCM) in blended cements
The EU MSCA-ETN SOCRATES1 project promotes circular economy

through valorization of industrial by-products including (but not limited
to) non-ferrous metallurgy (NFM) slags, municipal solid waste incinerator
(MSWI) bottom ash, jarosite and paper residue, in various applications. A
general flow sheet of this project is shown in Figure 2. Initially, precious
metals and critical elements are extracted for further refining while toxic
components are removed for containment. The remaining clean mineral
1
-Curie Actions (MSCA) European Training Network (ETN) for the
sustainable, zero-waste valorization of critical-metal-containing industrial process residues under the
EU Framework Program for Research and Innovation Horizon 2020 Grant Agreement No.721385
ASTOVEZA 2022 1
Introduction _________________________________________________
fraction, comprising at least 80-90 % of the total volume, is then valorized

as precursor for inorganic polymers, as catalysts, or as supplementary
cementitious materials (SCM) in blended cements [2]. The lattermost
encompasses the research work presented in this thesis manuscript. Having
undergone pre-processing prior to the addition in the dry-mix, these
-
Generally, when by-products are adapted as substitutes to cements, as
aggregates, or as precursors for other construction materials,
environmental and economic benefits are envisioned while maintaining
acceptable product performance [3,4]. If the SOCRATES residues prove
to be viable candidates, the impact of this valorization path would be
substantial, considering the following figures [5]:
8 % of the global anthropogenic CO2 is attributed to cement
manufacturing
30 billion metric tons of concrete is produced globally per year
25-50 % of projected global increase in demand for concrete in
2050
With these, it is easy to comprehend how sustainable binders have received
significant attention in Europe in the recent decades. Nonetheless, there
remain plenty of research gaps to be addressed not only in terms of the
variety of residues and cement systems investigated, but also in
understanding the properties and hydration mechanism of these binders.
Majority of the previous research works are focused on the addition of the
more traditional SCM (i.e., GGBFS2, coal combustion fly ash, and
limestone) in ordinary Portland cement (PC)-dominated binders [6,7].
hydration by acting as an inert filler in the early hours. This was due to the
enhanced nucleation and growth when the SCM added is finer than the
cement component; and to the higher effective water to cement ratio as less
proportion of cement is added in the blend [8,9]. In the early stage of
hydration, this effect is said to be mainly physical in nature considering the
often lower dissolution rate of SCM relative to the cement component. At
long term however, ions released from the dissolving SCM could
chemically alter the hydration mechanism. For instance, siliceous SCM
such as fly ashes contribute silica for the formation of strätlingite, C2ASH8.
In addition, the dissolution of lime and alkali from the SCM could also
2
ground granulated blast furnace slag (GGBFS)
2 ASTOVEZA 2022
_________________________________________________ Introduction
modify the pH of the pore solution, thereby modifying the solubility of the
phases during hydration [6,9].
Industrial Residues as SCM

The incorporation of various types of traditional SCM in cement-based
construction materials have long been widely adopted on a commercial
scale [4]. However, their limited and often localized availability remains
to be an important challenge in boosting their market share. This is further
threatened by the sustainability-motivated decline in the numbers of
carbon-intensive energy and iron manufacturing plants where most of the
traditional SCM are generated [10]. This opens the opportunity for
exploring alternative sources of SCM with the general preconditions being
that these sources should be relatively clean (depleted of metals and toxic
components), sustainable (low environmental footprint), of lower or
comparable-cost, and of substantial high volume. Although these main
conditions are aligned with those envisioned for the pre-treated
SOCRATES residues, it is obviously still critical to assess their reactivity
and performance in the binders.
Figure 3 Chemical compatibility of some of the SOCRATES residues (orange);

traditional SCM (gray) and cements (green). The *MgO and *Fe2O3 components are
exclusively accounted for the SOCRATES residues. Data were collated to create a
modified version of the ternary diagrams reported in [2,11 13].
The chemistry, crystal structure and other properties of the SOCRATES

residues could be very different compared to those of the traditional SCM.
ASTOVEZA 2022 3
Introduction _________________________________________________
For instance, whereas the latter consist mainly of CaO, SiO2 and Al2O3,
NFM slags and jarosite residue are much richer in iron and sulfates,
respectively. The difference in the chemistry among the SOCRATES
residues, traditional SCM and cements is depicted in a ternary diagram in
Figure 3. Although this diagram is already valuable in grouping the
different resources, there is more information not disclosed. For instance,
the phase where iron partitions in the NFM slag and jarosite is amorphous
versus crystalline, and predominantly Fe2+ versus Fe3+.
Although MSWI bottom ashes could have major (>70 wt%) proportion of
CaO, SiO2 and Al2O3, they typically have a much lower degree of
vitrification as opposed to fly ashes or the GGBFS. Apart from the
difference in chemistry and oxidation state of certain metals (that might
affect the glass forming ability), this is also due to the lower firing
temperature they have been exposed to (where only partial vitrification
occurred) as well as due to the slower cooling rate afterwards. On the other
hand, the substitution level with the commercially-marketed paper residue
[14] remains limited due to its relatively high free lime content.
Overall, the NFM slags are of particular interest as SCM among the
SOCRATES residues. This is mainly because of their expected highly
amorphous nature which is often linked to their reactivity [15]. In addition,
these slags have undergone high temperature pre-treatment such that the
heavy metals and toxic components are substantially depleted. However,
discussing the reactivity and the hydration mechanism of iron (Fe)-rich

slags in calcium aluminate cement (CAC)-based systems.
With this general perspective, the work defined in the context of the
SOCRATES project targets a missed opportunity not only in the
application stand point, but also in the fundamental understanding of the
fate of Fe during CAC hydration. The need for a thorough investigation on
the influence several properties to the performance of the binders will be
critical in assessing the valorization of the rest of the SOCRATES residues
as SCM in this particular cement system.
Calcium aluminate blended cements

CAC is a collective term adopted for high-alumina cements (typically
between 40 wt% and 80 wt% Al2O3) with monocalcium aluminate (CA) as
the principal component [16]. They are used in special applications such
as self levelling underlayment (SLU), refractory, and in aggressive
environments (Figure 4). These applications are driven by the unique
properties of CAC, such as of rapid strength development at a wide
4 ASTOVEZA 2022
_________________________________________________ Introduction
temperature range, and resistance to high temperature and chemically

aggressive conditions [17].
Figure 4 Some applications of CAC-based binders:

(a) self-levelling underlayment, (b) tile adhesive, (c) refractory lining [18]
In the building chemistry industry, CAC is typically combined with PC

and calcium sulfate in different proportions. Binders dominated by CAC
take advantage of the fast setting, rapid hardening, shrinkage
compensation, and self-drying capacity3 related to ettringite formation.
These binders could display substantial expansion while maintaining high
mechanical strength and durability [9]. On the other hand, PC-rich systems
are less expensive and more suitable for high volume applications. In these
systems, the minor proportion of CAC and calcium sulfate is primarily
responsible for the quick setting and relatively high early strength property
of the binder [19]. Whereas, the PC component hydrating at a slower rate
is known to contribute to the long term strength properties [20].
There are relatively few scientific papers in the open literature on the
addition of SCM in CAC-based blends. Even the existing works are limited
to the traditional SCM such as GGBFS and fly ashes [21 23]. In terms of
sustainability, a Life Cycle Assessment (LCA) comparing SLU blends [24]
has previously estimated a lower carbon footprint for CAC-rich ternary
binders (containing CAC, PC and calcium sulfate) versus a PC-rich blend
while maintaining a comparable applicative performance. In the end,
despite the significantly lower scale of annual production estimated for
CAC at 2 million tons versus that for PC at 4 giga tons [25], the niche
market of CAC should be regarded as a supplemental opportunity for
sustainable binders rather than a direct competitor of PC.
3
The ability of the binder to fully consume the mix water for hydration
ASTOVEZA 2022 5
Introduction _________________________________________________
Research aims and outline

Two main objectives were laid out in the context of the SOCRATES
project, as follows:
1. To evaluate the reactivity of the SOCRATES by-products in CAC-
based systems
2. To propose activation techniques for these by-products in order to
improve their reactivity, and the performance of the binder
Guided by these objectives, various reactivity tests were established
serving as a screening test in order to determine which among the residues
has the highest potential for valorization in this application. The evolution
of the physical and chemical properties of the selected residue during
hydration was subsequently investigated in greater detail.
The challenge of this research work lies in the complexity of the system.
The hydration of the ternary CAC, PC and calcium sulfate hemihydrate
(HH) binders involves complex reactions often difficult to distinguish as
they occur simultaneously. Furthermore, the diverse origin of the
SOCRATES residues cannot permit a comparison of the reactivity at equal
volume substitution. In addition, the abundance of Fe and the amorphous
nature of some of the residues, particularly the NFM slags, require
complementary characterization techniques in addition to the more
conventional options widely employed in cement systems (i.e., XRD and
TGA). Since this investigation involves the use of raw materials rather than
synthetic components, it is expected that certain properties could not be
directly linked to a single reaction. Nonetheless, the interest is directed
towards demonstrating the behavior of raw materials in real complex
systems closer to the industrial setting.
The thesis is organized into four chapters, as follows:
Chapter 1 begins with a discussion on the SOCRATES project to provide
a general outlook on the unit processes and industrial flow sheets where
the residues are generated. The properties and hydration reactions of binary
(CAC-calcium sulfate) and ternary (CAC-PC-calcium sulfate) binders are
then discussed based on literature reports. The state-of-the-art in SCM
addition in these cement systems is also described. Finally, relevant studies
on the fate of Fe based on the evolution of its speciation during hydration
are collated towards the end of the chapter.
The first half of Chapter 2 provides details on the materials and methods
of the study. The properties of the SOCRATES residues, based on their
origin, are highlighted. The characterization techniques employed
throughout the study are discussed in detail along with the protocols of the
tests. The second half of the chapter is dedicated to a screening test to
6 ASTOVEZA 2022
_________________________________________________ Introduction
compare the reactivity of the residues. It ends with a short case study
evaluating the use of jarosite as a possible sulfate source in the binary
CAC-HH blend. This chapter aims to address the first objective on the list
by demonstrating the superior reactivity of one of the Fe-rich NFM slags
among the residues tested. This selection narrowed down the focus of the
last two chapters of the manuscript.
The reactivity of the Fe-rich slag in the binary (CAC-HH) and ternary
(CAC-HH-PC) systems are presented in Chapters 3 and 4, respectively. A
rapid non-destructive technique for quantifying the degree of slag
hydration using tomography is demonstrated in these chapters. The
physical properties (strength, setting time, and dimensional stability) of the
binders are evaluated using standard mortar samples. The phase
assemblage evolution is followed from the early hours (isothermal
calorimetry) up to 1 year of curing (XRD, TGA, FTIR) using paste
samples. The results were compared to models obtained from (GEMS)
thermodynamic simulation. Finally, the fate of Fe during hydration is
described at the end of both chapters using advanced microscopy and
spectroscopy techniques. The second objective of the research is addressed
by examining the potential role as activators of the PC (in the ternary
system), sulfates (by varying the CAC/HH ratio), and lime (in the form of
the paper residue).
ASTOVEZA 2022 7
____________________________________________________ Chapter 1
Chapter 1
State-of-the-art
Contents
Introduction .. 11
1.1 The SOCRATES Project .... 13

1.2 Calcium aluminate blended cements . .. 17
1.1.1. Pure CAC systems 18
1.1.2. Binary CAC-C$ systems .. 21
1.1.3. Ternary CAC-PC-C$ systems ..... 24
1.3 Supplementary Cementitious Materials .... 27
1.3.1. 1.3.1. Traditional SCM ............... 27
1.3.2. 1.3.2. Non-conventional SCM ... . 30
1.4 The role of Fe in inorganic polymers and cement binders 31
1.4.1. Fe speciation in slag-containing inorganic polymers 33
1.4.2. Fe speciation in cementitious binders ...... 35
1.4.3. Other techniques to characterize and quantify the
reactivity of slags 37
Conclusion ... 41
ASTOVEZA 2022 9
____________________________________________________ Chapter 1
Introduction
This chapter provides a wider outlook on the SOCRATES Project; calcium
aluminate blended cements; supplementary cementitious materials (SCM)
in calcium aluminate cement (CAC)-based systems; and the reactivity of
non-ferrous metallurgy (NFM) slags. It is essential to begin the discussion
with the origin of the SOCRATES residues in order to understand their
properties. This also helps assess the environmental impact of their
valorization as SCM in blended cements. This discussion is followed by
the state-of-the-art on the properties and hydration mechanism of CAC-
based blends in pure CAC, binary CAC-calcium sulfate (C$) and ternary
CAC-C$-Portland cement (PC) systems.
Moreover, an overview of the incorporation of traditional SCM and
industrial by-products (bottom ash, jarosite and NFM slags) in cements is
provided highlighting the need for the latter alternative materials and the
state-of-the-art in this application. The discussion in this particular section
n Appendix A. Finally,
a brief literature review on the reactivity of Fe-rich NFM slags in cement
and inorganic polymer (IP) systems is provided at the end of the chapter.
Among the SOCRATES residues, the NFM slags are of particular interest
considering their relatively low heavy metals content and their amorphous
nature. Both of these are generally considered as favorable properties of
SCM and will be elaborated in more detail in this section. The discussion
on the fate of Fe during hydration will provide a critical background for
the interpretation of results in Chapters 3 and 4 which are dedicated to the
addition of NFM slags in CAC-based binders.
ASTOVEZA 2022 11
____________________________________________________ Chapter 1
1.1. The SOCRATES Project

While promoting circular economy through valorization of industrial
residues, the SOCRATES project also aims to develop alternative sources
of critical and economically-important metals [26] such as indium,
germanium, gallium and antimony. As illustrated in Figure 1.1, the
recovery of these metals from industrial residues extracted from landfills
or directly from primary (ores) and secondary (recycled) metal production
flow lies within the scope of the project. Through technospheric mining,
whereby minerals are recovered from alternative stocks often considered
as wastes, the project aims to close the loop towards circular economy.
Figure 1.1 The scope of the SOCRATES project (top part)

presented in a generalized life cycle flow sheet for metals and alloys [2]
It has been reported that there are between 150,000 and 500,000 landfills
throughout EU which could be a potential source of secondary raw
materials and energy [27]. The benefits from their extraction do not only
range from economic (minimizing the landfill remediation cost) and
ecologic (land reclamation and health hazards from associated pollution)
aspects but could also trigger novel technologies to facilitate the metal
extraction [28]. Case in point, a wide range of metallurgical processes
including plasma-, bio-, solvo-, electro- and ionometallurgy are explored
in the SOCRATES project both in extracting valuable metals and in
ASTOVEZA 2022 13
Chapter 1 ____________________________________________________
removing the hazardous components of the residues [29 33]. Some of the
industrial by-products investigated include: NFM slags, MSWI bottom
ashes, jarosite and paper-fabrication residues. Most of which have been
left in stockpiles or landfills for a long period of time for the lack of suitable
application and some due to their toxic components.
The NFM slags in this study are coming from secondary copper and lead-
zinc smelting processes typically operated at 1200-1300°C in reducing
environment. They are generally dominated by FeO and SiO2, and contain
moderate amounts of Al2O3 and CaO. The degree of vitrification is
determined by the cooling rate following the tapping4. Granulation is often
performed by quenching in water to yield higher glass fraction as opposed
to air-cooling conditions. When slowly-cooled, the mineralogy is often
dominated by fayalite (Fe2SiO4) along with other olivine and spinel group
phases. Compared to ferrous metallurgy slags, NFM slags are typically
richer in trace elements (i.e., Cu, Cr, Mn, Pb and Zn) that may pose
challenge to their reuse in construction applications. Case in point, heavy
metals have been reported to retard the precipitation of portlandite with the
reduction of pH resulting from the hydrolysis of metal ions [34].
One of the promising technologies to refine NFM slags, particularly
fayalitic copper slags, in the project is the submerged plasma fuming
process (Figure 1.2). It aims to recover Pb and Zn, and remove other
volatile components (i.e., Cl, F, Ge) using an energy-dense reducing gas
made from a combination of heated compressed air and natural gas. The
volatile components are precipitated in a post combustion duct and
recovered in the form of oxide dusts [35]. The resulting slag, being
generally metal-depleted, is being marketed since 2018 as a clean synthetic
mineral for sustainable building applications [36].
Another clean (metal-depleted) commercialized by-product assessed in
this project is the paper residue. It is derived from the de-inking sludge
during the recycling of paper wastes in a fluidized-bed incinerator. This
process is designed for waste-to-energy applications, such that
approximately 7 GWh of electricity (or equivalent to the estimated annual
energy requirement of 151,000 households) is generated annually from the
incineration of 140,000 tons of paper pulp residue in this process alone
[37]. The remaining lime-rich residue after the combustion is marketed as
a green binder displaying pozzolanic activity that can be applied in dosages
up to 20 wt% [14].
4
removal of slag by allowing it to flow in a tap hole at the lower region of a blast furnace
14 ASTOVEZA 2022
____________________________________________________ Chapter 1
Figure 1.2 Submerged plasma-driven slag fuming process [35]
The other SOCRATES by-products, MSWI bottom ash and jarosite

residue, generally require further pre-treatment to better adapt their
properties according to their intended application. For the latter, this step
is often critical to reduce its toxic and heavy metals components. Jarosite
(NaFe3(SO4)2(OH)6), in general, is a family of iron-hydroxysulphate
minerals formed when iron sulfides are oxidized in ore deposits [38].
Residues of jarosite could be formed as by-products of zinc
hydrometallurgy, often classified as hazardous waste due to their heavy
metals (Pb and Zn) content [39]. In the SOCRATES project, the elemental
sulphur-rich waste stream from direct zinc sulphide leaching and the
jarosite-rich precipitates from the other direct leaching reactors are filtered
majority of these residues are currently contained in landfills, pilot scale

pyrometallurgical trials have been initiated to recover the metals and
produce a cleaner slag more suitable for construction applications [40].
On the other hand, the MSWI bottom ash investigated is heterogeneous,
with inclusions of plastics, stones, bricks, glass, metals and other unburnt
materials. Without prior assortment (i.e., sieving and magnetic separation)
to remove these unburnt fraction, the irregularly-shaped particles when
used as SCM could increase its water absorption capacity and
ASTOVEZA 2022 15
Chapter 1 ____________________________________________________
consequently reduce its strength [41]. In addition, thermal treatment might

even be necessary in some cases (i.e., high metallic aluminum and bottle
glass in fines fraction) to vitrify the bottom ash and obtain a denser
microstructure [42].
After pre-treatment, three valorization pathways are explored for the
metal-depleted by-products of the SOCRATES project, namely: as
heterogeneous catalysts; (2) as precursors to inorganic polymers (IP); and
(3) as SCM in CAC-based binders. These applications are directed towards
high-value, high-volume valorization relative to the more conventional
utilization in sandblasting and in aggregates replacement. Unlike the latter
applications, the reactivity of the residues will be critical in their role as
catalysts and binders.
For catalysis, the residual matrix in the SOCRATES flow sheet could be
interesting for specific reactions5 due to their high FeO and CaO contents
[2,43]. They could serve as cheaper alternatives for precursors although
the influence of the impurities on the catalytic performance must be
rigorously investigated to corroborate this application. On the other hand,
utilization as precursors to construction materials offers the high volume
valorization especially in the case of IP where the viability of complete
substitution of aluminosilicates6 with similar NFM slags has been
previously demonstrated [44 46]. In these systems, the slags are mixed
with alkali silicates or hydroxides to form a 3D-connected Si-O networks.
The abundance of Fe in the glass component of the NFM slags favors its
dissolution in the alkaline medium and could be a favorable criterium
[34,47]. Similar
to CAC, the durability and environmental advantages of IP binders relative
to PC has been driving its demand in niche markets [48,49].
The third valorization pathway, for which this PhD work is dedicated,
revolves around the incorporation of the by-products as SCM in CAC-
based binders. As previously highlighted in the introduction, this
valorization potential not only aims to recycle the industrial by-products
but also to reduce the carbon footprint of cement. The specialized market
of CAC dedicated to fast repair, aggressive environment, and high
temperature applications is a supplemental opportunity to those of the PC
systems. The limited work in the open literature concerning the addition of
similar industrial by-products in CAC systems also supports the novelty of
this study. As an SCM, it will be critical to assess the chemical reactivity
5
in particular: selective catalytic reduction of NOx with NH3 ; propane dehydrogenation to
propylene and propylene epoxidation
6
kaolinite, fly ash and blast furnace slags more conventionally used as precursors to IP
16 ASTOVEZA 2022
____________________________________________________ Chapter 1
of the by-products and to perform a screening test to narrow down the

selection. Based on this selection, the physical properties and the hydration
mechanism of the by-product with the highest potential are explored in
more details aided by advanced characterization techniques.
1.2. Calcium aluminate blended cements

Cements based on calcium aluminates marked more than a century of full
in 1908. His process is based on the fusion of low-silica aluminous and

ferruginous raw materials, mainly bauxite, with limestone. Initially, the
demand for this special type of cement has been driven by the durability
issues encountered with PC, particularly in sulfate-rich environment. An
even greater volume demand for CAC was realized in 1916 during the First
World War as its rapid hardening properties made it suitable for
construction of gun emplacements [50]. Rigorous tests and studies that
followed led to the production of various CAC products that we now know
today.
Figure 1.3 Two manufacturing processes for CAC production [18]
There are two manufacturing processes for CAC production, namely:

fusion and sintering (Figure 1.3). Fusion is operated at temperatures above
1400°C in a reverberatory furnace using limestone and bauxite as the main
raw components. This process yields the more traditional CAC products
(i.e., Ciment Fondu®, Secar 51®) at the lower alumina range, typically
below 52 wt%, and could be darker in color due to the higher Fe content.
On the contrary, the white varieties of CAC (i.e., Secar 71®), richer in
alumina, are produced from the sintering of alumina and lime at
temperatures above 1500 °C in rotary kilns. The latter is primarily used for
ASTOVEZA 2022 17
Chapter 1 ____________________________________________________
refractory and decorative usage, while the former is more adapted for
higher volume construction purposes [18,19].
CAC is generally dominated7 by monocalcium aluminate (CA) and could
alternatively contain significant proportions of calcium di-aluminate (CA2)
and mayenite (C12A7
gehlenite (C2AS) and ferrite phase (C4AF) are typically present and the rest
is principally balanced by smaller percentages (<5 wt%) of belite (C2S),
perovskite (CT), and spinel (MgAl2O4).
1.2.1. Pure CAC systems

CAC hydration is highly dependent on temperature, in contrary to that of
the PC for which hydrates formed are said to be generally similar at a
broader temperature range [51]. This temperature dependence defines
whether the hydrates formed are metastable (CAH10, C2AH8) or stable
(C3AH6). The metastable hydrates form, according to reactions given in
Equation 1.1 and 1.2, at lower temperature [52]. On the other hand, stable
hydrogarnet, C3AH6, is formed at elevated temperature as represented in
Equation 1.3. In real systems however, the presence of impurities that can
act as nucleation sites, and the kinetic effects can influence the formation
of the hydrate phases therefore shifting the below indicated temperatures
of formation [19].
<15°C Equation 1.1
27-50°C Equation 1.2
>50°C Equation 1.3
The above reactions occur in the solution following a dissolution-

precipitation mechanism [53]. Ca2+ and Al(OH)4- ions are released as the
main component of CAC, CA, dissolves in contact with water. A layer of
Ca[Al(OH)4]2 is generated around the surface layer of CA as the ions are
hydrated until supersaturation where the hydroxylated phase dissolves
again to form calcium aluminate hydrates, as those presented in the above
equations. In these reactions, the AH3 formed can take an amorphous or
crystalline form (i.e., gibbsite, bayerite or nordstrandite). Setting is
triggered by the consumption of one of the reactants or by the limited space
for hydrate precipitation [9]. Other possible reactions during the hydration
of pure CAC systems originate from the hydration of the minor
components (< 10 wt% in total), such as alite (C3S), belite (C2S), tricalcium
7
approximately <60 wt% CA for fused CAC and 60-70 wt% CA for sintered CAC
18 ASTOVEZA 2022
____________________________________________________ Chapter 1
aluminate (C3A), ferrite (C4AF) phase, and calcium carbonate (Cc)8, listed
in Equation 1.4 to 1.10 [19,54].
Equation 1.4
Equation 1.5
Equation 1.6
Equation 1.7
Equation 1.8
Equation 1.9
Equation 1.10
where y=H2O/Si; and (A,F) signifies Al2O3-Fe2O3 substitution

Aside from the difference in the principal active phase compared to PC,
the pore water pH of pure CAC systems is 11.4-12.5, which is typically
much lower compared to the 12.6-14 for the PC system [55]. This lower
pH is considered as an advantage for certain applications such as in
engineered barrier systems for containing high-level radioactive waste
where a clay layer of bentonite is often integrated. This clay layer is
reported to be altered in contact with hyper-alkaline solutions leading to
the reduction in its swelling ability [56]. In this application, CAC-based
blends can be an ideal alternative as structural cast concrete, grouting,
sealing of plugs and lining of tunnels [57 59].
However, the main challenge limiting the direct application of pure CAC
is the inevitable phenomenon called conversion. It is well-known to be an
unfavorable, but manageable, phenomenon in pure CAC binder where the
metastable hydrates are converted to stable hydrates (Equation 1.11 and
1.12). These reactions are triggered by the nucleation of C3AH6 and could
take only a few minutes to occur at elevated temperatures, i.e., 80°C, or
several years below 30°C [51,60]. Since these hydrates exhibit differences
not only in morphology (Figure 1.4) but more importantly in density and
solubility, the dimensional stability and hence the mechanical strength of
hardened binder is compromised when conversion occurs [61]. In the early
ldings
in the UK, for which conversion was associated, triggered the restriction
for the use of pure CAC binder in construction applications [62 64].
8
originating either as a component of the raw cement; or produced as a carbonation product of C
and CH with the CO2 gas from the atmosphere during curing/sample preparation
ASTOVEZA 2022 19
Chapter 1 ____________________________________________________
Equation 1.11
Equation 1.12
Figure 1.4 SEM images of the hexagonal CAH10 (a) and cubic C3AH6 (b) [65]
In order to mitigate the conversion phenomenon, the hydrate phase

assemblage can be altered by the addition of: (1) SCM, (2) sulfate carrier
or (3) by introducing PC in the system. The incorporation of the more
conventional SCM, including GGBFS, microsilica and metakaolin, have
been widely investigated in the past [22,66 68] despite these studies being
limited to siliceous mineral additives. These previous works have
demonstrated that the addition of SCM could reduce the conversion by
favoring the formation of strätlingite (C2ASH8) over that of the metastable
aluminate hydrates. In addition, SCM could also increase the capillary
water absorption by altering the pore structure. They can contribute to the
long term mechanical strength hence improving the overall stability of the
CAC binders, according to these studies.
On the other hand, the addition of sulfates in the system is another
approach with the same goal of shifting the hydrate phase assemblage to
avoid conversion. An ettringite-binder system is created with the addition
of sulfate-containing substances (i.e., gypsum, anhydrite, hemihydrate or
mixes thereof) with the main reaction presented in Equation 1.13. The
self-drying capacity, high early strength, and expansive properties of CAC-
based binders particularly in self-levelling mortars are linked to this main
hydration reaction [69].
Equation 1.13
In a similar manner, the addition of PC further increases the complexity of

the system by modifying the hydrate phase assemblage. In these (CAC-
C$-PC) ternary systems, ettringite (C6A$3H32) and C-S-H are formed in
the early stages of hydration but the assemblage changes with the depletion
of the sulfate source [70]. The corresponding reactions depending on the
proportion of the components are elaborated in more detail in section 1.2.3.
20 ASTOVEZA 2022
____________________________________________________ Chapter 1
Despite this major setback in the utilization of pure CAC systems, it is

clear that there are several solutions currently available which have been
widely demonstrated to mitigate the conversion phenomenon. The interest
in CAC binders continue to grow with recent advances directed towards
amorphous calcium aluminate cements (ACA) having mayenite, C12A7, as
the main reactive phase instead of CA. These latest variants were reported
to have lower carbon footprint and display faster hydration kinetics
compared to the more conventional CAC based on CA [71,72].
1.2.2. Binary CAC-C$ systems

During the first hours of hydration, the amount of sulfates added influences
the phase assemblage, setting time, rheology, and porosity distribution
[19]. When designing the formulations, the CA/C$ molar ratio is important
in determining the main hydration reactions. In lower sulfate systems
(CA/C$ > 2), the formation of ettringite (C6A$3H32) and aluminum
hydroxide (AH3) according to Equation 1.13 is the dominant reaction
during the early ages of hydration. The minor components of CAC could
also contribute to ettringite formation according to Equation 1.14 and
1.15.
Equation 1.14
Equation 1.15
When the sulfates are depleted, ettringite transforms into calcium

monosulfoaluminate (C4A$H12) as it reacts with the remaining CA and
C3A from CAC [63] (Equation 1.16 and 1.17). The presence of calcium
carbonate could also contribute to the consumption of ettringite, leading to
the precipitation of monocarboaluminate (C4AcH11) or
hemicarboaluminate (C4Ac0.5H12) according to Equation 1.18 and 1.19,
respectively.
Equation 1.16
Equation 1.17
Equation 1.18
Equation 1.19
In high sulfate systems (CAC/C$ < 2), the ettringite formed in the early
hours of hydration remains stable overtime. However, the excessive
ASTOVEZA 2022 21
Chapter 1 ____________________________________________________
gypsum allows the formation of secondary ettringite at later ages as the

less reactive components of CAC (i.e., C4AF) hydrate. Over-expansion
leading to disintegration of mortar samples has been previously reported
in such high sulfate formulations, associated to the crystallization pressure
of ettringite forming at later ages [73].
-situ isothermal
calorimetry [9] revealed two main hydration peaks for the majority of
CA/HH ratios investigated (Figure 1.5). Using in-situ XRD, these peaks
were determined to correspond to: (1) the rapid dissolution of HH leading
to gypsum formation; and (2) the massive precipitation of ettringite, likely9
accompanied by aluminum hydroxide. In the low sulfate system with
CA/HH molar ratio of 4 (80CA-20HH), a third peak was exclusively
observed attributed to the monosulfoaluminate formation as the sulfates
were depleted, simultaneous to the continuous dissolution of CA.
Figure 1.5 Heat flow curves of formulations with different proportion of CA and HH
measured per 10 g of paste samples with 0.4 water to binder (w/b) ratio at 20°C using
ex-situ isothermal calorimeter [9]
This observed influence of the sulfate amount to the phase assemblage

during the first hours of hydration, was also reflected at long term through
the thermodynamic modelling of a similar binary system (CA-anhydrite)
in Saout et al., 2018 [74] in Figure 1.6. In this model, the highest water
demand coincides with the maximum volume of ettringite formed due to
the high amount of bound water in its structure. This yields the
aforementioned self-drying capacity of ettringite which is a valuable
9
the amorphous nature of AH3 in this formulation hindered its detection using XRD
22 ASTOVEZA 2022
____________________________________________________ Chapter 1
property in self-levelling concrete applications. It shortens the waiting time

by eradicating the need to completely evaporate the excess mix water prior
to flooring installation [75].
Figure 1.6 Phase assemblage evolution of CA-C$ (anhydrite) paste

derived using thermodynamic modelling at 25°C and 0.8 w/b ratio. The volume of phases
are presented on the primary y-axis; while the w/b ratio needed for the full hydration of
CA is on the secondary axis [74].
In terms of the mechanical strength, maximizing the formation of stable

ettringite is often targeted to formulate high strength binders [76].
Compared to the typical hexagonal platelets of monosulfoaluminates
(Figure 1.7 a), ettringite is often characterized as having a needle-like
morphology (Figure 1.7 b). Han et al., 2021 [77] reported that these
relatively10 low-density needles of ettringite are able to fill up macropores
during hydration, contributing to the compressive strength of the binder.
However, the crystal growth pressure accompanying ettringite formation
is also reported to cause expansion strains during the early hours of
hydration by enlarging the distances between the anhydrous cement
particles [79].
10
ettringite has the lowest (true) density at 1.78 g/cm3 among the other common hydrate phases in
these systems such as monosulfoaluminates (2.01 g/cm3); gibbsite (2.42 g/cm3) and strätlingite (1.94
g/cm3) [78]
ASTOVEZA 2022 23
Chapter 1 ____________________________________________________
Figure 1.7 SEM micrograph of: (a)synthesized monosulfoaluminate [80] and

(b) ettringite [81].
crucial role in controlling the setting time of the dry mix. The typical
sulfate carriers (gypsum, anhydrite and HH) readily hydrate and set into
calcium sulfate dihydrate in contact with water leading to an accelerated
rate of expansion and hardening [79]. For this reason, set retarders (i.e.,
citric acid, tartaric acid) are often introduced as admixtures in order to
counteract the influence of sulfates. Nguyen et al.,2019 [82] reported that
at a significant dosage of 2 wt%, citric acid does not only retard hydration
but could also contribute to the conversion of ettringite to
monosulfoaluminate; and influence the mechanical strength of mortars.
This was associated to the formation of citrate ligand around ettringite
crystals that temporarily inhibits its precipitation, consequently favoring
monosulfoaluminate formation [83]. Nonetheless, citric acid is more
typically used in smaller amounts (0.1-0.5 wt%) only to control the setting.
No evident influence to the phase assemblage or to the mechanical strength
should be expected at this dosage [82].
1.2.3. Ternary CAC-PC-C$ systems

Previous studies [84 87] in ternary CAC-PC-C$ binders have
hypothesized the governing hydration mechanism in these system.
Bizzozero and Scrivener (2014) enumerated the hydration reactions and
discussed their kinetics by following the phase assemblage evolution over
time in a CAC-rich system [85]. Their study demonstrated the formation
of ettringite and AH3, principally in the early hydration days in a similar
manner as in the binary CAC-C$ systems according to Equation 1.13. PC
could further contribute to this initial ettringite formation by providing
additional Ca2+ ions in the form of portlandite (CH) in a reaction described
in Equation 1.20. C-S-H is formed as the main hydration product in PC-
rich systems (>80 wt% PC) along with other minor components (i.e.,
hydrogarnet, strätlingite, monocarboaluminates). This is better
24 ASTOVEZA 2022
____________________________________________________ Chapter 1
demonstrated in the thermodynamic models of a PC-rich (a) and a CAC-

rich (b) ternary systems by Qoku (2017) in Figure 1.8.
Equation 1.20
Figure 1.8 Phase assemblage evolution of: (a) a PC-rich ternary system; and (b) a CAC-
rich ternary system, simulated using GEMS thermodynamic modelling at 20°C
and 1 bar pressure. Phases were abbreviated as follows: (Ms) monosulfoaluminate; (Hc)
hemicarboaluminate; (Mc) monocarboaluminate [19]
In both the (80 wt%) PC- and (65 wt%) CAC-rich formulations modelled
in Figure 1.8, ettringite was predicted to form in a significant quantity and
remained stable throughout the simulated 90 days of hydration. C-S-H and
CH, the main hydration products in pure PC systems, were only present in
the PC-rich system. Instead, the silicates from the minor PC added were
predicted to be incorporated in strätlingite for the CAC-rich formulation.
Furthermore, these models illustrate that CH and AH3 do not co-exist
ASTOVEZA 2022 25
Chapter 1 ____________________________________________________
thermodynamically; and that the formation of Hc and Mc were likely

associated to the PC, typically containing a small amount (< 2 wt%) of
calcite. The only significant discrepancy that the author noted between the
modelled and the experimental phase assemblages were linked to the
kinetic effect. In the thermodynamic models, the components were
predicted to dissolve and diffuse uniformly while this may not be exactly
the case in the real system [19].
Torréns-Martin et al., 2013 [86] and Nehring et al., 2018 [87] both
highlighted the significance of CAC/C$ and PC/CAC ratio to the kinetics
of hydration. They explained that in CAC-rich systems with sufficient11
C$, ettringite layer could form around the anhydrous grains provisionally
delaying the silicates (mainly C3S and C2S) hydration as it retards the
diffusion of ions by creating a temporary barrier. This mechanism was
observed delay in PC hydration. Using isothermal calorimetry, Torréns-

Martin et al., 2013 [86] associated this mechanism to the observed
temporary (3-6 hours) halt in the increase of the cumulative heat release
curve of a similar ternary binder. When the barrier layer is broken with the
continuous dissolution of ions, the cumulative heat release continued to
increase again [86]. In terms of mechanical performance, Bizzozero and
Scrivener (2014) supported that PC is less reactive in the early hydration
and only contributes to the long term (>28 days) strength of mortars [85].
Despite the numerous studies relating to this ternary binder system, the
previous investigations were mostly limited to PC-rich formulations,
where PC takes 70-85 wt% of the binder in the dry mix. While these blends
are more representative of the industrial high volume applications
considering the lower cost of PC, the adaptability of CAC-dominated
blends in specialized applications along with its conceivable
environmental advantage remains indispensable. Case in point, the
maximization of the self-drying capacity of CAC in flooring applications
can be used as a criterion to optimize the CAC/PC ratio when designing
the dry-mix formulations [88]. In terms of sustainability, a Life Cycle
Analysis Assessment (LCA) report comparing blends for SLU flooring
mortars [24] had previously estimated a lower carbon footprint for a CAC-
dominated ternary binder compared to an equivalent volume of a PC-rich
ternary blend. Both displayed comparable applicative performance,
notably in terms of mechanical strength.
In CAC-based blends, PC is also added at lower proportions (10-30 wt%)
to improve the reactivity of GGBFS [19,89] not only by increasing the pH
11
at the amount where ettringite formation is favored over that of Ms
26 ASTOVEZA 2022
____________________________________________________ Chapter 1
of the pore solution, but also by contributing ions (Ca2+, Al(OH)4- and SO42-
) that can participate during hydrate formation. Thermodynamic modelling
estimates pH values above 12 in pore solution of pure PC hydration [90];
whereas a pH less than 12 is expected for CAC-rich systems even reaching
10.5 with excess C$ [9]. This advantage is further supported by the
improved stability of ettringite formed at pH 12 reported in Van-Rompaey,
2006 [91].
1.3. Supplementary Cementitious Materials

Green cements or eco-efficient cements [4] are produced using alternative
materials or energy sources with lower environmental footprint for a
performance-comparable product. Among other major considerations in
assessing the viability of replacement, the cost and the technical
performance of the derived formulations are standing out. The substituting
ibution to the properties
of the blend and the hydration reactions, if any, are purely physical in
nature. On the other hand, the term SCM is used for materials exhibiting
hydraulic or pozzolanic properties during hydration [12].
1.3.1. Traditional SCM

Snellings et al., 2012 [12] provided a comprehensive review covering both
natural and artificial SCM. The natural SCM are extracted raw from the
nature, either of volcanic or sedimentary origin, among which pyroclastic
materials, pumices, zeolitic tuffs, and naturally burned clays are included.
On the other hand, artificial SCM refer to thermally-treated materials
typically generated as by-products of industrial processes. These could
range from slags, incineration ashes, burned residues, silica fumes to waste
glass. Among these numerous types of SCM, only three materials are
considered to take a major role as SCM in the market over the past 25 years
[4], briefly described in this reference as follows:
GGBFS. A by-product of ferrous smelting having a global
generation of above 300 Mt annually. Historically, metal-depleted
ferrous slag were naturally cooled at ambient temperature and once
hardened, were sold as aggregates employing minimal
comminution. With extensive research supporting the latent
hydraulicity of GGBFS and demonstrating its potential as SCM,
granulation was integrated in the operation to make this by-product
marketable.
ASTOVEZA 2022 27
Chapter 1 ____________________________________________________
Fly ash. A by-product of coal combustion for electricity generation.

