Vous êtes sur la page 1sur 392

CHARLES TREMBLAY

Utilisation du lithium pour contrer la RAS dans le


béton : efficacité face aux granulats réactifs canadiens,
mécanismes de réaction et essais accélérés d'évaluation

Thèse présentée
à la Faculté des études supérieures et postdoctorales de l'Université Laval
dans le cadre du programme de doctorat en Sciences de la Terre
pour l'obtention du grade de Philosophiae Doctor (Ph.D.)

DEPARTEMENT DE GEOLOGIE ET DE GENIE GEOLOGIQUE


FACULTÉ DES SCIENCES ET DE GÉNIE
UNIVERSITÉ LAVAL
QUÉBEC
2012

© Charles Tremblay, 2012


Résumé

Cette thèse porte sur l'étude de l'efficacité d'adjuvants à base de lithium pour contrer la
réaction alcalis-silice dans le béton (RAS). Les principaux objectifs étaient les suivants:
(1) évaluer la performance d'une solution de UNO3 et d'un verre de lithium pour contrer la
RAS avec l'aide de l'essai sur prismes de béton d'une durée de deux ans à 38°C
(CSA A23.2-14A ou ASTM C 1293) et à jusqu'à neuf mois à 60°C (essai modifié),
(2) modifier l'essai accéléré sur barres de mortier (CSA A23.2-25A ou ASTM C 1260) afin
de permettre l'évaluation de l'efficacité de ces adjuvants, et (3) étudier les mécanismes à
l'origine de l'efficacité du lithium. Pour ce faire, les essais de laboratoire mentionnés plus
haut ont été réalisés sur une sélection de granulats réactifs représentant différents faciès
(types de roches) à travers le Canada et les Etats-Unis.

Les essais sur prismes de béton démontrent que le LiNOa, à un ratio molaire [Li]/[Na+K]
de 0,74, permet de réduire l'expansion sous la limite de 0.04% après deux ans à 38°C, avec
six des douze granulats testés; trois autres granulats exigent un dosage entre 0,74 et 1,11.
Un dosage jusqu'à 1,11 n'est pas suffisant pour les trois derniers granulats. Le dosage
requis varie donc d'une source de granulats à l'autre et ne peut être relié au type de
granulats ou à son degré de réactivité. Le verre de lithium en poudre ne s'est pas du tout
avéré efficace. Les ajouts cimentaires testés en parallèle aux adjuvants à base de lithium (à
titre comparatif) se sont montrés efficaces s'ils ont une composition appropriée et s'ils sont
utilisés à un dosage adéquat. Leur combinaison avec le LiN03 n'a pas démontré d'efficacité
supplémentaire. L'expansion des éprouvettes de béton conservées pendant six mois à 60°C
présente une bonne corrélation avec celle des prismes testés pendant deux ans à 38°C, mais
seulement pour les mélanges témoins.

Une version modifiée de la méthode accélérée sur barres de mortier ASTM Cl260 (ou
CSA A23.2-25A), d'une durée de 28 jours, est proposée pour évaluer le ratio [Li]/[Na+K]
efficace à utiliser dans le béton pour contrer la RAS. Cette méthode prédit ce ratio pour les
granulats répondant relativement bien au LiN03 et identifie les granulats qui répondent
moins bien à ce produit (i.e. ceux qui nécessiteront l'utilisation d'un dosage
significativement supérieur au dosage « standard » de 0,74 recommandé par le
manufacturier.

Des analyses au microscope électronique à balayage (MEB), par spectrométrie de masse à


ionisation secondaire (SIMS), par diffraction X et sous lumière ultra-violette ont été
réalisées sur divers échantillons de bétons, de mortiers, de granulats et de particules
réactives. L'analyse comparative des résultats de ces essais suggèrent que la réduction ou la
suppression de la dissolution de la silice, et ce, pour une autre raison que la réduction du pH
ou la formation d'une couche protectrice sur les phases réactives, est le mécanisme le plus
probable pour expliquer l'efficacité du LiN03 contre la RAS. Ce mécanisme demeure
toutefois inexpliqué.
Abstract

This thesis is concerned by the effectiveness of lithium-based products against the alkali-
silica reaction (ASR). The main objectives were: (1) to evaluate the effectiveness of LiNÛ3
and of a lithium glass to counteract the ASR using concrete prism testing at 38°C for two
years (CSA A23.2-14A or ASTM C 1293) and at 60°C up to 9 months (modified method),
(2) to modify the accelerated mortar bar test (CSA A23.2-25A or ASTM C 1260) to
evaluate the effectiveness of lithium based admixtures, and (3) to study the mechanism
involved in the reduction of the ASR in the presence of lithium. In order to achieve the
above objectives, the laboratory tests listed before were made using a selection of reactive
aggregates of different facies (rock types) from across Canada and the USA.

The concrete prisms tests showed that L.N03 at a [Li]/[Na+K] molar ratio of 0.74, satisfied
the 2-year, 0.04% expansion limit criterion at 38°C for six ofthe twelve aggregates tested;
three aggregates required a ratio between 0.74 and 1.11, while a ratio of 1.11 was not
effective with the last three aggregates. The required dosage varied from one aggregate to
another and cannot be related to aggregate composition and degree of reactivity. The
lithium glass in powder form was not found effective at all. The supplementary cementing
materials also tested in parallel to the lithium products, for comp.arison purposes, were
found effective provided they had the appropriate composition and were used under
adequate dosage. The 6-month concrete prism expansion at 60°C correlated well with the
2-year concrete prism expansion at 38°C, however only for the control mixtures.

A modified version of the accelerated mortar bar method ASTM C 1260 (or
CSA A23.2-25A), requiring 28 days, is proposed to evaluate the effective [Li]/[Na+K] ratio
to be used in concrete. This method predicts this ratio for aggregates that respond well to
LiNÛ3 and identifies the aggregates that respond particularly badly to this product (i.e. that
would require significantly higher amounts of LiNÛ3 than the "standard" dosage of
0.74 proposed by the manufacturer).

Analyses using Scanning Electron Microscopy (SEM), Secondary Ion Mass Spectrometry
(SIMS), X-Ray Diffraction (XRD) and Fluorescence under UV light were performed on
different samples of concretes, mortars, aggregates, and a variety of reactive silica particles.
The comparative analysis of the results of the above tests suggested that the reduction or
elimination ofthe dissolution of silica due to another reason than pH reduction or formation
of a protective layer over the reactive particles, is the most likely mechanism explaining the
effectiveness of LiNU3 against ASR. However, this mechanism remains unexplained.
Avant-propos

Cette thèse de doctorat regroupe cinq articles scientifiques qui ont été publiés dans des
revues scientifiques avec jury d'examen. L'ordre des articles dans la thèse respecte la
continuité de cette recherche ainsi que les dates de publication.

Chacun de ces articles est le fruit d'un travail de collaboration entre l'auteur de la thèse, le
directeur et le co-directeur des travaux et les autres membres impliqués dans l'étude. Le
directeur de la thèse est le Dr Marc-André Bérubé, professeur associé du Département de
géologie et de génie géologique de l'Université Laval. Le co-directeur est le Dr Benoît
Fournier, professeur au Département de géologie et de génie géologique de l'Université
Laval. Les autres membres impliqués dans l'étude et ayant contribué aux divers articles
sont le Dr Michael D. A. Thomas, professeur au Département de génie civil à l'Université
du Nouveau-Brunswick, le Dr Kevin J. Folliard, professeur au Département de génie civil à
l'Université du Texas à Austin, Monsieur David B. Stokes directeur de la technologie du
béton pour la Division lithium de FMC Corporation et le Dr Pierre-Claver Nkinamubanzi,
chercheur scientifique, Institut de recherche en construction, Conseil national de recherche
du Canada.
Remerciements

En tout premier lieu, j'aimerais remercier très sincèrement mon directeur de recherche,
Dr Marc-André Bérubé, qui est à l'origine de ce travail. Je n'aurais pu trouver meilleure
personne sur le plan humain et professionnel pour me soutenir tout au long de mon projet
de doctorat. Il a toujours été disponible, d'une patience exemplaire et d'une très grande
générosité.

J'exprime ma profonde gratitude à mon co-directeur, Dr Benoit Fournier, pour son aide
appréciable sur le plan technique et ses précieux conseils durant toutes ces années. Je ne le
remercierai jamais assez pour sa gentillesse, ses compétences scientifiques ainsi que sa très
grande disponibilité.

Un merci aux autres partenaires du projet, les Dr Michael Thomas et Kevin Folliard, pour
leur aide.

Je voudrais remercier le Centre de Recherche sur les Infrastructures en Béton (CRIB) et le


Département de géologie et génie géologique de l'Université Laval. Ces organismes m'ont
permis d'utiliser les laboratoires et les équipements nécessaires pour mener à bien mon
projet de recherche.

J'aimerais remercier le Fonds Québécois de la Recherche sur la Nature et les Technologies


(FQRNT) pour le soutien financier (Bourses du ministère des Transports programme A4)
accordé pendant une partie de mes études supérieures, de même que les partenaires
financiers Adjuvants Euclid Canada inc, FMC ainsi que le ministère des Transports pour
leur support.

J'aimerais également remercier messieurs Nizar Smaoui, Martin Plante, Jean Frenette,
Pierre Therrien, Thomas Côté, Éric Cyr, ainsi que mesdames Anne Croteau et Marianne
Rochefort pour leur aide technique précieuse lors de la réalisation de ce projet de recherche.
Je remercie également le Concrete Durability Center (CDC), tout particulièrement
monsieur Jason Ideker pour m'avoir si bien accueilli lors de mon stage à Austin au Texas et
pour son amitié, ainsi que messieurs Thano Drimalas et Ryan Barborak pour l'aide apportée
et les nombreuses discussions sur le béton lors de mon séjour au Texas.

Je ne pourrais passer sous silence les encouragements de la part de ma compagne


Geneviève, mon fils Pierre-Antoine, ma fille Laurianne, mes parents Monique et Real, mon
frère Nicolas ainsi que mes nombreux ami(e)s. Merci à mes collègues, tout particulièrement
à Guylaine Laforest pour son aide précieuse.

Un merci au Laboratoires d'Expertises de Québec Ltée pour leur encouragement et leur


soutien durant les dernières années de mon doctorat.

Finalement, merci à toute l'équipe travaillant sur le lithium pour les nombreux moments
mémorables et inoubliables qu'on a passés ensemble, tout particulièrement lors des voyages
à Boise, à Boston, à Las Vegas, en Chine, au Texas, à San Francisco et à Washington.
Vil

Table des matières


Page

RÉSUMÉ ii

ABSTRACT iii

AVANT-PROPOS iv

REMERCIEMENTS v

TABLE DES MATIÈRES vii

LISTE DES FIGURES xi

LISTE DES TABLEAUX xvii

1 INTRODUCTION GÉNÉRALE 1

1.1 MISE EN CONTEXTE 1

1.2 PROBLÉMATIQUE 2
1.2.1 Efficacité des adjuvants à base de lithium 2
1.2.2 Efficacité des ajouts cimentaires 3
1.2.3 Méthodes d'évaluation de l'efficacité 4

1.3 OBJECTIFS 5

1.4 CONTRIBUTION SCIENTIFIQUE DES TRAVAUX 6

1.5 PUBLICATIONS SCIENTIFIQUES COMPOSANT LA THÈSE 8

2 REVUE DE LA LITTÉRATURE 10

2.1 LES CONDITIONS ESSENTIELLES À LA RAS 10


2.1.1 Les granulats réactifs 10
2.1.2 La teneur en alcalis 12
2.1.3 Létaux d'humidité 15
Vlll

Table des matières (suite)

Page

2.2 LES MÉCANISMES DE LA RÉACTION ALCALIS-SILICE 18


2.2.1 Dissolution et solubilité delà silice 19
2.2.2 Formation d'un gel silico-calco-alcalin 24
2.2.3 Les mécanismes d'expansion 25
2.2.3.1 Pression d'imbibition 25
2.2.3.2 Pression osmotique 26
2.2.3.3 Double couche électrique 27

2.3 LES AJOUTS CIMENTAIRES POUR CONTRER LE RAS 29


2.3.1 Les cendres volantes 29
2.3.2 Les laitiers granulés de hauts fourneaux (GGBFS) 32
2.3.3 La fumée de silice 33
2.3.4 Autres ajouts cimentaires 35

2.4 LES ADJUVANTS À BASE DE LITHIUM 35


2.4.1 Historique 35
2.4.2 Sources de lithium 36
2.4.3 Le lithium dans le béton 37
2.4.4 Diminution de la dissolution de la silice 42
2.4.5 Formation d'un silicate de lithium stable protégeant la silice 43
2.4.6 Diminution de la repolymérisation de la silice 44
2.4.7 Production d'un gel moins ou non gonflant 44
2.4.8 Théorie de la double couche 45

2.5 POINT DE VUE ÉCONOMIQUE DES MÉTHODES


DE PRÉVENTION DE LA RAS 46

2.6 RÉFÉRENCES 50

3 PRESENTATION DU PROGRAMME DE DOCTORAT 65

4 INFLUENCE DE PRODUITS À BASE DE LITHIUM PROPOSÉS


POUR CONTRER LA RAS SUR LA COMPOSITION DE LA
SOLUTION INTERSTITIELLE ET DES HYDRATES DU
CIMENT 67

4.1 INTRODUCTION 67

4.2 RÉSUMÉ 67
IX

Table des matières (suite)


Page

4.3 PUBLICATION SCIENTIFIQUE NO. 1 : INFLUENCE OF


LITHIUM-BASED PRODUCTS PROPOSED FOR COUNTERACTING
ASR ON THE CHEMISTRY OF PORE SOLUTION AND
CEMENT HYDRATES 68

5 UTILISATION DE L'ESSAI SUR PRISMES DE BÉTON POUR


ÉVALUER L'EFFICACITÉ DES ADJUVANTS À BASE
DE LITHIUM 97

5.1 INTRODUCTION 97

5.2 RÉSUMÉ 97

5.3 PUBLICATION SCIENTIFIQUE NO. 2 : EFFECTIVENESS OF


LITHIUM-BASED PRODUCTS IN CONCRETE MADE WITH
CANADIAN NATURAL AGGREGATES SUSCEPTIBLE
TO ALKALI-SILICA REACTIVITY 98

6 UTILISATION DE L'ESSAI ACCÉLÉRÉ SUR BARRES DE


MORTIER POUR ÉVALUER L'EFFICACITÉ DU LINO3
CONTRE LA RÉACTION ALCALIS-SILICE (PARTIES I ET II) .130

PARTIE II

6.1 INTRODUCTION 130

6.2 RÉSUMÉ DE LA PARTIE I 130

6.3 PUBLICATION SCIENTIFIQUE NO. 3 : USE OF THE ACCELERATED


MORTAR BAR TEST TO EVALUATE THE EFFECTIVENESS OF
LINO3 AGAINST ALKALI-SILICA REACTION - PART 1:
PORE SOLUTION CHEMISTRY AND INFLUENCE OF
VARIOUS PARAMETERS 131

PARTIE II

6.4 RÉSUMÉ DE LA PARTIE II 161

6.5 PUBLICATION SCIENTIFIQUE NO. 4 : USE OF THE ACCELERATED


MORTAR BAR TEST TO EVALUATE THE EFFECTIVENESS OF
LINO3 AGAINST ALKALI-SILICA REACTION - PART 2:
COMPARISON WITH RESULTS FROM THE CONCRETE
PRISM TEST 162
X

Table des matières (suite)

Page

7 LES MÉCANISMES À L'ORIGINE DE L'EFFICACITÉ

DU NITRATE DE LITHIUM 199

7.1 INTRODUCTION 199

7.2 RÉSUMÉ 199


7.3 PUBLICATION SCIENTIFIQUE NO. 5 : EXPERIMENTAL
INVESTIGATION OF THE MECHANISMS BY WHICH LIN0 3
IS EFFECTIVE AGAINST ASR 201

8 CONCLUSION GÉNÉRALE 247

9 RECOMMANDATIONS / PERSPECTIVES 249

ANNEXE A : ANALYSES CHIMIQUES DES GRANULATS ET

DESCRIPTIONS PÉTROGRAPHIQUES DÉTAILLÉES 252

ANNEXE B : MÉLANGES DE PÂTE DE CIMENT 258

ANNEXE C : ESSAIS SUR PRISMES DE BÉTON 264

ANNEXE D : ESSAIS SUR BARRES DE MORTIER 348


XI

Liste des figures

Page

Chapitre 2 Revue de la littérature

Figure 2.1 Évolution chimique de la solution interstitielle d'une pâte


de ciment contenant 1.0 Na2Ûé et un rapport E/C de 0.5
[selon Longuet et al. 1973] 14

Figure 2.2 Effet de l'humidité relative sur l'expansion du béton à 2 ans à 38°C

(essai CSA A23.2-14A) [tirée de Pedneault 1996] 16

Figure 2.3 Surface d'une particule de silice pure 19

Figure 2.4 Solubilité de la silice amorphe en fonction du pH


[tirée de Tang et Su-fen 1980] 21
Figure 2.5 Évolution (a) de la quantité de silice dissoute, (b) de la concentration
en ions alcalins en solution et (c) du pH en fonction de la quantité de
silice initiale dans différentes solutions d'hydroxydes alcalins
de concentration 1 mol/L (pH = 14,0) [tirée de Gaboriaud 1999] 22

Figure 2.6 Modèle de la double couche 28

Chapitre 4 Influence de produits à base de lithium proposés pour contrer la RAS


sur la composition de la solution interstitielle et des hydrates du ciment

Figure 4.1 Pore solution chemistry of cement pastes stored at 23°C:


A) [Na+]. B) [K+]. C) [Na++K+]. D) [Li+]. E) [Na++K++Li+].
F) [Li+]/[Na++K+] 90

Figure 4.2 Pore solution chemistry of cement pastes stored at 38°C:


A) [Na+]. B) [K+]. C) [Na++K+]. D) [Li+]. E) [Na++K++Li+].
F) [Li+]/[Na++K+] 91

Figure 4.3 Pore solution chemistry of cement pastes stored at 60°C:


A) [Na+]. B) [K+]. C) [Na++K+]. D) [Li+]. E) [Na++K++Li+].
F) [Li+]/[Na++K+] 92

Figure 4.4 Proportion of alkalies in the pore solution of cement pastes


stored at 23°C: A) [Na+]. B) [K+]. C) [Na++K+]. D) [Li+] 93
Xll

Liste des figures (suite)

Page

Figure 4.5 Proportion of alkalies in the pore solution of cement pastes stored
at 38°C: A) [Na+]. B) [K+]. C) [Na++K+]. D) [Li+] 94

Figure 4.6 Proportion of alkalies in the pore solution of cement pastes stored
at 60°C: A) [Na+]. B) [K+]. C) [Na++K+]. D) [Li+] 95

Figure 4.7 %Na2Ûe (including Li) in the cement hydrates present in the control
and L1NO3 cement pastes: A) 23°C. B) 38°C. C) 60°C 96

Chapitre 5 Utilisation de l'essai sur prismes de béton pour évaluer l'efficacité


des adjuvants à base de lithium

Figure 5.1 Effectiveness of lithium nitrate in the presence of all reactive aggregates
at 38°C (100°F). (A) NS1, A12, NB3, NS2. (B) NB2, All, QC2, BC.
(C) NB1, ONI, ON2, QC1 124

Figure 5.2 Effectiveness ofthe lithium glass in the presence of different reactive
aggregates at 38°C(100°F) 125

Figure 5.3 Effectiveness of SCM-LiN03 combinations in the presence of


different reactive aggregates at 38°C (100°F). (A) ONI. (B) ON2.
(C)NS1.(D)NB1 126

Figure 5.4 Expansions at 38°C and 60°C (100°F and 140°F) in the concrete prism test.
(A) After 1 year at 38°C (100°F) vs. 3 months at 60°C (140°F).
(B) After 2 years at 38°C (100°F) vs. 6 months at 60°C (140°F).
(C) After 2 years at 38°C (100°F) vs. 9 months at 60°C (140°F) 127

Figure 5.5 Expansion curves in the concrete prism test at 60°C (140°F).
(A) AL2 gravel. (B) NB1 limestone. (C) BC gravel. (D) AL1 gravel 128

Figure 5.6 pH ofthe concrete pore solution of a number of mixtures.


(A) At 38°C (100°F). (B) At 60°C (140°F) 129

Figure 5.7 [Li]/[Na+K] ofthe concrete pore solution of a


number of mixtures. (A) At 38°C (100°F). (B) At60°C (140°F) 129
Xlll

Liste des figures (suite)

Page

Chapitre 6 Utilisation de l'essai accéléré sur barres de mortier pour évaluer


l'efficacité du L-NO3 contre la réaction alcalis-silice (parties I et II)

Partie I

Figure 6.1 pH in the mortar pore and soak solutions (initially IN NaOH) as a
function of time in the AMBT, with lithium only in the soak solution
(A) or in bars (B) (ONI reactive limestone aggregate) 153

Figure 6.2 Na, K, and Li concentrations in the mortar pore and soak solutions
(initially IN NaOH) as a function of time in the AMBT, with (B, D, F)
and without (A, C, E) lithium in bars and different Li concentrations
in the soak solution (ONI reactive limestone aggregate) 155

Figure 6.3 Si concentration in the soak solution without (A) and with (B) Li
in bars (ONI reactive limestone aggregate) 155

Figure 6.4 Influence ofthe concentration ofthe NaOH soak solution on expansion
in the AMBT (no Li in bars neither in the soak solution) 156

Figure 6.5 Influence ofthe [Li]/[Na] and concentration in the NaOH soak
solution on expansion in the AMBT (no Li in bars) 156

Figure 6.6 Influence ofthe composition ofthe soak solution (NaOH or KOH)
on expansion in the AMBT for two aggregates (ONI and NM)
(no Li in bars) 157

Figure 6.7 Influence ofthe cement alkali content on expansion in the AMBT
(no Li in bars). (A) Expansion with time in absence of Li in the soak
solution (standard procedure). (B) 14-day expansion as a function of
the [Li]/[Na] in the soak solution 157

Figure 6.8 Influence ofthe water-to-cement ratio on expansion in the AMBT as a


function ofthe [Li]/[Na] in the soak solution (no Li in bars) 158

Figure 6.9 14-day expansion in the AMBT for different aggregates as a function
ofthe [Li]/[Na] in the soak solution 158

Figure 6.10 14-day expansion in the AMBT as a function ofthe [Li]/[Na]


in the soak solution with or without lithium in bars. (A) ONI.
(B) NM. (C) NSI, NB3, ON2, and TX. (D) NB1, NB2, BC, and AL2.
(E) QC, AL1, and NS2 160
XIV

Liste des figures (suite)

Page

Figure 6.11 Effect on the 14-day expansion in the AMBT of using a [Li]/[Na+K]
of 0.74 in bars, as a function of aggregate expansivity (i.e. 14-day
expansion ofthe control mixtures subjected to the standard procedure).... 160

Partie II

Figure 6.12 Minimum [Li]/[Na+K] needed in the CPT as a function of aggregate


expansivity in the CPT (i.e. 1-year expansion for the control mixture) 189

Figure 6.13 14-day and 28-day expansions in the AMBT for different aggregates
as a function ofthe [Li]/[Na] in the soak solution using a [Li]/[Na+K]
of0.74inbars 190

Figure 6.14 Aggregate expansivity in the AMBT (14-day expansion for the control
mixture) as a function of aggregate expansivity in the CPT
(i.e. 1-year expansion for the control mixture) 191

Figure 6.15 Minimum [Li]/[Na+K] needed in the CPT as a function of aggregate


expansivity in the AMBT (i.e. 14-day expansion for the control
mixture) 191

Figure 6.16 Minimum [Li]/[Na+K] needed in the CPT as a function ofthe 28-day
expansion in the AMBT in the absence of lithium in bars, for a
[Li]/[Na] ranging from 0.222 to 0.592 in the soak solution 192

Figure 6.17 Minimum [Li]/[Na+K] needed in the CPT as a function ofthe 28-day
expansion in the AMBT using a [Li]/[Na+K] of 0.74 in bars and a
[Li]/[Na] ranging from 0 to 0.444 in the soak solution 193

Figure 6.18 Minimum [Li]/[Na+K] needed in the CPT as a function ofthe minimum
[Li]/[Na] needed in the soak solution to satisfy the 0.10% expansion limit
criterion in the AMBT after 14 days (A, C) and 28 days (B, D), in the
absence (A, B) and in the presence (C, D) of lithium in bars at a
[Li]/[Na+K] of 0.74 194

Figure 6.19 Minimum [Li]/[Na+K] needed in the CPT as a function ofthe


reduction in expansion in the AMBT after 28 days in the absence
of lithium in bars and for a [Li]/[Na] r.anging from 0.222 to 0.592
in the soak solution, compared to the control mixtures 195
XV

Liste des figures (suite)


Page

Figure 6.20 Minimum [Li]/[Na+K] needed in the CPT as a function ofthe reduction
in expansion in the AMBT after 14 or 28 days using a [Li]/[Na+K]
of 0.74 in bars and a [Li]/[Na] ranging from 0 to 0.444 in the soak
solution, compared to the control mixtures 197

Figure 6.21 Potential process, using the AMBT, to determine the [Li]/[Na+K]
needed in concrete to prevent ASR with lithium nitrate 197

Figure 6.22 Minimum [Li]/[Na+K] needed in the CPT as a function ofthe reduction
in expansion in the AMBT after 14 or 28 days using a [Li]/[Na+K]
of 0.74 in bars and various [Li]/[Na] in the soak solution 198

Chapitre 7 Les mécanismes à l'origine de l'efficacité du nitrate de lithium

Figure 7.1 Micro-SIMS elemental scans on a polished section of a concrete


made with a highly reactive rhyolite and a [Li]/[Na+K] of 0.74,
after 2 years at 38°C and >95% R.H. (0.01 vol% expansion) 240

Figure 7.2 XRD analysis of a polished section of obsidian before (A) and after (B)
one hour of autoclaving in a IN LiOH solution at 300 psi or 2070 kPa
(350°C). (A) The obsidian is mostly composed of amorphous glass.
(B) The obsidian surface is covered by a layer of crystalline lithium
silicate Li 2 Si0 3 (ICDD card file 29-0829) 241

Figure 7.3 XRD analysis of a polished section of a highly reactive rhyolite


before (A) and after (B) one hour of autoclaving in a IN LiOH
solution at 300 psi or 2070 kPa (350°C). After autoclaving (B), a
thin crystalline lithium silicate Li 2 Si0 3 (ICDD card file 29-0828)
is detected on the rhyolite surface 242

Figure 7.4 SEM views at different scales ofthe Li-silicate layer (Li2Si03)
observed on the surface of an obsidian particle immersed for
6 months in a IN LiOH + 0.74N LiN0 3 solution at 80°C 243
XVI

Liste des figures (suite)

Page

Figure 7.5 Alkali concentrations in the soak solution over time when aggregate
powders (150-300 um) were immersed in various Li solutions at 60°C.
(A) All aggregates in 0.68N (Na,K)OH control solution.
(B) All aggregates in IN LiOH.
(C) Rhyolite in 0.8N (Na,K)OH + 0.28N LiN0 3 ([Li]/[Na+K]) = 0.35).
(D) Rhyolite in 0.8N (Na,K)OH + 0.59N LiNOs ([Li]/[Na+K]) = 0.74) ...244

Figure 7.6 Mass variation of opal and chalcedony particles immersed at 80°C
in solutions of IN NaOH (control) and IN NaOH + LiN03 in various
amounts ([Li]/[Na] from 0 to 1.0). (A) Chalcedony. (B) Opal.
(C) Opal and chalcedony particles immersed for 28 days in a solution
of IN NaOH + 0.5N LiN0 3 (chalcedony) or 1 .ON LiN0 3 (opal),
then immersed in the IN NaOH control solution 245

Figure 7.7 XRD analysis ofthe crystalline lithium silicate L_2Si03 (ICDD card
file 29-0828) detected on the surface of a chalcedony particle
after immersion for 28 days in a IN NaOH + 0.5N LÎNÛ3
solution at 80°C 246
xvii

Liste des tableaux

Page

Chapitre 2 Revue de la littérature

Tableau 2.1 Types de roches potentiellement réactifs aux alcalis du ciment 12

Tableau 2.2 Rayons ioniques des principaux alcalis 36

Tableau 2.3 Résultats des différentes études sur l'utilisation du lithium contre
la RAS (modifié à partir de Folliard et al. 2003) 39

Tableau 2.4 Théories proposées par différents auteurs pour expliquer l'effet

bénéfique du lithium contre la RAS 41

Tableau 2.5 Coûts d'utilisation du nitrate de lithium dans le béton 47

Tableau 2.6 Coûts moyens des ajouts cimentaires par rapport au ciment

dans chaque province canadienne [tiré de Bouzoubaa et Fournier 2003] ....47

Tableau 2.7 Coûts moyens du transport des matériaux 48

Tableau 2.8 Coûts en matériaux de diverses méthodes de prévention de la RAS


pour 1 m3 de béton de ciment 48
Chapitre 4 Influence de produits à base de lithium proposés pour contrer la RAS
sur la composition de la solution interstitielle et des hydrates du ciment

Table 4.1 Cement paste mixtures, storage temperature, and testing 87

Table 4.2 Total design alkali contents measured, water contents, and proportions of
pore
solution extracted 88

Table 4.3 Chemistry of pore solutions (all concentrations and ratios are in Mol/L
of pore solution) 89

Chapitre 5 Utilisation de l'essai sur prismes de béton pour évaluer l'efficacité


des adjuvants à base de lithium

Table 5.1 Origin, composition, and reactivity level ofthe coarse aggregates 116

Table 5.2 Chemical analysis ofthe cements and SCM's 116


XV111

Liste des tableaux (suite)

Page

Table 5.3 Concrete mixtures tested and expansion (%) at 2 years in the standard
CPT at 38°C (100°F) 117

Table 5.4 Petrography of the coarse aggregates 118

Table 5.5 Minimum [Li]/[Na+K] ratio to satisfy the 0.040% expansion limit
criterion in the concrete prism test (the results are classed by
increasing reactivity level ofthe coarse aggregates investigated
in the control mixtures) 119

Table 5.6 Expansion (%) at 6 months in the accelerated CPT at 60°C (140°F) 120

Table 5.7 Expansion (%) at 9 months in the accelerated CPT at 60°C (140°F) 121

Table 5.8 Correlation between the CPT at 60°C (140°F) vs 38°C (100°F) 122

Table 5.9 Chemical analysis of pore solutions expressed from concrete test prisms
after various periods of storage at 38 or 60°C (100 or 140°F) 123

Chapitre 6 Utilisation de l'essai accéléré sur barres de mortier pour évaluer

l'efficacité du LiN03 contre la réaction alcalis-silice (parties I et II)

Partie I

Table 6.1 Origin, rock types/minerai assemblages, and level of reactivity

ofthe aggregates 150

Table 6.2 Chemical analysis ofthe cements 150

Table 6.3 Experimental conditions and 14-day expansions in the AMBT 151

Table 6.4 Coefficient of variation ofthe 14-day expansion in the AMBT (control
mixtures and standard procedure with no Li in the soak solution) 152
Table 6.5 Influence ofthe initial 24-hr soaking in water on the 14-day expansion
in the AMBT 152
XIX

Liste des tableaux (suite)


Page

Partie II

Table 6.6 Origin, rock type, and reactivity of the coarse aggregates 181

Table 6.7 Concrete mixtures tested in the CPT and expansions obtained after one

(control mixtures) or two years (lithium mixtures) 181

Table 6.8 Experimental conditions and 14-day expansions in the AMBT 182

Table 6.9 28-day expansions in the AMBT 183

Table 6.10 Effect of Li in the mortar bars and/or in the soak solution on the 14- and
28-day expansions in the AMBT 184
Table 6.11 Minimum [Li]/[Na] needed in the soak solution to satisfy the 0.10%
expansion limit criterion in the AMBT after 14 and 28 days, with or
without Li in bars 185

Table 6.12 Coefficients of linear regression (R ) obtained between CPT and


AMBT results at 14 and 28 days for the various conditions tested
at Laval University 186

Table 6.13 Determination ofthe [Li]/[Na+K] needed in concrete based


on different limits after 28 days in the AMBT using a
[Li]/[Na+K] in the soak solution of 50% of this ratio in bars 187

Table 6.14 Determination ofthe [Li]/[Na+K] needed in concrete based


on the reduction/increase of expansion after 28 days in the AMBT
using a [Li]/[Na+K] of 0.74 in bars and a [Li]/[Na] of 0.148
in the soak solution 188
XX

Liste des tableaux (suite)


Page

Chapitre 7 Les mécanismes à l'origine de l'efficacité du nitrate de lithium

Table 7.1 Minerai phases detected by XRD on the polished surface of various
materials before and after immersion for one hour in a IN LiOH
solution at 350°C (300 psi or 2070 kPa in the autoclave) 237

Table 7.2 Li consumption (in wt%) after 28 days in the test solution when
aggregate powders (150-300 pm) were immersed in various Li
solutions at 60°C 237

Table 7.3 Mass variation (in wt%) of various aggregates and mineral phases
(varieties of silica, volcanic glass, and feldspars) after 28 days in
various Li solutions at 80°C 238

Table 7.4 Summary of the degree of agreement between the results of test series
nos. 1 to 4d and the mechanisms A to F proposed to explain the
effectiveness of lithium against ASR 239
1 Introduction générale
1.1 Mise en contexte

Pendant de nombreuses années, on a pensé que les granulats utilisés dans le béton de
ciment étaient pratiquement inertes, qu'ils n'étaient l'objet d'aucune interaction chimique
avec la pâte de ciment hydraté et qu'ils ne pouvaient être une cause de détérioration du
béton. Dans les années 40, en Californie, on a reconnu le premier cas de détérioration du
béton causé par une réaction chimique entre les granulats et la solution interstitielle du
béton [Stanton 1940]. Cette réaction, nommée « réaction alcalis-granulats » (RAG) à la
suite de sa découverte, produit une expansion du béton qui cause sa fissuration et sa
détérioration prématurée. La réaction alcalis-granulats consiste en une réaction chimique
entre certains types de granulats et les ions hydroxyles (OH) provenant de la solution
interstitielle de la pâte de ciment hydratée. Depuis, de nombreux cas ont été rapportés et, en
1989, la RAG venait au douzième rang sur une liste de seize causes de détérioration du
béton [OCDE 1989]. Par contre, l'importance des dommages causés par la RAG varie
beaucoup d'une région à l'autre en fonction de la composition des granulats et des ciments
utilisés, ainsi que des conditions climatiques. C'est pour cela que, dans certaines régions,
elle est une des principales causes de dégradation du béton. On retrouve principalement
deux types de RAG, soit la réaction alcalis-silice (RAS), qui est la plus répandue et dont il
est question dans ce travail, et la réaction alcalis-carbonate (RAC) qui se produit avec
certains granulats de calcaires dolomitiques argileux.

La réaction alcalis-silice (RAS) est une réaction chimique se produisant entre la solution
interstitielle fortement alcaline du béton et certaines phases siliceuses présentes dans les
granulats. La solution interstitielle du béton est une phase aqueuse qui est fortement basique
(pH > 13) et qui possède une concentration en ions alcalins élevée [Longuet et al. 1973]. Le
pH élevé de la solution interstitielle favorise la solubilisation de certaines phases siliceuses,
ce qui provoque la formation d'un gel silico-calco-alcalin. Le gel formé augmente de
volume avec l'eau absorbée provoquant des contraintes à l'intérieur des granulats et de la
pâte, ce qui se traduit par le gonflement et la fissuration du béton. Le gonflement du béton
peut facilement être de l'ordre de 0,1% à 0,2%, ce qui représente un allongement de 1 à 2
centimètres pour un élément de béton long de 10 mètres. L'ampleur de cette expansion sera
toutefois grandement influencée par la présence de confinement (interne - armatures; ou
externe) agissant sur l'élément affecté. Dans certains types d'ouvrages (barrages, trottoirs,
dalles, barrières médianes, etc.), ce niveau de gonflement peut être problématique si le
béton n'a pas l'espace disponible pour prendre librement de l'expansion. L'expansion
causant des mouvements différentiels entre les différents éléments d'une structure affectée
par la RAS est souvent le problème le plus sérieux à gérer, surtout dans le cas d'ouvrages
tels que des barrages ou certains éléments mobiles ne peuvent tolérer de tels mouvements.
La fissuration du béton lors de la RAG réduit considérablement sa résistance et ce, surtout
en traction. De plus, les fissures sont des zones privilégiées pour la pénétration des ions
chlorures qui peuvent entraîner la corrosion de l'acier d'armature.

Trois conditions simultanées sont essentielles au développement de la RAS au sein du


béton [Dolar-Mantuani 1983, Regourd 1983, Oberholster 1984]. Le granulat composant le
béton doit être réactif (phases réactives dans le granulat), l'humidité relative dans le béton
doit être assez élevée et la concentration en ions alcalins doit dépasser un seuil critique dans
la solution interstitielle.

1.2 Problématique

1.2.1 Efficacité des adjuvants à base de lithium

Certains travaux réalisés au début des années 50 [McCoy et Caldwell 1951] ont fait
ressortir l'effet bénéfique du lithium pour combattre la réaction alcalis-silice dans le béton.
La majorité de ces travaux ont montré que l'efficacité du lithium est fonction de la teneur
en alcalis du béton et de la quantité de lithium utilisée par rapport à ces alcalis. Un rapport
molaire [Li]/[Na+K] situé entre 0,6 et 1,0 suffirait à éliminer l'expansion pour la majorité
des granulats [Blackwell et al. 1997]. Le dosage efficace varie toutefois de façon
significative selon le granulat réactif à contrer. Aucune relation entre le degré de réactivité
des granulats et l'efficacité du lithium n'a pu être établie. Étonnamment, certains granulats
très réactifs sont faciles à contrôler alors que d'autres, beaucoup moins réactifs, nécessitent
un plus fort dosage. On sait également que le nitrate de lithium (LiN03) est plus efficace, à
dosage (ou rapport molaire) égal, que l'hydroxyde de lithium (LÎOH.H20), ce dernier
augmentant même parfois l'expansion lorsque utilisé à faible dosage [Diamond 1989], alors
que la performance d'un produit commercialisé depuis une dizaine d'années, un verre à
base de lithium, reste à démontrer. Toutefois, ce nouveau produit pourrait s'avérer
particulièrement efficace et utile pour des utilisations particulières comme le béton projeté
par voie sèche. Somme toute, les mécanismes d'action par lesquels le lithium permet de
réduire la RAG ne sont pas encore bien compris.

Le nitrate et l'hydroxyde de lithium ont été utilisés au cours des dernières années à des fins
préventives dans le cas de nouveaux ouvrages fabriqués avec des granulats réactifs, en
présence ou non de certaines quantités de cendres volantes [Folliard et al. 2003]. On a aussi
récemment utilisé de tels adjuvants afin de ralentir la détérioration d'ouvrages déjà atteints
de RAS, principalement aux Etats-Unis. Des solutions de lithium ont ainsi été appliquées
au béton par un processus d'imprégnation électrochimique, d'imprégnation sous vide ou par
simple épandage sur des pavages affectés de RAS [Folliard et al. 2008]. Les traitements de
surface et d'imprégnation sous vides se sont avérés inefficaces dû à la faible pénétration des
ions lithium dans le béton. La méthode d'imprégnation électrochimique a montré
significativement plus d'efficacité pour faire pénétrer le lithium dans le béton. Toutefois, il
a été remarqué que les ions sodium et potassium pouvaient ainsi se concentrer
considérablement près de l'acier d'armature, soulevant des questions supplémentaires quant
à l'effet bénéfique du lithium pour les éléments de béton armés.

1.2.2 Efficacité des ajouts cimentaires

Certains ajouts cimentaires comme les fumées de silice, les cendres volantes et les laitiers
de hauts-fourneaux sont considérés comme efficaces contre la RAS à la condition que leur
composition soit appropriée et qu'ils soient utilisés en quantités suffisantes en
remplacement du ciment. Généralement, pour les cendres volantes, celles contenant une
forte proportion en silice et une faible quantité en alcalis sont les plus efficaces, alors
qu'une quantité plus importante de cendre à forte teneur en calcium sera généralement
nécessaire pour obtenir un niveau adéquat de prévention contre la RAS. Par contre, dans
certaines circonstances (granulats très réactifs, humidité élevée et/ou ouvrage avec une
durée de vie très longue), le dosage requis pour contrer la RAS est élevé et risque de nuire à
la durabilité de l'ouvrage face aux cycles de gel/dégel. De plus, dans certaines régions,
l'approvisionnement de ces ajouts cimentaires peut être très coûteux dû à l'éloignement des
ressources et aux coûts de transport que cela implique. Dans ces cas, le lithium pourrait
s'avérer un moyen efficace de prévention contre la RAG. Par ailleurs, plusieurs études
suggèrent l'apport positif de l'utilisation combinée de certains ajouts cimentaires (cendres
volantes) et d'adjuvants à base de lithium [Thomas et al. 2000, Stokes et Manissero 2001].

1.2.3 Méthodes d'évaluation de l'efficacité

Dans les études antérieures aux années 90 portant sur l'utilisation des adjuvants à base de
lithium, on réalisait des essais sur barres de mortier avec des granulats hyper-réactifs
comme la cristobalite, l'opale, le pyrex ou une variété d'andésite riche en verre volcanique.
Or, au fil du temps, ces essais (ASTM C 227 et C 441) se sont révélés peu réalistes ou peu
fiables afin d'évaluer la performance de mesures préventives contre la RAS [Bérubé et
Duchesne 1992]. L'essai le plus fiable présentement utilisé et reconnu pour sa capacité à
démontrer l'efficacité des produits à base de lithium face à la RAS est l'essai sur prismes de
béton ASTM C 1293 ou CSA A23.2-28A. Les résultats obtenus avec cet essai, en présence
d'adjuvants à base de lithium, présentent une bonne corrélation avec les résultats obtenus
sur des blocs de grandes dimensions entreposés à l'extérieur [Fournier et al. 2003]. Cet
essai est réalisé à 38°C, dans l'air humide à plus de 95% d'humidité relative. Il a une durée
de deux ans lorsqu'il est utilisé pour évaluer l'efficacité d'adjuvants à base de lithium ou
d'ajouts cimentaires. Une telle durée est, dans bien des cas, considérée trop longue pour
l'industrie et il serait utile d'avoir un essai de plus courte durée permettant d'évaluer les
performances et le dosage requis.

Comme une augmentation de la température permet d'accélérer la cinétique de la RAS,


conserver les éprouvettes à plus haute température pourrait permettre de réduire la durée de
l'essai. Les travaux de DeGrosbois et Fontaine [2000], de Tourna et al. [2001] et de
Fournier et al. [2004] ont démontré qu'une version modifiée de l'essai ASTM C 1293
(température de 60°C) est une méthode adéquate pour tester la réactivité des granulats. Les
expansions entre deux et quatre mois à 60°C ont présenté une bonne corrélation avec les
expansions à un an à 38°C. Ainsi, cette augmentation de la température pourrait réduire la
période d'essai d'environ 75%, comme c'est le cas lorsque cette modification, encore peu
répandue, est utilisée pour tester des granulats. De plus, il a été démontré récemment que le
type de sable non réactif utilisé lors de l'essai a une importante influence sur l'expansion
finale à des températures de 38°C et de 60°C utilisées pour l'essai [Ideker et al. 2010].

Un autre essai, très populaire, est l'essai accéléré sur barres de mortier CSA A23.2-25A ou
ASTM C 1260, d'une durée de 14 jours. Il est très souvent utilisé afin de déterminer
l'efficacité des ajouts cimentaires face à la RAS (ASTM C 1567 ou CSA A23.2-28A)
[Thomas et al. 2007a]. Cet essai n'est toutefois pas approprié dans sa forme actuelle pour
évaluer l'efficacité des produits à base de lithium. En effet, l'immersion des barres dans la
solution de NaOH à 80°C entraîne le lessivage des ions lithium contenus dans la solution
interstitielle des mortiers, annulant ainsi tout effet positif. Des modifications à cet essai
devront être apportées avant que celui-ci permette de tester adéquatement l'efficacité des
adjuvants de lithium. De plus, l'essai modifié devra présenter une bonne corrélation avec
les résultats obtenus sur béton.

1.3 Objectifs

Jusqu'à maintenant, l'efficacité des adjuvants à base de lithium a montré qu'elle dépendait
fortement du ratio [Li]/[Na+K] utilisé dans le béton, mais aussi du type de produit à base de
lithium et du type de granulat réactif à contrer, sans que l'on comprenne exactement trop
bien pourquoi dans ce dernier cas. On constate donc que les mécanismes de réaction
impliquant le lithium ne sont pas encore bien compris. De plus, la plupart des études
réalisées sur le lithium en présence de granulats naturels se sont déroulées sur des granulats
provenant des États-Unis, souvent passablement différents en composition des granulats
communément reconnus comme réactifs au Canada. Il y a donc lieu de poursuivre les
recherches dans le cas des différents types de granulats réactifs canadiens et d'essayer
d'expliquer les causes de l'efficacité ou non de ces produits en présence de nos granulats.
À ce jour, le seul essai fiable disponible pour tester l'efficacité des adjuvants à base de
lithium est l'essai sur prismes de béton d'une durée de deux ans. Il est donc important pour
les producteurs de granulats et les fabricants d'adjuvants à base de lithium de disposer d'un
essai aussi fiable mais de plus courte durée.

Somme toute, le projet actuel de recherche porte sur l'utilisation des adjuvants à base de
lithium, en présence ou non d'ajouts cimentaires, pour contrer l'expansion du béton due à la
réactivité alcalis-silice (RAS). Les objectifs particuliers qui sont visés sont les suivants:

1. Déterminer l'efficacité de deux produits à base de lithium (le nitrate de lithium et le


verre de lithium) en présence d'une variété de granulats réactifs canadiens de
différentes compositions et de différents degrés de réactivité, et ce, en présence ou
non d'ajouts cimentaires (cendres volantes, laitiers de hauts fourneaux et ciment
ternaire) également considérés comme efficaces contre la RAS lorsqu'utilisés dans
des dosages appropriés;

2. Développer un essai rapide et fiable pour l'industrie afin de déterminer si un


adjuvant à base de lithium est efficace et à quel dosage;

3. Déterminer les mécanismes à l'origine de cette efficacité en tenant compte du


produit testé (type et dosage) et de la nature du granulat réactif à contrer
(composition et degré de réactivité).

1.4 Contribution scientifique des travaux

Bien que l'effet bénéfique du lithium pour contrer la RAS soit connu depuis les années 50,
l'utilisation des adjuvants à base de lithium demeure une des plus récentes innovations pour
la prévention chimique de la RAS dans le béton.

Bien que de nombreuses études aient démontré l'efficacité des adjuvants à base de lithium
et en particulier du nitrate de lithium, plusieurs zones grises restent à éclaircir afin d'utiliser
ces produits de façon éclairée et judicieuse. Les travaux de recherche réalisés dans le cadre
de ce doctorat permettront entre autres:
D'améliorer de façon significative l'ensemble des connaissances sur les
performances des adjuvants à base de lithium pour contrer la RAS;

D'évaluer l'efficacité de ces adjuvants face aux divers granulats réactifs canadiens
qui ont été, jusqu'à maintenant, très peu testés;

De proposer un nouvel essai ou un essai modifié permettant de prédire de façon


accélérée et fiable l'efficacité de ces produits;

De modifier, selon les résultats obtenus, les normes canadiennes existantes et ce, en
proposant un critère analytique décisionnel basé sur des dosages minimaux
d'adjuvants à base de lithium requis; ces dosages seraient fonction de la nature et du
degré de réactivité des granulats; à défaut de se conformer à ce critère, un essai
accéléré serait proposé afin d'évaluer la performance des produits à base de lithium.

Sous réserve d'une bonne utilisation des résultats de cette étude, ceux-ci contribueront à
augmenter la durée de vie utile des prochains ouvrages en béton. Malgré le fait que certains
granulats possèdent un degré de réactivité élevé, ces granulats pourraient être utilisés
adéquatement avec un dosage équilibré des adjuvants à base de lithium, ce qui diminuerait
les coûts monétaires et environnementaux du transport qu'exigerait l'approvisionnement en
granulats non-réactifs, en ciment à faible teneur en alcalis ou encore en ajouts cimentaires.
Dans certaines régions, les granulats non-réactifs se retrouvent parfois à des distances très
grandes. Si, dans ces régions, les sources d'ajouts cimentaires se retrouvent également
éloignées, ou si leur utilisation pouvait entraîner certains problèmes au niveau de la
durabilité du béton (résistance à l'écaillage...), l'utilisation des adjuvants chimiques pour
contrer la RAS pourrait devenir une solution très intéressante si ces produits étaient utilisés
adéquatement.
1.5 Publications scientifiques composant la thèse

Les articles suivants font partie intégrante de la présente thèse:

[1] BÉRUBÉ, M.A., TREMBLAY, C , FOURNIER, B., THOMAS, M.D.A. et D.B.


STOKES (2004): Influence of lithium products proposed for counteracting ASR on
the chemistry of cement hydrates and pore solution. Cement and Concrete
Research, Vol. 34, pp. 1645-1660.

[2] TREMBLAY, C , BÉRUBÉ, M.A., FOURNIER, B., THOMAS, M.D.A. et


FOLLIARD K.J. (2007): Effectiveness of lithium-based products in concrete made
with Canadian natural aggregates susceptible to alkali-silica reactivity. ACI
Materials Journal, Vol. 104, No. 2, pp. 195-205.

[3] TREMBLAY, C , BÉRUBÉ, M.A., FOURNIER, B., THOMAS, M.D.A.,


FOLLIARD K.J. et NKINAMUBANZI, P.C. (2008): Use ofthe accelerated mortar
bar test to evaluate the effectiveness ofLiNOi against alkali-silica reaction - Part
I: Influence of various parameters. Journal of ASTM International (JAI), Vol. 5,
No. 8, 17 p.

[4] TREMBLAY, C , BÉRUBÉ, M.A., FOURNIER, B., THOMAS, M.D.A.,


FOLLIARD K.J. et NKINAMUBANZI, P.C. (2008): Use ofthe accelerated mortar
bar test to evaluate the effectiveness of LiNOs against alkali-silica reaction - Part
II: Comparison with the concrete prism test results. Journal of ASTM International
(JAI), Vol. 5, No. 8, 21 p.

[5] TREMBLAY, G, BÉRUBÉ, M.A, FOURNIER, B. et THOMAS, M.DA. (2010):


Experimental study on the reaction mechanisms explaining the effectiveness of
LiNOs against ASR. Cement and Concrete Research, Vol. 40, pp. 583-597.
Les autres articles suivants ont été publiés durant les travaux de recherche. Malgré le fait
que ces publications portent sur des résultats obtenus lors de cette étude, ils ne font pas
partie intégrante de cette thèse afin d'éviter les répétitions et de conserver un ordre logique
dans la présentation des résultats.

TREMBLAY, C , BÉRUBÉ, M.A., FOURNIER, B. et THOMAS, M.D.A. (2004):


Performance of lithium-based products against ASR: application to Canadian aggregates.
lx CANMET/ACI International Conference on Recent Advances in Concrete Technology,
Las Vegas (USA), Mai 2004, pp. 247-268.

TREMBLAY, C , BÉRUBÉ, M.A., FOURNIER, B., THOMAS, M.D.A. and STOKES D.


(2004): Performance of lithium-based products against ASR: application to Canadian
aggregates, reaction mechanisms and testing. 12l International Conference on Alkali-
Aggregate Reaction (ICAAR), Beijing (Chine), Octobre 2004, pp. 668-677.

BÉRUBÉ, M.A. et TREMBLAY, C. (2004): Chemistry of pore solution expressed under


high pressure - Influence of various parameters and comparison with the hot-water
th

extraction method. 12 International Conference on Alkali-Aggregate Reaction (ICAAR),


Beijing (Chine), Octobre 2004, pp. 833-834.
TREMBLAY, C , BÉRUBÉ, M.A., FOURNIER, B., THOMAS, M.D.A. et STOKES D.
(2006): Effectiveness of lithium-based products in concrete made with Canadian
aggregates susceptible to ASR. Marc-André Bérubé symposium on alkali-aggregate
reactivity in concrete, Montréal (Canada), Mai 2006, pp. 217-240.

TREMBLAY, C , BÉRUBÉ, M.A., FOURNIER, B. et THOMAS, M.D.A. (2008):


Experimental investigation of the mechanisms by which LiNOs is effective against ASR.
13l International Conference on Alkali-Aggregate Reaction (ICAAR), Trondheim
(Norvège), Juin 2008, 10 p.
10

2 Revue de la littérature

2.1 Les conditions essentielles à la RAS

Tel que vu précédemment, trois conditions simultanées sont essentielles au développement


de la RAS au sein du béton. Le granulat (pierre et/ou sable) composant le béton doit être
réactif, la concentration en ions alcalins doit dépasser un seuil critique dans la solution
interstitielle et l'ouvrage en béton doit être exposé à une certaine humidité [Fournier et
Bérubé 2000]. Voyons maintenant en détail ces trois conditions.

2.1.1 Les gran ulats réactifs

La première condition essentielle pour qu'il y ait développement d'une réaction alcalis-
silice est la présence d'un granulat potentiellement réactif dans le béton. Un granulat est un
matériau granulaire, c'est-à-dire formé d'un ensemble de particules qui sont, dans le cas
présent, des particules de roches qui viennent principalement de deux sources, soit les
dépôts meubles, c'est-à-dire les sables et graviers, soit des roches massives concassées
(carrières). On regroupe les types de roches en trois grands groupes:

• Les roches ignées issues de la cristallisation d'un magma à l'intérieur (intrusives) ou


à l'extérieur (extrusives) de la surface de la croûte terrestre;

Les roches sédimentaires issues de l'interaction entre les processus de l'altération


superficielle, l'érosion, le transport et le dépôt de particules, et la diagenèse;

• Les roches métamorphiques résultant de la transformation de roches ignées


,sédimentaires ou déjà métamorphisées sous l'effet de températures et/ou de
pressions élevées.

On entend par granulats réactifs que les particules de roches contiennent des phases
minérales, généralement de la silice ou des silicates, instables aux pH très basiques (>13) de
la solution interstitielle du béton. Les roches étant composées d'un assemblage de différents
minéraux, tous les types de roches (ignées, métamorphiques ou sédimentaires) possèdent
des variétés pouvant contenir une ou des phases de silice dites « réactives ». Il faut bien
11

comprendre que toutes les formes de silice sont réactives, mais à des degrés très différents
d'une variété à l'autre. Ces phases de silice sont réactives selon leur degré de cristallinité et
leur texture. Les phases siliceuses considérées comme réactives sont souvent faiblement
cristallisées, hydratées, plus poreuses et assez solubles à pH élevé [Bérubé et al. 1988]. Le
mécanisme de réaction serait sensiblement le même, ce serait plutôt la vitesse de réaction
qui différencierait les variétés de silice entre-elles. Ainsi, les principales phases minérales
déjà reconnues comme réactives sont:

1. Silice faiblement cristallisée: verre volcanique et verre artificiel;

2. Silice métastable: cristobalite et tridymite;

3. Silice hydratée: opale;

4. Quartz cryptocristallin: calcédoine et chert;

5. Quartz microcristallin: verre volcanique dévitrifié et autres variétés de quartz


microcristallin;

6. Ciments sédimentaires: quartz recristallisé;

7. Quartz métamorphisé: quartz granulaire à extinction roulante.

Ces différentes variétés de silice ont des propriétés souvent différentes les unes des autres.
Le tableau suivant présente les types de roches pouvant contenir ces phases minérales et
reconnues comme réactives.
12

Tableau 2.1 ­ Types de roches potentiellement réactives aux alcalis du ciment

Phase minérale Types de roches


réactive et degré de
réactivité Ignées Sédimentaires Métamorphiques
Shales, grès, calcaires
Silice faiblement
Obsidienne silicifiés, chert, silex,
réactive

cristallisée
Très

argilite
Rhyolites à basaltes +
Silice métastable équivalents pyroclastiques

Cherts à calcédoine, grès


0» Quartz quartzi tiques,
Granites
cryptocristallin calcaires siliceux,
uR dolomies

U
Quartz Rhyolites à basaltes + Micaschistes,
e microcristallin équivalents pyroclastiques
Calcaire siliceux
quartzites

E Ciments Grès, orthoquartzites, Quartzites,


— sédimentaires siltstones schistes à chlorite
e» Schistes à quartz et
Quartz granulaire à Grès, grauwackes, chlorite,
o
■ * *

Granites
û­ extinction roulante arkoses gneiss granitiques à
granodioritiques

2.1.2 La teneur en alcalis

Ya présence d'une certaine quantité d'alcalis est essentielle afin que la RAS ait lieu. Les
alcalis proviennent principalement du ciment, mais peuvent aussi être fournis par les
granulats, les ajouts minéraux [Duchesne et Bérubé 1992], les additifs chimiques [Okada et
al. 1987] et même par des sources externes comme les sels déglaçants [Grattan­Bellew
1981, Rogers 1987]. On considère toutefois que la majorité des alcalis disponibles pour la
réaction proviennent du ciment portland. Diamond (1989) a en effet montré qu'il y avait
une relation directe entre la teneur en alcalis du ciment et le pH de la solution interstitielle
de pâtes de ciment durcies. Ces alcalis se retrouvent généralement sous forme de sulfates et
de carbonates. On exprime généralement la quantité d'alcalis d'un ciment en pourcentage
d'oxyde de sodium équivalent, ce qui se traduit par l'équation suivante:

%Na 2 O é = %Na 2 0 + 0,658»%K2O


13

Le facteur multipliant le pourcentage d'oxyde de potassium est le rapport des poids


molaires de l'oxyde de sodium sur l'oxyde de potassium (62 g/mole/94 g/mole). La plupart
des ciments portland utilisés en Amérique du Nord possèdent entre 0,4 et 1,1% de Na20é.
La quantité de ciment normalement utilisée varie généralement de 250 à 450 kg/m3 de
béton, selon l'utilisation. Ainsi, un mètre cube de béton contiendra normalement entre
I et 4,5 kg/m3 de Na2Oé. Si cette quantité ne dépasse pas un seuil critique, la RAS sera
très limitée même si les autres conditions sont présentes. Cette limite se situe pour
certains auteurs à environ 3 kg/m3 de Na2Oé [Folliard et al. 2003], tandis que la norme
CSA A23.2-27A recommande une limite de 2,4 kg/m3 pour la plupart des ouvrages.
Toutefois, cette même norme limite la teneur à 1,8 kg/m3 (et même 1,2 kg/m3) de Na20é
pour les ouvrages plus problématiques, c'est-à-dire massifs, d'une très longue durée de vie
et/ou exposés continuellement à l'humidité. On remarque donc que cette limite varie selon
la réactivité du granulat, le type d'ouvrage, la durée de vie et l'exposition à l'humidité des
ouvrages. On peut toutefois affirmer que la RAS se manifeste plus rarement lorsque le
contenu en alcalis du béton est sous la barre des 3 kg/m3 de Na20é.

II faut toutefois faire une nette distinction entre la quantité d'alcalis totaux et la quantité
d'alcalis actifs. Les alcalis totaux sont la quantité d'alcalis retrouvés dans le mélange
original de béton tandis que les alcalis actifs sont les alcalis retrouvés dans la solution
interstitielle du béton. Ce sont ces derniers qui jouent un rôle dans la réaction alcalis-
granulats. Lors de l'hydratation des matériaux cimentaires, une partie des alcalis du ciment
et des ajouts cimentaires est incorporée dans les hydrates tandis qu'une autre partie est
relâchée dans la solution interstitielle du béton. Environ 50% des alcalis du ciment se
retrouvent dans la solution interstitielle [Brouwers et Van Eijk, 2003].

La concentration des alcalis dans la solution interstitielle évolue toutefois avec le temps. La
figure 2.1 illustre l'évolution de la concentration des alcalis dans la solution interstitielle
d'une pâte de ciment. Cette concentration augmente progressivement dans les premiers
jours jusqu'à quelques semaines pour se stabiliser par la suite [Longuet et al. 1973]. Lors de
l'hydratation du ciment, il y a formation de silicates de calcium hydratés (CSH), de
portlandite et de sulfoaluminate (ettringite). Au départ, la solution interstitielle devient vite
saturée en ions calcium (Ca +) et en ion sulfates (SO42). La formation des CSH et de
14

l'ettringite consomme rapidement ces ions et il y a donc, par la suite, dissolution des
sulfates et des carbonates alcalins libérant ainsi les ions alcalins. L'augmentation de la
concentration en ions alcalins favorise la libération d'ions OH" afin de maintenir un
équilibre des charges dans la solution [Vernet 1992]. De ce fait, le pH devient plus élevé, ce
qui favorise l'attaque des phases siliceuses et donc la réaction alcalis­granulats.

Le maintien de l'humidité dans le béton permet donc aux matériaux cimentaires de


poursuivre leur hydratation, l'hydratation additionnelle et la chute correspondante de la
teneur en eau ayant pour effet d'augmenter la concentration des ions alcalins dans la
solution interstitielle et ainsi le pH (augmentation correspondant des ions OH").

0.80 JM I I iinii IIIII I

0.70
O)
£ 0.60
"5>
"5 0.50
E.
"g 0.40

■g 0.30
0)
o
o 0.20

0.10­

0.00
10 100 1000
Age (jours)

Figure 2.1 ­ Évolution chimique de la solution interstitielle d'une pâte de ciment


contenant 1.0 Na20e et un rapport E/C de 0.5 [selon Longuet et al. 1973].
15

2.1.3 Le taux d'humidité

Ya dernière condition essentielle pour que la réaction alcalis-silice se produise dans un


béton est que celui-ci soit exposé à une humidité assez élevée. Le rôle de l'eau dans la RAS
est très important [Larive, 1998]; l'eau permet la diffusion des espèces ioniques et entre
dans la composition même des produits causant l'expansion. En effet, l'humidité permet au
gel silico-calco-alcalin formé d'absorber de l'eau. Ce faisant, ce gel gonfle et provoque
ainsi l'expansion et la fissuration du béton. On considère généralement qu'une humidité
relative d'environ 80-85% est suffisante pour que la RAS se produise. Rappelons, toutefois,
avant de poursuivre, que le pourcentage d'humidité relative est le pourcentage d'humidité
ou de vapeur d'eau contenue dans l'air par rapport à la quantité maximum que peut contenir
cet air avant de se condenser, c'est-à-dire avant de se transformer en gouttelettes d'eau. Plus
la masse d'air est chaude, plus elle peut absorber de vapeur d'eau sans se condenser alors
que plus elle est froide, moins elle peut en contenir. Deux masses d'air de températures
différentes contiendront donc une quantité de vapeur d'eau différente, même si l'humidité
relative des deux masses d'air est équivalente.
16

T
80 90
% Humidité Relative

Figure 2.2 - Effet de l'humidité relative sur l'expansion du béton à 2 ans à 38°C (essai
CSA A23.2-14A) [tirée de Pedneault 1996].

On constate à la figure 2.2 que la limite d'expansion tolérée dans le cadre de l'essai sur
prismes de béton CSA A23.2-14A, soit 0,040% à un an, est généralement dépassée lorsque
le béton est soumis à des conditions d'humidité relative supérieures à 80% (90% pour le
grès de Potsdam), à 38°C. 11 faut toutefois comprendre qu'à une humidité relative plus
faible, la RAS a lieu mais à un rythme plus lent. De plus, même si un ouvrage se retrouve
dans une région moins humide (moins de 80% d'humidité), il est possible que la RAS se
développe si elle est en contact direct avec un sol humide ou avec une source externe d'eau
(barrage, béton de fondation au contact de la nappe phréatique, etc.). Dans ce cas, le taux
d'humidité dans l'ouvrage sera très élevé, sauf dans les quelques centimètres de la surface
de cet ouvrage. Il en est de même pour les ouvrages de béton de grandes dimensions, car
ceux-ci possèdent une quantité d'eau interne suffisante pour provoquer le gonflement du
béton. En effet, la plupart des bétons possèdent un rapport eau/ciment supérieur à celui
17

nécessaire pour hydrater tout le ciment. Par conséquent, il demeure une grande quantité
d'eau dans le béton, laquelle est disponible pour la RAS et ce, même sous un climat
désertique [Stark 1991].

Tel qu'expliqué auparavant, le taux d'humidité est toujours relatif à la température. Ainsi,
une température plus chaude permet de contenir une plus importante quantité d'humidité
dans l'air, et vice versa. Pour un même taux d'humidité, un ouvrage dans une région chaude
a plus de risque de développer de la RAS qu'un ouvrage soumis au même taux d'humidité,
mais à des températures plus froides. Toutefois, on doit aussi tenir compte que la vitesse de
réaction de la RAS augmente avec la température.
18

2.2 Les mécanismes de la réaction alcalis-silice

Les mécanismes impliqués lors de la réaction alcalis-silice sont encore mal compris, mais la
majorité des auteurs s'entendent pour dire que l'on peut séparer la réaction en trois grandes
étapes. Premièrement, il y a une étape de dissolution des phases siliceuses, suivie de la
formation d'un gel silico-calco-alcalin, et d'une dernière étape qui est le processus
d'expansion lui-même. Les mécanismes de réaction généralement acceptés pour la RAS
sont résumés ici et décrits plus en détail par la suite.

1. Dissolution de la silice. Il y a concentration des ions alcalins et hydroxyles dans la


solution interstitielle lors de la progression de l'hydratation, puis migration de ceux-
ci vers les phases réactives. Il y a par la suite réaction de neutralisation acide-base,
c'est-à-dire que les ions hydroxyles (OH) attaquent les groupes silanols (Si-OH),
abondants à la surface des phases siliceuses, et les ponts siloxanes (Si—O—Si), plus
résistants et situés plus en profondeur;

2. Formation d'un gel silico-calco-alcalin. Lorsque les réactions précédentes ont lieu,
les hydroxydes alcalins pénètrent de plus en plus profondément dans la structure de
la silice, attaquant progressivement la phase réactive du granulat et formant des
espèces silicatées plus ou moins polymérisées. Ces espèces réagissent ensuite avec
le calcium présent dans la solution interstitielle ou dans la pâte de ciment,
produisant ainsi un gel silico-calco-alcalin;

3. Expansion. La dernière étape est le mécanisme même de gonflement. Malgré le fait


que les auteurs ne s'entendent pas sur les mécanismes associés au phénomène de
gonflement même du béton, il existe trois théories principales. La première est que
le gel silico-calco-alcalin gonfle en absorbant de l'eau et crée ainsi des pressions
d'imbibition. Une deuxième théorie propose que ce sont plutôt les pressions
osmotiques provenant de la solution silico-alcaline ou de la silice résiduelle qui sont
à l'origine des gonflements. Une troisième théorie proposée afin d'expliquer
l'expansion fait appel à la répulsion électrostatique reliée au phénomène de la
double-couche électrique présente autour des particules chargées (gel ou silice).
19

2.2.1 Dissolution et solubilité de la silice

Premièrement, il faut faire la distinction entre la dissolution de la silice et la solubilité de


celle-ci. La dissolution est l'attaque de la silice par les ions hydroxyles qui mettent peu à
peu la silice sous forme d'espèces ioniques en solution, tandis que la solubilité est la
quantité maximale de silice que l'on peut dissoudre par unité de volume aux conditions
existantes (pH, température, pression, nature de la charge ionique, etc.).

Les premiers travaux importants sur la silice ont été publiés par lier [1956]. Ces travaux
portaient sur la solubilité, la polymérisation et les propriétés de surface de la silice. Ils sont
toutefois difficilement applicables à la RAS dans le béton, car ils ont été réalisés à des pH
beaucoup plus faibles.

Les premiers travaux portant sur la dissolution de la silice directement reliée à la réaction
alcalis-silice ont été réalisés par Dent-Glasser et Kataoka [1981] et poursuivis par d'autres
auteurs [Wijnen et al. 1989, Michaud 1995]. Ils ont permis d'identifier les réactions
chimiques gouvernant la dissolution de la silice en milieu fortement basique. La figure 2.3
montre les liens ioniques d'une particule de silice pure. Ainsi, en surface, on retrouve les
groupes silanols (Si-OH) et, en profondeur, les ponts siloxanes (Si - O - Si).

HO HO OH

Surlace sPr-Of s t ""O-Si.


'
PY î % J >

Figure 2.3 - Surface d'une particule de silice pure.


20

Il y a deux réactions principales lors de l'attaque de la silice par les ions hydroxyles:

Réaction 1 - Attaque des groupes silanols

Les ions hydroxyles (OH) attaquent les groupes silanols (Si-OH) qui abondent à la surface
des phases siliceuses. C'est ainsi qu'une molécule libre d'H20 sera libérée, alors qu'un ion
positif « M » (K+, Na+, Ca++) se combinera à la charge négative de l'atome d'oxygène de
surface (équation 1):

S i - O H + Off + (M+) => - Si - OM + + H 2 0 [1]

Réaction 2 - Attaque des ponts siloxanes

Pendant ce temps, les ponts siloxanes, plus solides (Si - O - Si), sont aussi attaqués par les
ions hydroxyles. Les siloxanes se brisent et leur charge négative est neutralisée par
un ion alcalin tout en libérant de la silice qui passe en solution sous forme
d'espèces ioniques silicatées (HiSi04, H3Si04', HzSiO/"), le tout dépendant du pH de la
solution (équation 2).

Si - O - Si + OH => 2 Si - OM + + HO- Si [2]

Dans une solution contenant seulement des ions hydroxyles, des ions alcalins (Na ou K) et
de la silice, il est possible de mettre en solution des quantités importantes de silice lorsque
celle-ci est en présence d'une solution hautement basique. On peut observer la solubilité de
la silice amorphe en fonction du pH à la figure 2.4. On voit qu'en passant d'un pH de 12 à
13, la solubilité de cette silice est d'environ dix fois supérieure. La solubilité de la silice
augmente aussi avec la température (en l'absence toutefois de portlandite ou de chaux, i.e.
d'une source d'ions OH- permettant d'empêcher le pH de chuter, comme dans le cas des
systèmes cimentaires).
21

.u 11

60­
__T
§
£50­
"^■^

»
u
% 40"
JO J
u I
■o 30­
I
> I
1 20­ I
3 I
10­

J >
. + — — — • ^ ­ * ^
8
o­ [ I I I I I I I I
a5 9.0 9.5 10.0 10.5 11.0 11.5 1Z0 1Z5 110
PH

Figure 2.4 ­ Solubilité de la silice amorphe en fonction du pH [tirée de Tang et Su­fen


1980].

Dent­Glasser et Kataoka [1981] ainsi que Gaboriaud [1999], plus récemment, ont démontré
que dans un tel système, il y a une quantité optimale de silice que l'on peut dissoudre dans
un volume donné de solution. Sur la figure 2.5, on peut voir cet optimum, situé à environ
2,4 mol/L de silice dans une solution IM. Les essais correspondants ont été réalisés en
ajoutant des quantités de silice solide (sous forme amorphe) à des solutions contenant
1 mole par litre de divers hydroxydes (en l'absence toutefois de portlandite, i.e. d'une
source d'ions OH­ empêchant le pH de chuter, comme dans le cas des systèmes
cimentaires).
22

2one1

1.1
1. :: •©•• N»OH
"j
0.9
>:!!|!!!M^. •A­ KOH
•l>­ LiOH
Ï °'8" V RbOH
•♦• C»OH
% 0.6­j
8 0.5­

1 ° 3 " (b) — * i : |
â 0.2­
i—■—p ­i—j­ 1 ' 1 «~
1 2 4

o­. NiOH
«■£ ■A­ KOH
»• LiOH
13­
▼• RbOH

% 12 H
'X, • ••■ & ' ( ^ ■ • • M | H
■«■• CtOH

' <»» ■■■A


. • . . Q. . * • • • » • • ^ • . • . . { • U l K l . X l • • • • t f H

(c)
10 ■«—r
i ­1—r ">—r
4
2
quantité de silice initiale introduite
(mole par litre de tolution)

Figure 2.5 ­ Évolution (a) de la quantité de silice dissoute, (b) de la concentration en ions
alcalins en solution et (c) du pH en fonction de la quantité de silice initiale
dans différentes solutions d'hydroxydes alcalins de concentration 1 mol/L
(pH = 14,0) [tirée de Gaboriaud 1999].
23

On peut séparer les graphiques de la figure 2.5 en deux zones : d'une part, la zone 1 où la
quantité de silice initiale introduite est complètement dissoute et, d'autre part, la zone 2 où
la silice n'est pas complètement dissoute. Ainsi, une certaine quantité de silice solide
demeure dans la solution. On peut voir à la figure 2.5a que la quantité de silice en solution,
dans la zone 1, est approximativement égale à la quantité de silice initialement introduite,
car toute la silice est dissoute. Toutefois, la quantité de silice dissoute diminue dans la
zone 2, même si de plus grandes quantités de silice sont ajoutées à la solution. Tel
qu'indiqué par Gaboriaud [1999], ce phénomène est attribuable à l'adsorption des alcalis
sur les surfaces chargées négativement de la silice non dissoute. En effet, à la figure 2.5b,
qui représente la concentration d'alcalis restante dans la solution, on remarque que la
quantité d'alcalis est toujours stable lorsqu'est complètement dissoute la silice (zone 1),
mais diminue rapidement dans la zone 2. Cette diminution est expliquée par le fait que les
alcalis sont adsorbés à la surface de la silice non dissoute et ne sont plus présents dans la
solution. On peut voir sur le même graphique que les ions alcalins semblent adsorbés
préférentiellement selon l'ordre dans le tableau périodique (équation 3).

Na < K < Li < Rb < Cs [3]

Le phénomène de « pessimum effect », où une quantité optimale de silice peut être dissoute
et créer un maximum d'expansion, est observé dans le béton lors d'essais accélérés sur
barres de mortier [Shayan 1992].

À la figure 2.5c, on peut observer que le pH diminue rapidement plus on dissout de silice,
ce qui est normal, car la dissolution de la silice est une réaction de neutralisation acide-
base. Selon les conditions expérimentales de Gaboriaud [1999], une valeur minimale de pH
d'environ 11 est atteinte peu importe la solution utilisée.

Il faut toutefois spécifier que dans un béton, la quantité de silice en solution est toujours
très faible dû à la présence des ions calcium qui provoquent la formation d'un gel de silice
dès que de la silice passe en solution.
24

2.2.2 Formation d'un gel silico-calco-alcalin

L'ion calcium, normalement présent dans la solution interstitielle d'un béton, joue un rôle
important dans la formation du gel de RAS. En effet, la présence de portlandite dans la pâte
de ciment permet de maintenir une certaine concentration de calcium en solution, en
fonction du pH de la solution interstitielle ainsi que de la température. Cette concentration
est toutefois assez importante pour déstabiliser la solution silico-alcaline. Dent-Glasser et
Kataoka [1981] ont étudié le rôle du calcium sur la RAS et sont arrivés à la conclusion que
la présence de Ca (OH)2 en saturation dans une solution limite la quantité de silice en
solution en favorisant la précipitation de C-S-H (silicate de calcium hydraté). Comme l'ion
calcium possède une double charge ionique, sa liaison avec les espèces silicatées serait
favorisée face aux ions alcalins selon la réaction suivante (équation 4):

Si - O- + Ca++ + H 2 0 => Si - O - Ca+ + O - Si + H 2 0 => Si - O - Ca - O - Si + H 2 0 [4]

Selon Gaboriaud et al. [1999], le calcium en solution aurait un rôle de cataliseur en


favorisant la liaison des ions silicates durant le processus de gélification.

Les analyses chimiques de ces C-S-H démontrent qu'ils contiennent aussi des ions alcalins
sous des proportions très variables. Toutefois, selon certains auteurs [Davies et Oberholster
1988, Lombardi et al. 1996], les ions alcalins présents dans le gel ne feraient pas partie
intégrante de la structure du gel et seraient plutôt présents dans la solution interstitielle qui,
pour sa part, est absorbée par ce dernier.

Diamond (1989) a proposé qu'en l'absence de calcium, la silice est simplement dissoute et
ne forme pas de gel. Selon Thomas [1998], une expansion significative apparaît seulement
si de la portlandite Ca(OH)2 est disponible, et les systèmes sans calcium disponible
démontreraient peu d'expansion. La silice ainsi dissoute demeure en solution. La présence
de l'ion calcium apparaît donc comme essentielle à la formation du gel gonflant.
25

2.2.3 Les mécanismes d'expansion

Si la majorité des scientifiques s'entendent sur les mécanismes à la base de la réaction


alcalis-silice, il n'en est pas de même pour le phénomène d'expansion associé à la RAS. On
s'entend pour dire que le produit de réaction est une substance colloïdale, soit un gel
(C-S-H) contenant du silicium, des alcalis et du calcium. Ce gel , de composition chimique
variable, possède différentes propriétés selon son contenu en alcalis et/ou en calcium
[Diamond 2000]. Par contre, les idées sont partagées quant aux mécanismes d'expansion de
ce gel. Trois principales théories sont proposées pour expliquer les gonflements
généralement observés sur les bétons atteints de RAS. Une théorie évoque la pression
d'imbibition, une autre évoque les pressions osmotiques tandis que la dernière fait appel à
la théorie de la double-couche électrique pour expliquer le gonflement.

2.2.3.1 Pression d'imbibition

Selon certains chercheurs [McGowan et Vivian 1952, Powers et Steinour 1955, Tang
1981], le gonflement du béton dû à la RAS serait attribuable au simple fait d'imbibition du
gel. Une pression d'imbibition se produit lorsqu'un corps s'imbibe de liquide. Dans le cas
de la RAS, le corps est le gel silico-calco-alcalin et le liquide, la solution interstitielle du
béton. Lorsque l'eau pénètre par imbibition dans le gel visqueux de silicate alcalin, les
pressions deviennent alors importantes au moment où les pores et les fissures sont remplis
de gel. Les pressions augmentent aussi si le taux d'imbibition du gel est plus grand que la
vitesse de déplacement du gel dans le milieu poreux. Lorsque les pressions dépassent la
résistance en traction de la pâte ou des granulats, il y a fissuration et gonflement du béton.
L'apparition de la fissuration facilite la diffusion et favorise ainsi la poursuite des réactions.
26

2.2.3.2 Pression osmotique

Cette théorie, basée sur les pressions osmotiques, a été proposée en premier lieu par Hansen
[1944] et soutenue depuis par différents chercheurs [Powers et Steinour 1955, Dent-Glasser
1979, Struble et Diamond 1981, Chatterji et al.1986]. On parle d'osmose lorsqu'il y a une
différence de concentration ionique entre deux solutions séparées par une membrane semi-
perméable, c'est-à-dire perméable uniquement à l'eau (ou au solvant de façon plus
générale) et imperméable aux solutés. Lorsque les deux solutions ne contiennent pas le
même nombre de particules dissoutes par unité de volume, on observe un mouvement d'eau
qui va tenter de compenser cette différence de concentration en diluant le compartiment le
plus concentré. Lors de la migration des ions, il y a apparition de pressions dites
osmotiques. Même s'il n'y a pas de consensus sur l'origine de la membrane semi-
perméable, la théorie des pressions osmotiques peut généralement être résumée ainsi:

Premièrement, il y a réaction superficielle entre les ions hydroxyles, alcalins et calcium


avec les phases siliceuses et formation d'un gel, comme vu précédemment. Par contre, ce
gel, un silicate de calcium hydraté alcalin, serait peu ou non gonflant. Ce gel ou tout
simplement la pâte de ciment [Dent-Glasser, 1979] formerait la membrane semi-perméable
en bordure de la phase siliceuse. À mesure que la réaction se poursuit à l'intérieur du
granulat, il se développe une différence de concentration des ions (Na+, K+, Ca^, OH" et la
silice dissoute) entre la solution à l'intérieur du granulat et celle à l'extérieur. Cette
différence de concentration produit une pression osmotique qui, lorsque supérieure à la
résistance en traction de la pâte de ciment et/ou du granulat, crée leur fissuration. Dans ce
modèle, le produit de réaction n'a pas besoin d'être expansif, car l'expansion est causée par
les pressions osmotiques.
27

2.2.3.3 Double couche électrique

La plus récente théorie afin d'expliquer les mécanismes d'expansion repose sur le modèle
de la double couche électrique appelé communément le modèle de Gouy-Chapman-Stern
[Gouy 1910, Chapman, 1913, Stern 1924]. La théorie s'applique au gel de silice qui est un
colloïde. Un colloïde se définit comme une substance en suspension par interaction
électrostatique avec les molécules d'eau. Les particules formant un colloïde sont très petites
et ont un diamètre entre 0,1 à 0,001 micromètre.

Les surfaces hydroxyles des oxydes et des silicates peuvent perdre ou gagner un proton,
résultant de la charge de surface qui varie selon le pH.

À faible pH, la surface hydroxyle gagne un proton : Si=OH + H+ —► Si=OH2+ et la


surface devient cationique (chargée positivement);

À pH élevé, la surface hydroxyle perd un proton : Si=OH + OH" —» Si=0" + H2O


et la surface devient anionique (chargée négativement).

Le produit formé lors de la réaction alcalis­silice est un gel de silice dans un milieu très
basique (pH > 12,5); la surface est donc chargée négativement. Les charges négatives
attirent les ions positifs libres dans la solution interstitielle. Ceux­ci sont étroitement collés
aux colloïdes et forment la couche liée ou de STERN (figure 2.6) qui attire à son tour des
anions accompagnés d'une faible quantité de cations: c'est la couche diffuse « diffuse
layer » ou de GOUY­CHAPMAN. Il y a donc formation d'une double couche ionique, l'une
accompagnant la particule lors de ses déplacements, l'autre se déplaçant indépendamment
ou avec un certain retard. La couche de STERN, immobile et dense, est formée de cations
hydratés (Ca, Na, K. Li, Al, etc.). L'habilité à pénétrer cette couche pour atteindre la
surface varie avec le rayon hydraté des cations.

Il existe entre les deux couches un potentiel électrostatique qui varie en fonction de la
distance par rapport à la surface du colloïde. Dans la couche liée, le potentiel décroît
linéairement, car les cations constitutifs sont empilés uniformément. En revanche, dans la
couche de GOUY­CHAPMAN, le potentiel électrostatique varie de manière non linéaire,
étant donné que la répartition ionique résulte d'un mélange aléatoire de cations et d'anions.
28

La valeur du potentiel à la surface de la couche de STERN est appelée « potentiel zêta ».


Les colloïdes étant chargés négativement, ce potentiel est négatif. Les particules de
potentiel zêta négatif se repoussent très fortement. C'est pourquoi les colloïdes sont très
stables et inhibent toute agglomération. La figure 2.6 illustre le principe de la double
couche électrique pour une surface chargée négativement.

©
© © ©
©
©
© ©
©
©
—I
Couche de Stem Couche diffuse de Gouy-Chapman

Figure 2.6 - Modèle de la double couche.

Chatterji et Kawamura [1992] ont été les premiers à proposer que cette théorie pourrait
jouer un rôle lors de la RAS, expliquant entre autres le contenu élevé en calcium dans les
produits de réaction. Toutefois, c'est Prezzi et al. [1997] qui ont proposé la théorie de la
double couche comme explication du gonflement du gel formé lors de la RAS. Selon cette
théorie, dans un système colloïdal (gel) donné, plus la valence des ions dans la double
couche est grande, ou plus la concentration de ces ions est grande, plus l'épaisseur de la
double couche est petite ainsi que les forces répulsives qui peuvent être produites en
présence de l'eau (solution aqueuse). Cela signifie que les ions monovalents (Na+ et K+)
produisent une double couche plus épaisse et plus de force de répulsion que les ions
bivalents (Ca++). Ainsi, un gel contenant du sodium serait plus expansif et causerait de plus
grandes pressions qu'un gel contenant uniquement du calcium. Les travaux de Rodrigues et
al. [1999 et 2001] ont aussi porté sur cette théorie pour expliquer le gonflement des gels
associés à la RAS. Cette théorie indique donc que le gonflement observé lors de la RAS est
provoqué par le gel lui même, comme la théorie de la pression d'imbibition, mais en
incluant l'importance de la nature des ions sur les pressions engendrées.
29

2.3 Les ajouts cimentaires pour contrer la RAS

Il est reconnu que l'introduction d'ajouts cimentaires (cendres volantes, laitiers de hauts
fourneaux, fumées de silice et pouzzolanes naturelles) dans les bétons, en addition ou en
substitution partielle au ciment Portland, si elles respectent certaines conditions qui seront
discutées, réduit ou annule l'expansion provoquée par la RAS. Les ajouts cimentaires sont
des matériaux qui, combinés avec du ciment Portland, influencent les propriétés du béton
durci par action hydraulique et/ou pouzzolanique. L'utilisation des ajouts cimentaires pour
contrer la RAS est la méthode de prévention la plus couramment utilisée lorsque des
granulats potentiellement réactifs sont utilisés dans le béton [Duchesne et Bérubé, 2000].
On utilise aussi parfois des combinaisons de plusieurs ajouts cimentaires qui, avec le
ciment portland, sont aussi très efficaces (ciments ternaires ou quaternaire, par exemple).
Les ajouts cimentaires jouent plusieurs rôles importants face à la RAS. Ils permettent, entre
autres, de diminuer la perméabilité du béton et influencent la composition (quantité
d'alcalis libres, pH) de la solution interstitielle. Les prochains chapitres discuteront de
l'efficacité des ajouts cimentaires et des mécanismes causant cette efficacité.

2.3.1 Les cendres volantes

Les cendres volantes sont des sous-produits de la combustion du charbon pulvérisé dans les
centrales thermiques. Autrefois, ces cendres étaient rejetées dans l'atmosphère avec la
fumée (gaz), mais elles sont depuis plusieurs années récupérées dans un but
environnemental. Les cendres volantes se présentent sous la forme de fines particules qui se
sont formées en suspension dans l'air et se sont solidifiées sous forme de sphères de verre
contenant quelques phases cristallines. Le diamètre des particules varie entre 1 et 150 um,
mais la majorité des particules ont moins de 45 |j.m [Berry et Malhotra 1986]. Elles sont
majoritairement composées de silice, de calcium, d'aluminium et de fer; mais elles
contiennent aussi des quantités variables d'alcalis et certaines autres impuretés, tel que le
charbon.
30

Selon la norme CSA A3001 [CSA 2008], les cendres volantes sont dites de classe F, CH ou
CI, selon leur composition chimique. Elles peuvent être hydrauliques et contenir plus de
20% de CaO. Les cendres volantes de type F sont pouzzolaniques. Elles présentent des
teneurs en CaO inférieures à 15% et réagissent plus lentement que les cendres de type CH.
Bien entendu, certaines cendres volantes contiennent une teneur intermédiaire de CaO
située entre 15 et 20%; on les nomme alors de type CI.

Dans le béton frais, le remplacement d'une partie du ciment par des cendres volantes
diminue la demande en eau du béton dû à la sphéricité des particules (faible surface
spécifique). Donc, pour une même maniabilité, le béton demandera une quantité d'eau
moindre. Comme les cendres volantes sont, selon leur composition, hydrauliques et/ou
pouzzolaniques, la résistance du béton en sera affectée. Aux jeunes âges, la résistance
mécanique sera réduite (± fortement selon la nature de la cendre), tandis qu'à plus long
terme, la résistance rattrapera et dépassera même celle d'un béton au ciment portland. Selon
la composition des cendres volantes et de la teneur utilisée (en remplacement du ciment),
les effets suivants peuvent être observés:

• Réduction de la durabilité à l'écaillage et aux cycles de gel-dégel pour des teneurs


en cendres volantes plutôt élevées (>25%) [Marchand et al. 2005];

• Diminution de la perméabilité [Knutsson, 2010] (donc meilleure résistance à la


corrosion des armatures;

• Influence sur l'entraînement de l'air dans le béton selon la quantité de carbone dans
la cendre [Hill et Folliard 2006].

Mais la propriété bénéfique la plus importante dans le cadre de cette étude est son effet sur
la RAS. En effet, l'efficacité des cendres volantes pour contrer la RAS est largement
connue. De nombreuses études ont démontré que l'efficacité des cendres volantes diminue
généralement avec son contenu en CaO et en alcalis [Shehata et Thomas 2000b, Duchesne
et Bérubé 1994a, 1994b]. Donc, les cendres volantes les plus efficaces pour contrecarrer la
RAS sont celles contenant une faible quantité de CaO (classe F) et une faible quantité
d'alcalis. Les cendres volantes à plus haute teneur en CaO doivent être utilisées à plus
31

grand dosage pour être efficaces. Bien que la teneur en alcalis joue un important rôle dans
l'efficacité des cendres volantes [Shayan et al. 1996], la majorité de ces alcalis sont peu
solubles, car ils ne sont pas sous forme de sulfate. Par contre, il a été démontré que
certaines cendres volantes avec des contenus élevés en CaO possèdent des alcalis plus
solubles pouvant contribuer à la réaction alcalis-silice [Mehta 1983].

Selon la norme CSA A23.2-27A [CSA 2009], pour un ouvrage massif soumis à l'humidité
et d'une durée de vie de plus de 100 ans, l'utilisation d'une cendre volante de classe F (avec
moins de 3% de Na2Ûé) à un dosage minimum de 35% est efficace pour contrer la RAS
avec un granulat fortement réactif. La quantité minimum requise est un peu plus élevée
pour les cendres volantes de classe CI (40%).

On explique l'efficacité des cendres volantes généralement par trois phénomènes:

Diminution de la perméabilité. La présence de cendres volantes favoriserait la


formation plus abondante de C-S-H, diminuant ainsi la perméabilité de la pâte et
donc la mobilité des ions (à l'intérieur même de la masse et de l'extérieur vers
l'intérieur). La réduction de perméabilité diminuerait aussi l'apport d'eau provenant
de l'extérieur qui, comme on l'a vu, contribue à la réaction et à l'expansion
[Zachoria2008];

• Diminution du pH. La formation de C-S-H en présence de cendres volantes


intégrerait plus d'alcalis, ce qui diminuerait le pH du béton. L'analyse de la solution
interstitielle de béton a bien démontré que l'ajout de cendres volantes réduit
l'alcalinité [Diamond 1981, Xu and al. 1995, Arano et Kawamura 2000, Duchesne
et Bérubé 2000]. Pour des cendres ayant une teneur en alcalis similaire, la quantité
d'alcalis intégrés dans les produits d'hydratation augmente lorsque le ratio
CaO/Si02 diminue [Shehata et Thomas 2000a]. Pour d'autres chercheurs
[Kawamura et Matsushita 2007], la diminution du pH peut être expliquée par la
théorie de la double couche électrique, tel que vu précédemment. La silice présente
dans les cendres volantes offre une très grande surface spécifique qui adsorberait les
alcalis, ce qui expliquerait la diminution du pH;
32

• Formation de gel moins dommageable. Lors de l'ajout de cendres volantes, les


réactions formant les C-S-H consomment de la portlandite (60% de cendres
volantes pourrait consommer toute la portlandite produite par l'hydratation du
ciment portland [Zhang et al. 2000]). Le gel formé lors de la RAS contiendrait
moins de Ca et serait de faible viscosité, ce qui réduirait les dommages [Arano et
Kawamura 2000]. Selon Yamamoto et Tsutomu [2000], l'activité pouzzolanique
joue un rôle important. Ainsi, plus l'activité pouzzolanique est élevée, plus la
portlandite est consommée et plus grand est l'effet bénéfique contre la RAS.
Toutefois, d'autres recherches [Duchesne et Bérubé 1994a] ont démontré qu'il
n'existe aucune relation entre la quantité de portlandite consommée et la réduction
de l'expansion dans le béton.

2.3.2 Les laitiers granulés de hauts fourneaux (GGBFS)

Les laitiers granulés de hauts fourneaux (« ground granulated blast furnace slag » -
GGBFS) sont des sous-produits industriels de la fabrication de la fonte et proviennent
principalement des résidus de la gangue du minerai de fer, de la coke et du fondant utilisé
(calcaire ou dolomite) [Kosmatka et al. 2011]. Lors de la fonte, le laitier se sépare
naturellement grâce à sa plus faible densité. Pour être actif, le laitier doit être granulé, c'est-
à-dire projeté à l'état fondu, dès sa sortie du haut fourneau, dans un violent courant d'eau
froide. Le produit est alors vitrifié et il est par la suite broyé à une finesse égale ou
inférieure à celle du ciment. Lorsqu'il est granulé, ce matériau possède des propriétés
hydrauliques et une composition chimique assez bien définies, imposées par la nature du
minerai et par la qualité de la fonte. Les principaux composés sont, dans l'ordre, les oxydes
de Si, de Ca, de Mg, d'Al, et le soufre.

Les GGBFS sont efficaces pour contrer la RAS mais à des dosages généralement plus
élevés que les cendres volantes [Hooton 2000]. Toutefois, ils présentent l'avantage d'avoir
des performances plus constantes que les cendres volantes. Selon la norme
CSA A23.2-27A- 2009, une quantité de plus de 60% en remplacement du ciment est
considérée comme efficace pour contrer la RAS pour tous les types d'ouvrages et toutes les
conditions d'exposition. On explique la performance des GGBFS par les mêmes trois
33

phénomènes que les cendres volantes. Par contre, le principal effet serait qu'ils réduisent
l'alcalinité de la solution interstitielle et le pH [Canham et al. 1987, Duchesne et Bérubé
1994b] par l'incorporation d'une plus grande quantité d'alcalis dans les produits
d'hydratation. De plus, selon Wang [1995], les GGBFS auraient un effet de dilution, de
stabilisation et d'immobilisation des alcalis.

2.3.3 La fumée de silice

La fumée de silice, également connue sous le nom de microsilice, est un sous-produit de la


réduction du quartz dans des fours à arc électriques lors de la production du silicium métal
ou d'alliages au silicium (ex : ferrosilicium) [Kosmatka et al. 2011]. Formée de particules
amorphes, elle est composée majoritairement de dioxyde de silice (85 à 97% de SiÛ2). Elle
possède une granulométrie très fine (particules de moins de 1 nm) et elle est considérée
comme très réactive pouzzolaniquement. Dû à sa grande finesse (surface spécifique
élevée), l'utilisation de fumée de silice joue un rôle important sur la rhéologie du béton
(demande en eau plus élevée). Malgré cela, la fumée de silice améliore de manière
considérable la résistance en compression, l'imperméabilité et la durabilité du béton.

L'efficacité de la fumée de silice contre la RAS est expliquée par la baisse rapide de la
concentration des ions OH" dans la solution interstitielle [Shehata et Thomas 2000a]. Il faut
toutefois spécifier que cette baisse rapide du pH est suivie d'une augmentation de celui-ci
dans le temps, contrairement aux cendres volantes et aux laitiers. Cette baisse rapide du pH
est expliquée par l'adsorption d'une grande quantité d'alcalis à la surface des particules
individuelles de fumée de silice. En effet, les particules individuelles possèdent une
importante surface spécifique chargée négativement permettant d'absorber des ions positifs,
ce que sont les alcalis, tel que vu précédemment dans la théorie de la double couche
électrique. Par la suite, les particules très fines de silice sont dissoutes peu à peu et libèrent
les alcalis, ce qui fait augmenter le pH.

Tel que vu précédemment avec les cendres volantes, certains affirment que l'efficacité de
cet ajout peut aussi être expliquée par le fait que la fumée de silice réduit la perméabilité,
augmente les propriétés mécaniques du béton et consomme une quantité importante de
34

portlandite (source des ions OH" pour la solution interstitielle) lors de la réaction
pouzzolanique. L'utilisation de la fumée de silice pour contrer la RAS est un moyen
efficace (selon le dosage utilisé), mais moins fréquent que l'utilisation des cendres volantes
et des laitiers granulés de hauts-fourneaux, dû au coût plus élevé et au problème de
rhéologie , lorsque ajoutée en trop grande quantité. En effet, plusieurs études [Duchesne et
Bérubé 1994a, Ramezanianpour et al. 2004] ont démontré qu'il faut souvent jusqu'à 10%
de fumée de silice pour diminuer adéquatement l'expansion. Un tel dosage peut être
excessif dans certains cas, car l'ajout d'eau ou de superplastifiant pour avoir une
maniabilité acceptable peut causer un autre problème, soit un retrait excessif du béton.
Selon la norme CSA A23.2-27A 2009, on peut calculer le dosage requis en fumées de silice
selon le type d'ouvrage et ses conditions d'exposition en fonction de la teneur en alcalis du
béton, mais on ne devrait jamais en utiliser moins de 7% lorsqu'elles correspondent à la
seule mesure préventive utilisée pour contrer les effets néfastes de la RAS.

Des études ont démontré qu'il est souvent préférable d'utiliser la fumée de silice en
combinaison avec d'autres ajouts cimentaires [Thomas et al. 1999]. L'utilisation de telles
combinaisons, comme les ciments ternaires, est de plus en plus fréquente. De plus, lorsque
la fumée de silice se retrouve pré-mélangée dans des produits cimentaires (ciments binaires,
ternaires ou quaternaires), cela permet d'avoir un meilleur contrôle sur l'agglomération des
particules. En effet, la fumée de silice utilisée dans le béton est souvent vendue sous forme
densifiée, car le coût de transport est beaucoup plus faible et la manutention en est
beaucoup simplifiée. Cette fumée densifiée consiste en des agglomérations de particules de
diamètres compris entre 10 um et quelques mm. Ces agglomérations ne sont souvent pas
toutes brisées lors du malaxage normal d'un béton et peuvent devenir des particules de
silice réactive, tels des granulats réactifs dans le béton [Rangaraju et Olek 2000, Diamond
et al. 2004]. Toutefois, selon certains autres chercheurs, ces agglomérations ne causeraient
pas d'expansion dans le béton et ne réduiraient pas les effets bénéfiques de la fumée de
silice [Gudmundsson et al. 2000].

1
Afin de résoudre ce problème, les bétons incorporant de la fumée de silice sont également superplastifiés
afin d'assurer une bonne maniabilité et ainsi une consolidation et une mise en place adéquates du béton
35

2.3.4 Autres ajouts cimentaires

Finalement, il existe plusieurs autres types d'ajouts minéraux que l'on peut utiliser en
remplacement du ciment pour contrer la RAS, comme le métakaolin (argile kaolinite
calcinée), la zéolite (silicate d'aluminium hydraté) naturelle ou modifiée ou encore les
pouzzolanes naturelles. Ainsi, à titre indicatif, Ramlochan et al. [2000] ont démontré que
10-15% de métakaolin hautement réactif, en remplacement du ciment, diminue
considérablement la concentration en ions OH" dans la solution interstitielle et est efficace
pour contrer la RAS. De même, il faudrait entre 15 et 20% de zéolite naturelle pour contrer
la RAS [Feng et al. 1998].

D'autres produits encore plus exotiques, tels que des cendres de balles de riz ayant un
contenu élevé en silice (85% de Si02), ont aussi démontré leur efficacité [Mehta et Folliard
1995, Hasparyk et al. 2001]. De même, de la fibre de verre broyée finement peut aussi être
efficace [Xu et al. 1995].

Finalement, la majorité de ces produits sont des poudres sèches finement broyées qui
agissent sur le béton de manière similaire, soit par une baisse marquée du pH de la solution
interstitielle et par une diminution de la perméabilité de la pâte de ciment ou par réaction
pouzzolanique consommant la portlandite. Par contre, il existe d'autres méthodes, telles que
l'ajout d'adjuvants liquides au béton frais, qui agissent complètement différemment pour
contrer la RAS. Ces adjuvants, à base de lithium, constituent le sujet principal de cette
recherche et font l'objet de la prochaine section.

2.4 Les adjuvants à base de lithium

2.4.1 Historique

Le lithium a été découvert en 1817, en Suède, par Johann August Arfvedson dans un
minéral identifié « pétalite » (LiAl(Si205)2). Il fut isolé 38 ans plus tard, soit en 1855, par
William Thomas Brande et Sir Humphrey Davis, par electrolyse d'un oxyde de lithium
(Li20). Mais le produit ne fut commercialisé qu'en 1923. Son nom vient du grec « lithos »
36

qui signifie « roche », car il fut découvert dans un minéral, tandis que les deux autres alcalis
(sodium et potassium) furent découverts dans des plantes.

Le lithium est un alcalin (groupe 1A avec le sodium et le potassium) et son numéro


atomique est 3. On le retrouve dans la nature sous la forme d'un mélange de deux isotopes,
soit 6Li (7,5%) et 7Li (92,5%), ce qui fait que sa masse atomique est de 6,941 g-mol"1
[David 2009]. Le Li est le solide le plus léger (0,53 g-cm"3 à 20 °C) et est très réactif, car il
a tendance à s'oxyder, du fait qu'il possède une valence de +1. De ce fait, il est très peu
répandu dans la nature; on le retrouve sous des teneurs d'environ 13 ppm dans la croûte
terrestre, de 0,2 ppm dans les océans et il est indétectable dans l'air.

L'ion lithium est exceptionnellement petit et a un ratio charge/rayon très élevé. Ainsi, ses
propriétés sont considérablement différentes des autres ions alcalins. Les principales
propriétés d'un ion sont fonction de sa charge, de son rayon ionique et de sa structure
électronique interne. Les rayons ioniques des principaux alcalis sont présentés au
tableau 2.2.

Tableau 2.2 - Rayons ioniques des principaux alcalis

Rayon ionique Rayon hydraté


Alcali
Angstrom
Lithium 0,60 3,40
Sodium 0,95 2,76
Potassium 1,33 2,32

2.4.2 Sources de lithium

Bien que le lithium se retrouve distribué un peu partout sur la planète terre, on le retrouve
plus concentré dans certains silicates de lithium tels que le spodumène, la pétalite, la
lépidolite et l'amblygonite, ainsi que dans certains types de minéraux argileux. De plus, on
retrouve le lithium dans des saumures ayant une grande concentration en carbonate de
lithium ou en chlorure de lithium. La production de lithium à partir de silicate de lithium est
beaucoup plus énergivore et coûteuse que celle à partir de saumures.
37

Selon les statistiques de 2006 [Ober 2007], le Chili est le plus grand producteur mondial de
carbonate de lithium depuis 1997, où il a dépassé les États-Unis. Ces derniers sont les plus
grands consommateurs de lithium et les plus grands producteurs de composés à base de
lithium de la planète.

La majorité de la production de lithium provient des saumures. On retrouve des saumures


en exploitation aux États-unis (dans le Nevada), au Chili, en Argentine, en Chine et en
Bolivie. On y exploite en majorité un carbonate de lithium ainsi qu'un chlorure de sodium
(pour la source en Argentine). Au Chili, on produit du carbonate de lithium ainsi que des
produits chimiques intermédiaires et spéciaux à base de lithium, par evaporation solaire des
saumures [Ressources Naturelles Canada 2005]. On remarque que le prix du lithium a chuté
de 4 470 $ U.S./tonne en 2000 à 1 590 $ U.S./tonne en 2002 [Ober 2007], grâce à la mise
en marché d'importantes exploitations de saumures, ce qui a, part le fait même, abaissé les
coûts de production du lithium.

D'autre pays, tels l'Australie, le Brésil, le Canada, le Portugal et le Zimbabwe, produisent


aussi des quantités significatives de lithium à partir de minerais [Jaskula 2010].

2.4.3 Le lithium dans le béton

L'utilisation de produits à base de lithium pour contrôler la RAS a été rapportée pour la
première fois par McCoy et Caldwell [1951]. Ces deux chercheurs avaient réalisé une
importante étude sur les adjuvants chimiques pouvant prévenir ou minimiser l'expansion
due à la RAS dans le béton. Dans cette étude, plus de cent substances différentes incluant
des sels, des acides, des huiles et certaines protéines et liquides organiques, ont été testées
afin de vérifier leur efficacité pour réduire l'expansion causée par la RAS. McCoy et
Caldwell ont utilisé l'essai ASTM C 227 comme essai pour mesurer l'influence de ces
produits sur l'expansion due à la RAS, avec un verre de Pyrex comme granulat réactif et
deux ciments contenant 0,57% et 1,15% de Na20é. Ils ont trouvé que certains sels
d'ammoniaque et de cuivre permettaient de réduire quelque peu l'expansion, mais ce sont
les composés à base de lithium qui se sont avérés les plus efficaces. Ils ont remarqué que
les composés à base de lithium (LiCl, Li2C03, LiF, Li2SÛ4, LiNÛ3 et L_2Si03), lorsque
38

utilisés en quantité suffisante, éliminent l'expansion après huit semaines d'entreposage


à 38°C.

Depuis l'étude de McCoy et Caldwell, quelques études additionnelles ont été réalisées sur
l'utilisation de produits à base de lithium pour contrer la RAS, mais c'est seulement à partir
des années 1990 qu'il y a eu une recrudescence de l'intérêt pour ces produits. On peut
aujourd'hui affirmer que les adjuvants chimiques à base de lithium utilisés dans le béton
frais sont un moyen efficace, lorsqu'uti lises à un dosage approprié, pour empêcher ou
minimiser l'expansion due à la RAS [Thomas et al. 2007b].

11 a été démontré que l'utilisation de ces adjuvants à base de lithium, à des dosages
normalement nécessaires pour contrer la RAS, influence peu les propriétés mécaniques du
béton durci et ses propriétés à l'état frais [Thomas et al. 2003]. Par ailleurs, une étude
récente [Millard et Kurtis 2008] a démontré que ces adjuvants peuvent accélérer la prise
initiale du béton et ce, jusqu'à 15 à 22% dans certains cas, et par la suite retarder la prise
après les premières 24 heures.