Despite its threefold annual production compared to GGBFS at
900 Mt, the heterogeneity of this material has kept its valorization
as SCM limited. Moreover, as a direct output of coal combustion
known to have a primary contribution to the anthropogenic CO2
emi
scrutiny [92].
Limestone. Formerly having a limited application as a filler, the
reported reactivity of this abundant natural SCM in the past decades
has expanded its application in the cement industry. In the recent
years, keen efforts have been dedicated to the development and
commercialization of a PC-based binder with substantial limestone
and calcined clay substitution, claiming a promising impact on
reducing the carbon footprint in the building sector [93,94].
The contribution of SCM during the first hours of hydration is said to be
limited to physical effect, given that their dissolution is often significantly
slower than the cement component in the blend [8]
mechanism introduced earlier in the discussion is commonly demonstrated
using heat flow curves from isothermal calorimetry, as presented in Figure
1.9 [95]. It shows that the 40 wt% replacement of PC with either slag or
fly ash enhanced the intensity of the main hydration peak, creating a
sharper acceleration slope and a prolonged deceleration period. As
explained earlier, this was attributed to the higher effective water to cement
ratio and the additional nucleation sites particularly for fine SCM [9]. The
displayed no significant chemical reactivity during the first 35 h of

hydration.
At longer hydration period, a reactive SCM is expected to influence the
mechanical strength evolution of the binder [96]. One of the methods used
to estimate the reactivity of slags is the hydraulic efficiency index, h, under
the French standard NF P 18 506 (1992)12. This value is calculated as the
ratio of the compressive strength of a (PC) mortar with 50 wt% SCM to
cement substitution, to that of a pure PC-binder mortar. The typical values
for GGBFS ranges from 0.6 to 0.85, after 28 days of hydration [97].
12
equivalent to the ASTM C 989 for GGBFS
28 ASTOVEZA 2022
____________________________________________________ Chapter 1
Figure 1.9 Heat flow curves of a pure PC system and blends with 40 wt% replacement
with SCM, calculated per gram of cement [95].
High level of substitution with SCM typically results in strength reduction,

especially in the early days of hydration when the SCM has not yet reacted
extensively. This is better illustrated in Figure 1.10 b where the
compressive strengths of PC-based mortars decrease in function of the
replacement level with GGBFS and fly ash. On the contrary, it can be seen
that the addition of very fine silica fume has an inverse effect, mainly
attributed to the higher packing density13 obtained from this binder
formulation [96].
A systematic investigation was performed in Lothenbach et al., 2011 [8]
connecting the properties of traditional SCM to the physical and chemical
properties of PC-based binders. In that study, the changes in pore solution
due to the SCM dissolution were linked to the mechanism of hydrate
formation and subsequently to the volume, porosity and durability of the
systems. It was highlighted that despite the reduced early strength at higher
replacement levels, the addition of SCM could ultimately contribute to the
durability aspects including: better control of alkali-silica reaction14;
increased chloride penetration resistance; sulfate resistance; or lower heat
of hydration.
13
pertains to the ability of the SCM to fill up the voids of the binder, associated to its PSD and the
morphology of the particles
14
deleterious expansion in concrete attributed to the formation of hygroscopic sodium silicate gels
from the reaction of highly alkaline cement paste, siliceous aggregates and water
ASTOVEZA 2022 29
Chapter 1 ____________________________________________________
Figure 1.10 (a) Particle size distribution of several SCM compared to that of a PC;
and (b) the compressive strengths in function of cement replacement level, relative to a
pure PC binder [96].
1.3.2. Non-conventional SCM

Despite recognizing numerous types of traditional SCM and their growing
application on a commercial scale, their limited and often localized
availability further triggered by the declining numbers of carbon-intensive
energy and steel manufacturing plants poses a critical challenge for a full
shift towards blended cements. To answer this demand for alternative
sources, the promising approach of the Enhanced Landfill Mining (ELFM)
concept [98] opens new pathways for potential types of novel SCM.
Landfill mining is a form of secondary mining whereby valuable materials,
such as high-value metal components, are extracted from ancillary sources,
particularly pre-existing landfills [99]. It is a long-existing concept elicited
by the diminishing supply of high grade ores, newly-developed metal
extraction technology and the increasing cost of land use.
30 ASTOVEZA 2022
____________________________________________________ Chapter 1
With an integrated sustainability target, the concept of ELFM conveys a

more rigorous approach on energy flow, social and ecological
considerations in landfill mining while placing the concept of circular
economy [100] in the limelight. ELFM offers a promising solution in
parallel with direct recycling of scraps and urban mining of end-of-life
products [27]. Prior to the utilization of the landfilled by-products, pre-
treatment is often necessary to recover their valuable components and to
better adapt their properties as SCM a concept aligned with the
framework of the SOCRATES project. Additionally, the pre-treatment
step provides a margin to engineer the by-products in order to induce or
increase their reactivity as SCM and to remove their heavy metals and
hazardous components making them more suitable for building
applications. This makes the remaining residues good candidates as
mineral additives: clean (metal-depleted), low-cost and high volume
materials. To have a substantial impact in reducing the carbon footprint by
replacing cements, these residues must effectively contribute to the
application-based performance of the binder thereby providing a direct
reduction to the clinker factor [4].
Some of the properties of the by-products investigated in the SOCRATES
project can differ substantially from those of the aforementioned
traditional SCM. The typical composition of similar by-products
published in literature [101 105] is presented in Table 1.1.
Table 1.1 Chemical composition of by-

published in literature [101 105]
bottom ash and jarosite residue relate
values below detection limit.
NFM slags MSWI jarosite
in wt% copper slag zinc slag lead slag bottom ash residue
CaO 5.2 14.5 10.5 38.1 6.9
SiO2 41 19.8 24.4 15.8 6.8
Al2O3 3.8 3.4 0.9 6.8
FeO or
44.8 37.2 46.8 4.2 32.1
Fe2O3
MgO 1.2 1.3 3.5 1.9
K2O 0.5 7.3 0.7
Na2O 4.1 0.2 0.6
SO3 1.1 4.4 6.6 1.5
TiO2 0.2 0.3
PbO 1.5 1.4 2
ZnO 14.2 9.2
others 3 8.4 0.7 28.2 33.2
Sum 100 100 100 100 100
ASTOVEZA 2022 31
Chapter 1 ____________________________________________________
Whereas the traditional SCM are typically aluminosilicates in

composition, NFM slags and jarosite residue are Fe-rich given their
production process described in section 1.1. Furthermore, the Fe
component in the NFM slags principally has 2+ oxidation state considering
the reducing atmosphere in smelting furnaces as opposed to the 3+ state in
cements [106]. On the other hand, MSWI bottom ashes usually have a
lower degree of vitrification as opposed to fly ashes considering the
cooling rates during their generation. The chemical composition and the
degree of vitrification are only two of the many factors [107,108]
considered to define the reactivity of residues and consequently their
potential for valorization as SCM.
A more detailed state-of-the-art review on the use of NFM slags, MSWI
bottom ashes and jarosite residues has been co-published by the author in
a conference article attached in Appendix A. The paper discusses the use
of these by-products as non-conventional SCM highlighting the advances
and opportunities in this valorization pathway both in the industrial and the
laboratory scale. Based on this review, their commercialization is still
widely limited to low-value applications particularly as aggregates in
concrete. Their generally lower reactivity, inhomogeneity, undesirable
components (i.e., heavy metals and glass) and other factors (i.e.,
availability of local supply) are among the biggest challenges. In addition,
the annual global production of NFM slags (30-40 Mt [103]), MSWI
bottom ashes (26 Mt [109]); and jarosite residues (6-7 Mt [110]) are
significantly lower than the traditional SCM, often limiting their use to
niche applications.
The review paper (Appendix A) also highlights the lack of publications in
the open literature dealing with the incorporation of these by-products in
specialized cement binders, such as in CAC systems. Oftentimes, the
previous studies evaluate their potential in pure PC or PC-rich systems.
Among the by-products discussed, the NFM slags could have the highest
potential as SCM considering their reported contribution to the mechanical
performance of the binders [28,101,111]; their relatively higher production
volume [112]; and the availability of pre-treatment installations to generate
a metal-depleted supply ideal for the construction market [35].
The second part of the paper presents the result of an LCA and exergy15
analysis on the use of NFM slag as a substitute to PC. The case study
demonstrated 26 % decrease in the exergy cost (in MW) mainly related to
the lower material circulation in the rotary kiln; and the 32 % lower Global
Warming Potential (in kg CO2) with the 30 wt% substitution of PC with
15
measure of work needed to perform a process or produce a material
32 ASTOVEZA 2022
____________________________________________________ Chapter 1
the slag. These sustainability measures further support the interest in the
valorization potential, particularly, of NFM slags as SCM. In this regard,
during hydration that could influence the physical and chemical properties
of these binders.
1.4. The role of Fe in inorganic polymers and cement binders
1.4.1. Fe speciation in slag-containing inorganic polymers

While the typical Fe in a conventional slag-containing cement systems are
mostly in their elemental form (for GGBFS) or in the 3+ oxidation state
(for C4AF) [113]16, those in the NFM slags are dominantly in the 2+ state
given the reducing atmosphere in the furnace during their generation [114].
The presence of several oxidation states (2+, 3+) and coordination numbers
(4, 5, 6) adds to the complexity of the hydration reactions of these NFM
slags [115]. In the recent years, publications investigating the Fe speciation
in NFM slags as binder precursors have been widely directed towards IP
[47,114 118].
Using Mössbauer spectroscopy, Onisei et al., 2018 [114] provided insights
on the atomistic structure of an industrially-quenched (80 wt% amorphous)
fayalite slag before and after polymerization. Three main Fe components
were derived from the deconvolution of the Mössbauer spectra of the raw
slag, namely: (1) Fe2+ in fayalite; (2) Fe2+ in a hercynite (FeAl2O4)-like
phase; and (3) Fe2+ in the glass state. During the polymerization, the Fe2+
component in the crystalline fraction only reacted to a very low extent. On
the other hand, those in the glass state were predominantly converted into
Fe3+ in an amorphous ferrifayalite-like phase most likely having a 5-fold
O-coordination [114]. These results were consistent with subsequent
works on synthetic slags in Peys et al., 2019 [116] further suggesting the
presence of an additional Fe3+ species as a minor component (15 %
absorption area) of the amorphous raw slag; and a new Fe2+ site in a
distorted octahedra as a reaction product. No new crystalline phase was
detected through XRD leaving a question as to the identity of this new
product formed despite its seeming resemblance to the structure of fayalite
minerals. Nonetheless, both studies demonstrated the role of the newly-
formed Fe3+ component as network former in the glassy silicate IP
network. Peys et al., 2019 [116] further elaborated that the observed
oxidation commences between 3 h to 6 h after mixing and progresses
16
The study only took the Fe speciation of the non-amorphous fraction in consideration
ASTOVEZA 2022 33
Chapter 1 ____________________________________________________
significantly within the first 24 h. This timeline was consistent with the
onset of the polymerization exotherm in the calorimetry performed in the
same study [116]. Peys et al., 2019 [115] considered O2 + 4 e- => 2 O2- and
Fe2+ => Fe3++ e- as possible half reactions governing this oxidation, the
result of their study on polymerization under N2 atmosphere revealed
oxygen imbalance in the system. Nonetheless, the study demonstrated that
the observed oxidation was only driven by the atmospheric conditions
during the later stages (>28 days), and not during the first (3) days after
mixing [115].
Figure 1.11 The structure of a synthetic NFM slag proposed in Siakati et al., 2020
[119] simulated based on MD simulations and PDF studies. The slag has a molar
composition of 0.33CaO:0.67FeO-SiO2 following (CaO+FeO)/SiO2 = 1. The [FeOx]
(where x = 5 or 6) and [SiO4] tetrahedra were represented in yellow and dark blue
planes, respectively
Using experimentally-obtained pair distribution functions (PDF) [120] in

conjunction with molecular dynamic (MD) simulations, Siakati et al., 2020
[119] proposed the glassy structure of a synthetic NFM slag graphically
represented in Figure 1.11. In this structure, Fe is mainly present in
tetrahedral (4-fold coordination) and pyramidal (5 or 6-fold coordination)
configurations. The randomly distributed Fe atoms tend to exhibit the
lower coordination with O when more Si atoms are present in the
neighboring environment, and the higher coordination with more Ca and
Fe atoms around [119]. Knowledge on the atomistic structure of the slag is
not only important in controlling its polymerization or hydration reactions,
but can also be decisive in optimizing the physical properties of the binder,
such as: its density and color [121]; and its setting and chemical shrinkage
[115,122].
1.4.2. Fe speciation in cementitious binders
34 ASTOVEZA 2022
____________________________________________________ Chapter 1
Fe in conventional cementitious systems can be contributed by the slag

(GGBFS), the clinker and the aggregates. The identification of Fe-bearing
phases using the classical (i.e., XRD and TGA) or even with the more
advanced techniques (i.e., X-ray absorption near edge structure (XANES)
spectroscopy) is made complicated by its poorly-defined coordination
geometry in glasses [123]. A recent investigation, Mancini et al., 2021
[124], discussed the evolution of Fe speciation in slag cements under
different curing conditions using synchroton-based techniques: XANES
and micro-XRF. The results demonstrated the oxidation of Fe2+ (mainly in
the form of C4AF) from the anhydrous blend to Fe3+ during hydration. The
most common17 Fe-containing hydration product was siliceous
hydrogarnet, (CaO)3(Al,Fe)2O3(SiO2)0.84(H2O)4.32, generally consistent
with the prediction based on thermodynamic modelling. It was explained
however that some of the small differences between the experimental and
the modelled phase assemblage were attributed to the kinetic effect and the
heterogeneity of the samples. Moreover, the corrosion (accompanied by
oxidation) of free Fe from the slag was characterized, revealing the
formation of several possible hydration products (i.e., Fe(OH) 2, FeOOH,
FeO, Fe2O3) determined by the abundance of O2 in the environment. They
also found that the oxidation was accelerated by the presence of chloride
ions in seawater, and was hindered by high pH in cement pore solutions
[124]. Considering the significant difference in the volume of free Fe and
its corrosion products [125,126], the reaction mechanism could be critical
in maintaining the dimensional stability of these binders.
Several other studies [106,113,127 134] on cement systems have
previously described the fate Fe from C4AF during hydration. Rose et al.,
2006 [127] emphasized that controlling the C4AF hydration could be
beneficial given its metal sorption capabilities; while Tao et al., 2021 [128]
reiterated its contribution to the wear and sulfate resistance of special
application cements. Changes in the coordination structure (bond type,
bond strength, polarity, etc.) are reported to be directly related to the
reactivity of C4AF [128]. This property could differentiate the reactivity
rate among crystals with varying extent of Fe-Al substitution. By
complementing Mössbauer spectroscopy and XRD results with atomistic
simulations18, Tao et al., 2021 [128] ranked the reactivity of cations in the
laboratory synthesized C4AF crystals from highest to lowest, as follows:
Ca > Al > Fe ions, implying that lower Fe-Al substitution begets higher
reactivity. They also concluded that tetrahedral coordination for both Al
17
except for the supersulphated slag cement (84.5 wt% slag, 15 wt% anhydrite and 0.5 wt% KOH)
due to the lower amount of Ca versus CEM II and III samples investigated in their study
18
quantum-mechanical simulation based on density functional theory
ASTOVEZA 2022 35
Chapter 1 ____________________________________________________
and Fe ions are more easily broken (more reactive) as opposed to

octahedral coordination.
Using Mössbauer spectroscopy, the distorted octahedral and tetrahedral
sites occupied by Fe3+ in C4AF were discerned, and the evolution of the
spectra revealed the incorporation of Fe in several hydrate phases
[129,130,135]. Figure 1.12 shows the evolution over time of the
Mössbauer spectrum of a pure PC sample (a) and the corresponding
deconvoluted models (b) [129]. The model supports the presence of two
Fe components from C4AF in the raw PC, decreasing in intensity over time
as a third doublet forms towards 14 days of hydration.
Figure 1.12 (a) Evolution of the Mössbauer spectrum of a pure PC hydrated

up to 48 days; and (b) its corresponding deconvoluted model with each doublet
representing an Fe component [129].
Aside from the aforementioned siliceous hydrogarnet, Fe uptake was

similarly observed in ettringite [131,136], monosulfate or monocarbonate
[127,132], and C-S-H [113,133], depending on the type and the abundance
of ions in the pore solution. In most of these phases, Fe-Al solid solution
was identified as the principal mechanism of incorporation. However,
these studies reported varying results regarding the formation of Fe
hydroxide from the hydration of C4AF. While some [127,128,132,134]
36 ASTOVEZA 2022
____________________________________________________ Chapter 1
supported the formation of amorphous Fe hydroxide phases (i.e., similar

to goethite and lepidocrocite), others [135,136] did not observe substantial
evidence of their presence. This could be explained by kinetic effects and
compositional variations creating a different environment during hydration
[124,132].
1.4.3. Other techniques to characterize and quantify the reactivity of slags

Fourier transform infrared spectroscopy (FTIR)
FTIR has been widely used together with Mössbauer spectroscopy in the
previously-cited investigations. It proves to be a practical complementary
technique with numerous advantages, including: (1) rapid results, obtained
after a few minutes; (2) only a few milligrams of sample needed; (3) no
need for specific sample preparation; (4) distinct peaks identified for
certain cement phases such as for portlandite; (5) proven applications in
IP, cements and coatings [137]. Using the Attenuated Total Reflection
(ATR) mode, FTIR has effectively demonstrated the reactivity of highly
amorphous Fe-rich slags characterized by a peak shift towards higher
wavenumbers of the Si-O stretching absorbance band (700-1200 cm-1) and
towards lower wavenumbers of the Si-O rocking band (400-500 cm-1) after
28 days of polymerization [114,116,117]. An illustration of this shift is
presented in Figure 1.13. In a wide variety of materials, this phenomenon
has been associated to the changes in the Si-O-T (where T could be Al, Fe
or Si) distances [138]. The incorporation of Fe3+ in the silicate networks is
reported to change the Si-O-Si linkages in the activating solution of IP
systems [116]. An idea on the degree of slag reactivity, relative to its glass
fraction, may also be derived in function of this peak shift, as shown in
Figure 1.13.
ASTOVEZA 2022 37
Chapter 1 ____________________________________________________
Figure 1.13 ATR-FTIR spectra of three synthetic Fe-rich (65-68 wt% FeO) slags
before (dotted lines) and after (solid lines) 28 days of polymerization modified from
Siakati et al., 2020 [117].
Selective dissolution
In a recent publication, Hallet et al., 2022 [139] adapted the ethylene
diamine tetra acetic acid (EDTA)-triethanolamine (TEA)-NaOH selective
dissolution test originally intended for fly ashes [140], to quantify the
degree of NFM slag reaction in PC-based blended cements. They
determined that almost 25 % of the slag has completely dissolved after 150
days of hydration and that the kinetics were comparable regardless of the
slag-PC substitution level (30 wt%, 50 wt% or 70 wt%) [139]. On the other
hand, a procedure using methanol-salicylic acid has been proposed for
GGBFS-CAC systems in Nakagawa et al., 1990 [141]. However, there has
been no methodology previously published for Fe-rich NFM slag in CAC-
Scanning Electron Microscopy (SEM)

Vayghan et al., 2021 [142] estimated the degree of slag hydration using
image analysis on the back scattered electron (BSE) micrographs of SEM.
This study on another fayalitic NFM slag in PC system (at 30 wt%
replacement level) estimated that 30 % of the slag has reacted after 90 days
of hydration. The methodology presented addressed the challenges
encountered in the more traditional techniques when dealing with the
highly amorphous slags (i.e., XRD with Rietveld refinement quantification
38 ASTOVEZA 2022
____________________________________________________ Chapter 1
[143]). It also took advantage of the wide compositional and density

difference between the Fe-rich slag and the cement matrix.
X-ray Computed Tomography (XCT)
Finally, another recently-published image-based quantification technique
for slag reactivity in a PC-based cement system utilizes XCT [144].
Collated figures taken from this study showing the main steps of image
analysis, in order to produce a 3D segmented render of the hardened paste
sample is shown in Figure 1.14.
Figure 1.14 Figures collated from the quantification of GGBFS hydration degree
using XCT in Wu et al., 2020 [144]. The reconstructed 2D image after the scan is shown
in (a). The grayscale threshold segmentation of the pixels under the green and red arrows
in (a) is illustrated in (b). The 2D (c) and 3D (d) renders after segmentation are also
presented.
The refined methodology introduced in Wu et al., 2020 [144]

differentiated GGBFS particles from the rest of the matrix through
grayscale threshold segmentation. Grayscale indices were assigned to a
pre-defined voxel according to the composition and density of phases, in a
similar manner to the aforementioned SEM-BSE technique but with the
transmission of X-rays instead [144]. This technique provides a rapid, non-
destructive alternative to SEM-BSE, eliminating the tedious sample
preparation steps including hydration stoppage and surface polishing. The
results of the study estimated that 56 wt% of the GGBFS was hydrated
ASTOVEZA 2022 39
Chapter 1 ____________________________________________________
after 28 days in a paste sample with 50 wt% PC replacement level at 0.5

w/b ratio. This methodology has not been previously adapted for cement
Considering the significant compositional and density difference attributed

to the Fe component of these slags, XCT is clearly a promising technique
to estimate the hydration degree.
40 ASTOVEZA 2022
____________________________________________________ Chapter 1
Conclusion
The SOCRATES project offers potential types of novel SCM in the form
of pre-treated industrial by-products, including: NFM slags, MSWI
bottom ashes, jarosite and paper residue. After the pre-extraction of
valuable components and the removal of heavy metals (i.e., fuming process
for NFM slags), the incorporation of the residual matrix in CAC-based
binders is evaluated in this study.
CAC is typically blended with calcium sulfate (C$) and PC in rapid-repair
building applications. The hydration products greatly depend upon the
ratio of these components in the mix. In general, the complete hydration of
a binary CAC-C$ system leads to an ettringite-rich binder given that
sufficient amount of sulfate is present. Whereas, the formation of
monosulfoaluminates is favored in low calcium sulfate systems. The phase
assemblage becomes more complex with the addition of PC where other
hydrates could form in greater proportions, such as C-S-H, hydrogarnet,
strätlingite, and monocarboaluminates. No previous study has been
published on the incorporation of residues similar to those of the
Among the residues investigated, the NFM slags are of particular interest
given their generally metal-depleted and amorphous nature. Previous
studies cited in this chapter, have demonstrated the reactivity of similar Fe-
rich slags in IP systems and of C4AF in cement systems. Using Mössbauer
spectroscopy, these studies consistently suggest the presence of several
oxidations states (2+, 3+) and coordination numbers (4, 5, 6) of Fe in both
systems. During polymerization or hydration, the oxidation process of Fe2+
to Fe3+ was discerned from the deconvolution of the Mössbauer spectra. Fe
uptake has been observed in several hydrate phases including siliceous
hydrogarnet, ettringite, monosulfate or monocarboaluminate, C-S-H and
even in several forms of Fe hydroxides.
Several techniques and methodologies have been employed to characterize
and even quantify slag reactivity. ATR-FTIR has been used in conjunction
with Mössbauer spectroscopy to indicate reactivity through distinctive
peak-shifts in the Si-O absorbance bands of Fe-rich slags. While the degree
of slag reaction in PC systems has been recently quantified using an
adapted EDTA-TEA-NaOH selective dissolution test. Finally, image
analyses from SEM-BSE micrographs and XCT 3D volumes also
demonstrated promising alternatives to quantify the degree of slag
reaction.
In the end, the challenge of this research work lies in the complexity of the
system. Reactions arising from the hydration of the ternary CAC, PC and
ASTOVEZA 2022 41
Chapter 1 ____________________________________________________
C$ binders are expectedly difficult to distinguish, as they occur

simultaneously. This is further aggravated by the heterogeneous
compositions of the SOCRATES residues considering their diverse
origins. A screening test that allows comparison of the by-
reactivity at equal volume substitution, must be established to define a
blend with the highest potential valorization in cement applications. The
influence of the SCM addition to the physical and chemical properties of
CAC-based blends must be followed from the early hours up to longer
curing ages. The abundance of Fe and the amorphous nature of some of the
residues, particularly the NFM slags, would require more advanced
characterization techniques such as TEM, FTIR and Mössbauer
spectroscopy, to complement the techniques more commonly employed in
cement systems (i.e., XRD, TGA). An adaptation of the recently-published
hydration degree estimation based on XCT and image analysis is also
interesting to develop, considering the distinct properties of the slag (i.e.,
high density, Fe-rich composition) relative to the cement components; and
the fast, non-destructive nature of the test.
42 ASTOVEZA 2022
____________________________________________________ Chapter 2
Chapter 2
Materials, methods and residue
selection
Contents
2.1 Materials and formulation 45
2.1.1. The Socrates residues 45
2.1.2. Cements .. .. 48
2.1.3. Other blend components 48
2.1.4. Dry mix formulations 48
2.2 Methods ... . 50
2.2.1. Basic characterization of raw materials 50
2.2.2. Physical properties 53
2.2.3. Thermal and thermo gravimetric analyses 56
2.2.4. Tomography 56
2.2.5. Spectroscopy 59
2.2.6. Microscopy ... 64
2.2.7. Other techniques 69
2.3 Selection of industrial residue 69
2.3.1. Electric conductivity measurements 69
2.3.2. Heat Release 70
2.3.3. Setting time and mechanical strength 71
2.4 Evaluating jarosite residue as a sulfate source 73
Summary 79
ASTOVEZA 2022 43
____________________________________________________ Chapter 2
2.1. Materials and formulation
2.1.1. The Socrates residues

Six by-products from the SOCRATES Project were investigated in this
study. A single batch of 10-20 kg of each material was received in buckets
originating from several industrial partners all around Europe. The
materials were homogenized and sampled to obtain a representative 3kg
portion. The samples were then oven-dried for 48 hours at 100°C prior to
grinding using a laboratory ball mill (if needed) until a comparable particle
size distribution (PSD) and/or specific surface area (SSA) as the reference
calcium aluminate cement (CAC) is obtained. A screening test was
performed as a preliminary step in order to determine which among these
by-products had the highest potential for valorization as supplementary
cementitious material (SCM) in a CAC blended formulation and therefore
will be the focus of the more extensive investigation that follows. These
by-products include:
Granulated Fe-rich slag (slag)

A by-product of a pilot-scale thermal treatment of residues from lead-zinc
metallurgy. The granulation process using water jets resulted to its glassy
structure. This was performed for the ease in handling and for the increased
reactivity when used in cementitious applications. Its dark color is a
manifestation of its high Fe content. Unless otherwise specified, the
manuscript.
Figure 2.1 Picture of the granulated Fe-rich slag as received
Slowly-cooled Fe-rich slag

A by-product originating from the same process as the granulated Fe-rich
slag but instead undergone air-cooling at an ambient temperature.
ASTOVEZA 2022 45
Chapter 2 ____________________________________________________
Figure 2.2 Picture of the slowly-cooled Fe-rich slag sample as received
Fumed fayalite slag (Koranel®)

A commercial product from Metallo marketed as clean synthetic mineral
for building applications. It is a granulated Fe-silicate originating from
secondary copper smelting where the slag is further subjected to a
submerged plasma zinc fuming process for Zn and Pb recovery. In effect,
the final granulated slag is expected to only have trace amounts of heavy
metals making it an ideal mineral additive for construction applications.
Figure 2.3 Picture of the fumed fayalite slag sample as received
Municipal solid-waste incinerator (MSWI) bottom ash

One of the principal residues produced from the incineration of municipal
waste. It consists of the coarser and denser fraction of the ashes typically
extracted from the bottom of the furnace. Often positioned in the lower
extremities, a portion of these materials could remain unburnt.
Consequently, the composition could be widely heterogeneous as observed
from the received materials where broken glasses and small fragments of
plastic wastes are still identifiable.
46 ASTOVEZA 2022
____________________________________________________ Chapter 2
Figure 2.4 Picture of the MSWI bottom ash sample as received
Jarosite residue
Dried filter cake coming from a combined stream of a direct zinc sulphide
leaching stage and direct leaching reactors. This residue is often contained
in specially constructed tailing ponds or industrial landfills depending on
the levels of hazardous components.
Figure 2.5 Picture of the jarosite residue sample as received
Paper residue (Top-Crete®)

A commercial product from CDEM Minerals BV marketed as a green
binder for concrete and cement industry. This white fine powder is
produced in a fluidized-bed incinerator that dehydrates de-inking sludge
formed from the recycling of paper wastes.
Figure 2.6 Picture of the paper residue sample as received
ASTOVEZA 2022 47
Chapter 2 ____________________________________________________
2.1.2. Cements
Three types of cements were used in this study, namely:
Ciment Fondu® and Secar 51® (CAC)

Two different varieties of calcium aluminate cement produced by Imerys
Aluminates through a fusion process. Throughout this study, this
Ordinary Portland cement, Milke® Premium (PC)

A CEM I 52.5 R type ordinary Portland cement manufactured by
Heidelberg Cement.
2.1.3. Other blend components
-hemihydrate, Prestia Selecta (HH)

Mineral additive manufactured by Lafarge, used as the primary sulfate
carrier in the blends. Throughout this study, this component is mostly
Millisil E400® (quartz filler)

Fine sand filler manufactured by Sibelco. In this study, it was used to
substitute SCM in mortars with the goal of accounting for filler effect.
Citric acid
Reagent grade admixture from Sigma-Aldrich, in granules/crystal form,
added to the dry-mix blends primarily as a set retarder.
2.1.4. Dry mix formulations

EN 196-1 standard mortars were prepared based on several dry-mix
formulations for various case studies involved this research. These mortars
were used in monitoring the physical properties (mechanical strength and
dimensional stability) of the blends. Their corresponding paste mixtures,
maintaining the same proportions of the binder components, were used for
further characterization (i.e., mineralogy, tomography, thermogravimetry,
and spectroscopy). The water to binder (w/b) ratio was kept constant at 0.5
for all the mortars and 0.65 for all the paste samples. The higher amount
of water used for the latter was directed towards compensating for the
water-curing condition followed only for the mortar samples. Further
details on these curing conditions will be given in the succeeding sections.
48 ASTOVEZA 2022
____________________________________________________ Chapter 2
- reference filler: the SOCRATES residue component of the binder

was replaced by an equivalent volume of the quartz filler; or
- reference cement: the binder consists solely of cements (CAC, PC)
and additives (HH, citric acid) preserving the same ratio of the
components as with the SCM-containing formulation
The first case was more often used for the mortars in this study in order to
distinguish any possible contribution to the physical properties of an SCM
as an inert filler. On the other hand, the second case was mostly used for
the pastes where the chemical reactivity is being investigated.
A screening test is presented at the end of this chapter (section 2.3) with
the binder compositions detailed in Table 2.1 to compare the reactivity of
the SOCRATES residues in a CAC-based system. In this test, the variety
of CAC used was Secar 51®. It is notable that the substitution by equal
volume resulted in varying weight percent composition of each SCM.
Table 2.1 Binder composition for the screening test where Fe-rich slag is abbreviated
as FS. Values are expressed in wt% except for the CAC/HH ratio
total CAC/HH
(in wt%) SCM/filler CAC HH
binder ratio
reference 0 82 18 100 4.5
quartz filler 24 62 14 100 4.5
granulated FS 30 57 13 100 4.5
slowly-cooled FS 29 58 13 100 4.5
fumed fayalite slag 29 58 13 100 4.5
bottom ash 24 62 14 100 4.5
In another case study under the screening test, the use of the jarosite residue
as a possible substitute to HH was explored. The binder composition used
is shown in Table 2.2. In this test, the variety of CAC used is Ciment
Fondu®.
Table 2.2 Binder composition for the case study on jarosite use as a sulfate source
in CAC- blended cement. Values are expressed in wt%
code (in wt%) jarosite CAC HH total binder

no jarosite reference 0 67 33 100
20JAR 20% substitution 7 67 26 100
Chapters 3 and 4 focused on the incorporation of the granulated Fe-rich

slag in CAC-HH binary and CAC-HH-PC ternary systems. The binder
formulations are detailed in Table 2.3.
ASTOVEZA 2022 49
Chapter 2 ____________________________________________________
Table 2.3 Composition of binders incorporating the granulated Fe-rich slag

where certain components are represented as follows: (TC) TopCrete® paper residue
and (PC) Portland cement. (CAC) Secar 51®
total CAC/HH
(in wt%) slag CAC HH TC PC
code binder ratio
A 4.5CAC-HH 30 57 13 0 0 100 4.5
B 1.6CAC-HH 30 43 27 0 0 100 1.6
A-TC 4.5CAC-HH-5TC 29 54 12 5 0 100 4.5
A-PC 4.5CAC-HH-30PC 30 33 7 0 30 100 4.5
B-PC 1.6CAC-HH-30PC 30 25 15 0 30 100 1.6
reference formulations:
REFcement (reference cement) no SCM nor quartz filler added
REFfiller (reference filler) slag replaced by equal volume of quartz
A shortened version of the sample name is provided in the first column.

Throughout this study, the full sample name and their shortened code may
be used interchangeably or together to occasionally recall the composition
of the binders. The reference formulations in this case are distinguished
either as reference cement or reference filler specified in the table above.
In all the binder formulations presented, a replacement level of 30 wt% for
most of the SCM was selected with the goal of ensuring that the
performance of the binders will not be significantly reduced at short term
by the substitution compared to the reference formulation. As already
mentioned in the literature review (section 1.3-1.4), these industrial
residues tend to be less reactive than the other blend components (CAC,
HH and PC). On the other hand, the risk of not being able to detect any
impact of the substitution makes it unpractical to go below the 30 wt%
level. Moreover, two CAC/HH ratios were set at 1.6 (high calcium sulfate
system) and 4.5 (low calcium sulfate system) with the goal of representing
the impact of sulfate proportion in the systems and its possible influence
on the reactivity of the slag. In the end, it is important to mention that the
selection of the binder compositions for this study was based on the results
of literature references and guided by unpublished internal laboratory tests.
2.2. Methods
2.2.1. Basic characterization of raw materials

The chemical composition of the raw materials presented in Table 2.4 was
measured using semi-quantitative Bruker S4 Pioneer X-ray fluorescence
(XRF).
50 ASTOVEZA 2022
____________________________________________________ Chapter 2
Table 2.4 Chemical analysis of raw materials obtained using XRF

where Fe-rich slag is abbreviated as FS. Other components not specified primarily
pertain to the oxides of Zn, Cu, Cr, P and Pb (more details in Table 2.5)
(in wt%) CaO Al2O3 SiO2 FeO/Fe2O3 SO3 MgO K2O Na2O others
granulated FS 12.5 4.8 25.9 46.0 0.7 2.2 0.6 4.0 3.3
slowly-cooled FS 17.6 6.0 31.1 33.4 0.7 3.0 0.5 4.8 2.9
fumed fayalite slag 4.2 10.3 30.6 42.4 0.8 1.3 0.2 2.1 8.1
bottom ash 18.3 11.4 42.9 9.9 2.9 2.3 1.3 5.6 5.4
jarosite residue 4.3 1.0 6.4 22.0 51.2 0.5 0.3 4.8 9.4
paper residue 68.1 11.0 15.2 1.1 0.4 0.5 0.4 0.5 3.0
Secar 51® 37.7 52.3 4.9 1.8 0.2 0.4 0.0 0.0 2.7
Ciment Fondu® 37.8 40.3 4.2 14.6 0.1 0.4 0.1 0.1 2.5
PC 67.1 3.9 20.9 1.4 4.3 0.7 0.8 0.3 0.6
HH 41.7 0.1 0.8 0.1 56.5 0.1 0.0 0.0 0.7
quartz filler 0.0 0.0 99.8 0.0 0.0 0.0 0.0 0.0 0.2
High Fe content is apparent for most of the SOCRATES residues except

for the siliceous bottom ash and the lime-rich paper residue. The difference
in the Fe content of the CAC used is also significant with a higher
contribution measured for Ciment Fondu® as opposed to the Secar 51®
having higher alumina content instead. The relevance of this difference is
further elaborated during the characterization of the hydrated binders in the
preceding sections. Although Fe oxide is expressed as Fe2O3 in XRD for
most of the materials, the oxidation state Fe2+ specifically for the slags is
more sensible given the reducing environment in the furnace during their
production as previously discussed in the preceding chapter. More
advanced characterization to better delineate the Fe speciation in the slag
is presented in section 2.2.5.
Table 2.4 is the summation of Zn,
Cu, Cr, P and Pb oxides. Among the SOCRATES residues, this value for
the fumed fayalite slag, bottom ash and the jarosite residue is strikingly
elevated with details shown in Table 2.5 posting an initial concern on their
direct addition as SCM in the dry mix. Concerns on the effect to the
hydration reactions and on the leachability of the heavy metals and toxic
components, already introduced in the preceding section, have to be
addressed. Despite this initial observation, however, it has to be reiterated
that most of these materials from the SOCRATES project will still undergo
pre-treatment process to reduce these minor components that are not ideal
for cementitious applications.
ASTOVEZA 2022 51
Chapter 2 ____________________________________________________
Table 2.5 Minor components of some of the raw materials not detailed in Table 2.4
(in wt%) ZnO CuO Cr2O3 P2O5 PbO others
fumed fayalite slag 2.3 0.5 1.6 1.6 0.0 2.1
bottom ash 0.5 0.4 0.1 1.5 0.1 2.8
jarosite residue 3.3 0.1 0.0 0.1 4.4 1.6
In addition, the mineralogy of the raw materials was determined using the
D8 Advance X-ray Diffractometer (XRD) from Bruker coupled with
LynxEye detetor. CuK (1.54 Å) radiation was used with a step size of
-80°.
DiffracEVA and TOPAS software were used for the phase identification
and quantification. The amorphous content of the slag samples was
determined by applying the method described in et al., 2001 [145]
where 10 wt% of analytical grade zincite is added as an internal standard
during sample preparation. The XRD diffractograms and phase
quantification are detailed in Appendix B. Generally, both the granulated
Fe-rich slag and the fumed fayalite slag exhibited high amorphous content
(>90 wt%). The slowly cooled Fe-rich slag yielded a much lower
amorphous content (13 wt%), balanced by fayalite and spinel in the
crystalline fraction. On the contrary, the other residues (bottom ash,
jarosite and paper residue) were dominated by several crystalline phases.
The density of the raw materials shown in Table 2.6 was obtained through
helium displacement using the Micrometrics AccuPyc II 1340 pycnometer
system. Particle size distribution (PSD) presented along was determined
using the Malvern Mastersizer 3000 laser diffraction system. Specific
surface area (SSA) was quantified using both a semi-automatic Blaine Air
Permeability Apparatus and the Micrometrics BET Tristar 3000.
Table 2.6 Density, particle size distribution and specific surface area of the raw materials
density PSD SSA
Blaine BET
g/cm3 d10 d50 d90
(in wt%) (cm2/g) (m2/g)
granulated FS 3.67 1.1 7.9 50.5 4370 1.352
slowly-cooled FS 3.47 0.9 17.5 84.6 3200 1.216
fumed fayalite slag 3.50 1.6 13.7 61.5 3460 1.363
bottom ash 2.70 0.9 11.2 53.3 6160 3.865
jarosite residue 2.91 0.6 5.4 104.0 5253 4.706
paper residue 2.80 0.6 4.0 38.7 7730 8.133
Secar 51® 3.02 1.1 11.6 52.3 4270 1.019
Ciment Fondu® 3.23 1.6 16.4 75.8 3300 0.876
PC 3.17 0.9 8.5 25.3 3412 0.992
HH 2.64 0.6 3.7 22.7 7580 3.519
quartz filler 2.66 1.2 12.4 40.6 4580 1.118
52 ASTOVEZA 2022
____________________________________________________ Chapter 2
The goal of the grinding step during the sample preparation was to obtain
a comparable PSD and SSA between the SOCRATES residues and the
components of the blends to be substituted (CAC, PC and HH). Although
the PSD plot of the raw materials presented in Appendix C could support
that the curves relatively fall within a comparable range, it can be seen in
Table 2.6 that the bottom ash, jarosite and the paper residue have
significantly higher SSA. This is significantly related to the previously-
discussed difference in mineralogy of the materials and should be
considered during the screening test in section 2.3.
2.2.2. Physical properties

Standard mortars (4 x 4 x 16 cm3) were prepared according the norm
EN 196-1 using a w/b of 0.5. These samples were used in monitoring the
physical properties of the formulations including setting time, mechanical
strength and dimensional variation.
Setting time
Initial and final setting times were determined using the Vicatronic
Automatic Vicat Machine at 20°C (AFNOR room) without submersion to
water during curing. 1.3g of citric acid (0.3 wt% of the binder) was added
to formulations B, A-TC, A-PC and B-PC (refer to Table 2.3) to retard the
setting and therefore provide sufficient time for placing the freshly-mixed
blends in the moulds. Such low amount of citric acid (0.1-0.5 wt%) is not
expected to influence the phase assemblage evolution nor the strength
development of the mortars in ettringite-based binders according to
Nguyen et al., 2019 [82].
Mechanical strength
Compressive and flexural strengths were measured using the 3R Syntris
and the 3R RP 10/300 ELC press respectively, following the EN 196-1
standard. After demoulding, the mortars were cured under water at 20° C.
Each data point taken for a specific curing age represents the average value
from three mortar samples for flexural strength or six half samples for the
compressive strength with the standard deviation shown as error bars.
In parallel, the water-curing condition after extraction from the moulds was
substituted by air-curing conditions at 20°C and 65% relative humidity for
some of the slag-incorporating blends and their respective references in
order to investigate the effect of the curing condition to the mechanical
strength of the mortars. The results presented in Appendix D support the
advantage of following the water-curing procedure in discerning the
possible contribution of the slag to the long term (>28 days) strength
development of the mortars.
ASTOVEZA 2022 53
Chapter 2 ____________________________________________________
In this study, two criteria were used to evaluate the strength development
of the mortars calculated according to Equation 2.1 and 2.2. SAIcement
refers to the ratio of the mechanical strength of an SCM-containing mortar
to that obtained from its reference cement mortar. On the other hand,
SAIfiller is the ratio of the mechanical strength of the slag-containing mortar
to that obtained from its reference filler mortar. The evolution of these
indices over the curing period was used as a preliminary indication of slag
reactivity in Chapters 3 and 4.
Equation 2.1
Equation 2.2
where: SAI is strength activity index of the SCM-containing mortars; and

mortars with
subscripts SCM (slag or filler), cement (reference formulation with no
SCM), slag (granulated Fe-rich slag) and (quartz) filler.
Dimensional variation
Considering the expansive nature of ettringite binders, the dimensional
stability of the mortars was monitored during curing. Two methods were
used, described as follows:
a. Using 4 x 4 x 16 cm3 standard mortars cured under water:
Figure 2.7 Standard mortar mounted in an extensometer
Length variation was followed with the aid of gauge studs cast into
the opposite ends of the mortars. At different curing ages, the same
mortars were taken out of the water bath to record the length using a
digital length comparator apparatus (extensometer, see Figure 2.7)
54 ASTOVEZA 2022
____________________________________________________ Chapter 2
with a resolution of 0.001

demoulding at 24 h were considered as the reference (zero change)
point for the measurements.
b. Using the Walter+Bai shrinkage test in air-curing condition:

The SWG H-400 device type was used to measure the length
variation of a triangular mortar sample with 7 x 7 x 7 cm3 base and
40 cm maximum length. Measurements were continuously recorded in-
situ, a few minutes after mixing, through displacement transducers
positioned at the opposite ends of the freshly-casted blends. In some
cases, the monitoring of the length could continue after extracting
(between 24 and 48 hours) the solidified samples using extensometers
similar to method (a).
Figure 2.8 Walter+Bai shrinkage test set up (a) and the dimensions of the triangular
prism sample after the in-situ measurements of dimensional change (b, c).
Lengths are shown in centimeters.
Preparation of paste samples for analyzing hydrated blends

The subsequent sections concern several techniques that were used not
only to characterize the raw materials but also to investigate the heat
release, the phase assemblage evolution and the microstructure of the
hydrated binders over time. For these analyses, paste samples were
prepared following the formulations presented in Tables 2.1-2.3. 100 g of
dry mix blend per batch was mixed with 65 g of water and homogenized
using a laboratory electric agitator (IKA RW 20 overhead stirrer with
centrifugal impeller) at 250 rpm and a mixing time of 2 min. The freshly-
ASTOVEZA 2022 55
Chapter 2 ____________________________________________________
prepared pastes were proportioned into smaller moulds, each representing

a curing age (i.e., 1, 7, 28 days). The containers were hermetically sealed
and agitated in a Turbula® mixer until the paste hardened, usually between
7-8 min. This protocol aimed to minimize, if not eliminate, the preferential
segregation of the higher density solid components of the blends
(especially the Fe-rich slags) prior to setting. Furthermore depending on
the technique, stopping the hydration in hardened pastes was executed
through solvent exchange by immersing the crushed samples in
isopropanol for approximately 5 days and subsequently washing with
diethyl ether to aid the drying. In most cases, further grinding to obtain a
particle size below 100 µm was necessary prior to the characterization.
2.2.3. Thermal and thermo gravimetric analyses
Isothermal calorimetry
Ex-situ isothermal calorimetry was performed at 20°C on 10 g of freshly-
prepared paste samples using the TAM air (TA instruments) device.
Thermogravimetric analysis (TGA)