Plusieurs de ces études ont porté sur la quantité de lithium minimale nécessaire pour contrer
la RAS. Cette quantité est évaluée sous forme d'un rapport molaire de la quantité de lithium
divisée par la quantité des autres alcalis, identifié par le rapport [Li/Na+K]. Ce rapport
permet de calculer les dosages des adjuvants à base de lithium que l'on doit utiliser dans le
béton pour contrer la RAS. Le tableau 2.3 présente un résumé des résultats des différentes
recherches sur le sujet.
39

Tableau 2.3 - Résultats des différentes études sur l'utilisation du lithium contre la RAS
(modifié à partir de Folliard et al. 2003)
Ration molaire minimum
Produit de
Études Essai Granulats réactifs pour contrer la RAS
lithium
[Li/Na+K]
LiCl, Li 2 C0 3 ,
M C C O Y et
ASTM C 227 Verre Pyrex LiF, Li 2 S0 4 , 0,74
CALDWELL 1951
LiN0 3 , Li 2 Si0 3
SAKAGUCHI et al. Verre Pyrex, sable LiOHH 2 0,
ASTM C 227 0,9
1989 andésite à pyroxene Li 2 C0 3 , LiN0 2
0,5% de la masse du ciment avec
L i O H H 2 0 , LiF,
OHAMAetal. 1989 Autoclave Opale amorphe LiF, 0,7% de la masse du ciment
Li 2 C0 3
avec LiOH.H.O
0,67 avec LiF
STARK 1992 ASTM C P 214' Andésite LiF, Li 2 C0 3
0,92 avec Li 2 C0 3
Rhyolite,
STARK et al. 1993 ASTMC 1260 LiOHH20, 0,75-1,0
gneiss granitique
DIAMOND et O N G Cristobalite, 1,2 avec la cristobalite
ASTM C 227 LiOH
1992 opale Beltane > 1,2 avec l'opale
QlNGHAN et al. 0,8 avec des mortiers à haute
Autoclave Andésite LiN0 3
1995 teneur en alcalis
LiOH H 2 0, LiF,
LUMLEY 1997 ASTMC 1293 Cristobalite 0,62
Li2COB
THOMAS et al. Variétés de granulats 0,52 à 0,74 avec LiN0 3
ASTMC 1293 LiOH, LiN0 3
2000 d'Angleterre 0,55 à 0.85 avec LiOH
Granulats canadiens LiN0 3 , 0,72 (LiN0 3 , avec Sudbury)
DURAND 2000 ASTMC 1293 (Sudbury, Potsdam, L i O H H 2 0 , LiF, 0,82 (LiOH H 2 0, LiF, Li 2 C0 3 ,
Sherbrooke) Li 2 C0 3 avec Sudbury)
NM (gravier), 0,75 (LiN0 3 avec NM)
FOURNIER et LiN0 3 ,
ASTMC 1293 Con (grès lithique), 0,93 (LiN0 3 avec Sp)
BÉRUBÉ2001 LiOHH20,
Sp (calcaire) 1.1 (LiOHH 2 0 avec Con)
KAWAMURA et ASTM C 227
Silex (flint) calciné LiOH, Li 2 C0 3 0,91
FUWA 2003 modifié
ASTMC 1260 Verre Pyrex, LiOHH20,
LANE 2002 0,93 avec LiN0 3
ASTMC 1293 quartz LiN0 3
M o et al. Granite 0,3 pour Na 2 O e < 2.5% (ciment)
Autoclave LiOHH20
2003 (qtz microcristallin) 0,6pourNa2Oe<3%
COLLINS et al. LiCl, LiOH, 0,6 avec LiOH, 0.9 avec LiCl, 0,8
ASTM C 227 Verre de borosilicate
2004 LiN0 3 avec LiN0 3
Jobe-Newman (sable
volcanique), 0,55 avec Jobe-Newman
BARBORAK
ASTMC 1293 Wright (sable lithique), LiN0 3 < 0,37 avec Wright
2005
Placitas (gravier > 0,74 avec Placitas
volcanique)
NS (grès lithique), > 1,48 avec NS
FENG et al. 2008 ASTMC 1293 LiN0 3
NB (rhyolite) 0,74 avec avec NB

1
Devenu ASTM Cl260.
40

Les différentes études démontrent qu'un rapport molaire minimum de [Li/Na+K] variant de
0,3 à 1,2 permet de contrer l'expansion due à la RAS. Même si on utilise différents produits
à base de lithium (LiN03, LiOHH 2 0, LiF, Li2C03, LiCl, Li2S04, LiN0 3 , LiN02, Li2Si03),
il semble que le rapport [Li/Na+K] minimum pour contrer l'expansion soit semblable pour
la majorité des produits, pour un granulat réactif donné, quoique légèrement supérieur pour
le LiOH comparativement au LiN03. Ces études prouvent donc l'efficacité des produits à
base de lithium pour contrer la RAS et on peut affirmer qu'un ratio [Li/Na+K] variant de
0,6 à 0,9 est généralement efficace selon les granulats. Cependant, il est nécessaire
d'ajouter que certains granulats semblent nécessiter des dosages plus élevés. De plus, la
teneur en alcalis (Na+K) du mélange de béton semble aussi avoir une influence. Les
mécanismes expliquant cette efficacité ne sont pas encore bien compris et plusieurs théories
ont été avancées, mais sans qu'aucune ne fasse l'unanimité. On peut retrouver au tableau
2.4 les diverses théories avancées par les chercheurs pour expliquer l'efficacité des
adjuvants à base de lithium.
41

Tableau 2.4 Théories proposées par différents auteurs pour expliquer l'effet bénéfique
du lithium contre la RAS
Études Théorie(s) avancée(s) afin d'expliquer l'efficacité du lithium
• Formation d'un silicate de lithium peu soluble et stable protégeant la silice
LAWRENCE et VIVIAN
réactive de l'attaque des autres alcalis.
1961
Diminution du taux de dissolution de la silice.
• Formation d'un produit à base de lithium non gonflant au profit d'un produit
SAKAGUCHI et al. 1989
gonflant (contenant plus de sodium et de potassium).
WiJNENetal. 1989 • Diminution du taux de dissolution de la silice.
DIAMOND et ONG 1992 • Formation d'un gel moins ou non gonflant.
STARK 1992
• Formation d'un gel moins ou non gonflant.
STARK et al. 1993
QINGHAN etal. 1995 • Stabilisation du gonflement du gel (gel moins gonflant).
PREZZietal. 1998 • Théorie de la double couche électrique.
• Limitation de la repolymérisation de la silice.
KURTIS et MONTEIRO
Agrégation favorisée, ce qui décroît la densité de charge de surface du gel (gel
2003
moins gonflant).
• Formation d'un silicate de lithium peu soluble et stable protégeant la silice
MITCHELL et al. 2004
réactive de l'attaque des autres alcalis.
• Diminution de la dissolution de la silice ou précipitation favorisée d'un produit
COLLINS et al. 2004 riche en silice (non gonflant).
Limitation de la repolymérisation de la silice.
Mo 2005 • Formation d'un produit non gonflant.
BROUXELet al. 2004 • Formation d'un gel plus cristallin et moins gonflant.
• Formation d'un produit non gonflant autour des granulats et agissant comme une
Lu et al, 2006
barrière protectrice.
Qi et WEN 2007 • Formation d'une couche protectrice autour de la silice.
FENG et al. 2008 • Formation d'une couche protectrice autour de la silice.
• Formation d'un produit non gonflant autour des granulats et agissant comme une
BULTEEL et al. 2008
barrière protectrice.

Nous pouvons donc regrouper les différentes conclusions des études en cinq mécanismes
pouvant expliquer l'efficacité des adjuvants à base de lithium:

1. Diminution de la dissolution de la silice;

2. Formation d'un silicate de lithium stable protégeant la silice;

3. Diminution de la repolymérisation de la silice;

4. Production d'un gel moins ou non gonflant;

Théorie de la double couche.

Ces différentes théories feront maintenant l'objet d'une revue plus approfondie.
42

2.4.4 Diminution de la dissolution de la silice

Lawrence et Vivian [1961] ont trouvé que le taux de dissolution de la silice varie en
fonction du type d'hydroxyde, suivant l'équation suivante:

LiOH < NaOH < KOH

Wijnen et al. [1989] ont trouvé la même relation et ont proposé que le taux de dissolution
décroît lorsque le rayon hydraté de l'ion alcalin est plus grand. Ces résultats suggèrent que
le lithium peut décroître le taux de dissolution de la silice, ce qui limiterait la formation
d'un gel expansif.

D'autre part, Lawrence et Vivian [1961] ont démontré qu'avec le temps, la concentration
de la silice dissoute pour chaque hydroxyde alcalin tend vers une même valeur et que le
lithium réduirait le taux de dissolution mais non la solubilité de la silice. Le taux de
dissolution de la silice varie donc avec le type d'hydroxyde employé, mais qu'en est-il dans
le béton? Lorsque du lithium est utilisé dans le béton, le pH (concentration en ion OH") est
sensiblement le même (sauf lors d'un ajout de LiOH). Est-ce que l'ajout d'une quantité de
lithium à pH constant (concentration fixe en ion OH") peut effectivement réduire le taux de
dissolution de la silice?

Natesaiyer et Hover [1992] ont testé l'effet de divers agents chimiques sur la solubilité de
la silice dans une solution de IN NaOH. Ils ont utilisé une version modifiée de l'essai
ASTM C 289, avec deux types de granulats, soit une opale et un calcaire (Spratt). Ils sont
arrivés à la conclusion que l'ajout de chlorure de lithium, à un ratio allant jusqu'à
0,36 [Li/Na], ne réduisait pas le taux de dissolution de la silice (opale). Pour le calcaire
Spratt, un dosage allant jusqu'à 0,025 [Li/Na] n'a pas eu d'influence sur le taux de
dissolution. Toutefois, ce dosage est probablement trop faible pour avoir une influence.
D'autre part, Collins et al. [2004] ont observé une diminution de la dissolution de silice
avec l'ajout de LiN0 3 et de LiCl, mais, par contre, une augmentation de cette même
dissolution avec l'ajout de LiOH dans une solution de 0,7M NaOH. Il a été suggéré que
43

l'augmentation de la dissolution de la silice par l'ajout de LiOH serait due à l'augmentation


dupH.

Il n'a pas été clairement démontré que le lithium joue un rôle important dans le taux de
dissolution de la silice, car s'il y a formation d'un silicate de lithium ou un autre
phénomène protégeant la silice, il est normal que la mesure du taux de dissolution soit
diminuée. Par contre, la plupart des auteurs s'entendent pour dire que le type d'hydroxyde
n'influence que très peu la solubilité de la silice (quantité totale pouvant être dissoute). On
peut voir à la figure 2.5a de la section 2.2.1 que la solubilité de la silice est la même quel
que soit le type d'hydroxyde, lorsque la silice est entièrement dissoute (zone 1). Par contre,
lorsque la silice n'est pas entièrement dissoute (zone 2), il semble que le type d'hydroxyde
joue un rôle sur la quantité de silice dissoute selon l'équation 5 suivante, laquelle est en
accord avec les résultats de Lawrence et Vivian [1961]:

NaOH > KOH > LiOH > RbOH > CsOH [5]

Toutefois, il est important de noter que la plupart des études sur le taux de dissolution et la
solubilité de la silice ont été réalisées sur des types de silice amorphe ou artificielle qui
représentent plus ou moins le type de silice réactive retrouvée dans les granulats naturels.

2.4.5 Formation d'un silicate de lithium stable protégeant la silice

Une autre théorie est que le lithium, en présence de silice réactive, formerait un produit
secondaire peu ou non expansif (silicate de lithium) autour des particules de silice réactive,
ce qui réduirait l'expansion causée par la RAS. Stark [1992] a proposé que ce produit doit
contenir une quantité minimum de lithium afin qu'il soit non gonflant. Collins et al. [2004]
ont trouvé que le lithium favorise la précipitation d'un produit riche en silice qui serait non
expansif. D'autre part, Lawrence et Vivian [1961] ont proposé qu'un silicate de lithium se
formerait lors de la RAS et que ce produit serait peu soluble et plus stable. Grâce à cette
stabilité, ce silicate protégerait la silice réactive des attaques des autres alcalis et des ions
OH".
44

Qi et Wen [2007] ont remarqué, à l'aide d'un microscope électronique à balayage, qu'une
couche protectrice se forme autour des grains d'opale lorsque ceux-ci sont immergés dans
une solution d'hydroxide de lithium.

2.4.6 Diminution de la repolymérisation de la silice

Une autre théorie proposée serait que le lithium limiterait ou diminuerait la


repolymérisation de la silice, ce qui diminuerait la quantité de gel formé. Utilisant un
microscope à transmission à rayons X afin d'étudier la réaction alcalis-silice dans une
solution artificielle contenant ou non du lithium, Kurtis et Monteiro [2003] ont observé
qu'il se formait plus de gel lorsqu'il y avait absence de lithium. Ainsi, ils ont proposé que
lorsque le lithium est présent en quantité suffisante, la répulsion entre les particules de
silice demeure élevée, ce qui implique que la repolymérisation de la silice sous forme d'un
gel expansif ne se produit pas. Par contre, cette théorie ne parvient pas à expliquer pourquoi
le lithium est consommé en présence de granulats réactifs [Sakaguchi et al. 1989, Collins et
al. 2004].

2.4.7 Production d'un gel moins ou non gonflant

Lawrence et Vivian [1961] ont montré que le gel de silice tend à être moins expansif avec
l'hydroxyde de sodium lorsque l'on augmente la concentration en hydroxyde de lithium
à 2 normale dans une solution à 2 normale d'hydroxyde de sodium. En mesurant la
concentration des alcalis dans la solution interstitielle de barres de mortier, extraite sous
pression, Sakaguchi et al. [1989] ont démontré que le lithium semble se consommer de
préférence aux autres alcalis. En effet, ils ont mesuré que la concentration en Li, dans des
barres de mortiers contenant au départ du lithium, décroît avec le temps tandis que les
concentrations en sodium et en potassium restent constantes. Par contre, dans les barres de
mortier sans lithium, les concentrations en sodium et en potassium décroissent avec le
temps. Les auteurs suggèrent que la réaction alcalis-lithium est favorisée par rapport à la
réaction alcalis-sodium ou alcalis-potassium.
45

Diamond et Ong [1992] ont démontré que plus un gel contient une grande proportion de
lithium par rapport au sodium et potassium, plus l'expansion de ce gel diminue. En accord
avec ces recherches, la formation d'un silicate de lithium (amorphe ou cristallin) non
expansif aux dépens d'un gel silico-calco-alcalin expansif pourrait expliquer l'efficacité du
lithium.

Certains chercheurs sont toutefois en désaccord avec cette théorie. Selon Chatterji [1989],
un système contenant Na+, K+, et Li+ favorise la réaction avec le sodium et le potassium,
mais pas avec le lithium. Les alcalis, chargés positivement, sont en compétition pour être
adsorbés à la surface de la silice, chargée négativement. Selon ces auteurs, l'affinité d'un
ion à être absorbé croît avec son rayon cationique, ce qui favoriserait les réactions avec le
sodium et le potassium.

2.4.8 Théorie de la double couche

Ya théorie de la double couche, communément appelée « electrical double layer (EDL) »,


soutient le phénomène d'adsorption à la surface des particules, tel que vu précédemment,
pour expliquer le gonflement lors de la RAS.

Tout d'abord, Prezzi et al. [1997] ont proposé d'utiliser cette théorie afin d'expliquer
l'expansion causée par le gel de la réaction alcalis-silice. Par la suite, Prezzi et al. [1998]
ont proposé cette même théorie pour expliquer l'effet des divers adjuvants chimiques
utilisés dans le béton (ou le mortier) pour contrer l'expansion associée à la réaction. Selon
cette théorie, deux facteurs jouent un rôle important dans le gonflement d'un gel de silice:
la valence et le rayon hydraté des ions présents dans la double couche. Cette théorie prédit,
d'une part, que plus la valence d'un ion est élevée, moins le gel gonflera et, d'autre part,
que plus le rayon hydraté d'un ion est élevé, plus le gel gonflera. Ainsi, cette théorie prédit
que l'expansion du gel augmentera selon la séquence suivante:

Al3+ < Ca2+ < Mg2+ < K+ < Na+ < Li+
46

Prezzi et al. [1998] ont utilisé ces divers ions sous forme de sels (chlorures et hydroxydes)
dans des barres de mortier (1 et 2 molarités) en présence d'une quartzite microcristalline
comme granulat réactif, ceci afin de mesurer leur effet sur l'expansion lors de l'essai
ASTM C 1260. Ils ont obtenu la relation prédite par la théorie sauf pour le cas du lithium
qui s'est retrouvé entre le magnésium et le potassium et qui, normalement (i.e. selon la
théorie), devrait produire plus d'expansion. Lors de cette même étude, on a utilisé différents
sels à 1 et 2 molarités dans les mortiers, ce qui est très élevé (environ 3 à 6%). L'utilisation
de ces adjuvants à de telles concentrations peut changer considérablement les propriétés des
mortiers. Des propriétés telles que la résistance en tension et la porosité sont des éléments
importants lors de l'essai accéléré sur barres de mortier. Il a été démontré par Mo [2005]
que l'ajout de LiOH à 1% diminue la résistance en compression des mortiers de 24% à
3 jours et de 38% à 28 jours.

La théorie de la double couche permet peut-être d'expliquer plusieurs observations lors du


gonflement du gel formé lors de la réaction alcalis-silice, mais elle est difficilement
utilisable pour expliquer l'efficacité du lithium.

2.5 Point de vue économique des méthodes de prévention de la RAS

Comme on a vu précédemment, il existe plusieurs méthodes efficaces pour contrer la RAS.


Un des principaux facteurs influençant la méthode privilégiée à utiliser dans le béton est
son coût additionnel sur la production du béton. Le prix d'une tonne de ciment portland de
type Gu varie d'une région à l'autre et se situe en moyenne entre 105 et 175 $/tonne
[Bouzoubaa et Fournier 2004]. Plus précisément, selon Ressources Naturelles Canada
[2005], le prix moyen du ciment au Canada a été de 117 $/tonne en 2006. Le prix d'un litre
de nitrate de lithium (30% en solution) est approximativement de 4,35 $ [Folliard et al.
2003] et le dosage recommandé par le fabricant est de 4,6 litres par kg de Na20e. Estimant
le coût des granulats et du sable à environ 8 $/tonne [Ressources Naturelles Canada 2005,
section « granulat »], on peut voir au tableau 2.5 le coût représenté par l'utilisation du
lithium.
47

Tableau 2.5 - Coûts d'utilisation du nitrate de lithium dans le béton


Béton avec un dosage
Béton sans lithium
Matériaux de[Lil/|Na+Kl=0.74
Quantité Prix Quantité Prix
Ciment (120$/1000 kg) 400 kg 48,00 $ 400 kg 48,00 $
Granulat (8$/1000 kg) 1850 kg 14,80$ 1850 kg 18,50$
... ... 1
Nitrate de lithium (4,35 $/litre) 13 litres 56,50 $
Coût par mètre cube 62,80 $ 123,00$
Utilisant un ciment contenant 0,7% de Na2Oe

Les données du tableau 2.5 révèlent que l'utilisation du lithium représente un coût
important des matériaux, soit une augmentation de près de 100% pour la production de
I m3 de béton. Il faut noter qu'au tableau 2.5 les coûts de transport des matériaux ne sont
pas compris, que ce soit pour le béton (entre l'usine de béton et le chantier), le ciment
(entre la cimenterie et l'usine de béton), l'adjuvant (entre le producteur et l'usine de béton)
ou le granulat (entre la carrière/gravière et l'usine).

II a été démontré auparavant que les ajouts cimentaires sont un des moyens efficaces pour
contrer la RAS. Le coût de ces matériaux (sans transport) par rapport au ciment est présenté
au tableau 2.6.

Tableau 2.6 - Coûts moyens des ajouts cimentaires par rapport au ciment dans chaque
province canadienne [tiré de Bouzoubaa et Fournier 2004]
Type d'ajout Coût moyen par rapport au ciment dans chaque province
cimcntairc
Atlantique Québec Ontario Saskatchewan Manitoba Alberta Colombie-Britannique
Cendre volante 40-45% 75-80% 50-60% 38-43% 43-45% 50-60% 56-70%
Laitier - 85% 80% - - 80-85% -
Fumée de silice 360-400% 480% 300-430% 350-430% 350-430% 390-450% 480%
Métakaolin - - - - - - 480%

L'analyse du tableau 2.6 révèle que l'utilisation de cendres volantes ou de laitiers de hauts
fourneaux en remplacement du ciment, lorsque ces produits sont de bonne qualité et situés
assez près, constitue un moyen abordable pour contrer la RAS.
48

Voyons maintenant divers scénarios pour un producteur de béton de ciment ayant à sa


portée un sable non réactif et une pierre fortement réactive. En supposant que les sources de
granulats (sable et gros granulats) sont situées à 10 km de l'usine et que la source de ciment
(teneur en alcalis de 0,70% de Na20e) la plus proche est à 100 km, on peut examiner les
méthodes préventives disponibles pour un coût équivalent à l'utilisation d'adjuvants à base
de lithium (le coût de transport du lithium étant négligeable celui-ci n'a pas été considéré).
Le coût de transport des divers constituants est calculé selon le tableau 2.7 provenant des
statistiques québécoises de 2008 [Ministère des Transports du Québec 2008].

Tableau 2.7 - Coûts moyens du transport des matériaux1


Tarifs tonne-kilomètre
Pour chaque kilomètre ou fraction de kilomètre additionnel
Du
Delà De 10 à De 30 à De 65 à Plus de
chargement
9,9 km 29,9 km 64,9 km 159,9 km 160 km
à 0,9 km
Gros granulats et sable 1,383$ 0,267 $ 0,252 $ 0,155$ 0,102$ 0,085 $

Autres matériaux en vrac 1,107$ 0,222 $ 0,201 $ 0,123$ 0,083 $ 0,067 $


Ministère des Transports du Québec 2008.

Tableau 2.8 - Coûts en matériaux de diverses méthodes de prévention de la RAS pour 1 m3


de béton de ciment
Utilisation d'un Utilisation d'un UtUisation
Utilisation de LiNQ 3 ciment à basse granulat non- de cendres
Matériaux teneur en alcalis réactif volantes 2,3
Quantité/ Quantité/ Quantité/ Quantité/
Prix Prix Prix Prix
distance distance distance distance
400 kg 48,00 $ 400 kg 48,00 $ 400 kg 48,00 $ 280 kg 33,60 $
Ciment (120$/1000 kg)
100 km 5,73$ 2176 km 61,76$ 100 km 5,73$ 100 km 4,01 $
Cendres volantes ... — — — ... ... 120 kg 10,80$
23
(90$/1000kg) — — — — — . . . 7502 km 61,35$
850 kg 6,80$ 850 kg 6,80$ 850 kg 6,80$ 850 kg 6,80$
Sable (8 $/l 000 kg)
10 km 3,31 $ 10 km 3,31$ 10 km 3,31 $ 10 km 3,31 $
Gros granulats 1000 kg 8,00$ 1000 kg 8,00$ 1000 kg 8,00$ 1000 kg 8,00$
(8$/1000kg) 10 km 3,89$ 10km 3,89$ 582 km 59,92 $ 10km 3,89$
Lithium (4,35$/litre) 12,88 litres 56,03 $' ... — — ... — ...
Total 131,76 S/m3 131,76 S/m3 131,76 $/m3 131,76 S/m3
Dosage recommandé par 1e fabricant de 4,6 litres/k g de Na2Oe;
Remplacement de 30% du ciment par les cendres volantes [CSA 2009];
3
Coût des cendres volantes équivalent à 75% de celui du ciment.
49

Le tableau 2.8 indique donc que l'utilisation d'un ciment à faible teneur en alcalis situé à
moins d'environ 2200 km est plus économique que l'utilisation d'un adjuvant à base de
lithium. L'utilisation d'un gros granulat non réactif demeure économique par rapport au
lithium si ce granulat se retrouve à moins d'environ 600 km. De plus, il est certain que si le
sable utilisé est également réactif, la distance sera diminuée d'environ la moitié.
L'utilisation de cendres volantes demeure avantageuse, par rapport à l'utilisation du
lithium, si la source se situe à moins d'environ 7500 km. Finalement, comme le révèle les
données du tableau 2.8, l'utilisation du lithium augmente considérablement les coûts des
matériaux entrant dans la fabrication du béton et d'autres méthodes de prévention
demeurent plus économiques. Toutefois, l'utilisation des adjuvants à base de lithium peut
être économiquement rentable dans les endroits plus isolés où il n'y a pas de sources de
granulats non réactifs et où les sources de cendres volantes, de laitier ou de ciment à faible
teneur en alcalis sont très éloignées. Toutefois, dans le calcul des coûts précédents, le
camionnage est utilisé comme moyen de transport; d'autres moyens de transport tels que
ferroviaire ou maritime pourraient être beaucoup plus économiques sur de longues
distances. Par conséquent, l'utilisation des adjuvants à base de lithium dans le béton comme
moyen de prévention de la RAS demeure donc un moyen de prévention, mais
économiquement rentable que dans de rares cas.
50

2.6 Références

ARANO, N. et KAWAMURA, M. (2000), Comparative consideration on the mechanisms


of ASR suppression due to différent mineral admixture. Il 1 International Conference on
Alkali-Aggregate Reaction (ICAAR), Québec (Canada), juin 2000, pp. 553-562.

BARBORAK, R.C. (2005), Using lithium compounds as an admixture in new concrete and
as a post-treatment for existing concrete affected by alkali-silica reaction, Thesis, Master
of science in engineering, The University of Texas at Austin, Austin (USA), 115 p.

BERRY, E.E. et MALHOTRA, V.M. (1986), Fly ash in concrete - Les cendres volantes
dans le béton, CANMET, Energy, Mines and Resources, Canada Centre for Mineral and
Energy Technology, SP85-3, 178 p.

BÉRUBÉ, M.A. et DUCHESNE, J. (1992), Evaluation of testing methods for assessing the
effectiveness of mineral admixtures in suppressing expansion due to alkali-aggregate
reaction. 4 th CANMET-ACI International Conference on Fly Ash, Silica Fume, Slag and
Natural Pozzolans in Concrete, Istanbul (Turquie), mai 1992, ACI SP-132, pp. 549-575.

BÉRUBÉ, M.A., FOURNIER, B., GRATTAN-BELLEW. P.E., VÉZINA, D., FRENETTE,


J., DUCHESNE, J., CHOQUETTE, M., OUELLET, C. et CLAVEAU, A. (1988), Les
réactions alcalis-granulats dans le béton : caractérisation des granulats, méthodes
préventives et inspection de bétons. Notes de cours, CRIB, Département de géologie et de
génie géologique, Université Laval, Québec, 338 p.

BLACKWELL, B.Q., THOMAS, M.D.A. et SUTHERLAND (1997), Use of lithium to


control expansion due to alkali-silica reaction in concrete containing U.K. aggregates.
4 th Int. Conf. CANMET/ACI on Concrete Durability, Sydney (Australie), ACI SP 170,
pp. 649-663.

BOUZOUBAA, N. et FOURNIER, B. (2004), Current situation of supplementary


cementitious materials (SCMs) in Canada. (SCMs) in Concrete Construction in Canada.
Canadian Journal of Civil Engineering, 32( 1 ): 129-143.
51

BROUWERS, H.J.H. et VAN EIJK, R.J. (2003), Alkali concentrations of pore solution in
hydrating OPC. Cement and Concrete Research, Vol. 33, pp. 191-196.

BROUXEL, M. DELABY, J.-B., DOUAY, F. et ALBAREDE, F. (2004), The inhibitor


effect of Li products on AAR affected concrete. 12th International Conference on Alkali-
Aggregate Reaction, Beijing (Chine), octobre 2004, pp. 625-628.

BULTEEL, D., GARCIA-DIAZ, E. et DÉGRUGILLIERS, P. (2008), Influence of lithium


hydroxide on alkali-silica reaction. 13* International Conference on Alkali-Aggregate
Reaction (ICAAR), Trondheim (Norvège), juin 2008, 10 p.

CANHAM, I., PAGE, C.L., NIXON, P.J. (1987), Aspects of the pore solution chemistry of
blended cements related to the control of alkali-silica reaction. Cement and Concrete
Research, Vol. 17, pp. 839-844.

CHAPMAN, D.L., (1913), A contribution to the theory of electrocapillarity. Philos. Mag.


1913; 2 5 : 475.

CHATTERJI, S. (1989), Mechanisms of alkali-silica reaction and expansion.


8th International Conference on Alkali-Aggregate Reaction, Kyoto (Japon), 1989,
pp. 101-106.

CHATTERJI, S. et KAWAMURA, M. (1992), Electrical double layer, ion transport and


reactions in hydrated cement paste. Cement and Concrete Research, Vol. 22, pp. 774-782.

CHATTERJI, S., JENSEN, A.D., THAULOW, N. et CHRISTENSEN, P. (1986), Studies


of alkali-silica reaction. Part 3. Mechanism by which NaCl and Ca(OH)2 affect the
reaction. Cement and Concrete Research, Vol. 16. pp. 246-254.

COLLINS, C.L., IDEKER, J.H., WILLIS, G.S. et KURTIS, K.E. (2004), Examination of
the effects of LiOH, LiCL, and LiN03 on alkali-silica reaction. Cement and Concrete
Research, Vol. 34, pp. 1403-1415.
52

CSA (2009), Standard CSA A23.1-09 Concrete materials and methods of concrete
constructions, and Standard CSA A23.2-09 Methods of test for concrete. Canadian
Standards Association, Mississauga, Ontario, Canada.

CSA (2008), Standard CSA A3001-08 Cementitious materials for use in concrete.
Canadian Standards Association, Mississauga, Ontario, Canada.

DAVID, R. L. (2009), Handbook of Chemistry and Physics, CRC Press Inc, 90e éd., 2804
P-

DAVIES, G. et OBERHOLSTER, R.E. (1988), Alkalis-silica reaction products and their


development. Cement and Concrete Research, Vol. 18, pp. 621-635.

DE GROSBOIS, M. et FONTAINE, E (2000), Performance of the 60°C-accelerated


concrete prism test for the evaluation of potential alkali-reactivity of concrete aggregates.
11 th International Conference on Alkali-Aggregate Reaction in Concrete, Québec (Canada),
juin 2000, pp. 277-286.

DENT-GLASSER, L.S. (1979), Osmotic pressure and the swelling of gels. Cement and
Concrete Research, Vol. 9, No. 4, pp. 515-517.

DENT-GLASSER, L.S. et KATAOKA, N. (1981), The chemistry of alkali-aggregate


reaction. Cement and Concrete Research, Vol. 11, pp. 1-9.

DIAMOND S. (1981), Effects of two Danish fly ashes on alkali contents ofpore solutions
of cement-fly ash pastes. Cement and Concrete Research, Vol. 11, No. 3, pp. 383-394.

DIAMOND S. (1989), ASR - Another look at mechanisms. 8th International Conference on


Alkali-Aggregate Reaction, K. Okada, S. Nishibayashi et M. Kawamura éds., Kyoto
(Japon), 1989, pp. 83-94.

DIAMOND, S. (2000), Chemistry and other characteristics of ASR gels. 11 th International


Conference on Alkali-Aggregate Reaction (ICAAR), Québec (Canada), juin 2000,
pp. 31-40.
53

DIAMOND, S. et ONG, S. (1992), The mechanisms of lithium effects on ASR.


9th International Conference on Alkali-Aggregate Reaction, London (UK), juillet 1992,
pp. 269-278.

DIAMOND, S., SAHU, S. et THAULOW, N. (2004), Reaction products of densified


silica fume agglomerates in concrete. Cement and Concrete Research, Vol. 34, No. 9,
pp. 1625-1632.

DOLAR-MANTUANI, L. (1983), Handbook of concrete aggregate - a pétrographie and


technological evaluation. Noyes Publication, New Jersey, 345 p.

DUCHESNE, J. et BÉRUBÉ, M.A. (1992), An autoclave mortar bar test for assessing
the effectiveness of mineral admixtures in suppressing expansion due to AAR.
9 International Conference, on Alkali-Aggregate Reaction, Londres (UK), The Concrete
Society, pp. 279-286.

DUCHESNE, J. et BÉRUBÉ, M.A. (1994a), The effectiveness of supplementary cementing


materials in suppressing expansion due to ASR: Another look at the reaction mechanisms -
Part 1: Concrete expansion and portlandite depletion. Cement and Concrete Research,
Vol. 24, pp. 73-82.

DUCHESNE, J. et BÉRUBÉ, M.A. (1994b), Tite effectiveness of supplementary


cementing materials in suppressing expansion due to ASR: Another look at the reaction
mechanisms - Part 2: Pore solution chemistry. Cement and Concrete Research, Vol. 24,
No. 2, pp. 221-230.

DUCHESNE, J. et BÉRUBÉ, M.A. (2000), Long term effectiveness of supplementary


cementing materials against ASR. 11 th International Conference on Alkali-Aggregate
Reaction (ICAAR), Québec (Canada), juin 2000, pp. 613-622.

DURAND, B. (2000), More results about the use of lithium salts and mineral admixture to
inhibit ASR in concrete. 11 th International Conference on Alkali-Aggregate Reaction
(ICAAR), Québec (Canada), juin 2000, pp. 623-632.
54

FENG, N., JIA, H. et HAO, T. (1998), Effect and mechanism of natural zeolite on
suppressing ASR in concrete. ACI SP 178-41, pp. 797-820.

FENG, X., THOMAS, M.D.A, BREMNER, T.W., FOLLIARD, K.J. et FOURNIER, B.


(2008), Summary of research on the effect of LiN03 on alkali-silica reaction in new
concrete. 13th International Conference on Alkali-Aggregate Reaction (ICAAR),
Trondheim (Norvège), juin 2008, 10 p.

FOLLIARD, K.J., THOMAS, M.D.A. et KURTIS, K.E. (2003), Guidelines for the use of
lithium to mitigate or prevent alkali-silica reaction (ASR). U.S. Department of
Transportation, Federal Highway Administration, Publication No. FHWA-RD-03-047,
86 p.

FOLLIARD, K.J, THOMAS, M.D.A., IDEKER, J.H., EAST, B. et FOURNIER, B. (2008),


Case studies of treating ASR-affected structures with lithium nitrate. 13l International
Conference on Alkali-Aggregate Reaction (ICAAR), Trondheim (Norvège), juin 2008,
10 p.

FOURNIER, B. et BÉRUBÉ M.-A. (2000), Alkali-aggregate reaction in concrete : a


review of basic concepts and engineering implications. Canadian Journal of Civil
Engineering, Vol. 27 No. 2, pp. 167-191.

FOURNIER, B. et BÉRUBÉ, M.-A. (2001), L'utilisation du lithium pour combattre la


réaction alkalis-silice (RAS). ACI, Progrès dans le domaine du béton, Montréal, 12 p.

FOURNIER, B., STOKES, D.B. et FERRO, A. (2003), Comparative field and laboratory
investigations on the use of supplementary cementing materials (SCMs) and lithium-based
admixtures to control expansion due to alkali-silica reaction (ASR) in concrete.
6th International CANMET/ACI Conference on Durability of Concrete, Thessaloniki
(Grèce), juin 2003, pp. 823-851.

FOURNIER, B., CHEVRIER, R., DE GROSBOIS, M., LISELLA, R., FOLLIARD, KJ.,
IDEKER, J., SHEHATA, M.,THOMAS, M.D.A., BAXTER, S. (2004), The accelerated
concrete prism test (60°C): variability of the test method and proposed expansion limits.
55

12th International Conference on Alkali-Aggregate Reaction, Beijing (Chine), octobre 2004,


pp.314-332.

GABORIAUD, F. (1999), Étude du rôle de l'ion alcalin au cours de la gélification des


solutions silico-alcalines déstabilisées par addition d'ions calcium, Thèse de doctorat,
Université de Bourgone, U.F.R Sciences et techniques, École Doctorat Carnot, 170 p.

GABORIAUD, F., NONAT, D., CHAUMONT, A. et CRAIEVICH, J. (1999), Aggregation


and gel formation in basic silico-calco-alkaline solutions studied: a SAXS, SANS, and ELS
study. Journal of Physical Chemistry B, Vol. 103, pp. 5775-5781.

GOUY, G. (1910), Sur la constitution de la charge électrique à la surface d'un electrolyte.


J. Phys. Theor., pp. 455-468.

GRATTAN-BELLEW, P.E. (1981), Canadian experience of alkali-expansivity in concrete.


5th International Conference on Alkali-Agregate Reaction, Cap Town (Afrique du Sud),
National Building Research InstituteCSIR, Papier S252/6, mars-avril 1981, l i p .

GUDMUNDSSON, G., HALFDANARSON, J. et MÔLLER, J. (2000), Effect of silica


fume properties on mitigation of ASR reactivity in concrete. II 1 International Conference
on Alkali-Aggregate Reaction (ICAAR), Québec (Canada), juin 2000, pp. 643-651.

HANSEN, W.C. (1944), Studies relating to the mechanism by which the alkali-aggregate
reaction proceeds in concrete. Journal ofthe American Concrete Institute, Vol. 15, No. 3,
pp. 213-217.

HASPARYK, P.N., MONTEIRO, P. J. M. et CARASEK H. (2001), Effect of silica


fume and rice husk ash on alkali-silica reaction. ACI Materials Journal, Vol. 97, No. 4,
pp. 486-492.

HILL, R. et FOLLIARD, K.J, (2006), The impact of fly ash on air-entrained concrete.
Bridge View, Issue 43, Spring 2006, pp. 5-6.
56

HOOTON, D.R. (2000), Canadian use of ground granulated blast-furnace slag as a


supplementary cementing material for enhanced performance of concrete. Canadian
Journal of Civil Engineering. Vol. 27, pp. 754-760.

IDEKER, J.H., EAST, B., FOLLIARD, K.J, THOMAS, M.D.A. et FOURNIER, B. (2010),
The current state of the accelerated concrete prism test. Cement and Concrete Research,
Vol. 40, pp. 550-555.

ILER, R.K. (1956), Tlte chemistry of silica: solubility, polymerization, colloid and surface
properties and biochemistry of silica. John Wiley & Sons, New-York (USA), 866 p.

JASKULA, B.W. (2010), 2008 Minerals yearbook - Lithium. Publications of the U.S.
Bureau of Mines and the U.S.Geological Survey, 10 p.

KAWAMURA, M. et FUWA, H. (2003), Effect of lithium salts on ASR gel composition


and expansion of mortars. Cement and Concrete Research, Vol. 33, pp. 913-919.

KAWAMURA, M. et MATSUSHITA, H. (2007), Suppressing effect of fly ash composition


on expansion of mortar due to alkali-silica reaction. ACI SP 242, pp. 279-292.

KNUTSSON, A. (2010), Freeze/thaw durability of concrete with fly ash. Master's Thesis,
Civil and Environmental Engineering, Chalmers University of Technology, Sweden, 99 p.

KOSMATKA, H.S., KERKHOFF, B., HOOTON, R.D. et MCGRATH R.J. (2011), Dosage
et contrôle des mélanges de béton. 8e édition, Association canadienne du ciment portland,
ISBN: 9781896553221, 411 p..

KURTIS, K.E. et MONTEIRO, P.J.M. (2003), Chemical additives to control expansion of


alkali-silica reaction gel: proposed mechanisms of control. Journal of Materials Sciences,
Vol. 38, pp. 2027-2036.

LANE, D.S. (2002), Laboratory investigation of lithium-bearing compounds for use in


concrete. VTRC 02-R16, Virginia Transportation Research Council, Commonwealth of
Virginia.
57

LARIVE, C. (1998), Apports combinés de l'expérimentation et de la modélisation à la


compréhension de Talcali-réaction et de ses effets mécaniques, Thèse de doctorat,
Laboratoire Central des Ponts et Chaussées, France, 305 p.

LAWRENCE, M. et VIVIAN, H.F. (1961), The reactions of various alkalies with silica.
Australian Journal of Applied Science, Vol. 12, pp. 96-103.

LOMBARDI J., PERRUCHOT A., MASSARD P. et LARIVE C. (1996), Étude comparée


de gels silico-calciques produits des réactions alcalis-granulats dans les bétons et de gels
synthétiques types. Cement and Concrete Research, Vol. 26, No. 4, pp. 623-631.

LONGUET, P., BURGLEN, L. et ZELWER, A. (1973), La phase liquide du ciment


hydraté. Revue des Matériaux de Construction, Vol. 676, pp. 35-41.

LU, D., MEI, L., XU, Z., TANG, M., MO, X. et FOURNIER, B. (2006), Alteration of
alkali reactive aggregates autoclaved in different alkali solutions and application to alkali-
aggregate reaction in concrete (II) expansion and microstructure of concrete
microbar.Cement and Concrete Research, Vol. 36, No. 6, pp. 1191-1200.

LUMLEY, J.S. (1997), ASR suppression by lithium compounds. Cement and Concrete
Research, Vol. 27, No. 2, pp. 235-44.

MARCHAND, J., JOLIN, M., MACHABÉE, Y. (2005), Influence of supplementary


cementing material on the de-icer salt scaling resistance of concrete - Ten years of field
performance. Cement Combinations for Durable Concrete, édité par R. K. Dhir, Thomas
Telford, pp. 579-590

McCOY, W.J. et CALDWELL, A.G. (1951), A new approach to inhibiting alkali-


aggregate reaction. ACI Materials Journal, Vol. 47, pp. 693-706.

McGOWAN, J.K. et VIVIAN, H.E. (1952), Studies in cement-aggregate


reaction: Correlation between crack development and expansion of mortars. Australian
Journal of Applied Science, Vol. 3, pp. 228-232.
58

MEHTA, P.K. (1983), Pozzolanic and cementitious by-products as mineral admixtures for
concrete - A critical review. 1st International Conference on the Use of Fly Ash, Silica
Fume, Slag and Other Mineral By-products in Concrete, Montebello (Canada), V.M.
Malhotra éd., ACI SP-79, pp 1-46.

MEHTA, P.K. et FOLLIARD, K.J. (1995), Rice husk ash - a unique supplementary
cementing material: durability aspects. ACI SP 154-28, Vol. 154, pp. 531-542.

MICHAUD, V. (1995), Simulation expérimentale de l'alcali-réaction des bétons. Étude de


l'influence des ions sulfate. Thèse de Doctorat. Université de Bourgogne, Dijon (France).

MILLARD, M.J. et KURTIS, K.E. (2008), Effects of lithium nitrate admixture on early-age
cement hydration. Cement and Concrete Research, Vol. 38, No. 4, pp. 500-510.

MINISTÈRE DES TRANSPORTS DU QUÉBEC (2008), Recueil des tarifs de camionnage


en vrac du Ministère des Transport, Vol. 3, 18 p.

MITCHELL, L.D., BEAUDOIN, JJ. et GRATTAN-BELLEW, P. (2004), The effect of


lithium hydroxide solution on alkali silica reaction gels created with opal. Cement and
Concrete Research, Vol. 34, pp. 641-649.

MO, X. (2005), Laboratory study of LiOH in inhibiting alkali-silica reaction at 20°C.


Cement and Concrete Research, Vol. 35. pp. 499-504.

MO, X., YU, C. et Xu, Z. (2003), Long-term effectiveness and mechanism of LiOH in
inhibiting alkali-silica reaction. Cement and Concrete Research, Vol. 33, pp.115-119.

NATESAIYER, K.C. et HOVER, K.C. (1992), Chemical agents for reducing solubility of
silica in IN sodium hydroxide, Cement and Concrete Research, Vol. 22, No. 24. pp.653-
662.

OBER, J.A. (2007), 2006 Minerals yearbook - Lithium. Publications ofthe U.S. Bureau of
Mines and the U.S.Geological Survey, 8 p.
59

OBERHOLSTER, R.E. (1984), Cracking of concrete resulting from expansive chemical


reactions. Symposium on cracking in concrete, Port Elizabeth (Afrique du Sud), 8 p.

OCDE (1989), Durabilité des ponts routiers en béton. Groupe d'experts scientifiques.
Organisation de coopération et de développement économiques, Paris (France).

OHAMA, Y., DEMURA, K. et KAKEGAWA, M. (1989), Inhibiting alkali-reaction with


chemical admixture. 8* International Conference on Alkali-Aggregate Reaction, K. Okada,
S. Nishibayashi et M. Kawamura éds., Kyoto (Japon), juillet 1989, pp. 253-258.

OKADA, K., AGAWA, T., ADACHI, M. et TAKAHASHI, K. (1987), Alkali-aggregate


reaction: a study on causative factors. l x International Conference on Alkali-Aggregate
Reaction, Ottawa (Canada), 1986, P.E. Grattan-Bellew éd., Noyes Publications,
pp. 347-350.

PEDNEAULT, A. (1996), Développement de procédures d'essai et d'analyse pour


l'évaluation du potentiel résiduel de réaction, d'expansion et de détérioration du béton
affecté de réactivité alcalis-silice. Mémoire de maîtrise, Université Laval, Québec
(Canada), 132 p.

POWERS, T.C. et STEINOUR, H.H. (1955), An interpretation ofpublished researches on


alkali-aggregate reactivity - Parts I and II. Journal of the American Concrete Institute,
Vol. 51, pp. 497 et 785.

PREZZI, M., MONTEIRO, P.J.M. et SPOSITO, G. (1997), Alkali-silica reaction - P a r t i :


Use of the double-layer theory to explain the behavior of reaction-product gels. ACI
Materials Journal, Vol. 94, No. 1, pp. 10-17.

PREZZI, M., MONTEIRO, P.J.M. et SPOSITO, G. (1998), Alkali-silica reaction - Part 2:


The effect of chemical admixtures. ACI Materials Journal, Vol. 95, No. 1, pp. 3-10.

QI, Y. et WEN, Z.-Y. (2007), Effects of lithium hydroxide on alkali silica reaction gels
created with opal. Construction and Building Materials, Vol. 21, No. 8, pp. 1656-1660.
60

QINGHAN, B., NISHIBAYASHI, S., XUEQUAN, W., YOSHINO, A., HONG, Z.,
TIECHENG, W. et MINGSHU, T. (1995), Preliminary study ofthe effects of LiN0 3 on
expansion of mortars subjected to alkali-silica reaction. Cement and Concrete Research,
Vol. 25, pp. 1647-1654.

RAMEZANIANPOUR, A.A., RAISS, G.A.M. et PARHIZKAR T. (2004), Influence of


silica fume on alkali-silica reaction mitigation. 12th International Conference on Alkali-
Aggregate Reaction, Beijing (Chine), octobre 2004, pp. 646-650.

RAMLOCHAN, T., THOMAS, M. et GRUBER, K.A. (2000), The effect of métakaolin on


alkali-silica reaction in concrete. Cement and Concrete Research, Vol. 30, pp. 339-344.

RANGARAJU, P.R. et OLEK, J. (2000), Evaluation of the potential of densified silica


fume to cause alkali-silica reaction in cementitious matrices using ASTM C 1260 test
procedure. Cement, Concrete and Aggregate, Vol. 22, No. 2, pp. 150-159.

REGOURD, M. (1983), Altération interne des bétons - la réactivité alcalis-granulats.


Travaux, No. 577, pp. 62-11.

RESSOURCES NATURELLES CANADA - SECTEUR DES MINÉRAUX ET


DES MÉTAUX (2005), Annuaire des minéraux du Canada. D. Panagapko éd.,
Section 28.1-28.17 (Granulats).

RESSOURCES NATURELLES CANADA - SECTEUR DES MINÉRAUX ET DES


MÉTAUX (2005), Annuaire des minéraux du Canada. Section 32.1-32.4 (Lithium).

RODRIGUES, F.A., MONTEIRO, P.J.M. et SPOSITO, G. (1999), The alkali-silica


reaction - The surface charge density of silica and its effect on expansive pressure. Cement
and Concrete Research, Vol. 29, No. 4, pp. 527-530.

RODRIGUES, F.A., MONTEIRO, P.J.M. et SPOSITO, G. (2001), The alkali-silica


reaction: the effect of monovalent and bivalent cations on surface charge density of opal.
Cement and Concrete Research, Vol. 31, No. 11, pp. 1549-1552.
61

ROGERS, C A . (1987), Testing Canadian aggregate for alkali reactivity. 7th International
Conference on Alkali-Aggregate Reaction, Ottawa (Canada), Noyes Publication, P.E
Grattan-Bellew éd., pp. 259-263.

SAKAGUCHI, Y., TAKAKURA, M., TAKAHIRO, H. et FUMINORI, T. (1989), The


inhibiting effect of lithium coumpounds on alkali-silica reaction. 8 International
Conference on Alkali-Aggregate Reaction, K. Okada, S. Nishibayashi et M. Kawamura
éds., Kyoto (Japon), juillet 1989, pp. 229-234.

SHAY AN, A. (1992), Tlie 'Pessimum' effect in an accelerated mortar bar test using lm
NaOH solution at 80°C. Cement and Concrete Research, Vol. 14, No. 4, pp. 249-255.

SHAY AN, A. DIGGINS, R. Et IVANUSEC, I. (1996), Effectiveness of fly ash in


preventing deleterious expansion due to alkali-aggregate reaction in normal and steam-
cured concrete. Cement and Concrete Research, Vol. 26, No. 1, pp. 153-164.

SHEHATA, M.H. et THOMAS, M.D.A (2000a), Tlie effects of silica fume and fly ash
hydration products on chemistry of the pore solution and portlandite consumption.
11 International Conference on Alkali-Aggregate Reaction (ICAAR), Québec (Canada),
juin 2000, pp. 753-762.

SHEHATA, M.H. et THOMAS, M.D.A (2000b), The effect offly ash composition on the
expansion of concrete due to alkali-silica reaction. Cement and Concrete Research,
Vol. 30, pp. 1063-1072.

STANTON, T.E. (1940), The expansion of concrete through reaction between cement
and aggregate. Proceedings of the American Society of Civil Engineers, Vol. 66,
pp. 1759-1786.

STARK, D., (1991), The moisture condition of field concrete exhibiting alkali-silica
reactivity. Second International Conference on Durability of Concrete, ACI Publication SP
126/52, pp 973-987.
62

STARK, D. (1992), Lithium salt admixtures - An alternative method to prevent expansive


alkali-silica reactivity. 9th International Conference on Alkali-aggregate Reaction (ICAAR),
Londres (UK), 1992, pp. 1017-1025.

STARK, D., MORGAN, B., OKAMOTO, P. et DIAMOND, S. (1993), Eliminating or


minimizing alkali-silica reactivity. SHRP-C-343, National Research Council, Washington
D.C.

STERN, O. (1924), Zeitschrift fur Elektrochemie and angewandte physikalische Chemie.


Elektrochem,Vol. 30, pp 508-516,

STOKES, D.B. et MANISSERO, C E . (2001), Use of lithium-based admixture for ASR


expansion. 14 Reunion Tecnica Asociacion Argentina de technologia del Hormigon,
Universidad Nacional del Centro de la Provincia de Buenos Aires, Argentine, 10 p.

STRUBLE L. et DIAMOND, S. (1981), Unstable swelling behaviour of alkali silica gels.


Cement and Concrete Research, Vol. 11, No. 4, pp. 611-617.

TANG, M. S. (1981), Some remarks about alkali-silica reactions. Cement and Concrete
Research, Vol. 11, pp. 477-478.

TANG, M.S. et SU-FEN, H. (1980), Effect of Ca(OH) 2 on Alkali-Silica Reaction.


8th International Congress of Cement Chemistry, Paris (France), Vol. 2, pp. 94-99.

THOMAS, M.D.A. (1998), Tlie role of calcium in alkali-silica reaction. Material Science
of Concrete (Sidney Diamond Symposium), The American Ceramic Society, Ohio, USA,
Special volume, pp. 325-337.

THOMAS, M.D.A., SHEHATA, M.H., SASHIPRAKASH, S.G., HOPKINS, D.S. et CAIL,


K., (1999), Use of ternary cementitious systems containing silica fume and fly ash in
concrete. Cement and Concrete Research, Vol. 29, pp. 1207-1214.

THOMAS, M.D.A., HOOPER, R. et STOKES, D. (2000), Use of lithium-containing


compounds to control expansion in concrete due to alkali-silica reaction. 11 International
63

Conference on Alkali-Aggregate Reaction (ICAAR), Québec (Canada), juin 2000,


pp.783-810.

THOMAS, M.D.A., STOKES, D. et RODGERS, T. (2003), The effect of lithium-based


admixtures. ACI SP 217-32, American Concrete Institute, pp. 483-498.

THOMAS, M.D.A., FOURNIER, B., FOLLIARD, K.J., SHEHATA, M., IDEKER J.H. et
ROGERS, C A . (2007a), Performance limits for evaluating supplementaiy cementing
materials using the accelerated mortar bar test. ACI Materials Journal, Vol. 104, No. 2, pp.
115-122.

THOMAS, M.D.A., FOURNIER, B., FOLLIARD, K.J., IDEKER J.H. et RESENDEZ,


Y. (2007b), The use of lithium to prevent or mitigate alkali-silica reaction in concrete
pavements and structures. U.S. Department of Transportation, Federal Highway
Administration, Publication no. FHWA-HRT-06-13, 50 p.

TOUMA, W.E., FOWLER, D.W. et CARRASQUILLO, R L . (2001), Alkali-silica reaction


in Portland cement concrete: testing methods and mitigation alternatives. Report ICAR
301-1F, International Center for Aggregates Research, Austin (USA).

VERNET, C. (1992), La durabilité des bétons (Stabilité chimique des hydrates -


Chapitre 5, Presse de l'école nationale des ponts et chaussées, France, pp. 129-170.

WANG, T. (1995), Effect of blast furnace slag in reducing expansion due to alkali-silica
reaction in concrete. ACI Materials Journal, Vol. 153, pp. 1069-1086.

WIJNEN, P.W.J.G., BEELEN, T.P.M., DE HAAN, J.W., RUMMENS, C.P.J., VAN DE


VEN, L.J.M. et VAN SANTEN, R.A. (1989), Silica gel dissolution in aqueous
alkali metal hydroxides studies by 29Si NMR. Journal of Non-Crystalline Solids, vol. 109,
pp. 85-94.

XU, G.J.Z., WATT, D.F. et HUDEC, P.P.T. (1995), Effectiveness of mineral admixtures in
reducing ASR expansion. Cement and Concrete Research, Vol. 25, No. 6, pp. 1225-1236.
64

YAMAMOTO, T. et TSUTOMU, K. (2000), Chemical assessment method for the


preventive effect of fly ash on alkali-silica reaction. 11 International Conference on
Alkali-Aggregate Reaction (ICAAR), Québec (Canada), juin 2000, pp. 753-762.

ZACHORIA, J. B., WILLIAM, S.C. et STANLEY, R.P. (2008), Alternate mitigation


materials for alkali-silica reaction (ASR) in concrete. Colorado Département of
Transportation, Report No CDOT-2008-10, novembre 2008, 74 p.

ZHANG, Y.M., SUN, W. et YAN, H.D. (2000), Hydration of high-volume fly ash cement
pastes. Cement and Concrete Composites, Vol 22, p. 445-452.
65

3 Présentation du programme de doctorat

Rappelons que les objectifs de cette thèse étaient multiples. Premièrement, il fallait
déterminer l'efficacité de deux produits à base de lithium (nitrate de lithium et verre de
lithium) en présence d'une variété de granulats réactifs et ce, avec ou sans ajouts
cimentaires (cendres volantes, laitiers de hauts fourneaux et ciment ternaire).
Deuxièmement, il y avait lieu de développer un essai rapide et fiable pour l'industrie afin
de déterminer l'efficacité des adjuvants à base de lithium face à la RAS et le dosage requis.
Troisièmement, il fallait préciser les mécanismes à l'origine de l'efficacité du lithium.

Pour atteindre ces objectifs, nous avons débuté par une étude sur l'influence du lithium sur
la composition de la solution interstitielle et des hydrates du ciment. Ces travaux, présentés
dans le premier article intitulé « Influence of lithium-based products proposed for
counteracting ASR on the chemistry ofpore solution and cement hydrates », ont permis de
bien comprendre l'influence du lithium. Cette étude a été cruciale pour la réalisation de la
suite du projet portant sur l'évaluation de l'efficacité des adjuvants à base de lithium et
aussi pour le développement d'un essai rapide sur barres de mortier.

Ainsi, à la suite du premier article, nous avons procédé à l'évaluation de l'efficacité


des adjuvants à base de lithium avec l'essai sur prismes de béton CSA A23.2-14A ou
ASTM C 1293, ce qui a conduit à la publication du deuxième article intitulé « Effectiveness
of lithium-based products in concrete made with Canadian natural aggregates susceptible
to alkali-silica reactivity ». Ces travaux sont le cœur même de cette thèse et les données
obtenues ont été mises à profit pour toute la suite du projet de doctorat.

Fort de ces données, le projet s'est poursuivi en étudiant la possibilité de modifier l'essai
accéléré sur barres de mortier CSA A23.2-25A ou ASTM C 1260 pour permettre d'évaluer
plus rapidement l'efficacité du lithium. Ces travaux ont mené à la publication des troisième
et quatrième articles intitulés «Use of the accelerated mortar bar test to evaluate the
effectiveness of L1NO3 against alkali-silica reaction - Part 1: Pore solution chemistry and
influence of various parameters» et «... Part 2: Comparison with results from the concrete
prism test».
66

Parallèlement au déroulement de tous ces travaux, de nombreux essais ont été réalisés afin
de comprendre les mécanismes par lesquels le lithium est efficace pour contrer la RAS.
L'utilisation des résultats de ces essais et les diverses observations réalisées tout au long de
cette étude a mené à la publication du dernier article intitulé « Experimental investigation of
the mechanisms by which L1NO3 is effective against ASR ».

Les chapitres suivants présenteront les articles dans leur ordre chronologique de
publication, tout en respectant la progression du projet de doctorat.

Les résultats de toutes les analyses chimiques, les descriptions pétrographiques de tous les
granulats utilisés, les fiches détaillées de tous les mélanges de pâte de ciment, de mortier et
de béton, ainsi que les résultats des mesures d'expansion sont présentés sous forme
d'annexés. Toutes ces informations pourraient s'avérer utiles pour la poursuite des études
sur le sujet.
67

4 Influence de produits à base de lithium proposés pour contrer la


RAS sur la composition de la solution interstitielle et des hydrates
du ciment

4.1 Introduction

Cette publication a été soumise à la revue Cement and Concrete Research le 8 octobre 2003
et acceptée le 25 mars 2004. L'article a été publié dans le numéro de septembre 2004 de
cette revue (volume 34, numéro 9) dédié à la mémoire de H.F.W. Taylor.

4.2 Résumé

Des pâtes avec un ciment à basse teneur et avec un ciment à haute teneur en alcalis ont été
fabriquées avec ou sans nitrate de lithium (LiNOs) ou un verre contenant du lithium. Un
rapport constant [Li]/[Na+K] de 0.74 a été utilisé lors d'ajouts de lithium dans les pâtes.
Les spécimens ont été placés dans des récipients scellés et entreposés à trois différentes
températures, soit 23, 38, et 60°C La solution interstitielle a été extraite et analysée après
3, 7, 28 et 91 jours. De plus, la teneur en eau résiduelle a été déterminée par séchage. Le
verre de Li réagit lentement, la [Li+] en solution augmentant progressivement avec le
temps, la température, la finesse (verre de lithium original vs. broyé plus finement) et la
[Na++K+]. L'utilisation du verre de Li (original) augmente le pH d'environ 0.1, et d'environ
0.2 lorsque plus finement broyé. En revanche, l'ajout de LiNÛ3 diminue le pH d'environ
0.1, en dépit de l'augmentation significative de la [Na + + K+] dans la solution interstitielle.
Plus le contenu total en Na20é (Li compris) est élevé dams le mélange original, plus la
quantité d'alcalis incorporée dans les hydrates de ciment est aussi élevée. Le ratio
[Li+]/[Na++K+] de la solution interstitielle représente environ la moitié du ratio initialement
utilisé dans le mélange (0.74), tandis que ce ratio dans les hydrates de ciment est toujours
de plus de 1.1. Le Li est l'alcali le plus fortement incorporé aux hydrates de ciment, tandis
que le K est le plus faiblement incorporé.
68

4.3 Publication scientifique no. 1

Influence of lithium-based products proposed for counteracting


ASR on the chemistry of pore solution and cement hydrates

M.A. Bérubé a, C. Tremblay b, B. Fournier 0 , M.D.A. Thomas d, D.B. Stokes


a
Professor, Département de géologie et de génie géologique, Université Laval, Québec, QC, Canada, G1K 7P4
b
PhD Student, Département de géologie et de génie géologique, Université Laval, Québec, QC, Canada, G1K. 7P4
c
Research Scientist, ICON/CANMET, Natural Resources Canada, 405 Rochester, Ottawa, ON, Canada, Kl A 0G1
d
Professor, Department of Civil Engineering, University of New Brunswick, Fredericton, NB, Canada, E3B 5A3
e
Manager, Concrete Technology, FMC Lithium Division, 449 N. Cox Rd., Gastonia, NC 28054, USA

ABSTRACT

Low- and high-alkali cement pastes were made with or without LiN03 or a Li-bearing
glass. The [Li]/[Na+K] molar ratio was kept constant to 0.74. The specimens were stored at
23, 38, and 60°C in sealed containers. After 3, 7, 28, and 91 days, their pore solutions were
extracted and analysed and their residual water contents were obtained by drying. The Li
glass was found to react quite slowly and the corresponding [Li+] in solution progressively
increased with time, temperature, fineness (as-received glass vs ground glass), and the
[Na++K+] concentration in solution. This glass increased the pH by about 0.1, and by about
0.2 after it was finely ground. In contrast, LiNÛ3 decreased the pH by about 0.1, despite
significantly increasing the [Na++K+] in the pore solution. The higher the total %Na20e
content (including Li) in the original mixtures, the higher the total alkali content
incorporated in the cement hydrates. The [Li+]/[Na++K+] ratio in solution was about half of
the initial ratio (0.74), while this ratio in the cement hydrates was always over 1.1. Li is the
alkali the most preferentially incorporated into the cement hydrates, while K is the least.

Keywords: Alkali-Silica Reaction (C); Alkalis (D); Lithium (D); Cement Paste (E); Pore
Solution (B)
69

1 Introduction

The alkali-silica reaction (ASR) is a chemical reaction between the highly basic
(pH > 13) and alkaline (NaOH, KOH) concrete pore solution and certain "reactive" silica
mineral species inside the aggregate particles. Following this internal chemical attack, the
reactive silica transforms into a silico-calco-alkaline gel that swells in the presence of
water, and the swelling may cause expansion of the reactive aggregate particles, stresses on
the surrounding cement paste, internal microcracking, expansion of the concrete element,
surface macrocracking, and significant reduction of the intended service life of the
structure. The extent of the reaction depends upon numerous factors; the reactivity of the
aggregate to counteract, the concrete alkali content, and the moisture content and humidity
conditions are the most critical. When reactive aggregates have to be used in concrete, the
most popular preventive measures consist of limiting the concrete alkali content and/or
using sufficient amounts of effective supplementary cementing materials (SCMs) or
chemical admixtures [1].

McCoy and Caldwell [2] first observed that certain lithium-based products were capable of
significantly reducing expansion due to ASR. After a limited number of studies confirmed
this point during the following decades [3-5], the interest for such products resurrected in
the early nineties under the SHRP program in the USA [6, 7]. More recent studies also
confirmed the positive effect of combining lithium compounds and fly ash [8-11]. At first,
the most commonly tested lithium-based products, i.e. LiOH, LiCl, LiF, LÎ2C03, and
LiNÛ3, seemed to produce relatively similar expansion reductions when used in concrete at
similar ([Li]/[Na+K]) molar ratios [12, 13]. However, a number of studies also
demonstrated that insufficient dosages of LiOH, LiC03, and LiF may actually increase
ASR expansion of concrete rather than reduce it, while even increasing the pH of the
concrete pore solution [7, 9, 14, 15]. On the other hand, L1NO3 appears not susceptible to
such a pessimum effect [1], while not significantly increasing the pH ofthe concrete pore
solution [15, 16]. A recent study performed at CANMET (Canada) with three highly-
reactive aggregates also suggests that LiN03 is generally more effective against ASR than
LiOH at similar molar ratios [11].
70

At first sight, the effectiveness of lithium for counteracting ASR is surprising since Li (like
Na and K) is part of the alkali group in the periodic table. As mentioned before, some
previous studies indicated that certain lithium salts may contribute in increasing the pH of
the concrete pore solution and may even increase expansion. Lithium may also be involved
in the reactions with reactive aggregates since it was found in the ASR gel [14, 17].
According to Mo et al. [17], because having a smaller ionic radius and a higher surface
charge density, the Li+ ions are more readily incorporated in the ASR product than Na+ or
K+ ions, and the lithium-bearing ASR product is crystalline and non-expansive. However,
the results of Mitchell et al. [18] rather suggest that ASR is inhibited and perhaps stopped
in the presence of lithium (LiOH in this case) and support the theory of the formation of a
protective layer of a non-expansive lithium-bearing silicate on the surface of the reactive
aggregate particles.

Most studies performed until now indicated that the effectiveness of lithium against ASR is
mainly affected by the concrete alkali content; such studies suggest that a dosage of
lithium-based products corresponding to [Li]/[Na+K] molar ratios ranging between 0.6 and
1.0 is sufficient to eliminate the expansion with most reactive aggregates [1, 4, 9, 10, 12,
19]. However, the effective lithium dosage may vary significantly from one reactive
aggregate to another, certainly as a function of their degree of reactivity, but also of their
mineralogical composition. For instance, LiN03 was found to be effective with some
extremely reactive aggregates, but relatively less effective with some less reactive ones
[11, 20]. So, the exact mechanism by which lithium can suppress ASR expansion is still not
fully understood [20].

Most experimental programs performed until the early nineties involved mortar bars
incorporating extremely reactive materials such as cristobalite, opal, and Pyrex. However,
the tests generally performed at the time (ASTM C 227 or C 441) are more and more being
considered unrealistic and/or unreliable when used for determining the effectiveness of
preventive measures against ASR [21]. More recently, extensive test programs involving
expansion tests on concrete specimens incorporating natural reactive aggregates have been
conducted in UK, USA, and Canada. In fact, testing concrete specimens in humid air at
38°C and >95% RH, for instance following the Concrete Prism Test CSA A23.2-14A or
71

ASTM C 1293, is considered the most reliable, or possibly the only reliable test method for
correctly determining the effectiveness of lithium-based products against ASR. As used in
Canada and USA, this test takes one year to evaluate the degree of reactivity of concrete
aggregates; however, the Canadian Standards Association stated that the testing period shall
be extended to two years when testing preventive measures against ASR [1]. Such a
duration is too long in many circumstances. Storing the test prisms at 60°C rather than
38°C may shorten the testing period, for instance about four times, as it does when such a
modification is used for testing aggregates [22]. On the other hand, the accelerated mortar
bar method CSA A23.2-25A or ASTM C 1260, which takes only two weeks for testing
both aggregate reactivity and effectiveness of SCMs against ASR, is not realistic for
evaluating the effectiveness of lithium-based products. The mortar bars in this test are
immersed in a IN NaOH solution at 80°C, which causes leaching through diffusion ofthe
lithium initially present in the bars, thus likely reducing the beneficial effect of lithium
against ASR [20]. Considering the rapidity and the great popularity of this test, some
researchers proposed adding lithium to the test solution [7]. At first, such an approach
appeared very interesting, but convincing results have not been obtained up to now. An
important question arises about the lithium concentration of the soak solution, which in
principle should be representative of the pore solution that prevails at equilibrium in real
concrete.

2 Research Significance

This study is part of an extensive research project whose objectives are: (1) to evaluate the
effectiveness of two lithium-based products (LiN03 and lithium-bearing glass) against a
variety of Canadian reactive aggregates with different composition and reactivity, in
combination or not with Canadian supplementary cementing materials (SCMs); (2) to
determine the reaction mechanisms that could explain such an effectiveness, taking into
account the nature of the reactive aggregate to counteract, and (3) to propose a rapid test
method for determining this effectiveness. As part of the investigations on the reaction
mechanisms, this study on cement pastes was undertaken in order to better assess the effect
of lithium on the hydraulic system by itself, i.e. in the absence of aggregates. This is
72

assessed as a function of the following parameters: (1) type of lithium-based product


(LiN03 and Li-glass); (2) concrete alkali content (low- and high-alkali cements); (3) nature
and origin of the alkali load (high-alkali cement vs low-alkali cement raised at the same
alkali level by adding NaOH or KOH to the mixture water); (4) temperature, and (5) time.
The principal objective is to establish the distribution of alkalis (Na, K, and Li) in the pore
solution and cement hydrates as function of all the parameters mentioned above. The
concrete alkali content, the origin and the nature of the alkalis, and the variation of
temperature were all investigated based on the fact that all test methods used for testing the
effectiveness of preventive measures against ASR, including the use of lithium-based
products, are performed using temperatures (38, 60, or 80°C) and alkali contents (NaOH
addition or immersion in IN NaOH solution) higher than in the field. The lithium-based
products may behave differently as a function of these parameters. The time factor is also
of importance in order to assess the chemical stability of the cement hydrates and of the
other reaction products formed.

3 Materials and Methods

3.1 Cements and Lithium Products

Two CSA Type 10 (ASTM Type I) cements were used, a low-alkali cement (0.51% Na2Ûe)
and a high-alkali cement (1.25% Na20e). The composition ofthe low-alkali cement is:
63.6% CaO, 19.7% Si0 2 , 4.74% A1203, 3.58% Fe 2 0 3 , 0.21% Ti0 2 , 1.35% MgO,
2.77% S0 3 , 0.20% Na 2 0, 0.47% K 2 0, 3.71% LOI (loss on ignition); the composition of
the high-alkali cement is: 62.7% CaO, 20.5% Si0 2 , 4.69% A1203, 3.23% Fe 2 0 3 ,
0.26% Ti0 2 , 3.72% MgO, 3.25% SO3, 0.43% Na 2 0, 1.23% K 2 0, 0.84% LOI. Two
commercially available lithium-based products were used: a nitrate solution containing
30% LiN03 by mass, and a Li-bearing glass containing about 20% Li 2 0 by mass and with
the following composition : 26.1% CaO, 50.8% Si0 2 , 1.51% A1203, 0.12% Fe 2 0 3 , 0.20%
Ti0 2 , 0.61 MgO, <0.1% SO3, 0.16% Na 2 0, 0.07% K 2 0, 0.20% Na2Oe, 19.6% Li 2 0,
0.79% LOI. The lithium glass was used both as received (14% < 10 pm) and after having
been finely ground (53% < 10 pm).
73

3.2 Preparation of Cement Paste Specimens and Storage Conditions

Low-alkali and high-alkali cement pastes were made with the two cements, with and
without the three lithium products (LiNÛ3 solution, as-received Li-glass, and ground
Li-glass), always using a water-to-cement ratio of 0.42 (Table 4.1). For a number of
mixtures made with the low-alkali cement, NaOH or KOH was added to the mixture water
in order to reacfrff.35% Na2Oe by mass of cement (based on Na and K only). For all
mixtures incorporating LiN03 or Li-glass, the total [Li]/[Na+K] molar ratio (including
cement, NaOH or KOH added, and LiN03 or Li-glass) was controlled at 0.74. The paste
specimens were cast into cylindrical plastic bottles, 40 mm in diameter by 65 mm in length,
with the bottles left opened and resting in a vertical position for 24 hours over water in
sealed plastic containers at 23°C. After this initial curing period, the plastic bottles were
capped; some containers remained at 23°C, while others were stored at 38°C (temperature
controlled-room) or 60°C (oven). Specimens of each mixture were subjected to all three
temperatures.

3.3 Periodic Measurements

At specific ages, i.e. 7, 28, and 91 days after the initial curing (and also after 3 days for the
specimens stored at 60°C), paste cylinders of each mixture were removed from their plastic
bottle, weighed (WO) and cooled at room temperature. Samples of pore solution were then
expressed under high pressure (1000 MPa). The equipment used has a cylindrical chamber
of 50 mm in diameter by 90 mm in length such that the cylindrical paste specimens could
be tested as is. After expression, the pore solutions were chemically analysed using ICP
(Inductively Coupled Plasma) for Na, K, and Li. Their OH" concentrations were determined
by direct titration with standardized HC1 down the phenolphtalein end point, and their pH
were calculated from these concentrations (= 14 - log(l/[OH"]). As mentioned above, all
pore solutions were expressed at room temperature and the above relationship was used for
calculating their pH; however, the actual concentrations prevailing at 38 and 60°C (for the
specimens stored at these temperatures) could be a little different from those measured at
74

23°C, because of variations of all solubility products with temperature, including the
dissociation product of water (pkw). The specimens tested were also weighed after
extraction (Wl) in order to determine, by difference, the amount of pore solution
extracted using the pore expression technique. They were then allowed to dry at 80°C
until mass equilibrium (W2). The % of pore water (extracted pore water + water
evaporated during drying) in the original specimens were then calculated
(100 • [WO - W2]/W0), and the percentage of the total pore solution extracted were
calculated (100 • [WO - W1]/[W0-W2]).

3.4 Distribution of Na, K, and Li in Pore Solutions and Cement Hydrates

The total amounts of Na, K, and Li in the original specimens were calculated from the
mixture proportionings (cement, water, NaOH, KOH, and lithium additions). The amounts
of Na, K, and Li in the pore solution and cement hydrates (however including the unreacted
lithium glass, when used), were calculated from the water content of the paste specimens,
assuming that the solution remaining inside the specimens after extraction, but evaporated
after drying, was chemically similar to the pore solution extracted and analysed. For a
better comparison between the three types of alkali involved (Na, K, Li), all results are
expressed as Mol/L (pore solutions) or Mol/kg (solids) of Na, K, and Li.

4 Results and Discussion

4.1 Pore Water Contents and Proportions of Pore Solution Expressed

As expected, the pore water content tends to decrease slightly with time, due to more
complete hydration, on average by about 2% between 7 and 91 days (Table 4.2). The
amounts of pore solution extracted represent between 5 and about 25% of the total water
content, but are often less than 10% after 91 days. A slightly larger percentage can
generally be extracted from the pastes stored at 60°C, especially after 91 days. Since all
paste mixtures were subjected to the same load of 1000 MPa, this could likely reflect a
coarser or more porous microstructure at 60°C, then the development of a more
compressible paste.
75

Globally, the pore water content and the proportion of extracted pore solution do not seem
to be influenced by the alkali content of the system. However, the pore water content is
often higher by about 1 % (on a total of around 18%) in the mixtures containing lithium
nitrate than in corresponding control mixtures, no matter the temperature and the time,
suggesting that the presence of nitrate ions in solution slightly affects the cement hydration.

Considering the relatively low percentages of pore solution extracted, even using a load of
1000 MPa, a question may be raised about the representativity ofthe samples extracted.
However, the pore solution that is chemically aggressive against reactive aggregates likely
corresponds to the solution located in the vicinity of the aggregate particles (porous
transition zone), which is possibly the solution the most easily and the most rapidly
extracted.

4.2 pH of Pore Solutions

The pH in the pore solutions, computed from the OH" concentrations, vary between
13.45 and 14.11 (Table 4.3); it seems to remain relatively constant with time, except
perhaps at 60°C, where a slight decrease of about 0.1 is generally observed with aging
pastes. The pH also tends to be slightly less, by about 0.2, at 60°C compared to 23°C, for
all mixtures; however, as mentioned before, the results were not corrected for temperature,
all pore solution expressions and OH" analyses having been made at 23°C

On average for the low-alkali mixtures (series 0.51% ± Li), the pH is about 0.3-0.4 lower
than for the high-alkali mixtures made with the same low-alkali cement but increased to
1.25% Na2Oe (series 0.51% + NaOH ± Li and 0.51% + KOH ± Li), as expected. Moreover,
the pH of the pore solution does not seem to be influenced by the type of alkali addition,
i.e. NaOH or KOH. However, it is slightly higher, i.e. by about 0.1, for the pastes made
with the high-alkali cement (series 1.25% ± Li) compared to those with the same %Na20e
(excluding Li) but made with the low-alkali cement plus NaOH or KOH. An hypothesis is
proposed later to explain this point.
76

On average for each series of mixtures, the pH is increased by about 0.1 in the presence of
the Li-glass as-received, and by about 0.2 in the presence ofthe ground glass (Table 4.3). It
must be mentioned that the Li-glass contains alkalies (0.20% Na20e), which were not
considered when batching, and that it dissolves more rapidly when it is ground, as
discussed hereafter.

On the other hand, the pH is decreased by about 0.1 in the presence of LiN03, despite a
higher overall [Na++K++Li+] concentration in the pore solution; this is surely related to the
presence of nitrate ions (not analyzed), which compete with the OH" ions. The results are
quite similar to those reported by Diamond [15], except that the pH was a little higher in the
presence of LiN03 compared with mixtures without lithium. This author suggests that
lithium added as LiN03, LiCl or Li2S04 would not generate LiOH in solution and would
not increase the pH, in contrast with other forms such as LiF, Li2C03 and LiOH.

4.3 Alkali Concentrations of Pore Solutions

For each storage temperature, the [Na+], [K+], [Na++K+], and [Na++K++Li+] of the pore
solution tend to slightly increase up to 28 days, then reach a steady state (Table 4.3 and
Figures 4.1 to 4.3). Moreover, the higher the initial Na and K content in the mixtures, the
higher the corresponding concentrations in the pore solution. The same trends are observed
for the [Li+] concentration for mixtures containing LiNÛ3; this may suggest that
corresponding lithium-bearing hydrates are stable with time as concerns releasing of
alkalies ions in the long term. However, in the presence of the Li-glass, the [Li+]
concentration tends to increase slowly with time due to the progressive dissolution of the
glass (Figures 4.ID, 4.2D, 4.3D).

The [Na+] is reduced if KOH is added to the mixture water (Table 4.3 and Figure 4.2A:
0.5% vs 0.5% + KOH; 0.5% + LiN0 3 vs 0.5% + KOH + LiN0 3 ). Interestingly, the [Na+] in
the pore solution of the high-alkali mixtures increases significantly when LiN03 is
added (Table 4.3 and Figures 4.1 A, 4.2A, 4.3A: 0.5% + KOH vs 0.5% + KOH + LiN0 3 ;
1.25% vs 1.25% + LiN0 3 ; 0.5% + NaOH vs 0.5% + NaOH + LiN0 3 ). This is attributed to
the fact that Li is the alkali preferentially incorporated into the cement hydrates. However,
77

such a phenomenon was not observed in the case of the low-alkali mixtures
(0.5% vs 0.5% + LiNÛ3), which suggests in turn that the competition between the various
types of alkali ions for the hydrates is less severe when the total alkali concentration is low;
the hydrates are then able to accommodate each type.

The [K+] is decreased when NaOH is added (Figures 4. IB, 4.2B, 4.3B:
0.5% vs 0.5% + NaOH; 0.5% + LiN0 3 vs 0.5% + NaOH + LiN0 3 ). As previously seen for
the [Na+], the [K+] in the high-alkali mixtures increases significantly in the presence of
LiN0 3 (Figures 4.1B, 4.2B, 4.3B: 0.5% + NaOH vs 0.5% + NaOH + LiN0 3 ; 1.25% vs
1.25% + LiN0 3 ; 0.5% + KOH vs 0.5% + KOH + LiN0 3 ).

Figures 4.1C, 4.2C, and 4.3C clearly show that the addition of LiN0 3 results in
significantly greater [Na++K+] concentrations in the pore solution of the high-alkali
mixtures. Interestingly, these concentrations are not the same for each of the three series of
high-alkali mixtures made with similar alkali contents. For the pastes with 1.25% Na2Oe
(by mass of cement), with or without lithium addition, the highest [Na++K+] concentrations
are obtained for the mixtures made with the high-alkali cement (1.25% Na2Oe). This is
followed by the mixtures made with the low-alkali cement (0.51% Na2Oe) plus KOH (see
Figure 4.2C). The mixtures made with the same cement plus NaOH exhibit the lowest
[Na++K+] concentrations in the pore solution (for the high alkali mixes). This suggests that,
for similar total alkali contents, alkalis added to the mixture water, all available at the early
stage of cement hydration, are more readily incorporated in the cement hydrates than alkalis
initially present in the cement, especially those as solid solution in C3A and C3S; however,
the unreleased cement alkalies are progressively released with time, then enrich the pore
solution and increase its pH. The results also suggest that Na is more easily incorporated
into the cement hydrates than K. Consequently, the use of a high-alkali cement could be
more aggressive towards reactive aggregates than a cement with a lower alkali content that
is increased at the same level by adding NaOH or KOH, since a relatively lower proportion
of the total alkalies is incorporated in the hydrates, then a higher proportion remains in the
pore solution.
78

4.4 Lithium Concentrations and [Li]/[Na+K] Molar Ratio of Pore Solutions

As mentioned before, when the Li-glass is used, the [Li+] tends to slowly and progressively
increase with time (Figures 4.ID, 4.2D, 4.3D). Moreover, the [Li+] concentrations are
significantly lower than those obtained for the mixtures incorporating similar molar
amounts of lithium as LiN03, with a few exceptions, i.e. after 91 days at 60°C (Figure
4.3D), and at 38°C when the glass is finely ground (Figure 4.2D). For the Li-glass mixtures,
the facts that the [Li+] increases with time (Figures 4.ID, 4.2D, 4.3D), temperature (Figures
4.3D vs 4.2D vs 4.ID), fineness (Figure 4.2D), and [Na++K+] (Figures 4.ID, 4.2D, 4.3D:
0.5% + LiG vs 0.5% + NaOH + LiG or 1.25% + LiG), clearly indicates that the Li-glass as-
received does not release lithium rapidly, at least under normal conditions of temperature
and pH. For the high-alkali mixtures incorporating LiNÛ3, the higher the original Li
content in the mixture, the higher the residual [Li+] in the pore solution, which is not
surprising. For a same Na2Oe content of 1.25% by mass of cement in the original mixture
(excluding lithium), the [Li+] is not significantly affected by the nature of the other alkalies
in the system (Figures 4.ID, 4.2D, 4.3D: 1.25% + LiN03= 0.5% + KOH + LiN03= 0.5% +
NaOH + LiN0 3 ).

For all mixtures incorporating LiN03 or Li-glass, the total [Li]/[Na+K] molar ratio was
controlled at 0.74. Figures 4.IF, 4.2F, and 4.3F indicate that this ratio in the pore solution is
significantly less than as added. For the mixtures made with LiN03, it generally ranges
between 0.30 and 0.45, whatever the temperature and the age of the specimens; in other
words, only half or less of the added lithium is found in the pore solution, as observed by
Diamond [15]. For the mixtures incorporating the Li-glass as-received, the ratio increases
progressively with time from almost 0 to about 0.15 at 23°C or 38°C, and from 0.05 to 0.35
at 60°C With the ground glass tested only at 38°C, this ratio is higher, increasing with time
from 0.15 to 0.45.

If the accelerated mortar bar test CSA A23.2-25A or ASTM C 1260 is used to evaluate the
effectiveness of soluble Li salts against ASR by incorporating lithium in the test solution,
the results obtained in this study suggest that this ratio in the test solution should be
79

adjusted with respect to the actual ratio observed at equilibrium in the pore solution of non-
immersed control specimens.

4.5 Proportions of Alkalies in Pore Solutions and Cement Hydrates

The results obtained for the proportions of Na, K, Na+K, and Li in the pore solution
are presented in Figures 4.4 (23°C), 4.5 (38°C), and 4.6 (60°C). The calculations are based
on the total (design) amounts of alkalis in the original mixtures (Table 4.2), the pore
water contents (Table 4.2), and the alkali concentrations in the extracted pore solutions
(Table 4.3). The following observations are drawn regarding the control mixtures
(0.5%, 0.5% + NaOH, 0.5% + KOH, and 1.25%) and the mixtures containing LiN0 3 .

Independent of temperature, the proportions of Na, K, and Na+K in the pore


solution of such mixtures tend to slightly increase up to 28 days, then to decrease
slightly. However, it must be noted that for each mixture, only one measurement
was made after 28 days;

Also independent of temperature, the proportion of Li in the pore solution of the


LiN03 mixtures generally tends to decrease more rapidly at early age, likely
because Li is readily available and incorporated in cement hydrates during this more
active phase of hydration. This result agrees with Diamond [15] who reported that
as much as half of the added lithium as LiN03 is incorporated in the cement
hydrates during the period of active hydration. The lithium so removed from the
pore solution is presumably not available for subsequent incorporation into ASR gel
[15] or to prevent ASR;

• On average for the LiNÛ3 mixtures, the proportions of Na, K, and Li in the pore
solution after 91 days are respectively about 55%, 80%, and 35% (Figures 4.4 to
4.6) and the effect of temperature is not very significant. It is clear that Li is
incorporated in greater proportions than Na in the cement hydrates, which in turn is
incorporated in greater proportions than K;
80

The effect of LiN03 on the other alkali ions in the pore solution is also clear in the
case of the high-alkali control and L1NO3 mixtures: in presence of LiNÛ3, the
proportions of Na and K in the pore solution significantly increase (Figures 4.4C,
4.5C, 4.6C for Na+K), which again supports the fact that Li is more preferentially
incorporated in hydrates than Na and K;

For the control and LiN03 mixtures made with the same total alkali content,
i.e. 1.25% Na2Oe (excluding Li), the proportions of Na, K, and Na+K in
the pore solution tend to be highest when using the high-alkali cement, lowest when
using the low-alkali cement plus NaOH, while the mixtures made with the same
cement plus KOH fall between or approach the high-alkali cement mixtures
(Figures 4.4C to 4.6C: 1.25% > 0.5% + KOH > 0.5% + NaOH; 1.25% + LiN0 3 =
0.5% + KOH + LiN03> 0.5% + NaOH + LiN0 3 ). As mentioned before, based on
the alkali concentrations, this suggests again that alkalies added as NaOH or KOH
are slightly more readily incorporated in the cement hydrates than alkalies initially
included in the anhydrous cement particles, and that Na is more incorporated
thanK.

For the mixtures containing Li-glass, the proportions of Li in the pore solution tend to
progressively increase with time (Figures 4.4D, 4.5D, 4.6D); moreover, the higher the
temperature (Figures 4.4D vs 4.5D vs 4.6D) and the finer the glass (Figure 4.5D), the
greater the rate of release of Li and the greater the ultimate proportion of Li in the pore
solution (i.e. after 91 days). These ultimate proportions for the glass as-received are around
9% at 23°C, 12%) at 38°C, and 23% at 60°C In the mixtures incorporating the ground glass,
tested only at 38°C, the proportions at 91 days range from 20 to 35%. These results clearly
demonstrate that the lithium glass as-received progressively dissolves at a relatively slow
rate in the pore solution, but that a fine grinding significantly improves its reactivity.
81

4.6 ILi]//[Na+K] Molar Ratio and Total Alkali Contents of Cement Hydrates

Starting with an overall [Li]/[Na+K] ratio of 0.74 (Table 4.2) and considering that this ratio
is usually less than 0.50 in the pore solutions (Table 4.3 and Figures 4.IF, 4.2F and 4.3F),
this ratio is necesssarily more than 0.74 in the solids. The high ratios observed for the
Li-glass mixtures are because there is still a large proportion of unreacted solid Li-glass
particles even after 91 days. However, for the LiNÛ3 mixtures, the above ratio corresponds
to the Li incorporated in the cement hydrates only. It varies between 1.1 and 1.6 for the
0.5% + LiN0 3 mixtures, 1.1 and 1.5 for the 0.5% + NaOH + LiN0 3 mixtures, 1.8 and 1.9
for the
0.5% + KOH + LiN0 3 mixtures, and 1.7 and 2.5 for the 1.25% + LiN0 3 mixtures. For the
three series of high-alkali mixtures, the ratio is thus highest with the high-alkali cement,
intermediate with the low-alkali cement plus KOH, and lowest with the same cement plus
NaOH. These results agree with the highest, intermediate, and lowest proportions of Na and
K in the corresponding pore solution, respectively. There is no clear variation of the
[Li]/[Na+K] ratio in hydrates with time or temperature.

The total %Na20e contents (including Li) in the cement hydrates of the control and LiN03
mixtures are illustrated in Figure 4.7. These contents are equal to the sum of the three
individual Na, K, and Li contents, expressed in moles of Na+K+Li per kg of solids
(hydrates), multiplied by 0.031. One can note that the %Na20e in hydrates is relatively
constant with time for each mixture, and not significantly affected by temperature. The
most important parameter is the total Na20e in the original mixture: the higher this content,
the higher the alkali content in the cement hydrates. A similar observation was made by
Duchesne and Bérubé [23] in a study involving a variety of SCMs (2 silica fumes, 3 fly
ashes, and 1 blast-furnace slag). Evidently, in the present study, the total Na20e content
(including Li) in the original mixtures is higher when LiN03 is added to the mixture. For
mixtures with the same total alkali contents without lithium addition (e.g. 1.25% using the
high alkali cement, the low alkali cement + NaOH or the low-alkali cement + KOH), the
alkali content ofthe cement hydrates is least with the high-alkali cement, intermediate with
the low-alkali cement plus KOH, and highest with the same cement plus NaOH. Once
again, this confirms that cement alkalis are less readily incorporated into the cement
82

hydrates than alkalis already available in the pore solution at the time of mixing (i.e. as
NaOH or KOH addition), and that Na is more incorporated than K (NaOH vs KOH
addition). However, the total alkali content in the hydrates of LiNÛ3 mixtures is more
constant, likely because Li takes the place available in the hydrates when the other alkalies
are less available.

5 Conclusion

This study on cement pastes was undertaken to better assess the effect of lithium on the
chemistry of the pore solution and cement hydrates of hydraulic systems, and the effect of
temperature. The following conclusions can be drawn.

The pH of the pore solution generally decreased by about 0.1 when adding LiN03,
while it increased by about 0.1 in the presence of a Li-bearing glass, and by about
0.2 when this glass was finely ground;

• For the mixtures made with the Li-glass, the [Li+] concentration increased with
time, temperature, glass fineness (glass as-received vs ground glass), and the
[Na++K+] concentration in the pore solution. The results obtained clearly
demonstrate that the Li-glass does not rapidly release lithium, at least under normal
conditions of temperature and pH;

• On average for the mixtures incorporating LiN03, the proportions of Na, K, and Li
in the pore solution after 91 days (the oldest specimens tested) were about 55%,
80%, and 35% ofthe original load, respectively, and the effect of temperature was
not very significant. It is clear that Li was incorporated into the cement hydrates in
greater proportions than Na and K, with Na being always incorporated in greater
proportions than K, even in the absence of Li;

• For all mixtures incorporating LiN03, the [Li+]/[Na++K+] molar ratio in the pore
solution was usually between 0.30 and 0.45, i.e. about half of the original value of
0.74. On the other hand, this ratio was always >1.1 in the cement hydrates. Both
ratios were not signific-antly influenced by temperature and time;
83

• The total %Na20e content (including Li) of the cement hydrates in the control and
LiN03 mixtures was relatively constant with time for each mixture, and was not
significantly affected by temperature. However, the higher the total Na2Oe content
in the original mixtures, the higher the amount of alkalies incorporated into the
cement hydrates;

• The overall alkali composition of the pore solution and the cement hydrates appear
not significantly influenced by temperature, at least between 23 and 60°C This
supports the efforts currently done in order to develop accelerated test methods at
relatively high temperatures for evaluating the effectiveness against ASR of Li-
based products, for instance a concrete prism test performed at 60°C;

• The proportion of pore water expressed under 1000 MPa represented only between
5 and 26% ofthe total water content, and was often less than 10% after 91 days. A
somewhat greater percentage of the pore solution was generally expressed from the
mixtures stored at 60°C, especially after 91 days, possibly because of a more easily
compressible microstructure.

6 Acknowledgements

This study was financially supported by Euclid Admixtures Canada Inc. and the National
Science and Engineering Research Council of Canada (NSERC).

7 References

[1] CSA A23.1-00, Concrete Materials and Methods of Concrete Construction,


Appendix B - Alkali-aggregate reaction, Canadian Standards Association, Rexdale,
Ontario, Canada, 2000.

[2] W.J. McCoy, A.G. Caldwell, New approach in inhibiting alkali-aggregate


expansion, ACI Materials Journal, 22 (9) (1951) 693-706.
84

[3] M. Lawrence, H.F. Vivian, The reactions of various alkalies with silica, Australian
Journal of Applied Science, 12 (1961) 96-103.

[4] Y. Sakaguchi, M. Takakura, A. Kitagawa, T. Hori, F. Tomosawa, M. Abe, The


inhibiting effect of lithium compounds on ASR, 8l International Conference on
AAR in Concrete, Kyoto, Japan, 1989, pp. 229-234.

[5] Y. Ohama, K. Demura, M. Kakegawa, Inhibiting alkali-aggregate reaction with


chemical admixtures, 8th International Conference on AAR in Concrete, Kyoto,
Japan, 1989, pp. 253-258.

[6] D. Stark, D., Lithium salt admixtures - an alternative method to prevent expansive
alkali-silica reactivity, 9l International Conference on AAR in Concrete, London,
UK, 1992, pp. 1017-1025.

[7] D. Stark, B. Morgan, P. Okamoto, S. Diamond, S., Eliminating or minimizing the


risk of alkali-silica reactivity, National Research Council, Strategic Highway
Research Program (SHRP), SHRP-C 343, 1993, 266 p.

[8] M.D.A. Thomas, D.B. Stokes, Use of a lithium-bearing admixture to suppress


expansion in concrete due to alkali-silica reaction, Transportation Research Record,
no. 1668, 1999, pp. 54-59.

[9] M.D.A. Thomas, R. Hooper, D.B. Stokes, Use of lithium-containing compounds to


control expansion in concrete due to ASR, 11 International Conference on AAR in
Concrete, Québec City, Canada, 2000, pp. 783-792.

[10] D.B. Stokes, CE. Manissero, Use of lithium-based admixture to control ASR
expansion, 14 Reunion Tecnica Asociacion Argentina de Tecnologia del
Hormigon, Universidad Nacional del Centro de la Provincia de Buenos Aires,
Argentina, 2001, 10 p.
85

[11] B. Fournier, D.B. Stokes, A. Ferro, Comparative field and laboratory investigations
on the use of supplementary cementing materials and lithium-based admixtures to
control expansion due to alkali-silica reaction in concrete, 6th International
CANMET/ACI Conference on Durability of Concrete, Thessaloniki, Greece, 2003,
pp.823-851.

[12] J.S. Lumley, ASR suppression by lithium compounds, Cement and Concrete
Research, 27 (2) (1997) 235-244.

[13] B. Durand, More results about the use of lithium salts and mineral admixtures to
inhibit ASR in concrete, 11th International Conference on AAR in Concrete, Québec
City, Canada, 2000, pp. 623-632.

[14] S. Diamond, S. Ong, The mechanisms of lithium effects on ASR, 9th International
Conference on AAR in Concrete, London, UK, 1992, pp. 269-278.

[15] S. Diamond, Unique response of LiN03 as an alkali silica reaction-preventive


admixture, Cement and Concrete Research, 29 (8) (1999) 1271-1275.

[16] D.B. Stokes, H.H. Wang, S. Diamond, S., A lithium-based admixture for ASR
tVi

control that does not increase the pore solution pH, 5 CANMET/ACI International
Conference on Superplastisizers and Other Chemical Admixtures in Concrete,
Roma, Italy, 1977, ACI SP 173, pp. 855-868.
[17] X. Mo, C. Yu, Z. Xu, Long-term effectiveness and mechanism of LiOH in
inhibiting alkali-silica reaction, Cement and Concrete Research, 33 (1) (2003)
115-119.

[18] L.D. Mitchell, J.J. Beaudoin, P. Grattan-Bellew, The effects of lithium hydroxyde
solution on alkali silica reaction gels created with opal, Cement and Concrete
Research, 34 (4) (2004) 641-649.
86

[19] B.Q. Blackwell, M.D.A. Thomas, A. Sutherland, A., Use of lithium to control
expansion due to alkali-silica reaction in concrete containing U.K. aggregates,
4 CANMET/ACI International Conference on Concrete Durability, Sydney,
Australia, 1997, ACI SP 170, pp. 649-663.

[20] K.J. Folliard, M.D.A. Thomas, K.E. Kurtis, Guidelines for use of lithium to mitigate
or prevent ASR, Publication No. FHWA-RD03-047, 2003, 61 p.

[21] M.A. Bérubé, J. Duchesne, Evaluation of testing methods for assessing the
effectiveness of mineral admixtures in suppressing expansion due to alkali-
aggregate reaction, 4th CANMET-ACI International Conference on Fly Ash, Silica
Fume, Slag and Natural Pozzolans in Concrete, Istanbul, Turkey, 1992,
ACI SP-132, pp. 549-575.

[22] M. De Grosbois, É. Fontaine, Performance ofthe 60°C-acccelerated concrete prism


test for the evaluation of potential alkali-reactivity of concrete aggregates,
11 * International Conference on AAR in Concrete, Québec City, Canada, 2000,
pp. 277-286.

[23] J. Duchesne, M.A. Bérubé, M.A., Effect of supplementary cementing materials on


the composition of cement hydration products, Advanced Cement Based Materials,
2 (2) (1994) 43-52.
87

Table 4.1 - Cemenl: paste mixtures, storage temperature, and testing


Total %Na2Oe a 0.51% 1.25% (with or without alkali addition)
Cement (%Na2Oe) 0.51% 0.51% 1.25%
Alkali addition none NaOH KOH none
Control pastes 0.51% 0.51%+NaOH 0.51%+KOH 1.25%
(23,38,60°C) (23,38,60°C) (38°C) (23,38,60°C)
LiNOj (LiN) 0.51%+LiN 0.51%+NaOH+LiN 0.51%+KOH+LiN 1.25%+LiN
pastes b (23,38,60°C) (23,38,60°C) (38°C) (23,38,60°C)
Li-glass (LiG) 0.51%+LiG 0.51%+NaOH+LiG 1.25%+LiG
pastes b (23,38,60°C) (23,38,60°C) (23,38,60°C)
Ground Li-glass 0.51%+LiGg 0.51%+NaOH+LiGg 1.25%+LiGg
(LiGg) pastes b (38°C) (38°C) (38°C)
Water/cement 0.42
Initial curing 24 hours in uncapped plastic bottles maintained over water in sealed containers at 23°C
Storage In capped plastic bottles maintained over water in sealed containers at 23, 38 or 60°C
Pore expression After 7, 28, and 91 days for pastes stored at 23 and 38°C;
and analysis after 3, 7, 28, and 91 days for pastes stored at 60°C
%Na 2 0 + 0.628 %K 2 0, by mass of cement;
Molar ratio Li/[Na+K] = 0.74.
88

Table 4.2 - Total design alkali contents measured, water contents, and proportions of pore
solution extracted
Alkalis Days Pore water content Water extracted
Paste (Mol/kg o f paste) (%) (% of total)
Na K Li Total Li/Na+K 23°C 38°C 60°C 23°C 38°C 60°C
0.51% 0.045 0.070 0 0.116 0 3 - - 18.0 - - 19
7 18.1 18.3 18.0 20 20 18
28 17.0 16.9 16.9 12 10 16
91 15.8 15.8 16.0 8 9 13
0.51% 0.045 0.070 0.086 0.201 0.74 3 - - 18.5 - - 17
+LiN a 7 19.1 18.5 18.3 21 18 18
28 18.0 17.6 17.2 12 10 12
91 16.8 16.7 17.1 8 8 15
0.51% 0.045 0.070 0.086 0.201 0.74 7 18.2 18.2 18.1 17 18 20
+LiGb 28 16.8 17.0 17.1 12 16 16
91 15.8 16.9 15.7 8 12 11
0.51% 0.045 0.070 0.086 0.201 0.74 7 - 19.0 - 18 - -
+LiGg c 28 - 17.9 - 12 - -
91 17.3 11
0.51% 0.214 0.070 0 0.284 0 3 - - 18.8 - - 20
+NaOH 7 18.4 18.4 18.3 17 16 20
28 17.9 17.8 17.4 12 11 14
91 16.7 17.0 16.4 7 11 12
0.51% 0.214 0.070 0.210 0.494 0.74 3 - - 21.2 - - 32
+NaOH 7 19.8 20.1 20.7 18 18 26
+LiN 28 19.2 19.4 19.7 14 16 20
91 18.2 17.5 19.5 8 6 19
0.51% 0.214 0.070 0.210 0.494 0.74 3 - - 19.0 - - 21
+NaOH 7 18.6 18.7 18.5 13 14 13
+LiG 28 18.4 17.9 17.7 12 15 20
91 16.8 17.1 16.7 7 8 11
0.51% 0.214 0.070 0.210 0.494 0.74 7 - 18.7 - - 16 -
+NaOH 28 - 17.7 - - 10 -
+LiGg 91 - 16.8 - - 9 -
0.51% 0.045 0.239 0 0.284 0 7 - 19.6 - - 14 -
+KOH 28 - 17.8 - - 10 -
91 - 17.7 - - 11 -
0.51% 0.045 0.239 0.210 0.494 0.74 7 - 19.8 - - 14 -
+KOH 28 - 18.6 - - 10 -
+LiN 91 - 17.9 - - 13 -
1.25% 0.100 0.184 0 0.284 0 3 - - (19) « - - -
7 18.0 18.1 18.0 17 17 17
28 16.8 17.1 16.1 11 8 12
91 15.7 16.2 16.2 10 15 13
1.25% 0.100 0.184 0.210 0.494 0.74 3 - - (20) d - - -
+LiN 7 18.7 18.8 19.2 17 18 19
28 17.9 17.2 16.9 12 12 15
91 16.7 16.5 16.9 9 12 15
1.25% 0.100 0.184 0.210 0.494 0.74 3 - - (18) d - - -
+LiG 7 17.5 17.5 17.2 16 14 15
28 17.1 16.9 16.1 10 11 10
91 15.6 •15.9 15.5 5 6 13
1.25% 0.100 0.184 0.210 0.494 0.74 7 - 18.5 - - 18 -
+LiGg 28 - 16.8 - 7 -
91 15.7 . 8
LiN03; Bulk Li-glass;
Li-glass ground at the cement fineness; Estimated value.
89

Table 4.3 - Chemistry of pore solutions (all concentrations and ratios are in Mol/L of pore
solution)
Paste Days 23°C 38°C i30°C
Na K Li Li/ pH Na K Li Li/ pH Na K Li Li/ pH
Na,K Na,K Na,K
0.51% 3 - - - - - - - - 0.121 0.238 - 13.45
7 0.128 0.285 - 13.67 0.136 0.293 - 13.60 0.131 0.248 - 13.51
28 0.170 0.345 - 13.75 0.154 0.319 - 13.75 0.135 0.255 - 13.51
91 0.160 0.329 - 13.76 0.142 0.302 - 13.45 0.135 0.261 - 13.45
0.51% 3 - - - - - - - - - 0.114 0.242 0.225 0.63 13.57
+LiN 7 0.113 0.264 0.168 0.45 13.67 0.134 0.298 0.160 0.37 13.67 0.131 0.263 0.216 0.55 13.51
28 0.156 0.324 0.217 0.45 0.144 0.318 0.178 0.39 - 0.134 0.271 0.165 0.41 13.54
91 0.137 0.300 0.198 0.45 13.58 0.148 0.334 0.196 0.41 13.76 0.128 0.269 0.206 0.52 13.58
0.51% 7 0.123 0.280 0.006 0.01 13.75 0.125 0.272 0.000 0.00 13.65 0.124 0.250 0.029 0.08 13.57
+LiG 28 0.135 0.302 0.006 0.01 - 0.130 0.278 0.010 0.02 - 0.128 0.256 0.060 0.16 -
91 0.157 0.336 0.050 0.10 13.76 0.181 0.369 0.077 0.14 - 0.127 0.265 0.128 0.33 13.45
0.51% 7 - - - - - 0.123 0.253 0.069 0.18 13.75 - - - - -
+LiGg 28 - - - - - 0.119 0.253 0.086 0.23 13.75 - - - - -
91 - - - - - 0.125 0.272 0.180 0.45 13.68 - - - - -
0.51% 3 - - - - - - - - - - 0.523 0.181 - - 13.74
+NaOH 7 0.485 0.187 - - 13.92 0.416 0.156 - - 13.87 0.494 0.179 - - 13.81
28 0.611 0.238 - - 13.97 0.595 0.217 - - 13.92 0.588 0.199 - - 13.76
91 0.603 0.247 - - 13.93 0.589 0.224 - - 13.82 0.566 0.198 - - 13.76
0.51% 3 - - - - - - - - - - 0.655 0.232 0.429 0.48 13.78
+NaOH 7 0.631 0.251 0.432 0.49 13.92 0.686 0.266 0.451 0.47 13.92 0.630 0.228 0.392 0.46 -
28 0.693 0.278 0.399 0.41 13.87 0.743 0.286 0.433 0.42 13.80 0.716 0.259 0.397 0.41 13.69
+LiN 91 0.701 0.313 0.421 0.42 13.88 0.742 0.306 0.438 0.42 13.82 0.676 0.256 0.401 0.43 13.58
0.51% 3 - - - - - - - - - - 0.515 0.179 0.075 0.11 13.87
+NaOH 7 0.606 0.230 0.035 0.04 14.01 0.576 0.207 0.027 0.03 13.93 0.457 0.162 0.054 0.09 13.87
28 0.621 0.231 0.067 0.08 13.83 0.599 0.216 0.063 0.08 14.01 0.646 0.218 0.117 0.14 13.75
+LiG 91 0.577 0.234 0.110 0.14 13.98 0.591 0.225 0.131 0.16 13.93 0.590 0.209 0.234 0.29 13.82
0.51% 7 - - - - 0.573 0.215 0.165 0.21 13.97 - - - -
+NaOH 28 - - - - 0.605 0.223 0.232 0.28 - - - - -
91 - - - - - 0.575 0.226 0.318 0.40 13.98 - - - -
+LiGg
0.51% 7 - - - - 0.088 0.780 13.93 - - - -
+KOH 28 - - - - 0.092 0.816 13.88 - - - -
91 - - - - 0.100 0.790 13.93 - - - -
0.51% 7 - - - - 0.102 0.956 0.401 0.38 13.82 - - - -
+KOH 28 - - - - 0.113 1.054 0.434 0.37 13.82 - - - -
91 - - - - - 0.122 1.066 0.424 0.36 13.76 - - - -
+LiN
1.25% 3 0.227 0.645
7 0.256 0.806 - - 14.05 0.240 0.711 - - 14.05 0.244 0.672 - - 13.92
28 0.289 0.857 - - 14.11 0.271 0.742 - - 14.01 0.266 0.703 - - 13.78
91 0.262 0.758 - - 13.98 0.280 0.777 - - 13.98 0.265 0.716 - - 13.76
1.25% 3 - - - - - - - - - - 0.301 0.870 0.430 0.37 -
+LiN 7 0.287 0.941 0.476 0.39 14.05 0.303 0.949 0.464 0.37 13.81 0.316 0.891 0.429 0.36 13.81
28 0.300 0.935 0.409 0.33 14.01 0.319 0.929 0.439 0.35 13.92 0.342 0.918 0.393 0.31 13.71
91 0.321 1.023 0.408 0.30 13.93 0.310 0.917 0.403 0.33 13.93 0.326 0.995 0.357 0.27 13.68
1.25% 3 0.242 0.693 0.068 0.07
7 0245 0758
+LiG °053 °05 " 0276
°- 804 0090
°08 1401 0202 0583 0065 008 1387

28 0.268 0.787 0.063 0.06 14.08 0.287 0.778 0.076 0.07 14.05 0.259 0.700 0.107 0.11 13.83
91 0.281 0.808 0.097 0.09 14.06 0.276 0.753 0.136 0.13 13.98 0.257 0.690 0.195 0.21 13.82
1.25% 7 . . . 0.205 0.605 0.195 0.24 14.05 - - - -
+LiGg 28 0.292 0.792 0.254 0.23 14.08
91 0.302 0.832 0.300 0.26 14.06
90

At 23°C
! I I I 4 I I 1 I I I I
0.8­ 1.1

0S%»N»nH + LINQ3 ,
0 7­

•""' . 0.5% + NaOH


d 0.6
o
S
is 0.5 ­
Z
•s
c 0.4­
_o
'■C 1.25% ­tUNOi! 14
n 0.3 ­ — 1 ~~
■g
0) ­
•—~~~ T_5% '
u
c
O 0.2 • 0.5%
o J_^—" 0.5% ♦ LiN03
0.1

A
0.0
0 7 14 21 28 35 42 49 56 63 70 77 84 91 0 7 14 21 28 35 42 49 56 63 70 77 84 91
A g e (days) A g e (days)
I I I
1.4 ­

1.3 _^15%___kfflfi3
1.2 ­

■5 1.1 ­
f 1.0­ 0 5%* NaO[l±L_N03 ~~ a
L

* 0.9 *~"~^ , 0.5%* NaOH ,'


_i 0.3
■5 0.8

"5 0.7 ­ c
o
•2 0.6 ­
a
Ë _____±UN03
t <l*%
b 0.5 ­ c
*­<^^~ 0.5%fLiNO3 "
S 0.4 S
c
° 0.3 o
(J 0.1
0.2

0.1 ­
C
0.0 ­ 0.0
0 7 14 21 28 35 42 49 56 63 70 77 84 91 0 7 14 21 28 35 42 49 56 63 70 77 84 91
A g e (days) A g e (days)
_ i i i i i i i _

21 28 35 42 49 56 63 70 77 84 91 0 7 14 21 28 35 42 49 56 63 70 77 84 91
A g e (days) A g e (days)

Figure 4.1 ­ Pore solution chemistry of cement pastes stored at 23°C: A) [Na+]. B) [K+].
C) [Na++K+]. D) [Li""]. E) [Na++K++Li+]. F) [Li+]/[Na++K+].
91

At 38°C
I I I I I I I I I I I
1.11 i
1
^ ^ * 0.5% ♦ KOH + LIN03
1.0
*:~ . 1.25% + LiN03
0.9 ­
, 0.5% . KOH
^0.8­
•._ —"* 1.25%
^0.7­

"5 0.6

•2 0.5
IB
| 0.4­ 0.5%
, 0.5%4»LINO3 /
u
O 0.3
*
O • 0.5% ♦ NaOH + L1N03
0.2 ^^___­­^* 0.5% ♦ NaOH

0.1 ­
B
~i 1 r~ 0.0­ i 1 1 1 1 1 1 1 1 1 1 1­
0 7 14 21 28 35 42 49 56 63 70 77 84 91 0 7 14 21 28 35 42 49 56 63 70 77 84 91
age (days)
Age (days)
1.4
0.5
115%____lNfi3­
0.5% * NaOH + LINQ3
1.2 _ 5% * KOH + LIN03
0.5% ♦ NaOH ♦ LiN03

1" lL5%_±J_QfcL
+
ID
0.8

| 0.6
S 0.5%»LiNO3 .;
§ 0.4 0.5%
c
o
o
0.2­

0.0 i \ \ \ l i ] ^ r

0 7 14 21 28 35 42 49 56 63 70 77 84 91 0 7 14 21 28 35 42 49 56 63 70 77 84 91
Age (days) Age (days)
1.8 0.6

1.6 0.5% . KOH ♦ UN03


0.5
0.5% ♦ NaOH + LINÛ3
O 1.4

ÎU _A2¥bjL!Miss_

­_ IS

is LO g 0.3
0.8 o
o
s a
2 0.6 <* 0.2
c
8
o 0.1
0.2

0.0 0.0
0 7 14 21 28 35 42 49 56 63 70 77 84 91 0 7 14 21 28 35 42 49 56 63 70 77 84 91
Age (days) A g e (days)

Figure 4.2 ­ Pore solution chemistry of cement pastes stored at 38°C: A) [Na+]. B) [K+].
C) [Na++K+]. D) [Li+]. E) [Na++K++Li+]. F) [Li+]/[Na++K+].
92

I I I I I I
At 60°C
i i i
0.8 H 1.1

1.0
0.7­ ­—SJ^±HMQtijtUfm__^
­—,: V25% •UN03_
0.9
?0.6 «5% * NaOH
. J

HI I) H
E. o
is 0.5 ­
F 07
_J,25%
­£
z
"S O 0.6
c
o
0.4 ­ c
o
■a 1_B% +1 i n n i IS
( ) ■ >

II
•g 0.3 ­ c 0.4
1.25% o
u o
c
c o 0.3
O 0.2 ­
0.5%
­ o
0 5% ♦ UN03
o 0.2
0.5% ♦ NaOH
0.1 ­ •=­* 0.5% ♦ UN03 0.1
0.5% + NaOH ♦ LIN03
A B
0.0 0.0 ~i 1 1 1 i r­
0 7 14 21 28 35 42 49 56 63 70 77 84 91 7 14 21 28 35 42 49 56 63 70 77 84 91
Age (days) Age (days)
1.4 _l I 1 I L_

1.2

1__ " "S^^NiÔHTÛNM


J.25%
".
+
is
0.8 0.5% * NaOH

I 0.6
|
8 0.4
c
o
o
0.2

0.0 n 1 1 ] ; ^ 1 1 1 r~
7 14 21 28 35 42 49 56 63 70 77 84 91 0 7 14 21 28 35 42 49 56 63 70 77 84 91
Age (days) Age (days)

T 1 i 1 1 1 1 r~
7 14 21 28 35 42 49 56 63 70 77 84 91 0 7 14 21 28 35 42 49 56 63 70 77 84 91
Age (days) Age (Days)

Figure 4.3 ­ Pore solution chemistry of cement pastes stored at 60°C: A) [Na+]. B) [K+].
C) [Na++K+]. D) [Li+]. E) [Na++K++Li+]. F) [Li+]/[Na++K+].
93

At 23°C
100 100 ­J 1 1 L_

1^53__LUN____
90 90

o 80 | 80
3 D
O O
in
70 ■ 70
o
a. »
o
S^A_­_îfiH_a___!3_,r ça 60
__
50 o
50

40 40

B
30 30 ­i 1 1 ! 1 1 1 1 1 1 1 r
7 14 21 28 35 42 49 56 63 70 77 84 91 0 7 14 21 28 35 42 49 56 63 70 77 84 91
Age (days) Age (days)

50

40
c
o
_3 UNOT
S 30
2
o
a
c
~ 20

10-

7 14 21 28 35 42 49 56 63 70 77 84 91 0 7 14 21 28 35 42 49 56 63 70 77 84 91
Age 9days) Age (days)

Figure 4.4 ­ Proportion of alkalies in the pore solution of cement pastes stored at 23°C:
A) [Na+]. B) [K+]. C) [Na++K+]. D) [Li"*].
94

100

90

.1 80
3
O
IA
70

60

50

40

30
0 7 14 21 28 35 42 49 56 63 70 77 84 91 0 7 14 21 28 35 42 49 56 63 70 77 84 91
Age (days) Age (days)

50

40
c
o

8 30
S
o
a
20
■6

10

n 1 1 ' 1 1 — T-
7 14 21 28 35 42 49 56 63 70 77 84 91 0 7 14 21 28 35 42 49 56 63 70 77 84 91
Age 9days) Age (days)

Figure 4.5 - Proportion of alkalies in the pore solution of cement pastes stored at 38°C:
A) [Na+]. B) [K+]. C) [Na++K+]. D) [Li+].
95

AT 60°C
100

90

c
o 80

70

^-____NaOH^UN03

£ 60
_J_25%+JJN03^

50 ___5%_HjN03

40

30
0 7 14 21 28 35 42 49 X 63 70 77 84 91 7 14 21 28 35 42 49 56 63 70 77 84 91
Age (days) Age (days)

0 7 14 21 28 35 42 49 56 63 70 77 84 91 0 7 14 21 28 35 42 49 56 63 70 77 84 91
Age (days) Age (days)

Figure 4.6 - Proportion of alkalies in the pore solution of cement pastes stored at 60°C:
A) [Na+]. B) [K+]. C) [Na++K+]. D) [Li+].
96

23°C 1.0
38°Ç
1.0

0.9 ___________a____u_io__ 0.9

0.8 0.8
1.25% f I iNfl­l

S 0.7 0.7

■O 0.6 ■o 0.6
.e
0.5 .E 0.5 n ■;»/. ♦ KOH
S
O
N 0.4
§ 0.4
0.5% ♦ LINO„
0.3 0.3

0.2 0.2
_»_____5%­
0.1 0.1
B
0.0 — . 1 1 1 1 j ■ | 4 1 1 , , 1 4 [ ­ ­ , , , , 1 _ 0.0
O 7 14 21 28 35 42 49 56 63 70 77 84 91 10 20 30 40 50 60 70 80 90 100
Age (days) Age (days)

60°C
1.0

0.9

0.8
1.25% ♦ LiN03

0.7

0.6

0.5% * UNQ3

0.3

0.2

0.1

0.0 ­i 1 1 1 1 1 1 \ \ 1 1 r
7 14 21 28 35 42 49 56 63 70 77 84 91
Age (days)

Figure 4.7 ­ %Na20e (including Li) in the cement hydrates present in the control and
LiN0 3 cement pastes: A) 23°C. B) 38°C. C) 60°C.
97

5 Utilisation de l'essai sur prismes de béton pour évaluer l'efficacité


des adjuvants à base de lithium

5.1 Introduction

Cette publication a été soumise à la revue ACI Materials Journal le 17 avril 2006 et
acceptée le 4 août 2006. Elle a été publiée dans le numéro des mois de mars et avril 2007 de
cette revue (volume 104, numéro 2).