50 mg ground (<100 µm) paste samples were used for the
thermogravimetric analysis following solvent exchange to stop the
hydration. All tests were performed under N2 atmosphere at a heating rate
of 30 °C/min. Differential thermogravimetric analysis (DTG) curves were
generated from the treated data obtained using the Mettler Toledo
TGA/DSC simultaneous thermal analyzer. The values of bound water
content correspond to the weight loss measured between 50°C and 500°C
per 100 g of anhydrous blend at 0.65 w/b ratio (65 g water), calculated
according to the following equation:
Equation 2.3
where: bound water is expressed in grams (g) and mx represents the mass
of the sample at x = 50°C or 500°C.
2.2.4. Tomography
A methodology to estimate the degree of slag hydration using a
combination of X-ray computed tomography (XCT) and volume analysis
was introduced in this study. Micro-CT images of hardened slag-
containing paste samples were acquired using the Phoenix Nanotom S
56 ASTOVEZA 2022
____________________________________________________ Chapter 2
XCT (GE Inspection Technologies). This technique takes advantage of the

relatively high density and atomic number of Fe resulting in higher
greyscale value assignment to the reconstructed slag particles. After 24
3
hours of curing in plastic containers, ) of the
hardened paste wrapped in Parafilm® tape was mounted into the sample
holder for the scan. For the main results highlighted in the next chapters,
the hydration of a single sample per blend was followed over time
eliminating the need to perform solvent exchange to stop the hydration.
Figure 2.9 Picture of the XCT unit at the Materials Science Department of KU Leuven
used in this study (a) and of a hardened paste wrapped in Parafilm® and mounted in the
sample holder for XCT (b)
A 360° rotation scan using a step size of 0.15° was performed to obtain
tomographic projections. The X-ray source was regulated at 90 kV
(voltage) and 170 µA (current) in the nano focus mode 1. Tungsten on
CVD synthetic diamond was installed as a target and a 0.1mm thick copper
filter was mounted to harden the incident X-ray beams. A balance to
maximize the scan area while maintaining the precision lead to the
selection of 2.15 µm as the isotropic voxel resolution. This means that any
change in the scan properties below the particle size 2.15 µm was not
detected at this resolution. 2400 projections were taken with 500 ms
exposure time yielding a total scan time of 1 hour and 26 min per sample.
The position of the sample was saved for the next scan after a pre-
determined curing period.
Image reconstruction was performed in the Datos|x software producing
cross-sectional stack images. Using the 16-bit greyscale reconstructed
projections, threshold segmentation was done in the Bruker CT-Analyser
(CTAn) v 1.18.8.0+ software. A standard series of operations for image
correction described in Appendix E was followed in order to produce
binary image stack and calculate the total volume occupied by either of the
components (1 or 0) in the scan volume. This was then uploaded in the
ASTOVEZA 2022 57
Chapter 2 ____________________________________________________
Bruker CTVox v 3.3 software to generate 3D volume renders of the

samples at different curing ages. The total slag hydration degree was
estimated using the total percent volume of the slag after 1 day of hydration
as the zero reactivity reference. It follows the assumption that the degree
of slag hydration after 1 day is negligible and that all the slag dissolved
participated in the hydration reactions. Therefore, the total percent of
hydrated slag was obtained using Equation 2.4.
Equation 2.4
where: is degree of hydration; V is the total volume of the slag; and i is

the curing time for which the hydration degree is being estimated.
A sensitivity analysis was performed to demonstrate the effect of the lower
greyscale threshold limit assigned to the slag particles during the
segmentation, to the estimated total volume of the slag. Moreover, a
reference hardened paste sample (hydration is stopped) was scanned in the
beginning of each test for a curing period in order to correct displacement
of the greyscale values due to any possible time-dependent changes in the
X-ray beam intensity.
Finally, the same methodology was adapted to hardened paste samples
(formulation A, see section 2.1.4) where: (1) the volume of interest (VOI)
is changed from cubic to cylindrical; (2) and solvent exchange was
performed to stop the hydration after 1 and 90 days of curing. In effect,
instead of following the slag hydration degree of a single paste sample, two
different samples of the same formulation were scanned for 1 day and 90
days. The estimated hydration degrees were plotted together with those
obtained from the 9 mm3 cubic VOI without stopping the hydration. This
is done to demonstrate the precision of the estimation using XCT even
when the procedures are modified.
The development of the protocols for this technique was critical in this
research work faced with the difficulty of quantifying the dissolution of
the highly amorphous Fe-rich slag using the more conventional techniques
(i.e., dissolution test, XRD quantification). The protocols developed could
be relevant not only for CAC-based systems but also for other binder
systems (i.e., PC-based blends) where the components have distinct
differences in density and composition.
2.2.5. Spectroscopy
57
Fe Mössbauer spectroscopy
This technique was used to elucidate the atomic environment of Fe, being
58 ASTOVEZA 2022
____________________________________________________ Chapter 2
the main component of the slag, and its evolution during the hydration.
57
Fe Mössbauer spectroscopy (WissEl GmbH) was operated using a
57
CoRh source and a metallic iron foil for the velocity scale calibration.
Although most of the spectra were measured at room temperature, some
samples were also analyzed at 77K to extricate overlapping signatures of
different Fe species during the fitting. The evaluation of the spectra was
performed by least-square minimization fitting of lines using the IMSG
software.
The Mössbauer spectra obtained from the raw slag is presented in Figure
2.10. Different Fe components were also fitted in this figure, using the
parameters presented in Table 2.7. More details on the model can be found
in Appendix F.
Figure 2.10 Deconvoluted 57Fe Mössbauer spectra of the slag (granulated FS)
measured at room temperature
Table 2.7 57Fe Mössbauer hyperfine parameters of the fitting proposed for the slag
shown in Figure 2.10. The different parameters are abbreviated as follows: isomer shift
(IS); central value of the quadruple splitting (QS); central value of the hyperfine magnetic
field (MF); total Gaussian-
hyperfine
The typical errors for the IS and QS is ± 0.02 mm/s and ± 5% for the AA
IS QS MF AA
Component (mm/s) (mm/s) (kOe) (mm/s or kOe) (%)
G1 1.14 1.91 0.00 0.42 30
G2 0.79 1.73 0.00 0.51 40
G3 0.38 1.04 0.00 0.17 9
M1 0.91 0.67 0.00 0.20 18
M2 -0.08 -0.01 330.93 14.65 3
ASTOVEZA 2022 59
Chapter 2 ____________________________________________________
total 100
The fitting demonstrates the diversity of Fe speciation in the slag despite
its seemingly simple crystallography composed of an amorphous phase (92
wt%). Components very similar to those found in Fe-rich slags in Siakati
et al., 2020 [117] and Peys et al., 2019 [116] were identified. Based on the
combination of Mössbauer parameters, the compositions were assigned as
follows: two doublets of Fe2+ (G1 and G2) and a doublet of Fe3+ (G3) in
the glass fraction; a doublet of Fe2+ (M1) and a sextet magnetic component
(M2) both in the crystalline fraction. The components were selected based
on the initial characterization of the slag (section 2.2.1). The reducing
environment in the furnace during the slag generation evidently resulted to
the dominance of the Fe2+ components in the system. The wide spread of
the quadruple s
incorporated in the amorphous fraction. The presence of two Fe2+
contributions in a similar slag has been previously reported to be a result
of the high irregularity of the Fe ions in the glass network [117] and
characterized by the asymmetric velocity resonant intensities. The higher
IS and QS values of G1 indicate the higher coordination number of Fe in
this component.
Furthermore, a minor Fe3+ component (G3) in the glass fraction was also
likely to be present in octahedral coordination. Wüstite was assigned as the
component M1 with the fitted area fixed according to the measured amount
of wüstite using XRD-Rietveld in the preceding section. On the contrary,
addition of the minor magnetic component (M2) improved the fitting of
the experimental spectrum despite the fact that no ferromagnetic minerals
were detected by using the other characterization techniques. Nonetheless,
it must be noted that the value could fall well within the measurement error
± 5% for AA in the fitting.
In addition, Figure 2.11 and Figure 2.12 show the spectra obtained for the
raw cement samples. Despite the indication from XRD (Appendix B) that
Fe in both of these materials is part of the C4AF crystal, a clear difference
in the spectra is obtained. That of the pure CAC has a similar pattern to the
spectra published in Honeybourne, 1980 [146] for a high alumina cement.
The primary contribution was attributed to a non-stoichiometric C4AF, but
also the presence of minor amounts of wüstite and magnetite was noted.
On the other hand, the spectra of the raw PC resembled that previously
obtained in Rai et al., 2009 [147] associated to the contribution of C4AF in
anhydrous PC. The noisy signals were caused by the significantly low Fe
content of these materials (< 2 wt% Fe2O3, Table 2.4) relative to the raw
slag. It follows that the signature of Fe that will be detected in the hydrated
pastes following the blends in section 2.1.4, can be assumed to almost
exclusively originate from the slag rather than the cement components.
60 ASTOVEZA 2022
____________________________________________________ Chapter 2
Nevertheless obtaining these spectra, although with noisy signals,

remained essential in discerning any possible contribution of the cements
during the fitting.
Figure 2.11 Deconvoluted 57Fe Mössbauer spectra of the CAC (Secar 51®)
Figure 2.12 Deconvoluted 57Fe Mössbauer spectra of the PC

The evolution of the Mössbauer spectra of hydrated pastes is presented in

Chapters 3 and 4. In addition, the slag hydration degrees of formulation A
paste samples at different curing ages were also estimated from the
absorption areas of the unreacted slag components, using the simplified
equation below:
Equation 2.5
ASTOVEZA 2022 61
Chapter 2 ____________________________________________________
where: is degree of slag hydration; AA is the absorption area obtained

from the fitting of the Mössbauer spectra; i
is estimated. More details on this calculation are provided in Appendix H.
Attenuated Total Reflection Fourier Transform Infrared Spectroscopy

(ATR-FTIR)
The ATR-FTIR spectra were obtained using the Bruker Alpha-P
spectrometer with diamond crystal. Each measurement consisted of 32
scans with a resolution of 4 cm-1 and a range from 380 to 4000 cm-1. The
raw results were expressed in transmittance and were then converted to
absorbance for the analyses. The average of three runs was used to plot
each spectra. The primary objective of this technique in this study is to
attempt to demonstrate the reactivity of the slag through the Si-O peak shift
similar to what has been previously reported for inorganic polymers [117],
elaborated in Chapter 1. This characteristic was claimed [117] to be related
to the partial oxidation of the Fe2+ in the slag to Fe3+ as it participates in
the silicate network forming/modification during the polymerization. In
addition, the phase assemblage evolution of the hydrated pastes were also
followed using this technique. The results obtained were interpreted in
conjunction with those of the XRD and TGA.
The ATR-FTIR spectra of the main raw materials used in Chapters 3 and
4 are presented in this section. The assignment of the bands was based on
literature data [86, 137, 165, 171, 172, 200] on pure and synthetic systems,
better elaborated in sections 3.3.3 and 4.3.3.
For the slag (Figure 2.13), the Si-O-T (where T could be Al, Fe or Si)
asymmetric stretching vibrations at 900 cm-1 and Si-O-T asymmetric
bending vibrations at 470 cm-1 were detected.
Figure 2.13 ATR-FTIR of the raw (granulated FS) slag
A more complex spectrum (Figure 2.14) is obtained for the CAC

(Secar 51®) consisting of Si-O peaks (around 1030 cm-1 and 415 cm-1) and
several Al-O peaks (in the region between 521-786 cm-1).
62 ASTOVEZA 2022
____________________________________________________ Chapter 2
Figure 2.14 ATR-FTIR of Secar 51®
For the PC, two main peaks were identified positioned at 880 cm-1 and 520
cm-1 attributed to the multiple coinciding peaks of Si-O asymmetric
stretching vibration bands of C3S and C2S.
Figure 2.15 ATR-FTIR of the PC
Finally for the HH, distinct S-O peaks at: 1150 and 1115 cm-1 (mode v3),
1006 cm-1 (mode v1), 653 and 593 cm-1 (mode v4); and O-H peaks at 1620
cm-1, 3660 cm-1 and 3550 cm-1 (mode v2, v3 and v1 respectively) were
identified.
Figure 2.16 ATR-FTIR of the calcium sulfate hemihydrate (HH)
ASTOVEZA 2022 63
Chapter 2 ____________________________________________________
2.2.6. Microscopy
Optical Microscope
An initial observation of the hardened paste samples under an optical
microscope (Nikon Eclipse LV150) revealed the distinct features of the
slag particles. Under polarized light (Figure 2.17 a), the dark color of the
raw slag particles makes them readily distinguishable from the other
components present such that a simple image analysis using ImageJ can be
used to reconstruct binary images as demonstrated in Figure 2.17 b.
Following the concept and assumptions described in section 2.2.4, a
similar methodology for calculating the degree of slag hydration can be
performed based on the change in the total slag volume over time.
However, the main drawback in using images from the optical microscope
is the limited resolution. In order to address this issue, the use of SEM-
BSE micrographs was opted in this study. Nonetheless, these initial
observations using the optical microscope could be useful for a rapid
evaluation for future studies.
Figure 2.17 (a) Micrograph of a slag-containing hydrated mortar sample

observed under an optical microscope in polarized light. (b) Binary image resulting from
image processing using the Image J software.
Scanning Electron Microscope (SEM)

The ThermoFischer Quanta 250 FEG scanning electron microscope was
used to analyze the microstructures of the slag-containing hydrated blends
in this study. After stopping the hydration through solvent exchange,
fragments of hardened paste and mortar samples were impregnated under
vacuum with an epoxy resin (EPOXYPACK 30). This was followed by
progressive polishing from 9 to ¼ µm aided by a diamond spray (Struers
DP-Lubricant Brown, alcohol-based with polyethylene glycol). The scan
was performed at an accelerating voltage of 20 kV using the backscattered
electron (BSE) and the energy dispersive X-ray analysis (EDS) detectors.
64 ASTOVEZA 2022
____________________________________________________ Chapter 2
In Figure 2.18, images from optical microscope are presented side by side
with that of the SEM in BSE mode. Both techniques revealed the distinct
dendritic surface of the raw slag particles, a notable feature rendered by
the Fe-O dendrites in granulated slags [148]. This facilitated the
identification of the slag particles under the microscope in this study and
consequently aided the selection of image analysis operations for the
quantification of the degree of slag hydration.
Figure 2.18 Micrograph of a slag-containing hydrated mortar sample observed under an

optical microscope in (a) polarized light and (b) reflected light. (c) SEM-BSE image
focused on a raw slag particle.
Image analysis was employed on SEM-BSE micrographs of formulation A

hardened paste samples after 1 and 90 days of hydration. The operations
followed are described in detail in Appendix G. The degree of slag
hydration was evaluated following the same assumptions as those using
tomography (section 2.2.4), however calculating the total slag area
(Equation 2.6) instead of the slag volume for the latter.
Equation 2.6
where: is degree of hydration; A is the total area of the slag. In this case,
the degree of slag hydration is estimated after 90 days of hydration.
A comparison of the values of the degree of slag hydration obtained using
image analysis of SEM-BSE micrographs versus that obtained from the
volume analysis of tomography is presented in Chapter 3.
Transmission Electron Microscope (TEM)

The CM200 transmission electron microscope from FEI equipped with an
EDS Si (Li) EDAX system, operated at 200 kV, was used to characterize
the hydrate phase assemblage on the nano-scale. After stopping the
hydration, finely ground paste samples were dispersed in acetone. The
suspension was left in an ultrasonic bath for at least 1 min to aid the
dispersion. Few drops were extracted, placed in a copper grid and mounted
ASTOVEZA 2022 65
Chapter 2 ____________________________________________________
for analysis. Energy dispersive X-ray spectroscopy (EDS) was used to

determine the chemical analysis of the selected areas.
The technique was also used to observe the differences in morphologies
with a reduced intermixing of phases in the same analytical volume relative
to SEM. This interest was supported by the investigation of Fe uptake on
C-S-H in Siramanont et al., 2021 [133] where changes in morphology were
observed corresponding to different extents of Fe incorporation (Figure
2.19). Those with lower Fe uptake were characterized by typical foil-like
morphology for C-S-H, while the formation of Fe-rich spherical
nanoparticles distributed around the C-S-H foils was noted at higher
uptake [133].
Figure 2.19 TEM images of pure synthetic C-S-H (a); C-S-H with low, 0.010-0.014
mmol/g, Fe uptake; (b) and C-S-H with high, 0.039-0.048 mmol/g, Fe uptake
modified from Siramanont et al., 2021 [133].
Scanning Transmission Electron Microscope (STEM)

The JEM-ARM 200F Cold FEG scanning transmission electron
microscope equipped with a spherical aberration (Cs) probe corrector was
operated at 200 kV. EDS spectra and X-maps were recorded on a Jeol
spectrometer (SDD, Jeol DRY SD 30 GV). This technique was employed
to facilitate the search for the Fe-bearing hydrate phases, similar to the
method previously presented in Siramanont et al., 2021 [133]. A modified
version of Fe-maps from their study shown in Figure 2.20, demonstrated
the role of STEM in assessing whether Fe is incorporated in the structure
of the C-S-H (more uniformly distributed) or in another phase evolving
from C-S-H during hydration (presence of high concentration zones).
In Chapters 3 and 4, the percentages associated with the TEM and STEM
results pertain to the atomic composition of the marked areas estimated
from EDS analysis. It must be emphasized however that these percentages
should not be interpreted as the absolute values, considering the semi-
quantitative nature of the analysis. Rather, they are indicators of the
average proportions of elements interacting with the electron beam. Light
(low atomic mass) elements such as O, C and H are often erroneously
66 ASTOVEZA 2022
____________________________________________________ Chapter 2
estimated in EDS analysis. Considering that these elements are significant

components of the hydrate phases, the absolute composition of the areas
analyzed could not be accurately determined.
Figure 2.20 STEM-EDS Fe-maps of C-S-H foils possibly with Fe incorporated in the
structure (a); and C-S-H foils intermixed with Fe-rich spherical nanoparticles,
possibly ferrihydrites or hydrogarnet. The images were modified from
Siramanont et al., 2021 [133].
In addition, some phases, particularly the hydrated ones, could disintegrate

during or after the scans which may result in differences between the
theoretical and experimental composition of the identified phases. A
pertinent example of this is the analysis of S and O in ettringite. Ettringite
6+
forms a hexagonal prismatic crystal consisting of Ca6[Al2(OH)12 2O]
columns bonded through electrostatic interactions with {(SO4)3 2O}
in the inter-column channels, illustrated in Figure 2.21. With this
configuration, the sulfate tetrahedra could potentially disintegrate along
with the water molecules when bombarded with electron beams during the
EDS analysis. Moreover, the SO42- ions in ettringite are easily substituted
by OH-, CO32- [149] and even by other oxy-anions (i.e., CrO4 , SeO4 ,
SeO3 ) [150]. The presence of these several solid solutions, in addition to
the potential amorphization of ettringite with pressure or at longer ages
[151], could be attributed to lower S proportions as opposed to what is
theoretical expected for ettringite.
Nonetheless, expressing the values in terms of atomic percentages allows
calculation of molar/atomic ratios of the relatively heavier elements
especially Ca, Si, and Al. When presented in ratios, they become more
ASTOVEZA 2022 67
Chapter 2 ____________________________________________________
reliable references for phase identification in conjunction with the

characteristic morphology of some minerals. This is particularly the case
for ettringite, considering its tendency to maintain its distinctive
needle/rod-like microstructure even at wide ranges (> 65%) of solid
substitution of its sulfate ions [150].
Figure 2.21 Crystal structure of ettringite projected on to the (0001) plane.

Ca atoms are shown as blue circles, oxygen atoms in red, aluminum atoms in light blue,
sulfate tetrahedral in yellow and hydrogen atoms in grey [152,153].
2.2.7. Other techniques
Electric conductivity
Electrical conductivity measurements were made using the Mettler-Toledo
SevenMulti for 0.1 g/L and 0.5 g/L suspensions with no pH adjustment.
The temperature was controlled at 20°C using a 1 L thermostatic beaker
filled with 900 mL of deionized water (pH 7.1) where the solid is
introduced. Measurements were taken at 5 s interval and the suspension
was constantly agitated at 500 rpm. The beaker was sealed using
Parafilm® to limit atmospheric carbonation. A schematic diagram of the
set-up adapted from Maach, 2016 [154] is presented in Figure 2.22.
ICP-AES
5 mL solution was extracted after 30 min of conductivity test using a
syringe equipped with 0.2 µm polyethersulfone filter. The filtrate was
stabilized for ICP-AES analysis by adding 5 vol% nitric acid (67%,
NORMATOM®).
68 ASTOVEZA 2022
____________________________________________________ Chapter 2
Figure 2.22 Schematic diagram of the electric conductivity set-up

modified from Maach, 2018 [154]
2.3. Selection of industrial residue

In this section, the potential of the SOCRATES residues as SCM in a CAC-
HH binary blend is assessed. Electric conductivity measurement was
performed on the raw residues. Isothermal calorimetry, setting time and
compressive strength tests used the binder formulations described in Table
2.1. At the end of the section, a separate assessment involving the jarosite
residue is presented exploring its potential as a sulfate source therefore
substituting HH in the system. Following the formulations described in
Table 2.2, the effect of the substitution level to the compressive strength
evolution and to the microstructure of the blends was evaluated.
2.3.1. Electric conductivity measurements

Measurements from the SOCRATES residues are plotted in Figure 2.23
along with that of the reference cement binder. The passage of ions in the
solution was assumed to be related to the dissolution of the residues.
Consequently, this technique was considered as a possible measure of
SCM reactivity a concept that has been previously demonstrated [155].
In Figure 2.23 a, it is evident that the paper residue readily dissolves into
the solution which can be attributed to its high free lime content (Appendix
B, Figure B.6). With a conductivity more than twice as high as the cement
binder, it can be deduced that the amount of paper residue addition should
be limited in order to mitigate the risk of flash setting. On the contrary, the
conductivity of the other SOCRATES residues is significantly lower than
ASTOVEZA 2022 69
Chapter 2 ____________________________________________________
the reference binder. In Figure 2.23 b, a closer look on the trend shows
that although the overall degree of conductivity is closely comparable
among these other residues, the kinetics appear to have a particular pattern
for the highly amorphous slags. Their conductivities continue to increase
throughout the first 30 min unlike the other residues which appear to
stabilize even within the first 2 min of the test.
Figure 2.23 Conductivity measurements on raw materials using 0.5 g/L suspension
in deionized water at 20°C
Considering these initial results along with the mineralogical composition

of the residues (Appendix B), it was decided to evaluate the valorization
potential of the jarosite and paper residues separately. These residues are
sulfate- and lime-rich, respectively, making them distinctive compared to
the other SOCRATES residues.
2.3.2. Heat Release

The result of the isothermal calorimetry test on paste samples is presented
in Figure 2.24. Three main peaks in the heat flow curves (Figure 2.24 a)
similar to those previously identified [9] in low sulfate CAC binders were
observed. These peaks have been associated to: (a-b) the fast dissolution
of HH leading to rapid gypsum precipitation; (b-c) massive ettringite
precipitation resulting from CA and gypsum reaction; and (c-d)
monosulfoaluminate formation with the depletion of gypsum. Comparable
calorimetry results were identified independently of the type of residue
added and therefore during the first hours of reaction, their addition does
not appear to significantly alter the hydration of CAC-HH.
70 ASTOVEZA 2022
____________________________________________________ Chapter 2
Figure 2.24 (a) Heat flow and (b) cumulative heat release of paste samples for the
screening test
On the other hand, the cumulative heat release graph shown in Figure 2.24
b demonstrates the overall decrease in heat release with the addition of the
SOCRATES residues. Only the blend incorporating bottom ash has
achieved a comparable level of heat release as the reference formulation
after more than 90 hours of hydration. This could be related to the
relatively high specific surface area (Table 2.6) of the bottom ash, possibly
providing more nucleation sites for hydration. For the other residues, the
differences in heat values may not be sufficient enough to draw conclusive
comparisons.
2.3.3. Setting time and mechanical strength

The setting time and compressive strength evolution of standard mortars
are presented in Figure 2.25 and 2.26, respectively.
Figure 2.25 Initial and final setting time of mixtures for the screening test
The addition of the residues generally, but not substantially, delayed the
setting compared to the reference cement formulation. Only the blend
ASTOVEZA 2022 71
Chapter 2 ____________________________________________________
incorporating the bottom ash displayed a shorter final setting time.

Although the time reduction is minimal, this could once again be linked to
the fineness of the bottom ash supposedly affecting the early hydration
reactions as suggested in the preceding section. On the other hand, the
highest delay in setting was noted for the blend incorporating the slowly-
cooled FS. Since the fast setting property of CAC is among its valued
features, such delay might be an issue and therefore, necessitates
admixtures for set control.
The strength evolution was used as the main criterion for the screening test.
In the first 28 days of hydration, the strength values of the residue-
containing blends (Figure 2.25) were merely comparable to those obtained
from the reference with quartz filler. Despite the overall decrease in
strength compared to the reference cement formulation, the evolution of
the SAIcement values particularly for the amorphous slags (granulated FS
and fumed fayalite slag) could be considered as one of the possible
manifestations of slag hydration at long term. At this point however, it is
not clear whether the strength development is directly proportional to the
reactivity of the residues. Nonetheless, the trends observed demonstrated
the possible long term contribution of the residue addition to the physical
properties of blend.
Figure 2.26 (a) Compressive strength evolution and (b) strength activity index (SAI) of
standard mortars for the screening test
Considering the results from the raw material characterization along with
the properties of blends presented in this section, the highly amorphous
slags were determined to have the highest potential as SCM (at 30 wt%
addition) in the CAC-rich system. Consequently, Chapters 2 and 3 will be
focusing on the properties and hydration mechanism of blends
incorporating the granulated FS (slag). In addition, the challenge in
improving the early age performance of these blends by activating the slag
will also be addressed in the next chapters.
72 ASTOVEZA 2022
____________________________________________________ Chapter 2
2.4. Evaluating jarosite residue as a sulfate source

The high sulfate composition (>50 wt% SO3, Table 2.4) of the jarosite
residue draws the interest in its potential valorization as a sulfate source in
the CAC-HH binder. However, the mineralogy of this residue (Appendix
B, Figure B.5) is complex with several crystalline phases contributing to
the sulfates content. An initial assessment is presented in this section to
demonstrate the relevance of its mineralogy to the properties of the blend.
The compressive strength evolution of the blends (Table 2.2) with
different levels of jarosite residue substitution is shown in Figure 2.27.
The substitution clearly resulted in a decrease in strength of the standard
mortar samples. The drop was more evident in the first day of curing. This
became less significant at 7 and 28 days for 20 wt% and 50 wt%
substitution. Although there are many factors influencing the strength of a
blend, this observation could be interpreted as a possible indication that the
hydration of the residue could contribute to the strength development albeit
at a slower rate relative to the HH.
Figure 2.27 Compressive strength evolution of blends with jarosite residue addition
Based on the XRD patterns of the hydrated pastes (Figure 2.28) after 28
days of curing, the phase assemblage of the reference blend is dominated
by ettringite and gibbsite with small amount of monosulfoaluminate. As
the jarosite residue is added in higher quantities, the peaks associated to
ettringite diminish. However, the patterns generally remained consistent
until the HH was completely substituted by the residue (100JAR). At
100 wt% substitution, the peaks associated to the raw jarosite residue
(Appendix B, Figure B.5) particularly those of jarosite, gypsum and quartz
became more apparent. The height of the main ettringite peak detected at
and appeared to be slightly shifted
to 9.12°. As mentioned in Chapter 1, these shifts are often attributed to Al-
Fe solid substitution [131] which could be a possibility considering the
ASTOVEZA 2022 73
Chapter 2 ____________________________________________________
high Fe content (22 wt% Fe2O3) of the jarosite residue. Moreover, an

9.5-12°) was also detected in
the 100JAR blend. The absence of distinct peaks of the pure CAC
hydration products (i.e., C2AH8, CAH10) could signify the participation of
the jarosite residue in altering the hydration reactions of the CAC
component. These observations are consistent with the compressive
strength evolution (Figure 2.27) demonstrating the close resemblance in
the properties of the blends at 28 days for 20 wt% and 50 wt% substitution.
Figure 2.28 XRD patterns of hydrated pastes with different substitution level of
jarosite residue after 28 days of curing. The phases are abbreviated as follows:
(Ett) ettringite, (Ms) monosulfoaluminate, (Gib) gibbsite, (CA) monocalcium
aluminate, (Jar) jarosite, (Gyp) gypsum, (Sul) elemental sulfur, and (Qua) quartz.
Only the principal diffractogram peaks are labelled.
Additionally, a dissolution test was performed to compare the passage of

ions in the solution between HH and the jarosite residue. The methodology
for electric conductivity measurement followed by ICP-AES detailed in
section 2.2.7 was executed. The conductivity measurements (Figure 2.29)
demonstrated the faster dissolution of HH in the solution. Using the values
of sulfur (S) ion concentration after 30 min obtained from ICP analysis,
the percentage of dissolved S was calculated relative to the respective total
S content of the raw materials. Despite their comparable SO3 content
(Table 2.4), the difference in mineralogy proved to be an important factor
affecting the dissolution of ions. The sulfates from the jarosite residue may
74 ASTOVEZA 2022
____________________________________________________ Chapter 2
not be as readily available to participate in hydration reactions in the

solution compared to those of the HH.
Figure 2.29 Conductivity of the jarosite residue versus HH

using 0.1 g/L suspension in deionized water at 20°C
Comparing the SEM images of the paste samples at 0 wt% (a) and 100 wt%
(b) substitution levels (Figure 2.30), ettringite needles are observed to be
less abundant as HH is replaced by the jarosite residue.
Figure 2.30 SEM-SE images of (a) no jarosite and (b) 100JAR paste samples
after 3 days of curing. Encircled in red are needle-like structures generally associated to
ettringite.
It can be inferred that the lower amount of sulfates available in the solution
lead to decreased ettringite formation an observation consistent with the
phase assemblage evolution obtained through XRD. As previously
mentioned in Chapter 1, maximizing ettringite formation is often desirable
in sulfated CAC systems due to its strength-giving properties [9,156]
which could be associated, although not directly correlated, with the
obtained strength values of the blends containing the jarosite residue.
ASTOVEZA 2022 75
Chapter 2 ____________________________________________________
Other ions (Na, Al, Si, K, Mg, P and Fe) were also released from the
jarosite residue during the dissolution test quantified in Figure 2.31. One
of the biggest concerns in the valorization of the jarosite residue as SCM
is the possible release of toxic elements and heavy metals. Although these
components were not analyzed in ICP for this study, the dissolution of the
other ions could signal that the jarosite residue is indeed reactive in this
system and therefore the toxic components should be monitored more
closely. For instance, the sulfur content of the raw residue alone is already
alarming (20-30 wt% see Appendix B, Figure B.5). It is clear that this
residue is not the typical jarosite by-product previously reported to have
potential use as raw meal substitute to gypsum [157]; or as mineral additive
in non-structural concrete [158]. Going back to the source of this by-
product (section 2.2.1), the complexity of the composition of this residue
can be traced from combining the streams of different leaching processes
into one filtration unit. In fact, the sulfur content can be totally avoided by
dedicating separate filtration lines for each reactors. This solution, of
course, may not consider the other practical aspects of the production.
Figure 2.31 Comparison of the ion concentrations after the dissolution test
measured using ICP-AES
Despite the promising strength evolution at lower substitution rates, the

overall evaluation does not support the incorporation of the untreated
jarosite residue in the CAC-HH binder system primarily due to its high S
(elemental), Pb and Zn contents. Extensive analysis of the toxicity of the
residue and its leaching characteristics remains indispensable in evaluating
its valorization potential. Pre-treatment is recommended to better tailor the
properties of this residue to cementitious applications.
76 ASTOVEZA 2022
Chapter 2 ____________________________________________________
Summary
This chapter details the materials and methods used in this study. Residues
coming from a single production batch were used throughout this study to
maintain uniform results. They were homogenized, sampled, dried and
grinded to obtain a PSD as close as possible to that measured for the CAC.
The characterization of the residues revealed a diverse mineralogy related
to their origin. Their main properties are highlighted below:
Granulated FS
An iron-rich highly (92 wt%) amorphous slag originating from lead-zinc
production. Mössbauer spectroscopy revealed that the Fe speciation in slag
is dominated by Fe2+. The Si-O bonds in this material were detected using
ATR-FTIR. When added in CAC-rich blended cement, the dark color, high
density and Fe content of the unreacted slag particles could facilitate its
differentiation from the rest of the cement matrix.
Slowly-cooled FS
An iron-rich slag with the same origin as the granulated FS but more
crystalline in nature (only 13 wt% amorphous) due to the slower cooling
rate employed. Its composition is dominated by fayalite and spinel.
Fumed fayalite slag
An iron-rich highly (94 wt%) amorphous slag originating from secondary
copper production. Although the basic properties are closely comparable
to those of the granulated FS, this batch of material still contains 2.3 wt%
of ZnO.
Bottom ash
A siliceous by-product of a municipal waste incinerator with a highly
heterogenous composition mainly consisting of quartz, calcite, muscovite
and wollastonite. Fragments of unburnt components (i.e., plastics, glass)
were observed in the sample received. Its SSA was much higher than those
of the slags in this study despite displaying a similar PSD.
Jarosite residue
A sulfate-rich (51 wt% SO3) by-product of zinc leaching. Its mineralogy is
dominated by jarosite, gypsum and sulfur. Its substantial Fe content
(22 wt% Fe2O3) is linked to the jarosite component. Significant traces of
Pb and Zn were also detected. Similar to the bottom ash, high SSA value
was also obtained.
Paper residue
A lime-rich (68 wt% CaO) by-product of paper recycling. No further
grinding was executed due to its very fine nature as-received. The electric
78 ASTOVEZA 2022
____________________________________________________ Chapter 2
conductivity measurements performed demonstrated its rapid and

extensive dissolution in water.
Based on these properties, the Fe-rich slags and the bottom ash were added
at 30 wt% to a binary blend of CAC-HH with the goal of evaluating their
potential as SCM. A screening test consisting of electric conductivity
measurements, isothermal calorimetry, setting time and compressive
strength evaluation was performed. Among the residues investigated, the
highly amorphous slags displayed the best potential mainly supported by
the strength evolution of standard mortars. These slags displayed long term
(>28 days) strength contribution with SAIcement values reaching up to as
much as 0.9. Considering the results of the screening test along with the
raw materials characterization, the granulated FS was selected as the main
SCM for further investigation in Chapters 3 and 4.
Moreover, this chapter introduced various techniques for investigating the
physical properties (i.e., mechanical strength, dimensional stability, setting
time); phase assemblage evolution (XRD, TGA); the dissolution rate of
slag (XCT, SEM); and the fate of Fe during hydration (Mössbauer
spectroscopy, TEM, STEM). At the end of this chapter, the jarosite residue
was evaluated as a sulfate source substituting HH in the CAC-rich blend.
The strength evolution, phase assemblage, microstructure and dissolution
of ions were examined. Despite the possibility of mitigating the observed
strength reduction by controlling the substitution level, the overall findings
do not support the incorporation of the untreated residue without prior
toxicity evaluation. Pre-treatment will also be necessary to convert the
sulfur component of the residue and to tailor its properties in order to
improve its potential as SCM.
ASTOVEZA 2022 79
____________________________________________________ Chapter 3
Chapter 3
Fe-rich slag addition in binary CAC-
HH binder
Contents
Introduction . 83
3.1 Physical properties .. 84

3.1.1. Low calcium sulfate system . 84
3.1.2. High calcium sulfate system .. 87
3.2 Quantifying the degree of slag hydration . 89
3.2.1. SEM-BSE . 89
3.2.2. Tomography . 92
3.3 Phase assemblage evolution over time .. 96
3.3.1. Thermodynamic modelling 96
3.3.2. Early hydration reactions . 99
3.3.3. Long term hydration . 100
3.4 The fate of Fe during hydration . 108
3.4.1. TEM . 108
3.4.2. STEM 112
3.4.3. Mössbauer spectroscopy 120
3.5 Activation through paper residue addition 129
Conclusion 135
ASTOVEZA 2022 81
____________________________________________________ Chapter 3
Introduction
In this chapter, the addition of the highly amorphous Fe-rich non-ferrous
metallurgy (NFM) slag in calcium aluminate cement (CAC)-calcium
sulfate hemihydrate (HH) binder systems was investigated. In Chapter 2,
this slag was identified to have the best potential as a supplementary
cementitious material (SCM). The slag displayed contribution to the long
term (>28 days) strength development of the standard mortars. In addition,
its leachates (CuO and ZnO) were determined to be both below the
detection limit, in contrary to the 2.3 wt% ZnO measured in a similar
granulated slag among the selection.
The properties of formulations, A, B, and A-TC (Chapter 2, Table 2.3)
were examined in this chapter. All of these formulations contained
30 wt% of slag. Formulations A and B have CAC/HH ratio of 4.5 and
1.6, respectively. On the other hand, formulation A-TC has CAC/HH ratio
of 4.5 similar to that of the formulation A, but with 5 wt% addition of the
paper residue (TC). This lime-rich (68 wt% CaO) material is highly soluble
in water and is already being commercially marketed as a mineral additive
for cement and concrete. Considering these properties, it has been
evaluated separately as a possible activator in formulation A, aiming to
induce slag dissolution and contribute Ca2+ ions during the early age
(<28 days).
Several techniques were employed to characterize the properties of the
blends for up to 1 year. Mechanical strength and dimensional stability were
monitored using mortar samples. The degree of slag dissolution was
estimated using SEM-BSE, X-ray Computed Tomography (XCT) and
Mössbauer spectroscopy. The phase assemblage of the binary systems
were simulated using thermodynamic modelling in function of the CAC
proportion. Experimentally, several techniques, including isothermal
calorimetry, XRD, TGA, and ATR-FTIR were used to follow the evolution
of the phase assemblage using paste samples. Finally, the fate of the Fe
from the slag was described based on the results of Mössbauer
spectroscopy, TEM and scanning transmission electron microscope
(STEM).
The methodology for the characterization techniques was specifically
adapted to the distinct properties of the slag primarily related to its high Fe
content (i.e., high density, high atomic mass, Mössbauer effect). For this
reason, the CAC used was changed from Ciment Fondu® in the screening
test to Secar 51®. The latter contains less Fe2O3 (< 2 wt%) and was
therefore expected to interfere less with the signals of the Fe from the slag.
ASTOVEZA 2022 83
Chapter 3 ____________________________________________________
3.1. Physical properties

The physical properties of two blends representing a low (formulation A)
and a high (formulation B) calcium sulfate system are presented in this
section.
3.1.1. Low calcium sulfate system

The initial and final setting times of formulation (A) 4.5CAC-HH mortar
blend and its references are presented in Figure 3.1. It can be seen that the
addition of slag resulted in a minimal delay in the setting compared to the
reference cement formulation. Changes in setting time with the addition of
SCM have been previously [112,159] linked to the following factors:
Reactivity of the SCM during the first hours of hydration (typically
in the presence of high lime and sulfate content)
Lower effective w/b ratio
Absorption/adsorption of admixtures by SCM with high SSA
Filler effect mechanism
[9] property is
often attributed to the fast hydration of the CAC component. With the
addition of the slag, the reduced amount of the CAC content can be another
reason for the delayed setting along with the factors listed above.
Figure 3.1 Initial and final setting time of formulation A and its references
On the other hand, the reference filler formulation displayed a more
significant setting retardation despite following an equal volume
substitution as the slag. Considering that the quartz filler and the slag have
a closely comparable PSD and SSA (Chapter 2, Table 2.6), this difference
the early hydration was not limited to that of an inert filler.
84 ASTOVEZA 2022
____________________________________________________ Chapter 3
The compressive (Figure 3.2) and flexural (Figure 3.3) strength evolution
of formulation A and its references were followed using standard mortars
over 1 year of curing under water. The evolution of the SAIcement values are
presented in the secondary axes (right).
Figure 3.2 Compressive strength of formulation A (primary axis) and its SAI evolution
(secondary axis)
Figure 3.3 Flexural strength of formulation A (primary axis) and its SAI evolution
(secondary axis)
Similar to the observed trend during the screening test in Chapter 2, the
addition of the slag also clearly resulted in the overall decrease in strength
compared to the reference cement formulation. The same effect has been
previously reported in PC systems during the early hydration stage (<28
days) of a similar NFM slag [160] and as well as other SCM [8]. Although
this trend was less evident in terms of the flexural strength, the decrease in
the compressive strength reached as much as 31% in the first day of curing
with SAI value 0.64. It was not until after more than 28 days that the
compressive strengths of formulation A and its reference became
comparable at 39 MPa versus 41 MPa, respectively. This was exhibited by
the abrupt increase in the SAI values from 28 to 90 days. This could be
ASTOVEZA 2022 85
Chapter 3 ____________________________________________________
linked to slower dissolution of the slag compared to the other binder

components (CAC and HH), demonstrated already during the screening
test in Chapter 2 (Chapter 2, Figure 2.23). In general, slower dissolution
rate influencing the strength evolution was reported to be prominent even
among traditional SCM [12].
Nevertheless, following the SAI trends for the compressive strength
(Figure 3.2
long term strength of the blend in a degree different from that of the quartz
filler. Beyond 28 days, the strength values of the slag-containing
formulation were comparable to the reference cement. On the other hand,
the SAI values of the reference filler mortar remained consistent.
Another property that could support the potential of the slag as SCM in
this system was the dimensional stability. Figure 3.4 compares the length
variation of formulation A to its reference. The measurements from the
first 24 hours were obtained using the Walter+Bai shrinkage test in air-
curing condition. Further measurements using the same samples were
taken using an extensometer after demoulding and curing under water.
Figure 3.4 Length variation of formulation A and its reference obtained from Walter+Bai
shrinkage test for the first 24 hours of hydration. Measurements were continued using an
extensometer for the demoulded samples cured under water.
It can be seen that the addition of the slag did not significantly alter the
dimensional stability of the blend. An overall shrinkage was observed
during the first 24 h for both formulations. The mechanism of ettringite
formation was hypothesized to slightly alter trend, such
that shrinkage was counteracted in the reference cement formulation by the
fast formation of interlocking ettringite from 6 to 18 h. On the contrary,
the slag-containing blend displayed a more abrupt shrinkage at this period,
which could be an indication of less amount of expansive hydrates formed.
86 ASTOVEZA 2022
____________________________________________________ Chapter 3
Beyond 24 hours, curing under water triggered a slight expansion for both
formulations. Not only was additional water provided to stimulate
hydration in this step, but also drying shrinkage was avoided throughout
the remaining duration of curing.
The overall dimensional stability of formulation A was reflected through
the progressive strength development in mortar samples. The addition of
the slag even appeared to have positively contributed to the stability
keeping the values closer to zero in Figure 3.4. In the end, it should be
considered that despite this observation, the scale of the length
measurements prese
within the confidence interval of the test and consequently may not provide
conclusive quantitative comparison.
3.1.2. High calcium sulfate system

The physical properties of formulation (B) 1.6CAC-HH with and without
slag, are discussed separately since 0.3 wt% of citric acid was added in this
dry mix to retard the setting. Recall from Chapter 2 that for some of the
formulations in this study, setting was delayed in order to provide
sufficient time for placing the freshly-mixed blends in moulds. In this case,
the initial and final setting time recorded for formulation B was 38 and 46
min, respectively. As mentioned earlier, however, the influence of such a
low amount of citric acid was expected to be limited within the first few
hours of hydration and should not significantly alter the long term chemical
properties of the blend.
The compressive and flexural strength evolution of formulation B and its
references are presented in the primary axes of Figure 3.5 and Figure 3.6,
respectively. The evolution of SAIcement values are shown in the secondary
axes.
ASTOVEZA 2022 87
Chapter 3 ____________________________________________________
Figure 3.5 Compressive strength of formulation B (primary axis) and its

SAI evolution (secondary axis)
Figure 3.6 Flexural strength of formulation B (primary axis) and its

SAI evolution (secondary axis)
The strength values in this high calcium sulfate system were considerably
higher than those obtained from the low calcium sulfate formulation, B.
However, the SAI values were generally lower implying that the observed
increase in strength was more likely associated to the hydration of CAC
and HH, producing more ettringite, rather than an increased rate or extent
of slag hydration. An indication of this was the observed expansion of
mortars in Figure 3.7. It will also be confirmed from the techniques in the
following section.
Figure 3.7 Length variation of formulation B and its references

measured for standard mortars cured under water
88 ASTOVEZA 2022
____________________________________________________ Chapter 3
While an overall shrinkage was observed in the low-sulfate formulation

(A), it was the opposite case in the high-sulfate system (B) reaching as
much as 2.5 mm/m expansion for the reference cement mortar. It is clear
that the space-filling property [161] of the low density hydrate, ettringite,
has directly contributed to the improved early strengths in formulation B.
Furthermore, the addition of the slag also appeared to stabilize the blend
resulting in lower expansion compared to the reference cement
formulation. Although the differences in the expansion degrees between
the slag-containing and its reference cement mortar were more significant
(higher than the confidence interval) this time, it could be argued that this
was mainly due to the filler effect, looking at the length variations of the
reference filler mortars in Figure 3.7.
in formulation B was more significant between the 28 and the 90 days

where the slope of the SAI is the steepest. These results also suggest that
the CAC/HH ratio used did not have a significant influence on the
reactivity of the slag. This could be confirmed by the estimations of the
degree of slag dissolution in the next section.
3.2. Quantifying the degree of slag hydration

In this section, two techniques were used to estimate the degree of slag
hydration, namely: (1) image analysis of SEM-BSE micrographs; and (2)
volume analysis of XCT reconstructed 3D renders. As previously stated in
Chapter 2, these methodologies were based on the assumption that slag
dissolution was equivalent to reactivity. This means that the total amount
of dissolved slag has directly participated in the hydration reactions. In
addition, it was also assumed that the slag was unreactive in the first 24
hours due to the absence of data during this period.
3.2.1. SEM-BSE
The SEM was initially used in this study to characterize the microstructure
of the blends. Figure 3.8 (a) shows an SEM-BSE micrograph of
formulation A mortar sample after 90 days of hydration. The slag particles
appear the brightest due to their high Fe content (high atomic number), and
their boundaries remained intact with no apparent formation of hydration
rim around the surface of the particles. On the contrary, the hydration of
CAC particles was more evident with their dissolved surfaces gradually
blending into the matrix. The lower reactivity of the slag versus the other
binder components (CAC and HH) was further supported by the EDS maps
(Figure 3.8 b-f) revealing the massive dispersion of Ca, Al and S
ASTOVEZA 2022 89
Chapter 3 ____________________________________________________
throughout the interstices, while Fe and Si are localized in the slag and
sand particles, respectively.
Figure 3.8 (a) SEM-BSE micrograph of formulation A mortar after 90 days of hydration
along with (b-f) the corresponding EDS maps for Fe, Ca, Al, S and Si
Looking at an SEM micrograph of a reference cement mortar (no slag) in

Figure 3.9, localized hydrate-rich areas are observed where CAC particles
are almost completely dissolved into the matrix. This was not the case for
the corresponding slag-containing mortar sample even after 90 days of
curing. In addition, the presence of the coarse sand particles greatly diluted
the binder in the images leading to the decision to use paste samples instead
for the image analysis for slag hydration degree quantification.
Sample image analysis of slag-containing paste micrographs are shown in
Figure 3.10. These were among the micrographs taken from polished
sections of formulation A where hydration was stopped after 1 day (a) and
90 days (b) of curing.
90 ASTOVEZA 2022
____________________________________________________ Chapter 3
Figure 3.9 (a) SEM-BSE micrograph of formulation A reference cement mortar

after 90 days of hydration; and (b) a focused image of the area marked in red showing a
localized hydrate-rich area
Figure 3.10 SEM-BSE micrograph of formulation A pastes after (a) 1 day and (b) 90 days
and their corresponding binary images obtained using Image J software
Based on the image analysis, the total slag area and degree of hydration
after 90 days of curing were estimated. The results are presented in Table
3.1. Despite the abrupt increase in SAI values (Figure 3.2) after 90 days
of hydration, it turns out that only about 14 % of the slag has reacted at this
point. Otherwise, this could also signify that the properties of paste and
mortar samples at any given curing time were not directly comparable,
particularly considering the difference in their interfacial transition zones
[9] resulting from the w/b ratios used.
ASTOVEZA 2022 91
Chapter 3 ____________________________________________________
Table 3.1 Estimated slag area and hydration degree using SEM-BSE + image analysis
curing time total slag area degree of slag hydration
(days) (µm2) (%)
1 5.63 0
90 4.83 14.2
Although the huge compositional difference between the slag and the rest
of the components of the hydrated paste greatly favored this methodology,
certain assumptions and challenges encountered must be carefully
considered. For instance, the highly contrasting abrasion resistance of the
blend components (i.e., slag versus CAC) resulted in uneven surfaces
during polishing. The hydrate products and the unreacted CAC were more
easily scratched compared to the unreacted slag particles leaving the latter
embossed on the surface. This was manifested in the micrographs as
shadows around the slag particles which could significantly influence the
image analysis. Another important issue was representativeness. The need
to prepare polished sections required the use of separate samples to be
analyzed per curing time. This means that we were not actually following
the dissolution of the same slag particles over time, rather we were simply
assuming that the total areas detected were representative of each curing
time. Considering that we were working on the micro-scale, the randomly
selected micrographs might not accurately represent the overall slag
distribution. In fact, areas where slag particles appeared to be agglomerated
and areas with particularly low amount of slag were observed as shown in
Figure 3.10 (b) 3 and 4, respectively.
In order to address the above-mentioned challenges, a non-destructive
technique using a combination of XCT and volume analysis was discussed
in the next section. Although the methodology revolved around the similar
principle (following the changes in the slag area/ volume over time), the
need for preparing polished sections and using different samples per curing
time was eliminated.
3.2.2. Tomography
The degree of slag hydration of formulations A and B were followed after
1, 28, 90, 180 and 365 days using a single hardened paste sample per
formulation. The threshold segmentation to select the slag particles was
optimized, guided by the observations from the SEM (EDS point ID). The
major steps involved were illustrated in Figure 3.11. In this figure, a red
arrow is drawn in the 2D reconstructed images (a). The plot profiles (b)
corresponding to the pixels under the arrows demonstrate that the selection
of the lower threshold value for the slag particles was greatly facilitated by
the distinct difference in the range of grayscale index between the slag and
the rest of the components. The binary images (c) were obtained following
92 ASTOVEZA 2022
____________________________________________________ Chapter 3
the operations described in Appendix E. Using 1400 binary images per

sample, the 3D renders (d) of the undissolved slag particles (shown in
white) were built.
Figure 3.11 Illustration of the steps involved in the volume analysis of XCT scans
There were two important details to consider in analyzing the XCT results:
1. Variations in the scanned VOI was inevitable despite using the same
paste samples over time mainly due to the dimensional changes
during hydration. For instance, the particles in the edges could move
in (shrinkage) or out (expansion) of the VOI.
2. The voxel resolution of the scans was set to 2.15 µm in order to
cover a sufficiently representative volume while optimizing the scan
time. This means however that the slag particles with a diameter
ASTOVEZA 2022 93
Chapter 3 ____________________________________________________
<2.15 µm could not be accurately detected unless it occupies i.e.,

70-90 vol% of the voxel, variable according to the threshold
segmentation limits imposed during volume reconstruction. This
could decrease the accuracy of slag hydration degree estimation,
considering that about 20 wt% of the slag particles fall below this
size according to the PSD in Appendix C. The direct consequence
of increasing the resolution or reducing the voxel size is either an
increased scan time, or a smaller sample size. The former is not ideal
due to the on-going hydration, while the latter might yield less
representative estimation.
Figure 3.12 shows the estimated slag hydration degree (primary axis;
vertical left) and their respective total slag volumes (secondary axis;
vertical right) obtained from the XCT-volume analysis. The estimated slag
volume varying between 6.0-6.5 % for the two formulations at 1 day seems
to be reasonable compared to the 8.4 vol% estimated (Appendix H) using
the density of the components without accounting for the porosity. Based
on these results, the rate and extent of slag reactivity in the low-sulfate
system (formulation A) appeared to be 6% higher than in the high-sulfate
system (formulation B). Indeed, the higher strength values obtained for
formulation B mortars were not proportional to the slag reactivity, rather
more related to the hydration of the CAC in a high-sulfate system.
Moreover, the convergin
28 days for the two formulations could support the earlier observation
curing (<90 days) was not significant enough to contribute to the strength
evolution, regardless of the CAC/HH ratio.
Figure 3.12 Estimated slag volume and hydration degree using XCT + volume analysis
Considering that only 40-50 % of the slag has reacted even after 1 year,
the interest in exploring the reactivity of slag in modified systems was later
94 ASTOVEZA 2022
____________________________________________________ Chapter 3
addressed in this study with the addition of the lime-rich paper residue
(Chapter 3, section 3.5) and PC (Chapter 4).
Despite the concern on the risk of underestimation due to the voxel
resolution, the estimated slag hydration degree (27 %) still ended up higher
than that (14 %) from the image analysis using SEM micrographs.
Although the two techniques were not directly comparable (i.e., volume-
versus area-based calculation; number of images processed), these findings
still open questions on the sensitivity of these image-based calculations.
Sensitivity Analysis
The influence of the grayscale threshold segmentation and the use of
different samples (hydration stopped) per curing time were evaluated to
test the sensitivity of the XCT methodology employed. These were among
perception in this methodology.

For the first factor, a ±5% deviation was introduced to the lower grayscale
threshold assigned to the slag based on the plot profiles (i.e., Figure 3.11
b) of formulation A. This aimed to quantify the influence of the selected
threshold to the estimated total volume of the slag. In effect, it was highly
of the slag particles) during the scans compared to the rest of the
components (image contrast).
For the second factor, samples of formulation A hardened pastes were
scanned after stopping the hydration at 1 and 90 days (one sample per
curing time). The VOI used for volume analyses was also changed to
cylinder (Figure 3.13) as opposed to the previously-presented cubic
samples. Smaller volumes (9 mm3 for the cubic VOI and 6.28 mm3 for the
cylindrical VOI) were scanned and reconstructed.
Figure 3.13 Illustration of XCT analysis using cylindrical volumes

The results obtained are presented in Figure 3.14. The ±5 % deviation in
the lower threshold values only equated to a difference of ±0.2 % in
volume and consequently ±3 % in slag hydration degree. This low
sensitivity to threshold segmentation was consistent with the visual
ASTOVEZA 2022 95
Chapter 3 ____________________________________________________
observation that the slag particles were easily distinguishable from the rest
of the components both in the SEM and XCT images.
Figure 3.14 Estimated (a) slag total volume and (b) slag hydration degree
from the sensitivity analysis for formulation A
On the other hand, the use of different samples per curing time and a
smaller VOI resulted in -0.43% volume or -5% slag hydration degree
difference. This highlights the advantage of using XCT over SEM, as a
rapid non-destructive test that could autonomously take a large sample size
while minimizing factors that could impose human bias.
In the end, selecting the more suitable methodology is a matter of
balancing several factors to satisfy the goal of the study. The focus must
be directed towards the rate rather than the absolute values of the slag
hydration to avoid misinterpretations. In any case, quantification must be
supported by complementary techniques (i.e., selective dissolution)
especially if determining the absolute values is crucial.
3.3. Phase assemblage evolution over time

After estimating the degree of slag dissolution, the phase assemblages of
the systems were simulated in function of CAC proportion using
thermodynamic modelling. The evolution of the phases was also
characterized experimentally for formulations A and B using several
techniques from the early hours of hydration up to a period of 1 year.
3.3.1. Thermodynamic modelling

The thermodynamic models of the binary CAC-HH systems without and
with the 30 wt% slag incorporation are shown in Figure 3.15 and Figure
3.16, respectively. All simulations were based on 20° C, 1 bar pressure
96 ASTOVEZA 2022
____________________________________________________ Chapter 3
condition and were calculated at infinite curing time (until stability). Other
assumptions along with the codes used to generate these models are
detailed in Appendix I. The models were calculated in function of the CAC
content (x-axis), which could be interpreted in relation to the CAC/HH
ratios (1.6 and 4.5) chosen in this study.
In Figure 3.15, it can be seen that gypsum is formed with the excessive
HH at lower (< 53%) CAC percentages. On the opposite extremity,
hydrogarnet is formed from the hydration of the remaining pure CAC when
sulfates are depleted. As the model does not consider the kinetics of
hydration, the formation of metastable calcium aluminate hydrates
mentioned in Chapter 1 is experimentally expected, prior to their ultimate
conversion to hydrogarnet at longer curing ages. Moreover, the phase
assemblage of reference formulations A and B are shown to be dominated
by monosulfoaluminate and ettringite, respectively. This model better
illustrates the interest in the CAC/HH ratios selected in this study. These
formulations both avoid regions of excessive sulfates marked by gypsum
formation, which could have detrimental effects (i.e., cracking, spalling)
related to secondary ettringite formation [162].
In both reference formulations, the silicates from the CAC are expected to
be incorporated in strätlingite, while the Fe from C4AF partakes in
siliceous hydrogarnet formation. It can also be seen that formulation B
dwells within the region of maximum hydrate formation by mass, such that
the 65 g of water added to the 100 g of anhydrous blend is barely sufficient
for full hydration.
Changes in the phase assemblage are evident with the addition of 30 wt%
slag in the system (Figure 3.16). As indicated in Appendix I, only 50 vol%
of the amorphous fraction of the slag was assumed to react during the
modelling, based on the XCT estimations after 1 year of hydration. The
black region, along with the minor (<1 at%) components of CAC and HH.
Fe appears to be incorporated in ferrihydrite at lower CAC percentages.
The dominant Fe-bearing phase at higher percentages is still Fe-siliceous
hydrogarnet represented in the GEMS in several phases as follows:
C3FS0.84H4.32
C3FS1.34H3.32
C3(A,F)S0.84H (solid-solution between Al and Fe)
ASTOVEZA 2022 97
Chapter 3 ____________________________________________________
Figure 3.15 Phase assemblage of the reference (no slag) binary binder calculated from
GEMS thermodynamic modelling software. CAC/HH content is varied in the x-axis
Figure 3.16 Phase assemblage of the binary binder incorporating 30% slag with varying
CAC and HH proportions, calculated from GEMS thermodynamic modelling software
98 ASTOVEZA 2022
____________________________________________________ Chapter 3
Unlike the model of the reference formulation, the slag-containing

formulation A is dominated by Fe-siliceous hydrogarnet and ettringite
instead of monosulfoaluminate. Even prior to experimentation, some
deviation of the model from the actual result was already anticipated given
the slower kinetics of slag dissolution. This brings forward the importance
of kinetics as a factor in validating and later on improving thermodynamic
models of cement hydration incorporating SCM.
3.3.2. Early hydration reactions

The first hours of hydration were characterized using ex-situ isothermal
calorimetry at 20°C. Note that for both formulations, no citric acid was
added to the pastes prepared for this test. The heat flow and cumulative
heat release curves are shown in Figure 3.17 and Figure 3.18,
respectively.
Figure 3.17 Heat flow measured ex-situ using isothermal calorimetry

for (a) formulation A and (b) formulation B pastes
The three main peaks in the heat flow curves (Figure 3.17 a) reported for
low-sulfate CAC-based systems [9] and similar to those presented in the
screening test (section 2.3.2) were observed for formulation A. On the
other hand, only two main hydration peaks (Figure 3.18 b) were found for
the high-sulfate system represented by formulation B. In this case, the third
peak associated to monosulfoaluminate formation was not detected. This
signifies that the abundance of sulfates in formulation B favors the
formation of ettringite and gibbsite over that of the monosulfoaluminates
in the early age of hydration.
In both formulations, the addition of the slag reduced the peak intensities
relative to the reference formulations but did not seem to alter the early
ASTOVEZA 2022 99
Chapter 3 ____________________________________________________
hydration reactions. In terms of reactivity, the trends are consistent with

those of the physical properties of the mortars discussed in section 3.1.
Recall that the slag appeared to be unreactive and even caused an overall
reduction in strength of mortars during the first 28 days of curing. In
addition, the more extensive ettringite formation depicted in the isothermal
calorimetry curves (a-b region of formulation B versus b-c region of
formulation A, Figure 3.17 ) could be linked to the overall expansion
observed in formulation B not observed in formulation A. Nonetheless, the
high heat release in the first 2 hours due to the extensive ettringite
formation did not result to any significant difference in the cumulative heat
release curves (Figure 3.18) at longer time.
Figure 3.18 Cumulative heat release measured using isothermal calorimetry
3.3.3. Long term hydration

Based on the XRD patterns in Figure 3.19, the hydrate phase assemblage
of formulation A mainly consisted of monosulfoaluminate, ettringite and
strätlingite. Although ettringite appeared to dominate the system during the
first 7 days, the lower amount of the remaining sulfates has likely resulted
in the conversion of ettringite to monosulfoaluminate in the later ages,
exhibited through the diminishing peaks of ettringite. However, clearly
detecting the diffraction peaks of monosulfoaluminate proved to be a
challenge due to its poorly-ordered crystal structure and the possible
variability in its composition [163]. This case was also true for gibbsite
which could be partly amorphous in these systems, highlighting the
significance of complementary techniques such as TGA. In interpreting the
XRD patterns, the peak intensities must be interpreted on a relative rather
than a quantitative basis considering the presence of several amorphous
phases in the system.
100 ASTOVEZA 2022

____________________________________________________ Chapter 3
Figure 3.19 XRD patterns of formulation A (solid black lines) and its reference
measured over 1 year of curing. Phases are abbreviated as follows: (Str) strätlingite,
(Ett) ettringite, (Ms) monosulfoaluminate, (Geh) gehlenite, (Hdg) hydrogarnet and (Wue)
wüstite
Unlike in the slag-containing formulation, hydrogarnet was detected after
1 year of curing in the reference formulation. It has been demonstrated
[163,164] that the formation of hydrogarnet can destabilize several AFm
phases in low-sulfate systems below 50°C. Based on thermodynamic
modelling [164], AFm only becomes stable when hydrogarnet is
suppressed under the assumption that its formation is not
thermodynamically-favored in the early age of hydration. This instability
could be one of the reasons behind the increasing peak intensities from 90
days to 1 year giving the impression of ettringite re-formation observed in
the reference formulation.
Strätlingite peaks were consistently present from 7 days to 1 year in the
slag-containing formulation which could have contributed to its strength
development, supporting previous studies [165,166] where the strength-
giving property of strätlingite has been demonstrated. In general, the
addition of siliceous SCM inducing strätlingite formation has been
associated to denser pore structures [165,167]. At the same time, the
preferential formation of hydrates (i.e., ettringite, strätlingite) other than
ASTOVEZA 2022 101

Chapter 3 ____________________________________________________
the metastable calcium aluminate phases (i.e., CAH10, C2AH8) with the
addition of SCM and calcium sulfates inhibits the conversion phenomenon
[168] typically considered undesirable in pure CAC systems. The acidic
nature of the Fe-rich slag could also slow down the conversion reactions
[169]. On the contrary, the appearance of strätlingite peaks in the reference
formulation occurred later between 7 and 28 days, diminishing again with
the formation of hydrogarnet and ettringite from 180 days.
Moreover, gehlenite appeared to be consumed faster in the reference
formulation with peaks greatly reduced beyond 7 days. This was not the
case for the slag-containing formulation likely due to the additional
silicates contributed by the slag in the system. On the other hand, wüstite
from the slag seemed to be unreactive, displaying consistent peak
intensities throughout the hydration period. This agrees with the findings
from a previous study [117] suggesting the preferential dissolution of the
amorphous component of similar slags over that of its crystalline
component.
Contrary to the more complex phase assemblage evolution of formulation
A and its reference, less changes were observed between the XRD patterns
of formulation B and its reference shown in Figure 3.20. The high-sulfate
system was indeed dominated by ettringite, stable throughout the hydration
period. This result agrees with the initial findings based on the strength
development and heat release. For both formulations, however, no new Fe-
containing hydrate was detected based on the XRD patterns alone. Despite
previous studies [131,170] suggesting the solid substitution of Fe to Al in
hydrates such as ettringite, monosulfoaluminate or siliceous hydrogarnet,
there was no clear peak shifts in the diffractograms presented that could
established at this point based on the XCT results, this could be an initial
indication that the Fe could be incorporated instead in an amorphous or in
a nano-crystalline hydrated phase. In order to unravel the fate of the Fe in
this system, further complementary techniques were employed.
102 ASTOVEZA 2022

____________________________________________________ Chapter 3
Figure 3.20 XRD patterns of formulation B (solid blue lines) and its reference
measured over 1 year of curing. Phases are abbreviated as follows: (Ett) ettringite,
(AFm) AFm, (Ms) monosulfoaluminate, (Gib) gibbsite, (Geh) gehlenite, (Hdg)
hydrogarnet, (Gyp) gypsum and (Wue) wüstite
The evolution of the DTG curves of formulation A and B are presented in

Figure 3.21. The dominant phase constitutions were consistent with those
observed from XRD. Despite the difficulty in distinguishing between the
overlapping peaks in certain temperature ranges, it was clear that
formulation A has a more complex hydrate assembly compared to
formulation B. Moreover, TGA confirmed the substantial amounts of AH3
and AFm phases (mainly monosulfoaluminate and strätlingite) which were
less evident using XRD due to their poor crystallinity.
Following the bound water content in Figure 3.22, the high value obtained
for formulation B was clearly due to its dominant hydrate, ettringite. This
supports the earlier assumptions relating the phase assemblage to the high
strength values and the expansive nature of mortars representing the high-
sulfate system. Nonetheless, the increasing bound water contents in both
formulations confirms continuous formation of hydrates, also reflected
through a progressive strength development over time.
ASTOVEZA 2022 103

Chapter 3 ____________________________________________________
Figure 3.21 Evolution of the DTG curves of (a) formulation A and (b) formulation B
along with their corresponding references. Slag-containing formulations are shown in
solid lines. Phases are abbreviated as follows: (Ett) ettringite, (AFm) AFm, (Str)
strätlingite, (Gib) gibbsite, (Hdg) hydrogarnet
Figure 3.22 Total bound water content of formulations A, B and their references
In an attempt to describe the slag hydration using a less conventional

technique in characterizing cement hydration, the hydrated paste samples
were analyzed using ATR-FTIR. From the preceding chapters, results from
previous study [117] have demonstrated formation of new bonds, depicted
by peak displacements in the ATR-FTIR spectrum. A shift in the Si-O
104 ASTOVEZA 2022

____________________________________________________ Chapter 3
asymmetric stretching band (700-1200 cm-1) towards the higher

wavenumber accompanied by the less apparent shift of the Si-O rocking
band (400-500 cm-1) towards a lower wavenumber were associated with
the partial oxidation Fe2+ to Fe3+ and with the incorporation of the Fe in the
polymerized network [116,117]. Although this mechanism was described
using a similar Fe-rich NFM slag, the binder investigated was an inorganic
polymer as opposed to the blended cements in this study.
The evolution of ATR-FTIR spectra of formulations A and B are presented
in Figure 3.23 and Figure 3.24, respectively. Based on literature
references, the band assignments of the identified spectral peaks in these
figures were listed in Table 3.2.
Figure 3.23 ATR-FTIR spectra of formulation A (solid black lines) and its reference.
identified.
Compared to the inorganic polymers investigated in Siakati et al.,
2020 [117], the cement systems in this study displayed a more complex
spectra due to the co-existence of various crystalline and amorphous
phases. The general phase assemblages of the two formulations were
ASTOVEZA 2022 105

Chapter 3 ____________________________________________________
consistent to those identified using XRD and TGA. The low-sulfate

system, formulation A, displayed signatures of ettringite,
monosulfoaluminate, gibbsite and strätlingite. A hydrogarnet peak (Al-O)
at around 810 cm-1 was also detected at 1 year only for the reference
formulation. Unfortunately, the Si-O peaks (930 cm-1 and 470 cm-1)
detected in the raw slag (Chapter 2, Figure 2.13) were hardly identifiable
in these systems. Despite the 30 wt% addition in the binders, the signatures
of the amorphous slag (low intensity broad peaks) were easily masked by
those of the crystalline components (high intensity sharp peaks).
Nonetheless, regions (Figure 3.23 a-c) where substantial changes in the
spectra were present, could be interpreted as indications of slag reactivity
considering the initial findings from XRD and TGA. In these areas, most
of the changes occur beyond 28 days of hydration. Recall that only about
10% of the slag (Figure 3.12) has dissolved at 28 days according to the
XCT analysis. One of the marked differences in patterns between the slag-
containing and the reference formulation is in region b at around 960-
980 cm-1 where the Al-O bands of gibbsite and strätlingite coincide. This
is around the same region where the Si-O stretching band of the raw slag
is expected. These marked differences in regions generally associated to
the Al-O bands, could be an initial indication that the slag hydration is
altering the formation of the Al-containing hydrates.
Figure 3.24 ATR-FTIR spectra of formulation B (solid blue lines) and its reference.
identified
As expected, the spectra (Figure 3.24) of the high-sulfate system exhibited

a less complex constitution dominated by distinct peaks assigned to
ettringite and gibbsite. The peak at 523 cm-1 where monosulfoaluminate
106 ASTOVEZA 2022

____________________________________________________ Chapter 3
(among other minor phases) was assigned is also less pronounced in

formulation B compared to that in formulation A (Figure 3.23). Moreover,
no discernible difference between the reference and the slag-containing
formulations nor throughout the curing period (28 days) was observed in
formulation B. This could support the overall lower slag reactivity in the
high-sulfate system estimated through XCT analysis (Figure 3.12).
Table 3.2 FTIR band assignments based on the distinct spectral peaks
observed from the hydrated paste samples
Primary Wavenumber
Bond Reference
contributor (cm-1)
O-H Ms/Ett/Str 3432 Horgnies et al., 2013 [137]
1660 Horgnies et al., 2013 [137]
S-O Ett/Ms 1115 Horgnies et al., 2013 [137]
Si-O-Al Str 1150 Okoronkwo and Glasser, 2016 [165]
710 Zapanta et al., 2020 [171]
Si-O anhydrous silicates 521 Torréns-Martin et al., 2013 [86]
Al-O Gib 1021 Schroeder, 2002 [172]
972 Schroeder, 2002 [172]
751 Schroeder, 2002 [172]
Str 965 Okoronkwo and Glasser, 2016 [165]
Hdg 810 Horgnies et al., 2013 [137]
Ett 857 Horgnies et al., 2013 [137]
Gib/Ms/Str/Hdg 523-537 Horgnies et al., 2013 [137]
Hdg 521 Torréns-Martin et al., 2013 [86]
In the end, the complexity of the ATR-FTIR spectra proved to be a huge

challenge in characterizing the phase assemblage evolution of the blended
-O bands could not be clearly
detected and therefore, no clear indication on the hydration mechanism of
the slag could be derived from this technique alone. Nevertheless, the
resulting phase assemblages were consistent with those obtained from the
more conventional techniques, XRD and TGA. Considering that each
analysis in ATR-FTIR does not require tedious sample preparations, could
readily take solid, paste or liquid samples, and takes less than a minute to
yield a spectrum, it could remain as a practical option for quick analyses.
3.4. The fate of Fe during hydration

In this section, the fate of Fe, the main component of the slag, was
investigated through TEM, STEM and Mössbauer spectroscopy. Although
ASTOVEZA 2022 107
Chapter 3 ____________________________________________________
the general phase assemblages of formulations A and B were already

described in the previous section using XRD, TGA and ATR-FTIR,
complementary techniques were still needed to elucidate the hydration
mechanism of the slag.
3.4.1. TEM
The TEM images of formulation A paste after 1 day of hydration are
presented in Figure 3.25. The observed phases were consistent with the
assemblage described in the preceding section. Several individual rods of
ettringite (A2 and A6); plate of anhydrous CA (A4); Al-rich gels (A1, A5,
A7 and A8) similar to gibbsite; and intermixed phases (i.e., A3) were
observed. No significant amount of Fe was detected in any hydrated phase
at this stage of hydration. Moreover, the amount of S detected through EDS
analysis in rod/needle-like phases labelled as ettringite was significantly
lower compared to what could be theoretically expected. This discrepancy
is linked to the uncertainties associated to the EDS analysis particularly for
S in ettringite, discussed in section 2.2.6.
Figure 3.25 TEM images and the atomic composition of marked areas in the (A)
4.5CAC- HH paste sample after 1 day of hydration
At longer curing ages (>28 days) when the degree of slag hydration became
more significant, Fe was detected in the Al-rich gels with a slightly higher
amount of Si compared to the gels observed in the early days of hydration.
Figure 3.26 shows the general morphologies (a) and the Fe-containing
intermixed phases (b-c) observed in formulation A. At 90 days, more plates
108 ASTOVEZA 2022
____________________________________________________ Chapter 3
of strätlingite (A10) intermixed with Al-rich gels were observed, which

were n
used to describe mostly the Fe-containing intermixed phase considering its
morphology and poor crystallinity. Although the gel was clearly a part of
the solid binder, the use of this term was adapted from Cuesta et al., 2017
[173] and related literature [140,174] where amorphous aluminum
hydroxide gels were characterized in calcium sulfoaluminate cement
systems.
Figure 3.26 TEM images and the atomic composition of marked areas in the (A)
4.5CAC-HH paste sample after long term curing
For some of the Fe-bearing gel phases (i.e., A12), the Fe signal was notably
higher than those observed in other gels (i.e., A9 and A11). To better
understand their nature, the diffraction mode (Figure 3.27) of TEM was
used to characterize gels at (a) 180 days and (b) 1 year, both containing
diffraction spots compared to the A13 gel at 180 days. The former could
be interpreted to have some crystalline component/s [175] intermixed with
gel phase/s; while the latter was clearly more poorly-structured [176]. The
possibility of having erroneously analyzed unreacted slag particles as part
of the gels was excluded, not only due to the spot size (only a few hundred
nanometers) but also due to the sensitivity (Figure 3.28) of these gels.
Even the gels at 1 year (A16 and A17) were immediately destroyed after
the EDS analysis, suggesting that they were, most likely, hydrated phases.
ASTOVEZA 2022 109

Chapter 3 ____________________________________________________
Figure 3.27 TEM images, diffraction patterns and the atomic composition of Fe-
containing gels in the (A) 4.5CAC-HH paste sample
Figure 3.28 TEM images before and after EDS analysis of Fe-containing gels in the (A)
4.5CAC-HH paste sample
Figure 3.29 TEM images and the atomic composition of crystalline phases in the (A)
4.5CAC-HH paste sample
110 ASTOVEZA 2022

____________________________________________________ Chapter 3
Contrary to the typical hydrated phases previously reported [170,177] to

incorporate Fe in cement systems (i.e., ettringite, strätlingite, AFm), no
significant amount of Fe was detected in ettringite (A2, A19, A20) and
strätlingite (A10, A18) identified in formulation A. Moreover, dendrites of
wüstite (A21) were observed intact after long curing ages, supporting the
observation in the previous section that the crystalline fraction of the slag
displayed no significant reactivity even after 1 year of hydration.
Similar findings on the fate of Fe were obtained from the TEM analyses
(Figure 3.30) of formulation B. Although a higher quantity of ettringite
rods was observed in the general phase assemblage (Figure 3.30 a) of this
high-sulfate sample, Fe was only found to be incorporated in an Al-rich
intermixed gel phase as in formulation A. At longer curing time, 1-2 at%
of Mg was generally associated with some of these Fe-bearing phases,
which could be an initial indication of an LDH phase formation similar to
what was reported in Hallet et al., 2022 [139]. More details on this phase
are presented along with the STEM results in the next section.
Figure 3.30 TEM images and the atomic composition of marked areas in the (B) 1.6CAC-
HH paste sample
ASTOVEZA 2022 111

Chapter 3 ____________________________________________________
3.4.2. STEM
Although TEM facilitated the identification of individual phases with
relatively less intermixing compared to the previous attempt using SEM
(section 3.2.1), the homogeneity of the Fe-bearing phase observed in TEM
was still in question. STEM was therefore employed to create elemental
maps focusing on these phases.
In Figure 3.31, the distribution of Fe suggests that the phase was

comprised of at least two phases: an amorphous gel (A23/A25) and a phase
(A24) with clearer diffraction spots. Fe appeared to be homogenously
distributed at lower concentration (5 at%) in the amorphous gel. A higher
incorporation of Fe (19 at%) was noted for the other phase (A24)
intermixed within this gel. For this specific sample, it was observed that
the increase in Fe content was accompanied by a decrease in Al proportion
which could be a possible indication that these ions were occupying the
same sites in the lattice of the intermixed phases. Nonetheless, the
possibility that the high-Fe region (A-24) actually corresponds to a
component of the raw slag could not be fully excluded, despite the noted
sensitivity of the phase to EDS analysis only observed on hydrated phases
during this study.
Another area of the same sample was examined and presented in Figure
3.32. The Fe content detected for the amorphous hydrate gels (A27 and
A28) remained below 12%, consistent with the general trend observed
from the TEM and STEM analyses. In addition, unreacted slag particles
(i.e., A26, A29) were detected. The EDS analysis (A26) and diffraction
pattern (EDS-A29) suggest that these particles could be wüstite rather than
the amorphous fraction of the raw slag, supporting the earlier observations
that the mineral fraction of the slag was less (if at all) reactive during 1
year of hydration.
112 ASTOVEZA 2022

____________________________________________________ Chapter 3
Figure 3.31 STEM elemental mapping, diffraction patterns and atomic composition
of marked areas in the (A) 4.5CAC-HH paste sample after 180 days of curing
ASTOVEZA 2022 113

Chapter 3 ____________________________________________________
Figure 3.32 STEM elemental mapping, diffraction pattern and atomic composition
of marked areas in the (A) 4.5CAC-HH paste sample after 180 days of curing
The compositions of the Fe-containing intermixed phases were plotted in

a Ca-Si-Al ternary diagram in Figure 3.33. Only the gels with 1-12 at% Fe
were selected in order to avoid analytical volumes with strong intermixing
with non Fe-bearing hydrated phases (<1 at% Fe) or with the unreacted
slag (> 29 at%). Additionally, values in the ternary diagram were also
expressed in atomic percentages, considering that comparison based on
atomic ratios rather than the absolute amounts has been cited [178] to
reduce errors associated to oxygen quantification in TEM-EDS analyses.
114 ASTOVEZA 2022

____________________________________________________ Chapter 3
Figure 3.33 Ternary diagram of Fe-containing hydrate gels with 1-12% Fe

observed using TEM and STEM. The values were calculated from the Ca, Si and Al
atomic composition from EDS analyses normalized to 100%. Reference points for
gibbsite, strätlingite, katoite and Si-hydrogarnet were added in the diagram based on
their empirical formula.
The ternary diagram in Figure 3.33 shows the distinct Al-rich cluster
occupied by the Fe-bearing hydrate gels. In view of the limited number
points, however, no definite conclusion could be drawn to compare the
variation in the composition between the gels in formulation A or B, nor
among the samples at different curing ages. Reference points for gibbsite,
strätlingite, katoite and Si-hydrogarnet were also added in the diagram
based on their empirical formula. The composition of the hydrate gels was
observed to fall between that of gibbsite and stätlingite/katoite, but
remained distinct from these pure phases.
Further attempts were made using STEM after 1 year of hydration to
identify the Fe-containing phase which exemplified some diffraction spots
in previous analyses (i.e., EDS-A24, Figure 3.31). Fe maps and diffraction
patterns of the Fe-bearing intermixed phases observed in formulation A are
presented in Figure 3.34 and Figure 3.35. The inhomogeneity of Fe
distribution is clear from the maps, consistent with the observation from
180 days. A distinct rod-like to lamellar morphology was also observed for
the phase with higher Fe content (i.e., A30 and A31, Figure 3.34) in the
updated 1 year sample. In an attempt to measure the interatomic spacing
(d-spacing) for the identification of this phase, a more focused STEM
image was taken as shown in Figure 3.36.
ASTOVEZA 2022 115

Chapter 3 ____________________________________________________
Figure 3.34 STEM image focused on an Fe-rich region (a);

along with its corresponding Fe map from EDS analysis (b); and the diffraction pattern
of marked area A-31 from the (A) 4.5CAC-HH paste sample after 1 year of curing
Figure 3.35 STEM image focused on an Fe-rich region (a); along with its corresponding
annular dark field image (b); Fe map from EDS analysis (c); and the diffraction pattern
of marked areas from the (A) 4.5CAC-HH paste sample after 1 year of curing
116 ASTOVEZA 2022

____________________________________________________ Chapter 3
Figure 3.36 STEM images focused on the needle-like Fe-rich phase (a, b and d),
along with a grayscale intensity plot (c) of the pixels under the yellow double-arrow in d.
The distances in c represent the atomic layer spacing of the unidentified phase from the
(A) 4.5CAC-HH paste sample after 1 year of curing
The d-spacing of the unidentified high Fe-bearing needle-like phase was

estimated at around 0.72 nm or equivalent to 7.2 Å using the grayscale
intensity plot (c) in Figure 3.36. Unfortunately, it was not possible to
perform EDS analysis on this phase due to its size (only a few nanometers)
and sensitivity (easily disintegrated), causing them to move out of focus
during the imaging. The other Fe-bearing needle-like phases (i.e., A30 and
A31) were still intermixed with Al-rich gels and, therefore, would not yield
the actual composition of the pure needle-like phase. Nonetheless, the
distinct morphology, the estimated d-spacing, and the general idea on the
chemistry of this Fe-bearing phase could be compared to those of some
relevant phases previously observed and reported in literature. Figure 3.37
shows some TEM images of Fe hydroxides (a and b) and an Fe-bearing
LDH phase (c), appearing to have a similar morphology to those observed
in the 1 year sample.
ASTOVEZA 2022 117

Chapter 3 ____________________________________________________
Figure 3.37 TEM images of needle-like Fe hydroxide phases (a, b);

and a lamellar layered double hydroxide (LDH) phase similar to green rust, modified
from literature references [179 181].
Three main possible phases were considered for the identity of the Fe-
bearing needle-like phase, described as follows:
i. Al-Fe hydroxide:
After 28 days, Fe signals were detected in Al(OH)3 gels with increasing
proportions of Ca and Si. In literature, amorphous Al(OH)3 were
observed to form polycrystalline "whiskers" morphology shown in the
TEM images in Figure 3.38 a. The different Fe hydroxides (Figure
3.37 a and b; Figure 3.38 b) demonstrated a nano-crystalline needle
morphology, similar to those observed in the 1 year samples. The
mechanism of transformation from the amorphous end-product
ferrihydrite to the needle-like goethite nano-crystals over time, has been
extensively discussed in Bazilevskaya, 2009 [182]. In her study, the
difference in the kinetics of ion dissolution was attributed to the
eventual growth of goethite needles within a poorly-crystalline
ferrihydrite matrix, thereby creating polycrystalline ring diffraction
patterns.
The d-spacing of Al and Fe hydroxides in literature were reported to be
typically below 4 Å [183,184]. However, it has been previously
118 ASTOVEZA 2022

____________________________________________________ Chapter 3
demonstrated that with an uptake of ion/s, this spacing could drastically

change (i.e., 7.4 Å with Cl- uptake in Fe3+O(OH,Cl), Figure 3.37 b
[180]).
Figure 3.38 TEM images of (a) Al(OH)3 with its inset selected-area electron diffraction
(SAED) pattern [183]; and of goethite, Fe(OH)O, with its inset SAED pattern [184].
ii. Fe-LDH, [MII(1-x)MIIIx(OH)2]x+. [(x/n) An-, m H2O]x-:

The formation of this phase during the hydration of an NFM slag in a
PC-based system has been experimentally demonstrated in Hallet et al.,
2022 [139]. When taken in a perpendicular angle to the synthetic LDH
planes, its TEM image in Figure 3.37 could also resemble the
morphology of the Fe-bearing phase obtained in this study. The
presence of this phase might explain the Mg detected in some of the
samples (i.e., A30 and A31 in Figure 3.34). In the cationic hydroxide
layers, Mg, Fe, (Ca, Al) ions octahedrally-coordinated with OH-, could
take the MII and MIII roles as seen in previous studies [181,185]. On the
other hand, water molecules and CO2 occupy the anionic interlayers.
Recall that the detection of the C is problematic with the EDS analysis
due to its low atomic number, which is likely to result in its
underestimation.
It was notable that the (Mg+Fe)/Al molar ratio at approximately 1.3
obtained in A30 and A31 was much lower than that of the synthetic
(Mg, Fe)2+-Al3+ LDH (hydrotalcite-like) structure at 3 [186]. This
difference could be associated to the high Al content either due to the
nature of the system itself (CAC-rich), or to the intermixing with the
Al-rich gels within the same analytical volume. Similar deviations have
been reported by Hallet et al., 2022 [139] due to the latter reason.
However, this could not directly explain the Si content detected in the
samples which has not been reported to partake in an LDH structure. In
terms of the d-spacing, values at 7.42-7.48 Å were reported for the Fe-
containing LDH [170,185], making it closer to the 7.2 Å obtained for
ASTOVEZA 2022 119

Chapter 3 ____________________________________________________
the 1 year sample as opposed to those of the Al-Fe hydroxides. One of

the factors that these differences in the d-spacing among the Fe-LDH
phases can be attributed to was the oxidation state of Fe. The different
proportions of Fe2+ and Fe3+
configuration, such that a closer interlayer spacing is expected in the
oxidized state [139].
iii. Fe-Si hydrogarnet

According to the thermodynamic model in section 3.3.1, this should be
the most thermodynamically-favored Fe-bearing phase at long term
(infinite time). Given the seemingly increasing proportion of Ca and Si
towards 1 year, the formation of this phase at longer curing time could
not be eliminated. In terms of d-spacing, the reported values for C3FH6
and C3FSH4 are at 4.49 Å and 5.14 Å, respectively [132], much lower
than those observed for the unidentified Fe-bearing phase observed in
this study. Although an uptake of Mg2+ in the Ca2+ sites can occur in
hydrogarnets [132], the Al content of the unidentified phase remains
significantly elevated. Moreover, a needle-like or lamellar morphology
has not been previously obse
in typically cubic (Chapter 1, Figure 1.4) hydrogarnets.
Among the three possible phases presented, Al-Fe hydroxide and Fe-LDH
seem to have the closest characteristics with the phase in question.
Unfortunately, the data obtained from TEM and STEM were proven not
sufficient to draw a concrete conclusion as to its identity, particularly
hindered by the intermixing of phases even on the nano-scale. This was
aggravated by the sensitivity of the hydrates resulting in their immediate
disintegration during the EDS analysis.
3.4.3. Mössbauer spectroscopy

To further describe the fate of Fe during slag hydration, 57Fe Mössbauer
spectroscopy was performed on the paste samples. The evolution of the
spectra of formulation A over 1 year of curing is presented in Figure 3.39.
Additionally, the spectra of formulation B at 28 days and the raw slag were
superimposed in the same figure for comparison.
It is evident from the general trends that the atomic environment of Fe
changes as the slag hydrates over time. In particular, the most drastic
transformation took place after more than 7 days (Figure 3.39) coinciding
with the changes in the phase assemblage in section 3.3 (start of transition
from ettringite- to AFm-rich system). The relative intensities of the
transmittance diminished over time as the highly amorphous raw slag was
diluted by the other components of the blend (CAC, HH) and its Fe was
120 ASTOVEZA 2022

____________________________________________________ Chapter 3
incorporated into the Al-rich hydrate gel, as seen from the TEM and STEM
results. At 28 days, the estimated slag hydration degrees of formulation A
and B were similar (10 % and 9 %, respectively, Figure 3.12) which could
explain the close resemblance of their Mössbauer spectra.
Figure 3.39 57Fe Mössbauer spectrum evolution of formulation A sample over 1 year
of curing. Superimposed are the spectra of formulation B after 28 days of hydration and
of the raw slag.
ASTOVEZA 2022 121

Chapter 3 ____________________________________________________
Two fitting procedures were explored in analyzing the spectra of

formulation A, producing a complex (Model 1) and a simplified model
(Model 2), described as follows:
Model 1, performed using the IMSG software, adopted most of the Fe

speciation of the raw slag presented in Chapter 2 (section 2.2.5)
comprising of two Fe2+ doublets (G1 and G2) and one Fe3+ doublet (G3)
in the glass fraction, and a doublet of Fe2+ (M1) for wüstite. The sextet
magnetic component (M2) was excluded for the paste samples because
its signature was almost negligible after the dilution. While the
hyperfine parameters of the new hydrates were allowed to vary, those
of the raw slag were fixed for all the samples. Based on the findings
from the other characterization techniques suggesting that wüstite was
unreactive throughout the curing period studied, its absorption area was
fixed at 18.2 % which was the value obtained from the raw slag fitting
(section 2.2.5). Note that this value was similar to the contribution of
wüstite estimated in Appendix J, assuming that the Mössbauer spectra
directly reflect the weight percentages of its components. Moreover, the
intensities and absorption areas of the glass components were made
interdependent, maintaining a similar ratio as hydration proceeds. This
was based on the assumption that the amorphous fraction of the slag
dissolves uniformly as slag reacts.
Model 2, performed using the WinNormos® software, offered a
simplified approach using only three components for the fitting of the
hydrated paste spectra consisting of three doublets: one for Fe2+ in the
amorphous fraction of the slag (G4), one for wüstite (M1), and one for
the Fe3+ mainly in the hydration product (P3). This model was presented
to challenge the integrity of the general trends (qualitative) observed in
Model 1 by allowing variations in the intensities and hyperfine
parameters of all components without imposing restrictions. For
instance, the absorption area of M1 (wüstite) was not fixed like in
Model 1. It must be noted that Model 2 was not made to challenge the
absolute value of each Fe component estimated from Model 1.
The deconvoluted spectra of formulation A paste after 28 days of
hydration, derived from the two models, are presented in Figure 3.40. The
Mössbauer hyperfine parameters used for the fitting procedures are
detailed in Table 3.3 (Model 1) and Table 3.4 (Model 2). Supplementary
figures obtained during the modelling can be found in Appendix J.
122 ASTOVEZA 2022

____________________________________________________ Chapter 3
Figure 3.40 57Fe Mössbauer spectrum of formulation A and its components

derived from a (a) complex and a (b) simplified modelling approach
ASTOVEZA 2022 123

Chapter 3 ____________________________________________________
Table 3.3 57Fe Mössbauer hyperfine parameters used for Model 1.