5.2 Résumé

Pour évaluer l'efficacité des adjuvants à base de lithium pour contrer la réaction alcalis-
silice (ASR), un total de 87 mélanges de béton ont été fabriqués avec 12 granulats réactifs
de différentes compositions et degrés de réactivité, utilisant divers dosages de nitrate de
lithium (LiNC^) ou d'un verre à base de lithium (Li-glass) en combinaison ou non avec des
ajouts cimentaires (SCM). Les prismes de béton ont été testés à 38°C (100°F) selon la
norme CSA A23.2-14A ou ASTM C 1293, mais aussi à 60°C (140°F), ceci pour évaluer la
possibilité d'accélérer l'essai. L'utilisation du LiNCh au dosage recommandé par le
manufacturier ([Li]/[Na+K] = 0.74) permet de satisfaire à la limite d'expansion de 0.040%
après 2 ans à 38°C (CSA A23.2-28A) pour 6 des 12 granulats testés; trois granulats
requérant un dosage compris entre 0.74 et 1.11, alors qu'un dosage de 1.11 n'est pas
suffisant pour les trois autres granulats. Le verre de lithium s'avère inefficace. Le ciment
ternaire testé, à base de laitier et de fumées de silice, s'est avéré efficace, tout comme les
cendres volantes et le laitier testés lorsque leur composition et leur dosage sont adéquats.
La plupart des combinaisons de LiN03-SCM ne montrent pas d'effet synergique. Le dosage
requis en LiNC.3 n'est pas fonction du degré de réactivité du granulat. L'expansion à 60°C
après 6 mois présente une assez bonne corrélation avec l'expansion à 38°C après 2 ans pour
les mélanges témoins et les mélanges incorporant des ajouts cimentaires, mais pas pour les
mélanges avec du LiNC»3.
98

5.3 Publication scientifique no. 2

Effectiveness of lithium-based products in concrete made with Canadian


natural aggregates susceptible to alkali-silica reactivity

C. Tremblay *, M.A. Bérubé**, B. Fournier***, M.D.A. Thomas**** and K.J.


Folliard*****

PhD Candidate and Associate Professor, Département de géologie et de génie géologique, Université Laval, Québec
City, QCG1K7P4, Canada
*" Research Scientist, CANMET-MTL, Ottawa, ON, Kl A 0G1, Canada
****Professor, Department of Civil Engineering, University of New Brunswick, Fredericton, NB, E3B 5A3, Canada,
*****Professor, Department of Civil Engineering, University of Texas at Austin, Austin, TX 78712

ABSTRACT

To evaluate the effectiveness of lithium-based products to counteract ASR, a total of 87


concrete mixtures were made incorporating 12 reactive aggregates of various types and
degrees of ASR, and using various dosages of LiNCh and Li-glass in combination or not
with supplementary cementing materials (SCM). The concrete prisms were tested at 38°C
(100°F), according to CSA A23.2-14A or ASTM C 1293, and also at 60°C (140°F) to
evaluate the possibility of accelerating the testing procedure. Using LiN(_>3 at the
[Li]/[Na+K] ratio of 0.74 recommended by the manufacturer, satisfied the 2-year,
0.04% expansion limit criterion (CSA A23.2-28A) at 38°C (100°F) for 6 aggregates; three
aggregates required a ratio between 0.74 and 1.11, while a ratio 1.11 was not effective with
the other three aggregates. The lithium glass was not effective. The ternary silica fume/slag
cement tested was effective and the fly ashes and slag as well, provided they were in
sufficient quantities and they had a proper composition. Most LiNC»3-SCM combinations
did not show significant synergetic effect. The required LiNC>3 dosage is not related to the
aggregate reactivity level. The 6-month expansion at 60°C (140°F) correlated well with the
2-year expansion at 38°C (100°F) for the control and SCM mixtures, but not for the LiNÛ3
mixtures.

Keywords: Aggregates, Alkali-silica reaction, Concrete, Lithium-based products,


Supplementary cementing materials.
99

1 Introduction

Alkali-silica reaction (ASR) affects the durability and serviceability of concrete structures
in many countries around the world. This chemical reaction between some siliceous
aggregates and the highly basic concrete pore solution produces a swelling gel that causes
the expansion and premature deterioration of the affected concrete elements. Alkali-silica
reactive aggregates can be used in concrete provided adequate preventive measure(s) is(are)
used. Commonly used preventive methods consist of limiting the concrete alkali content
and/or using sufficient amounts of effective supplementary cementing materials (SCM),
1 9

such as fly ash, silica fume, and/or slag ' , and/or chemical admixtures. In 1951, McCoy
and Caldwell were the first to observe that certain lithium-based products were capable of
significantly reducing the expansion due to ASR. Since then, other researchers confirmed
that a number of lithium-based products (LiOH-HiO, LiCl, LiF, Li2C03, and LiNC^) can
suppress the ASR expansion when used at a sufficient dosage4' 5' 6' 7 ' 8 . The efficacy of
LiN03 was recognized as it appeared not to be susceptible to pessimum effects, while not
increasing the pH of the pore solution in contrast to LiOH and some other lithium salts . A
recent study at CANMET (Canada) also suggested that LiNC>3 is generally more effective
against ASR than LiOH.FbO at similar [Li]/[Na+K] molar ratios10. Another study showed
that soda-lime glasses containing lithium can be manufactured which appeared as effective
in suppressing ASR as lithium nitrate11.
The alkali content of concrete is a critical parameter for the ASR and also impacts
significantly on the efficacy of lithium-based products. Most previous studies indicated that
a dosage of lithium-based products resulting in a [Li]/[Na+K] molar ratio between 0.6 and
1.0 was sufficient to control the expansion generated by the majority of reactive
aggregates2' 12' 13 ' 14 ' 15 ' 16 . However, the effective dosage may vary significantly from one
reactive aggregate to another, namely as a function of the degree of reactivity and
mineralogy. For instance, lithium nitrate was found to be more effective with some
extremely reactive aggregates, but relatively less effective at similar dosages with some
other less reactive ones 17 ' 18 . The possibility of combining lithium admixture and SCM has
also been investigated and proved to have a beneficial effect at further reducing ASR
expansion17.
100

It is the authors' opinion that the Concrete Prism Test CSA A23.2-14A19 or
ASTM C 1293 20 (CPT), performed at 38°C (100°F) and >95% RH, is currently the only
laboratory test method for reliably determining the effectiveness of lithium-based products
against ASR, at least in moderate to high-alkali concrete systems. In Canada and the USA,
the test duration is one year when evaluating the degree of reactivity of concrete
aggregates; however, CSA A23.2-28A21 states that the testing period shall be extended to
two years when evaluating preventive measures against ASR. Storing the test prisms at
60°C (140°F) rather than 38°C (100°F) has the potential to shorten the testing period;
however, the reliability of this accelerated version of the CPT is yet to be determined for
the lithium-based products. The original work of De Grosbois and Fontaine and other
researchers showed a good correlation between concrete prism test expansions after one
year at 38°C (100°F) and after three months at 60°C (140°F), when the test is used to
evaluate the potential alkali-reactivity of aggregates. It would be interesting to determine
what testing period at 60°C (140°F) would best correlate with the 2-year expansion at 38°C
(100°F) when evaluating the potential of SCM and lithium-based admixtures to control
ASR expansion.

2 Research significance and objectives

Over the past few decades, many studies have shown the effectiveness of lithium against
S 17

ASR " . However, the reaction mechanisms involved are still not well-understood.
Moreover, the effectiveness of lithium nitrate must be confirmed in the presence of a wider
variety of reactive aggregates and supplementary cementing materials, while a new lithium
product (Li glass) is now available on the market. Also, there is a need for an accelerated
test for assessing the effectiveness of lithium products against ASR. Accordingly, the main
objectives of this study are : (1) to evaluate the effectiveness of two lithium-based products
(nitrate and glass) against a wide variety of (Canadian) reactive aggregates; (2), to better
understand the reaction mechanisms involved; (3) to evaluate the effectiveness of a number
of supplementary cementing materials in combination or not with lithium-based products,
and (4) to verify the possibility of accelerating the concrete prism test procedure used for
determining this effectiveness, namely by performing this test at 60°C (140°F).
101

3 Experimental procedure

3.1 Aggregates, cements, SCM's, and lithium-based products

Very few studies were performed so far to evaluate the effectiveness of lithium-based
products in the presence of Canadian reactive aggregates10' 17. Twelve coarse aggregates
from various parts of the country and representing a range of composition and reactivity
levels were selected for this study (Table 5.1). These aggregates were used in concrete
mixtures incorporating a non-reactive granitic sand from the Québec City area and a CSA
Type GU (ASTM Type I) high-alkali cement (Table 5.2), in combination or not with two
different types of Li-based products (Table 5.3); the latter consist of a solution containing
30% LiN03 by mass, and a crushed glass essentially composed of Si02 (50.8%),
CaO (26.1%), AI2O3 (1.5%), and Li02 (19.6% by mass). The lithium glass is homogeneous;
it was used as supplied (14% < 10 pm (0.00394 in)) and after being finely ground (53% <
10 pm (0.00394 in)).

In addition to the above, concrete mixtures were also made with a ternary (silica fume /
slag) blended cement (Table 5.2) and two reactive aggregates from Eastern Canada (Table
5.3). Three fly ashes from Alberta, Nova Scotia, and Ontario, and one blast-furnace slag
from Ontario (Table 5.2) were also used in concrete mixtures incorporating local reactive
aggregates (Table 5.3).

3.2 Petrography of the coarse aggregates

The pétrographie examination of all aggregates was performed using a modified version of
24
the ASTM C 295 (Pétrographie Examination of Aggregates for Concrete). This test
evaluates the 'pétrographie potential' of alkali-reactivity through the examination of thin
sections of the aggregates under the optical microscope. Each aggregate was
macroscopically separated into different rock types or pétrographie facies and one thin
section of each facies was prepared and examined under the pétrographie microscope.
102

3.3 Concrete prism tests

Concrete prisms were made with the selected coarse aggregates, the non-reactive sand, the
CSA Type GU (ASTM Type I) cement or the silica fume/slag blended cement, and various
amounts of the two lithium-based products and/or the SCM's (Table 5.3). In accordance
with the CSA A23.2-14A19 or ASTM C 1293 20 Concrete Prism Test (CPT) procedure, the
total amount of cementitious materials used was 420 kg/m (26.25 lbs/ft ), while the alkali
content in the concrete mixture was increased to 1.25% Na20e by mass of portland cement
in the concrete mixture, by adding NaOH to the mixture water. The water-to-cementing
materials ratio varied from 0.38 to 0.42 (slump from 50 to 80 mm (1.97 to 3.15 in), which
falls within the range specified by CSA A23.2-28A21 when the CPT is used for testing
preventive measures against ASR.

The test prisms were stored above water in sealed plastic containers, a first set of three
prisms in a room at 38°C (100°F), and a second set of three prisms in testing cabinets at
60°C (140°F). Their expansion was measured up to two years at 38°C (100°F) and up to
9 months at 60°C (140°F). A total of 87 mixtures were made and test prisms tested both at
38°C (100°F) and 60°C (140°F) (Table 5.3). The "lithium" mixtures contained 0, 50, 75,
100, 125 or 150% ofthe standard dosage recommended by the manufacturer; this dosage
corresponds to a [Li]/[Na+K] molar ratio of 0.74 with respect to the total alkalies (Na+K)
contributed by the portland cement and the NaOH addition to the mixture water.

3.4 Compressive strength

The compressive strengh of each combination tested was measured following


ASTM C 39 25, on three cylinders, 100 mm (3.94 in) in diameter by 200 mm (7.87 in) in
length, after 28 days of curing in a fog room at 23°C (73°F).
103

3.5 Chemistry of concrete pore solution

To determine the distribution of ionic species in the pore solution of the hydraulic systems
investigated, the concrete pore solution was extracted at different times or expansion levels
under high pressure (1000 MPa (145 000 psi)). The equipment used in this study had a
cylindrical chamber of 50 mm (1.97 in) in diameter by 90 mm (3.54 in) in length and the
pore solutions were expressed from broken pieces of concrete. After expression, the pore
solutions were chemically analyzed using ICP (Initiation Chromatographic Plasma) for Na,
K, and Li. The OH* concentration was determined by direct titration with standardized
hydrochloric acid (HC1) down to the phenolphthalein end point, and their pH calculated
from these concentrations (= 14 + log[OH"]).

4 Experimental results and discussion

4.1 Petrography ofthe coarse aggregates

Table 5.4 shows the pétrographie composition of each coarse aggregate with the
potentially-reactive mineral phases. The degree of reactivity is given in accordance with
CSA A23.2-27A , i.e. based on the expansion level at 1 year in the concrete prism test at
38°C (100°F): expansion <0.040% : non reactive; 0.040 < expansion < 0.120% : moderately
reactive; expansion > 0.120% : highly reactive. As can be seen in Table 5.4, the aggregates
tested vary widely in composition across the country, including rock types of all origins, i.e.
sedimentary (e.g. siliceous limestone, greywacke, sandstone, dolostone, siltstone), igneous
(rhyolite, granite, basalt), and metamorphic (schists, quartzite). In most cases, the reactive
material is identified to be micro- to cryptocrystalline quartz.

4.2 Compressive strength

The results obtained for the compressive strength at 28 days showed no significant
differences between mixtures with and without lithium admixture. With respect to the
control mixtures, on average, the addition of lithium resulted in a decrease of 0.1 % in
strength, while the ternary cement induced an increase of about 10% in strength. The main
104

differences in strength are related to the type of aggregate (e.g. limestone vs. granitic), the
shape ofthe aggregate particles (i.e. quarried vs. gravel), and the use ofthe ternary cement
(increased compressive strength).

4.3 Concrete prism expansion at 38°C (100°F)

The limit recommended by CSA A23.2-28A21 when the CPT at 38°C (100°F) is used for
evaluating the efficacy of preventive measures against ASR is 0.040% after 2 years (rather
than 0.040% after one year when testing aggregates for potential ASR) 2.

4.3.1 Effectiveness of lithium nitrate

The lithium nitrate solution was used at various dosages in the presence of all reactive
aggregates. Table 5.3 gives the expansion (%) of the test prisms after 2 years at
38°C (100°F). The use of lithium nitrate invariably decreased the expansion for every
aggregate investigated, however at different rates (Figure 5.1). The minimum [Li]/[Na+K]
ratio required to reduce the expansion to under 0.040% after 2 years at 38°C (100°F) is
established by interpolation (unless the highest ratio used was not satisfying the above
criterion at 38°C (100°F), as shown in Figure 5.1). This minimum effective molar ratio was
found to vary from 0.56 to more than 1.11 at 38°C (100°F) (Table 5.5). These variations are
slightly larger than in other studies (0.6 to l)5"17. However, artificial aggregates or highly-
reactive natural aggregates were used in several of the previous studies while a larger
variety of natural aggregates was tested in the present study. It is clear from this study that
the effectiveness of lithium nitrate greatly varies from one reactive aggregate to another
(Table 5.5). Furthermore, it was found that this effectiveness does not necessarily correlate
with the degree of reactivity (i.e. expansion level after one year at 38°C (100°F) for the
control concrete) (see Table 5.5) nor with the type of reactive aggregate to counteract.
Unexpected data were obtained with the aggregate NS1 where higher expansion was
obtained for test prisms cast from the 150% LiN mixture (0.082% at two years - Table 5.3)
compared to the 125% LiN mixture (0.046% at two years - Table 5.3). There is no
105

explanation for this behavior at this stage and the mixtures are being repeated. However,
97

Feng and al. had found that a 200% LiN dosage was still ineffective in controlling
concrete prism expansion (i.e. < 0.040% at two years) with the NS1 aggregate.

4.3.2 Effectiveness of Li-glass

In the absence of SCM, the use of 75% or 100% ofthe standard dosage of Li-glass did not
result in significant reduction of concrete prism expansions; actually, test prisms
incorporating Li glass expanded similarly and sometimes even more than those cast from
the corresponding control mixtures (Figure 5.2; Tables 5.3 and 5.6). Even once ground
finer (53% < 10 pm (0.00394 in)), the lithium glass did not appear to be significantly more
effective in reducing concrete prism expansion.

4.3.3 Effectiveness ofthe SCM's with and without lithium

Figure 5.3A shows that 20% high-alkali fly ash FA3 from Ontario used with the ONI
limestone is not effective to reduce the expansion under 0.040% after 2 years at 38°C
(100°F) (expansion of 0.123%). The addition of lithium nitrate at a 0.56 [Li]/[Na+K] molar
ratio to the 20% FA3 concrete did not contribute to further reduce expansion (expansion of
0.119%). On the other hand, the use of 20% fly ash FA2 with the NS1 greywacke (Figure
5.3C), and 20% fly ash FA1 with the All and A12 gravels (Table 5.3), was effective by
itself in controlling expansion under the 0.040% expansion limit after 2 years at
38°C (100°F). The addition of 75% ofthe standard dose of lithium nitrate to the above 20%
fly ash concretes did not contribute to further decrease the expansion.

With the highly-reactive ONI limestone, in the absence of lithium, the higher the slag
content, the lower the expansion, with 35% slag reducing the expansion to just about the
0.040% expansion level after 2 years at 38°C (100°F) (Figure 5.3A). The addition of
lithium nitrate at a [Li]/[Na+K| molar ratio of 0.74 to 25% slag concrete was effective in
reducing the expansion below the 0.040% expansion level at 2 years; this is particularly
interesting considering that neither the 25% slag nor a 0.74 molar ratio of LiNÛ3 were
106

effective by themselves in controlling expansion. With both the ONI limestone (Figure
5.3A) and ON2 gravel (Figure 5.3B), smaller dosages (0.56 and/or 0.37 ratios) of lithium
nitrate decreased only slightly the expansion in cementitious systems incorporating 25%
(ONI and ON2) or 35% (ONI) slag. The use of lithium glass (LiG) in combination with
slag or fly ash generally resulted in slight further reduction of expansion compared to the
corresponding systems with slag and fly ash only. However, with the exception ofthe 20%
FA + 75% LiG and 25% Slag + 50% LiG concrete systems, the beneficial effect of lithium
nitrate (LiN) in ternary systems was similar or better than that of LiG when used at the
same dosage. The differences are however very small.

The ternary slag/silica fume cement was found to be very effective by itself with the four
aggregates tested (Figures 5.3A to 5.3D). The addition of lithium nitrate to the above
system at a [Li]/[Na+K] ratio of 0.37 (i.e. 50% ofthe standard dose) did not significantly
affect the results, the important expansion reduction observed (compared with the control
mixture made with an ordinary portland cement without lithium) being essentially
attributed to the presence of the very effective ternary cement. The addition of lithium to
the ternary cement even slightly increased the expansion for three of the four aggregates
tested. We cannot explain this behavior for the time being.

In conclusion, supplementary cementing materials can be effective in controlling ASR


expansion provided they are used in sufficient amounts. The objective of using
combinations of lithium nitrate and SCM is to take advantage of a potential synergetic
effect between the two products and to provide further protection in the long term through
the combined effect of SCM and lithium. In the case ofthe mixtures made in this study, the
addition of lithium nitrate in cementitious systems already incorporating enough SCM to
control ASR expansion by itself (e.g. 35% slag with ONI, ternary cement with all
aggregates, 20% FA2) did not contribute to significantly further reduce the expansion; it is
thus unclear whether or not the addition of lithium nitrate would potentially provide further
protection in the long-term to the above systems. On the other hand, the use of lithium
nitrate in more marginal systems, e.g. in combination with 25% slag with aggregates ONI
and ON2, did show some additional benefit by further reducing the concrete prism
expansion.
107

4.4 Comparison between the results at 38°C (100°F) and 60°C (140°F)

The limit suggested when the CPT at 60°C (140°F) is used for testing the potential alkali-
reactivity of aggregates is 0.040% after 3 months22'23. No limit is currently proposed when
the accelerated CPT is used for testing preventive measures against ASR. Considering that
the time limit changes from one to two years when evaluating the effectiveness of
preventive measures using the standard CPT (at 38°C (100°F)), it was first considered to
double the testing period, i.e. from 3 to 6 months, when the CPT at 60°C (140°F) is used
for testing preventive measures. Tables 5.6 and 5.7 gives the expansion (%) of the test
prisms after 6 and 9 months at 60°C (140°F).

Table 5.5 shows that the minimum effective molar ratio in concrete varies from 0.30 to
>1.11 at 60°C (140°F) (i.e. when using a 0.040% expansion limit at 6 months in the CPT).
This ratio seems slightly lower at 60°C (140°F) than at 38°C (100°F) (between 0.56 to
>1.11). Figure 5.4 compares the expansions after 1 year at 38°C (100°F) vs. those after
3 months at 60°C (140°F) (A), and the expansions after 2 years at 38°C (100°F) vs. those
after 6 months at 60°C (140°F) (B). For the control concretes (no lithium or SCM) and the
concretes containing the Li-glass, the correlations are fairly good between the two
temperatures in terms of both tests resulted in the acceptance or the rejection of the
mixtures under test with respect to the expansion limits used. The correlations are not as
good for the mixtures containing lithium nitrate with or without SCM. Numerous mixtures
do not pass the test at 38°C (100°F) (i.e. expansions > 0.040% after 2 years) but pass the
test at 60°C (140°F) (i.e. expansions < 0.040% after 6 months). The lithium nitrate could be
more effective at higher temperature or, following the more rapid alkali leaching from the
test prisms at 60°C (140°F) (B. Fournier, comm.. pers. 2010), the aggressive alkalies
(sodium and potassium) could be leached relatively more readily than lithium. These
explanations are hypothetical for the time being.

Figure 5.4C compares the 2-year expansions at 38°C (100°F) with the 9-month expansions
at 60°C (140°F). When using a limit of 0.040% in both tests, the correlation seems better;
however, some mixtures do not pass the test at 38°C (100°F) but do so at 60°C (140°F). A
better correlation is obtained, however, when using a limit of 0.030% at 60°C (140°F).
108

Table 5.8 gives the % of correlation between the two tests and the % of safe prediction
when using the 60°C (140°F) test. The % of correlation corresponds to the % of all
mixtures that pass the CPT at 60°C (140°F) and 38°C (100°F) or that fail at both
temperatures. The % of safe prediction includes all ofthe above plus all other mixtures that
pass at 38°C (100°F) but fail at 60°C (140°F). In actual fact, the most important thing is
that all mixtures that fail in the CPT at 38°C (100°F) also fail in the 60°C (140°F) test if
this test is used to predict the effectiveness of lithium. Table 5.8 shows that for the mixtures
with LiN03, the 2-year expansions at 38°C (100°F) correlate in 75% ofthe cases with the
6-month expansions at 60°C (140°F), when using a limit of 0.040% in both tests, but the %
of safe prediction with the 60°C (140°F) test is only 75%; actually 7 among the 28 LiN0 3
mixtures tested (up to 6 months at 60°C (140°F)) pass the test at 60°C (140°F) but fail at
38°C (100°F), using a limit of 0.040% in both tests. The 9-month expansions at
60°C (140°F) correlate at 82% with the 2-year expansions at 38°C (100°F), still using a
limit of 0.040% in both tests, and the % of safe prediction is also 82%. When the expansion
limit is lowered to 0.030% in the 60°C (140°F) test, the correlation remains relatively good
(73%) but the % of safe prediction is significantly increased (95%), with only one mixture
amongst the 22 LiN0 3 mixtures tested (up to 9 months at 60°C) passing the 60°C (140°F)
test while failing at 38°C (100°F). (In actual fact, 6 of the 28 LiN0 3 mixtures were
unfortunately not tested up to 9 months at 60°C (140°F), the test being stopped after
6 months in the early stages ofthe testing program).

It is important to note that, for a number of mixtures, especially the control mixtures, the
expansion at 60°C (140°F) levelled off after 9 months (as illustrated in Figure 5.5), likely
due to excessive alkali leaching.

4.5 Chemistry of concrete pore solutions

Table 5.9 gives the results of the chemical analysis of pore solutions expressed from
concrete test prisms after various periods of storage at 38 or 60°C (100 or 140°F). Figures
5.6A and 5.6B do not show any clear correlation between the expansion due to ASR and
the pH in the pore solution. However, on average, the pH tends to decrease slightly with
time. This is not surprising considering that ASR consumes OH* ions (reaction process of
109

reactive silica) and alkali ions (incorporation in ASR products), and that alkali leaching
from the test prisms occurs with time.

For the mixtures incorporating lithium nitrate, the [Li]/[Na+K] molar ratio in the pore
solution decreases with time compared to the original design ratio. Table 5.9 shows that this
ratio at early age (7 to 28 days) is approximately 0.43, 0.34, and 0.23 for mixtures with
original (or "design") ratios of 0.92, 0.74, and 0.56, respectively. This confirms that about
only half or less of the added lithium is thus found in the pore solution, as reported by
9R 90

previous researchs ' . The high proportion of Li incorporated in the cement hydrates can
be explained by the fact that lithium is readily available in the mixture water and is thus
incorporated in the cement hydrates during the early phase of hydration in greater
proportions than the other alkalis.
Figures 5.7A and 5.7B also show that the [Li]/[Na+K] ratio decreases over time in the
presence of reactive aggregates. The possible explanations for this behaviour are that
lithium ions are more rapidly consumed by the reactions, more rapidly leached with time
than the other alkali ions or preferentialy incorporated in the cement hydrates as the
hydratation continues. However, since Figure 5.7A shows that the [Li]/[Na+K] does not
decrease (even slightly increasing) with time in the presence of non-reactive aggregates, it
is more likely that lithium is more preferentially consumed and thus available in lower
amounts in the pore solution compared to the other alkalis.

For the reactive (ONI) and non-reactive (NF) concretes incorporating the Li-glass, the
[Li]/[Na+K] in the mixture with an original design ratio of 0.74 is about 0.1 at 91 days and
actually increases with time, which suggests that Li is slowly released with time
(Table 5.9). The relative amount of lithium is higher in the pore solution of the concrete
made with non-reactive aggregates compared to the concretes made with reactive
aggregates. This may be due to the increased leaching or consumption of lithium in the
reaction processes for the reactive concretes; also, the lithium available progressively to the
pore solution did not result in expansion reduction. The low release rate of lithium by the
lithium glass tested likely explains its limited effectiveness against ASR.
110

5 Conclusion

Concrete mixtures incorporating 12 different Canadian reactive aggregates plus one non-
reactive aggregate, and various amounts of lithium-based admixtures (LiN03 or Li-glass)
were subjected to concrete prisms expansion tests in humid air at 38°C (100°F) (standard
test) and 60°C (140°F) (accelerated test). At different times, the pore solution of a number
ofthe above concretes was extracted under high pressure and chemically analyzed.

1. Based on a 0.040% expansion limit criterion at 2 years (38°C (100°F)), the standard
dosage of LiNÛ3 ([Li]/[Na+K] = 0.74) was effective in the presence of 6 of the
12 reactive aggregates tested (ratio needed was between 0.56 and 0.71); three
aggregates required a higher LiNÛ3 dosage (ratio between 0.75 and 1.04), while a
1.11 ratio (maximum tested in this study) was not sufficient in the presence of three
ofthe reactive aggregates tested;

2. The results obtained in this study confirmed that the effective dosage of
LiNÛ3 is not related to the degree of reactivity of the reactive aggregates
to counteract; neither could it be directly correlated to the pétrographie
nature or rock type involved. It seems thus difficult to predict the response of an
aggregate to lithium based on either its reactivity or mineralogy which makes it
impossible to develop prescriptive guidelines for lithium at this time. In other
words, the dosage of lithium required for a given reactive aggregate has to be
determined by appropriate performance testing; at this time, it is the authors'
opinion that the only reliable test method for that purpose is the concrete prism test
CSA A23.2-14A or ASTM C 1293, at least in moderate to high-alkali concrete
systems as used in this study;

3. The Li-glass tested was clearly not effective against the various alkali-silica reactive
aggregates tested, and not significantly more effective when more finely ground.
The chemistry of the concrete pore solution suggests that most of the lithium is not
released by this glass even after 365 days at 38°C (100°F), and is thus not available
to prevent excessive expansion due to ASR;
Ill

4. The silica fume/slag blended cement was found to be effective to control ASR
expansion, as well as the supplementary cementing materials tested (slag and fly
ashes), provided their composition and the quantity used as cement replacement are
adequate. The addition of LiN03 to cementitious systems incorporating fly ash or
slag did not seem to offer significant advantage;

5. If the accelerated concrete prism test at 60°C (140°F) is used for evaluating the
effectiveness of LiN03 against ASR, in order to reject all mixtures that do not
satisfy to the standard CPT at 38°C (100°F) (2-year expansion limit criterion of
0.040%), it seems necessary to use a 9-month expansion limit of 0.030% for
acceptance, and even less. Indeed, using such a quite severe limit which rejects
5 out of the 9 LiNÛ3 mixtures that passed in the standard CPT (i.e. at 38°C
(100°F)), one out ofthe 13 LiNÛ3 mixtures that failed in the standard test even
passed at 60°C(140°F);

6. The presence of L1NO3 in the range tested in this study (i.e. up to 150% of the
standard dosage) did not deleteriously affect the compressive strengh of concrete;

7. Li ions are incorporated in greater proportions than Na and K ions in the cement
hydrates. At equilibrium, the [Li]/[Na+K] ratio in the concrete pore solution is
typically less than 50% ofthe original ratio used for designing the mixtures;

8. In the presence of L.NO3, the [Li] and [Li]/[Na+K] tend to decrease with time in the
presence of reactive aggregates (in other words with ASR), which suggests that Li
ions are progressively and more rapidly incorporated into reaction products than Na
and K ions.

6 Acknowledgements

This study was financially supported by Euclid Admixtures Canada Inc. and the National
Science and Engineering Research Council of Canada (NSERC).
112

7 References

[1] Fournier, B., Bérubé, M.A., Rogers, C.A., "CSA standard practice to evaluate
potential alkali-reactivity of aggregates and to select preventive measures against
AAR in new concrete structures," Proceedings l l ' International Conference on
AAR in Concrete, Québec City, Canada, June 2000, pp. 633-642.

[2] CSA (2004) : Standard CSA A23.1-04 "Concrete materials and methods of concrete
constructions, and Standard CSA A23.2-04," Methods of test for concrete.
Canadian Standards Association, Mississauga, Ontario, Canada, 2004.

[3] McCoy, W.J., Caldwell, A.G., "New approach in inhibiting alkali-aggregate


expansion," ACI Materials Journal,22 (9), 1951, pp. 693-706.

[4] Lawrence, M., Vivian, H.F.. "The reactions of various alkalies with silica,"
Australian Journal of Applied Science, 1961, pp. 96-103.

[5] Sakaguchi, Y., Takakura, M., Kitagawa, A., Hori, T., Tomosawa, F., Abe, M., "The
inhibiting effect of lithium compounds on ASR," Proceedings 8" International
Conference on AAR in Concrete, Kyoto, Japan, 1989, pp. 229-234.

[6] Ohama, Y., Demura, K., Kakegawa, M., "Inhibiting alkali-aggregate reaction with
chemical admixtures," Proceedings 8th International Conference on AAR in
Concrete, Kyoto, Japan, 1989, pp. 253-258.

[7] Stark, D., "Lithium salt admixtures - an alternative method to prevent expansive
alkali-silica reactivity," Proceedings 91 International Conference on AAR in
Concrete, London, UK, 1992, pp. 1017-1025.

[8] Stark, D., Morgan, B., Okamoto, P., Diamond, S., "Eliminating or minimizing the
risk of alkali-silica reactivity," National Research Council, Strategic Highway
Research Program (SHRP), SHRP-C 343, Washington D.C, 1993, 266 p.
113

[9] Stokes, D.B., Wang, H.H., Diamond, S., "A lithium-based admixture for ASR
control that does not increase the pore solution pH," Proceedings 5th CANMET/ACI
International Conference on Superplastisizers and Other Chemical Admixtures in
Concrete, Roma, Italy, ACI SP 173, 1997, pp. 855-868.

[10] Fournier, B., Bérubé, M.A., "Utilisation du lithium pour combattre la réaction
alcalis-silice," Proceedings ACI Progress in Concrete, Montréal, Canada, 2001,
12 p.

[11] Baxter, S.Z., Stokes, D.B., Manissero, CE., "A Lithium-based pozzolan for ASR
control," Proceedings 11" International Conference on AAR in Concrete, Québec
City, Canada, June 2000, pp. 573-582.

[12] Thomas, M.D.A., Hooper, R., Stokes, D., "Use of lithium-containing compounds to
th

control expansion in concrete due to ASR," Proceedings 11 International


Conference on AAR in Concrete, Québec City, Canada, June 2000, pp. 783-792.
[13] Stokes, D.B., Manissero, CE., "Use of lithium-based admixture to control ASR
expansion," Proceedings 14'Reunion Tecnica Asociacion Argentina de Tecnologia
del Hormigon, Universidad Nacional del Centro de la Provincia de Buenos Aires,
Argentina, 2001, 10 p.

[14] Lumley, J.S., "ASR suppression by lithium compounds," Cement and Concrete
Research, 27 (2), 1997, pp. 235-244.

[15] Blackwell, B.Q., Thomas, M.D.A. and Sutherland, A., "Use of Lithium to Control
Expansion Due to ASR in Concrete Containing U.K. Aggregates," Proceedings of
International Conference. CANMET/ACI Durability of Concrete, Sydney, Australia,
ACI SP 170, V.M. Malhotra Editor, August 1997, pp. 649-663.

[16] Durand, B., "More results about the use of lithium salts and mineral admisture to
th
inhibit
inhibit ASR
ASR in
in concrete,"
concrete," Proceedings
Proceedings 11
11 International Conference on AAR in
Internatio
Concrete, Québec City, Canada, June 2000, pp. 623-632
114

[17] Fournier, B., Stokes, D.B., Ferro, A., "Comparative field and laboratory
investigations on the use of supplementary cementing materials (SCM's) and
lithium-based admixtures to control expansion due to alkali-silica reaction (ASR) in
concrete," Proceedings 6th International CANMET/ACI Conference on Durability of
Concrete, Thessaloniki, Greece, 2003, pp. 823-851.

[18] Folliard, K.J., Thomas, M.D.A., Kurtis, K.E., "Guidelines for use of lithium to
mitigate or prevent ASR," Publication No. FHWA-RD03-047, 2003, 61 p.

[19] CSA A23.2-14A, " Potential expansivity of aggregates (procedure for length change
due to alkali-aggregate reaction in concrete prisms at 38 °C)," Methods of test for
concrete. Canadian Standards Association, Mississauga, Ontario, Canada, 2004,
pp. 246-256.

[20] ASTM C 1293, " Standard Test Method for Determination of Length Change of
Concrete Due to Alkali- Silica Reaction," Annual Book of ASTM Standards,
Vol. 04.02, Concrete and Aggregates, 2005.

[21] CSA A23.2-28A, " Standard practice for laboratory testing to demonstrate the
effectiveness of supplementary cementing materials and lithium-based admixtures
to prevent alkali-silica reaction in concrete," Methods of test for concrete. Canadian
Standards Association, Mississauga, Ontario, Canada, 2004, pp. 327-330.

[22] De Grosbois, M., Fontaine, E., "Performance of the 60°C-accelerated concrete


prism test for the evaluation of potential alkali-reactivity of concrete aggregates,"
th

Proceedings 11 International Conference on AAR in Concrete, Québec City,


Canada, June 2000, pp. 277-286.
115

[23] Fournier, B., Chevrier, R., DeGrosbois, M., Lisella, R., Folliard, K., Ideker, J.,
Shehata, M., Thomas, M. and Baxter, S., "The accelerated concrete prism test
(60C) : variability of the test method and proposed expansion limits," 12th
International Conference on AAR in Concrete, Beijing, China, 2004, pp. 314-323.

[24] ASTM C 295, "Standard Guide for Pétrographie Examination of Aggregates for
Concrete," Annual Book of ASTM Standards , Vol.04.02, Concrete and Aggregates,
2003.

[25] ASTM C 39, "Standard Test Method for Compressive Strength of Cylindrical
Concrete Specimens," Annual Book of ASTM Standards, Vol.04.02, Concrete and
Aggregates, 2005.

[26] CSA A23.2-27A, " Standard practice to identify degree of alkali-reactivity of


aggregates and to identify measures to avoid deleterious expansion in concrete,"
Methods of test for concrete. Canadian Standards Association, Mississauga,
Ontario, Canada, 2004, pp. 317-326.

[27] Feng, X., Thomas, M.D.A., Bremner, T.W., Balcom, B.J., and Folliard, K.J.,
"Effect of LiN03 on ASR-Expansion and Dissolution of Silica," Marc-André
Bérubé Symposium on Alkali-Aggregate Reactivity in Concrete, May 2006,
Montréal, Canada, 2006, pp. 139-152.

[28] Diamond, S., "Unique response of LiN03 as an alkali silica reaction-preventive


admixture," Cement and Concrete Research, 29 (8), 1999, pp. 1271-1275.

[29] Bérubé, M.A., Tremblay, C , Fournier, B., Thomas, M.D.A., Stokes, D., "Influence
of lithium products proposed for counteracting alkali-silica reactivity on the
chemistry of cement hydrates and pore solution," Cement and Concrete Research,
34, 2004, pp. 1645-1660.
116

Table 5.1 - Origin, composition, and reactivity level ofthe coarse aggregates
Origin Identification Rock type Degree of ASR*
NS1 Greywacke High
Nova Scotia
NS2 Granite Moderate
NB1 Limestone High
New Brunswick NB2 Rhyolite High
NB3 Greywacke Moderate
QC1 Dolostone Moderate
Québec
QC2 Chloritic schist Moderate
Siliceous
ONI (Spratt) High
Ontario limestone
ON2 Polygenic gravel Moderate
AL1 Polygenic gravel Moderate
Alberta
AL2 Polygenic gravel Moderate
British Columbia BC Polygenic gravel High
Newfoundland NF Pure limestone Non-reactive
* Based on CSA A23.2-27A (One-year expansion in CSA test method A23.2-14A:
0-0.04%: non-reactive; 0.04-0.12% : moderately reactive; >0.12% : highly reactive).

Table 5.2 - Chemical analysis ofthe cements and SCM's


FA2
Type GU Ternary* Slag FA1 FA3
(Nova
Oxides cement cement (Ontario) (Alberta) (Ontario)
Scotia)
% b y mass
Si0 2 19.70 26.5 36.7 52.3 47.9 36.8
CaO 62.70 53.1 34.3 13.4 2.51 16.1
A1203 5.28 6.10 10.0 23.7 24.6 19.6
Fe 2 0 3 2.34 1.96 0.60 4.32 18.5 5.44
MgO 2.79 4.06 15.3 1.22 1.26 3.68
SO3 3.13 3.95 2.26 0.11 0.57 3.55
Na 2 0 0.26 0.30 0.48 3.30 0.55 9.46
K20 0.85 0.84 0.44 0.51 2.81 0.50
LOI 2.28 1.45 1.56 0.98 0.71 0.49
Na2Oe 0.82 0.85 0.77 3.64 2.40 9.79
* Composed by 74% of Type GU cement, 21 % of slag, and 5% of silica fume.
117

Table 5.3 - Concrete mixtures tested and expansion (%) at 2 years in the standard CPT at
38°C(100°F)

Mixture Aggregate
NSl NS2 NB1 NB2 NB3 QC1QC2 ONI ON2 ALI AL2 BC NF Total
Control .181 .110 .220 .234 .118 .122 .112 .172 .180 .113 .149 .142 .013 13
150%LiN(LiNO3) .082 .076 .024 .032 4
125% LiN .046 .064 .065 .049 .016 .026 .042 7
100% LiN .131 .073 .013 .031 .022 .068 .085 .015 .102 .056 .010 .011 12
75% LiN .034 .086 .103 .076 .133 5
100% LiG (Li glass) .175 .279 .144 .099 .189 .147 .148 .012 8
100% LiG (ground) .170 1
75 % LiG .110 .170 2
20 % FA* (fly ash) .027 123 .014 .005 4
20% FA* + 75% LiN .033 119 .020 3
20% FA* + 50% LiN .000 1
20% FA* + 75% LiG .021 1
50% Slag 012 1
35% Slag 041 .016 2
25% Slag .099 .067 2
35 % Slag + 75% LiN ,034 1
35% Slag+ 50% LiN .035 1
35% Slag+125% LiG .033 1
35%Slag + 75%LiG .033 1
35% Slag + 50% LiG .036 1
25% Slag+100% LiN .040 1
25% Slag + 75% LiN .081 1
25% Slag + 50% LiN .093 .051 2
25% Slag+100% LiG .078 1
25% Slag + 75% LiG .095 1
25% Slag + 50% LiG .080 .080 2
Ternary cement 017 .018 .014 .009 4
Ter.cem. + 50% LiN .040 .026 .020 .006 4
Total 10 2 6 3 4 6 24 12 5 7 2 3 87
* FA1 with AL aggregates, FA2 with ONI and F A3 with NSl.
118

Table 5.4 - Petrography ofthe coarse aggregates


Aggregates
(main rock type) and Pétrographie facies (%) Reactive phases (%)
Degree of reactivity*
NS 1 (greywacke) Greywacke (77%) Microcrystalline quartz (10%)
High Argillaceous greywacke (23%) Microcrystalline quartz (2-3%)
NS2 Pink granite (82%) Microcrystalline quartz (10%)
(granite) Altered granite (9%) Microcrystalline quartz (5%), mylonite (highly reactive)
Moderate Sandstone (8%) Quartz undul. extint. (2%), micro.-crypto. quartz (10%)
Grey granite (1%) Microcrystalline quartz (5-10%)
NB 1 Dark calcilutite (98%) Microcrystalline quartz (5%)
(limestone) Fine-grained calcarenite (1%) Micro.-cryptocrystalline quartz (5%)
High Grey calcilutite (1%) Microcrystalline quartz (5-8%)
NB2 (rhyolite) Rhyolite (99%) Devitrified glass (85%)
High Coarser-grained rhyolite (1%) Microcrystalline quartz (5%)
Carbonate sandstone (23%) Microcrystalline quartz (10%)
Basalt (23%) No reactive mineral phase
NB3 Argillite(18%) Microcrystalline quartz (60%)
(gravel) Sericite schist ( 12%) Microcrystalline quartz (1%)
Moderate Altered felsic rock (9%) Microcrystalline quartz (<1%)
Meta-quartzite (8%) Quartz undul. extinc. (95%), microcrystalline quartz (5%)
Granite (7%) No reactive mineral phase
QC1 (dolostone) Dolostone (97%) Microcrystalline quartz (1-2%)
Moderate Clayey dolostone (3%) Microcrystalline quartz (2-3%)
QC2 Chloritic schist (63%) Microcrystalline quartz (40%)
(chloritic schist) Sericite schist (24%) Microcrystalline quartz (70%)
Moderate Cristalline quartz (13%) Micro- and cryptocristalline quartz (60%)
ONI Fine-grained calcarenite (65%) Cryptocrystalline quartz (>25%)
(siliceous limestone) Coarse-gr. calcarenite (34%) No reactive mineral phase
High Calcilutite (1%) Cryptocrystalline quartz (25%)
Green greywacke (60%) Microcrystalline quartz (15%)
ON2 Quartz sandstone (23%) Quartz cement (30%)
(polygenic gravel) Granite (10%) Microcrystalline quartz, quartz undul. extinct. (2%)
Moderate Quartz sandstone (5%) Quartz undul. extinct. (75%), quartz cement
Red quartzite (2%) Micro- to cryptocristalline quartz (90%)
Sandstone (35%) Microcrystalline quartz (5%), quartz overgrowths (5%)
Greywacke (23%) Microcrystalline quartz (5-10%)
All Pink sandstone ( 14%) No reactive mineral phase
(polygenic gravel) Calcilutite (10%) Microcrystalline quartz (2%)
Moderate Altered basalt (8%) Microcrystalline quartz (1%), chalcedony (<1%)
Green sandstone (8%) No reactive mineral phase
Granite (2%) No reactive mineral phase
Quartz sandstone (67%) Microcrystalline quartz (2-3%)
A12 Sandstone with chert (14%) Chalcedony (2%), microcrystalline quartz (80%)
Claystone(10%) Micro.-cryptocrystalline quartz (15%)
(polygenic gravel)
Chert (6%) Cryptocrystalline quartz (95%)
Moderate
Cristalline quartz (2%) Quartz with undulatory extinction (5%)
Siliceous siltstone ( 1 %) Micro- to cryptocrystalline quartz (45%),chalcedony ( 1 %)
Pink greywacke (35 %) No reactive mineral phase
Quartz sandstone ( 17%) No reactive mineral phase
BC
Chert (14%) Micro- to cryptocrystalline quartz (20%)
(polygenic gravel)
Gneiss (14%) Quartz with undulatory extinction (3%)
High
Basalt (12%) No reactive mineral phase
Quartzite (8%) Sutured quartz and quartz with undul. extinct. (100%)
NF (pure limestone) Calcilutite (98%) No reactive mineral phase
Non-Reactive Calcarenite (2%) Microcrystalline quartz (2%)
* Based on CSA A23.2-27A-2004 (One-year expansion in CSA test method A23.2-14A:
0-0.040% : non-reactive; 0.040-0.120% : moderately reactive; >0.120% : highly reactive).
119

able 5.5 - Minimum [Li]/[Na+K] ratio to satisfy the 0.040% expansion limit criterion in
the concrete prism test (the results are classed by increasing reactivity level
ofthe coarse aggregates investigated in the control mixtures)
Expansion Minimum Minimum
(%) of [Li]/[Na+K] to [Li]/[Na+K] to satisfy
Reactivitj
Aggregates control at 1 satisfy the 2-year the 6-month CPT
Level
year and CPT criterion at criterion at 60°C
3 8 C (100°F) 38°C (100°F) (140°F)
NS2 granite .029 0.56 0.43
QC2 chloritic
.082 >0.94 >0.94
schist
Moderately NB3 gravel .087 0.71 0.30
reactive QCl
.100 0.61 0.70
dolostone
AL1 gravel .103 0.91 0.58
AL2 gravel .113 0.97 0.67
ON2 gravel .122 0.66 0.50
BC gravel .131 0.58 0.44
NSl
.142 >1.11 >1.11
greywacke
Highly
ONI
reactive .162 1.04 0.65
limestone
NB2 rhyolite .172 0.63 0.67
NB1
.199 >1.11 >1.11
limestone
120

Table 5.6 - Expansion (%) at 6 months in the accelerated CPT at 60°C (140°F)

Mixture Aggregate
NSl NS2 NB1 NB2 NB3 QC1QC2 OMON2AL1AL2 BC NF
Control .198 .083 .180 .279 .052 .104 .126 .165 .157 .102 .120 .079 .019
150%LiN(LiNO 3 ) .054 .078 .034 .019
125% LiN .041 .077 .071 .021 .023 .035 .039
100% LiN .053 .065 .018 .026 .036 .071 .021 .016 .024 .034 .012 .001
75% LiN .027 .029 .092 .024 .051
100% LiG (Li glass) .198 .213 .119 .113 .149 .171 .122 .009
100% LiG (ground) .121
75 % LiG .138 .168
20 % FA* (fly ash) .021 .114 .009.006
20% FA* + 75% LiN .017 .103 .011
20% FA* + 50% LiN .006
20% FA* + 75% LiG .027
50 % Slag .000
35% Slag .020 .001
25% Slag .048 .006
35 % Slag + 75% LiN .008
35% Slag + 50% LiN .013
35% S l a g + 1 2 5 % LiG .010
35% Slag + 75% LiG .011
35% Slag + 50% LiG .019
2 5 % S l a g + 1 0 0 % LiN .018
2 5 % Slag + 75% LiN .035
2 5 % Slag + 50% LiN .025 .008
25% S l a g + 1 0 0 % LiG .040
2 5 % Slag + 75% LiG .046
25% Slag + 50% LiG .051 .012
Ternary cement .012 .010 .006 .006
Tern. cem. + 50% LiN .014 .015 .006 .000
FA1 with AL aggregates, FA2 with ONI and FA3 with NSl.
121

Table 5.7 - Expansion (%) at 9 months in the accelerated CPT at 60°C(140°F)

Mixture Aggregate
NSl NS2 NB1 NB2 NB3 QCl QC2 ONI ON2 ALI AL2 BC NF
Control NA .090 .199 .303 .060 NA NA NA NA .106 .130 .080 NA
150%LiN(LiNO 3 ) .073 .106 .038 .019
125% LiN NA .091 .088 NA .034 .038 .043
100% LiN NA .086 .016 .033 NA NA NA .027 .028 .038 .012 NA
75% LiN .031 .038 NA .033 .055
100% LiG (Li glass) .209 .229 NA NA NA NA .131 NA
100% LiG (ground) .135
75 % LiG NA NA
20 % FA* (fly ash) .030 .126 .017 .015
20% FA* + 75% LiN .027 .125 .017
20% FA* + 50% LiN .008
20% FA* + 75% LiG .043
50 % Slag NA
35% Slag NA NA
25% Slag .061 NA
35 % Slag + 75% LiN NA
35% Slag + 50% LiN NA
35% S l a g + 1 2 5 % LiG NA
35% Slag + 75% LiG NA
35% Slag + 50% LiG NA
25% S l a g + 1 0 0 % LiN NA
25% Slag + 75% LiN NA
25% Slag + 50% LiN NA NA
25% S l a g + 1 0 0 % LiG NA
25% Slag + 75% LiG NA
25% Slag + 50% LiG NA .012
Ternary cement .014 .025 .010 .010
Tern. Cem. + 50% LiN .019 .026 .010 .000
NA Non available;
* FA1 with AL aggregates, FA2 with ONI and FA3 with NSl.
122

Table 5.8 - Correlation between the CPT at 60°C (140°F) vs 38°C (100°F)
% of safe prediction
CPT vs AMBT Concrete type % of correlation* with CPT at 60°C
(140°F) **
Control 12/13 = 92% 12/13 = 92%
LiN0 3 2 0 / 2 8 = 71% 22 / 28 = 79%
lyearat38°C(100°F)
Liglass 11 /11 = 100% 11/11 = 100%
vs. 3 months at 60°C
SCM's 12/13 = 92% 12/13 = 92%
(140°F)
SCM's + LiN0 3 12/14 = 86% 12/14 = 86%
(limits at 0.04%)
SCM's + Liglass 4 / 8 = 50% 4 / 8 = 50%
Total 71 / 87 = 82% 73 / 87 = 84%
Control 13/13 = 100% 13/13 = 100%
LiN0 3 2 1 / 2 8 = 75% 21 / 28 = 75%
2 years at 38°C
Liglass 11/11 = 100% 11/11 = 100%
(100°F) vs. 6 months at
SCM's 11/13 = 85% 11 /13 = 85%
60°C (140°F)
SCM's + LiN0 3 10/14 = 71% 10/14 = 71%
(limits at 0.04%)
SCM's + Liglass 7 / 8 = 88% 7 / 8 = 88%
Total 73 / 87 = 84% 73 / 87 = 84%
Control 13/13 = 100% 13/13 = 100%
LiN0 3 18/22 = 82% 18/22 = 82%
2 years at 38°C
Liglass 11 /11 = 100% 11/11 = 100%
(100°F) vs. 9 months at
SCM's 9/9=100% 9/9=100%
60°C (140°F)
SCM's + LiN0 3 7 / 8 = 88% 7 / 8 = 88%
(limits at 0.04%)
SCM's + Liglass 3 / 4 = 75% 4/4=100%
Total 61 / 67 = 9 1 % 62 / 67 = 93%
Control 13/13 = 100% 13/13 = 100%
2 years at 38°C L1NO3 16/22 = 73% 21/22 = 95%
(100°F) (limit at Liglass 11/11 = 100% 11/11 = 100%
0.04%) vs. 9 months at SCM's 9 / 9 = 100% 9/9=100%
60°C (140°F) (limit at SCM's + LiN0 3 7 / 8 = 88% 7 / 8 = 88%
0.03%) SCM's + Liglass 3 / 4 = 75% 4/4=100%
Total 59 / 67 = 88% 65 / 67 - 97%
* % of all mixtures passing at 38 and 60°C (100 and 140°F) + failing at 38 and60°C(100andl40°F).
** % of all mixtures passing at 38 and 60°C (100 and 140°F) + failing at 38 and60°C(100andl40°F) +
passing at 38°C (100°F) while failing at 60°C (140°F).
123

Table 5.9 - Chemical analysis of pore solutions expressed from concrete test prisms after
various periods of storage at 38 or 60°C (100 or 140°F)
Ratio
T°C Age Expan- Concentration in Ratio
Aggregates [Li]/[Na+K] sion pore solution (Mol/1) [Li]/[Na PH
/°F (days)
in concrete (%) Na K Li +K1
0 91 .002 .411 .327 0 13.74
NF
0 365 .007 .324 .293 0 — 13.76
pure 38/ .74 91 .002 .505 .422 .308 .332 13.65
limestone 100 .74 365 .008 .354 .321 .236 .349 13.76
(Non- .74 (Liglass) 91 .004 .509 .358 .085 .098 13.80
reactive) .74 (Liglass) 365 .008 .283 .257 .105 .194 13.76
0 91 .010 .408 .292 0 — 13.72
0 365 .122 .256 .159 0 — 13.58
38/
.74 28 .006 .394 .362 .253 .335 13.61
100
.74 91 .006 .465 .343 .246 .305 13.56
.74 365 .013 .350 .254 .165 .273 13.45
ON2 0 7 .004 .361 .312 0 ... 13.71
gravel 0 28 .030 .353 .273 0 ... 13.65
(Moderately 0 91 .124 .263 .165 0 — 13.58
reactive) 60 / .56 7 .004 .410 .361 .178 .231 13.56
140 .56 28 .014 .414 .334 .154 .206 13.48
.56 91 .022 .270 .189 .096 .208 13.45
.92 7 .006 .429 .391 .360 .439 13.53
.92 28 .018 .470 .368 .323 .386 13.57
.92 91 .022 .322 .222 .194 .356 13.45
.92 7 .000 .442 .435 .400 .456 —
QC2 38/
.92 28 .003 .484 .465 .406 .428 13.69
chloritic 100 .92 91 .017 .401 .376 .305 .393 —
schist
.92 7 .007 .469 .434 .379 .417 13.66
(Moderately 60/ .92 28 .025 .443 .362 .286 .355 13.51
reactive) 140 .92 91 .047 .346 .251 .186 .311 13.45
0 91 .066 .335 .225 0 13.60
ONI (Spratt) 0 365 .162 .189 .126 0 13.58
siliceous
J i l l V V - v \ I >?

limestone 38/ .74 91 .009 .461 .325 .248 .316 13.56


100 .74 365 .058 .228 .150 .095 .252 13.58
(Highly
reactive) .74 (Liglass) 91 .080 .334 .220 .063 .114 13.65
.74 (Liglass) 365 .178 .161 .178 .041 .146 13.45
124

0.24 • ' I I i !
l ' 1
' -
O
0.2-
\
■a
c
10
0.16- %

CM
0.12 - -
____LL_Mjvel

C 0.08 - -;
o
s
c
re 0 . 0 4 -
a
X
IU •

T ' i ' i—'—i ' r ' I ' i i i i i i i '


0.185 0.37 0.555 0.74 0.925 1.11 0.185 0.37 0.555 0.74 0.925 1.11
Ratio [Li]/[Na+K] in concrete Ratio [Li]/[Na+K] in concrete

0.24

0 0.185 0.37 0.555 0.74 0.925 1.11


Ratio [Li]/[Na+K] in concrete

Figure 5.1 - Effectiveness of lithium nitrate in the presence of all reactive aggregates at
38°C (100°F). (A) NSl, A12, NB3, NS2. (B) NB2, All, QC2, BC. (C) NB1,
ONl,ON2,QCl.
125

' 1 ' 1 ' ON1 '


NS1
8
"D
! O N 1 ground Li­glass
i 0.16 ­
"%2_ ­
? _. AU

«M
<£i­­"""
A
£ 0.12
C i. __QÇ2___
o

1 ■ ^ ­
X
111
0.08 ■;

i ' l ' i ' i ' i '


0.185 0.37 0.555 0.74 0.925 1.11
Ratio [Li)/[Na+K] in concrete

Figure 5.2 ­ Effectiveness of the lithium glass in the presence of different reactive
aggregates at 38°C (100°F).
126

B
0.24 0.24
~i ' i ' r
ON1 limestone ON2 gravel

Slag
lTernary^ement
fernarycém®"1
r
I ' I ' i ' 1 ' I -i 1 r-
0.185 0.37 0.555 0.74 0.925 1.11 0.185 0.37 0.555 0.74 0.925 1.11
Ratio [Li]/[Na+K] in concrete Ratio [Li]/[Na+K] in concrete
C D
0.24 0.24
n ' i ' i
NS1 greywacke NB1 limestone

-i | , | , -
~l—'—l—'—i—' l <~
0.185 0.37 0.555 0.74 0.925 1.11 0.185 0.37 0.555 0.74 0.925 1.11
Ratio [Li]/[Na+K] in concrete Ratio [Li]/[Na+K] in concrete

Figure 5.3 - Effectiveness of SCM-LiN03 combinations in the presence of different


reactive aggregates at 38°C (100°F). (A) ONI. (B) ON2. (C) NSl. (D) NB1.
127

B
v.to I I I I I
1 ' 1 ' 1 ' 1 i j u._o — 1 i | , | i 1 J
1
0 Control
. | . *|
0 Control
! O LIN03
ô o­ 2 4 " 1
4 Liglass
­ ~ 0_24­ 0 LIN03
0 o 4 Liglass
o
eo ■ SCM's
• / ­ o
o " ■ SCM's
<
^o 0.2­
i Q SCM's + LiN03
__ VI 0.2­ D SCM's + LIN03
** S
J_ .c
B
0
C c
/ • ­
• A •4 __ O
E 0.16­

E 0.16­
•1 —
« <s
re /^ ♦ *re
_ « / ­
g 0.12­ ! ♦ / ♦ ­ 0.12­
>% + « ­
c • ♦ c ­
O ' * S o /o
M
c
0.08­
— 8 0.08­
&r * * ­
re
a
• a
re
x 0
w 0.04­ UJ 0.04­ * ô °" T t " ■ > • •
­ ^ ■O 0*0
V ­
*0 B
0 ti 1
i ' i ' i i 0 ­ 1 1
1 ' !
W ^ 0.04
1 0.08
i ' 0.12 0.16 0.2 0.24 0.28
1 ' 1 ' '
0 0.04 0.08 0.12 0.16 0.2 0.24 0.28
Expansion (%) at 1 year (38°C) Expansion (%) at 2 years (38°C)
C
0.2B

n '—i—'i ' i—'—r


0 0.04 0.08 0.12 0.16 0.2 0.24 0.28
Expansion (%) at 2 years (38°C)

Figure 5.4 ­ Expansions at 38°C and 60°C (100°F and 140°F) in the concrete prism test.
(A) After 1 year at 38°C (100°F) vs. 3 months at 60°C (140°F). (B) After
2 years at 38°C (100°F) vs. 6 months at 60°C (140°F). (C) After 2 years at
38°C (100°F) vs. 9 months at 60°C (140°F).
128

91 182 273 91 182 273


Age (days) Age (days)

0.20

0.18 BC Gravel

0.16 -

[U]/[Na+iq=0.74

~i r~
91 182 273 91 182 273
A g e (days) Age (days)

Figure 5.5- Expansion curves in the concrete prism test at 60°C (140°F). (A) AL2
gravel. (B) NB1 limestone. (C) BC gravel. (D) AL1 gravel.
129

B
13.8 I i ■ i ■ i
Mixtures at 38°C Mixtures at 60°C
­
13.7­ 13.7­

9 7 days
• 91 days ▲ 28 days ­
▲ 365 days • ■ 91 days

ï.13.6 13.6­ ­



13.5 13.5­
A
m ■
13.4 ~"I ' I ' I ' I <~ l ' l I '
0.04 0.08 0.12 0.16 0.2 0.04 0.08 0.12 0.16
Expansion (%) Expansion (%)

Figure 5.6 ­ pH of the concrete pore solution of a number of mixtures. (A) At 38°C
(100°F). (B)At60°C(140°F).

B
0.45 0.45
i i | i i i | , i i | i i i
i i r i—i—r
Mixtures at 38°C with original Mixtures at 60°C with original
Ratio [Li]/[Na+k]=0.74 Ratio [Li]/[Na+k]=0.92
0.40 0.40

+
re
0.35 ■=■ 0.35
NF(nonreactiveL

O
re
s. K

0.30 0.30 H

0.25 0.25 i i—i—i i i i i i i i r


182 273 364 0 7 28 56 91
Age(days) Age (days)

Figure 5.7 ­ [Li]/[Na+K] of the concrete pore solution of a number of mixtures.


(A) At 38°C (100°F). (B) At 60°C (140°F).
130

6 Utilisation de l'essai accéléré sur barres de mortier pour évaluer


l'efficacité du LiN0 3 contre la réaction alcalis-silice (parties I et II)
6.1 Introduction

Cette publication a été soumise en deux parties à la revue Journal of ASTM International
(JAI) le 9 juillet 2007 pour la première partie et le 15 juillet 2007 pour la seconde partie.
Les articles ont été acceptés et ont été publiés en septembre 2008 (volume 5, numéro 8).

PARTIE I

6.2 Résumé de la partie I

Jusqu'à maintenant, le seul essai fiable pour évaluer l'efficacité du nitrate de lithium contre
la réaction alcalis-silice (RAS) dans le béton est l'essai sur prismes de béton (CPT)
CSA A23.2-14A ou ASTM C 1293, d'une durée prolongée à deux ans. Dans sa forme
actuelle, l'essai accéléré sur barres de mortier (AMBT) CSA A23.2-25A ou ASTM C 1260,
communément utilisé, ne permet pas d'évaluer cette efficacité. Pour qu'il en soit ainsi, il
doit être modifié de manière à améliorer sa fiabilité. La première partie de cette étude
(Part I), qui implique une grande variété de granulats réactifs du Canada et des États-Unis,
vise à évaluer la distribution de divers ions (OH", Na+, K+, Li+, et silice) entre la solution
interstitielle des barres de mortier et la solution de trempage de l'AMBT. De plus,
l'influence d'un certain nombre de paramètres expérimentaux sur l'expansion des barres de
mortier avec ou sans nitrate de lithium est étudiée, soit la présence ou non de lithium dans
les barres de mortier d'origine, la concentration de Li dans la solution de trempage, la
concentration et la composition de cette solution (NaOH ou KOH), le contenu en alcalis du
ciment utilisé, le ratio eau/ciment et le trempage dans l'eau pure durant les premières
24 heures. La deuxième partie de cette étude (Part II) comparera les résultats obtenus avec
l'AMBT avec ceux de l'essai sur prismes de béton (CPT) et proposera certaines
modifications à l'AMBT lorsque cet essai est utilisé pour évaluer l'efficacité du nitrate de
lithium contre la RAS et ce, de façon à obtenir la meilleure corrélation possible avec les
résultats plus réalistes du CPT.
131

6.3 Publication scientifique no. 3

Use of the accelerated mortar bar test to evaluate the


effectiveness of UNO3 against alkali-silica reaction
Part 1 : Pore solution chemistry and influence of various parameters

C. Tremblay 8 , M.A. Bérubé b, B. Fournier c , M.D. Thomas d, K.J. Folliard e and P. C.


Nkinamubanzi

a
PhD student, Département de géologie et de génie géologique, Université Laval, Québec, QC, Canada, GI K 7P4
b
Associate Professor, Département de géologie et de génie géologique, Université Laval, Québec, QC, Canada, G1K 7P4
c
Assistant Professor, Département de géologie et de génie géologique, Université Laval, Québec, QC, Canada, G1K. 7P4
d
Professor, Department of Civil Engineering, University of New Brunswick, Fredericton, NB, Canada, E3B 5 A3
d
Professor, Department of Civil Engineering, University of Texas, Austin, TX 78712
f
Research Scientist, CANMET Mining and Minerai Sciences Laboratories, Natural Resources Canada, ,Ottawa, ON,
Canada Kl A 0E4.

ABSTRACT

For the time being, the only reliable test method to evaluate the effectiveness of lithium
nitrate against alkali-silica reaction (ASR) in concrete is the concrete prism test (CPT)
CSA A23.2-14A or ASTM C 1293, extended to two years. In its actual form, the more
commonly-used accelerated mortar bar test (AMBT) CSA A23.2-25A or ASTM C 1260 is
not able to predict this effectiveness and needs to be modified to improve it's reliability.
Part I of this study, which involves a large variety of reactive aggregates from Canada and
the U.S.A., aims to evaluate the partition of various ions (OH", Na+, K+, Li+, and silica
species) between the mortar pore solution and the soak solution in the AMBT, and the
effect of a number of experimental parameters on the expansion of mortar bars
with/without lithium nitrate, e.g. the presence of lithium in the original mortar bars, the Li
concentration in the soak solution, the concentration and the composition (NaOH vs KOH)
of the soak solution, the cement alkali content, the water-to-cement ratio, and the initial
24-hr soaking in pure water. The second part of this study (Part II) compares the above
AMBT results with the CPT results for the same aggregates and aims to propose
modifications to the AMBT when testing lithium nitrate against ASR to better correlate
with the most realistic CPT results.

Keywords: Accelerated mortar bar test, alkali-aggregate reaction, lithium nitrate, pore
solution
132

1 Introduction

In many countries, alkali-silica reaction (ASR) is a major cause of premature concrete


deterioration that sometimes evolves into serious serviceability problems. ASR is a reaction
between the alkali (K+, Na+) and hydroxyl (OH) ions in the concrete pore solution, and
some silica species in the aggregate particles. This reaction forms a swelling gel, which
absorbs water, swells, causes stresses and microcracks in concrete, and reduces its
mechanical and durability properties. About ten years after the first cases of ASR reported
by Stanton [1], McCoy and Caldwell [2] found that certain lithium-based products can
significantly reduce and even control the expansion due to ASR. A limited number of
studies confirmed these findings during the four following decades [3-5] but it is really in
the 1990's that the interest for such products has peaked.

The most common lithium-based products tested, i.e. LiOH»H20, LiCl, LiF, Li2C03, and
LiNÛ3, reduce expansion when used in mortar or concrete. Most studies suggested that
using lithium-based admixtures with [Li]/[Na+K] molar ratios between 0.6 and 1.2
(depending on the lithium product used and the aggregate type) can inhibit deleterious ASR
expansion [6-15]. However, a number of studies also demonstrated that insufficient dosages
of LiOH, LiCl, L_2C03, and LiF may increase ASR expansion rather than reducing it, most
likely by increasing the pH (with LiOH) ofthe concrete pore solution [8, 11, 16]. On the
other hand, LiNÛ3 appears non-susceptible to a pessimum effect by not increasing the pH
ofthe pore solution [17]. A recent study [18] using a highly-reactive volcanic aggregate
also showed that LiNÛ3 is more effective in controlling ASR expansion than LiOH»H20 at
similar [Li]/[Na+K] molar ratios. LiN03 has recently become the most common lithium salt
used for suppressing ASR [19]. However, further investigations showed that although
L1NO3 was effective in controlling ASR expansion with some varieties of highly-reactive
aggregates of volcanic origin, it was significantly less effective at similar molar ratios with
some other varieties of moderately- to highly-reactive aggregates such as schists, siliceous
limestones, and greywackes [14, 20, 21]. For the time being, the mechanisms explaining the
effectiveness of lithium admixtures against ASR are not fully understood [18, 20].
133

Until the early 1990's, most ofthe experimental programs on lithium-based admixtures
involved mortar bars incorporating extremely reactive materials such as cristobalite, opal,
Pyrex and volcanic rock. Also, the test procedures generally used at the time (e.g.
ASTM C 227) are more and more considered unreliable for predicting the effectiveness of
preventive measures against ASR with natural aggregates [22]. More recently, programs
involving expansion tests on concrete specimens incorporating natural reactive aggregates
have been conducted, particularly in the U.K., U.S.A., and Canada [9-12, 14, 15, 21]. In
fact, the Concrete Prism Test (CPT) CSA A23.2-14A (or ASTM C 1293) is considered by
most researchers and practitioners as the most reliable standard test to determine the
potential alkali-reactivity of concrete aggregates; this test is carried out over a one-year
period. However, CSA A23.2-28A states that the testing period shall be extended to two
years when evaluating preventive measures against ASR. The comparison of field
performance and laboratory expansion data indicates that a 2-year, 0.04% expansion limit
criterion in the CPT is generally a good indicator of field performance of concrete
incorporating lithium admixture [14]. Unfortunately, such a test duration is judged too long
in many circumstances, and there is a need for a more rapid test that still provides reliable
results.

In 1985, Oberholster and Davies [23] proposed the Accelerated Mortar Bar Test (AMBT),
which since then has become a standard test in many countries (e.g. CSA A23.2-25A,
ASTM C 1260, RILEM AAR-04). This test is very popular mostly because it allows
assessing the potential alkali-reactivity of aggregates in only two weeks. It can also be used
for determining the effectiveness of supplementary cementing materials (SCM) against
ASR in accordance with CSA A23.2-28A or ASTM C 1567. With the objective of
accelerating the reactions, the AMBT differs from the CPT in many respects: the
aggregates are crushed to sand size, a temperature of 80 °C is used, and the mortar bars are
immersed in a IN NaOH solution. Because ofthe above differences in testing regimes and
conditions, numerous test programs have shown that, even in the absence of SCM or
Li-based admixtures, the 1-year expansion levels obtained in the CPT do not necessarily
correlate well with that obtained after 14 days in the AMBT, with the latter being severe for
many aggregates with good field performance.
134

In addition, the AMBT is not appropriate in its actual form for evaluating the effectiveness
of lithium-based admixtures against ASR. The main reason is that a large proportion of the
lithium added to the mortar actually diffuses out of the bars into the NaOH soak solution
thus reducing its efficacy to control expansion due to ASR. Because of that, some
researchers proposed adding lithium nitrate to the soak solution to keep some sort of
chemical balance between the bar and the storage solution [8]. The question then becomes
what are the appropriate lithium concentrations to be used in the bars and in the soak
solution that would best simulate the conditions prevailing in concrete. Laboratory
investigations performed on concrete specimens have shown that about half of the lithium
originally introduced in the concrete at the time of batching is bound into the cement
hydrates (thus unavailable to control ASR) and that the [Li]/[Na+K] molar ratio in the
concrete pore solution is actually less than half of the original design ratio [11, 24]. Using
similar [Li]/[Na+K] molar ratios in the soak solution and in the mortar would thus
overestimate the effectiveness that such a dosage of the lithium admixture would have in
concrete [14, 21]. Research is required to optimize the test conditions in a modified version
of the AMBT so that the results would better correlate with those obtained in the more
realistic concrete prism test.

2 Research significance

Since the pioneering work of McCoy and Caldwell [2], various studies have shown that the
minimum "effective" [Li]/[Na+K] molar ratio needed in concrete incorporating alkali-silica
reactive aggregates depends on the alkali content, the type of lithium-based admixture, and
the type of reactive aggregate to counteract (degree of reactivity and composition as well).

The authors believe that, currently, the only laboratory test method for reliably determining
the effectiveness of lithium-based admixtures against ASR is the concrete prism test, which
takes 2 years to complete. The AMBT must be modified for testing the effectiveness of
lithium-based admixtures against ASR. The first phase of this study, described in this paper
(Part I), involved a large variety of reactive aggregates from Canada and the U.S.A., the
main objectives were: (1), to evaluate the relative concentration of various ions
135

(OH", Na+, K+, Li+, and silica species) in the soak and mortar pore solutions over time; and
(2), to evaluate the influence of a number of experimental parameters on the expansion in
the "modified" AMBT, e.g. the presence of lithium in the original mortar bars, the Li
concentration in the soak solution, the concentration and the type (NaOH vs KOH) of the
soak solution, the cement alkali content, the water-to-cement ratio, and the initial 24-hr
soaking in pure water. The second part of this study (Part II) compares the AMBT results
with those obtained in the CPT for the variety of aggregates and dosages of lithium, and
proposes a modified AMBT to evaluate the efficacy of lithium-based admixtures against
ASR [25].

3 Materials and methods

3.1 Aggregates, cements, and lithium admixtures

Eleven reactive coarse aggregates of different compositions and degrees of reactivity from
all across Canada were selected for this study, in addition to two reactive aggregates from
the U.S.A. (Table 6.1). These aggregates were used in mortars with 5 different CSA Type
GU (ASTM Type I) cements with alkali contents ranging from 0.25% to 1.25% Na20e
(Table 6.2). A liquid Li-based admixture was used containing 30% LiN03.

3.2 Accelerated mortar bar tests

AMBT's were performed following CSA A23.2-25A (similar to ASTM C 1260), using the
five cements mentioned above. Reagent grade NaOH and KOH chemicals (pellets) were
used for the preparation ofthe mortar bar soak solutions.

For the mortar mixtures with lithium, lithium nitrate was used in the fresh mortar at two
different [Li]/[Na+K] molar ratios, i.e. 0.74 and 1.11 (i.e. 100 and 150% ofthe standard
dosage, respectively), based on the amounts of sodium and potassium in the cement used.
The mortar bars were prepared with a water-to-cement ratio of 0.50, although some
mixtures were also made with ratios of 0.45 and 0.55 to evaluate the effect of this
parameter on expansion. After demolding (at 24 hours), the bars were immersed in water at
136

80°C for another 24 hours, then measured at 80°C for their initial length and immersed in
the test solution at 80°C. Some tests were also made without pre-soaking in water, once
again to evaluate the effect of that parameter on expansion. A IN NaOH soak solution was
generally used, which contained lithium nitrate at different lithium concentrations
([Li]/[Na] ranging from 0 to 0.89). NaOH solutions of different concentrations (0.25 to 4N)
and a IN KOH solution were also tested (Table 6.3).

Length change measurements of the mortar bars were carried out up to 28 days. A total of
180 tests were performed at Laval University corresponding to different sets of mortar
mixtures and soaking conditions (Table 6.3).

3.3 Chemistry of mortar pore and soak solutions

For this series of tests, seven different mixtures containing the same reactive limestone
aggregate (ONI) were selected. The bars were all made with a control high-alkali cement
(CI - Table 6.2), a fixed water-to-cement ratio of 0.50, but with various lithium contents in
the bars and/or the soak solution. The pore solution of the mortar bars was extracted under
high pressure (1000 MPa) at different ages (after demolding, after 24 hours in water, and
after 1, 3, 7, 14, and 28 days in the soak solution).

After expression, the mortar pore solutions and the soak solutions were chemically
analyzed using Inductively Coupled Plasma for Na, K, Li, and Si. The OH" concentration
was determined by direct titration with HC1 down the phenolphthalein end point, and the
pH was calculated (pH = 14 + log [OH"]). For each mixture and each time, the pore solution
extracted and analyzed represents the average composition of the corresponding sampled
bar, but concentration gradients are likely present along the cross section of this bar,
particularly at early ages.
137

4 Results and discussion

4.1 Chemistry of mortar pore and soak solutions

Initially, the mortar pore solution contains many ions, e.g. Na+, K+, Ca2+, OH", and SO4 ",
and the soak solution contains Na and OH" ions, while Li and NO3 " ions can be present in
both solutions, depending whether or not lithium nitrate is added to the mortar and/or to the
soak solution. Concentrations in both solutions tend to equilibrate with time through
diffusion. It is important to note that the volume of soak solution is substantially greater
than the volume ofthe mortar pore solution.

Figure 6.1 shows the mortar expansions and the pH variations with time in the mortar pore
and soak solutions. On the time scale, -1,0, and 1 to 28 days, correspond to the time at
which the bars were demolded, the time at which they were removed from water at 80°C
(24 hours later) and placed in NaOH solution, and the time of immersion in the IN NaOH
solution at 80°C, respectively. The pH of the mortar pore solution after demolding was
between 13.64 and 13.71, with (Figure 6.IB) or without lithium (Figure 6.1A), but it
dropped to about 13.2 after 24 hours of immersion in water. The 24-hr soaking in water
thus significantly decreased the pH in the mortar pore solution. For both systems, i.e. with
and without lithium, the pH rapidly increased to 13.74-13.78 after one day in the IN NaOH
solution and increased slowly afterwards to reach the pH value ofthe soak solution between
14 (mortar without lithium) and 28 days (mortar with lithium). Such an equilibrium
between the mortar pore solution and the soak solution at 14 days was also reported in
another study [26]. One can also observe in Figure 6.1 that the initial NaOH soak solution
has a pH of 14 at 0 day, which is normal for a concentration of IN; however, the pH was
found to decrease slowly with time, to reach between 13.85 and 13.88 at 28 days, likely
because OH" ions have penetrated in the mortar bars while some have been consumed
through the ASR (dissolution of silica and formation of gel). Slightly lower pH values were
measured in the mortar pore solution over the course of the test for the bars that showed
significantly larger expansion.
138

Figure 6.2 shows the evolution of the Na, K, and Li concentrations in the mortar pore and
soak solutions over time and as a function of mortar bar expansion. Because the initial soak
solution does not contain K, the [K] decreased rapidly in the mortar pore solution to reach
very low values after 7 to 14 days, for all mixtures (Figure 6.2A). For its part, the [Na] in
the mortar pore solution exceeded that of the soak solution after approximately 3 days
(Figure 6.2A). After 14 days, the [Na] in the mortar bar solution was higher by about 31%
than that in the soak solution, compared to about 28% in a previous study [26].

When lithium nitrate was present only in the mortar bars (Figure 6.2B), the [Li] in the
mortar pore solution decreased rapidly and reached very low values after 14 days. When
lithium nitrate was added to the soak solution at a low [Li]/[Na] of 0.22 (Figures 6.2C and
6.2D), the [Li] in the mortar pore solution increased rapidly during the first few days
(following the first 24 hours of immersion in water) but never reached the concentration of
the soak solution, stabilizing at about 0.10 to 0.12 mol/L compared to about 0.2 mol/L for
the soak solution; this was true whether Li was used in the mortar mix (Figure 6.2D) or not
(Figure 6.2C). This could be explained by the fact that lithium is consumed more rapidly by
the chemical reactions involved (cement hydration, ASR or other reactions) than the rate at
which it penetrates into the bars. When more lithium was added to the soak solution
([Li]/[Na] = 0.44; Figures 6.2E and 6.2F), the expansion was lower and the [Li] in the
mortar pore solution increased constantly to almost reach that of the soak solution at
28 days. It is interesting to note that, after 7 days of testing, the [Li] in the mortar pore
solution was similar (-0,3 mol/1) whether Li has been used in the mortar mixture (Figure
6.2F) or not (Figure 6.2E). Li ions appear to penetrate into the bars more slowly than Na
ions, possibly because of the larger hydrated radius of lithium [27], which reduces its
mobility compared to sodium or potassium ions.

Figure 6.3 shows the silicon concentration in the soak solution. It is clear that when there is
no lithium in the soak solution, the silicon concentration increases significantly with time.
On the other hand, silicon is not detected (or is present only in very small amounts) when
lithium is added to the soak solution, despite the fact that significant expansions are still
obtained for some mixtures. The above observations are true whether Li was used in the
mortar mix (Figure 6.3B) or not (Figure 6.3A). When lithium is used in the soak solution,
139

but not in sufficient quantity to prevent significant expansion, a white product precipitated
in the containers. X-ray diffraction showed that this product is amorphous (not crystallized)
and its chemical analysis gave: 55% Si0 2 , 29% Li 2 0, 4.9% CaO, 1.6% A1203, 0.9% Na 2 0,
0.5% K 2 0, and 7.0% loss on ignition.

4.2 Influence of the concentration and composition of the soak solution

In the study by Oberholster and Davies [23], the influence of the concentration of the
NaOH soak solution on expansion in the AMBT was investigated and the IN solution gave
the largest expansion with the set of aggregates tested. For their part, Fournier and Bérubé
[27] found that the expansion increases when increasing the concentration from 0.5 to
1.5 N in the presence of a set of reactive limestone aggregates from Québec (Canada). In
the present study, it appeared important to know the influence of the concentration of the
soak solution on the effectiveness of lithium nitrate against ASR.

Figure 6.4A shows this influence in the presence of three aggregates covering a range of
reactivity level. For the NM gravel and QC chloritic schist, a pessimum effect is observed
with maximum 14-day expansions being obtained in 0.5 and 2 N NaOH solutions,
respectively. On the other hand, with the ONI reactive siliceous limestone, the expansion
was found to increase regularly with increasing concentration in the soak solution, at least
up to 4 N. This limestone is similar to those tested by Fournier and Bérubé [28], with the
same conclusion. It then appears that the pessimum concentration depends upon the type of
aggregate tested. Moreover, the results indicate that the position ofthe maximum expansion
with the NM gravel (Figure 6.4B) and QC chloritic schist (not illustrated) does not change
with time.

For the two aggregates ONI and NM showing different behaviors in Figure 6.4, the
influence of lithium in the soak solution was investigated using various NaOH
concentrations (Figure 6.5). The results show that, for both aggregates, the lower the NaOH
concentration in the soak solution, the higher the rate of expansion reduction as a function
ofthe [Li]/[Na] ratio in the soak solution (slope of curves in Figure 6.5), and the lower the
minimum ratio required to satisfy the 14-day, 0.10% expansion limit criterion. It is
140

important to note that, for a given [Li]/[Na] in the soak solution, the amount of lithium is
4 times less in the soak solution at 0.25 N NaOH than in the one at IN NaOH, but that the
quantity of aggressive OH" ions is also 4 times less. Despite a same [Li]/[Na] in the soak
solution, the expansion thus vary significantly as a function of the NaOH concentration of
the soak solution, and this effect seems to vary from one aggregate to another.

To determine the influence of the type of alkali hydroxide in the soak solution, two
aggregates were subjected to the AMBT using IN NaOH and KOH soak solutions, with
and without lithium (Figure 6.6). The 14-day expansion ofthe control mixtures was lower
in KOH than in NaOH, by 32% for the ONI limestone and by 35% for the NM gravel. The
effectiveness of lithium follows similar trends with the two solutions with both aggregates.
However, more aggregates need to be tested before definite conclusions can be drawn.

4.3 Influence of the cement alkali content

Fournier and Bérubé [28] found that using different CSA Type GU (former CSA Type 10)
(ASTM Type I) cements induces minor differences in the 14-day expansion developed by
the reactive limestone ONI in the AMBT. These differences were within the precision of
the test and were not clearly related to any ofthe chemical or physical characteristics ofthe
cements tested. On the other hand, Simon and Wathne [29] found that in some cases, the
expansion measured in the AMBT may vary significantly for different cements. The results
of this study showed no correlation between the MgO cement content and the expansion.

In the present study, five different cements were tested in the presence of the limestone
ONI. The results obtained in the absence of lithium (standard AMBT) are shown in
Figure 6.7A. The corresponding coefficient of variation (for 14-day expansions) is 19%,
which is relatively close to the multi-laboratory coefficient of variation of about 15%
reported by Rogers [30], but clearly higher than the value of 7.3% obtained for the 6 control
mixtures made with the same aggregate and the same cement (CI in Table 6.4) and the
within-laboratory coefficient of variations ranging from 5 to 10% reported by various
authors [31]. The cement used thus seems to influence the expansion in the AMBT, in the
absence of lithium. However, the expansion differences obtained in the standard test could
141

not be explained by variations in the cement composition, including the alkali and MgO
contents. On the other hand, Figure 6.7B shows that the lower the cement alkali content, the
lower the minimum [Li]/[Na] required in the soak solution to satisfy the 14-day, 0.10%
expansion limit criterion. This agrees with Mo et al. [32] who found that a smaller
[Li]/[Na+K] ratio is needed when the alkali content is lower in mortar bars subjected to
autoclaving.

The results obtained also agree with Barborak [33] who found that the cement alkali
content affects significantly the 14-day expansion in the AMBT when the [Li]/[Na+K] ratio
is matched between the bars and the soak solution.

Based on the above results, it seems thus preferable to use a cement that falls within a
restricted range of alkali content when testing the effectiveness of lithium nitrate against
ASR using the AMBT.

4.4 Influence of the water-to-cement ratio

AMBT's were made with the ONI limestone using three different water-to-cement ratios
(0.45, 0.50, and 0.55). In the absence of lithium (standard procedure), the lowest
expansions were obtained using a ratio of 0.45, followed by 0.55, then by 0.50. One can see
in Figure 6.8 that, with no lithium in bars, the lower the water-to-cement ratio, the higher
the minimum [Li]/[Na] needed in the soak solution to satisfy the 14-day, 0.10% expansion
limit criterion. This result can be explained by the fact that lithium penetrates less rapidly in
mortar bars with a lower permeability, i.e. made with a lower water-to-cement ratio.
However, considering that all differences observed may fall within the precision ofthe test,
the results suggest that a small variation in the water-to-cement ratio does not influence
significantly the effectiveness of lithium, while a higher variation may affect the results.

It seems thus preferable to use a constant water-to-cement ratio when testing the
effectiveness of lithium nitrate against ASR using the AMBT.
142

4.5 Influence of the initial 24-hr soaking in water

It has been mentioned above that the initial 24-hr soaking in water significantly decreased
the pH in the mortar pore solution. To determine the influence of this initial soaking on
expansion, seven mixtures were tested, with and without soaking. The results in Table 6.5
show that the 24-hour water soaking resulted in increased 14-day expansions for all three
control mixtures subjected to the standard procedure (i.e. no lithium in bars neither in the
soak solution); however, it always decreased the expansion in the presence of lithium in the
soak solution (with or without lithium in bars). More work is needed to explain the above
behavior.

4.6 Influence of the addition of lithium on expansion in the AMBT

4.6.1 Li in the soak solution (no Li in bars)

In this part ofthe present study, all 13 reactive aggregates were subjected to the AMBT
with no lithium in bars but using different concentrations of lithium nitrate in the soak
solution ([Li[/[Na] varying from 0 to 0.89) (Table 6.3). The bars were all made with a
control high-alkali cement (CI - Table 6.2) and a fixed water-to-cement ratio of 0.50.

The 14-day expansion results are given in Table 6.3. The results for 9 different aggregates
are shown in Figure 6.9 as a function ofthe [Li[/[Na] molar ratio in the soak solution. The
results indicate that the 14-day expansion decreases almost linearly as a function ofthe Li
concentration in the soak solution, and at about the same rate (slope) for all aggregates
tested.

4.6.2 Li in bars and in the soak solution

Figure 6.10 shows the expansion results obtained after 14 days for mortar bars made with
different aggregates, with or without lithium nitrate, and immersed in IN NaOH soak
solutions of various [Li]/[Na]. The original [Li]/[Na+K] in the bars were 0 and 0.74 (also
143

1.11 for the ONI and NM aggregates; Table 6.3). Once again, the bars were all made with a
control high-alkali cement (CI - Table 6.2) and a fixed water-to-cement ratio of 0.50.

First of all, one can note that the relationship between the mortar expansion and the
[Li]/[Na] in the soak solution, which is quite linear in the absence of lithium in bars (e.g.
Figure 6.9), departs from linearity when lithium is used in bars at a [Li]/[Na+K] of 0.74 or
1.11 (e.g. Figures 6.10A to 6.10E). The presence of lithium in bars actually decreased the
quantity of lithium (i.e. [Li]/[Na]) required in the soak solution to reduce the 14-day
expansion in the AMBT below the common expansion limit (0.10%) for all reactive
aggregates tested, as observed in Figures 6.10A (slight decrease) and 6.10B (significant
decrease).

4.6.3 Li in bars (no Li in the soak solution)

As can be seen in Table 6.3, using lithium in bars only (at a molar ratio of 0.74) could not
reduce mortar bar expansions below the 14-day, 0.10% expansion limit criterion for any of
the aggregates tested; actually, the lowest expansion obtained was 0.203% (Table 6.3).
Figure 6.11 shows that adding lithium at a [Li]/[Na+K] ratio of 0.74 in mortar (without
lithium in the soak solution) actually contributed at decreasing expansion, with respect to
the control mixtures, for about 50% of the aggregates tested; the addition also appeared
more effective (i.e. expansion reductions) for those aggregates with the highest expansions
in the standard AMBT, actually for all those with expansions > 0.35% after 14 days. On the
other hand, the presence of lithium in bars only increased the 14-day expansion for seven
aggregates (i.e. NSl, NS2, NB1, QC, ONI, ALI, AL2) (e.g. Figure 6.11); interestingly, the
above seven aggregates were all reactive aggregates for which LiN03 was less effective in
concrete, as demonstrated in Part II of this study [24]. Such a behavior, which cannot be
explained for the time being, is not observed for the other 6 reactive aggregates tested (i.e.
NB2, NB3, ON2, BC, NM, TX) (e.g. Table 6.3).
144

5 Conclusion

The results obtained in Part I of this study on the use of the AMBT for determining the
effectiveness of lithium nitrate against ASR allow the following conclusions:

When lithium was present in small quantities in the soak solution (e.g.
[Li]/[Na] = 0.22), the Li concentration in the mortar pore solution did not reach
equilibrium with the soak solution, whether L1NO3 was used in mortar bars
([Li]/[Na+K] = 0.74) or not. However, when the [Li]/[Na] was increased to 0.44 in
the soak solution, the Li concentration in the mortar pore solution increased
constantly to reach about the same level as that in the soak solution after 28 days;
the fact of adding LiN03 in the mortar bars did not seem to make a significant
difference on the process;

When lithium is added to the soak solution, even at relatively low concentrations
(e.g. [Li]/[Na] - 0.22), negligible or very low concentrations of silica were observed
in this solution;

For identical [Li]/[Na] in the NaOH soak solution, the addition of lithium nitrate in
the solution was found to be relatively more effective at low NaOH
concencentrations, for the two aggregates tested;

The lower the alkali content of the cement used, the lower the minimum [Li]/[Na]
needed in the soak solution to satisfy the 14-day, 0.10% expansion limit criterion
(conclusion based on tests performed on one reactive aggregate only). This would
support the use of cements within a limited range of Na20e for evaluating the
effectiveness of lithium-based admixtures using the AMBT;

• The water-to-cement ratio seems to have a minor influence on the effectiveness of


lithium nitrate in the conditions ofthe AMBT;

• The initial 24-hr soaking of mortar bars in water results in higher expansions in the
absence of lithium (control mixtures), but lower expansions when lithium is used in
the NaOH soak solution (with or without lithium in bars);
145

• The 14-day expansion in the AMBT remained important for all 13 reactive
aggregates tested (> 0.203%), when LiN03 was used in the mortar bars ([Li]/[Na,K]
of 0,74) but not in the soak solution. In the absence of Li in the soak solution, the
addition of LiN03 to the mortar mixture actually increased the 14-day expansion
(compared to the control mixtures) for all reactive aggregates for which LiN03 was
less effective in concrete, as demonstrated in Part II of this study [24];

In the absence of lithium in bars, for all reactive aggregates tested, the 14-day
expansion in the AMBT linearly decreased at a similar rate as a function of the
[Li]/[Na] ratio in the IN NaOH soak solution;

• For all aggregates tested, the use of a [Li]/[Na+K] of 0.74 in the mortar bars
decreased the [Li]/[Na] needed in the IN NaOH soak solution to reach a 14-day
0.10 % expansion limit criterion. However, the relative impact on expansion of
adding lithium to the mortar mixture was different from one aggregate to another.

In the second part of this study [24], the AMBT results are compared to those obtained
from concrete prisms tests CSA A23.2-14A (or ASTM C 1293) performed on the same
reactive aggregates and a modified test procedure is recommended on the use ofthe AMBT
to evaluate the effectiveness of lithium nitrate against ASR.

Lithium nitrate was the only lithium salt investigated in this study; the conclusions thus
obtained may not apply to other lithium salts.

6 Acknowledgements

This study was financially supported by Euclid Admixtures Canada Inc. and the National
Science and Engineering Research Council of Canada (NSERC).
146

7 References

[1] Stanton, T.E. (1940): The expansion of concrete through reaction between
cement and aggregate. Procceedings of the American Society of Civil Engineers,
66: 1759-1786.

[2] McCoy, W.J., Caldwell, A.G. (1951): New approach in inhibiting alkali-aggregate
expansion. ACI Materials Journal, 22 (9): 693-706.

[3] Lawrence, M., Vivian, H.F. (1961): The reactions of various alkalies with silica.
Australian Journal of Applied Science, 12: 96.

[4] Sakaguchi, Y., Takakura, M., Kitagawa, A., Hori, T., Tomosawa, F., Abe, M.
(1989): The inhibiting effect of lithium compounds on ASR. 8th International
Conference on AAR in Concrete, Kyoto, Japan, pp. 229-234.

[5] Ohama, Y., Demura, K., Kakegawa, M. (1989): Inhibiting alkali-aggregate reaction
with chemical admixtures. 8th International Conference on AAR in Concrete, Kyoto,
Japan, pp.253-258.

[6] Collins, CL., Ideker, J.H., Willis, G.S., Kurtis, K.E. (2004): Examination of the
effects of LiOH, LiCL, and LiN0 3 on alkali-silica reaction. Cement and Concrete
Research, 34: 1403-1415.

[7] Stark, D. (1992): Lithium salt admixtures: an alternative method to prevent


expansive alkali-silica reactivity. 9l International Conference on AAR in Concrete,
London, UK, pp. 1017-1025.

[8] Stark, D., Morgan, B., Okamoto, P., Diamond, S. (1993): Eliminating or minimizing
the risk of alkali-silica reactivity. National Research Council, Strategic Highway
Research Program (SHRP), SHRP-C 343, Washington D.C, 266 p.

[9] Lumley, J.S. (1997): ASR suppression by lithium compounds. Cement and Concrete
Research, 27 (2): 235-244.
147

[10] Durand, B. (2000): More results about the use of lithium salts and mineral
admixtures to inhibit ASR in concrete. 11th International Conference on AAR in
Concrete, Québec City, Canada, pp. 623-632.

[11] Diamond, S., Ong, S. (1992): The mechanisms of lithium effects on ASR.
9th International Conference on AAR in Concrete, London, UK, pp. 269-278.

[12] Blackwell, B.Q., Thomas, M.D.A., Sutherland, A. (1997): Use of lithium to control
expansion due to alkali-silica reaction in concrete containing U.K. aggregates.
4th CANMET/ACI International Conference on Concrete Durability, Sydney,
Australia, ACI SP 170, pp. 649-663.

[13] Kawamura, M., Fuwa, H. (2003): Effect of lithium salts on ASR gel composition
and expansion of mortars. Cement and Concrete Research, 33: 913-919.

[14] Fournier, B., Stokes D., Ferro A. (2003): Comparative field and laboratory
investigations on the use of supplementary cementing materials (SCMs) and
lithium-based admixtures to control expansion due to alkali-silica reaction (ASR) in
concrete. 6th CANMET/ACI International Conference on Durability of Concrete,
Thessaloniki, Greece, pp. 823-851.

[15] Lane, D.S. (2002): Laboratory investigation of lithium-bearing compounds for use
in concrete. Virginia Transportation Research Council, VTRC 02-R16.

[16] Stokes, D.B., Manissero, C.E. (2001): Use of lithium-based admixture to control
ASR expansion. 14l Reunion Tecnica Asociacion Argentina de Tecnologia del
Hormigon, Universidad Nacional del Centro de la Provincia de Buenos Aires,
Argentina, 10 p.

[17] Stokes, D.B., Wang, H.H., Diamond, S. (1997): A lithium-based admixture for ASR
control that does not increase the pore solution pH. 5th CANMET/ACI International
Conference on Superplastisizers and Other Chemical Admixtures in Concrete,
Roma, Italy, ACI SP 173, pp. 855-868.
148

[18] Fournier, B., Bérubé, M.A. (2001): Utilisation du lithium pour combattre la réaction
alcalis-silice. ACI Progress in Concrete 2001, Montréal, Canada, 12 p.

[19] Diamond, S. (1999): Unique response of L1NO3 as an alkali silica reaction-


preventive admixture. Cement and Concrete Research, 29 (8): 1271-1275.

[20] Folliard, K.J., Thomas, M.D.A., Fournier, B., Kurtis, K.E. and Ideker J.H. (2006):
Interim Recommendations for the Use of Lithium to Mitigate or Prevent Alkali-
Silica Reaction (ASR). Publication No. FHWA-HRT-06-073, 84 p.

[21] Stokes, D.B., Baxter, S., Thomas, M.D.A., Hill, R. (2003): Test methods for
evaluating the efficacy of chemical admixtures for controlling expansion due to
ASR. 7th CANMET/ACI International Conference on Superplastisizers and Other
Chemical Admixtures in Concrete, Berlin, Germany, pp. 393-408.

[22] Bérubé, M.A., Duchesne, J. (1992): Evaluation of testing methods for assessing the
effectiveness of mineral admixtures in suppressing expansion due to alkali-
aggregate reaction. 4th CANMET/ACI International Conference on Fly Ash, Silica
Fume, Slag and Natural Pozzolans in Concrete, Istanbul, Turkey, ACI SP-132,
pp. 549-575.

[23] Oberholster, R.E., Davies, G. (1985): An accelerated method for testing the
potential alkali reactivity of siliceous aggregates. Cement and Concrete Research,
16: 181-189.

[24] Bérubé, M.A., Tremblay, C , Fournier, B., Thomas, M.D.A., Stokes, D. (2004):
Influence of lithium products proposed for counteracting alkali-silica reactivity on
the chemistry of cement hydrates and pore solution. Cement and Concrete Research,
34: 1645-1660.

[25] Tremblay, C , Bérubé, M.A., Fournier, B., Thomas, M.D.A., Folliard, K.J. (2008):
Use ofthe accelerated mortar bar test to evaluate the effectiveness of LiNÛ3 against
alkali-silica reaction - Part II: Comparison with results from the concrete prism test.
Journal of ASTM Internationnal (JAI), Vol 5, No. 8, September, 21 p.
149

[26] Bérubé M.A., Duchesne J., Chouinard, D. (1995): Why the accelerated mortar bar
method ASTM C 1260 is reliable for evaluating the effectiveness of supplementary
cementing materials in suppressing expansion due to alkali-silica reactivity.
Cement, Concrete and Aggregates, 17 (1): 26-34.

[27] Mary Virginia Orna, Alkali Metals, A SourceBook Module,Version 1.0 1993.
http://chemmovies.unl.edu/chem source pdf/alkali metals.pdf

[28] Fournier, B., Bérubé, M.A. (1991): Application ofthe NBRI accelerated mortar bar
test to siliceous carbonate aggregates produced in the St. Lawrence Lowlands
(Québec, Canada): Part 1: Influence of various parameters on the test results.
Cement and Concrete Research, 21 (5): 853-862.

[29] Simon, M.J., Wathne, L.G. (2000): Effect of cement and added NaOH on measured
expansions in ASTM C 1260 tests. 11th International Conference on AAR in
Concrete, Québec City, Canada, pp. 483-492.

[30] Rogers, CA. (1999): Multi-laboratory study of the accelerated mortar bar test
(ASTM C 1260) for alkali-silica reaction. Cement, Concrete, and aggregates, 21 (2):
185-194.

[31] Fournier, B. & Bérubé, M.A. (2000): Alkali-Aggregate Reaction in Concrete:


A Review of Basic Concepts and Engineering Implications. Canadian Journal of
Civil Engineering, 27(2): 167-191.

[ 32] Mo, X., Yu, C , Xu, Z. (2003): Long-term effectiveness and mechanism of LiOH in
inhibiting alkali-silica reaction. Cement and Concrete Research, 33 (1): 115-119.

[33] Barborak, R.C. (2005): Using lithium compounds as an admixture in new concrete
and as a post-treatment for existing concrete affected by alkali-silica reaction.
M.Sc.Eng., University of Texas, Austin, U.S.A., 129 p.
150

Table 6.1 - Origin, rock types/mineral assemblages, and level of reactivity ofthe
aggregates

Origin Degree of
Identification Rock type Reactive phases/rocks
reactivity3
CANADA
Nova Scotia NSl Greywacke High Microcrystalline quartz
NS2 Granitic Marginal Microcrystalline quartz
New Brunswick NB1 Limestone High Microcrystalline quartz
NB2 Rhyolite High Devitrified glass
NB3 Greywacke/argilite Moderate Microcrystalline quartz
Microcrystalline quartz
Québec QC Chloritic schist Moderate Cryptocrystalline quartz
Ontario ONI (Spratt) Siliceous limestone High Cryptocrystalline quartz
Microcrystalline quartz
ON2 (Sudbury) Polygenic gravel Moderate Quartz cement
Alberta AL1 Polygenic gravel Moderate Microcrystalline quartz
Microcrystalline quartz
AL2 Polygenic gravel Moderate Cryptocrystalline quartz
Micro- to cryptocrystalline
British Colombia BC Polygenic gravel High quartz
U.S.A.
New-Mexico NM Volcanic gravel High Rhyolite, quartz, granite
Texas TX Volcanic sand High Quartz, chert,
Based on the one-year expansion in the Concrete Prism Test CSA A23.2-14A or ASTM C 1293:
<0.040 %: non-reactive; 0.040-0.120 %: moderately reactive; >0.120%: highly reactive.