The abbreviations are designated as follows: isomer shift (IS); central value of the
quadruple splitting (QS); central value of the hyperfine magnetic field (MF); total
Gaussian-
he component (AA). The typical errors for IS
and QS is ± 0.02 mm/s; and ± 5% for A
IS QS MF AA
Sample Component (mm/s) (mm/s) (kOe) (mm/s or kOe) (%)
7 days (G1) Fe2+ 1.15 1.91 0.00 0.42 29
(G2) Fe2+ 0.79 1.73 0.00 0.51 39
(G3) Fe3+ 0.37 1.04 0.00 0.17 8
(M1) wüstite 0.91 0.66 0.00 0.20 18
(P1) Fe3+ hydrated 0.21 0.25 0.00 0.25 3
(P2) Fe2+ hydrated 0.82 0.97 0.00 0.40 3
28 days (G1) Fe2+ 1.15 1.91 0.00 0.42 15
(G2) Fe2+ 0.79 1.73 0.00 0.51 18
(G3) Fe3+ 0.37 1.04 0.00 0.17 4
(M1) wüstite 0.91 0.66 0.00 0.20 18
(P1) Fe3+ hydrated 0.32 0.83 0.00 0.28 18
(P2) Fe2+ hydrated 0.89 0.93 0.00 0.39 27
90 days (G1) Fe2+ 1.15 1.91 0.00 0.42 14
(G2) Fe2+ 0.79 1.73 0.00 0.51 16
(G3) Fe3+ 0.37 1.04 0.00 0.17 3
(M1) wüstite 0.91 0.66 0.00 0.20 18
(P1) Fe3+ hydrated 0.24 0.75 0.00 0.35 22
(P2) Fe2+ hydrated 0.83 0.91 0.00 0.45 27
180 days (G1) Fe2+ 1.15 1.91 0.00 0.42 10
(G2) Fe2+ 0.79 1.73 0.00 0.51 13
(G3) Fe3+ 0.37 1.04 0.00 0.17 3
(M1) wüstite 0.91 0.66 0.00 0.20 18
(P1) Fe3+ hydrated 0.27 0.84 0.00 0.35 34
(P2) Fe2+ hydrated 0.84 0.87 0.00 0.34 22
Table 3.4 57Fe Mössbauer hyperfine parameters used for Model 2.

The abbreviations are designated as follows: isomer shift (IS); central value of the
quadruple splitting (QS); central value of the hyperfine magnetic field (MF); total
Gaussian-
IS QS MF AA
Sample Component (mm/s) (mm/s) (kOe) (mm/s or kOe) (%)
raw slag (G4) Fe2+ 1.10 2.12 0.00 0.57 39
124 ASTOVEZA 2022

____________________________________________________ Chapter 3
(M1) wüstite 1.10 0.95 0.00 0.78 53

(P3) Fe3+ 0.15 0.17 0.00 0.30 6
(M2) magnetic 0.02 0.00 33.30 2
7 days (G4) Fe2+ 0.95 2.10 0.00 0.53 39
(M1) wüstite 0.88 0.97 0.00 0.75 50
(P3) Fe3+ 0.14 0.20 0.00 0.31 10
28 days (G4) Fe2+ 1.00 1.92 0.00 0.57 22
(M1) wüstite 1.00 0.68 0.00 0.57 55
(P3) Fe3+ 0.13 0.47 0.00 0.46 23
90 days (G4) Fe2+ 0.91 2.21 0.00 0.53 20
(M1) wüstite 1.03 0.59 0.00 0.54 49
(P3) Fe3+ 0.25 0.37 0.00 0.41 30
180 days (G4) Fe2+ 0.91 2.21 0.00 0.52 12
(M1) wüstite 1.03 0.59 0.00 0.54 48
(P3) Fe3+ 0.22 0.51 0.00 0.51 39
In Model 1, two new doublets (P1 and P2) were introduced as products of
hydration. The parameters of the P1 component at 7 days, displaying
relatively low IS and QS values, could represent a phase with a similar
structure as the Fe(OH)3 gels reported in PC cement systems [106,135,187]
which in time develops into Fe-monosulfate [129,136]. This
transformation is accompanied by an increased distortion of the atoms
surrounding the Fe signaled by the higher QS values at later ages (<7 days).
However, given the nature (poor to no crystal structure) of the Fe-bearing
hydrate gel observed from TEM and STEM, it was difficult to clearly
postulate the exact atomic environment of the new Fe-containing phases
detected. As Fe is relatively scarce in conventional cement systems,
especially in CAC-based binders, there is no sufficient Mössbauer data on
Fe-containing hydrates, at the moment, to compare with those obtained in
this study. Although the hyperfine parameters of P1 could only be
associated to that of Fe-monosulfate given these constraints, it was clear
from the EDS analyses of TEM and STEM that the Fe-bearing hydrates
could not be monosulfate considering the extremely low (<1.5 at.%)
amount of S detected. Nonetheless, the atomic environment (Fe-Al solid
substitution) and oxidation state (Fe3+) of Fe could be very similar, which
is also the case for the ferrihydrite gels versus the Al-rich hydrate gels at
early ages.
Furthermore, the hyperfine parameters of P2 in Model 1 was even more
intriguing with the combination of IS and QS values that has never been
systems. In particular, Fe2+ in 4- or 5-fold coordination with oxygen is
ASTOVEZA 2022 125

Chapter 3 ____________________________________________________
considered rare [188]. The closest structure that could be compared to this
component is one of the doublets of staurolite
((Fe,Mg)2(Al,Fe)9O6[SiO4]4(O,OH)2) reported in Dyar et al., 1991 [189].
This doublet is in the exceptionally low end of the range for 4-fold
coordinated Fe2+ arising from the delocalization of electrons due to charge
transfer phenomena (i.e., Al3+ Fe2+ or Fe 2+ Fe substitutions) [189]. In
-formed component P2 versus
the Fe2+ in G1 and G2 from the raw slag could suggest a generally more
ordered environment of Fe. This could well be the case if the component
P2 majorly corresponds to the unidentified Fe-bearing phase detected in
STEM, such that the Fe2+ is transitioning from an amorphous state in the
raw slag.
In Model 2, the Fe in the raw slag was assumed to be dominantly Fe2+ (G4)
with an environment comparable to those in fayalite minerals [114,190].
The hyperfine parameters of the doublet P3 varied significantly as the
component shifts from representing a minor Fe3+ proportion of the raw
-bearing hydration product in the paste
samples. This was expected given the simplified approach with minimal
restrictions imposed on the hyperfine parameters and absorption areas, as
described earlier, for Model 2. As hydration proceeds, the best fit obtained
for the spectra suggests P3 parameters closer to those of the Fe-bearing
hydrates in cement systems represented by P1 in Model 1. In addition,
resulted in the overestimation of the M1 (wüstite) component.

Nevertheless, it is important to highlight that the exact values of the
Mössbauer parameters are difficult to directly relate to the theoretical Fe
environment in crystal structures notably due to the presence of long-range
interactions and cation substitutions [189]. As mentioned earlier, the
interest in this application is to observe the general trends that could
elaborate a better understanding of the evolution of Fe speciation during
slag hydration. For this purpose, the evolution of the absorption areas
estimated from the two models using the formulation A paste samples is
presented in Figure 3.41. Supplementary figures obtained during the
modelling may be referred to in Appendix J.
The diminishing absorption areas over time (Figure 3.41) of the
components representing the glass fraction of the slag (G1, G2 and G3 for
Model 1 and G4 for Model 2) were consistent for both models suggesting
slag dissolution. New Fe-bearing phases were generated as the slag
hydrates, where Fe was potentially in octahedral coordination with oxygen
in the 3+ oxidation state or in tetrahedral coordination with oxygen in the
2+ oxidation state. Despite the overestimation of wüstite in Model 2, the
overall steady values of its absorption areas, even when no fitting
126 ASTOVEZA 2022
____________________________________________________ Chapter 3
constraint was applied for the different samples, supported the assumption
Figure 3.41 Absorption areas (AA) calculated from the two modelling approaches
By following the changes in the sum of the absorption areas (AA) of the
reactive components of the raw slag (G1, G2 and G3 for Model 1 and G4
for Model 2), the slag hydration degrees of formulation A at different
curing ages were estimated (Figure 3.42). Similar to the assumption
applied in the XCT analysis in section 3.2.2, this estimation equates slag
dissolution to slag hydration. Note that the original equation for the slag
hydration degree presented in Chapter 2 (section 2.2.5) needed correction
due to the apparent overestimation of wüstite especially in Model 2. The
details of these calculations can be referred to in Appendix H.
ASTOVEZA 2022 127

Chapter 3 ____________________________________________________
Figure 3.42 Slag hydration degree in formulation A paste (primary axis)

estimated based on the changes in the total absorption areas (secondary axis) of the
unreacted slag for Models 1 and 2.
The estimated values of the slag hydration degree converge at longer

curing ages despite the steady difference in the values of the absorption
areas between the two models. These estimated values were also higher
compared those estimated from the XCT analysis (section 3.2.2). Although
the values from these two techniques could not be directly equated, one of
the main factors that could explain the lower values obtained from the XCT
was the voxel resolution setting, 2.15 µm. The result from the Mössbauer
spectra fitting proved to be essential in confirming this issue. Nonetheless,
the abrupt changes in the values between 7 and 28 days (Figure 3.42) call
for further refinement of the Mössbauer fitting procedures. Although it was
clearly based on the general trends alone (Figure 3.39) that a significant
transformation occurred between these curing times, the resulting
hyperfine parameters (Table 3.3 and Table 3.4) from the best fit models
were also considerably altered. Yet, they were presumed to represent the
same components due to the lack of concrete information on the actual
structure of the Fe-bearing hydrates.
128 ASTOVEZA 2022

____________________________________________________ Chapter 3
3.5. Activation through paper residue addition

The combination of various techniques provided abundant information
essential in characterizing and estimating the degree of slag hydration in
the CAC-based ternary systems investigated. The results presented earlier
in this chapter, did not only confirm slag reactivity but also demonstrated
its long term positive contribution to compressive strength in
dimensionally-stable mortars. Despite these encouraging results, one of the
critical challenges that needs to be addressed in order to increase its
potential for application is the activation at the early ages (<28 days) of
hydration.
In Chapter 2 (section 2.3), the reactivity of several SOCRATES residues
was evaluated in a screening test. The distinct mineralogy of the lime-rich
paper residue (Top-Crete®) yielding considerably high conductivity
values in water, supported its promising potential as a green binder for
which it is currently being marketed for. When added at 23 wt% in a CAC-
HH binder system (CAC/HH = 1.6), high-intensity rapid heat release
(Figure 3.43 a) in the first hour of hydration was observed. Compared to
a formulation with an equivalent volume of slag, the hydration peaks also
appeared earlier, likely indicating faster reaction kinetics. However, the
cumulative heat release curves (Figure 3.43 b) show that the total released
was, in fact, lower for the paste with the paper residue from about 4 hours
onwards.
Figure 3.43 (a) Heat flow and (b) cumulative heat release of pastes containing 23 wt%
paper residue (in blue) and 30 wt% slag (in black).
One concern on the use of this paper residue as SCM is its elevated
(23 wt%) free lime content (Appendix B) excessively surpassing the 1 or
ASTOVEZA 2022 129

Chapter 3 ____________________________________________________
2.519 wt% limit imposed by the EN-450 [191]. Although this limit was
intended for fly ash addition, De Shepper et al., 2011 [192] reiterated the
significance of maintaining the free lime content below 3 wt% in cements
to prevent undesirable effects such as volume expansion, flash setting and
reduced mechanical strength. At 23 wt% addition in the binder, the
effective free lime content was 5.2 wt% from the paper residue alone. For
this reason, this residue was evaluated separately from the other potential
SCM in this study.
In an attempt to increase the slag reactivity in formulation A, the paper
residue was added at 5 wt%, following the blend formulation A-TC (Table
2.3, where TC stands for Top-Crete® paper residue). Even with this small
amount of paper residue, the addition of 0.3 wt% citric acid was needed to
prolong the setting time (Figure 3.44) allowing the freshly-mixed blends
to be properly placed in the moulds. The properties of this blend were then
compared to those of formulation A, and a reference formulation
(reference no TC) with the citric acid addition but without the paper
residue.
Figure 3.44 Initial and final setting time of formulations A-TC; reference no TC; and
formulation A
The compressive (Figure 3.45) and flexural (Figure 3.46) strength

evolution along with the SAI values demonstrated the significant
contribution of the paper residue addition to the strength of the mortar
samples. However, the addition of the citric acid alone also appeared to
have contributed to this strength increase based on the values of the
reference formulation. For instance, a compressive strength of 27 MPa was
obtained without the citric acid addition, while this reached 38 MPa with
the citric acid after 28 days of hydration (Figure 3.45). The benefit of the
19
when a complementary autoclave expansion test yields satisfactory results
130 ASTOVEZA 2022

____________________________________________________ Chapter 3
citric acid addition in improving the workability has been reported [193
195] to have an impact on the mechanical strength, mainly by altering the
microstructure and the porosity during the early age hydration.
Nonetheless, as previously highlighted in Chapter 2, such a low amount of
addition (0.3 wt%) for the binders in this study should not have an
influence in the long term phase assemblage of the binders [82].
Figure 3.45 Compressive strength (primary axis) and SAI evolution (secondary axis) of
formulation A-TC (blue), its reference formulation (gray) and formulation A (black).
SAI was calculated by dividing the strength values of A-TC and the reference by that of
the formulation A.
Figure 3.46 Flexural strength (primary axis) and SAI evolution (secondary axis) of
formulation A-TC (blue), its reference formulation (gray) and formulation A (black).
SAI was calculated by dividing the strength values of A-TC and the reference by that of
the formulation A.
The SAI values of A-TC, particularly those for the compressive strength
(Figure 3.45), suggest that the positive contribution of the paper residue
addition to the strength was most pronounced in the first day of curing,
progressively diminishing over time. This could be related to the length
variation in Figure 3.47 indicating a period of expansion from the 5th hour
ASTOVEZA 2022 131

Chapter 3 ____________________________________________________
or an overall period of shrinkage suppression between 5 and 15 hours.

While the paper residue was dissolving extensively during this period
according to the electric conductivity measurements in section 2.3.1, a
more dimensionally-stable binder was simultaneously obtained using
mortar samples at around the same period, leading to high strength values
at early ages of hydration.
Figure 3.47 Length variation of formulations A-TC (blue) and A (black) obtained from
Walter+Bai shrinkage test
Furthermore, the slight increase in slag hydration degree was validated by

the result of the XCT analysis presented in Figure 3.48. Note however that
the differences in the values from those obtained for formulation A, were
close to the ± 3 % approximate error range of XCT determined from the
sensitivity analysis in section 3.2.2. In addition, no significant difference
in the Mössbauer spectra (Figure 3.49 b) between the two formulations
was observed at 28 days.
Figure 3.48 The slag hydration degree (primary axis) and slag total volume of
formulations A-TC and A estimated from XCT + volume analysis
132 ASTOVEZA 2022

____________________________________________________ Chapter 3
Figure 3.49 (a) XRD patterns; (b) Mössbauer spectra; (c) DTG curves; and (d) total
bound water evolution of formulations A-TC, its reference and formulation A.
comprised of gehlenite and monosulfoaluminate.
Despite the minimal addition, the early hydration reactions of the paper
residue appeared to have altered the phase assemblage in the first 28 days
by favoring the formation of AFm phases (likely monosulfoaluminates
given the peak locations), evident from the regions marked in red in the
XRD (a) and DTG (c) patterns in Figure 3.49. This is also consistent with
the generally lower total bound water content (d) of formulation A-TC at
1 and 7 days. Similar to the previous observations in formulation A, Fe
was detected only in Al-rich hydrated phase at long curing times (Figure
3.50). After 1 year, the abundance of these hydrate gel phases was apparent
throughout the TEM frames observed.
ASTOVEZA 2022 133

Chapter 3 ____________________________________________________
Figure 3.50 TEM images and the atomic composition of marked areas in the (A-TC)
4.5CAC-HH-5TC paste sample
While the observed increase in strength in the high-sulfate system
(formulation B) was attributed to the increased ettringite formation, that in
A-TC could be partly attributed to a marginally higher degree of slag
hydration based on the XCT estimations (Figure 3.48). These initial
results could boost the interest in the valorization of the paper residue as
an SCM and an activator for slag. In the end, the mix design selection will
be critical in maximizing the synergistic benefits from the slag and the
paper residue.
134 ASTOVEZA 2022

____________________________________________________ Chapter 3
Conclusion
This chapter demonstrated the reactivity of the slag in binary CAC-HH
systems. The long term (>28 days) contribution of the slag to the
mechanical strength of mortars was confirmed by the evolution of SAI
values increasing over time for both the low (formulation A) and the high
(formulation B) sulfate systems. Although both produced dimensionally-
stable mortars, an overall shrinkage was observed in formulation A as
opposed to the expansive nature of formulation B. This was associated to
the difference in the phase assemblage of these formulations, generally
described using XRD, TGA and ATR-FTIR as follows:
Formulation (A) 4.5CAC-HH: dominated by monosulfoaluminate;
ettringite and gibbsite content decreased over time while strätlingite
formed significantly from 7 days onwards
Formulation (B) 1.6CAC-HH: dominated by ettringite and gibbsite;
with small amounts of strätlingite and monosulfoaluminates
Despite the higher strength values of formulation B mortars, the rate of
slag dissolution remained comparable to that of the low-sulfate system (A),
according to XCT. After 1 year, approximately 49 wt% of the slag
dissolved in formulation A and 43 wt% in B. These estimations were lower
compared to those obtained from Mössbauer spectroscopy. For instance,
56 wt% of the slag was estimated to have reacted after 180 days based on
the latter, while this value was only at 36 % using XCT. This difference
was mainly attributed to the voxel resolution setting (2.15 µm) in XCT,
which could have underestimated the actual values.
No Fe-containing hydrated phase was detected from XRD, TGA and ATR-
FTIR. Using TEM and STEM, Fe was found to be incorporated at low
2-5 at%) in an amorphous phase rich in Al. As the slag
dissolved more significantly after 90 days, an unidentified needle-like
phase with a higher Fe
contained small amounts of Ca, Si and Mg, and could be similar to Al-Fe-
hydroxides or Fe-LDH. Unfortunately, the techniques presented proved to
be insufficient for identifying this phase, given the problems with
intermixing on the nano-scale, and with the sensitivity of hydrates to EDS
analysis. Nonetheless, it was clear that Fe was not detected in the more
conventional hydrated phases, such as ettringite, monosulfoaluminate or
strätlingite.
Moreover, the evolution of the Mössbauer spectra of formulation A
provided strong evidence of the changes in the atomic environment of Fe
during hydration. The 4/5-fold coordinated Fe2+ dominating the raw slag
was transformed into a combination of Fe3+ and Fe2+ in octahedral and 4/5-
ASTOVEZA 2022 135

Chapter 3 ____________________________________________________
fold coordination, respectively. It was however difficult to postulate the

exact atomic environment of the new Fe-containing phases detected
through TEM and STEM, needed to validate the fitting models for the
Mössbauer spectra. The hyperfine parameters of the new Fe3+ component
were similar to those of the ferrihydrite gels in cement systems, which in
time develops into Fe-monosulfate. On the other hand, the new Fe2+
component had parameters comparable to a rare doublet of staurolite
((Fe,Mg)2(Al,Fe)9O6[SiO4]4(O,OH)2). It must be emphasized that the
author does not imply the presence of this mineral in the system. Rather,
the rare Fe environment associated to the charge transfer phenomena (i.e.,
Al3+ Fe2+ or Fe 2+ Fe substitutions) in staurolite was similar to what was
observed in the new component.
At the end of this chapter, the activating effect to the slag of the lime-rich
paper residue (added at 5 wt%) was investigated. The increase in the
strength of mortars was accompanied by high SAI values, particularly in
the first 28 days. An increased AFm formation was also observed during
this period, despite estimating only a minimal augmentation (2-5 %) in the
slag hydration degree using XCT. These overall encouraging results could
support the need to perform more trials in order to better optimize the mix
design and maximize the benefits from the addition of the paper residue.
136 ASTOVEZA 2022

____________________________________________________ Chapter 4
Chapter 4
Fe-rich slag addition in ternary CAC-
HH-PC binder
Contents
Introduction 139
4.1 Physical properties 140

4.1.1. Low calcium sulfate system 141
4.1.2. High calcium sulfate system 143
4.2 Quantifying the degree of slag hydration using XCT ....... 147
4.3 Phase assemblage evolution over time 151
4.3.1. Thermodynamic modelling 151
4.3.2. Early hydration reactions 155
4.3.3. Long term hydration 157
4.4 The fate of Fe during hydration 165
4.4.1. TEM 166
4.4.2. STEM 170
4.4.3. Mössbauer spectroscopy 175
Conclusion . . 179
ASTOVEZA 2022 137

____________________________________________________ Chapter 4
Introduction
In this chapter, the reactivity of the Fe-rich non-ferrous metallurgy (NFM)
slag, selected during the screening test in Chapter 2, was investigated in
ternary calcium aluminate cement (CAC)-calcium sulfate hemihydrate
(HH)-Portland cement (PC) binders. The incorporation of PC in these
CAC-rich systems serves two main purposes:
To emulate a system (CAC-HH-PC) closer to the industrial
formulations of CAC-based blends for rapid repair applications
To evaluate the potential of PC as an activator of the slag
Although there have been several studies on the ternary CAC-HH-PC
binders, detailed in Chapter 1, these preceding investigations were limited
to PC-rich (60-80 wt%) systems. By introducing the slag in formulations
dominated by CAC and HH, the lower carbon footprint and special
properties (i.e., self-drying capacity, rapid hardening) of such binders are
better maximized. Moreover, PC is evaluated as a possible activator of the
slag in these ternary binders. In Chapter 3, the addition of 5 wt% paper
residue slightly increased the rate of slag dissolution while significantly
enhancing the mechanical strength of the mortars. However, due to the
exceptionally elevated free lime content (23 wt%) of this residue, higher
incorporation was not recommended. On the contrary, PC can provide
additional Ca2+ and increase the alkalinity of the hydrated system without
significantly contributing to the free lime content of the raw blend.
The properties of formulations A-PC and B-PC (Table 2.3) were examined
in this chapter. Both of these formulations contained 30 wt% slag and 30
wt% PC. The CAC/HH ratio in formulation A-PC was 4.5 just like in
formulation A. The CAC/HH ratio in formulation B-PC was 1.6 adapted
from formulation B. Several techniques used in Chapter 3 were similarly
employed in this chapter to characterize the hydrated blends over time.
Mechanical strength and dimensional stability were monitored using
mortar samples. The early hydration reactions were compared using
isothermal calorimetry. The degree of slag hydration was estimated using
X-ray Computed Tomography (XCT). The phase assemblage evolution
was followed using XRD, TGA and ATR-FTIR of paste samples. Finally,
Mössbauer spectroscopy, TEM and STEM were used to describe the fate
of Fe during the hydration.
ASTOVEZA 2022 139

Chapter 4 ____________________________________________________
4.1. Physical properties

The physical properties of two ternary blends representing a low
(formulation A-PC) and a high (formulation B-PC) calcium sulfate system
are presented in this section. The addition of PC resulted in a shorter setting
time preventing the freshly-mixed blends to be properly placed in the
moulds. For this reason, 0.3 wt% citric acid was added in A-PC, B-PC and
their references. This is the similar citric acid dosage used in A-TC and B
in the previous chapter. Figure 4.1 shows the initial and final setting times
obtained from these formulations.
Figure 4.1 Initial and final setting time of formulation A-PC, B-PC and their references
Similar to the observation from the previous formulations (A, B and A-

TC), the addition of the slag resulted in a minimal delay (1-6 min) in the
setting time of mortars. Considering that the slag was unreactive during the
first hour of hydration (assumed based on the results from Chapter 3), it is
likely that the higher effective w/b ratio was one of the main factors
causing this prolonged setting consistent with the observations in Stefanou
and Larsinos, 1981 [196] and Dodson, 1994 [197]. Although this agrees
with the general trends observed with the addition of SCM at low (<
40 wt%) replacement levels [112], Juenger et al., 2008 [198] performed a
number of elaborated studies showing conflicting trends depending on slag
composition, fineness, curing conditions and replacement level. The
changes in setting time observed were also almost negligible compared to
the 20-30 min differences reported in the aforementioned studies.
Nonetheless, this very minimal delay in setting resulting from the slag
addition could have a positive implication in minimizing the amount of
citric acid needed in order to obtain a comparable setting time as the
reference cement mortars.
140 ASTOVEZA 2022

____________________________________________________ Chapter 4
4.1.1. Low calcium sulfate system

The compressive and flexural strength evolution of formulation A-PC and
its references are presented in Figure 4.2 and Figure 4.3, respectively. The
evolution of SAI values are shown in the secondary axes.
Figure 4.2 Compressive strength of formulation A-PC (primary axis)

and its SAI evolution (secondary axis)
Figure 4.3 Flexural strength of formulation A-PC (primary axis)

Overall, the strength was improved relative to formulation A with no PC

(section 3.1.1) especially at 28 days. This was even more evident with the
SAIcement values exceeding 1 as hydration proceeds. It implies that the slag-
containing mortars were able to surpass the strength values of their
reference cement mortars. In Figure 4.2, the increasing difference between
the SAIcement values of the slag-containing formulation and that of the
reference filler clearly indicates that the incorporation of slag positively
contributed to the strength development of mortars. In addition, SAIfiller
was plotted in the same figures to confirm whether the SAIcement reflected
ASTOVEZA 2022 141

Chapter 4 ____________________________________________________
the trends correctly or the values were overestimated due to a possible

excessive expansion in the reference cement mortars leading to cracks and
compromised strength values. Contrary to the minimal length variations
observed earlier in the binary CAC-HH binders, more significant
expansion was detected in the ternary systems. Figure 4.4 shows the length
variations in A-PC obtained from Walter+Bai shrinkage test (a) and water-
cured standard mortar samples (b).
Figure 4.4 Length variation of formulation A-PC and its references

obtained from Walter+Bai shrinkage (a) and water-cured mortars (b)
142 ASTOVEZA 2022

____________________________________________________ Chapter 4
After a high and rapid expansion during the first hour, the momentary
period of shrinkage between the second and third hour in the Walter+Bai
test (Figure 4.4 a) could be an initial indication of ettringite to
monosulfoaluminate conversion as the sulfates were depleted in the
system. This was however simultaneous to other factors affecting the
dimensional changes at early hours such as self-desiccation and drying
shrinkage.
Despite the continuous expansion of the blends during the 2 months of
curing under water (reaching up to 2.3 mm/m for the reference cement),
the mortars appeared to be generally stable and exhibited strength
development. The slag addition seemed to have improved the dimensional
stability of the blend by mitigating the expansion, consistent with its
behavior in the binary systems. With no visible macro cracks signaling
excessive expansion in this low calcium sulfate system (see Figure 4.8),
the SAIcement values were deemed to be acceptable. Nonetheless, the
increasing SAIfiller values confirmed the positive contribution of the slag
addition to the strength, as a reactive, rather than an inert, component.
4.1.2. High calcium sulfate system

In Chapter 3, increasing the HH content in a binary CAC-HH system up to
1.6 CAC/HH ratio, resulted in significantly higher strength values of
mortars. Nonetheless, this was explained by the increase in ettringite
formation predominantly from CAC and HH reaction, rather a
manifestation of increased slag reactivity.
With the PC addition in a high calcium sulfate system (B-PC), the strength
values (Figure 4.5 and Figure 4.6) obtained were rather comparable to
those in the low-sulfate system, A-PC. However, excessive expansion
(Figure 4.7) was observed specifically in the reference formulations
(reference filler and reference cement) resulting in poor strength
development. For this reason, there were no strength data points beyond
28 days as the mortars no longer fitted in the extensometer. In Figure 4.8,
actual images of A-PC and B-PC mortars with their respective references
are shown after 45 days of curing under water.
Due to the excessive expansion in the reference mortars, the resulting
SAIcement values were disproportionate and therefore did not represent the
slag contribution to the strength as intended. In this case, SAIfiller was a
better criterion to delineate the influence of the slag as a reactive SCM. At
least in the first 28 days, no macro cracks were observed in the reference
mortars. Once again, the results highlight that the addition of the slag
stabilized the dimensional variation in these blends, consistent with the
previous observations for formulations A, B, A-TC and A-PC. The
ASTOVEZA 2022 143

Chapter 4 ____________________________________________________
mitigation of the destructive expansion is evident in the rapid setting and

hardening properties of B-PC reaching > 20 MPa just after 1 day of curing
and as much as 49 MPa after 28 days.
Figure 4.5 Compressive strength of formulation B-PC (primary axis)

Figure 4.6 Flexural strength of formulation B-PC (primary axis)

Furthermore, the momentary period of shrinkage between 2 and 3 h

detected in A-PC (Figure 4.4 a) was not observed in B-PC. It is most likely
that the high amount of sulfates in the latter system resulted in the
formation of stable ettringite, contrary to those converting to
monosulfoaluminate in the former. It can also be observed in B-PC (Figure
4.7 a) that the rate of expansion was slower compared to what was
observed in the reference cement with a steep length variation curve in the
first 3 h of hydration. It follows that the addition of the slag did not only
mitigate the overall length variation at long term, but also reduced the
kinetics of expansion in air-curing conditions.
144 ASTOVEZA 2022

____________________________________________________ Chapter 4
Less difference in the trends was observed for underwater curing condition
(Figure 4.7 b), where the shape of the curves remained consistent even at
varying expansion degrees. Relative to what was observed from the
reference fillers, it can be said that the stability of the slag-containing
mortars was not exclusively due to binder dilution (less CAC and HH
proportion in the mortars). Instead, the dissolution of the slag likely had an
impact on the phase assemblage and/or the microstructure, manifesting
through the dimensional stability.
Macro cracks started to become apparent after 45 days for the reference
mortars cured under water, where the length variations surpassed
13.7 mm/m (Figure 4.7 b). It is important to mention that despite the
absence of macro cracks and the superior mechanical strength performance
of the slag-containing mortars, the expansion observed in this formulation,
reaching 8 mm/m, was far beyond the standard expansion limits reported
for sulfated PC and CSA systems typically between 0.04 % (0.4 mm/m)
and 0.5 % (5 mm/m), respectively [85,199]. Nonetheless, this degree of
expansion was still below the reported crack threshold limit of 2.25 % [85]
in binary CAC-HH systems, demonstrating the exceptional property of
these CAC-rich blends as expansive binders. Considering that there was
no extensive mix design optimization (i.e., to determine the ideal w/b ratio
and the proportion of each component) and the admixture used was limited
to citric acid, the rapid setting and hardening properties and the expansive
nature of the slag-containing formulations could be considered as
remarkable properties for fast repair applications.
ASTOVEZA 2022 145

Chapter 4 ____________________________________________________
Figure 4.7 Length variation of formulation B-PC and its references

obtained from Walter+Bai shrinkage (a) and water-cured mortars (b)
146 ASTOVEZA 2022

____________________________________________________ Chapter 4
Figure 4.8 Pictures of A-PC and B-PC mortars with their references
after 45 days of curing under water
4.2. Quantifying the degree of slag hydration using XCT

The degree of slag hydration of formulations A-PC and B-PC were
estimated using a combination of XCT and volume analysis as previously
employed in section 3.2. Figure 4.9 illustrates the main operations
involved in the analysis including: (a) 2D reconstruction of the XCT scans;
(b) threshold segmentation to extricate the unreacted slag particles; (c)
creation of 2D binary images based on the threshold segmentation; and (d)
generation of 3D volume renders interpolated from 1400 frames of 2D
binary images per sample.
ASTOVEZA 2022 147

Chapter 4 ____________________________________________________
Figure 4.9 Illustration of the steps involved in the volume analysis of XCT scans
using samples after 28 days of hydration
148 ASTOVEZA 2022

____________________________________________________ Chapter 4
The plot profiles (b) shown in Figure 4.9 correspond to the 16-bit
grayscale indices of the pixels under the red arrow drawn in their
corresponding 2D reconstructed images (a). The same lower threshold
index limit was used as in the samples from the binary CAC-HH binders
in the previous section. The parameters of the scans and the image analyses
were likewise kept constant for all the samples in this study. It is important
to reiterate that the estimation uses the total volume of the unreacted slag
particles at 1 day as the zero reactivity reference (see section 2.2.4) such
that any slag dissolution occurring before this period were assumed to be
negligible and consequently not accounted in the estimations.
Similar to the scans for formulations A, B and A-TC, high grayscale
contrasts were observed between the slag particles and the rest of the
cement and hydrate matrix even with the addition of PC. It was evident
that the slightly higher density of PC (3.11 g/cm3) compared to that of the
CAC (3.02 g/cm3
(1.4 % Fe2O3 for PC versus 1.8 % Fe2O3 for CAC) and thus yielding a
grayscale index comparable to that of the latter. The combination of these
factors preserved the advantage of the distinct properties of the slag in
using XCT to estimate the degree of slag hydration.
Figure 4.10 presents the estimated degree of slag hydration in
formulations A-PC and B-PC along with those in formulations A, B and
A-TC discussed in the preceding chapter. Note that the estimation was only
performed after 28 days for formulation A-PC and thus the absence of
some data points in the figure.
Figure 4.10 Estimated slag volume and hydration degree using XCT + volume analysis.
ASTOVEZA 2022 149

Chapter 4 ____________________________________________________
Overall, higher degree of slag hydration was estimated for the formulations
incorporating PC as illustrated in Figure 4.10. In the first 28 days, the XCT
results suggest a considerable increase from 9 % (B) to 17 % (B-PC) of the
amount of the dissolved slag with the addition of 30 wt% PC while keeping
the CAC/HH ratio constant at 1.6. Although this increase was less
significant between formulations A and A-PC (4.5 CAC/HH ratio), the
seemingly improved slag reactivity might be considered as a good
direction in improving the 28-day performance of the binders to address
the goal of early-age slag activation. The mechanism could be similar to
what was observed for the paper residue in section 3.5, with the PC
increasing the pH and contributing ions, particularly Ca2+, which
participate in hydrates formation. However, the increased reactivity
became less pronounced beyond 28 days with the differences approaching
50%) towards 1 year.
It is also critical to take into account the possible consequence of the
observed expansion in these PC-containing binders. Although the paste
samples were not subjected to underwater curing, unlike the mortars, the
w/b ratio used was higher (0.65 vs 0.50) for the pastes, failing to exclude
the possibility of expansive behavior. Substantial expansion occurring
after the first day of hydration could result in overestimation of slag
hydration degree as the unreacted slag particles initially located close to
the borders of the assigned scan volume may be slightly displaced out of
the frames. In effect, less unreacted slag volume is estimated not because
of the increased slag reactivity but because of the expansion. In order to
address this possible issue in future analyses of expansive binders, it is
recommended to run a full scan of a smaller sample, rather than setting
fixed dimensions of scan volume which is more ideal for bigger samples
with irregular dimensions.
Nonetheless, the impact of paper residue addition (where shrinkage is
observed) to the estimated slag hydration degree is a clear indication that
the increase in lime content and pH were beneficial in marginally
improving the early-age reactivity of slag according to the XCT
estimations. A similar case is expected with the PC addition despite any
possible contribution of the expansion to the estimated values.
Furthermore, higher slag reactivity was noted in the ternary formulation
with higher sulfate content (B-PC) in the first 28 days of hydration. This is
in contrary with the overall trend observed in the binary systems (A and
B). With these observations, the phase assemblage evolution of both the
systems are discussed in the subsequent section in order to further describe
the possible impact of the blend proportions to the reactivity of the slag.
150 ASTOVEZA 2022

____________________________________________________ Chapter 4
4.3. Phase assemblage evolution over time
4.3.1. Thermodynamic modelling

The thermodynamic models of the ternary CAC-HH-PC systems without
and with the 30 wt% slag addition are presented in Figure 4.11 and Figure
4.12, respectively. The models were generated in function of CAC content
(x-axis), which could be interpreted in relation to the CAC/HH ratio. Only
50 wt% of the amorphous fraction of the slag was assumed to hydrate
based on the estimations derived from XCT. The rest of the unreacted slag
(<1 wt%) components of CAC, HH and PC. More details on the

assumptions and the codes used to generate the models are detailed in
Appendix I.
According to the models (Figure 4.11 and Figure 4.12), gypsum is formed
with the excessive HH at lower CAC proportions: <21 % for the reference,
and <11 % for the slag-containing sample. In the reference formulation
(Figure 4.11), the model suggests a phase assemblage dominated by
strätlingite and monosulfoaluminate for the low-sulfate system, A-PC; and
an assemblage dominated by strätlingite and ettringite for the high-sulfate
system, B-PC. Comparing this reference formulation model with that of
the binary systems in section 3.3.1, it can be said that the addition of PC
results in the increased quantity of strätlingite. Furthermore, the model
suggests that a minor amount of C-S-H is only formed in B-PC.
Despite the addition of 30 wt% PC in the modelled system, the typical
hydration product of pure PC, portlandite (CH), was not predicted to form
in regions with more than 15 wt% CAC. This is because CH and gibbsite
cannot co-exist thermodynamically and thus favoring their conversion into
other hydrate products, as demonstrated in Bizzozero and Scrivener, 2014
[85]. Some of the possible reactions are presented in Equation 4.1-4.3.
ASTOVEZA 2022 151

Chapter 4 ____________________________________________________
Figure 4.11 Phase assemblage of the reference (no slag) ternary CAC-HH-PC binder
with varying CAC and HH proportions calculated from thermodynamic modelling
Figure 4.12 Phase assemblage of the ternary binder incorporating 30 wt% slag
and 30 wt% PC with varying CAC and HH proportions calculated from thermodynamic
modelling. Only 50 wt% of the amorphous fraction of the slag was assumed to hydrate.
152 ASTOVEZA 2022

____________________________________________________ Chapter 4
In Equation 4.1, CH and gibbsite react with C$Hx to form additional

ettringite. This case is likely to occur if all the anhydrous components
simultaneously hydrate, therefore not considering the actual kinetics of the
reactions. The late reaction of PC relative to CAC and HH is however
expected as discussed in the previous section, such that it is more likely
that the sulfates are already consumed before the reaction indicated could
take place. Instead the reactions in Equation 4.2 and Equation 4.3,
suggesting the formation of monosulfoaluminates and strätlingite are more
realistic, experimentally.
Equation 4.1
Equation 4.2
Equation 4.3
where x is equal to 2 for gypsum, 0.5 for hemihydrate and 0 for anhydrite
The models suggest that the addition of slag favors C-S-H formation over
that of ettringite and strätlingite, throughout a wide range of CAC/HH ratio
(Figure 4.12). This is even more evident in the high-sulfate formulation
B-PC for which the model displays a significant proportion of C-S-H along
with monosulfoaluminates (Ms) and strätlingite as dominant phases. In
the A-PC model, on the other hand, the assemblage is dominated by
strätlingite, Ms, and Fe-containing siliceous hydrogarnet. In both
formulations, Fe is incorporated in siliceous hydrogarnet and in an Fe-
containing Ms (Fe-Ms).
- -
4A$H12
1 wt%) Fe-Ms, C4(A,F)$H12. In the latter, the solid solution allows a
variation of the Al/(Al+Fe) ratio between 0 and 0.45. The stability of this
phase is described in detail in Dilnesa et al., 2011 [132]. As this phase was
not suggested to form in the previous models for the binary CAC-HH
systems (Chapter 3), its formation is likely associated to the increased in
Ms (Equation 4.2) and strätlingite (Equation 4.3) proportions resulting
from the instability of CH and gibbsite with the addition PC. In addition,
minor amounts of calcite and O-H hydrotalcite were consistently present
throughout the range to accommodate the C and Mg, respectively, coming
from the PC.
ASTOVEZA 2022 153

Chapter 4 ____________________________________________________
Figure 4.13 Composition by volume of the reference (no slag) ternary binder
with varying CAC and HH proportions simulated from GEMS.
Figure 4.14 Composition by volume of the ternary binder incorporating 30 wt% slag
and 30 wt% PC with varying CAC and HH proportions calculated from GEMS.
154 ASTOVEZA 2022

____________________________________________________ Chapter 4
In terms of volumetric composition (Figure 4.14), the maximum total solid

volume coincides with the peak of ettringite formation at around 11 wt%
CAC 29 wt% HH. This is explained by the expansive behavior of the
low-density hydrate ettringite, previously described in section 1.2.2. The
huge difference between the total solid volume of the reference (Figure
4.13) and the slag-containing (Figure 4.14) formulations, can explain the
excessive expansion observed for the reference mortars in section 4.1.
On the other hand, there is only a slight difference in the total solid volume
between formulations A-PC and B-PC, with the former being slightly more
expansive according to the model. This is in contrary to the more extensive
expansion observed experimentally in formulation B-PC. This suggests
that the model does not entirely reflect the actual phase assemblage
obtained experimentally. This could be similar to the observed deviation
in the models for formulation A and B in Chapter 3. It is important to
reiterate that the GEMS models presented here do not consider the kinetics
of hydration. In particular, the phase assemblage shown are the
thermodynamically stable phases in equilibrium state (given an infinite
reaction time). For this reason, the possibility of eventual conversion, after
long hydration ages, into more closely resembling these modelled phase
assemblages could not be excluded.
4.3.2. Early hydration reactions

The heat flow and cumulative heat release curves of the two ternary binders
are shown in Figure 4.15 and Figure 4.16, respectively.
Figure 4.15 Heat flow measured ex-situ using isothermal calorimetry for A-PC
(solid black curve) and B-PC (dotted blue curve) pastes
ASTOVEZA 2022 155

Chapter 4 ____________________________________________________
Figure 4.16 Cumulative heat release measured using isothermal calorimetry
With the addition of PC in the system, it became more difficult to associate

each exothermic peak to a single hydration reaction, as opposed to
formulations A and B in the preceding chapter. Nevertheless, three main
peaks are recognizable in both formulations which likely correspond to the
main peaks identified in CAC-HH-slag binders in section 3.3.
The fast dissolution of several anhydrous phases, principally the HH, in
the first two hours of hydration is marked by the first exothermic peaks in
both formulations. During the first six hours, extensive ettringite formation
is anticipated in these systems supported by the expansive property
previously observed in mortar samples. Consequently, it can be deduced
that the second exothermic peak is primarily linked to ettringite formation.
Afterwards, the formation of AFm phases follows as the sulfates are
depleted, particularly in A-PC, possibly constituting the third exothermic
peak. Approximately during the same period, the PC components react in
parallel, liberating Ca2+ ions, subsequently raising the pH of the pore
solution. Silicates from the PC also contribute to the formation of hydrate
phases, C-S-H and strätlingite, later in the hydration.
In similar ternary binders, CAC hydration is believed to delay the
hydration of PC through the mechanism of ettringite barrier formation
[86]. It is proposed that ettringite, resulting from the reaction of CA and
calcium sulfate, forms a film-like layer around the anhydrous grains
provisionally delaying silicates (mainly C3S and C2S) hydration. This
retardation in the diffusion process could be associated to the momentary
plateaus marked (1 and 2) in the cumulative heat release curves (Figure
4.16). The reaction occurs in multiple stages and the heat release is
resumed when the barrier layers are broken with the destabilization of
ettringite.
156 ASTOVEZA 2022

____________________________________________________ Chapter 4
Despite the differences in the heat release curves (Figure 4.15) between
formulations A-PC and B-PC, their cumulative heat release beyond 40
hours remain comparable. This observation is consistent with the trends
observed between the low (A) and high (B) sulfate systems in the binary
blends in Chapter 3.
4.3.3. Long term hydration

The addition of the PC in the system clearly modified the phase assemblage
formed after hydration looking at the XRD, TGA and ATR-FTIR results
presented in this section. The interpretation of the trends was guided by the
characterization of pure PC paste, hydrated in the same conditions as the
A-PC and B-PC samples, with results presented in Appendix K.
Consistent with thermodynamic modelling, strätlingite appears to take a
major proportion of formulation A- ith
well- Figure 4.17). The
signature of Si-O-Al asymmetric stretching band at 710 cm-1 (Figure 4.20)
associated with strätlingite is also exclusively present in this formulation
affirming the significant difference in the quantity of strätlingite between
the low- and high-sulfate blends. Although barely noticeable at a quick
glance, changes in the phase assemblage are generally more pronounced
between 7 and 28 days for formulation A-PC. These include the variations
-11° along with calcite formation
(29.5°) in XRD (Figure 4.17); and the increasing amount of C-S-H as a
bump below 100°C forms in the DTG profile (Figure 4.19) between 7 and
28 days. The increase in AFm phases could be related to the conversion of
ettringite to monosulfoaluminate as the sulfate is depleted in this low-
sulfate blend. On the other hand, the delayed hydration of PC previously
suggested by the cumulative heat release curve could explain the increase
in C-S-H at later ages. The sharp peak majorly associated to calcite could
be a result of carbonation during sample preparation.
ASTOVEZA 2022 157

Chapter 4 ____________________________________________________
Figure 4.17 XRD patterns of formulation A-PC (solid black lines) and its reference.
Phases are abbreviated as follows: (Str) strätlingite, (Ett) ettringite, (Ms)
monosulfoaluminate, (Geh) gehlenite, (Cc) calcite, (CSH) C-S-H, (CH) Portlandite,
(Hdg) hydrogarnet and (Wue) wüstite. Note that some of the phases are indicated only to
mark the position of their main peaks, despite their absence in the phase assemblage
Despite the variations mentioned above, there was no drastic change in the
phase assemblage nor was there an appearance of new peaks that could be
directly associated to the hydration products of the slag in A-PC. Only
14 % of the slag has reacted after 28 days according to the XCT results
(Figure 4.10). This is likewise reflected through the stable ATR-FTIR
spectra over time (Figure 4.20) for this formulation. On the contrary, the
changes in formulation B-PC, which contains higher amount of sulfates,
are more apparent particularly between 1 and 7 days. The actual phase
assemblage also appears to be ettringite-rich, in contrary to the
thermodynamically-predicted system rich in monosulfoaluminates. This
could be explained by the lower slag reaction in the real system yielding
an assemblage closer to what was observed for the reference formulation.
Nonetheless, the model was able to predict the instability of CH and
gibbsite in these systems.
Strong ettringite peaks from the XRD patterns (Figure 4.18) alone are
clearly evident, primarily resulting from the fast hydration of CA (from the
CAC) and calcium sulfate (from the HH). The ettringite formed appears to
158 ASTOVEZA 2022

____________________________________________________ Chapter 4
be generally stable throughout the curing period. Strätlingite forms

between 1 and 7 days marked by the appearance of XRD peaks at 7° and
21.2° for the slag-containing formulation. In the same period, drastic
changes in the XRD pattern between 28°-30° where several silicate phases
coincide, is noted. These changes could be relevant to slag reactivity
considering the findings in section 3.3, indicating the effect of the added
silicates from the slag to the formation of strätlingite and subsequently, the
reduced consumption of gehlenite from the CAC.
Figure 4.18 XRD patterns of formulation B-PC (solid blue lines) and its reference.
Phases are abbreviated as follows: (Ett) ettringite, (AFm) AFm, (Ms)
monosulfoaluminate, (Gib) gibbsite, (Geh) gehlenite, (Gyp) gypsum and (Wue) wüstite.
Similar to the observation in formulation A-PC, the formation of C-S-H in

B-PC is indicated by the peak shift in DTG (Figure 4.19) resulting to
-S-H (90-140°C) peaks at 165 days. Note
that this distinction was guided by the DTG curves obtained from pure PC
systems presented in Appendix K which delineates the ettringite and C-S-
H peaks, accordingly.
ASTOVEZA 2022 159

Chapter 4 ____________________________________________________
Figure 4.19 DTG curves of (a) formulation A-PC and (b) formulation B-PC
along with their corresponding references. Slag-containing formulations are shown in
solid lines. Phases are abbreviated as follows: (Ett) ettringite, (CSH) C-S-H, (AFm) AFm,
(Str) strätlingite, (Gib) gibbsite, (Hdg) hydrogarnet
Figure 4.20 ATR-FTIR spectra of formulation A-PC (solid black lines) and its reference.
phases identified.
160 ASTOVEZA 2022

____________________________________________________ Chapter 4
The ATR-FTIR spectra of formulation B-PC (Figure 4.21) are dominated

by the asymmetric stretching band of S-O for ettringite (and minimally for
C-S-H) at 1115 cm-1. Moreover, an additional contributor to the 965 cm-1
peak for both the PC-containing formulations is the Si-O-Si stretching
vibration band assigned to C-S-H. This coincides with the previously
observed Al-O vibration band in the binary systems. Overall, the phase
assemblages deduced from these spectra are generally consistent with
those identified according to the XRD and DTG patterns, as follows:
A-PC: dominated by strätlingite and monosulfate; with minor
amount of C-S-H
B-PC: dominated by ettringite; with strätlingite content increasing
over time; minor C-S-H content
Figure 4.21 ATR-FTIR spectra of formulation B-PC (solid blue lines) and its reference.
phases identified.
Table 4.1 FTIR band assignments based on the distinct spectral peaks observed
Primary Wavenumber
Bond Reference
contributor (cm-1)
O-H Ms/Ett/Str 3432 Horgnies et al., 2013 [137]
1660 Horgnies et al., 2013 [137]
S-O Ett/Ms 1115 Horgnies et al., 2013 [137]
Si-O-Al Str 1150 Okoronkwo and Glasser , 2016 [165]
710 Zapanta et al., 2020 [171]
ASTOVEZA 2022 161

Chapter 4 ____________________________________________________
Si-O C-S-H 965 Guan et al., 2013 [200]

anhydrous silicates 521 Torréns-Martin et al., 2013 [86]
Al-O Gib 1021 Schroeder, 2002 [172]
972 Schroeder, 2002 [172]
751 Schroeder, 2002 [172]
Str 965 Okoronkwo and Glasser, 2016 [165]
Hdg 810 Horgnies et al., 2013 [137]
Ett 857 Horgnies et al., 2013 [137]
Gib/Ms/Str/Hdg 523-537 Horgnies et al., 2013 [137]
Hdg 521 Torréns-Martin et al., 2013 [86]
Having estimated the highest slag hydration degree using XCT for B-PC
as against all the other formulations studied, the ATR-FTIR spectrum of
the raw slag is superimposed to that of formulation B-PC at 1 day in Figure
4.21
that of the B-PC in order to estimate the possible contribution of the
unreacted slag to the peak intensities of the hydrated blends. The
unadjusted spectrum of the raw slag was previously presented in Figure
2.13. Notice that the Si-O asymmetric stretching band of the raw slag at
around 930 cm-1 coincided with the region (between 857 and 965 cm-1)
where the difference between the spectra of the reference and the slag-
containing formulations are observed at 1 day. Assuming that the slag has
barely reacted at this period, it could be assumed that the raw slag is
contributing to the spectrum of the blend in a similar manner to the one
illustrated.
In section 1.4.3, it was highlighted that an evidence of slag reactivity was
depicted in Siakati et al., 2020 [117] principally through peak shift of the
Si-O asymmetric stretching band (700-1200 cm-1) towards a higher
wavenumber. The disappearance of the gaps (red double arrows) in the
superimposed spectra in Figure 4.22 could be interpreted as a possible
consequence of a similar peak displacement towards the higher
wavenumber, causing the unreacted slag peaks to be concealed in the
adjacent areas. Although this could have been accompanied by a shift of
the Si-O rocking band (400-500 cm-1) towards a lower wavenumber, it
could not be clearly distinguished due to the very weak intensity of the
second peak after the dilution in the blend.
162 ASTOVEZA 2022

____________________________________________________ Chapter 4
Figure 4.22 Superimposed ATR-FTIR spectra of the raw slag (red) to those of the B-PC
at different curing times. It aims to highlight how the gaps formed over time (red double
arrows), could be an evidence of the proposed peak-displacement scenario signifying slag
hydration.
Figure 4.23 -FTIR spectrum superimposed to that of the B-PC

to demonstrate a possible configuration of reactivity-related peak shift
ASTOVEZA 2022 163

Chapter 4 ____________________________________________________
A hypothetical scenario demonstrating the possible peak shift is illustrated

in Figure 4.23 for formulation B-PC where the highest degree of slag
hydration after 28 days was observed. Between 1 and 7 days, a 60 cm-1
shift towards a higher wavenumber is presumed to allow the slag spectra
to fit into that of the blend. This fitting could be more relevant if the
throughout the hydration period considering its expected poor reactivity

during the first 7 days.
To reiterate, such a peak shift has been previously linked [116,117] to the
partial oxidation of Fe2+ to Fe3+ accompanying the change in the overall
atomic environment of Fe during polymerization. For the cement blends
investigated in this study, this could correspond to the formation of Fe-
containing hydrates, such as those suggested by the thermodynamic
models or those previously detected in Chapter 3. Similar to the binary
systems (formulations A and B), most of the discernible changes were
concentrated in the Si-O and Al-O regions particularly between 857 cm-1
and 1021 cm-1 coincident with the dominant peak of the raw slag. This
hydrate phase assemblage.

Although the evolution of the phase assemblages over time observed from
the ATR-FTIR followed the trend of those from XRD and TGA, it can be
said that the results obtained did not particularly contribute to a clearer
for the ATR-FTIR results obtained in Chapter 3. The demonstrated peak

shift in Figure 4.23 remains hypothetical and further tests ideally using
less complex cement systems or even synthetic systems are recommended
to confirm the proposed fitting. Nevertheless, this technique could be
considered as an alternative or exploratory step to XRD and TGA, for a
rapid (2 min) non-destructive testing when a limited (0.05-0.1 g) amount
of sample is available.
Lastly, the total bound water contents of the ternary binders, A-PC and B-
PC, were plotted together with their respective references in Figure 4.24.
Overall, the higher bound water contents of the reference formulations
confirm the prediction from the thermodynamic models (section 4.3.1)
where a higher total mass of hydrates is estimated to form relative to the
slag-containing pastes. Furthermore, the late reaction of the slag (versus
the cement components) is reflected through the differences in the
increments of the values. The reference formulations appear to have
obtained their maximum hydration earlier on followed by only minimal
increase in the total bound water content. Whereas the slag-containing
formulations displayed more significant increments per curing time. The
drastic increase in the bound water content of formulation B-PC between
164 ASTOVEZA 2022
____________________________________________________ Chapter 4
28 days (34 g) and 165 days (47 g) is in agreement with the slag hydration
degree quantified though XCT, discussed earlier in section 4.2. For this
reason, a greater focus is dedicated on this formulation in the subsequent
sections for the investigation on the fate of Fe particularly between 28 to
165 days of curing.
Figure 4.24 Total bound water content of formulations A-PC, B-PC and their references
4.4. The fate of Fe during hydration

In this section, the fate of Fe, the main component of the slag, is
investigated through TEM, STEM and Mössbauer spectroscopy. These
advanced techniques proved to be essential in elucidating a better
understanding of the hydration mechanism of the slag in Chapter 3. With
the addition of the PC, the phase assemblages of the ternary (CAC-HH-
PC) systems have evidently changed compared to those of the binary
(CAC-HH) systems. PC contributed Ca and Si components leading to the
more extensive strätlingite formation and C-S-H generation, while
increasing the alkalinity of the system, and thereby influencing slag
dissolution.
Although the characterization techniques in the previous section were able
to describe the general phase assemblage of formulations A-PC and B-PC,
no clear indication was derived as to the fate of Fe upon slag dissolution.
The formation of Fe-containing hydrogarnet and/or monosulfoaluminates,
as predicted from thermodynamic modelling, could not be confirmed from
XRD, TGA nor FTIR alone due to the complexity of the systems with
overlapping signatures. Neither can the formation of similar Fe-containing
hydrated gels, as those observed in the binary systems in Chapter 3, could
be confirmed mainly due to their poor crystallinity. Analyses through
ASTOVEZA 2022 165

Chapter 4 ____________________________________________________
TEM, STEM and Mössbauer spectroscopy in the subsequent sections are

expected to provide more information on the nature of the hydrated phase/s
incorporating Fe.
4.4.1. TEM
The general phase assemblage of formulation A-PC is reflected in the TEM
images shown in Figure 4.25 and Figure 4.26. In these images, the
abundance of strätlingite is apparent with its distinct plate-like morphology
and siliceous composition marked in blue. Most of the particles are
agglomerated in Figure 4.25 a and b, making it difficult for them to be
analyzed individually.
Figure 4.25 TEM images and the atomic composition of marked areas in the (A-PC)
Figure 4.26 TEM images and the atomic composition of marked areas in the (A-PC)
4.5CAC-HH-30PC paste sample after 28 days of hydration
166 ASTOVEZA 2022

____________________________________________________ Chapter 4
Significant amounts of hydrated gels are also detected throughout the

samples exhibiting a combination of foil-like or fibrous (i.e., A-PC 1-3)
and granular (B-PC 1, 3 and 6) morphologies. Similar forms have been
previously observed for C-S-H morphologies according to the
investigation in Techer et al., 2012 [201]. Having detected Ca, Al, and Si
in these hydrated gels, it is convenient to associate them to C-S-H at first
glance. The selected electron diffraction pattern for B-PC 1 (Figure 4.27)
also confirmed the poorly-crystalline to amorphous nature of this phase
with the absence of clear diffraction spots. However, compared to C-S-H,
the Al estimated was significantly higher depicted in the ternary diagram
(Figure 4.32).
No significant amount of Fe was detected in the crystalline hydrate phases
such as in strätlingite (i.e., A-PC 4-6), nor in ettringite (i.e., A-PC 7-8) even
in the high calcium sulfate system (i.e., B-PC 14, Figure 4.28). This is
consistent with the findings from the binary system in Chapter 3. Less than
1 at% of Fe was estimated in all these phases (Figure 4.26 and Figure
4.28), which could well fall within the analytical sensitivity of the device
or could be influenced by the hydrated gels superimposed within the target
area. In addition, monosulfoaluminates and siliceous hydrogarnet
predicted to incorporate Fe based on thermodynamic modelling, could not
be clearly identified from the TEM results.
Instead, Fe was detected in the hydrated gels rich in Ca, Si and Al. The
addition of PC appeared to have altered the composition of these Fe-
containing hydrates as opposed to the Al-rich gels previously observed in
the binary systems. It is postulated that the kinetics of hydration hindered
the incorporation of Fe in the crystalline phases considering their relatively
faster formation and stability early in the hydration. Whereas Fe is released
slowly as the slag dissolves over time.
Some of the TEM images of formulation B-PC after 28 and 150 days of
curing are presented in Figure 4.27 and Figure 4.28, respectively. As
expected, less particles with plate-like morphology (encircled in blue)
mainly associated to strätlingite due their silicates content, were observed.
Instead, more ettringite rods (marked in orange) were present as a
consequence of higher amount of sulfate in this formulation. In the first 28
days of curing, the Fe content in the intermixed Ca-Al-Si rich hydrated gels
was mostly limited to < 4 at%. The diffraction pattern of B-PC 1
demonstrates the amorphous nature of the phase. Similar to the Al-rich gels
in the binary blends in Chapter 3, these hydrated gels exhibit inconsistent
thickness with certain regions appearing to be darker in the images than
the rest of the matrix (i.e., B-PC 3). For this reason, the possibility of nano-
crystals inclusion comparable to what was presented in the binary systems
ASTOVEZA 2022 167

Chapter 4 ____________________________________________________
could not be excluded. In addition, the Fe-containing gels also displayed

high sensitivity to the analysis and were easily disintegrated after the scans,
similar to what was demonstrated in section 3.4.1. This is critical to take
into consideration when interpreting the variation in the chemical
composition of the gels. When disintegrated, it is possible that certain
components have been volatilized resulting in their reduced atomic
percentage.
Figure 4.27 TEM images, diffraction pattern and the atomic composition of Fe-
containing gels in the (B-PC) 1.6CAC-HH-30PC paste sample after 28 days of hydration
168 ASTOVEZA 2022

____________________________________________________ Chapter 4
Figure 4.28 TEM images and the atomic composition of marked areas in the (B-PC)
1.6CAC-HH-30PC paste sample after 150 days of hydration
The Fe content after 150 days of curing was generally elevated as the
hydrated gels (B-PC 8-11) appear to be thicker in Figure 4.28. This was
accompanied by an increase in the atomic percentages of Ca and Si almost
twice as those obtained for the gels at 28 days. It is probable that the late
dissolution of PC and slag contributed to this increase. Since CAC was
reacting immediately in the early days of curing, the Al content of the gels
remained stable between 28 and 150 days.
Once again, no significant amount of Fe was detected in ettringite (B-PC
14) even after longer curing time. Unreacted wüstite was detected at 150
days (Figure 4.28 B-PC 13) exhibiting its distinct dendritic morphology.
This supports the earlier remark that only the amorphous fraction of the
slag is dissolving during the hydration. Adjacent to the Fe-rich wüstite (35
at% Fe in Figure 4.28 e), a hydrated gel (B-
distinguished with a highly contrasting Fe signal. This demonstrates the
precision of the EDS in analyzing different phases on a nano-scale.
Although the intermixing of phases due to the complex nature of the
ASTOVEZA 2022 169

Chapter 4 ____________________________________________________
samples could not be completely eliminated, TEM still yielded more

precise analysis compared to the SEM used in section 3.2.1.
4.4.2. STEM
In Chapter 3, the elemental maps obtained from STEM demonstrated the
non-uniform distribution of Fe in the hydrated gels. It has driven the
proposition that there are at least two Fe-containing hydrate phases in the
CAC-based systems: (1) an amorphous part with a lower Fe concentration;
(2) and an unidentified nano-crystalline phase comprising the darker
regions of the gels, with a higher Fe content. Although the TEM results in
the preceding section indicated a different composition of the gels in the
ternary binders, the inhomogeneity of these gels provided an initial
evidence of the possible presence of intermixed phases congruent to the
binary systems.
STEM bright field images of a generally homogenous gel (B-PC 15-16)

and an intermixed gel (B-PC 17-18) along with their corresponding
elemental maps are presented in Figure 4.29 and Figure 4.30,
respectively. The uniform distribution of Fe in gels with lower Fe content
(< 4 at%) is evident in the maps of Figure 4.29 a. This clearly confirms
that Fe is indeed incorporated in the gels instead of being simply measured
from a neighboring unreacted slag particle. On the contrary, Fe was clearly
more concentrated in the darker region of the gel in B-PC 17 compared to
the B-PC 18 region in Figure 4.30. The proportions of Ca, Si and Al
remained consistent despite the huge difference in the Fe content.
To further highlight the absence of Fe in the crystalline hydrates,

particularly in ettringite (the dominant hydrate phase in B-PC), Fe and S
maps of STEM bright field images containing rods of ettringite (B-PC 19
and B-PC 20) are shown in Figure 4.31. Both the maps and the EDS
analyses confirmed the absence of Fe in ettringite. The rapid formation of
ettringite during the first hours of hydration and its stability over time
driven by the abundance of sulfate in the system, were assumed to inhibit
the incorporation of Fe in this phase.
170 ASTOVEZA 2022

____________________________________________________ Chapter 4
Figure 4.29 STEM elemental mapping of a hydrated gel phase in the (B-PC)
1.6CAC-HH-30PC paste sample after 150 days of curing. In analyzing the STEM
elemental maps, note that the color intensities were independent for each element.
ASTOVEZA 2022 171

Chapter 4 ____________________________________________________
Figure 4.30 STEM elemental mapping for a selected area in the (B-PC)
1.6CAC-HH-30PC paste sample after 150 days of curing.
172 ASTOVEZA 2022

____________________________________________________ Chapter 4
Figure 4.31 STEM images and their corresponding Fe and S maps from (B-PC)
1.6CAC- HH-30PC paste sample after 150 days of curing.
The composition of the Fe-containing hydrated gels were plotted in a Ca-

Si-Al ternary diagram in Figure 4.32, following the similar conditions
applied to the binary system. Only the gels with 1-12 at% Fe were selected
in order to minimize the inclusion of analytical volumes where strong
intermixing of phases could influence the composition. Values were also
expressed in atomic percentages to reduce errors associated to O
quantification in TEM-EDS analyses [178].
The difference in composition of the hydrated gels between the binary (no
PC) and ternary (with PC) blends is clearly depicted in Figure 4.32, as
each system occupies a distinct cluster in the diagram. The hydrated gels
in the ternary blends contained considerably lower proportions of Al.
Instead, they were enriched in Ca and Si approaching the region between
the Si-hydrogarnet and the C-A-S-H region reported in Portland cement
systems [202]. As the hydration advanced from 28 to 150 days for
formulation B-PC, the TEM results alone appear to suggest the tendency
to move towards the region of lower Ca content in the ternary diagram. A
relevant trend on decreasing Ca/Si ratio in C-S-H has been previously
associated to the addition of siliceous SCM [11,203,204]. This trend could
be related to the possible incorporation of Fe in the interlayer of C-A-S-H
gels demonstrated in Mancini et al., 2020 [113] using 29Si NMR and XAS
ASTOVEZA 2022 173

Chapter 4 ____________________________________________________
spectroscopy in a co-precipitation experiment. As explained in their study,

both Fe2+ and Fe3+ could compete with Ca2+ ions octahedrally-coordinated
to the adjacent Si tetrahedra in the interlayer of C-S-H at high Ca/Si ratios
(1.2-1.5). This has been demonstrated despite the substantial difference in
the ionic radii of these cations (i.e., Ca2+ with r= 1.08-1.20 Å; Fe2+/3+ with
0.64-0.86 Å), exemplifying the remarkable stoichiometric and structural
variability of C-S-H [113,124,205].
Figure 4.32 Ternary diagram of Fe-containing hydrated gels with 1-12 at% Fe observed
using TEM and STEM. The values were calculated from the Ca, Si and Al atomic
composition from EDS analyses normalized to 100 at%. Reference points for, gibbsite,
strätlingite, katoite and Si-hydrogarnet were added in the diagram based on their
empirical formula while the C-A-S-H region was estimated based on the reported stable
composition in Portland cement system [202]. The area covered by the Al-rich gels in
However, additional STEM analyses plotted in Figure 4.32 suggest

instead that the composition of the amorphous gels were rather more
consistent over time with Ca/Si ratio between 1.2 and 2.2. The Al content
of the gels was also significantly higher compared to the previously
obtained [202,205] experimental composition of C-S-H. It is possible that
the deviation observed was influenced by the presence of the nano-
crystalline phase intermixed with the amorphous hydrated gel. In the end,
it is difficult to establish concrete trends as to the changes in the
composition of the Fe-bearing hydrated gels over time considering the
limited number of data points and the complexity of the system.
The challenge remains when the phases are superimposed or intermixed
within the same analytical volume. This would support the importance of
working with synthetic systems with less components as a complementary
characterization of the Fe-containing hydrated gels to be considered in
future works. Nonetheless, demonstrating the fate of Fe using the actual
174 ASTOVEZA 2022
____________________________________________________ Chapter 4
raw materials has contributed to a critical outlook on an industrial point of

view. It is possible that other factors, such as the presence of minor
impurities and the kinetics of hydration, could have an impact to the
hydration mechanism of the slag causing the difference in the fate of Fe
from the collated reports in Chapter 1 on synthetic and modelled systems,
versus that in the blends presented in this study.
4.4.3. Mössbauer spectroscopy

Mössbauer spectroscopy was used in the preceding chapter to describe the
changes in the oxidation state and the atomic environment of Fe during the
slag hydration in binary CAC-HH systems. In addition, a quantification of
hydration degree was likewise presented using a fitting model to
deconvolute the spectra.
The evolution of the spectra of B-PC paste after 1, 7 and 28 days of
hydration is presented in Figure 4.33. Note that the focus for further
analysis was dedicated to this formulation due to the indication of higher
degree of slag hydration from the previously presented results. The spectra
of the raw slag and PC are also shown in this figure for reference. Unlike
the slow transition observed in the binary systems (section 3.4.3) during
the first 28 days, the difference in the spectrum of the raw slag to that of
B-PC immediately after 1 day is already remarkable. This could support
the faster rate of slag dissolution proposed based on the other techniques
highlighting the activating effect of PC addition.
A comparable transition was only observed between 7 and 90 days for the
binary systems. The difference in the spectrum could also not be explained
simply by the contribution of the Fe from the PC. Not only that this
contribution is minimal (1.4 % Fe2O3 in pure PC corresponding to only
2.9 % of the total Fe in the blend) to begin with, but it is also clear from
Figure 4.33 that the positions of the PC peaks do not correspond to the
main peaks of the hydrated paste. Moreover, the spectrum of B-PC at 1
day closely resembles that of the binary blend (B) at 28 days (Figure 3.39),
while the B-PC spectrum at 28 days is similar to the binary blend (A)
spectrum at 90 days. This further supports the suggested faster kinetics of
slag hydration when the PC is added in the system.
The spectrum of the blend at each curing time was deconvoluted following
the fitting procedures for Model 1 in section 3.4.3. This model utilizes four
Fe components for the unreacted fraction of the blend and two Fe
components after hydration. The former consists of two Fe2+ doublets (G1
and G2) and one Fe3+ doublet (G3) in the glass fraction; and a doublet of
Fe2+ (M1) for wüstite. On the other hand, the reacted components include
a doublet of Fe3+ (P1) and Fe2+ (P2). In addition, a sextet magnetic
ASTOVEZA 2022 175

Chapter 4 ____________________________________________________
component (M2) was used only for the raw slag fitting. The modeled
spectrum of the B-PC paste after 28 days is shown in Figure 4.34.
Figure 4.33 57Fe Mössbauer spectrum evolution of formulation B-PC after 28 days
along with the spectra of the raw slag and the raw PC used in the formulations. Running
average per 10 data points was used to fit a curve for the PC spectrum (section 2.2.5) in
order to minimize the noisy signature resulting from its low (1.4% Fe2O3) Fe content.
176 ASTOVEZA 2022

____________________________________________________ Chapter 4
Figure 4.34 57Fe Mössbauer spectrum of formulation B-PC at 28 days and its
deconvoluted components
The absorption areas (AA) obtained from the fitted models at different
curing time are presented in Figure 4.35. Using the same equation
elaborated in Appendix J, the degree of slag hydration was estimated based
on the changes in the sum of the AA of the unreacted slag components.
The results of this estimation is presented in Figure 4.36.
Figure 4.35 Evolution of B-PC absorption areas (AA) calculated based on the models.
Values at 0 day corresponds to the fitting of the raw slag.
The results confirm the earlier remark on the resemblance of the B-PC
spectrum at 28 days (with 42 % slag hydration degree) to that of the A
spectrum at 90 days (with 48 % slag hydration degree). The slower
ASTOVEZA 2022 177

Chapter 4 ____________________________________________________
increase in the values between 1 and 28 days relative to the trend for the
binary systems in section 3.4.3 suggests that a similar extent of dissolution
was achieved faster in the ternary blends with PC. This is evident from the
steep slope between 0 and 1 day in Figure 4.36.
Figure 4.36 Slag hydration degree of formulation B-PC (primary axis) estimated based
on the changes in the total absorption areas (secondary axis) of the unreacted slag
Nonetheless, it is important to remark the huge difference in the slag
hydration degree values obtained from this estimation compared to those
from the XCT in section 4.2. As mentioned for binary systems,
underestimation from XCT could be linked to two main factors: (1) the
assumption that the slag is completely unreactive in the first 24 hours of
hydration; and (2) the 2.15 µm voxel resolution setting which excluded the
contribution of the finer slag particles. This could be addressed in the
future works by performing faster scans immediately after the paste
hardens and several others within the first 24 hours of curing. Sensitivity
analysis on the voxel resolution setting could also be performed in order to
quantify the influence of this factor to the estimation.
Finally, the components of the fitting procedure followed for the
Mössbauer spectroscopy could be further refined and adapted for the
ternary systems. Although Fe was detected in intermixed hydrated gels in
both the binary and ternary systems, the difference in their compositions
merits a re-evaluation of the fitted components, P1 and P2. Further
characterization preferably using synthetic systems with less components
will be a critical prerequisite in identifying the better adapted components
for the re-evaluation.
178 ASTOVEZA 2022

____________________________________________________ Chapter 4
Conclusion
This chapter demonstrated the reactivity of the slag in ternary CAC-HH-
PC systems. The binders exhibited an expansive nature, particularly those
of formulation B-PC which had the higher sulfate content. In comparison
to the reference mortars, the addition of the slag proved to be beneficial in
mitigating excessive expansion such that macro cracks and consequently
the decrease in mechanical strength were avoided even when the mortars
were cured under water. In addition, promising SAI values (>1) exceeding
those obtained in the binary system were achieved. High compressive
strengths at around 20 MPa were obtained at 1 day, which increased to
more than 40 MPa at 28 days for the slag-containing mortars. However,
the addition of the PC significantly accelerated the setting time prompting
the need for 0.3 wt% citric acid addition to provide sufficient time for the
mixtures to be placed in the moulds. This resulted in setting time ranging
from 20-40 min.
Estimation of the degree of slag hydration using Mössbauer spectroscopy
and XCT both supported the faster kinetics of slag hydration with the
addition of PC. The reactivity of the slag in the first 28 days of hydration
was reflected through the drastic changes in the Mössbauer spectra,
particularly of the high-sulfate formulation. On the other hand, the XCT
results showed that although the slag was more reactive for the ternary
binders in the first 28 days, the difference in the hydration degree was less
pronounced at a longer curing time. Despite the consistency in the trends
observed between these two techniques, however, significant differences
in the estimated values were linked to two main factors: (1) the assumption
that the slag is completely unreactive in the first 24 hours of hydration; and
(2) the 2.15 µm voxel resolution setting which excluded the contribution
of the finer particles.
The phase assemblages obtained experimentally based on the XRD, TGA
and ATR-FTIR results could be summarized as follows:
Formulation (A-PC) 4.5CAC-HH-30PC: dominated by

strätlingite and monosulfate; with minor amount of C-S-H
Formulation (B-PC) 1.6CAC-HH-30PC: dominated by ettringite;
with strätlingite amount increasing over time; minor C-S-H
content
According to the thermodynamic models, the addition of the slag should
favor a more extensive formation of monosulfates and C-S-H. However,
the experimental results demonstrated an extensive ettringite formation
instead, especially for the high-sulfate system. This difference could be
explained by the kinetics of the hydration, not considered in the models.
ASTOVEZA 2022 179

Chapter 4 ____________________________________________________
Ettringite is formed in the early ages of hydration and remains generally

stable throughout the 165 days covered by the study. Furthermore,
although the models suggest the incorporation of Fe in
monosulfoaluminates and siliceous hydrogarnet, there was no sufficient
evidence confirming the uptake of Fe in these phases.
Using TEM, no significant amount of Fe was detected in the more
conventional hydrated phases such as ettringite, monosulfoaluminate or
strätlingite, similar to the findings in the binary systems. Instead, Fe was
found to be incorporated in Ca-Al-Si-rich intermixed hydrated gel.
Elemental maps from STEM suggest that Fe could be in at least two
hydrate phases: (1) an amorphous gel with a lower (1-4 at%) Fe
concentration; (2) and an unidentified nano-crystalline phase comprising
the gels with higher (5-13 at%) Fe content. Nevertheless, the possibility
that the Fe-bearing intermixed phases could evolve into other hydrate
phases (i.e., Si-hydrogarnet, C-A-S-H) over time could not be excluded.
Although a huge gap still needs to be filled towards fully unravelling the
ts supported its potential
valorization as an SCM in an expansive CAC-HH-PC ternary binder. The
chapter also demonstrated that PC could moderately act as an activator of
the slag theoretically by increasing the alkalinity of the system and by
providing additional elements (Ca and Si) that participate in the formation
of Fe-bearing hydrates.
180 ASTOVEZA 2022

_____________________________________Conclusions & Perspectives
Conclusions and Perspectives

The EU MSCA-ETN SOCRATES Project promotes circular economy by
creating near-zero-waste flow sheets for the valorization of industrial by-
products. After the extraction of high-value components and the removal
of toxic compounds, the use of the remaining residues as supplementary
cementitious materials (SCM) in calcium aluminate cement (CAC)-based
binders was explored.
The evolution of the mechanical strength of standard mortars containing
highly-amorphous Fe-rich slags (30 wt% replacement) demonstrated their
positive contribution to this property of the blend. For this reason, the main
focus of this study was dedicated to the best-performing NFM slag in
binary CAC - calcium sulfate hemihydrate (HH) and ternary CAC HH
Portland cement (PC) blends. The principal formulations evaluated are
presented in Table 5.1. The influence of the sulfate proportion to the
properties of the binders was also investigated, using a low- (4.5 CAC/HH
ratio) and a high- (1.6 CAC/HH ratio) calcium sulfate systems.
Table 5.1 Composition of binders incorporating the granulated Fe-rich slag

total CAC/HH
(in wt%) slag CAC HH PC
code binder ratio
A 4.5CAC-HH 30 57 13 0 100 4.5
B 1.6CAC-HH 30 43 27 0 100 1.6
A-PC 4.5CAC-HH-30PC 30 33 7 30 100 4.5
B-PC 1.6CAC-HH-30PC 30 25 15 30 100 1.6
reference formulations:
(reference cement) no SCM nor quartz filler added
(reference filler) slag replaced by equal volume of quartz
Physical Properties
As presented in Figure 5.1, high compressive strengths were obtained for
mortars incorporating the slag and 1.3 wt% citric acid. The latter, initially
added to delay the setting, also assisted in the early strength development.
Overall, higher compressive strengths were achieved in the high-sulfate
systems due to the more extensive ettringite formation. However, the
expansive nature of ettringite proved to be problematic in reference cement
mortars containing PC (formulation B-PC) for which excessive expansion
eventually led to macro cracks (Figure 5.2) and loss in strength. The
addition of slag mitigated this expansion, producing more dimensionally-
stable mortars. This is likely governed by the dilution effect (physical
ASTOVEZA 2022 181

Conclusions & Perspectives ____________________________________
factor) in the early hydration ages, and by the resulting phase assemblage
(chemical contribution of the slag) at long term.
Figure 5.1 Summary of the compressive strength results
Figure 5.2 Pictures of the B-PC and B (high-sulfate) mortars after 45 days of curing
under water
182 ASTOVEZA 2022

Quantifying the degree of slag hydration

A methodology to estimate the degree of the slag hydration was developed
based on image analysis using X-ray computed tomography (XCT). The
technique took advantage of the contrasting X-ray signals from the high-
density Fe-rich slag and the cement matrix. The results were compared to
those obtained from Mössbauer spectroscopy in Figure 5.3. The latter was
based on the relative amounts of the absorption areas attributed to the
newly-formed Fe component during slag hydration.
Figure 5.3 Consolidated slag hydration degree estimated using XCT and Mössbauer
spectroscopy
Both results supported the faster kinetics of slag hydration with the
addition of PC. At longer curing period however, the estimated values all
converged towards 50 %. The generally lower values obtained from XCT
were linked to two main factors: (1) the assumption made in XCT that the
slag is completely unreactive in the first 24 hours of hydration; and (2) the
2.15 µm voxel resolution setting in XCT which excluded the contribution
of the finer slag particles or of other phases present at this scale.
Phase assemblage evolution

The phase assemblages of the slag-containing formulations were simulated
using GEMS thermodynamic modelling and were compared to those
observed experimentally at long term (>165 days) using XRD, TGA and
ASTOVEZA 2022 183

ATR-FTIR. Only 50 % of the amorphous fraction of the slag was assumed

to react in GEMS following the estimations from XCT for the experimental
samples. A summary of the results is provided in Table 5.2.
Table 5.2 Consolidated phase assemblage based on thermodynamic modelling
and experimental results (after >165 days). Monosulfoaluminate is abbreviated as Ms.
thermodynamic modelling experimental results
components: major minor major minor
ettringite,
ettringite, gibbsite,
A Ms gibbsite,
hydrogarnet strätlingite
strätlingite
hydrogarnet, ettringite, Ms,
B ettringite
gibbsite gibbsite strätlingite
strätlingite, hydrogarnet, strätlingite,

A-PC C-S-H
Ms C-S-H Ms
strätlingite, strätlingite,
Ms,
B-PC ettringite, ettringite C-S-H,
C-S-H
hydrogarnet Ms
While the experimental results are generally in agreement with the models,
the differences were attributed predominantly to the kinetics of hydration,
not accounted for during the simulation. Experimentally, ettringite
dominated the high-sulfate systems in the early hydration ages, while the
slag and PC exhibited slower hydration kinetics relative to the CAC and
HH. The ettringite formed was stable over time, therefore modifying the
proportion of the ions available for hydration at later ages. This could also
be the reason why the Fe-containing hydrogarnet and Fe-
monosulfoaluminate predicted in the models were not detected
experimentally.
Following the experimental phase assemblage evolution over time, the
influence of the slag was manifested in the silicates (i.e., strätlingite,
gehlenite, C-S-H) regions where the major changes were observed, relative
to the reference formulations. The crystalline component of the slag,
wüstite, also appeared unreactive over time, demonstrating that only the
glass fraction was dissolving throughout the 1-year-curing period.
Moreover, another evidence of slag reactivity was proposed using the
ATR-FTIR spectra through a hypothetical peak-displacement of the raw
-O asymmetric stretching band at 930 cm-1. However, the
complexity of the systems studied and the difficulty in characterizing the
amorphous phases did not allow a conclusive answer as to the fate of Fe
using the more common characterization techniques in cement hydration.
184 ASTOVEZA 2022

The fate of Fe during hydration

Fe was detected using TEM and STEM mainly in hydrated amorphous gel
phases grouped into two compositional clusters, as shown in Figure 5.4.
The gels were richer in Al in the binary systems, while the proportions of
Ca and Si were higher in the ternary systems. This difference was due to
the contribution of PC in the latter.
Figure 5.4 Consolidated ternary diagram of Fe-containing hydrate gels with 1-12% Fe
obtained from (TEM-) EDS analyses. Reference points for, gibbsite, strätlingite, katoite,
Si-hydrogarnet and C-A-S-H were added from literature data [202].
At longer curing ages, the TEM results revealed the tendency of

composition to move towards regions of lower Ca content in the ternary
diagram. On the other hand, the STEM maps revealed the formation of an
Fe-rich nano-crystalline phase in the binary system after longer curing
period (>180 days). This unidentified phase exhibited a rod/needle-like
morphology with a d-spacing of approximately 7.2 Å. The phase matching
these characteristics the closest would be Al-Fe hydroxide or Fe-
containing layered double hydroxide (LDH). The latter could further
explain the significant amount of Mg detected. Unfortunately, these
techniques alone were not sufficient in revealing the identity of this Fe-
bearing phase. Complementary techniques or methodology are needed for
this purpose. Nonetheless, the possibility of its ultimate transformation at
long term into Fe-containing hydrogarnet or Fe-monosulfates, as suggested
by the thermodynamic models, could not be eliminated.
In addition to the above, the evolution of the Mössbauer spectra during
hydration clearly demonstrated the change in the atomic environment of
ASTOVEZA 2022 185

Fe as new components emerged. The fittings from two independent models

consistently suggested that the hydration of the slag was accompanied by
Fe oxidation from the 2+ to 3+ state and a change from tetrahedral to
octahedral coordination. The new Fe components displayed Mössbauer
parameters similar to those reported for ferrihydrite gels and Fe-
monosulfates in cement systems. One of the two Mössbauer models further
suggested an additional 4- to 5-fold coordinated Fe2+ component in a
hydrated phase. The closest structure was comparable to the Fe
environment in a rare doublet of straurolite,
(Fe,Mg)2(Al,Fe)9O6[SiO4]4(O,OH)2, arising from the delocalization of
electrons due to a charge transfer phenomenon (i.e., Al3+ Fe2+ or Fe 2+
Fe substitutions). The decrease in the total Gaussian-type spreading of the
consequence of the transition from amorphous to nano-crystalline state,

and related to the formation of the unidentified Fe-bearing nano-crystalline
phase detected using STEM.
Other SOCRATES residues

The potential use of the jarosite residue as a sulfate source, and the paper
residue as an activator in binary CAC-HH systems were explored in
separate case studies.
The results showed that when the jarosite residue replaced the calcium
sulfate source by up to 50 wt%, high compressive strength values of 26
MPa after 1 day and 60 MPa after 28 days of hydration were attained. This
was the case despite the lower soluble sulfur in the jarosite residue based
on conductivity measurements. This implies that the change in the phase
assemblage brought by the reaction of sulfur was not drastic enough to
manifest physically. However, its environmental behavior needs to be
assessed and potentially develop pre-treatment processes, considering the
presence of significant heavy metals and elemental sulfur in its
components. On the other hand, the addition of 5 wt% lime-rich paper
residue to induce slag dissolution, accompanied by 1.3 wt% citric acid for
set control, significantly increased (> 10 MPa difference) the mechanical
strength of the CAC-HH-slag blend. This has been associated to the
anticipated increase in pH with the addition the lime-rich residue, and
improved packing of the blend prior to setting. While only a minimal
augmentation (2-5 %) in the slag hydration degree was estimated using
XCT, the addition of the paper residue appeared to influence the phase
assemblage with an increased AFm formation.
186 ASTOVEZA 2022