Table 6.2 - Chemical analysis ofthe cements


Cement CI C2 C3 C4 C5
Oxide (% by mass)
Si0 2 19.70 19.70 20.30 19.74 19.57
CaO 62.70 63.60 62.40 64.79 60.82
A1203 5.28 4.74 4.73 4.56 4.88
Fe 2 0 3 2.34 3.58 3.21 3.45 2.91
MgO 2.79 1.35 3.70 0.91 2.52
so3 3.13 2.77 3.21 2.45 3.32
Na 2 0 0.26 0.20 0.44 0.14 0.27
K20 0.85 0.21 1.24 0.16 0.97
LOI 2.28 3.71 0.90 3.17 2.82
Na2Oe 0.82 0.51 1.25 0.25 0.91
Bogue potential composition (% by mass)
C3S 58 64 54 71 43
C2S 13 8 18 3 28
C3A 10 6 7 6 2
C4AF 7 11 10 11 16
151

Table 6.3 - Experimental conditions and 14-day expansions in the AMBT


Aggregate -> NS1NS2NB1NB2NB3 QC ONI ON2AL1AL2 BC NM TX
Cement -> CI CI CI CI CI CI CI C2 C3 C4 C5 CI CI CI CI CI CI
[Li]/[Na+K] [Li]/[Na]in
in bars soak solution
0.000a .273.197.246 .414 .295.152.330.390 .320 .229.292 .238 .286.224 .375 .966 .656
repeated 2b l b 5b 2^ lb
0.037 .367
0.074 .369
0.111 .349
0 0.148 .186
0.222 .215.103 .097.234.228.227.095 .145.229 .818.468
0.296 .031 .125.144 .187
0.370 .081.062.083 .031 .177
0.444 .069 .030.140.025 .017.127.115 .618.281
0.518 .078 .099 .135.014
0.592 .035 .013.056 .065 .029 .607
0.740 .054
0.888 .043
0.000 .309.208.331 .266.274.205.421 .203.319.275 .362 .728 .441
0.148 .370.127.235 .278.174.477 .158.328.274.327.618.361
0.74 0.296 .197.050.063.034.114.054.258 .024.205.205.072.401.173
0.444 .047 .031 .027.071 .045.099.021.101.027
0.592 .024 .033 .019
0.000 .468 .848_
0.148 .491 .669
1.11
0.296 .345 .503
0.444 .099 .168
Other tests (number) c 1 2 28 19
11 7 8 7 6 9 52 3 3 4 1 6 9 7 7 33 8
Number of tests
181
For each aggregate, the expansion value indicated corresponds to the first control test performed;
For each corresponding aggregate, number of times the (control) test has been repeated;
Mixtures with another soak solution (KOH vs. NaOH), other concentrations in the NaOH soak solution
(0.25 to 4N vs. IN), other water-to-cement ratios (0.45, 0.55 vs. 0.50) in the mortar bars, and with/without
the initial 24-hr soaking in pure water.
152

Table 6.4 - Coefficient of variation ofthe 14-day expansion in the AMBT (control
mixtures and standard procedure with no Li in the soak solution)
Aggregate NSl QCl ONI AL1 NM
greywacke schist limestone gravel gravel
Number of mixtures with the same
3 2 6 3 2
cement C1 (3 bars per mixture)
14-day expansion values 0.273 0.152 0.330 0.286 0.966
0.311 0.138 0.305 0.253 0.920
0.285 0.296 0.232
0.317
0.339
0.277
Average expansion 0.290 0.145 0.311 0.257 0.943
Standard deviation 0.019 0.010 0.023 0.027 0.033
Coefficient of variation (%) 6.7 6.8 7.3 10.6 3.4
Average coefficient of variation (%) 7.3 a
Weighed with respect to the number of test performed with each aggregate.

Table 6.5 - Influence ofthe initial 24-hr soaking in water on the 14-day expansion in the
AMBT
[Li]/[Na+K] 14-day expansion Effect of soaking
Aggregate
in bars in soak soaking in water no soaking on expansion
ONI 0 0 0.330 0.252 +31%
0 0.444 0.140 0.165 -15%
0 0.592 0.056 0.089 -37%
0.74 0.370 0.082 0.125 -34%
NB2 0 0 0.414 0.348 +19%
0 0.296 0.125 0.206 -39%
NSl 0 0 0.311 0.285 +9%
153

m—i—i—r

No Li in bars, [Li]/[Na]=0.44
in soak solution
[Li]/[Na+K]=0.74 in bars
and no Li in soak solution
13.0 -rn—i ! ! 13.0 —
i—!—i—n—i—r^—i—n—i—i—i—i—r-
7 14 21 28 7 14 21 28
Age (days) Age (days)

Figure 6.1 - p H in the mortar pore and soak solutions (initially IN NaOH) as a function of
time in the AMBT, with lithium only in the soak solution (A) or in bars
(B) (ONI reactive limestone aggregate).
154

[Li]/[Na+K]=0.74 in bars
and no Li in soak solution
S OJ-
No Li in bars and
no Li in soak solution 0.2 \ Mortar pore solution

o
K in mortar pore solution o
0.1
Soak solution

r i i i i i—pr
14 21 28 14 21 28
Age (days) Age (days)

No Li in bar and [Li]/[Na]=0.22 [Li]/[Na+K]=0.74 in bars and


in soak solution 2" 0.4 [Li]/[Na]=0.22 in soak solution

14
Age (days)

Figure 6.2 - Na, K, and Li concentrations in the mortar pore and soak solutions (initially
IN NaOH) as a function of time in the AMBT, with (B and D) and without
(A and C) lithium in bars with different Li concentrations in the soak
solution (ONI reactive limestone aggregate).
155

No Li in bars and [Li]/[Na+K]=0.74 in bars and


[Li]/[Na]=0.44 in soak solution [Li]/[Na]=0.44 in soak solution

14 21 28 7 14 21 28
Age (days) Age (days)

Figure 6.2 (suite) - Na, K, and Li concentrations in the mortar pore and soak solutions
(initially IN NaOH) as a function of time in the AMBT, with (F) and
without (E) lithium in bars with different Li concentrations in the
soak solution (ONI reactive limestone aggregate).

B
40
[Li]/[Na+K]=0.74 in bars

7 14 14
Age (days) Age (days)

Figure 6.3 - Si concentration in the soak solution without (A) and with (B) Li in bars
(ONI reactive limestone aggregate).
156

T
0 1 2 3 4 0 0.5 1 1.5 2 2.5 3
Concentration of the soak solution (Mol/L of NaOH) Concentration of the soak solution (Mol/L of NaOH)

Figure 6.4 - Influence ofthe concentration ofthe NaOH soak solution on expansion in the
AMBT (no Li in bars neither in the soak solution).

i—'—r
0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5
Ratio [Li]/[Na] in soak solution Ratio [Li]/[Na] in soak solution

Figure 6.5 - Influence ofthe [Li]/[Na] and concentration in the NaOH soak solution on
expansion in the AMBT (no Li in bars).
157

0.0 0.1 0.2 0.3 0.4 0.5


■Ratio [Li]/[Na or K] in soak solution

Figure 6.6 ­ Influence of the composition of the soak solution (NaOH or KOH) on
expansion in the AMBT for two aggregates (ONI and NM) (no Li in bars).

ON1 limestone •/ ^ C 2 0.51%Na,O.

«C1 0.82%Na2O.
x^»C31.25%Na,0.
0.3­ ^ / * C 5 0.91%Na_Or
Expansion (%

^ • C 4 0.25%Na­O.
ro
o

0.1 ­

I | I I I I I I
n ' i ' r
7 14 21 0.0 0.1 0.2 0.3 04 0.5
Age (days) Ratio [Li]/[Na] in soak solution

Figure 6.7 ­ Influence of the cement alkali content on expansion in the AMBT (no Li in
bars). (A) Expansion with time in absence of Li in the soak solution (standard
procedure). (B) 14­day expansion as a function ofthe [Li]/[Na] in the soak
solution.
158

0.5 ~~l ' I ' 1 ' I ' 1 —


ON1 limestone
with different W/C ratios
0.4
■ ■ ■ VWO0.45
• • • W/O0.50
♦ ♦ ♦ WO0.55

0.1 0.2 0.3 0.4 0.5 0.6


Ratio [Li]/[Na] in soak solution

Figure 6.8 ­ Influence ofthe water­to­cement ratio on expansion in the AMBT as a


function ofthe [Li]/[Na] in the soak solution (no Li in bars).

B
­ 1 —i—'—r . ' I ' I ' I ' I ' I i

No Li in bars 0.9 No Li in bars ­

55
0.8 ­ ^£\
**"fc\ ­
r 0.7­ ­
>. 0 . 6 ­
■S ­
­­.^^

^15 0.5­ ­_
8(A o . 4 :<
2. 0 . 3 ­ . ­
x
111 *
^ ­ ^ 3 j _^^s_
0.2­
<5_w ^~~~­­^^
* ' ^
0.1 ­ ?ïte_7rr^^*r^r­^
­i | i—[— ' I ■ i < i ■ i ■ i
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.1 0.2 03 0.4 0.5 0.6
Ratio [Li]/[Na] in soak solution Ratio [Li]/[Na] in soak solution

Figure 6.9 ­ 14­day expansion in the AMBT for different aggregates as a function ofthe
[Li]/[Na] in the soak solution.
159

1.0
i—'—i '—i—' i ' r
ON1 Limestone
0.8

[LiJ/[Na+K] = 0.74 in bars

[Li]/[Na+K] = 1.11 in bars

-I | I | ! | I— ' 1 r- i—'—i ' i ' i ' r


0 0.1 0.2 03 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
Ratio [Li]/[Na] i n soak s o l u t i o n Ratio [Li]/[Na] i n soak s o l u t i o n

0.5 "■ I ' I ' I ' I


[Li]/[Na+K]=0.74 in bars

^ r T
0.2 03 0.4 0.5 0.6 0.1 0.2 0.3 0.4 0.5
■Ratio [Li]/[Na] i n soak s o l u t i o n Ratio [Li]/[Na] i n soak s o l u t i o n

Figure 6.10- 14-day expansion in the AM BT as a function ofthe [Li]/[Na] ratio in the
soak solution with or without lithium in bars. (A) ONI. (B) NM .
(C) NSl, NB3, ON2, and TX. (D) NB1, NB2, BC, and AL2.
160

n ' r
0.1 0.2 0.3 0.4 0.5
Ratio [Li]/[Na] in soak solution

Figure 6.10 (suite) - 14-day expansion in the AMBT as a function ofthe [Li]/[Na] ratio in
the soak solution with or without lithium in bars. (E) QC, AL1, and
NS2.

«.o n
I <-
g No lithium
B
S
in soak solution
£ 20.0

O)
c

O 0.0
c
.2
g
Si
X
« -20.0
§

-40.0
1.0 0.2 0.4 0.6 0.8 1.0
14-day expans on in the control test (%)

Figure 6.11 - Effect on the 14-day expansion in the AMBT of using a [Li]/[Na+K] of 0.74
in bars, as a function of aggregate expansivity (i.e. 14-day expansion ofthe
control mixtures subjected to the standard procedure).
161

PARTIE II

6.4 Résumé de la partie II

Jusqu'à maintenant, l'essai sur prismes de béton (CPT) CSA A23.2-14A ou ASTM C 1293
demeure l'essai le plus fiable pour évaluer l'efficacité du nitrate de lithium contre la
réaction alcalis-silice (RAS). Toutefois, la prolongation de la période d'essai à deux ans
limite son utilisation dans la pratique. Dans sa forme actuelle, l'essai accéléré sur barres de
mortier (AMBT) CSA A23.2-25A ou ASTM C 1260, quoique fort populaire par sa rapidité,
ne permet pas d'évaluer cette efficacité. Pour qu'il en soit ainsi, il doit être modifié de
manière à augmenter sa fiabilité. La première partie de cette étude (Part I), qui implique
une grande variété de granulats réactifs du Canada et des États-Unis, visait à évaluer
l'influence d'un certain nombre de paramètres sur l'expansion des barres de mortier. Cette
seconde partie (Part II) compare les résultats obtenus avec diverses modifications de
l'AMBT avec ceux obtenus pour les mêmes granulats avec le CPT, l'objectif étant de
proposer des modifications à l'essai accéléré afin de déterminer la quantité minimale de
nitrate de lithium nécessaire pour empêcher l'expansion due à la réaction alcalis-silice dans
le béton. Les résultats obtenus montrent que l'efficacité du nitrate de lithium varie
grandement d'un granulat réactif à un autre et ne peut-être reliée au degré de réactivité ou à
la nature pétrographique des granulats. Une méthode sécuritaire pour prédire le ratio
[Li]/[Na+K] efficace à utiliser dans le béton est proposée. Cette méthode implique deux
essais AMBT dont un mélange où le nitrate de lithium est incorporé à la fois dans les barres
de mortier et dans la solution de trempage, et un autre mélange témoin sans lithium. Cette
méthode permet de déterminer un ratio [Li]/[Na+K] efficace pour les granulats qui
répondent relativement bien au lithium. Elle permet aussi d'identifier les granulats avec
lesquels le lithium performe moins bien et pour lesquels l'essai sur prismes de béton est
recommandé pour évaluer le dosage minimum en lithium à utiliser pour contrôler la RAS.
162

6.5 Publication scientifique no. 4

Use of the accelerated mortar bar test to evaluate the


effectiveness of IJNO3 against alkali-silica reaction
Part 2: Comparison with results from the concrete prism test

C. Tremblay", M.A. Bérubé b, B. Fournier 0 , M.D. Thomas d , K.J. Folliard e and P. C.


N kinamu ban/i

* PhD student, Département de géologie et de génie géologique, Université Laval, Québec, QC, Canada, G1K 7P4
b
Associate Professor, Département de géologie et de génie géologique, Université Laval, Québec, QC, Canada, G1K 7P4
c
Assistant Professor, Département de géologie et de génie géologique, Université Laval, Québec, QC, Canada, GI K 7P4
d
Professor, Department of Civil Engineering, University of New Brunswick, Fredericton, NB, Canada, E3B 5A3
d
Professor, Department of Civil Engineering, University of Texas, Austin, TX 78712
f
Research Scientist, CANMET Mining and Minerai Sciences Laboratories, Natural Resources Canada, 3484 Limebank
Road,Ottawa, ON, Canada Kl A 0E4.

ABSTRACT

For the time being, the concrete prism test (CPT) CSA A23.2-14A or ASTM C 1293
remains the most reliable test method to evaluate the effectiveness of lithium nitrate against
alkali-silica reaction (ASR) in concrete; however, the extended testing period of two years
has often limited its acceptance by practitioners. In its actual form, the popular accelerated
mortar bar test (AMBT) CSA A23.2-25A or ASTM C 1260 cannot be used to predict this
effectiveness, thus needs to be modified accordingly. Part I of this study looked at the
influence of a number of parameters on the effectiveness of lithium to control expansion of
mortar bars incorporating a variety of reactive aggregates from Canada and the U.S.A. The
second part of this study (Part II) compares the results obtained in modified versions ofthe
AMBT with those from the CPT performed on the same aggregates, with the objective of
proposing the best accelerated test procedure for determining the minimum amount of
lithium nitrate necessary to counteract ASR expansion in concrete. The results obtained in
this study have shown that the effectiveness of lithium nitrate greatly varies from one
reactive aggregate to another while not being correlated with the degree of reactivity nor
the pétrographie nature of the reactive aggregates to counteract. A safe method of
predicting the effective [Li]/[Na+K] to used in concrete is proposed, which uses two
AMBT's, one of which involves adding lithium to both the mortar bar and the soak
solution. It allows the prediction of an effective [Li]/[Na+K] for aggregates that respond
relatively well to lithium. The method allows the identification of aggregates that respond
particularly badly to the lithium, for which the concrete prism test is recommended for
evaluating the minimum lithium dosage to use for ASR control.

Keywords: Accelerated mortar bar test, alkali-aggregate reaction, lithium nitrate, concrete
prism test
163

1 Introduction

The most popular accelerated test to evaluate the potential alkali-silica reactivity of
aggregates in concrete is the Accelerated Mortar Bar Test (AMBT) ASTM C 1260 or
CSA A23.2-25A, first proposed in the mid 1980's by Oberholster and Davies [1]. This test,
which involves immersion of mortar bars in a IN NaOH solution at 80°C, became very
popular mostly because it can provide an assessment of the potential alkali-silica reactivity
of concrete aggregates in only two weeks. It can also be used for determining the
effectiveness of supplementary cementing materials (SCM) against ASR, in
accordance with CSA A23.2-28A or ASTM CI 567.

As mentioned in the first part of this study [2], the AMBT is not appropriate in its current
form for evaluating the effectiveness of lithium-based products against ASR. The main
reason is that a large proportion of the lithium ions originally added to the mortar mixture
under test actually diffuses out ofthe bars into the IN NaOH soak solution. The AMBT
thus need to be modified accordingly. Some researchers like Stark et al. [3] proposed
adding lithium nitrate to the soak solution to keep some sort of chemical balance in the
system. Based on preliminary industry recommendations, Folliard et al. [4] first suggested
to use the same [Li]/[Na+K] molar ratio in the mortar bars and the soak solution; however,
based on the results of additional laboratory investigations, the authors revised their
recommendations [5] recognizing that, along with Stokes et al. [6] and Barborak [7], such a
modification overestimates the effectiveness of lithium nitrate compared with expansion
results obtained in the Concrete Prism Test (CPT). The reason for the "amplified"
beneficial effect of lithium when used in such amounts in the soak solution is explained by
the fact that the actual [Li]/[Na+K] prevailing after cement hydration in the mortar/concrete
pore solution is less than 50% of the original design ratio, i.e. between 41 and 47%
according to Tremblay et al. [9]. Consequently, using a [Li]/[Na] of 0.74 in the soak
solution would be in a sense like using a [Li]/[Na+K] "design" ratio of about 1.5 (i.e. 0.74 x
2) or about 200% the standard dosage recommended by the manufacturer in the concrete.
The fact of controlling mortar bar expansion below the expansion limit (e.g. 0.10%) at 14
or 28 days when using a [Li]/[Na] standard dosage of 0.74 in the soak solution would
consequently not be an indication that such a dosage would be effective in concrete; this
164

could indeed result in the acceptance of many reactive aggregates which fail in the CPT
using the same [Li]/[Na+K] design ratio.

Based on the above observations, it was suggested that, in order to reach equilibrium, the
appropriate lithium concentration to use in the AMBT soak solution could be about half of
the original "design" ratio in the mortar bars. Using a [Li]/[Na+K] in the soak solution of
50% of the design ratio in the mortar bars, Barborak [7] indeed obtained a good correlation
with the concrete prism expansion data (ASTM C 1293 or CSA A23.2-14A) for 3 out ofthe
5 aggregates tested, while 35% of the design ratio, in the soak solution, correlated better
with the CPT expansion data for the last two aggregates. As pointed out in the literature and
illustrated in Part I of this study, reactive aggregates were often found to react differently as
a function of the concentration of the soak solution in the AMBT. Furthermore, Borborak
[7] observed that the response of different aggregates to lithium in the AMBT conditions
was in line with the concentration of the soak solution at which maximum mortar bar
expansion (without lithium) was observed. The authors thus proposed the following two-
stage approach for evaluating the efficacy of lithium using the AMBT:

1. Test the aggregates in accelerated mortar bar tests at different NaOH normalities
with 0.25N increments (no lithium used). This allows separating two classes of
aggregates, i.e. those for which expansion peaks under and above IN NaOH;

2. The safe concentration of lithium to use in concrete could then be found by running
AMBT using a IN NaOH soak solution in which the lithium dosage would be either
50 or 35% ofthe design molar ratio, whether the expansion ofthe aggregate under
test peaks below or above IN NaOH (determined in stage 1), respectively.

This method seems to be valid for the 5 aggregates tested in that study [7].

The effect of a number of experimental parameters on the expansion of mortar bars,


with/without lithium nitrate in the bars and/or in the soak solution, were investigated in the
first part of this study [2]. The parameters investigated in the presence of a variety of
reactive aggregates from Canada and the U.S.A., were the following: [Li]-to-[Na] molar
ratios in the soak solution (0 to 0.89), [Li]-to-[Na+K] molar ratios in the mortar bars
165

(0, 0.74, 1.11), normality (0 to 4N) and nature (NaOH, KOH) of the soak solution,
cement alkali content (5 cements from 0.25 to 1.25% Na20e), water-to-cement ratio
(0.45, 0.50, 0.55), and initial 24-hr soaking in pure water or not (as normally used in the
ASTM C 1260 test). It was found that the effectiveness of lithium nitrate against ASR
expansion in the AMBT was mostly affected by the quantity of lithium in the soak solution,
the presence or not of lithium in the mortar bars, and the alkali content of the cement (in
order of decreasing importance).

2 Research significance

The authors believe that the only existing test method that can reliably evaluate the
effectiveness of lithium-based admixtures in controlling expansion due to ASR with natural
reactive aggregates is the CPT, which takes 2 years. It was demonstrated that the results
obtained in the CPT are generally well-correlated with field performance [11];
consequently, it appears critical that, in the absence of long-term field performance data,
the results obtained in an accelereted test are well-correlated with those obtained in the
CPT. In order to establish conditions that would allow using the AMBT for evaluating the
efficacy of lithium against ASR, a number of modifications to the AMBT have been
investigated in the first part of this study [2], which involved 14 reactive aggregates from
Canada and the U.S.A. with different compositions and degrees of reactivity. The main
objective of this second part of the study was, based on the comparison between AMBT
and CPT results, to propose modifications to the conventional AMBT for more adequately
determining the effectiveness of lithium nitrate against ASR. Such modifications could then
be evaluated for potential standardization (e.g. ASTM, CSA, AASHTO) when a sufficient
number of aggregates would have been tested using the approach(es) proposed.
166

3 Materials and methods

3.1 Concrete prism tests performed at Laval University

As part of an extensive study to evaluate the effectiveness of lithium-based products


(nitrate and glass), in combination or not with supplementary cementing materials [9], to
control expansion due to ASR, concrete prisms were made with a non-reactive sand, eleven
Canadian coarse reactive aggregates of various compositions and degrees of reactivity
(Table 6.6), a CSA Type GU (or ASTM Type 1) cement containing 0.82% Na2Oe (cement
CI in Table 6.2 in [2]), then subjected to the CPT. The "lithium" mixtures contained
lithium nitrate at [Li]-to-[Na+K] molar ratios ranging from 0.37 to 1.11, which correspond
to 50 to 150% of the standard dosage (0.74) recommended by the manufacturer [4,5]
(Table 6.7). The effectiveness of lithium nitrate against ASR was determined in accordance
with Standard Practice CSA A23.2-28A, i.e. using a 2-year, 0.040% expansion limit
criterion for acceptance.

3.2 Accelerated mortar bar tests performed at Laval University

Materials and methods pertinent to the AMBT's are described in Part I [2]. These tests were
performed on the same eleven reactive aggregates from Canada plus two highly- reactive
aggregates from the U.S.A. (Table 6.6: NM and TX). In accordance with CSA A23.2-25A
or ASTM C 1260, the aggregates were used to make mortar bars with the same cement
mentioned before (0.82% Na2Ûe). A liquid Li-based admixture containing 30% LiN03 was
either used or not (control mixtures) to make the mortar bars, corresponding to different
[Li]-to-[Na+K] design molar ratios (0, 0.74, and 1.11). The mortar bars were immersed and
their length change monitored up to 28 days in IN NaOH soak solutions containing various
concentrations of lithium nitrate ([Li]-to-[Na] molar ratios ranging from 0 to 0.89). The
results from 110 (Table 6.8) out ofthe 180 AMBT's presented in Part I (see Table 6.3 in
[2]) and corresponding to different sets of mortar mixtures and soaking conditions are
considered hereafter for comparison with the CPT results.
167

3.3 Accelerated mortar bar and concrete prism tests performed in other
laboratories

All CPT's made with the three reactive aggregates from the U.S.A. (i.e. NM, TX, and WR)
(Table 6.7) and all AMBT's made with the aggregate WR (Table 6.8) were performed at
the University of Texas, Austin (U.S.A.). For comparison purposes, AMBT results obtained
at CANMET-MTL (Ottawa, Canada) for aggregates NSl, NB1, NB3,QC, ONI, ON2, AL1,
AL2, BC and TX (unpublished data) and at the University of Texas for aggregates ONI,
ON2, NM, and TX [7] are also considered hereafter (Table 6.8).

4 Results and discussion

4.1 CPT results

The CPT results are shown in Table 6.7 (expansions at one year for control mixtures and at
two years for mixtures incorporating lithium nitrate). The last column in Table 6.7 gives,
for each aggregate tested, the minimum [Li]-to-[Na+K] design molar ratio needed to satisfy
the 2-year, 0.040% expansion limit criterion used in the CPT when testing preventive
measures against ASR, according to CSA A23.2-28A. This minimum ratio is obtained by
interpolation between two tested [Li]/[Na+K], unless the highest ratio tested did not satisfy
the above criterion (sign > in Table 6.7). As clearly shown in Table 6.7 and Figure 6.12,
this minimum ratio cannot be related to the degree of reactivity ofthe aggregates tested, i.e.
to the 1-year expansion ofthe control concrete mixture; for instance, a low ratio of 0.55
seems efficient in the presence of the extremely-reactive TX sand, while a much higher
ratio in excess of 0.93 is required with the moderately-reactive chloritic schist QC. Also,
the effectiveness of lithium nitrate (in the CPT) cannot be related to the pétrographie nature
ofthe reactive aggregate under test.
168

4.2 AMBT results

The AMBT results are presented in Tables 6.8 (14 days) and 6.9 (28 days). The results
obtained at Laval University have already been discussed, to some extent, in Part I [2].
Results obtained at CANMET-MTL and at the University of Texas, for all [Li]/[Na] greater
than 0 in the soak solution, have been obtained by linear interpolation between two ratios
tested in these laboratories.

For the benefit ofthe following discussion, Table 6.10 gives, with respect to the control
tests (no Li in bars nor in the soak solution), the reductions (or increases) in the AMBT
expansion after 14 and 28 days for each important series of tests corresponding to
the following [Li]/[Na+K] in the bars and [Li]/[Na] in the IN NaOH soak solution,
respectively: i.e. 0/0.222, 0/0.444, 0/0.592, 0.74/0, 0.74/0.148, 0.74/0.296, and 0.74/0.444.

Table 6.11 gives the minimum [Li]/[Na] needed in the IN NaOH soak solution to satisfy a
0.10% expansion limit criterion in the AMBT after 14 and 28 days, in the absence
([Li]/[Na+K] = 0) or in the presence ([Li]/[Na+K] = 0.74) of lithium in the bars. For each
aggregate and each ratio in bars (i.e. 0 and 0.74), the minimum [Li]/[Na] needed in the soak
solution at 14 (or 28) days is obtained by extrapolation on a plot of expansion at 14 (or 28)
days against the [Li]/[Na] in the soak solution, as illustrated in Figure 6.13 for a
[Li]/[Na+K] of 0.74 in bars. Table 6.11 also gives the ratio between the minimum
[Li]/[Na+K] needed in the CPT and the minimum [Li]/[Na] needed in the AMBT soak
solution (with and without lithium in bars).

4.3 Methods of comparison between CPT and AMBT results

In the following, the CPT results (1-year expansions of control mixtures and minimum
[Li]/[Na+K] needed to satisfy the 2-year, 0.040% expansion limit criterion) will be
compared to the following results obtained in the AMBT after 14 and 28 days:

1. Expansions for the control tests (i.e. no Li in bars nor in the soak solution);

2. Expansions with lithium in bars and/or in the soak solution (various concentrations);
169

3. Minimum [Li]/[Na] needed in the soak solution to satisfy a 0.10% expansion limit
criterion after 14 or 28 days (with or without lithium in bars), and;

4. Expansion reductions/increases when using lithium in bars and/or in the soak


solution (various concentrations).

For the benefit ofthe following discussion, Table 6.12 gives the coefficients of linear
regression (R2) obtained between the above parameters for the AMBT's performed by the
three participating laboratories. Except for the control tests, the 28-day results usually show
similar or better correlations than the 14-day results; most figures presented hereafter will
thus be drawn using the 28-day results.

4.4 Comparison based on expansion results in the AMBT after 14 and 28 days

4.4.1 Control tests (no lithium in bars nor in the soak solution)

As shown in Figure 6.14, for the control mortar mixtures (i.e. no Li in bars) subjected to the
standard AMBT procedure (i.e. no Li in the IN NaOH soak solution), the 14-day expansion
in the AMBT does not correlate well with the 1-year expansion in the CPT for the control
mixtures. Moreover, Figure 6.15 shows that there is no correlation between the minimum
[Li]-to-[Na+K] design molar ratio needed to satisfy the 0.040%, 2-year expansion limit
criterion in the CPT (Table 6.7), and the 14-day expansion ofthe control mixtures in the
AMBT.

4.4.2 Tests with lithium in bars and/or in the soak solution

Figures 6.16 (no Li in bars) and 6.17 ([Li]/[Na+K] = 0.74 in bars) show the correlations
between the minimum [Li]/[Na+K] needed in the CPT and the 28-day expansion in the
AMBT for two [Li]/[Na+K] in bars (i.e. 0 or 0.74) and a range of [Li]/[Na] in the soak
solution (i.e. from 0 up to to 0.592). All correlations are very poor, with linear regression
coefficients between 0.00 and 0.20 (Table 6.12).
170

The "best" correlation of that set (R2 = 0.20) is obtained at 28 days when using a
[Li]/[Na+K] of 0.74 in bars and a [Li]/[Na] of 0.296 in the soak solution (Figure 6.17C). It
is interesting to note that such a concentration in the soak solution is close to that measured
at early age in the pore solution of concrete prisms incorporating a standard dosage
(0.74 molar ratio) of lithium nitrate [9]. Based on the aggregates tested in this study, Figure
6.17C suggests that using the above test conditions in the AMBT, an expansion of less than
0.15% after 28 days would include three reactive aggregates that could be counteracted in
concrete when using the standard [Li]/[Na+K] dosage of 0.74 (i.e. lower left quadrant in
Figure 6.17C). However, such a 0.74 molar ratio is also sufficient in concrete prisms
incorporating four other reactive aggregates but that expand more than 0.20% after 28 days
(including WR) (lower right quadrant in Figure 6.17C), even almost sufficient for the
extremely-reactive NM gravel (needed ratio of 0.77 close to 0.74), which causes an AMBT
expansion over 0.80%. The above criterion (0.15% at 28 days) would thus be severe for
many aggregates. Consequently, considering the expansion results obtained in the AMBT
at 28 days (Figures 6.16 and 6.17) and at 14 days as well (not illustrated), using lithium-to-
alkali molar ratios of 0.74 in bars and 0.296 in the soak solution, although promising, does
not appear fully adequate for determining the minimum [Li]/[Na+K] needed in concrete in
the presence of reactive aggregates.

4.5 Comparison based on the minimum [Li]/[Na] needed in the soak solution to
satisfy a 0.10% expansion limit criterion in the AMBT after 14 or 28 days

4.5.1 Tests in the absence of lithium in bars

As part of this study, it was found that for most aggregates tested, the use of increasing
dosages of lithium nitrate in the soak solution (only) resulted in almost linear reductions of
the 14-day expansion ofthe mortar bars and that the rate of expansion reduction was fairly
similar for all aggregates tested. Figures 6.18A and 6.18B show that, in the absence of
lithium in bars, there is a very poor correlation between the [Li]/[Na] ratio needed in the
soak solution to satisfy the 0.10% expansion limit criterion in the AMBT and the minimum
[Li]/[Na+K] needed in the CPT (Table 6.7) (Table 6.12: R2 = 0.00 and 0.02, at 14 and 28
days, respectively). According to these results, the approach of using lithium in the AMBT
171

soak solution only, does not seem to be appropriate for determining the minimum dosage of
lithium nitrate in the presence of reactive aggregates.

4.5.2 Tests with a [Li]/[Na+K] ofO. 74 in bars

When lithium is used in bars at a 0.74 lithium-to-alkali molar ratio, Figure 6.13 shows that
the minimum [Li]/[Na] needed in the soak solution to satisfy the 0.10% expansion limit
criterion after 14 or 28 days of testing varies from one aggregate to another. Figures 6.18C
and 6.18D present the minimum [Li]/[Na+K] needed in the CPT plotted against the above
[Li]/[Na] minima in the soak solution. The correlations are slightly better than in the
absence of lithium in bars (Figures 6.18A and 6.18B), particularly at 28 days, but remain
poor with linear regression coefficients R2 of only 0.19 and 0.25 after 14 and 28 days,
respectively (Table 6.12). Based on the aggregates tested in this study, Figure 6.18D
suggests that, when using a [Li]/[Na+K] of 0.74 in the bars, a reactive aggregate requiring a
[Li]-to-[Na] molar ratio of less than 0.35 in the soak solution to control the mortar bar
expansion to < 0.10% at 28 days (3 aggregates in this study) could be counteracted in
concrete when using the standard [Li]/[Na+K] dosage of 0.74 (i.e. lower left quadrant in
Figure 6.18D). However, Figure 6.18D also shows that 4 other aggregates that can be
controlled using a 0.74 molar ratio in the CPT need a minimum [Li]/[Na] in the soak
solution of over 0.35 to satisfy the 28-day criterion (i.e. lower right quadrant in Figure
6.18D), thus indicating that the above testing approach (i.e. 0.74 molar ratio in bars and
0.35 molar ratio in the soak solution) would be severe for the above aggregates.

4.6 Comparison based on expansion reduction in the AMBT after 14 and 28 days

4.6.1 Tests in the absence of lithium in bars

Table 6.10 and Figure 6.19 show that, in the absence of lithium in bars, using lithium in the
soak solution at a [Li]/[Na] of 0.222 (lowest ratio tested) or more, always results in
expansion reduction after 14 and 28 days with respect to the control tests (no Li in bars
neither in the soak solution), except in one case (NSl at 28 days tested at CANMET using a
172

[Li]/[Na] of 0.222 in the soak solution - Figure 6.19A). However, the correlations between
the minimum [Li]/[Na+K] in the CPT and the expansion reduction in the AMBT remains
•y

poor, with linear regression coefficients R varying between 0.01 and 0.12 (Table 6.12).
None of the corresponding relationships can be used to evaluate the minimum [Li]/[Na+K]
needed in concrete to counteract ASR.

4.6.2 Tests with a [Li]/[Na +K] ofO. 74 in bars

Table 6.10 and Figure 6.20 show the effect on the AMBT expansion (in terms of reductions
or increases at 14 and 28 days) when lithium nitrate is added in bars at a [Li]/[Na+K] of
0.74, without (Figures 6.20A and 6.20B) or with (Figures 6.20C to 6.20F) lithium in the
soak solution. The correlations are generally much better than all those reported so far (R
from 0.35 to 0.79 - Table 6.12). The best correlations are obtained when using a [Li]/[Na]
of 0.148 (i.e. 20% ofthe standard dosage) in the soak solution (R2 = 0.73 and 0.79 at 14 and
28 days, respectively; Table 6.12 and Figures 6.20C and 6.20D).

Tests in the absence of lithium in the soak solution - As already mentioned in Part I [2],
adding lithium nitrate in the bars only, at a molar ratio of 0.74, contributed at decreasing the
14-day expansion (i.e. compared to the control) for about half of the aggregates tested
(Figure 6.20A); such an addition, however, never contributed at reducing expansion under
0.10% after 28 days (Figure 6.17A and Table 6.9: expansion > 0.335%) and even after
14 days (Table 6.8: expansion > 0.203 %). It is interesting to note in Figure 6.20A that the
14-day expansion decreased for 7 out of the 8 reactive aggregates (including WR) for
which a [Li]/[Na+K] of 0.80 or less was effective in concrete (i.e. lower left quadrant in
Figure 6.20A), which is closed to the standard dosage of 0.74 (all except NS2 which
presents a slight expansion increase of 6%). On the other hand, adding lithium in bars
resulted in increased 14-day expansions for the other 6 reactive aggregates that required
higher lithium dosage in the CPT (i.e. upper right quadrant in Figure 6.20A). This
correlation thus holds for 13 out ofthe 14 aggregates tested. Therefore, for a given reactive
aggregate tested in the AMBT, the use of lithium nitrate only in the bars may thus give a
certain idea about the effectiveness of this admixture against ASR in concrete. In other
words, a [Li]/[Na+K] of about 0.80 should be sufficient for controlling deleterious
173

expansion in concrete when the use of lithium nitrate in mortar bars at a [Li]/[Na+K] of
0.74 (with no Li in the soak solution) does not increase the 14-day expansion with respect
to the control mixture subjected to the standard AMBT procedure. However, such a
criterion holds for the set of reactive aggregates tested. Moreover, it remains to explain why
the presence of lithium nitrate in bars (only) increased the expansion in the AMBT for
many of the reactive aggregates tested with respect to the control tests performed in the
absence of lithium.

Tests in the presence of lithium in the soak solution - The higher the [Li]/[Na] in the
soak solution (from 0 up to 0.444), the lower the number of aggregates showing expansion
increases in the AMBT (Figures 6.20A to 6.20F), this number falling to zero with the
highest ratio tested (0.74 : 0.444; Figure 6.20F). As mentioned earlier, the best correlations
of all (R = 0.73 and 0.79 at 14 and 28 days, respectively) were obtained when using a
[Li]/[Na+K] of 0.74 in bars and a [Li]/[Na] of 0.148 in the soak solution (Table 6.12 and
Figures 6.20C and 6.20D). Figure 6.20D indeed shows that the 28-day expansion decreased
by 10% or more for all 8 reactive aggregates (including WR) for which a [Li]/[Na+K] of
0.80 or less was effective in concrete (Table 6.7). In the meantime, the 28-day expansion
decreased by less than 10% or increased for all 6 other reactive aggregates tested that
required a higher lithium dosage in the CPT (upper right quadrant in Figure 6.20D). This
correlation holds for all 14 aggregates tested using a [Li]/[Na] of 0.148 in the soak solution,
which is better than in the absence of lithium in the soak solution (Figure 6.20A). Again,
such a correlation holds for the set of reactive aggregates tested and it remains to explain
why the presence of lithium nitrate in bars and in the soak solution (up to a [Li]/[Na] of
0.296) often increases the expansion in the AMBT with respect to the control tests.

4.7 Tests using a [Li]/[Na] in the soak solution half of the [Li|/|Na+K] in bars

An alternative approach for estimating the minimum lithium content in concrete consists in
performing a series of AMBT's using a range of [Li]/[Na] in the IN NaOH soak solution
but always equal to 50% ofthe [Li]/[Na+K] used in bars [13]. This approach is based on
the fact that about 50% of the lithium added to concrete is incorporated in cement hydrates
174

with about half of the lithium remaining available in the concrete pore solution to
counteract ASR [10]. Using such testing conditions, different 28-day expansion limit
criteria for acceptance were evaluated. Table 6.13 shows the corresponding results obtained
at Laval University, CANMET-MTL, and the University of Texas for the fourteen reactive
aggregates tested in this study.

The regression coefficient R2 obtained for the linear relationship between the minimum
[Li]/[Na+K] needed in the CPT and the minimum ratio needed in the AMBT to satisfy the
above criteria were still found to be poor (i.e. between 0.23 and 0.39). Moreover, Table
6.13 shows that no single limit criterion in the AMBT allows to reproduce the minimum
molar ratio needed in concrete. Also, the method underestimated the [Li]/[Na+K] needed in
the CPT for many aggregates when using an expansion limit criterion of 0.08%, 0.10% or
0.12% at 28 days. Therefore, a decision made on the above experimental approach and an
expansion limit criterion of let's say 0.080% would result in insufficient lithium dosage in
concrete for three ofthe aggregates tested (i.e. NSl, NB1 and QC schist). To be safe for the
majority ofthe aggregates (13/14), an expansion limit criterion of 0.04% at 28 days should
be used. Such an expansion limit criterion ratio appears excessive for many of the reactive
aggregates tested. Despite the fact that this method is based on the results of concrete pore
solution evaluation, the results obtained with different expansion limit criteria at 28 days
are inconclusive. Moreover, this method requires a large number of mixtures and it remains
complex.

4.8 Discussion and recommendations

The results obtained in this study have shown that it is difficult to determine the
specific/minimum [Li]/[Na+K] needed to control ASR expansion in concrete based on a
specific expansion limit criterion (e.g. 0.08 or 0.10%... after 14 or 28 days) in the AMBT
performed using specific amounts of lithium in bars and/or in the soak solution. The
determination of the minimum [Li]/[Na] needed in the AMBT soak solution to satisfy a
specific expansion limit criterion requires many tests in soak solutions of various [Li]/[Na]
175

ratios; also, the relative response in the AMBT conditions varies significantly from one
aggregate to another.

However, the consideration of a criterion based on expansion reduction in AMBT's


incorporating lithium, for instance using a [Li]/[Na+K] of 0.74 in bars and a [Li]/[Na] of
0.148 in the soak solution, was found to provide a better and easier way of determining the
minimum [Li]/[Na+K] needed in concrete incorporating ASR reactive aggregates.
The corresponding correlation is actually slightly better when using the 28-day results
(R2 = 0.79) than the 14-day results (R2 = 0.73) (Figures 6.20D vs. 6.20C). Moreover, it is
possible that the above conditions are not optimized, considering that only two [Li]/[Na+K]
ratios in bars have been tested (i.e. 0 and 0.74) while the best [Li]/[Na] in the soak solution
could be slightly lower or higher than 0.148.

Nevertheless, based on the experimental conditions and reactive aggregates tested in this
study, a procedure (Figures 6.21 and 6.22) is proposed for the determination of the
minimum [Li]/[Na+K] needed in the CPT based on the 28-day results from two AMBT's:

1. A control test (control mixture subjected to the standard ASTM C 1260 test
procedure extended to 28 days), and;

2. A 28-d test performed using a [Li]/[Na+K] of 0.74 in bars and a [Li]/[Na] of 0.148
in the soak solution.

The 28 day expansion for the control is named El and the 28-day expansion for the mixture
with lithium is E2. If ((E2-E1)/E1) is less than 0.1, the equation in Figure 6.21 can be used
to calculate the amount of lithium to use in concrete. If, on the other hand, ((E2-E1)/E1) >
0.1, it means that we are in presence of an aggregate for which "conventional" dosages of
lithium were not found to be effective in controlling ASR expansion and the use of concrete
prism test is recommended for establishing the ratio to use. This approach reduces the risk
of underestimating the lithium dosage, based on the AMBT, in the case of reactive
aggregates that do not respond well to "conventional" dosages in concrete (like QC and
NSl). For example, Feng [14] found that a [Li]/[Na+K]=1.48 is still not sufficient to
control deleterious concrete prism expansion with the NSl aggregate. Larger than
176

"conventional" dosages of lithium are also needed for the aggregates QC and NB1. It is
very important that the proposed method is able to identify such aggregates and to
determine the appropriate method for establishing the minimum lithium dosage for ASR
control, which is what the approach proposed in Figures 6.21, 6.22 and Table 6.14 does.

On the other hand, in the presence of a reactive aggregate that responds well to lithium, the
following equation (1.0 + 0.7 x ((E2-E1)/E1), which is based on the upper limit ofthe
figure 6.20D, can be used to predict the effective ratio to use in concrete.

The approach described above offers a relatively simple, reliable and rapid way of
establishing safe dosages of lithium for ASR control. It requires only two mortar bar tests.
The use of this modified version ofthe accelerated mortar bar test allows to predict lithium-
to-alkali molar ratios of up to 1.07 and the minimum ratio predicted with this method for all
the 14 aggregates tested is 0.68 (NB2 and ON2). This approach is safe for all the
14 reactive aggregates tested in this study. However, it is important that the variability of
the test (repeatability / reproducibility) ) be further evaluated when the AMBT is performed
in the presence of lithium nitrate.

5 Conclusion

This study concerns the use of the Accelerated Mortar Bar Test (AMBT) for determining
the effectiveness of lithium nitrate against ASR in concrete. The results of AMBT's
performed on 14 reactive aggregates from Canada and the U.S.A. using various lithium
dosages in bars and/or in the IN NaOH soak solution have been compared with the results
from Concrete Prisms Tests (CPT) performed on the same aggregates using various lithium
dosages. The following conclusions can be drawn:

• A relatively poor correlation was observed between the potential expansivity of the
aggregates tested in the CPT (i.e. 1-year expansion for the control mixtures) and in
the AMBT (i.e. 14-day expansion for the control tests in the absence of lithium);

• Moreover, there is no correlation between the minimum [Li]/[Na+K] needed in the


CPT and the expansivity (or reactivity level) ofthe aggregates in the AMBT;
177

• The use of lithium nitrate in the mortar bars at a [Li]/[Na] of 0.74, in the absence of
lithium in the soak solution, contributed at increasing the 14-day expansion in the
AMBT for about half of the reactive aggregates tested, most of them requiring a
[Li]/[Na+K] design ratio of 0.74 or more to control ASR expansion in concrete;

• The specific/minimum [Li]/[Na+K] needed in concrete incorporating the reactive


aggregates tested could not be determined based on a specific expansion limit
criterion (e.g. expansion limit criterion between 0.04 or 0.12% after 14 or 28 days)
in AMBT's performed using specific amounts of lithium in the mortar bars and/or
in the IN NaOH soak solution;

However, significantly better correlations were obtained between the minimum


[Li]/[Na+K] needed in the CPT and the amount of expansion reduction in AMBT's
performed using a [Li]/[Na+K] of 0.74 in bars, and a [Li]/[Na] ranging from 0 to
•y

0.30 in the soak solution (R from 0.43 to 0.79). Moreover, the correlations are
usually better at 28 days than at 14 days and the best correlation (R2 = 0.79) was
obtained at 28 days with a [Li]/[Na] of 0.15 in the soak solution. Based on this best
correlation, a relatively simple and safe method could be used to evaluate the
minimum ratio of lithium nitrate to use in concrete to conteract the ASR. This
process stay safe for all the 14 reactive aggregates tested in this study;

• The minimum lithium nitrate dosage in concrete calculated using the equation
provides a safe estimate of the amount of lithium required for all aggregates tested;
however, it also overestimates the minimum required molar ratio for a certain
number of the above aggregates. Moreover, other reactive aggregates, not tested in
this study, could fall outside the field limited by this "safe" relationship, which
should thus be used with circumspection.

Future work is in progress to validate this "modified" version of the AMBT with a larger
number of reactive aggregates and to evaluate the coefficient of experimental variability.
Provided the results are conclusive, the "modified" version could possibly be proposed
for standardization. For the time being, it is our belief that the concrete prism test
178

CSA A23.2-14A or ASTM C 1293 remains the most reliable method for determining the
amount of lithium nitrate needed in concrete incorporating reactive aggregates.

6 Acknowledgements

This study was financially supported by Euclid Admixtures Canada Inc. and the National
Science and Engineering Research Council of Canada (NSERC).

7 References

[1] Oberholster, R.E., Davies, G. (1985): An accelerated method for testing the
potential alkali reactivity of siliceous aggregates. Cement and Concrete Research,
16: 181-189.

[2] Tremblay, C , Bérubé, M.A., Fournier, B., Thomas, M.D.A., Folliard, K.J. (2007):
Use ofthe accelerated mortar bar test to evaluate the effectiveness of LiN03 against
alkali-silica reaction - Part 1 : Pore solution chemistry and influence of various
parameters. Journal of ASTM International,Vol. 5, No. 8, September, 21 p.

[3] Stark, D., Morgan, B., Okamoto, P., Diamond, S. (1993): Eliminating or
minimizing the risk of alkali-silica reactivity. National Research Council, Strategic
Highway Research Program (SHRP), SHRP-C 343, Washington D.C, 266 p.

[4] Folliard, K.J., Thomas, M.D.A., Kurtis, K. (2003): Guidelines for the use of lithium
to mitigate or prevent ASR. Publication No. FHWA-RD-03-047, Federal Highway
Administration, U.S. Department of Transportation, Washington D.C, July 2003.

[5] Folliard, K.J., Thomas, M.D.A., Fournier, B., Kurtis, K. and Ideker, J.H. (2006):
Interim recommendations for the use of lithium to mitigate or prevent Alkali-Silica
Reaction (ASR). Publication No. FHWA-RD-06-073, Federal Highway
Administration, U.S. Department of Transportation, Washington D.C, July 2006.
179

[6] Stokes, D.B., Baxter, S., Thomas, M.D.A., Hill, R. (2003): Test methods for
evaluating the efficacy of chemical admixtures for controlling expansion due to
ASR. Proceeding of the 7th CANMET/ACI International Conference on
Superplastisizers and Other Chemical Admixtures in Concrete, Berlin, Germany,
pp. 393-408.

[7] Barborak, R.C. (2005): Using lithium compounds as an admixture in new concrete
and as a post-treatment for existing concrete affected by alkali-silica reaction.
M.Sc.Eng., University of Texas, Austin, U.S.A., 129 p.

[8] Diamond, S., Ong, S. (1992): The mechanisms of lithium effects on ASR.
Proceeding of the 9th International Conference on AAR in Concrete, London, UK,
pp. 269-278.

[9] Tremblay, C , Bérubé, M.A., Fournier, B., Thomas, M.D.A., Folliard, K.J. (2007):
Effectiveness of lithium-based products in concrete made with Canadian aggregates
susceptible to ASR. ACI Materials Journal, Vol. 104 (2), pp. 195-205.

[10] Bérubé, M.A., Tremblay, C , Fournier, B., Thomas, M.D.A., Stokes, D. (2004):
Influence of lithium products proposed for counteracting alkali-silica reactivity on
the chemistry of cement hydrates and pore solution. Cement and Concrete
Research, 34: 1645-1660.

[11] Fournier, B., Stokes, D.B., Ferro, A., Comparative field and laboratory
investigations on the use of supplementary cementing materials (SCM's) and
lithium-based admixtures to control expansion due to alkali-silica reaction (ASR) in
concrete. Proceedings 6 International CANMET/ACI Conference on Durability of
Concrete, Thessaloniki, Greece, 2003, pp. 823-851.

[12] Durand, B. (2000): More results about the use of lithium salts and mineral
admixtures to inhibit ASR in concrete. Proceeding of the 11 International
Conference on AAR in Concrete, Québec City, Canada, pp. 623-632.
180

[13] CRD-C662-10 (2010): Determining the potential alkali-silica reactivity of


combinations of cementitious materials, lithium nitrate admixture and aggregate
(accelerated mortar bar method), COE, janvier 2010, 6 p.

[14] Feng. X. (2007): Effects and mechanisms of lithium nitrate on controlling alkali-
silica reaction, PH.D. Thesis, University of New Brunswick, Canada, 229 p.
181

Table 6.6 - Origin, rock type, and reactivity ofthe coarse aggregates
Origin Identification Rock type Degree of ASR a
CANADA
Nova Scotia NSl Greywacke High
NS2 Granitic Marginal
New Brunswick NB1 Limestone High
NB2 Rhyolite High
NB3 Greywacke/argilite Moderate
Québec QC Chloritic schist Moderate
Ontario ONI (Spratt) Siliceous limestone High
ON2 (Sudbury) Polygenic gravel High
AL1 Polygenic gravel Moderate
AL2 Polygenic gravel Moderate
British Colombia BC Polygenic gravel High
U.S.A.
New-Mexico NM Volcanic gravel High
Texas TX Volcanic sand High
Texas WR Polygenic sand High
Based on the 1-yr expansion in the Concrete Prism Test CSA A23.2-14A or
ASTM C 1293: <0.040%: non-reactive; 0.040-0.120%: moderately reactive;
>0.120%: highly reactive.