Future Outlook
The extensive characterization of the physical and mineralogical properties
of the CAC-based binders incorporating some of the SOCRATES residues,
demonstrated the positive potential of this valorization pathway.
Nonetheless, better understanding of the hydration mechanism of the slag
will be needed to eventually adapt the properties of the binders to their
target applications.
Future works should focus on complementary analyses to identify the Fe-
bearing nano-crystalline phase detected using STEM. This could be done
by employing other advanced characterization methods such as nuclear
magnetic resonance (NMR) or X-ray absorption (XAS) spectroscopy,
which have been previously used [113,123,206] in studying the uptake of
Fe in hydrated phases in cement systems. The identification could also be
facilitated by reproducing the phase synthetically and at higher
concentrations, in order to minimize the problem of intermixing on a nano-
scale.
Moreover, the methodology developed for quantifying the degree of slag
hydration using XCT could be optimized by reducing the voxel resolution
and including the finer fractions in the estimation. Although this would
imply a scan with longer duration or smaller volume. A refined selective
dissolution procedure based on salicylic acid-methanol [141] or NaOH-
EDTA [139,207] could also be developed to validate the estimations.
The models used for the fitting of the Mössbauer spectra could be
improved by finding the compromise between the simplistic and complex
models in order to reflect the reality more accurately. The selection of the
fitting parameters could be supported by complementary techniques, such
as titration using H2SO4, HF, and H3BO3 [208] to confirm the oxidation
state of Fe before and after hydration.
Optimizing the proportions of CAC, HH and PC aided by design of
experiment (DOE) could also potentially improve the reactivity of the slag
and the overall properties of the blend. From an industrial point of view,
this could be beneficial as the applicative performance, along with
economic factors, are crucial considerations. Finally, the long-term
durability of the optimized formulation in different chemical environments
must be tested to better understand the behavior of the blend and adapt it
according to the application.
ASTOVEZA 2022 187

__________________________________________________ Appendices
Appendices
A. Life cycle assessment and exergy analysis on the use of industrial
by-products as SCM in PC system
The paper attached in this section was presented by the author at the Forum
of Young Researchers in Sustainable Building 2019 (YRSB19) in Prague,
Czech Republic on July 2, 2019. This work focused on the addition of
fumed fayalite slag (Koranel®), one of the SOCRATES residues, in a pure
PC system. These results were excluded from the main part of the thesis
which involved CAC-rich binders.
ASTOVEZA 2022 189

Appendices __________________________________________________
190 ASTOVEZA 2022

__________________________________________________ Appendices
ASTOVEZA 2022 191

Appendices __________________________________________________
192 ASTOVEZA 2022

__________________________________________________ Appendices
ASTOVEZA 2022 193

Appendices __________________________________________________
194 ASTOVEZA 2022

__________________________________________________ Appendices
ASTOVEZA 2022 195

Appendices __________________________________________________
196 ASTOVEZA 2022

__________________________________________________ Appendices
ASTOVEZA 2022 197

Appendices __________________________________________________
198 ASTOVEZA 2022

__________________________________________________ Appendices
ASTOVEZA 2022 199

Appendices __________________________________________________
200 ASTOVEZA 2022

__________________________________________________ Appendices
ASTOVEZA 2022 201

Appendices __________________________________________________
202 ASTOVEZA 2022

__________________________________________________ Appendices
B. Mineralogy of the raw materials using XRD

This section contains the XRD patterns of the raw materials obtained
following the methodology described in chapter 2.2.1. The references used
for peak identification were taken from an internal database for the
cements, and from the inorganic crystal structure database of NIST, 2020
[209] for the other raw materials. Note that only the ranges of values are
indicated for some of the SOCRATES residues due to confidentiality
agreements. The composition of the commercial products may vary
according to their production batch.
The SOCRATES residues

granulated Fe-rich slag
Figure B.1 XRD pattern and Rietveld analysis of the granulated Fe-rich slag abbreviated
slowly-cooled Fe-rich slag
Figure B.2 XRD pattern and Rietveld analysis of the slowly-cooled Fe-rich slag
-
ASTOVEZA 2022 203

Appendices __________________________________________________
fumed fayalite slag (Koranel®)
Figure B.3 XRD pattern and Rietveld analysis of the fumed fayalite slag (Koranel®)
MSWI bottom ash
Figure B.4 XRD pattern and Rietveld analysis of the bottom ash
jarosite residue
Figure B.5 XRD pattern and Rietveld analysis of the jarosite residue
204 ASTOVEZA 2022

__________________________________________________ Appendices
paper residue (TopCrete®)
Figure B.6 XRD pattern and Rietveld analysis of the paper residue
Cements
CAC, Secar 51®
Figure B.7 XRD pattern and Rietveld analysis of the CAC (Secar 51®) abbreviated as
CAC, Ciment Fondu®
Figure B.8 XRD pattern and Rietveld analysis of the CAC (Ciment Fondue®)
abbreviated
ASTOVEZA 2022 205

Appendices __________________________________________________
Ordinary Portland cement, OPC Milke®
Figure B.9 XRD pattern and Rietveld analysis of the ordinary Portland cement
Other blend components

-Hemihydrate®
Figure B.10 XRD pattern and Rietveld analysis of the calcium sulfate hemihydrate
quartz filler, Silice E400®
Figure B.11 XRD pattern and Rietveld analysis of the quartz filler abbreviated
in this study
206 ASTOVEZA 2022

__________________________________________________ Appendices
C. Particle size distribution of raw materials

The particle size distribution of the raw materials are plotted altogether in
Figure C.1 below.
Figure C.1 The particle size distribution of the raw materials
ASTOVEZA 2022 207

Appendices __________________________________________________
D. Effect of curing condition to the mechanical strength of the

mortars
In this section, the mechanical strength of mortars are compared between
air-curing and under water curing conditions both at 20°C. Measurements
from the slag-containing blends (A) 4.5CAC-HH and (A-TC) 4.5 CAC-
HH-5TC (refer to Table 2.3); and their reference formulations (no SCM)
are presented in Figure D.1 and D.2. Each data point represents the
average value taken from three mortar samples. Error bars are partly
presented for some of the data points for the valuation of the significance
of the deviations observed among the measurements. The dotted diagonal
line represents the values where the strengths measured between the two
curing conditions are equal.
Figure D.1 Comparison of the compressive strengths of air-cured versus water-cured

mortar samples. The dotted diagonal line represents values at 1:1 ratio
In the early stage of hydration (<90 days), no significant difference is

observed between the compressive strengths of air- and water-cured
mortars for all the formulations. This could support that there is sufficient
208 ASTOVEZA 2022

__________________________________________________ Appendices
amount of water at w/b = 0.5 for the hydration reactions in the

monosulfoaluminate-dominated system (section 3.3 and 3.5). However, as
the slag starts to react more significantly from 90 days (section 3.2), it
could be argued that more water is consumed during hydration to produce
phases that contributes to the strength of the water-cured mortars. This
deviation towards values above the line (ratio 1:1) at longer curing time is
more evident for the slag containing samples circled in red in Figure D.1.
Figure D.2 Comparison of the flexural strengths of air-cured versus water-cured mortar
samples. The dotted diagonal line represents values at 1:1 ratio
The difference observed is less pronounced for the flexural strengths

(Figure D.2) most likely due to the smaller range of values measured from
all the samples. Nonetheless, the observations in this section highlight the
importance of the water-curing condition in discerning the possible long
term contribution of the slag hydration to the properties of the mortars.
ASTOVEZA 2022 209

Appendices __________________________________________________
E. Image processing operations performed for the quantification of

slag hydration degree using XCT
Image processing of the greyscale reconstructed stack images to obtain
their binary stacks is executed in the the Bruker CT-Analyzer (v 1.18.8.0+)
software. The series of operations programmed and performed consistently
to all the converted cross sections at all curing ages consisted of the
following:
Step 1: Filtering (median\round\radius 1, unsharp mask\round\radius 1)
Step 2: Global thresholding (global\83-255)
Step 3: Despeckle (remove white speckle\ less than 20 pixels)
Step 4: 3D Analysis (check: number of objects)
Step 5: Save Bitmaps
All these steps are pre-programmed in the software and can only be
customized using general pre-set properties specified inside the
parentheses in each step enumerated above. These filters were selected
during pre-trials guided by the SEM micrographs observed under BSE and
EDS detectors.
Figure E.1 Sample image correction in Brucker CTAn software
210 ASTOVEZA 2022

__________________________________________________ Appendices
F. Mössbauer Spectroscopy Parameters

Using the combination of Mössbauer parameters, the Fe speciations and
their atomic environment in the raw slag sample were hypothesized. The
description of the relevant Mössbauer parameters presented in this section
were taken from Bill, 2013 [210] and Yoshida et al., 2013 [211]. An
illustration is shown below for their graphical representation.
Figure F.1 Illustration of IS, QS and MF using the Mössbauer spectra of wüstite (a) and
magnetite (b)
Isomer shift (IS)

Refers to the shift of resonance spectrum resulting from the change in the
electron density around the Mössbauer nucleus. Graphically, it is depicted
as the distance between the peaks and the zero on the relative velocity
scale. For 57Fe Mössbauer, this value describes the valence state of Fe
atom.
Quadruple splitting (QS)
Describes the asymmetry of the electric filed around the Mössbauer
-transitions
for the 57Fe with electric quadruple interaction. Graphically, this is
represented by the difference of the velocity values corresponding to the
peaks of the doublet.
Hyperfine magnetic field (MF)
A property exclusive to magnetic materials. The energy levels in the Fe
nucleus splits into multiple components (6 components for magnetite) due
to the presence of magnetic field.
ASTOVEZA 2022 211

Appendices __________________________________________________
Total Gaussian-
Statistical parameter related to the QS derived from the least square
minimization fitting in the IMSG software. Gaussian-type symmetric
spreading of the QS values was allowed during the fitting.
Total area (A)
The total area fitted to each Fe component proportional to the quantity
(percent composition) that they represent in the modelled spectrum.
During the fitting, the typical errors for the IS and QS were ± 0.02 mm/s
and ± 5 % for the A. The components selected were based on the results
from complementary characterization techniques and literature data on
synthetic systems. Chi-squared test was used to evaluate the best fit for
each dataset evaluated.
212 ASTOVEZA 2022

__________________________________________________ Appendices
G. Estimation of slag hydration degree using SEM-BSE and image

analysis
Using the ImageJ software integrated with the MorpholibJ plug-in, the
image correction and the total slag area quantification using the SEM-BSE
micrographs is achieved following the operations described below:
Step 1: Adjust threshold (155-254)
Step 2: Area opening: remove particles below 100 pixels (MorpholibJ)
Step 3: Fill holes (MorpholibJ)
Step 4: Opening: removing long particles (square/3, MorpholibJ)
Step 5: Area opening: remove particles below 100 pixels (MorpholibJ)
Step 6: Dilation (square/1, MorpholibJ)
Step 7: Set the scale according to the known distance
Step 8: Calculate total area in binary image (Analyze Particles)
All these steps are consistently followed in the software. The figure below
demonstrates these operations:
Figure G.1 SEM-BSE image of formulation A.

Some of the slag particles are marked in yellow identified with the help of EDS analyses
ASTOVEZA 2022 213

Appendices __________________________________________________
Figure G.2 Sample image correction using Image J software
214 ASTOVEZA 2022

__________________________________________________ Appendices
H. Estimation of the theoretical slag volume in paste sample for XCT

Estimated for 100 g of anhydrous paste prior to hydration and without
accounting for porosity:
density,
wt, g wt, % vol, cm3 vol, %
g/cm3
CAC 57.3 34.7 3.0 11.4 19.5
HH 12.7 7.7 2.6 3.0 5.1
slag 30.0 18.2 3.7 5.0 8.4
water 65.0 39.4 1.0 39.4 67.0
total 165.0 100.0 10.3 58.8 100.0
ASTOVEZA 2022 215

Appendices __________________________________________________
I. Supplementary data for thermodynamic modelling

GEMS thermodynamic models calculate the chemical speciation from a
bulk composition in the equilibrium state, by minimizing the Gibbs free
energy of a system at a given temperature and pressure.
The following were applied during the modelling:
The databases psi-nagra, cemdata 18.01 and claysor 18-12 v0.1 were
activated in creating the Project files
The w/b ratio used for the models was 0.65 corresponding to the
experimental value used for the paste samples. The only exception
made was for the reference formulation in the binary system (section
3.3.1) where the w/b ratio was increased to 0.80 in order to provide
sufficient water for complete hydration, particularly in regions between
50-60 wt% CAC content.
CAC, HH and PC components were introduced in their mineral forms
Some minor (< 1 wt%) components of the raw materials (i.e., ZnO,
CuO, P2O5 from the slag) were considered unreactive during the
The components
used could be referred to in the codes provided.
Wüstite from the slag was considered unreactive and was added in the
was allowed to react roughly guided by the estimations from XCT after
1 year of hydration.
Fe-bearing hydrates that were permitted in the equilibrium state include
CSH, ettringite, monosulfates, hydrogarnet, monocarbonate,
ferrihydrites and other AFm and AFt phases. Phases that were
deactivated included pyrite, goethite, elemental iron, hematite,
magnetite, etc.
The mineralogical composition (CAC, HH and PC) and chemical assay
(slag) of each component presented in Chapter 2 were entered in the
modC array columns.
For some models, extra O2 component was added to aid the balance
CAC-HH (reference) system

In the Process Simulators of GEMS, the code used for the Controls section
was as follows:
$modC0 CAC modC1 HH
xa_[{Al2O3}] =: 0.01* (modC[0][0]*cNu+ modC[0][1]*(100-cNu));
216 ASTOVEZA 2022

__________________________________________________ Appendices
xa_[{C2S}] =: 0.01* (modC[1][0]*cNu+ modC[1][1]*(100-cNu));

xa_[{CA}] =: 0.01* (modC[2][0]*cNu+ modC[2][1]*(100-cNu));
xa_[{CaCO3}] =: 0.01* (modC[3][0]*cNu+ modC[3][1]*(100-cNu));
xa_[{CaO}] =: 0.01* (modC[4][0]*cNu+ modC[4][1]*(100-cNu));
xa_[{CaSO4_05H2O}] =: 0.01* (modC[5][0]*cNu+ modC[5][1]*(100-cNu));
xa_[{MgO}] =: 0.01* (modC[6][0]*cNu+ modC[6][1]*(100-cNu));
xa_[{SiO2}] =: 0.01* (modC[7][0]*cNu+ modC[7][1]*(100-cNu));
xa_[{C4AF}] =: 0.01* (modC[8][0]*cNu+ modC[8][1]*(100-cNu));
xa_[{K2O}] =: 0.01* (modC[9][0]*cNu+ modC[9][1]*(100-cNu));
xa_[{Aqua}] =: 65;
xa_[{O2}] =: 10;
In the Sampling section of the Process Simulators, the following code was
used to calculate the phase assemblage composition in weight percentages:
xp[J] =: J;
yp[J][0] =: phM[{CSHQ}];
yp[J][1] =: phM[{straetlingite}];
yp[J][2] =: phM[{ettringite}];
yp[J][3] =: phM[{SO4_OH_AFm}];
yp[J][4] =: phM[{OH_SO4_AFm}];
yp[J][5] =: phM[{SO4_CO3_AFt}];
yp[J][6] =: phM[{CO3_SO4_AFt}];
yp[J][7] =: phM[{Gibbsite}];
yp[J][8] =: phM[{C3AH6}];
yp[J][9] =: phM[{CAH10}];
yp[J][10] =: phM[{C4AH11}];
yp[J][11] =: phM[{C4AsH12}];
yp[J][12] =: phM[{Calcite}];
yp[J][13] =: phM[{C3FS0.84H4.32}];
yp[J][14] =: phM[{C3FS1.34H3.32}];
yp[J][15] =: phM[{Gypsum}];
yp[J][16] =: phM[{hemihydrate}];
yp[J][17] =: phM[{Ferrihydrite-am}];
yp[J][18] =: phM[{OH-hydrotalcite}];
yp[J][19] =: phM[{aq_gen}];
yp[J][20] =: 0.01* (modC[10][0]*cNu+ modC[10][1]*(100-cNu));
CAC-HH-slag system
ASTOVEZA 2022 217

Appendices __________________________________________________
was as follows:
$modC0 S51 modC1 HH modC2 slag
xa_[{Al2O3}] =: 0.01*(cNu*modC[0][0]+modC[0][1]*(70-cNu)+modC[0][2]*30);
xa_[{C2S}] =: 0.01*(cNu*modC[1][0]+modC[1][1]*(70-cNu)+modC[1][2]*30);
xa_[{CA}] =: 0.01*(cNu*modC[2][0]+modC[2][1]*(70-cNu)+modC[2][2]*30);
xa_[{CaCO3}] =: 0.01*(cNu*modC[3][0]+modC[3][1]*(70-cNu)+modC[3][2]*30);
xa_[{CaO}] =: 0.01*(cNu*modC[4][0]+modC[4][1]*(70-cNu)+modC[4][2]*30);
xa_[{CaSO4_05H2O}] =: 0.01*(cNu*modC[5][0]+modC[5][1]*(70-
cNu)+modC[5][2]*30);
xa_[{SiO2}] =: 0.01*(cNu*modC[6][0]+modC[6][1]*(70-cNu)+modC[6][2]*30);
xa_[{C4AF}] =: 0.01*(cNu*modC[7][0]+modC[7][1]*(70-cNu)+modC[7][2]*30);
xa_[{FeO}] =: 0.01*(cNu*modC[8][0]+modC[8][1]*(70-cNu)+modC[8][2]*30);
xa_[{MgO}] =: 0.01*(cNu*modC[9][0]+modC[9][1]*(70-cNu)+modC[9][2]*30);
xa_[{Na2O}] =: 0.01*(cNu*modC[10][0]+modC[10][1]*(70-cNu)+modC[10][2]*30);
xa_[{K2O}] =: 0.01*(cNu*modC[11][0]+modC[11][1]*(70-cNu)+modC[11][2]*30);
xa_[{Aqua}] =: 65;
xa_[{O2}] =: 10;
CAC-HH-PC (reference) system

was as follows:
$modC0 S51 modC1 HH modC2 PC
xa_[{Gypsum}] =:0.01*(cNu*modC[0][0]+modC[0][1]*(70-cNu)+modC[0][2]*30);
xa_[{C3A}] =: 0.01*(cNu*modC[1][0]+modC[1][1]*(70-cNu)+modC[1][2]*30);
xa_[{Al2O3}] =: 0.01*(cNu*modC[3][0]+modC[3][1]*(70-cNu)+modC[3][2]*30);
xa_[{CA}] =: 0.01*(cNu*modC[5][0]+modC[5][1]*(70-cNu)+modC[5][2]*30);
xa_[{CaCO3}] =: 0.01*(cNu*modC[6][0]+modC[6][1]*(70-cNu)+modC[6][2]*30);
xa_[{CaO}] =: 0.01*(cNu*modC[7][0]+modC[7][1]*(70-cNu)+modC[7][2]*30);
xa_[{CaSO4_05H2O}] =: 0.01*(cNu*modC[8][0]+modC[8][1]*(70-
cNu)+modC[8][2]*30);
xa_[{SiO2}] =: 0.01*(cNu*modC[9][0]+modC[9][1]*(70-cNu)+modC[9][2]*30);
xa_[{C4AF}] =: 0.01*(cNu*modC[10][0]+modC[10][1]*(70-cNu)+modC[10][2]*30);
xa_[{K2O}] =: 0.01*(cNu*modC[11][0]+modC[11][1]*(70-cNu)+modC[11][2]*30);
xa_[{Aqua}] =: 65;
218 ASTOVEZA 2022

__________________________________________________ Appendices
was as follows:
xp[J] =: J;
yp[J][6] =: phM[{C3AH6}];
yp[J][7] =: phM[{Portlandite}];
yp[J][8] =: phM[{C3(AF)S0.84H}];
yp[J][10] =: phM[{Ferrihydrite-mc}];
yp[J][11] =: 0.01*(cNu*modC[12][0]+modC[12][1]*(70-cNu)+modC[12][2]*30);
This was changed to the following code to obtain the composition in

volume percentages instead:
xp[J] =: J;
yp[J][0] =: phVol[{Gypsum}];
yp[J][1] =: phVol[{ettringite}];
yp[J][2] =: phVol[{SO4_OH_AFm}];
yp[J][3] =: phVol[{straetlingite}];
yp[J][4] =: phVol[{Gibbsite}];
yp[J][5] =: phVol[{CSHQ}];
yp[J][6] =: phVol[{C3AH6}];
yp[J][7] =: phVol[{Portlandite}];
yp[J][8] =: phVol[{C3(AF)S0.84H}];
yp[J][9] =: phVol[{Ferrihydrite-am}];
yp[J][10] =: phVol[{Ferrihydrite-mc}];
yp[J][11] =: 0.01*((cNu*modC[12][0]+modC[12][1]*(70-cNu)+modC[12][2]*30)/3);
yp[J][12] =: phVol[{aq_gen}];
CAC-HH-PC-slag system
was as follows:
$modC0 Secar51 modC1 HH modC2PC mod3slag
ASTOVEZA 2022 219

Appendices __________________________________________________
xa_[{Gypsum}] =:0.01*(cNu*modC[0][0]+modC[0][1]*(40-cNu)+modC[0][2]*30+
modC[0][3]*30);
xa_[{C3A}] =: 0.01*(cNu*modC[1][0]+modC[1][1]*(40-cNu)+modC[1][2]*30+
modC[1][3]*30);
xa_[{C3S}] =: 0.01*(cNu*modC[2][0]+modC[2][1]*(40-cNu)+modC[2][2]*30+
modC[2][3]*30);
xa_[{Al2O3}] =: 0.01*(cNu*modC[3][0]+modC[3][1]*(40-cNu)+modC[3][2]*30+
modC[3][3]*30);
xa_[{C2S}] =: 0.01*(cNu*modC[4][0]+modC[4][1]*(40-cNu)+modC[4][2]*30+
modC[4][3]*30);
xa_[{CA}] =: 0.01*(cNu*modC[5][0]+modC[5][1]*(40-cNu)+modC[5][2]*30+
modC[5][3]*30);
xa_[{CaCO3}] =: 0.01*(cNu*modC[6][0]+modC[6][1]*(40-cNu)+modC[6][2]*30+
modC[6][3]*30);
xa_[{CaO}] =: 0.01*(cNu*modC[7][0]+modC[7][1]*(40-cNu)+modC[7][2]*30+
modC[7][3]*30);
xa_[{CaSO4_05H2O}] =: 0.01*(cNu*modC[8][0]+modC[8][1]*(40-cNu)+
modC[8][2]*30+modC[8][3]*30);
xa_[{SiO2}] =: 0.01*(cNu*modC[9][0]+modC[9][1]*(40-cNu)+modC[9][2]*30+
modC[9][3]*30);
xa_[{C4AF}] =: 0.01*(cNu*modC[10][0]+modC[10][1]*(40-cNu)+modC[10][2]*30+
modC[10][3]*30);
xa_[{FeO}] =: 0.01*(cNu*modC[11][0]+modC[11][1]*(40-
cNu)+modC[11][2]*30+modC[11][3]*30);
xa_[{MgO}] =: 0.01*(cNu*modC[12][0]+modC[12][1]*(40-
cNu)+modC[12][2]*30+modC[12][3]*30);
xa_[{Na2O}] =: 0.01*(cNu*modC[13][0]+modC[13][1]*(40-
cNu)+modC[13][2]*30+modC[13][3]*30);
xa_[{K2O}] =: 0.01*(cNu*modC[14][0]+modC[14][1]*(40-
cNu)+modC[14][2]*30+modC[14][3]*30);
xa_[{Aqua}] =: 65;
xa_[{O2}] =: 25;
In the Sampling section of the Process Simulators, the following code was
used to calculate the phase assemblage composition in weight percentages:
xp[J] =: J;
yp[J][2] =: phM[{ettringite-AlFe}];
yp[J][4] =: phM[{monosulph-AlFe}];
220 ASTOVEZA 2022

__________________________________________________ Appendices
yp[J][8] =: phM[{CNASH}];
yp[J][9] =: phM[{C3AH6}];
yp[J][10] =: phM[{Portlandite}];
yp[J][11] =: phM[{C3(AF)S0.84H}];
yp[J][13] =: phM[{Calcite}];
yp[J][14] =: phM[{C3FS0.84H4.32}];
yp[J][15] =: phM[{C3FS1.34H3.32}];
yp[J][16] =: phM[{MgAl-OH-LDH}];
yp[J][17] =: phM[{OH-hydrotalcite}];
yp[J][18]] =: 0.01*(cNu*modC[15][0]+modC[15][1]*(40-cNu)+modC[15][2]*30+
modC[15][3]*30);
This was changed to the following code to obtain the composition in

volume percentages instead:
xp[J] =: J;
yp[J][0] =: phVol[{Gypsum}];
yp[J][1] =: phVol[{ettringite}];
yp[J][2] =: phVol[{ettringite-AlFe}];
yp[J][3] =: phVol[{SO4_OH_AFm}];
yp[J][4] =: phVol[{monosulph-AlFe}];
yp[J][5] =: phVol[{straetlingite}];
yp[J][6] =: phVol[{Gibbsite}];
yp[J][7] =: phVol[{CSHQ}];
yp[J][8] =: phVol[{CNASH}];
yp[J][9] =: phVol[{C3AH6}];
yp[J][10] =: phVol[{Portlandite}];
yp[J][11] =: phVol[{C3(AF)S0.84H}];
yp[J][12] =: phVol[{Ferrihydrite-am}];
yp[J][13] =: phVol[{Calcite}];
yp[J][14] =: phVol[{C3FS0.84H4.32}];
yp[J][15] =: phVol[{C3FS1.34H3.32}];
yp[J][16] =: phVol[{MgAl-OH-LDH}];
yp[J][17] =: phVol[{OH-hydrotalcite}];
yp[J][18]] =: 0.01*(((cNu*modC[15][0]+modC[15][1]*(40-cNu)+modC[15][2]*30+
modC[15][3]*30))/3.5);
yp[J][19] =: phVol[{aq_gen}];
J. Supplementary data for Mössbauer spectroscopy
ASTOVEZA 2022 221

Appendices __________________________________________________
This appendix contains supplementary calculations and graphs used in the

Mössbauer spectra fitting discussed in Chapters 3 and 4.
Estimating the expected absorption area of wüstite

Assuming that the wüstite from slag is not participating in hydration based
on the findings from several characterization techniques, its absorption
area is expected to be consistent throughout the curing period and could be
roughly estimated as follows:
where: AAT is the estimated theoretical absorption area; W is weight; and

the Fe contributions are calculated by multiplying the Fe2O3 analysis of
slag and CAC from XRD to their corresponding weights in 100 g of paste.
The estimated area increases up to 17.4 % when the Fe contribution of the
CAC is not considered (given the poor signals presented in Chapter 2
section 2.2.5), which makes the value closer to the 18.2 % obtained from
the Mössbauer spectrum of the raw slag. Other source of deviation could
be attributed to the fact that Fe was measured only as Fe2O3 in the XRD to
be consistent with the cement samples, whereas it is clear that various other
speciation exist.
Calculating the slag hydration degree based on the absorption areas

As mentioned in the Methodology (Chapter 2), the slag hydration degree
was estimated based on the changes in the total absorption areas assigned
to the components of the unreacted slag particles (Equation 2.5). These
components consist of G1, G2, G3 and M1 for Model 1; and G4 and M1
for Model 2. Using the absorption areas obtained for these components
(section 3.4.3), the results were plotted in Figure J.1. However, these
values are problematic due to the overestimation of wüstite, especially in
Model 2, as discussed in Chapter 3. In fact, when the M1 (wüstite)
component was eliminated from the calculations for both models, the
resulting hydration degrees became more comparable (Figure J.2).
222 ASTOVEZA 2022

__________________________________________________ Appendices
Figure J.1 The uncorrected slag hydration degree (primary axis)

and the sum of AA of the unreacted slag components (secondary axis)
Figure J.2
mponents (secondary axis)
In order to correct these values, the previously estimated theoretical

absorption area of wüstite (16.3wt%) in the raw slag was added in the
original equation as follows:
ASTOVEZA 2022 223

Appendices __________________________________________________
Or simply:
where: AA is the summation of the

absorption areas; Gx are the glass components G1-G3 for Model 1 and G4
for Model 2; AAT,wüstite is equal to 16.3% as shown in the preceding section
of this appendix; and i
Note that this calculation assumes that wüstite is unreactive throughout the
1 year curing and that the signature from the Fe in the CAC is negligible.
Both assumptions were supported by the findings from the other
techniques.
Additional figures from Model 1
Figure J.3 Mössbauer spectrum and components of formulation A

(after 7 days of curing) derived using Model 1
224 ASTOVEZA 2022

__________________________________________________ Appendices


ASTOVEZA 2022 225

Appendices __________________________________________________
Additional figures from Model 2
Figure J.6 Mössbauer spectrum and components of the raw slag derived using Model 2

226 ASTOVEZA 2022

__________________________________________________ Appendices


ASTOVEZA 2022 227

Appendices __________________________________________________
Figure J.10 Mössbauer spectrum and components of formulation B

Figure J.11 Mössbauer spectrum and components of formulation A-TC (after 28 days of
curing) derived using Model 2
228 ASTOVEZA 2022

__________________________________________________ Appendices
K. Hydration of pure PC system

Pure PC was hydrated at 0.6 w/c ratio following the same procedure for
the paste samples described in section 2.1.4. XRD, TGA, and ATR-FTIR
were used to characterize the phase assemblage of the hydrated samples
after 1, 7 and 28 days of hydration.
Figure K.1 XRD patterns of pure PC paste after 1, 7 and 28 days of hydration
From both the XRD and DTG patterns alone, it is clear that the hydrated
system is dominated by portlandite (CH) and the amorphous C-S-H (CSH).
The sharp diffraction peak of the CH easily masked the peaks of the minor
hydrate phases such as calcite, ettringite and other calcium aluminate
hydrate phases. The DTG curves were critical in characterizing the
presence of amorphous C-S-H and in delineating the position of its peak
relative to ettringite: C-S-H at 90-140°C, and ettringite at around 140°C.
The latter proves to be essential in interpreting the patterns of formulations
A-PC and B-PC in chapter 4. In addition, the peaks at around 172°C could
be assigned to the formation of minor amounts of monosulfoaluminates. A
small shoulder at 92°C could be due to residual water or isopropanol
during the sample preparation.
The ATR-FTIR spectra are dominated by the strong Si-O peak at 950 cm-
1
assigned to the asymmetric stretching vibration band of C-S-H. Out of
plane and in plane vibration bands between 450 and 500 cm-1 were also
assigned to C-S-H. The sharp peaks at around 3600 cm-1 are the signature
of the O-H band in portlandite. The diminishing peak at around 1110cm-1
ASTOVEZA 2022 229

Appendices __________________________________________________
most likely corresponds to the Si-O asymmetric vibration bands of the

anhydrous alite and belite phases of PC.
Figure K.2 DTG curves of pure PC paste after 1, 7 and 28 days of hydration
Figure K.3 ATR-FTIR spectra of pure PC paste after 1, 7 and 28 days of hydration
230 ASTOVEZA 2022

___________________________________________ List of Publications
List of Publications
Part of the results of this work has been previously presented by the author
in the following publications:
Journal articles
J. Astoveza; R. Trauchessec; R. Soth ; Y. Pontikes, Properties of
Calcium Aluminate Blended Cement incorporating Fe-rich Slag:
Evolution over a Curing Period of 1 Year . Journal of Construction and
Building Materials. (2021).
https://doi.org/10.1016/j.conbuildmat.2021.122569
J. Astoveza; R. Trauchessec; S. Migot-Choux; R. Soth; Y.
Pontikes, Fe-rich slag addition in ternary Portland cement, aluminate
cement and calcium sulfate binders. Journal of Cement and Concrete
Research (2022).
https://doi.org/10.1016/j.cemconres.2021.106689
Conferences
6th International Slag Valorisation Symposium. Mechelen
(Belgium). 1-5 April 2019 poster presentation and conference paper
entitled: Assessing the reactivity of industrial by-products in calcium
aluminate cement-based formulations
Forum of Young Researchers in Sustainable Building 2019.
Prague (Czech Republic). 2-4 July 2019 oral presentation and
conference paper entitled: Industrial By-Products as Non-Conventional
Supplementary Cementitious Material.
International Process Metallurgy Symposium. Espoo (Finland). 5-
6 November 2019 oral presentation entitled: Quantifying the Degree
of Fe-rich Slag Hydration in Calcium Aluminate Blended Cement by
Image Analysis of SEM-BSE and XCT.
Concrete Solutions towards Carbon Neutral Construction by 2050,
Advanced Materials for Sustainable Infrastructure Development,
Gordon Research Conference. Ventura (USA), 22-23 February 2020
poster presentation entitled: Calcium Aluminate Blended Cement
incorporating Fe-rich Slag: Overview and Characterization
Techniques.
ASTOVEZA 2022 231

List of Publications ____________________________________________
43èmes Journées du Groupe Francophone de Spectrométrie

Mössbauer, Nancy (France). 18-19 May 2022 - poster and oral
presentation entitled: Étu
Seminars
6 June
2019 - poster presentation entitled: Calcium Aluminate Blended
Cements Incorporating Engineered Residues.
The submission of a third journal article is planned for the third quarter of
2022. The paper will cover the results of the STEM and Mössbauer
spectroscopy on formulation (A) 4.5CAC-HH.
232 ASTOVEZA 2022

___________________________________________________ References
References
[1] European Commission, Circular Economy Action Plan: The
European Green Deal, (2021), https://ec.europa.eu/.
[2] SOCRATES EU MSCA-ETN, Project proposal: Innovative training
networks (ITN) call: H2020-MSCA-ITN-2016, Part B, (2016),
https://etn-socrates.eu/.
[3] United Nations Environement Program (UNEP), Buildings and
climate change: Summary for decision-makers, (2009),
www.unep.fr/scp/sun.
[4] K.L. Scrivener, V.M. John, E.M. Gartner, Eco-efficient cements:
Potential economically viable solutions for a low-CO2 cement-
based materials industry, Cement and Concrete Research 114 (2018)
2 26.
[5] P.J.M. Monteiro, S.A. Miller, A. Horvath, Towards sustainable
concrete, Nature materials 16 (2017) 698 699.
[6] A. Vollpracht, B. Lothenbach, R. Snellings, J. Haufe, The pore
solution of blended cements: a review, Mater Struct 49 (2016)
3341 3367.
[7] S. Samad, A. Shah, M.C. Limbachiya, Strength development
characteristics of concrete produced with blended cement using
ground granulated blast furnace slag (GGBS) under various curing
1213.
[8] B. Lothenbach, K. Scrivener, R.D. Hooton, Supplementary
cementitious materials, Cement and Concrete Research 41 (2011)
1244 1256.
[9] J. Bizzozero, Hydration and dimensional stability of calcium
aluminate cement based systems, (Doctorate thesis): Faculté des
sciences et techniques de l'ingénieur, Laboratoire des matériaux de
constrcution, Lausanne, EPFL, (2014).
[10] International Energy Agency, Technology Roadmap: Low-Carbon
Transition in the Cement Industry, (2018),
https://iea.blob.core.windows.net/.
[11] J.E. Rossen, Composition and morphology of C-A-S-H in pastes of
alite and cement blended with supplementary cementitious
materials, Lausanne, EPFL, (2014).
ASTOVEZA 2022 233

References ___________________________________________________
[12] R. Snellings, G. Mertens, J. Elsen, Supplementary Cementitious

Materials, Reviews in Mineralogy and Geochemistry 74 (2012)
211 278.
[13] D. Marchon, S. Kawashima, H. Bessaies-Bey, S. Mantellato, S. Ng,
Hydration and rheology control of concrete for digital fabrication:
Potential admixtures and cement chemistry, Cement and Concrete
Research 112 (2018) 96 110.
[14] AVR, TopCrete: Binder properties. Composition, particle size and
basic properties, (2021), https://www.avr.nl/.
[15] W. Chen, H.J.H. Brouwers, The reaction of slag in cement, theory
and computer modelling: Materials Science, (2006).
[16] M.F. Zawrah, A.B. Shehata, E.A. Kishar, R.N. Yamani, Synthesis,
hydration and sintering of calcium aluminate nanopowder for
advanced applications, Comptes Rendus Chimie 14 (2011) 611
618.
[17] P.C. Hewlett, F.M. Lea, Lea's chemistry of cement and concrete, 4th
ed., Butterworth-Heinemann, (2003).
[18] Imerys, CAC Process and Mineralogy. Construction Academy
(internal document), (2018), https://www.imerys.com/.
[19] E. Qoku, Characterization and Quantification of Crystalline and
Amorphous Phase Assemblage in Ternary Binders During
Hydration, (Doctorate thesis): Faculty of Mechanical, Process and
Energy Engineering, Technische Universität Bergakademie
Freiberg, (2019).
[20] L. Xu, P. Wang, G. de Schutter, G. Wu, Effect of calcium aluminate
cement variety on the hydration of portland cement in blended
system, J. Wuhan Univ. Technol.-Mat. Sci. Edit. 29 (2014) 751
756.
[21] A.M. Dunster, F. Moulinier, K.C. Quillin, G.J. Osborne, Durability
of concrete made with calcium aluminate cement and ground
granulated blastfurnace slag in sulfate and marine environments, in:
C.H. Fentiman, R.J. Mangabhai, K. Scrivener (Eds.), Calcium
Aluminate Cements: Proceedings of a centenary conference, IHS
BRE Press, (2008).
[22] A.J. Majumdar, R.N. Edmonds, B. Singh, Hydration of Secar 71
aluminous cement in presence of granulated blast furnace slag,
Cement and Concrete Research 20 (1990) 7 14.
234 ASTOVEZA 2022

___________________________________________________ References
[23] L. Fernández-Carrasco, E. Vázquez, Reactions of fly ash with

calcium aluminate cement and calcium sulphate, Fuel 88 (2009)
1533 1538.
[24] E. Henry-Lanier, B. Espinosa, C. Eychenne-Baron, Environmental
footprint of calcium aluminate cement (CAC), in: C.H. Fentiman,
R.J. Mangabhai, K. Scrivener (Eds.), Calcium aluminates:
Proceedings of the international conference 2014, Palais des Papes,
Avignon, France, 18-21 May 2014 / edited by Charles Fentiman,
Raman Mangabhai and Karen Scrivener, IHS, Bracknell, Berkshire,
(2014), pp. 3 16.
[25] U.S. Geological Survey, Minerals Commodity Summaries, (2014),
http://minerals.usgs.gov/minerals/pubs/commodity/cement/.
[26] European Commission (main authors: G. Magnus, M. Grohol),
Report on critical raw materials and the circular economy, (2018),
http://publications.europa.eu/.
[27] P.T. Jones, D. Geysen, Y. Tielemans, S. van Passel, Y. Pontikes, B.
Blanpain, M. Quaghebeur, N. Hoekstra, Enhanced Landfill Mining
in view of multiple resource recovery: a critical review, Journal of
Cleaner Production 55 (2013) 45 55.
[28] European Commission, European Enhanced Landfill Mining
Consortium: Internal Market, Industry, Entrepreneurship and SMEs,
(2014), https://ec.europa.eu/growth/sectors/raw-materials/.
[29] -M. Hannula, K. Yliniemi, M.
Lundström, Electrochemical recovery of minor concentrations of
gold from cyanide-free cupric chloride leaching solutions, Journal of
[30] P. Altinkaya, J. Mäkinen, P. Kinnunen, E. Kolehmainen, M.
Haapalainen, M. Lundström, Effect of biological pretreatment on
metal extraction from flotation tailings for chloride leaching,
Minerals Engineering 129 (2018) 47 53.
[31] T. Palden, M. Regadío, B. Onghena, K. Binnemans, Selective Metal
Recovery from Jarosite Residue by Leaching with Acid-Equilibrated
Ionic Liquids and Precipitation-Stripping, ACS Sustainable Chem.
Eng. 7 (2019) 4239 4246.
[32] S. Spathariotis, N. Peeters, K.S. Ryder, A.P. Abbott, K. Binnemans,
S. Riaño, Separation of iron(iii ), zinc(ii ) and lead(ii ) from a
choline chloride ethylene glycol deep eutectic solvent by solvent
extraction, RSC Adv. 10 (2020) 33161 33170.
ASTOVEZA 2022 235

References ___________________________________________________
[33] I.M. Pateli, A.P. Abbott, G.R.T. Jenkin, J.M. Hartley,

Electrochemical oxidation as alternative for dissolution of metal
oxides in deep eutectic solvents, Green Chem. 22 (2020) 8360
8368.
[34] N.M. Piatak, M.B. Parsons, R.R. Seal, Characteristics and
environmental aspects of slag: A review, Applied Geochemistry 57
(2015) 236 266.
[35] M. Chintinne, C. Geenen, B. Coletti, S. Smets, Production of clean
slag at Metallo Belgium, in: European Metallurgical Conference:
Leipzig, (2017), pp. 1463 1474.
[36] NEW-MINE, Green slag valorisation is now an industrial success
story, (2019), https://new-mine.eu/green-slag-valorisation-
industrial-success-story/.
[37] H. Middelweerd, AVR pioniert met grootschalige CO2-afvang,
(2020), https://www.change.inc/retail/avr-pioniert-met-
grootschalige-co2-afvang-34249.
[38] P. Cogram, Jarosite, in: Reference Module in Earth Systems and
Environmental Sciences, Elsevier, (2018).
[39] J. Li, H. Ma, Zinc and lead recovery from jarosite residues produced
in zinc hydrometallurgy by vacuum reduction and distillation, Green
Processing and Synthesis 7 (2018) 552 557.
[40] M. Rämä, S. Nurmi, A. Jokilaakso, L. Klemettinen, P. Taskinen, J.
Salminen, Thermal Processing of Jarosite Leach Residue for a Safe
Disposable Slag and Valuable Metals Recovery, Metals 8 (2018)
744.
[41] J. An, J. Kim, B.H. Nam, Investigation on Impacts of Municipal
Solid Waste Incineration Bottom Ash on Cement Hydration, ACI
Materials Journal 114 (2017).
[42] an Cheng, Effect of incinerator bottom ash properties on mechanical
and pore size of blended cement mortars, Materials & Design (1980-
2015) 36 (2012) 859 864.
[43] H.-C. Chiu, W.-H. Huang, L.-C. Hsu, Y.-G. Lin, Y.-H. Lai, C.-Y.
Lin, Calcium containing iron oxide as an efficient and robust
catalyst in (photo-)electrocatalytic water oxidation at neutral pH,
Sustainable Energy Fuels 2 (2018) 271 279.
[44] Y. Pontikes, L. Machiels, S. Onisei, L. Pandelaers, D. Geysen, P.T.
Jones, B. Blanpain, Slags with a high Al and Fe content as
precursors for inorganic polymers, Applied Clay Science 73 (2013)
93 102.
236 ASTOVEZA 2022

___________________________________________________ References
[45] I. Maragkos, I.P. Giannopoulou, D. Panias, Synthesis of ferronickel

slag-based geopolymers, Minerals Engineering 22 (2009) 196 203.
[46] Y. Pontikes, Introducing the Extraordinary Leuven Cement: Raw
Materials, Process, Performance, and First Real-Life Applications,
in: G. Gaustad, C. Fleuriault, M. Gökelma, J.A. Howarter, R.
Kirchain, K. Ma, C. Meskers, N.R. Neelameggham, E. Olivetti,
A.C. Powell, F. Tesfaye, D. Verhulst, M. Zhang (Eds.), REWAS
2019, Springer International Publishing, Cham, 2019, pp. 165 166.
[47] A. Peys, Inorganic polymers from CaO-FeO-SiO2 slags: Processing,
reaction mechanism and molecular structure, (Doctorate thesis),
(2018).
[48] A. Peys, M. Peeters, A. Katsiki, L. Kriskova, H. Rahier, Y.
Pontikes, Performance and Durability of Fe-Rich Inorganic Polymer
Composites with Basalt Fibers, in: W.M. Kriven, N.P. Bansal, M.
Kusnezoff, T. Ohji, Y. Zhou, K. Il Moon, J. Matyas, K. Shimamura,
S. Kirihara, S. Gupta, J. Wang (Eds.), Proceedings of the 41st
International Conference on Advanced Ceramics and Composites,
John Wiley & Sons, Inc, Hoboken, NJ, USA, (2018), pp. 229 238.
[49] J.L. Provis, J.S.J. van Deventer, Alkali activated materials: State-of-
the-art report, RILEM TC 224-AAM / John L. Provis, Jannie S.J.
van Deventer, editors, Springer, Dordrecht, (2014).
[50] C.H. Fentiman, R.J. Mangabhai, K. Scrivener (Eds.), Calcium
Aluminate Cements: Proceedings of a centenary conference, IHS
BRE Press, (2008).
[51] H. Fryda, E. Charpentier, J.M. Bertino, Accelerated test for
conversion of calcium aluminate cement concrete, in: C.H.
Fentiman, R.J. Mangabhai, K. Scrivener (Eds.), Calcium Aluminate
Cements: Proceedings of a centenary conference, IHS BRE Press,
(2008).
[52] C. Parr, F. Simonin, B. Touzo, C. Wöhrmeyer, B. Valdelièvre, A.
Namba, The impact of calcium aluminate cement hydration upon
the properties of refractory castables: Technical report presented in
TARJ meeting, Ako, Japan (2004).
[53] J. Bensted, Calcium aluminate cements: Structure and performance
of cements (2002) 114 138.
[54] J. Bizzozero, K. Scrivener, Hydration and microstructure of rapid-
strength binders based on OPC accelerated by early ettringite
formation, in: C.H. Fentiman, R.J. Mangabhai, K. Scrivener (Eds.),
Calcium aluminates: Proceedings of the international conference
2014, Palais des Papes, Avignon, France, 18-21 May 2014 / edited
ASTOVEZA 2022 237

References ___________________________________________________
by Charles Fentiman, Raman Mangabhai and Karen Scrivener, IHS,

Bracknell, Berkshire, (2014).
[55] M.C. Alonso, J.L. García Calvo, A. Hidalgo, L. Fernández Luco,
Development and application of low-pH concretes for structural
purposes in geological repository systems, in: Geological
Repository Systems for Safe Disposal of Spent Nuclear Fuels and
Radioactive Waste, Elsevier, (2010), pp. 286 322.
[56]
Applied Clay Science 21 (2002) 257 269.

[57] A. Hidalgo, I. Llorente, C. Andrade, Physico-chemical
characterization of some low pH concrete using silica fume and
aluminous cement: Proceedings: Workshop on qualification of low
pH cements for geological repository, Stockholm, Sweden (2003)
15 16.
[58] H. Rojo, A.C. Scheinost, B. Lothenbach, A. Laube, E. Wieland, J.
Tits, Retention of selenium by calcium aluminate hydrate (AFm)
phases under strongly-reducing radioactive waste repository
conditions, Dalton transactions (Cambridge, England 2003) 47
(2018) 4209 4218.
[59] P. Swift, H. Kinoshita, N.C. Collier, C.A. Utton, Phosphate
modified calcium aluminate cement for radioactive waste
encapsulation, Advances in Applied Ceramics 112 (2013) 1 8.
[60] R.J. Collins, W. Gutt, Research on long-term properties of high
alumina cement concrete, Magazine of Concrete Research 40 (1988)
195 208.
[61] H.G. Midgley, A. Midgley, The conversion of high alumina cement,
Magazine of Concrete Research 27 (1975) 59 77.
[62] Concrete Society, Calcium aluminate cements in construction: A re-
assessment (Technical report), (1997).
[63] D. Zhang, X. Cai, L. Hu, Effect of Curing Temperature on
Hydration of Calcium Aluminate Cement Calcium Sulfate
Limestone System, J. Mater. Civ. Eng. 30 (2018) 6018011.
[64] P.J. French, R.G. Montgomery, T.D. Robson, High concrete
strength within the hour. Concrete Society, (1971),
http://worldcat.org/issn/00105317.
[65] Chinchón, General study of alkaline
hydrolysis in calcium aluminate cement mortars under a broad range
238 ASTOVEZA 2022

___________________________________________________ References
of experimental conditions, Cement and Concrete Research 30

(2000) 1689 1699.
[66] M. Collepardi, S. Monosi, P. Piccioli, The influence of pozzolanic
materials on the mechanical stability of aluminous cement, Cement
[67] A. Hidalgo, J.L. García, M.C. Alonso, L. Fernández, C. Andrade,
Microstructure development in mixes of calcium aluminate cement
with silica fume or fly ash, J Therm Anal Calorim 96 (2009) 335
345.
[68] M. Idrees, O. Ekincioglu, M.S. Sonyal, Hydration behavior of
calcium aluminate cement mortars with mineral admixtures at
different curing temperatures, Construction and Building Materials
285 (2021) 122839.
[69] F.P. Glasser, L. Zhang, Q. Zhou, Reactions of aluminate cements
with calcium sulphate, in: R.J. Mangabhai, F.P. Glasser (Eds.),
Proceedings of the international conference on calcium aluminate
cement: IOM Communications, London, UK, (2001), pp. 551 564.
[70] H.F.W. Taylor, Cement chemistry, 2nd edition: Academic Press,
(1990).
[71] Danielle van nes, Amorphous Calcium Aluminate Cement (ACA) is
on of the latest developments in cement chemistry: Caltra Nederland
B.V., Unpublished, (2021).
[72] Z. He, Y. Li, The Influence of Mayenite Employed as a Functional
Component on Hydration Properties of Ordinary Portland Cement,
Materials (Basel, Switzerland) 11 (2018).
[73] J. Bizzozero, C. Gosselin, K.L. Scrivener, Expansion mechanisms in
calcium aluminate and sulfoaluminate systems with calcium sulfate,
[74] G. Le Saoût, B. Lothenbach, P. Taquet, H. Fryda, F. Winnefeld,
Hydration of calcium aluminate cement blended with anhydrite,
Advances in Cement Research 30 (2018) 24 36.
[75] A. Anderberg, Moisture properties of self-levelling flooring
compounds: Division of Building Materials, LTH, Lund University
(2004).
[76]
système ternaire [Ciment Alumineux Sulfate de Calcium Laitier
de Haut Fourneau] à haute teneur en sulfate de calcium De
ASTOVEZA 2022 239

References ___________________________________________________
Laboratoire de Génie Civil et d'Ingénierie Environnementale, INSA

de Lyon, (2015).
[77] Y. Han, J. Xia, H. Chang, J. Xu, The Influence Mechanism of
Ettringite Crystals and Microstructure Characteristics on the
Strength of Calcium-Based Stabilized Soil, Materials (Basel,
Switzerland) 14 (2021).
[78] M. Balonis, F.P. Glasser, The density of cement phases, Cement and
Concrete Research 39 (2009) 733 739.
[79] B. Lee, G. Kim, J. Nam, K. Lee, G. Kim, S. Lee, K. Shin, T.
-Calcium Sulfate Hemihydrate on Setting,
Compressive Strength, and Shrinkage Strain of Cement Mortar,
Materials (Basel, Switzerland) 12 (2019).
[80] T. Matschei, B. Lothenbach, F.P. Glasser, Thermodynamic
properties of Portland cement hydrates in the system CaO Al2O3
SiO2 CaSO4 CaCO3 H2O, Cement and Concrete Research 37
(2007) 1379 1410.
[81] J. Plank, C. Hirsch, Impact of zeta potential of early cement
hydration phases on superplasticizer adsorption, Cement and
[82] H. Nguyen, P. Kinnunen, K. Gijbels, V. Carvelli, H. Sreenivasan,
A.M. Kantola, V.-V. Telkki, W. Schroeyers, M. Illikainen,
Ettringite-based binder from ladle slag and gypsum The effect of
citric acid on fresh and hardened state properties, Cement and
Concrete Research 123 (2019) 105800.
[83] H. Nguyen, W. Kunther, K. Gijbels, P. Samyn, V. Carvelli, M.
Illikainen, P. Kinnunen, On the retardation mechanisms of citric
acid in ettringite-based binders, Cement and Concrete Research 140
(2021) 106315.
[84] E.G. Moffatt, M.D.A. Thomas, Durability of Rapid-Strength
Concrete Produced with Ettringite-Based Binders, ACI Materials
Journal 115 (2018).
[85] J. Bizzozero, K. Scrivener, Hydration and microstructure of rapid-
strength binders based on OPC accelerated by early ettringite
formation, in: C.H. Fentiman, R.J. Mangabhai, K. Scrivener (Eds.),
Calcium aluminates: Proceedings of the international conference
2014, Palais des Papes, Avignon, France, 18-21 May 2014 / edited
by Charles Fentiman, Raman Mangabhai and Karen Scrivener, IHS,
Bracknell, Berkshire, (2014).
[86] D. Torréns-Martín, L. Fernández-Carrasco, M.T. Blanco-Varela,
Conduction calorimetric studies of ternary binders based on
240 ASTOVEZA 2022

___________________________________________________ References
Portland cement, calcium aluminate cement and calcium sulphate, J

Therm Anal Calorim 114 (2013) 799 807.
[87] J. Nehring, J. Neubauer, S. Berger, F. Goetz-Neunhoeffer,
Acceleration of OPC by CAC in binary and ternary systems: The
role of pore solution chemistry, Cement and Concrete Research 107
(2018) 264 274.
[88] A.C.A. Rego, F.A. Cardoso, R.G. Pileggi, Ternary system Portland
cement-calcium aluminate cement-calcium sulfate applied to self-
leveling mortar: a literature review, Cerâmica 67 (2021) 65 82.
[89] M. Burteaux, "Laitiers de haut fourneau," Techniques de l'Ingénieur.
no. M 7 425 (1995).
[90] Y. Abzaev, A. Gnyrya, S. Korobkov, K. Gauss, A. Boyarintsev, S.
Tomrachev, Thermodynamic modeling of Portland cement without
mineral additives, J. Phys.: Conf. Ser. 1145 (2019) 12016.
[91] G. Van-Rompaey, Etude de la réactivité des ciments riches en
laitier, à basse température et à temps court, sans ajout chloruré
(Doctorate thesis): Université Libre de Bruxelles (2006).
[92] K. Scrivener, V. John, E. Gartner (Eds.), Eco-efficient cements:
Potential, economically viable solutions for a low-CO2, cement-
based materials industry, 2016.
[93] S. Sánchez Berriel, A. Favier, E. Rosa Domínguez, I.R. Sánchez
Machado, U. Heierli, K. Scrivener, F. Martirena Hernández, G.
Habert, Assessing the environmental and economic potential of
Limestone Calcined Clay Cement in Cuba, Journal of Cleaner
Production 124 (2016) 361 369.
[94] Y. Cancio Díaz, S. Sánchez Berriel, U. Heierli, A.R. Favier, I.R.
Sánchez Machado, K.L. Scrivener, J.F. Martirena Hernández, G.
Habert, Limestone calcined clay cement as a low-carbon solution to
meet expanding cement demand in emerging economies,
Development Engineering 2 (2017) 82 91.
[95] E.M.J. Berodier, Impact of the Supplementary Cementitious
Materials on the kinetics and microstructuraldevelopment of cement
hydration: (Doctorate thesis), École Polythechnique Fédérale de
Lausanne, (2015).
[96] I. Mehdipour, K.H. Khayat, Effect of Supplementary Cementitious
Material Content and Binder Dispersion on Packing Density and
Compressive Strength of Sustainable Cement Paste: Technical
report, ACI Materials Journal (2016) 361 369.
ASTOVEZA 2022 241

References ___________________________________________________
[97] M. Behim, M. Beddar, P. Clastres, Reactivity of Granulated Blast

Furnace Slag, Slovak Journal of Civil Engineering 21 (2013) 7 14.
[98] P.T. Jones, D. Geysen, Y. Tielemans, S. van Passel, Y. Pontikes, B.
Blanpain, M. Quaghebeur, N. Hoekstra, Enhanced Landfill Mining
in view of multiple resource recovery: A critical review, Journal of
[99] J. Krook, N. Svensson, M. Eklund, Landfill mining: a critical review
of two decades of research, Waste management (New York, N.Y.)
32 (2012) 513 520.
[100] EU, Communication from the commission to the European
parliament, the council, the European economic and social
committee and the committee of the regions: Closing the loop - An
EU action plan for the Circular Economy Brussels COM 614 final,
(2015).
[101] M. Penpolcharoen, Utilization of secondary lead slag as
construction material, Cement and Concrete Research 35 (2005)
1050 1055.
[102] C. Morrison, R. Hooper, K. Lardner, The use of ferro-silicate slag
from ISF zinc production as a sand replacement in concrete, Cement
[103] B. Gorai, R.K. Jana, Premchand, Characteristics and utilisation of
copper slag a review, Resources, Conservation and Recycling 39
(2003) 299 313.
[104] P. Asokan, M. Saxena, S.R. Asolekar, Hazardous jarosite use in
developing non-hazardous product for engineering application,
Journal of hazardous materials 137 (2006) 1589 1599.
[105] A. Wongsa, K. Boonserm, C. Waisurasingha, V. Sata, P.
Chindaprasirt, Use of municipal solid waste incinerator (MSWI)
bottom ash in high calcium fly ash geopolymer matrix, Journal of
[106] K.S. Harchand, Vishwamittar, K. Chandra, Infrared and Mössbauer
study of two Indian cements, Cement and Concrete Research 10
(1980) 243 252.
[107] F. Engström, Y. Pontikes, D. Geysen, P.T. Jones, B. Björkman, B.
Blanpain, Review: Hot stage engineering to improve slag
valorisation options, in: P.T. Jones, Y. Pontikes, J. Elsen, O. Cizer,
L. Boehme, T. Van Gerven, D. Geysen, M. Go, B. Blanpain (Eds.),
Slag Valorisation Sympossium: The transition to sustainable
materials management, (2011), pp. 231 249.
242 ASTOVEZA 2022

___________________________________________________ References
[108] A.S. Brand, E.O. Fanijo, A Review of the Influence of Steel

Furnace Slag Type on the Properties of Cementitious Composites,
Applied Sciences 10 (2020) 8210.
[109] A.M. Joseph, R. Snellings, P. van den Heede, S. Matthys, N. de
Belie, The Use of Municipal Solid Waste Incineration Ash in
Various Building Materials: A Belgian Point of View, Materials
(Basel, Switzerland) 11 (2018).
[110]
Sto
Valorization from Jarosite-PbAg Sludge, Minerals 11 (2021) 255.
[111] USGS National Minerals Information Center, Mineral Commodity
Summaries 2018: Iron and Steel Slag, (2018),
https://minerals.usgs.gov/minerals/pubs/commodity/iron_&_steel_sl
ag/mcs-2018-fesla.pdf.
[112] A.A. Ramezanianpour, Cement Replacement Materials: Properties,
Durability, Sustainability, 1st ed., Springer Berlin Heidelberg;
Imprint: Springer, Berlin Heidelberg, (2014).
[113] A. Mancini, E. Wieland, G. Geng, R. Dähn, J. Skibsted, B. Wehrli,
B. Lothenbach, Fe(III) uptake by calcium silicate hydrates, Applied
Geochemistry 113 (2020) 104460.
[114] S. Onisei, A.P. Douvalis, A. Malfliet, A. Peys, Y. Pontikes,
Inorganic polymers made of fayalite slag: On the microstructure and
behavior of Fe, Journal of the American Ceramic Society 101
(2018) 2245 2257.
[115] A. Peys, A.P. Douvalis, C. Siakati, H. Rahier, B. Blanpain, Y.
Pontikes, The influence of air and temperature on the reaction
mechanism and molecular structure of Fe-silicate inorganic
polymers, Journal of Non-Crystalline Solids 526 (2019) 119675.
[116] A. Peys, A.P. Douvalis, V. Hallet, H. Rahier, B. Blanpain, Y.
Pontikes, Inorganic Polymers From CaO-FeOx-SiO2 Slag: The Start
of Oxidation of Fe and the Formation of a Mixed Valence Binder,
Front. Mater. 6 (2019).
[117] C. Siakati, A.P. Douvalis, P. Ziogas, A. Peys, Y. Pontikes, Impact
of the solidification path of FeO x SiO 2 slags on the resultant
inorganic polymers, Journal of the American Ceramic Society 103
(2020) 2173 2184.
[118] C. Siakati, A.P. Douvalis, V. Hallet, A. Peys, Y. Pontikes,
Influence of CaO/FeO ratio on the formation mechanism and
ASTOVEZA 2022 243

References ___________________________________________________
properties of alkali-activated Fe-rich slags, Cement and Concrete

Research 146 (2021) 106466.
[119] C. Siakati, R. Macchieraldo, B. Kirchner, F. Tielens, A. Peys, D.
Seveno, Y. Pontikes, Unraveling the nano-structure of a glassy
CaO-FeO-SiO2 slag by molecular dynamics simulations, Journal of
Non-Crystalline Solids 528 (2020) 119771.
[120] A. Peys, C.E. White, D. Olds, H. Rahier, B. Blanpain, Y. Pontikes,
Molecular structure of CaO FeO x SiO 2 glassy slags and resultant
inorganic polymer binders, Journal of the American Ceramic
Society 101 (2018) 5846 5857.
[121] R.M. Cornell, U. Schwertmann, The iron oxides: Structure,
properties, reactions, occurences and uses / R.M. Cornell, U.
Schwertmann, 2nd ed., Wiley-VCH, Weinheim, Cambridge, (2003).
[122] G.E. Brown Jr., F. Farges, G. Calas, Chapter 9. X-RAY
SCATTERING AND X-RAY SPECTROSCOPY STUDIES OF
SILICATE MELTS, in: J.F. Stebbins, P.F. McMillan, D.B.
Dingwell (Eds.), Structure, Dynamics, and Properties of Silicate
Melts, De Gruyter, 1995, pp. 317 410.
[123] M. Wilke, F. Farges, P.-E. Petit, G.E. Brown, F. Martin, Oxidation
state and coordination of Fe in minerals: An Fe K- XANES
spectroscopic study, American Mineralogist 86 (2001) 714 730.
[124] A. Mancini, E. Wieland, G. Geng, B. Lothenbach, B. Wehrli, R.
Dähn, Fe(II) interaction with cement phases: Method development,
wet chemical studies and X-ray absorption spectroscopy, Journal of
colloid and interface science 588 (2021) 692 704.
[125] C.N. Singman, Atomic volume and allotropy of the elements, J.
Chem. Educ. 61 (1984) 137.
[126] S.J. Jaffer, C.M. Hansson, Chloride-induced corrosion products of
steel in cracked-concrete subjected to different loading conditions,
[127] J. Rose, A. Bénard, S. El Mrabet, A. Masion, I. Moulin, V. Briois,
L. Olivi, J.-Y. Bottero, Evolution of iron speciation during hydration
of C4AF, Waste management (New York, N.Y.) 26 (2006) 720
724.
[128] Y. Tao, D. Wan, W. Zhang, F. Wang, S. Hu, Intrinsic reactivity and
dissolution characteristics of tetracalcium aluminoferrite, Cement
and Concrete Research 146 (2021) 106485.
244 ASTOVEZA 2022

___________________________________________________ References
[129] V. Lilkov, O. Petrov, Y. Tzvetanova, P. Savov, M. Kadiyski,

Mössbauer, XRD, and Complex Thermal Analysis of the Hydration
of Cement with Fly Ash, Journal of Spectroscopy 2013 (2013) 1 9.
[130] H.K. Choudhary, A.V. Anupama, R. Kumar, M.E. Panzi, S.
Matteppanavar, B.N. Sherikar, B. Sahoo, Observation of phase
transformations in cement during hydration, Construction and
Building Materials 101 (2015) 122 129.
[131] G. Möschner, B. Lothenbach, J. Rose, A. Ulrich, R. Figi, R.
Kretzschmar, Solubility of Fe ettringite
(Ca6[Fe(OH)6]2(SO4)3·26H2O), Geochimica et Cosmochimica
Acta 72 (2008) 1 18.
[132] B.Z. Dilnesa, B. Lothenbach, G. Le Saout, G. Renaudin, A.
Mesbah, Y. Filinchuk, A. Wichser, E. Wieland, Iron in carbonate
containing AFm phases, Cement and Concrete Research 41 (2011)
311 323.
[133] J. Siramanont, B.J. Walder, L. Emsley, P. Bowen, Iron
incorporation in synthetic precipitated calcium silicate hydrates,
Cement and Concrete Research 142 (2021) 106365.
[134] K. Mei, X. Cheng, T. Gu, Y. Zheng, P. Gong, B. Li, C. Zhang, L.
Zhang, B. Dai, Effects of Fe and Al ions during hydrogen sulphide
(H2S)-induced corrosion of tetracalcium aluminoferrite (C4AF) and
tricalcium aluminate (C3A), Journal of hazardous materials 403
(2021) 123928.
[135] K.S. Harchand, R. Kumar, K. Chandra, Vishwamittar, Mössbauer
and X-ray investigations of some portland cements, Cement and
[136] M. Sharma, Vishwamittar, K.S. Harchand, D. Raj, Mössbauer and
X-ray diffraction studies of two dry and hydrated portland cements
and their clinker, Cement and Concrete Research 21 (1991) 484
488.
[137] M. Horgnies, J.J. Chen, C. Bouillon, Overview about the use of
Fourier Transform Infrared spectroscopy to study cementitious
materials, in: C.A. Brebbia, A.A. Mammoli, A. Klemm (Eds.),
Materials Characterisation VI, WIT PressSouthampton, UK, (2013),
pp. 251 262.
[138] C.M.B. Henderson, D. Taylor, Infrared spectra of
aluminogermanate- and aluminate-sodalites and a re-examination of
position of the main i.r. absorption band for compounds with
ASTOVEZA 2022 245

References ___________________________________________________
framework structures, Spectrochimica Acta Part A: Molecular

Spectroscopy 35 (1979) 929 935.
[139] V. Hallet, M.T. Pedersen, B. Lothenbach, F. Winnefeld, N. de
Belie, Y. Pontikes, Hydration of blended cement with high volume
iron-rich slag from non-ferrous metallurgy, Cement and Concrete
Research 151 (2022) 106624.
[140] M.B. Haha, K. de Weerdt, B. Lothenbach, Quantification of the
degree of reaction of fly ash, Cement and Concrete Research 40
(2010) 1620 1629.
[141] K. Nakagawa, I. Terashima, K. Asaga, M. Daimon, A study of
hydration of amorphous calcium aluminate by selective dissolution
analysis, Cement and Concrete Research 20 (1990) 655 661.
[142] A. Gholizadeh Vayghan, L. Horckmans, R. Snellings, A. Peys, P.
Teck, J. Maier, B. Friedrich, K. Klejnowska, Use of Treated Non-
Ferrous Metallurgical Slags as Supplementary Cementitious
Materials in Cementitious Mixtures, Applied Sciences 11 (2021)
4028.
[143] S. Park, S.Y. Abate, H.K. Lee, H.-K. Kim, On the quantification of
degrees of reaction and hydration of sodium silicate-activated slag
cements, Mater Struct 53 (2020).
[144] Z. Wu, Y. Wei, S. Wang, J. Chen, Application of X-Ray Micro-CT
for Quantifying Degree of Hydration of Slag-Blended Cement Paste,
J. Mater. Civ. Eng. 32 (2020) 4020008.
[145] -Ray Diffraction Analysis of Clay-
Bearing Rocks from Random Preparations, Clays and Clay Minerals
49 (2001) 514 528.
[146] C.L. Honeybourne, An interpretation of the room temperature 57Fe
- Mossbauer spectrum of Grey high alumina cement, Cement and
[147] S. Rai, S. Kurian, V.N. Dwivedi, S.S. Das, N.B. Singh, N.S.
Gajbhiye, Mössbauer and calorimetric studies of portland cement
hydration in the presence of black gram pulse, Hyperfine Interact
188 (2009) 19 24.
[148] Y. Feng, J. Kero, Q. Yang, Q. Chen, F. Engström, C. Samuelsson,
C. Qi, Mechanical Activation of Granulated Copper Slag and Its
Influence on Hydration Heat and Compressive Strength of Blended
Cement, Materials (Basel, Switzerland) 12 (2019).
246 ASTOVEZA 2022

___________________________________________________ References
[149] H. Poellmann, H.-J. Kuzel, R. Wenda, Solid solution of ettringites

- in 3CaO.A12O3.32H2O,
[150] W. Wang, Y. Shao, H. Hou, M. Zhou, Synthesis and
thermodynamic properties of arsenate and sulfate-arsenate ettringite
structure phases, PloS one 12 (2017) e0182160.
[151]
Brandon, R.L. Hohlfelder, Cement Microstructures and Durability
in Ancient Roman Seawater Concretes, in: J. Válek, J.J. Hughes,
C.J.W.P. Groot (Eds.), Historic Mortars, Springer Netherlands,
Dordrecht, (2012), pp. 49 76.
[152] A.E. Moore, H.F.W. Taylor, Crystal structure of ettringite, Acta
Crystallogr B Struct Crystallogr Cryst Chem 26 (1970) 386 393.
[153] S.M. Clark, B. Colas, M. Kunz, S. Speziale, P.J.M. Monteiro,
Effect of pressure on the crystal structure of ettringite, Cement and
[154] N. Maach, Etude experimentale de l'hydratation de sulfates de
calcium et de mayenite (C12A7 et TEP) en suspension: Projet:
Modelisation cinetique de l'hydratation des systemes cimentaires,
Internal Technical Report, Kerneos Aluminate Technologies (2016)
1 88.
[155] J. Payá, M.V. Borrachero, J. Monzó, E. Peris-Mora, F. Amahjour,
Enhanced conductivity measurement techniques for evaluation of
fly ash pozzolanic activity, Cement and Concrete Research 31
(2001) 41 49.
[156] I. Kharchenco, V. Alekseev, Effect of ettringite morphology on the
properties of expanding cement systems, E3S Web Conf. 110 (2019)
1037.
[157] M. Katsioti, P. Boura, S. Agatzini, P.E. Tsakiridis, P. Oustadakis,
Use of jarosite/alunite precipitate as a substitute for gypsum in
Portland cement, Cement and Concrete Composites 27 (2005) 3 9.
[158] T. Chen, J. Dutrizac, Jarofix: Addressing iron disposal in the zinc
industry, The Journal of the Minerals, Metals & Materials Society
53 (2001) 32 35.
[159] N. Dave, A.K. Misra, A. Srivastava, S.K. Kaushik, Setting time and
standard consistency of quaternary binders: The influence of
cementitious material addition and mixing, International Journal of
Sustainable Built Environment 6 (2017) 30 36.
ASTOVEZA 2022 247

References ___________________________________________________
[160] V. Hallet, N. de Belie, Y. Pontikes, The impact of slag fineness on

the reactivity of blended cements with high-volume non-ferrous
metallurgy slag, Construction and Building Materials 257 (2020)
119400.
[161] I. Fernandes, M.A.T.M. Broekmans, F. Noronha, Petrography and
Geochemical Analysis for the Forensic Assessment of Concrete
Damage, in: K. Ritz, L. Dawson, D. Miller (Eds.), Criminal and
Environmental Soil Forensics, Springer Netherlands, Dordrecht,
(2009), pp. 163 180.
[162] B. Chen, Study of an ettringite-based thermochemical energy
storage for buildings: (Doctorate thesis) Thermics [physics.class-
ph], Université de Lyon (2020).
[163] P. Feng, C. Miao, J.W. Bullard, Factors influencing the stability of
AFm and AFt in the Ca-Al-S-O-H system at 25 °C, Journal of the
American Ceramic Society 99 (2016) 1031 1041.
[164] D. Damidot, B. Lothenbach, D. Herfort, F.P. Glasser,
Thermodynamics and cement science, Cement and Concrete
Research 41 (2011) 679 695.
[165] M.U. Okoronkwo, F.P. Glasser, Stability of strätlingite in the
CASH system, Mater Struct 49 (2016) 4305 4318.
[166] H.G. Midgley, P. Bhaskara Rao, Formation of stratlingite,
2CaO.SiO2.Al2O3.8H2O, in relation to the hydration of high
alumina cement, Cement and Concrete Research 8 (1978) 169 172.
[167] Y. Xi, D.D. Siemer, B.E. Scheetz, Strength development, hydration
reaction and pore structure of autoclaved slag cement with added
silica fume, Cement and Concrete Research 27 (1997) 75 82.
[168] M.A. Chavda, H. Kinoshita, J.L. Provis, Modification of a calcium
aluminate cement system to prevent conversion to cubic hydrates
and minimise corrosion of encapsulated aluminium metal, in: 32nd
Cement and Concrete Science Conference, Queen's University
Belfast, (2012).
[169] A. Fernández-Jiménez, T. Vázquez, A. Palomo, Effect of Sodium
Silicate on Calcium Aluminate Cement Hydration in Highly
Alkaline Media: A Microstructural Characterization, Journal of the
American Ceramic Society 94 (2011) 1297 1303.
[170] B.Z. Dilnesa, E. Wieland, B. Lothenbach, R. Dähn, K.L. Scrivener,
Fe-containing phases in hydrated cements, Cement and Concrete
Research 58 (2014) 45 55.
248 ASTOVEZA 2022

___________________________________________________ References
[171] J.F. Zapata, H.A. Colorado, M.A. Gomez, Effect of high

temperature and additions of silica on the microstructure and
properties of calcium aluminate cement pastes, Journal of
Sustainable Cement-Based Materials 9 (2020) 323 349.
[172] P.A. Schroeder, Infrared Spectroscopy in Clay Science Schroeder,
P.A., in: A.C. Rule, S. Guggenheim (Eds.), Teaching clay science,
Clay Minerals Society, Aurora, CO, (2002), pp. 182 205.
[173] A. Cuesta, A.G. de La Torre, I. Santacruz, P. Trtik, J.C. da Silva,
A. Diaz, M. Holler, M.A.G. Aranda, Chemistry and Mass Density of
Aluminum Hydroxide Gel in Eco-Cements by Ptychographic X-ray
Computed Tomography, The Journal of Physical Chemistry 121
(2017) 3044 3054.
[174] Y. Zhang, J. Chang, Microstructural evolution of aluminum
hydroxide gel during the hydration of calcium sulfoaluminate under
different alkali concentrations, Construction and Building Materials
180 (2018) 655 664.
[175] C. Phrompet, C. Sriwong, S. Maensiri, P. Chindaprasirt, C.
Ruttanapun, Optical and dielectric properties of nano-sized
tricalcium aluminate hexahydrate (C3AH6) cement, Construction
and Building Materials 179 (2018) 57 65.
[176] N.C.M. Marty, S. Grangeon, F. Warmont, C. Lerouge, Alteration
of nanocrystalline calcium silicate hydrate (C-S-H) at pH 9.2 and
room temperature: a combined mineralogical and chemical study,
Mineralogical Magazine 79 (2015) 437 458.
[177] G. Möschner, B. Lothenbach, F. Winnefeld, A. Ulrich, R. Figi, R.
Kretzschmar, Solid solution between Al-ettringite and Fe-ettringite
Research 39 (2009) 482 489.

[178] J.E. Rossen, K.L. Scrivener, Optimization of SEM-EDS to
determine the C A S H composition in matured cement paste
samples, Materials Characterization 123 (2017) 294 306.
[179] J. Song, S.-Y. Jia, B. Yu, S.-H. Wu, X. Han, Formation of iron
(hydr)oxides during the abiotic oxidation of Fe(II) in the presence of
arsenate, Journal of hazardous materials 294 (2015) 70 79.
[180] C. Frandsen, B.A. Legg, L.R. Comolli, H. Zhang, B. Gilbert, E.
Johnson, J.F. Banfield, Aggregation-induced growth and
-FeOOH nanorods to micron- -Fe 2 O 3
spindles, CrystEngComm 16 (2014) 1451 1458.
ASTOVEZA 2022 249

References ___________________________________________________
[181] F. Bocher, A. Géhin, C. Ruby, J. Ghanbaja, M. Abdelmoula, J.-

M.R. Génin, Coprecipitation of Fe(II III) hydroxycarbonate green
rust stabilised by phosphate adsorption, Solid State Sciences 6
(2004) 117 124.
[182] E. Bazilevskaya, Iron and aluminum hydroxide nanoparticles in the
environment: From nano-scale to field processes: (Doctorate thesis),
Soil and Science Biogeochemistry, The Pennsylvania State
University (2009).
[183] S. Kanehira, S. Kanamori, K. Nagashima, T. Saeki, H. Visbal, T.
Fukui, K. Hirao, Controllable hydrogen release via aluminum
powder corrosion in calcium hydroxide solutions, Journal of Asian
Ceramic Societies 1 (2013) 296 303.
[184] N. Karimian, S.G. Johnston, E.D. Burton, Iron and sulfur cycling in
acid sulfate soil wetlands under dynamic redox conditions: A
review, Chemosphere 197 (2018) 803 816.
[185] J.-M.R. Génin, S.J. Mills, A.G. Christy, O. Guérin, A.J. Herbillon,
E. Kuzmann, G. Ona-Nguema, C. Ruby, C. Upadhyay,
Mössbauerite, Fe 63+ O 4 (OH) 8 [CO 3 ]·3H 2 O, the fully
-Michel Bay, France,
Mineral. mag. 78 (2014) 447 465.
[186] A. de Roy, C. Forano, J.P. Besse, Layered Double Hydroxides:
Synthesis and Post-Synthesis Modification, in: V. Rives (Ed.),
Layered Double Hydroxides: Present and Future: Nova Science
Publishers, Inc., (2006), pp. 1 39.
[187] T. Liang, Y. Nanru, Hydration products of calcium aluminoferrite
in the presence of gypsum, Cement and Concrete Research 24
(1994) 150 158.
[188] M.D. Dyar, D.G. Agresti, M.W. Schaefer, C.A. Grant, E.C. Sklute,
Mössbauer spectroscopy of Earth and planetary materials, Annual
Review of Earth and Planetary Sciences 34 (2006) 83 125.
[189] M.D. Dyar, C.L. Perry, C.R. Rebbert, B.L. Dutrow, M.J.
Holdaway, H.M. Lang, Mossbauer spectroscopy of synthetic and
naturally occurring staurolite: American Mineralogist, 76 (1-2), 27-
41, (1991), https://digitalcommons.lsu.edu/geo_pubs/742.
[190] M.D. Dyar, E.C. Sklute, O.N. Menzies, P.A. Bland, D. Lindsley, T.
Glotch, M.D. Lane, M.W. Schaefer, B. Wopenka, R. Klima, J.L.
Bishop, T. Hiroi, C. Pieters, J. Sunshine, Spectroscopic
characteristics of synthetic olivine: An integrated multi-wavelength
and multi-technique approach, American Mineralogist 94 (2009)
883 898.
250 ASTOVEZA 2022

___________________________________________________ References
[191] K. Kaewmanee, P. Krammart, T. Sumranwanich, P.

Choktaweekarn, S. Tangtermsirikul, Effect of free lime content on
properties of cement fly ash mixtures, Construction and Building
Materials 38 (2013) 829 836.
[192] M. De Schepper, L. Vernimmen, N. De Belie, K. De Buysser, I.
Van Driessche, The assessment of clinker and cement regenerated
from completely recyclable concrete, in: XIII ICCC International
Congress on the Chemistry of Cement: Madrid, Spain, (2011).
[193] J. Purnomo, S. Sumarni, I.N. Saputro, Effect of citric acid on
setting-time and compressive strength of concrete, IOP Conference
Series: Materials Science and Engineering 578 (2019) 12077.
[194] M. Lanzón, P.A. García-Ruiz, Effect of citric acid on setting
inhibition and mechanical properties of gypsum building plasters,
Construction and Building Materials 28 (2012) 506 511.
[195] R. Salomão, V.C. Pandolfelli, Citric acid as anti-hydration additive
for magnesia containing refractory castables, Ceramics International
37 (2011) 1839 1842.
[196] G.D. Stefanou, C. Larsinos, Influence of mixing water on the
setting time of concrete, International Journal of Cement
Composites and Lightweight Concrete 3 (1981) 45 48.
[197] V. Dodson, Time of Setting, Significance of Tests and Properties of
Concrete and Concrete-Making Materials: STP 169C, ASTM, West
Conshohocken, Pa., 83, (1994),
https://www.concreteconstruction.net/.
[198] M. Juenger, M. Won, D. Fowler, C.W. Suh, A. Edson, Effects of
supplementary cementing materials on the setting time and early
strength of concrete: Materials Science, Engineering, University of
Texas at Austin, (2008).
[199] M. Cyr, R. Pouhet, Resistance to alkali-aggregate reaction (AAR)
of alkali-activated cement-based binders, in: Handbook of Alkali-
Activated Cements, Mortars and Concretes, Elsevier, (2015), pp.
397 422.
[200] W. Guan, F. Ji, Q. Chen, P. Yan, L. Pei, Synthesis and Enhanced
Phosphate Recovery Property of Porous Calcium Silicate Hydrate
Using Polyethyleneglycol as Pore-Generation Agent, Materials
(Basel, Switzerland) (2013) 2846 2861.
[201] I. Techer, D. Bartier, P. Boulvais, E. Tinseau, K. Suchorski, J.
Cabrera, A. Dauzères, Tracing interactions between natural argillites
and hyper-alkaline fluids from engineered cement paste and
ASTOVEZA 2022 251

References ___________________________________________________
concrete: Chemical and isotopic monitoring of a 15-years old deep-

disposal analogue, Applied Geochemistry 27 (2012) 1384 1402.
[202] I. Garcia-Lodeiro, A. Palomo, A. Fernández-Jiménez, D.E.
Macphee, Compatibility studies between N-A-S-H and C-A-S-H
gels. Study in the ternary diagram Na2O CaO Al2O3 SiO2 H2O,
[203] F. Avet, E. Boehm-Courjault, K. Scrivener, Investigation of C-A-S-
H composition, morphology and density in Limestone Calcined
Clay Cement (LC3), Cement and Concrete Research 115 (2019) 70
79.
[204] F. Deschner, F. Winnefeld, B. Lothenbach, S. Seufert, P. Schwesig,
S. Dittrich, F. Goetz-Neunhoeffer, J. Neubauer, Hydration of
Portland cement with high replacement by siliceous fly ash, Cement
[205] F. Ziegler, A.M. Scheidegger, C.A. Johnson, R. Dähn, E. Wieland,
Sorption mechanisms of zinc to calcium silicate hydrate: X-ray
absorption fine structure (XAFS) investigation, Environmental
science & technology 35 (2001) 1550 1555.
[206] S.L. Poulsen, V. Kocaba, G. Le Saoût, H.J. Jakobsen, K.L.
Scrivener, J. Skibsted, Improved quantification of alite and belite in
anhydrous Portland cements by (29)Si MAS NMR: effects of
paramagnetic ions, Solid state nuclear magnetic resonance 36 (2009)
32 44.
[207] E. Demoulian, C. Vernet, F. Hawthorn, P. Gourdin, Determination
de la teneur en laitier dans les ciments par dissolutions sélectives:
7th International Congress on the chemistry of cement (1980) II,
111-151.
[208] P. CLOSE, H.M. SHEPHERD, C.H. DRUMMOND,
Determination of Several Valences of Iron, Arsenic and Antimony,
and Selenium in Glass, Journal of the American Ceramic Society 41
(1958) 455 460.
[209] NIST, Inorganic Crystal Structure Database (ICSD) Web Site,
(2020), https://icsd.nist.gov/.
[210] E. Bill, 57Fe-Mössbauer Spectroscopy and Basic Interpretation of
Mössbauer Parameters, in: Practical Approaches to Biological
Inorganic Chemistry, Elsevier, (2013), pp. 109 130.
[211] Y. Yoshida, G. Langouche, Mössbauer Spectroscopy, Springer
Berlin Heidelberg, Berlin, Heidelberg, (2013).
252 ASTOVEZA 2022

Calcium Aluminate Blended

Uploaded by

Copyright:

Available Formats

Calcium Aluminate Blended

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Calcium Aluminate Blended

Uploaded by

Copyright:

Available Formats

Calcium Aluminate Blended

Other members of the jury:

Dissertation presented in partial

« sciences des matériaux »

Ciments composés à base d'aluminate de calcium

Par: Jennifer ASTOVEZA

Thèse soutenue publiquement le 15 juin 2022 à Louvain, Belgique

© 2022 KU Leuven Faculty of Engineering Science

ASTOVEZA 2022 iii

détaillée de la réactivité. Les mortiers standards incorporant de laitiers ont

ASTOVEZA 2022 vii

Mössbauer, de la microscopie électronique à transmission (MET) et de la

viii ASTOVEZA 2022

Ciments composés à base d'aluminate de

toxiques. La fraction minérale restante, qui représente plus de 80-90% du

-produits comme substituts aux ciments, ou

avantages économiques et environnementaux [1,2]. Si les résidus du projet

matériaux de construction visés [3]:

dernières décennies. Néanmoins, pour contrôler le comportement physique

ciment car leur nature hautement amorphe est généralement reconnue

présent d'études traitant de la réactivité et du mécanisme d'hydratation des

laitiers riches en oxyde de fer (FeO-Fe2O3) dans les systèmes à base de

Figure 2 Compositions chimiques de certains résidus de SOCRATES (orange) ; additions

étudiant notamment, sur le plan fondamental et appliqué, les mécanismes

l'influence de divers paramètres (formules, minéralogie, etc.) sur les

En effet, l'hydratation des liants ternaires (ciment alumineux - ciment

donné que cette étude implique l'utilisation de matières premières réelles

pouvoir analyser leur comportement dans des systèmes complexes, de

xii ASTOVEZA 2022

de fayalite (produit prétraité) ; une cendre d'incinération de déchets

Pour identifier et sélectionner ces résidus, il a été réalisé : i) un test de

que le laitier granulé de fayalite est composé de 46 % de FeO. La fraction

réactivité. Les résidus de jarosite et ceux issus du recyclage du papier sont

ASTOVEZA 2022 xiii

Tableau 1 Composition de liants incorporant le laitier riche en fer granulé

xiv ASTOVEZA 2022

étude a été faite avec un taux

jours. La faible quantité d'acide citrique, initialement ajouté pour retarder

été obtenues dans les systèmes à haute teneur en sulfate, grâce à la

Figure 4 Résistance à la compression des mortiers incorporant le laitier

Quantification du degré d'hydratation du laitier

L'estimation du degré d'hydratation du laitier à l'aide de la spectroscopie

converge vers 50 %. Les valeurs généralement plus faibles obtenues à

xvi ASTOVEZA 2022

Figure 6 Principales étapes de l'estimation du degré d'hydratation du laitier à l'aide de

Figure 7 Degré d'hydratation du laitier estimé à l'aide de la XCT et la

ASTOVEZA 2022 xvii

Évolution de l'assemblage des phases

contenant du laitier ont été prédites à l'aide d'une modélisation

50 %. Un résumé des résultats est présenté dans le Tableau 2.

Tableau 2 Résumé des assemblages de phases obtenus à partir de la modélisation

modélisation thermodynamique résultats expérimentaux

expérimentalement observée au jeune âge dans les systèmes à haute teneur

xviii ASTOVEZA 2022

strätlingite, gehlénite, C-S-H) où des changements majeurs ont été

Mécanisme de l'hydratation du fer

ASTOVEZA 2022 xix

période de durcissement plus longue (> 180 jours), les cartographies

transforme à termes en hydrogrenat ou en monosulfoaluminate contenant

Figure 9 Diagramme ternaire de composition atomique (At.%) de gels d'hydrates avec 1-