Table 6.7 - Concrete mixtures tested in the CPT and expansions obtained after one (control
mixtures) or two years (lithium mixtures)
[Li]/[Na+K] molar ratio Minimum
Aggregate 0 (control) 0.37 0.56 0.74 0.93 1.11 [Li]/[Na+K] to satisfy
1-year exp. (%) 2-year expansion (%) the CPT criteriona
LAVAL UNIVERSITY
NSl greywacke 0.142 0.131 0.046 0.082 >l.ll/>1.48b
NS2 granitic 0.029 0.034 0.56
NB1 limestone 0.179 0.073 0.064 0.076 >1.11
NB2 rhyolite 0.172 0.013 0.63/0.67"
NB3 gravel 0.087 0.086 0.031 0.71
QC chloritic 0.082 0.103 0.068 0.065 >0.93/X).72 d
ONI limestone 0.162 0.085 0.049 0.032 1.04
ON2 gravel 0.122 0.076 0.015 0.016 0.66/0.65 d
AL1 gravel 0.103 0.102 0.026 0.91
AL2 gravel 0.113 0.133 0.056 0.042 0.97
BC gravel 0.131 0.010 0.58
UNIVERSITY OF TEXAS c
NM gravel 0.203 0.256 0.183 0.046 0.77
TX sand 0.586 0.113 0.037 0.038 0.55
WRsand 0.188 0.037 0.028 0.032 0.37
Expansion <0.040% after 2 years; values interpolated from two tested [Li]/[Na+K] ratios unless the highest
ratio tested (following sign >) did not satisfy the above criterion;
From Feng [13]; other cement and non-reactive fine aggregate;
From Barborak [7]; other cement and non-reactive coarse (with TX, WR) or fine (with NM) aggregates;
From Durand [12]; other cement and non-reactive fine aggregate. With QC [Li]/[Na+K] =1.66 fail with LiF,
or Li2C03 or LiOH admixture.
182

Table 6.8 - Experimental conditions and 14-day expansions in the AMBT


Testing conditions Aggregate
[Li]/[Na+K] [Li]/[Na] in soak
NSl NS2 NB1 NB2 NB3 QC iONI ON2 AL1 AL2 BC NM TX WR
in bars solution
0.000 .273.197 .246 .414 .295 .152 .330 .238 .286 .224 .375 .966 .656
0.000a .293 .351 .341 .226 .307 .277 .278 .234 .489 .995
0.000" .43 .31 1.45 .71 .42
0.037 .367
0.074 .369
0.111 .349
0.148 .186
0.222 .215.103 .097 .234 .145 .229 .818 .468
0.222 ac .280 .179 .258 .227
0.222 b c .33 .19 1.34 .64 .36
0 0.296 .031 .125 .144 .187
0.370 .081 .062 .083 .177
0.444 .069 .030 .140 .017 .127 .115 .618 .281
0.444 ac .180 .054 .154 .045
0.444 " c .19 .02 1.06 .35 .20
0.518 .078 .099
0.592 .035 .013 .056 .065 .029 .607
0.592 " .148 .105
0.592 b c .06 .01 .73 .21 .08
0.740 .054
0.888 .043
0.278 ' .122 .270 .030 .172 .145 .059 .232
0.555
0.278" .05 .95 .27 .18
0.000 .309
.208 .331 .266 .274 .205 .421 .203 .319 .275 .362 .728 .441
crC') b
.387 .436 .321 .313 .529 .243 .320 .258 .356
.60 .30
.603
1.39 .63 .34
0.148 .370.127 .235 .20 Ie .278 .174 .477 .158 .328 .274 .327 .618 .361
O" .364 .330 .442 .128 .428
(") bc .56 .15 1.06 .46 .22
0.296 .197.050 .063 .034 .114 .054 .258 .024 .205 .205 .072 .401 .173
0.74 (")" .345 .345 .366 .027
r)b.c .22 .02 .73 .36 .06
0.370 s .163 .055 .045 .056 .246 .016 .060 .047 .023 .048
0.370b .02 .57 .09 .04
0.-444 .047 .031 .027 .071 .045 .099 .021 .101 .027
(") a .067 .048 .123 .013

o"
0.592 .024
.05 .02 .37 .07 .03
.033 .019
0.463 a .050 .037 .049 .074 .027 .042 .018 .017
0.925
0.463 " .04 .13 .05 .01
0.000 .468 .848
0.148 .491 .669
0.296 .345 .503
1.11
0.444 .099 .168
0.555" .038 .040 .041 .047 .030
0.555" .04 .03
19 7 13 6 11 19 40 25 13 11 12 25 25 11
XT
ier of tests
Numt 110 at Laval; 72 at CANMET-MTLa; 53 at University of Texas b
a
Tests performed at CANMET-MTL using a different cement (Na2Oeq of 0.93%) (unpublished data);
Test performed at University of Texas using a different cement; from Barborak [7];
c
Values are interpolated from two tested [Li]/[Na] ratios in the soak solution.
183

Table 6.9 - Experimental conditions and 28-day expansions in the AMBT


Testing conditions Aggregate
[Li]/[Na+K] [Li]/[Na] in soak
NSl NS2 NB1 NB2 NB3 QC ONI ON2 ALI AL2 BC NM TX WR
in bars solution
0.000 .453 .338 .398 .670 .490 .291 .618 .407 .478 .384 .543 1.19 .792
0.000 a .501 .630 .573 .426 .585 .481 .484 .370 .766 1.29
0.000 b .77 .49 1.91 .87 .58
0.037 .593
0.074 .619
0.111 .633
0.148 .307
0.222 .396 .212 .227 .446 .297 .433 1.040 .553
r)a.c .531 .386 .524 .431
r) b.c .57 .30 1.72 .78 .50
0 0.296 .108 .294 .302 .365
0.370 .174 .222 .235 .310
0.-444 .238 .060 .280 .054 .249 .171 .679 .321
(")*•. .460 .136 .325 .115
r) b.c .29 .06 1.24 .42 .27
0.518 .206 .184
0.592 .042 .030 .089 .120 .045 .716
(")" .275 .232
r) b.c .08 .02 .81 .25 .11
0.740 .066
0.888 .047

™r .361 .436 .120 .233 .346 .262


.16
.450
1.27 .41
0.000 .464 .335 .499 .430 .399 .349 .655 .335 .456 .387 .499 1.026 .653
(")' .678 .680 .482 .488 .782 .384 .476 .359 .495 .933
(")b .85 .45 1.89 .83 .51
0.148 .531 .278 .518 .365c .423 .335 .673 .350 .475 .410 .436 .889 .538
(") ac .612 .502 .759 .260 .703
(") bc .86 .28 1.49 .61 .35
0.296 .485 .050 .232 .089 .327 .278 .603 .050 .356 .305 .236 .746 .343
0.74 (") a c .530 .515 .739 .115
(,)b.c .50 .07 1.09 .32 .20
0.370 a .415 .199 .279 .262 .690 .026 .148 .236 .054 .291
Of .04 .88 .23 .12
0.-444 .288 .050 .042 .181 .176 .210 .049 .353 .109
(")" .317 .134 .363 .023
r) b.c .08 .03 .60 .16 .08
0.592 .036 .053 .039
.340 .065 .061 .174 .099 .041 .025
• # ' .06 .49 .11 .03
0.000 .647 1.336
0.148 .726 1.035
.672 .841
1.11 °296
0.444 .262 .443
0.555 a .089 .044 .050 .078 .057
b .05 .05
(")
19 7 13 6 10 19 40 25 13 11 12 25 25 10
Number of tests
110 at Laval; 72 at CANMET-MTL ".! 53 at University of Texas b
Tests performed at CANMET-MTL using a different cement; from Fournier (unpublished data);
Test performed at University of Texas using a different cement; from Barborak [7];
Values are interpolated from two tested [Li]/[Na] ratios in the soak solution.
184

Table 6.10 -Effect of Li in the mortar bars and/or in the soak solution on the 14- and
2 8-day expansions in the AMBT
Reduction (-) or increase (+)in expansion iri the AMBT (%)a
A (TOT(*OAt(*Q
[Li]/[Na+K] in bars: [Li]/[Na] in the soak solution
/
^&&IC&dlcs> 0:i 3.59 0:1 3.44 0 :0.22 0.74: 0.44 0.74: 0.30 0.74 : 0.15 0.74 :0
14d 28d 14d 28d 14d 28d 14d 28d 14d 28d 14d 28d 14d 28d
LAVAL UNIVERSITY
NSl greywacke -75 -47 -21 -13 -83 -36 -28 +7 +36 + 17 + 13 +2
NS2 granitic -48 -37 -75 -85 -36 -18 +6 -1
NB1 limestone -86 -89 -87 -87 -74 -42 -4 +30 +35 +25
NB2 rhyolite -92 -87 -51 -46 -36 -36
NB3 gravel -61 -33 -6 -14 -7 -19
QC schist -91 -90 -80 -79 -36 -22 -82 -86 -64 -4 +14 + 15 +35 +20
ONI limestone -83 -86 -58 -55 -29 -28 -78 -71 -22 -2 +45 +6 +28 + 13
ON2 gravel -93 -87 -39 -27 -90 -88 -34 -14 -15 -18
AL1 gravel -56 -48 -20 -9 -84 -63 -28 -26 + 15 -1 + 12 -5
AL2 gravel -71 -69 -49 -55 -56 -45 -8 -21 +22 +7 +23 -1
BC gravel -92 -92 -94 -91 -81 -57 -13 -20 -3 -8
NM gravel -37 -40 -36 -43 -15 -13 -90 -70 -58 -37 -36 -25 -25 -14
TX sand -57 -59 -29 -30 -96 -86 -74 -57 -45 -32 -33 -18
CANMET-MTL
NSl greywacke -49 -45 -39 -8 -4 +6 -77 -37 + 18 +6 +24 +22 +32 +35
NB1 limestone +24 -8
NB3 gravel -6 -16
QC schist -76 -68 -21 -9 -79 -69 +53 +21 +46 +18 +38 +15
ONI limestone -66 -60 -50 -44 -16 -10 -60 -38 +19 +26 +44 +30 +72 +34
ON2 gravel -84 -76 -18 -10 -95 -95 -90 -76 -54 -46 -12 -20
AL1 gravel +15 -2
AL2 gravel +10 -3
BC gravel -27 -35
TX sand -57 -45 -39 -31
UNIVERSITY OF TEXAS
ONI limestone -86 -90 -56 -62 -23 -26 -88 -90 -49 -35 +30 +12 +40 + 10
ON2 gravel -97 -96 -94 -88 -39 -39 -94 -94 -94 -86 -52 -43 -3 -8
NM gravel -50 -58 -27 -35 -8 -10 -74 -69 -50 -43 -27 -22 -4 -1
TX sand -70 -71 -51 -52 -10 -10 -90 -82 -49 -63 -35 -30 -11 -5
Wright sand -81 -81 -52 -53 -14 -14 -93 -86 -86 -66 -48 -40 -19 -12
With respect to the control mixtures (no Li in bars) subjected to the standard procedure (no Li in the soak
solution).
185

Table 6.11-Minimum [Li]/[Na] needed in the soak solution to satisfy


the 0.10% expansion limit criterion in the AMBT after 14 and
28 days, with or without Li in bars
No Li in bars [Li]/[Na+K] = 0.74 in bars Minimum [Li]/[Na+K] to
Minimum [Li]/[Na] in Minimum [Li]/[Na] in satisfy the CPT criterion
Aggregates
soak solution3 soak solution3 Expansion limit criterion
14 d 28 d 14 d 28 d <0.040% after 2 years
LAVAL UNIVERSITY
NSl greywacke 0.40 0.63 0.38 0.56 >1.11 />1.48 b
NS2 granitic 0.22 0.30 0.19 0.27 0.56
NB1 limestone 0.33 0.50 0.27 0.41 >1.11
NB2 rhyolite 0.31 0.47 0.21 0.30 0.63/0.67 b
NB3 gravel 0.36 0.52 0.30 0.44 0.71
QC schist 0.22 0.26 0.24 0.41 >0.93/>0.72 d
ONI limestone 0.51 0.58 0.42 0.48 1.04
ON2 gravel 0.30 0.40 0.22 0.27 0.66/0.65 d
AL1 gravel 0.50 0.62 0.39 0.50 0.91
AL2 gravel 0.49 0.65 0.44 0.59 0.97
BC gravel 0.49 0.55 0.27 0.40 0.58
NM gravel >0.80 >0.80 0.44 0.56 0.77
TX sand 0.65 0.67 0.38 0.45 0.55
CANMET-MTL
NSl greywacke 0.54 0.72 0.41 0.68 >1.11 />1.48 b
QC schist 0.35 0.46 0.36 0.46 >0.93/>0.72 d
ONI limestone 0.62 0.85 0.47 0.70 1.04
ON2 gravel 0.35 0.45 0.18 0.29 0.66/0.65 d
UNIVERSITY OF TEXAS
ONI limestone 0.53 0.55 0.35 0.36 1.04
ON2 gravel 0.31 0.36 0.20 0.28 0.66/0.65 d
NM gravel 1.10 1.10 0.54 0.65 0.77
TX sand 0.71 0.72 0.36 0.50 0.55
WR sand 0.54 0.60 0.30 0.41 0.37
Average 0.51 0.60 0.34 0.46 —
Lowest value 0.22 0.26 0.18 0.27 0.35
Highest value 1.10 1.10 0.54 0.70 >1.48
Expansion limit criterion: <0.10% after 14 or 28 days;
From Feng [13]; other cement and non-reactive fine aggregate;
From Barborak [7]; other cement and non-reactive coarse (with TX, WR) or fine (with NM) aggregates;
From Durand [12]; other cement and non-reactive fine aggregate. With QC [Li]/[Na+K] =1.66 fail with
LiF, or Li 2 C0 3 or LiOH admixture.
186

Table 6.12 -Coefficients of linear regression (R2) obtained between CPT and AMBT
results at 14 and 28 days for the various conditions tested at Laval
University
Compared variables [Li]/[Na+K] [Li]/[Na] in Nbr of R2
CPT (Y) AMBT (X) in bars soak solution results 14d 28d
Expansion for
Expansion for
control tests
control mixtures 0 0 22 0.21 0.14
(Figures 6.14 for 14d)
(1 year)
Expansion for
control tests 0 0 22 0.06 0.02
(Figures 6.15 for 14d)
0 0.222 17 0.03 0.00
0 0.444 17 0.03 0.00
Expansion with 0.04 0.01
0 0.592 13
lithium in bars and/or in the
Minimum 0.74 0 22 0.00 0.01
soak solution
[Li]/[Na+K] 0.74 0.148 22 0.05 0.08
(Figures 6.16 and 6.17)
needed in 0.74 0.296 22 0.09 0.20
concretea 0.74 0.444 18 0.00 0.07
0 0.222 17 0.04 0.11
0 0.444 17 0.01 0.12
Expansion reduction with 0.592 0.01 0.02
0 13
lithium in bars and/or in the
0.74 0 28 0.65 0.57
soak solution r
0.148
0.74 0.296 22 0.43 0.61
0.74 0.444 18 0.35 0.35
0 variable 22 0.00 0.02
Limit of 0.10% -
Minimum 0.74 variable 22 0.19 0.25
[Li]/[Na] Limit of 0.04% 14 0.39
needed in Limit of 0.06% 14 0.24
half of the -
the soak Limit of 0.08% Variable 14 0.23
ratio in bars - —
solution" Limit of 0.10% 14 0.23
Limit of 0.12% 14 0.24
To satisfy the 2 •year, 0.040% expansion limit criterion;
b
To satisfy an expansion limit criterion at 14 or 28 days.
187

Table 6.13 - Determination ofthe [Li]/[Na+K] needed in concrete based on


different limit after 28 days in the AMBT using a [Li]/[Na+K]
in the soak solution of 50% of this ratio in bars

Minimum
Minimum [Li]/[Na+K]
[Li]/[Na+K]
needed in the AMBT
needed in the CPT
Aggregates Expansion limit criterion after 28 days
Expansion limit
rritprirm <0 040%
V111C1H.'11 ^ \ J . \ J ^ \ J / o
< < < < <
after 2 years
0.04% 0.06% 0.08% 0.10% 0.12%
NSl greywacke 1.15 1.13 >l.lland>1.48a
NS2 granitic 0.67 * 0.62 0.58 0.54 0.49 0.56
NB1 limestone 1.35 0.85 >1.11
NB2 rhyolite 0.71 0.68 0.66 0.64 0.62 0.63 and 0.67 a
NB3 gravel 1.46 1.42 1.37 1.34 1.28 0.71
QC schist 1.30 0.94 | 0.87 >0.93and>0.72 b
ONI limestone 1.18 1.14 1.10 1.07 | 1.03 1.04
ON2 gravel 0.71 0.67 | 0.56 0.66 and 0.65 b
AL1 gravel 1.18 1.09 1.01 " 0.92 0.84 0.91
AL2 gravel 1.17 1.13 1.07 1.03 0.99 0.97
BC gravel 0.94 0.74 0.72 0.70 0.68 0.58
NM gravel 1.12 1.11 1.10 1.09 1.08 0.77°
TX sand 1.03 1.00 0.97 0.94 0.91 0.55 c
WR sand 0.90 0.86 0.82 0.78 0.74 0.37 c
2
Coefficient (R ) 0.39 0.24 0.23 0.23 0.24
Underestimation 1/14 2/14 4/14 5/14 8/14
a
CPT from Feng [13]; other cement and non-reactive fine aggregate;
b
CPT from Durand [12]; other cement and non-reactive fine aggregate;
c
CPT from Barborak [7]; other cement and non-reactive coarse (with TX, WR) or fine (with NM)
aggregates.
188

Table 6.14 -Determination of the [Li]/[Na+K] needed in concrete based on the


reduction/increase of expansion after 28 days in the AMBT using a
[Li]/[Na+K] of 0.74 in bars and a [Li]/[Na] of 0.148 in the soak solution
2 8-day expansion Minimum [Li]/[Na+K] needed to satisfy
in the AMBT (%) Expansion the CPT (Expansion <0.040% after 2 years)
variation Calculated
Aggregates
Control Test with By the the equation Observed with the
test (El) Li(E2) (E2-E1/E1) 1.0 + 0 . 7 0 x ( E 2 - E l ) / E l CPT
If(E2-El)/El > 0.1 use CPT
NSl greywacke1 0.453 0.531 0,17 Use CPT >l.lland>1.48a
2
NSl greywacke 0.501 0.612 0,22 Use CPT >1.11 and>1.48 a
1
NS2 granitic 0.338 0.278 -0,18 0.87 0.56
NB1 limestone1 0.398 0.518 0,3 Use CPT >1.11
NB2 rhyolite1 0.670 0.365 -0,46 0.68 0.63 and 0.67a
1
NB3 gravel 0.490 0.423 -0,14 0.90 0.71c
1
QC schist 0.291 0.335 0,15 Use CPT >0.93 and >0.72 b
QC schist2 0.426 0.502 0,18 Use CPT >0.93and>0.72 b
ONI limestone ' 0.618 0.673 0,09 1.06 1.04
ONI limestone2 0.585 0.759 0,3 Use CPT 1.04
ONI limestone3 0.77 0.86 0,12 Use CPT 1.04
ON2 gravel ' 0.407 0.350 -0,14 0.90 0.66 and 0.65 b
ON2 gravel2 0.481 0.260 -0,46 0.68 0.66 and 0.65 b
3
ON2 gravel 0.49 0.28 -0,43 0.70 0.66 and 0.65 b
1
AL1 gravel 0.478 0.475 -0,01 0.99 0.91
AL2 gravel1 0.384 0.410 0,07 1.05 0.97
BC gravel1 0.543 0.436 -0,2 0.86 0.58
NM gravel1 1.191 0.889 -0,25 0.83 0.77 c
NM gravel3 1.91 1.49 -0,22 0.85 0.77 c
1
TX sand 0.792 0.538 -0,32 0.78 0.55 c
2
TX sand 1.286 0.703 -0,43 0.70 0.55 c
3
TX sand 0.87 0.61 -0,3 0.79 0.55 c
3
WR sand 0.58 0.35 -0,39 0.73 0.37 c
2
Coefficient (R ) 0.79
Underestimation 0/23
AMBT from Laval University;
AMBT from CANMET (unpublished data);
AMBT from Barborak [7];
CPT from Feng [13]; other cement and non-reactive fine aggregate;
CPT from Durand [12]; other cement and non-reactive fine aggregate;
CPT from Barborak [7]; other cement and non-reactive coarse (with TX, WR) or fine (with NM)
aggregates.
189

1.2 i i i i i i i i

1.1 U
1
• U Laval

° *• 0.9
t; + UTexas

Ç«
î h 0.8­
IS ^
■*=• S 0.7
| | 0.6

|S o.5 H

0.4

0.3 i i i i i i I i
0 0.1 0.2 0.3 0.4 0.5 0.6
1­year expansion for the control mixture (%)

Figure 6.12 ­ Minimum [Li]/[Na+K] needed in the CPT as a function of aggregate


expansivity in the CPT (i.e. 1­year expansion for the control mixture).
190

| , | , | , | , 1 ' i ' i—' i ' i ' r


0.9 ­ BC gravel ­ TX sand
0.8­ ­
0.7 ­ —
S? ­
~c 0 . 6 ^J
O
«ô 0.5 ­« ­
C

LU ** — — ­ . \ %■
0.3­ ~
0.2­

0.1 ­

i ; i | < | i | 1 —| , 1 , 1 i | i | r
0.1 0.2 0.3 0.4 0.5 0.6 0.1 0.2 0.3 0.4 0.5 0.6
Ratio [Li]/[Na] in soak solution Ratio [Li]/[Na] in soak solution
D
1.0 r
' 1 ' I ' 1 ~
0.9 QC chloritic schist
0.8

0.7

i—'—i ' r
0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 0.5 0.6
Ratio [Li]/[Na] in soak solution Ratio [Li]/[Na] in soak solution

Figure 6.13 ­ 14­day and 28­day expansions in the AMBT for different aggregates as a
function ofthe [Li]/[Na] in the soak solution using a [Li]/[Na+K] of 0.74 in
bars.
191

1.5 i i i i i i i
r ' "V T
1.4
1.3 • ULaval
+ UTexas
1.2­ ■ Can met
P
m 1.1
S
1
Si 0.9
gS 0.8
x 'É
0.7
« E
R 0.6
0.5
0.4
0.3
0.2 . ■"*'•
0.1

0 0.1 0.2 0.3 0.4 0.5 0.6


1­year expansion for the control mixture (CPT) (%)

Figure 6.14­ Aggregate expansivity in the AMBT (14­day expansion for the control
mixture) as a function of aggregate expansivity in the CPT (i.e. 1­year
expansion for the control mixture).

1.2 I ' ! I ' I ' I ' I ' I


NS1
1.1 <Jb
• ULaval
o UTexas
se 1 +
■ Can met
gs ^ .
0.9 ­

p
■£(­ 0.8

£2 0.7 ON2
CE

E 0.6 TX
3 _
3±;
c '• 0.5

0.4

0.3
i |—i—j—.—|—,—p­i |—i—|—i—r^
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
14­day expansion for the control mixture (AMBT) (%)

Figure 6.15 ­ Minimum [Li]/[Na+K] needed in the CPT as a function of aggregate


expansivity in the AMBT (i.e. 14­day expansion for the control mixture).
192

B
1.2 1.2 T
I ' I ' I I ' I

1.1
0 : 0.222 1.1 'NS1 0 : 0.444
ON1
C «o 1 S 1
o i­
c 3
■ • ULaval S2 ON1 • ULaval
■ Canmet SS A Canmet
g S. 0.9 + UTexas S > 0.9 iQC + UTexas

^ t ­ 0.8
X>
«■=•_> 0.7 " £ 0.7
N2
4S)°
E S °­ 6 E 0.6

1*0.5
3 _
ES* 0.5
<G>
s
0.4 0.4­

0.3 r |— 0.3 1
I ' I ' I ' I "■" T
i ' i T ^ i ' i i ' i r~
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
28­day e x p a n s i o n in the AMBT (%) 28­day e x p a n s i o n in the AMBT (%)

1.2 I l _' I ' I

1.1
0 : 0.592
a
I
• ULaval
c n
O » t f °N1 ■ Canmet
+ UTexas

i n 0.8 NM

C»0.6
­S.
M 0.5
0.4­

0.3
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
28­day expansion in the AMBT (%)

Figure 6.16 ­ Minimum [Li]/[Na+K] needed in the CPT as a fonction of the 28­day
expansion in the AMBT in the absence of lithium in bars, for a [Li]/[Na]
ranging from 0.222 to 0.592 in the soak solution.
193

1.2 M ' I ' I ' I ' I ' I ' I ' I ' I I ' I
S 1
1.1
1^ 0.74 : 0 74 : 0.148
S 1
b2
c
8 cfc£3c
8>0.9

IS 1
ON2___"
C_SZ NM

s? 2 0.7
E­S 0 ­ 6 Tx

Eo
ci * 0.5
• ULaval m ULaval
0.4 ■ Canmet ■ Canmet
■f UTexas + UTexas!
0.3 1
I ' I ' I ' I ' I ' I ' I ' I I ' I ' I ' I ' I ' I • I ' I ' I
0.0 0.2 0.4 0.6 0.8 1.0 1.2 14 1.6 1.8 2.0 0.0 02 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
28­day expansion in the AMBT (%) 28­day expansion in the AMBT (%)

1.2 1
I ' I ' I ' I I ' I ' I ' 1 ' '
N
1.1 l l 4 > 0­74 : 0.296 0.74 : 0.444

"S C+*
Sc
2 1­
o S•

° *" 0.9 Q » NM ■£>

^>ON2
0.6 TX
*="« 0.7
a •
1 fi 0.8
• ULaval • ULaval
0.4 ■ Canmet ■ Canmet
+ UTexas + UTexas
0.3 1
I ' I ' I ' I ' ! ' I ' I ' I I ' I ' I ' I
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 ^A 1.6 1.8 2.0
28­day expansion, in the AMBT (%) 28­day expansion, in the AMBT (%)

Figure 6.17 ­ Minimum [Li]/[Na+K] needed in the CPT as a function of the 28­day
expansion in the AMBT using a [Li]/[Na+K] of 0.74 in bars and a [Li]/[Na]
ranging from 0 to 0.444 in the soak solution.
194

B
No Li in bars ­ 1 4 d No Li in bars ­ 28 d
1.2 1.2
T~ i—i—r —i i r
1.1 1.1 ­
* ct_i=Ns
MS1
ON1
0) <3S ONi
•> 1 1
b2 1­
c
8 8. 0.9 8 0.9­
g­T» QC ~ f c = > m

?"0.B­
fi­ NM

S I °­7 ON2_ sr © o.7 ON2 •

E IS °­6 E 8 °­6
E o nc • / CS­t­
I 2 0.5 ■5 — 0.5 —
• ULaval • U ­Laval
0.4 ■ Canmet 0.4
■ Canmet
+ UTexas + UTexas,
0.3 i i f i i—i—i i i r 0.3
i r ■f—r T T T T
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.0 0.1 0 2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
M i n i m u m [Li]/[Na] i n soak s o l u t i o n M i n i m u m [Li]/[Na] i n soak s o l u t i o n
t o pass t h e 0 , 1 % l i m i t i n the AMBT t o pass t h e 0 , 1 % limit in t h e AMBT

D
1.2
0.74 [Lil/fNa+K] in bar ­ 1 4 d 0.74 fLil/fNa+K1 in bar ­ 28 d

1.1 H

o
1
b2
c n
O S 0.9­
o >>
c^
0.8

It
f_T.C
0.7 ON2
_i • *
0.6 TX
is. 0.5 H C3>
Eo
ç **
a 0.4
sy • ULaval
■ Canmet
+ UTexas
0.3
i i f—i—i—i—i—i—i—r ~i—i—I i—i—I—i—r
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
M i n i m u m [Li]/[Na] i n soak s o l u t i o n M i n i m u m [Li]/[Na] in soak s o l u t i o n
t o pass t h e 0 , 1 % limit i n t h e AMBT t o pass the 0 , 1 % limit i n the A M B T

Figure 6.18 ­ Minimum [Li]/[Na+K] needed in the CPT as a fonction of the minimum
[Li]/[Na] needed in the soak solution to satisfy the 0.10% expansion limit
criterion in the AMBT after 14 days (A, C) and 28 days (B, D), in the
absence (A, B) and in the presence (C, D) of lithium in bars at a [Li]/[Na+K]
of 0.74.
195

B
0 : 0.222 0 : 0.444
1.2 1.2 i r
1—I—I—\ i—i r
0)
1.1 Ci NSI 3_
o
1.1 H
ONl
<fc NS1 _i>
% C__»0N1~4J> o CT»
be 1 biï i
1 | 0.9 H S IS 0.9
cï^fc * #
o >>
0.8
z" ON2
0.7 *=■ » 0.7 H

la
0.6
ii"­ TX

E o 0.5 |S0.5
• ULaval • ULaval
0.4 ■ Canmet 0.4 ■ Canmet
4 UTexas + UTexas
0.3 0.3
n i i i i i i i i r i i i i i i i—i i r
­100 ­90 ­80 ­70 ­60 ­50 ­40 ­30 ­20 ­10 0 10 ­100 ­90 ­80 ­70 ­60 ­50 ­40 ­30 ­20 ­10 0 10
Difference (%) in expansion after 28 days (AMBT) Difference (%) in expansion after 28 days (AMBT)

0 : 0.592
l l I I l I I I

1.1 ­ NS1 ^J__>


ON1 !___>
S in i—
S» 1 •

S ? 10
ft '8" C3­ NM O
•3; » 0.7 ­

||0.6­
+
|fi0.5­
s • ULaval Y
0.4­ Hh * Canmet T
+ UTexas j
1 l l l l l l I I I
­100 ­90 ­80 ­70 ­60 ­50 ­40 ­30 ­20 ­10 0 10
Difference (%) in expansion after 28 days (AMBT)

Figure 6.19­ Minimum [Li]/[Na+K] needed in the CPT as a function ofthe reduction in
expansion in the AMBT after 28 days in the absence of lithium in bars and
for a [Li]/[Na] ranging from 0.222 to 0.592 in the soak solution, compared
to the control mixtures.
196

B
0.74:0(14d) 0.744 : 00((28 d)
1.2 I / I ' I ' I '
'j ' r
1.1 NB4_E5 <t_­fcj. 5
NS1
0)
/ c 3 j j ON1 " p b
Si! 1­
o i
° ~ 0.9 ­

ft"8
« = • • 0.7

IA
E 8 °­6
g 0.6­
a c_rr^5^fr N S 2 | S 0.5
^WEES*'
■2 0.5
f • ULaval I
• ULaval 0.4 ■ Canmet
0.4­ WJ^/ ■ Canmet + UTexas
+ UTexas 0.3 I ' I ' I■+­
'' I ' I ' I ' I ' I ' I ' I ' I ' I
0.3 I ' I' I ' I ' I ' I ' I ' I ' I ' I ' I ' I 50 ­40 ­30 ­20 ­10 0 10 20 30 40 50 60 70 80
­50 ­40 ­30 ­20 ­10 0 10 20 30 40 50 60 70 80 Difference (%) i n expansion (AMBT)

D
0.74: 0.148 (14 d) 0.74 : 0.148 (28 d)
1.2 l~r

, i criN
? sifc _
fl>
1.1 ~3
C_EONU__4 1
&e i g j ~ + QN1 p_r>

gs _ #
• QC
084
ft ■C3H>NM! / i "
7
S J o­ CS'
ON2
6
« 3%'
¥ § o­
|So.5H
• ULaval • ULaval
0.4 ■ Canmet 0.4 ■ Canmet
+ UTexas
♦' + UTexas
0.3 I ' I ' I ' I ' I ' I ' I ' I ' I ' I 0.3 I ' I ' I ' I ' I ' I ' I ' I ' I
­60 ­50 ­40 ­30 ­20 ­10 0 10 20 30 40 50 60 ­50 ­40 ­30 ­20 ­10 0 10 20 30 40 50
Difference (%) i n e x p a n s i o n (AMBT) Difference (%) in expansion (AMBT)

Figure 6.20 ­ Minimum [Li]/[Na+K] needed in the CPT as a function of the reduction in
expansion in the AMBT after 14 or 28 days using a [Li]/[Na+K] of 0.74 in
bars and a [Li]/[Na] ranging from 0 to 0.444 in the soak solution compared
to the control mixtures.
197

E
0.74 : 0.296 (28 d) 0.74 : 0.444 (28 d)
■ i ■ i ■ i/ ■ i ' 1 ' 1 ' 1 ' / 1.2
NS1
1.1 ­ 1.1 •<_fcj
fl) »ON1
/«CEI ONI» m>/ <t
num [Li]/[Na+K] in concri

­»
> pass the CPT a 2 years

■S II I 1

0.9
io
o

' NM _
B 0.8 NM
, 1 , 1 ,
p o p

/ON2 • ­ ■t. £ 0.7 ^ O N 2

Ë | 0.6
■g * 0.5 ­ / — |2o.5
S
/
' • ULaval 1 " ii • ULaval
0.4­ ■ Canmet _ 0.4 H ■ Canmet
+ UTexas j + UTexas
0.3­ , ( , | 1 | 1 | 1 | i 1 i 1 i 0.3 ­f­ T T ^T ■¥­
­100 ­80 ­60 ­40 ­20 0 20 40 60 ­100 ­80 ­60 ­40 ­20 0 20 40 60
Difference (%) in expansion (AMBT) Difference (%) in expansion (AMBT)

Figure 6.20 (Suite) ­ Minimum [Li]/[Na+K] needed in the CPT as a function of the
reduction in expansion in the AMBT after 14 or 28 days using a
[Li]/[Na+K] of 0.74 in bars and a [Li]/[Na] ranging from 0 to 0.444
in the soak solution compared to the control mixtures.

Begin by testing the aggregate with the following two mixtures:


1. Control mixture (expansion at 28 days = El)
2. Mixture with lithium: [Li]/[Na+K] = 0.74 in bar and [Li]/[Na]=0.148 in
soak solution (expansion at 28 days = E2)

1
Isthe((E2­El)/El)<0.1

1
1
Yes No

Use the following equation:


1.0 +0.7x((E2­El)/El) = /.a./o Use concrete prism test to
The Ratio = [Li]/[Na+K] to use in evaluate the ratio to use
concrete (lithium nitrate admixture)

Figure 6.21 ­ Potential process, using the AMBT, to determine the [Li]/[Na+K] needed in
concrete to prevent ASR with lithium nitrate.
198

• ULaval
■ Canmet
+ UTexas
I ' I ' I ' I ' I ' I ' I ' I ' I ' I
-0.6 -0.5 -0.4 -0.3 - 0 2 -0.1 0.0 0.1 0 2 0.3 0.4 0
(E2-E,)IE,

Figure 6.22 - Minimum [Li]/[Na+K] needed in the CPT as a function of the reduction in
expansion in the AMBT after 14 or 28 days using a [Li]/[Na+K] of 0.74 in
bars and various [Li]/[Na] in the soak solution.
199

7 Les mécanismes à l'origine de l'efficacité du nitrate de lithium

7.1 Introduction

Cette publication a été soumise à la revue Cement and Concrete Research le 5 novembre
2008 et acceptée le 24 septembre 2009. L'article a été publié en 2010 dans un numéro
spécial de cette revue (volume 40, numéro 4) regroupant un certain nombre d'articles
sélectionnés parmi ceux présentés à la 13th International Conference on AAR (ICAAR),
conférence tenue en Norvège en juin 2008.

7.2 Résumé

Plusieurs séries d'expériences ont été menées sur des pâtes de ciments, des bétons fabriqués
avec une variété de granulats réactifs, des spécimens composites de pâte de ciment et de
particules de granulats et une variété de particules de granulats réactifs naturels et de phases
minérales immergés dans différentes solutions contenant du lithium. L'objectif principal
était de déterminer quel(s) mécanisme(s) permet(tent) de mieux expliquer l'efficacité du
LiNÛ3 contre la réaction alcalis-silice (RAS) et les variations de cette efficacité en fonction
du type de granulat réactif à contrer. Les principales conclusions sont les suivantes: (1), le
pH dans la solution interstitielle du béton n'est pas diminué de manière significative en
présence de LiNC^; (2), la quantité de silice dissoute dans la solution interstitielle est
toujours faible et n'est pas influencée par la présence de LiNCh, réfutant le mécanisme
proposé relatif à une plus grande solubilité de la silice en présence de lithium; (3), le seul
produit de réaction observé dans les bétons contenant du LiNC.3 ressemble au gel classique
de la RAS et son abondance est proportionnelle à l'expansion du béton, ce qui suggère qu'il
est expansif et qu'il contient du lithium, réfutant les mécanismes proposés relatifs à la
formation hâtive d'un produit cristallin ou d'un gel amorphe contenant Si et Li et qui serait
non ou moins expansif; (4), aucun produit de réaction formé tôt et recouvrant les grains de
silice réactive ou de granulats, produit qui pourrait agir comme une barrière physique
contre la dissolution de la silice, n'a été observé en présence de LiNÛ3, sauf pour un certain
nombre de matériaux réactifs mais après immersion dans une solution de IN LiOH à 350°C
en autoclave (également à 80°C pour l'obsidienne); (5), parmi tous les mécanismes
200

considérés dans cette étude, une plus grande stabilité chimique de la silice en présence de
lithium, pour une autre raison que la réduction du pH ou la formation hâtive d'un produit
protecteur sur les phases réactives, est celui qui permet d'expliquer le mieux l'efficacité du
LiNQ3 contre la RAS.
201

7.3 Publication scientifique no. 5

Experimental investigation of the mechanisms


by which LiN0 3 is effective against ASR

C. Tremblay 3 , M.A. Bérubé b , B. Fournier c , M.D. Thomas d , K.J. Folliarde


a
PhD Student, Département de géologie et de génie géologique, Université Laval, Québec, QC, Canada, G1K 7P4
b
Associate Professor, Département de géologie et de génie géologique, Université Laval, Québec, QC, Canada, G1K 7P4
c
Professor, Département de géologie et de génie géologique, Université Laval, Québec, QC, Canada, G1K 7P4
d
Professor, Department of Civil Engineering, University of New Brunswick, Fredericton, NB, Canada, E3B 5 A3
d
Professor, Department of Civil Engineering, University of Texas, Austin, U.S.A., TX 78712

ABSTRACT

Various series of experiments were carried out on cement pastes, concretes made with a
variety of reactive aggregates, composite specimens made of cement paste and reactive
aggregate particles, and a variety of reactive natural aggregates and mineral phases
immersed in various Li-bearing solutions. The main objective was to determine which
mechanisms(s) better explain(s) the effectiveness of LiNC>3 against ASR and variations in
this effectiveness as well with the type of reactive aggregate to counteract. The principal
conclusions are the following: (1), the pH in the concrete pore solution does not
significantly decrease in the presence of LiN(_>3; (2), the concentration of silica in the pore
solution is always low and not affected by the presence of LiNC>3, which does not support
the mechanism relating to higher solubility of silica in the presence of lithium; (3), the only
reaction product observed in the LiNCVbearing concretes looks like classical ASR gel and
its abundance is proportional to concrete expansion, thus is likely expansive while likely
containing lithium; this does not support the mechanisms relating to formation of a non or
less expansive Si-Li crystalline product or amorphous gel; (4), early-formed reaction
products coating the reactive silica grains or aggregate particles, which could act as a
physical barrier against further chemical attack of silica, were not observed in the LiNCh-
bearing concretes, but only for a number of reactive materials after immersion in IN LiOH
at 350°C in the autoclave (also at 80°C for obsidian); (5), higher chemical stability of silica
due to another reason than pH reduction or early formation of a protective coating over the
reactive phases, is the mechanism among those considered in this study that better explains
the effectiveness of LiNÛ3 against ASR.

Keywords: Alkali-silica reaction; concrete expansion testing; lithium nitrate; lithium


inhibition mechanisms; reaction products.
202

1 Introduction

A few days after mixing, the micropores in hardened concrete are filled with a highly basic
(i.e., pH > 12.5) solution that consists mainly of dissolved alkali hydroxides (K+, Na+, and
OH" ions) with minor amounts of other elements (e.g., Si, Ca+2, S04"2) [1-3]. Poorly
crystalline or métastable silica minerals (e.g. opal, tridymite, cristobalite), volcanic glass,
micro- to cryptocrystalline quartz, and some varieties of macrogranular quartz within the
aggregate particles are chemically unstable and react deleteriously in such high pH
environment, sometimes inducing the premature distress of the affected concrete elements.
This phenomenon is known as "alkali-silica reaction" (ASR).

The principal source of alkalies in conventional concrete is portland cement. The higher the
dosage and the alkali content ofthe cement, the higher the total concrete alkali content, the
higher the pH (or the OH ion concentration) in the pore solution, and the higher the risk of
ASR in presence of potentially reactive aggregates. The chemical attack of reactive silica
by the highly basic pore fluid results in the formation of an hygroscopic "alkali-silica gel"
[3,4]. Portlandite releases OH" ions in the concrete pore solution in order to satisfy
equilibrium between the cations (mostly alkali ions) and the anions (mostly OH" ions).
Differences in free energy between this gel and the pore fluid would then induce water and
various ionic species in the pore fluid to flow into this gel. The Ca+2 ions also play a
critical role in the process [1,3]. Tensile stresses build up and microcracking occurs when
the pressure generated at localized sites exceeds the tensile strength of the aggregate
particles and of the cement paste. Concrete expansion develops with associated
microcracking in the reactive aggregate particles and the cement paste.

Despite its name, the "alkali-silica gel" contains significant amounts of calcium [5].
Immersion tests in alkaline solution performed over the years in our laboratory on various
calcium-free reactive silica materials (e.g. synthetic silica gel, quartzitic sandstone with
quartz overgrowth and so on) clearly demonstrated that large amounts of silica can dissolve
and remain in solution until Ca(OH)2 is incorporated into the system, which thus results in
precipitation of a calcium-alkali-silica gel. Diamond [1] also observed that opal just
dissolved in pure NaOH solution, then concluded that calcium (e.g. portlandite) is required
203

for the formation of an expansive silica gel. It is also generally considered that other
crystalline products of ASR are not deleteriously expansive [1]. It thus appears that the
expansive/deleterious character of ASR products is largely influenced by their composition
and degree of crystallinity.

McCoy and Caldwell [6] first proposed that the alkali-silica reactivity (ASR) could be
controlled by using lithium admixtures. Since then, many studies showed that various
lithium salts can reduce ASR expansion in concrete when used in sufficient amounts [7].
LiNÛ3 proved to be one of the most effective salts. However, its effectiveness varies with
the concrete alkali content and the type of reactive aggregate to counteract, irrespective of
its inherent degree of expansivity (in control concretes) [8,9]. In actual fact, the
mechanisms involved are still not well understood.

Various mechanisms can be proposed to explain the beneficial effect of LiN03 against
ASR, which can be grouped in two main categories with respect to the two conditions
required for deleterious expansion to occur in the presence of reactive silica, i.e. chemical
attack of the silica by the highly basic concrete pore solution and formation of an
expansive calcium-alkali-silica gel:

1. The chemical stability of reactive silica is increased in the presence of LiNOy due
to: (a) pH decrease in the concrete pore solution (mechanism A), (b) some other
change(s) in the chemistry of this solution (mechanism B). and/or (c) early
formation, at the surface of or somehow surrounding the reactive silica grains or
aggregate particles, of a Si-Li reaction product, crystalline or amorphous, which
acts as a physical barrier against further reaction (mechanism C);

2. The chemical stability of reactive silica remains almost unchanged in the presence
of LJNQ3, but concrete expansion is significantly reduced or suppressed, due to:
(a) formation of a crystalline and non expansive Si-Li reaction product (mechanism
D), (b) formation of a Si-Li amorphous gel which is, however, non expansive or
much less expansive than the classical expansive ASR gel (mechanism E), or
(c) higher solubility of silica, which thus allows most ofthe reacted silica to remain
in solution without forming an expansive reaction product (mechanism F).
204

The mechanisms A to F are discussed hereafter.

1.1 Mechanism A - Increased chemical stability of reactive silica due to pH


decrease in the concrete pore solution

In the ASR, the aggressive ionic species against the reactive silica are the hydroxyl ions. As
mentioned above, the portlandite (Ca(OH)2) releases OH" ions in the concrete pore solution
in order to satisfy equilibrium between the cations (mostly alkali ions) and the anions
(mostly OH" ions), while the calcium ions play a critical role in the formation of an
expansive gel. It could be suggested that the pH ofthe concrete pore solution is reduced in
the presence of Li+ cations . However, a number of studies already proved that the pH is not
significantly modified in the presence of LiNÛ3 [10,11,12]. For instance, in the study by
Bérubé et al. [12] on cement pastes, the pH decreased by about 0.1 only in the presence of
LiNÛ3 with respect to control pastes. This was observed despite a higher overall
[Na+K+Li] concentration in the pore solution, which was related to the addition of nitrate
ions (not analyzed), which likely compete with the OH" ions. The findings of Bérubé et al.
[12] were very similar to the results reported by Diamond [11].

1.2 Mechanism B - Increased chemical stability of reactive silica due to some


change (other than pH) in the chemistry of the concrete pore solution

Lawrence and Vivian [13] found that the dissolution of silica is influenced by the type of
hydroxide involved, being lower in the presence of LiOH than with NaOH, the highest rate
being observed with KOH. Wijnen et al. [14] also observed that the dissolution of
amorphous silica gel was greatly affected by the type of alkali hydroxide used and
suggested that the dissolution rate decreases as the hydrated ionic radius increases, e.g.
from K+ (smallest radius) to Na+ to Li+ (largest radius). For their part, Collins et al. [15]
observed a reduction in silica dissolution in the presence of LiN03 and LiCl, but an
increase in the presence of LiOH (mixed with a solution of 0.7M NaOH). It is likely that
this increase was due to pH increase due to the addition of LiOH, as observed by Diamond
[11]. In the study by Feng et al. [16], the amounts of silica dissolved and sodium consumed
decreased with respect to the NaOH control solution when ground samples of various
205

reactive aggregates were immersed in a NaOH+LiN03 solution. However, the authors [16]
were not able to correlate the effectiveness of lithium, variable from one reactive aggregate
to another, irrespective of their inherent degree of reactivity (i.e. expansion in control
concretes), with the reduction of dissolved silica. All of the above results suggest that
lithium can decrease the dissolution of silica, then limiting the rate of formation of the
expansive ASR gel.

1.3 Mechanism C - Increased chemical stability of reactive silica due to early


formation, in contact with the reactive silica grains or aggregate particles, of a
Si-Li reaction product which acts as a physical barrier against further reaction

This protective product could be either crystalline or amorphous (i.e. a gel). In the study by
Mitchell et al. [17], opal particles were covered with a reaction product (possibly a
Li-silicate) after immersion in a IN LiOH solution oversaturated with respect to Ca(OH)2
(solid lime in excess). These authors suggested that ASR is inhibited and perhaps stopped
in LiOH due to the formation of a protective layer of such a non expansive lithium-bearing
silicate on the surface ofthe reactive aggregate particles. Yin and Wen [18] also observed a
layer of a product different from the classical ASR gel on the surface of opal particles
immersed in LiOH, which could hypothetically acts as a protective barrier. Lawrence and
Vivian [13] concluded that a lithium silicate likely forms in the presence of LiOH and opal,
and that this product is less soluble and more stable than the classical ASR gel (which
theory partly related to mechanism D below). Due to its stability, these authors thus
suggested that this product may form em insoluble coating that protects the reactive silica
from further attack by other alkali hydroxides.

1.4 Mechanisms D and E - Chemical stability of reactive silica almost unchanged


or remaining significant but concrete expansion significantly reduced or
suppressed due to formation of a non or less expansive Si-Li reaction product,
crystalline (mechanism D) or amorphous (mechanism E)

The most common mechanism proposed to explain the effectiveness of lithium against
ASR is that the swelling capacity of the ASR gel formed is modified in the presence of
206

lithium ions, which gel would be non or less expansive. This product could be either
crystalline or amorphous (i.e. a non or less expansive gel).

Sakaguchi et al. [19] measured the concentrations of alkalis (K, Na, and Li) in the pore
solution of mortar bars expressed under high pressure. For mortar bars containing lithium,
they observed that Li ions were consumed in greater proportion than K and Na ions, and
that the [Li] decreased with time while the [K] and [Na] remained constant. On the other
hand, the [K] and [Na] decreased with time in control mortar bars without lithium. The
authors [19] thus suggested the formation of a non expansive lithium-bearing product in
place of a more expansive product containing greater amounts of Na and K. In the study by
Diamond and Ong [20], mortar bars incorporating LiOH and reactive silica (Beltane opal
and cristobalite) did not expand but were showing gel when observed under ultraviolet light
after being treated with uranyl acetate. This reaction product was then considered non
expansive. This study [20] thus also suggests that the effectiveness of lithium against ASR
is related to the formation of a non expansive (amorphous or crystalline) lithium silicate in
place ofthe classical expansive silico-calco-alkaline gel formed in the absence of lithium.
According to Mo et al. [21], because having a smaller (unhydrated) ionic radius and a
higher surface charge density, the Li+ ions are more readily incorporated in the ASR
reaction product than the Na+ and K ions, and the lithium-bearing ASR product formed is
crystalline and non expansive. Kurtis and Monteiro [22] advanced that the beneficial effect
of lithium against ASR expansion is attributed to reduction in the surface charge density of
the alkali-silicate gel formed in the presence of lithium. In the presence of LiN03, Collins
et al. [15] observed the formation of a silica-rich product which could be non expansive. In
a solution of 0.7N NaOH containing LiOH at different [Li]/[Na], Yin and Wen [18] also
concluded that the dissolution of silica from opal grains was reduced with respect to a
lithium-free solution and, based on SEM observation, that a compact protective layer made
of a lithium-silica gel was formed on the surface ofthe opal grains.

The very well-established electrical double-layer theory (EDL) controls the adsorption of
ions near/on the surface of any particle. The colloidal particles of ASR gel are assumed to
be negatively charged, therefore tend to be surrounded by a positively charged electrical
double layer, the thickness of which depending on the surface charge, the ionic composition
207

of the solution and the cations adsorbed. Prezzi et al. [23] used this theory to explain the
expansion due to the ASR gel, then to explain the effect of various chemical additives used
to counteract ASR [24]. According to these authors, two factors would play an important
role in the swelling of a gel: (1), the valence and (2), the hydrated radius ofthe cations
present in the double layer: the higher the valence of a given cation, the lesser the gel
swells, and the larger the hydrated radius of this cation, the more the gel swells. Based on
these two parameters, the authors [24] considered that the expansion of the gel should
increase according to the following sequence of adsorbed cations: Al3+ < Ca2+ < Mg2+ < K+
< Na+ < Li+. Accordingly, Li ions should then result in a gel more expansive than gels
containing Na or K, which clearly does not agree with the well-known beneficial effect of
lithium. The authors [24] thus suggested that the anomalous position of Li+ in the above
sequence with respect to reality could be related to the fact that Li ions are primarily
sequestered by the cement hydrates, thus are not available for the ASR.

1.5 Mechanism F - Chemical stability of reactive silica almost unchanged or


remaining significant but concrete expansion significantly reduced or
suppressed due to a higher solubility of silica, which thus mostly remains in
solution without forming an expansive gel

Another possible mechanism is that Li ions significantly increase the solubility of the
reactive silica in the concrete pore solution. Using an X-ray microscope to observe the ASR
occurring in solutions containing lithium or not, Kurtis and Monteiro [22] observed that a
lesser amount of gel was formed in the presence of lithium. The authors thus proposed that
when lithium is present in sufficient quantity, the repulsion between the colloidal particles
of silica remains high, thus preventing their polymerization and the consequent formation
of an expansive silica gel.

This experimental study was undertaken with the main objective of determining the most
probable mechanism(s) explaining the beneficial effect of LiN03 in controlling expansion
due to ASR. A better understanding of the actual mechanisms involved would greatly help
in making decisions regarding the use of L1NO3 in concrete incorporating aggregates
susceptible to ASR, taking particularly into account the nature of the reactive aggregate to
counteract.
208

2 Materials and methods

Various series of experiments have been carried out in the presence of LiN03 or not
(controls) (1) on concrete specimens incorporating various aggregates of different
compositions and degrees of reactivity (pore solution chemistry and microanalysis), (2) on
pure cement pastes (pore solution chemistry), (3) on composite specimens made of cement
paste and a variety of reactive aggregates (immersion tests in NaOH+LiN03 solutions), and
(4) on particles of a variety of reactive aggregates, silica species, silicates, and volcanic
glass (immersion tests in various alkaline and Li-bearing solutions). Lithium was used as:
(1) a solution containing 30 wt.% LiN03 (Euclid Lifetime N), and (2) a solid pulverized
product containing at least 98 wt.% of LiOH-hbO (Sigma-Aldrich).

2.1 Pore solution chemistry of cement paste and concrete specimens (test series
no. 1)

As described elsewhere, cement paste cylinders, 40 mm in diameter by 65 mm in length


[12] and concrete prisms, 75 by 75 by 300 mm in size [9], were made with or without
LiNÛ3 (aqueous solution added to the mixture water), using a constant water-to-cement
ratio of 0.42, different [Li]/[Na+K] molar ratios (0 for controls, 0.56, 0.74, 1.11), different
cements (0.51, 0.82, 1.25 wt.% Na20e), and different alkali contents with respect to the
cement (0.51 wt.% Na20e using the low-alkali cement; 1.25 wt.% Na20 e using the low-
alkali or the medium-alkali cement with NaOH or KOH addition, or the high-alkali
cement). Twelve ASR-reactive natural aggregates (4 gravels, two siliceous limestones, two
greywackes, one dolostone, one chloritic schist, one rhyolite, and one granitic rock) and
one non reactive limestone aggregate, 5 to 20 mm in size, were used for making the
concrete specimens according to CSA A23.2-14A [25] (similar to ASTM C 1293 [26]). The
cement paste and concrete specimens were stored above water in sealed containers at
various temperatures (23, 38, and 60°C for cement pastes; 38 and 60°C for concretes). At
different times (3, 7, 28, and 91 days for cement pastes; up to one year for concretes stored
at 38°C, and up to 91 days for concretes tested at 60°C), the pore solution was expressed
under high pressure from the specimens (1000 MPa for cement pastes; 1400 MPa for
concretes) using the procedure first described by Longuet et al. [2] and commonly used
209

since then. The equipment used for expression in this study had a cylindrical chamber of
50 mm (1.97 in) in diameter by 90 mm (3.54 in) in length and the pore solutions were
expressed from broken pieces of concrete or from entire paste cylinders. The solutions were
diluted by 100X (20X for the analysis of Si) and chemically analyzed for Na, K, Li, and Si,
by ICP-OES (model Perkin Elmer). Taking in account the dilution factor, the minimum
detection limits are estimated to 0.4, 5, 10, and 0.5 ppm for Si, Na, K, and Li, respectively.
The precision are estimated to 0.007, 0.009, 0.005 and 0,030 mol/1 for Si, Na, K and Li.
The OH" concentration was also determined by direct titration with standardized HC1 down
to the phenolphtalein end point, and the pH values were calculated accordingly.

2.2 Visual examination and microanalysis of concrete specimens (test series no. 2)

2.2.1 Visual examination and SEM observations

At different times/expansions, the concrete specimens tested in series no. 1 were visually
examined under the stereomicroscope. Representative fragments of several concretes
incorporating various reactive aggregates and containing LiN03 or not, were then examined
in more details under the scanning electron microscope (SEM) (model JEOL JSM 840) at a
beam voltage of 15 kV and using an EDXA (energy dispersive X-Ray analyzer) for
chemical analysis.

2.2.2 Visual examination under ultraviolet light after uranyl acetate treatment

After 2 years at 38°C or 6 months at 60°C, a large number (42) of the above concrete
prisms, made with 12 reactive and one non-reactive aggregates, and incorporating LiN03 at
a [Li]/[Na+K] of 0.74 or 0 (control concretes) [9], were broken perpendicular to their length
in order to obtain a quite even broken surface which was then wetted with a solution of
uranyl acetate, allowed to stand for a few minutes, then thoroughly washed with running
water to remove the excess of solution, and observed under ultraviolet light, following
procedure AASHTO T 299-93 [27]. In this test, alkalies present in amorphous gel from
210

ASR are removed and replaced by uranyl ions then allowing the gel to exhibit a green
fluorescent color when exposed to ultraviolet light.

2.2.3 SIMS microanalysis

After 2 years at 38°C, polished sections of two concretes were analyzed using the Micro-
SIMS apparatus (Secondary Ion Mass Spectrometer equipped with a microbeam) of
CANMET (Ottawa, Canada), for assessing the spatial distribution of lithium in concrete
and particularly the presence of Li-bearing reaction products of any type (i.e. crystalline or
amorphous) or composition (e.g. calcium content). These two concretes, both containing
LiNÛ3 at a [Li]/[Na+K] of 0.74 and not expanding after 2 years, were made with a
non-reactive pure limestone and a highly-reactive rhyolite, respectively. In this technique,
the emission signals correspond to ions that volatilized from the surface of the specimen
tested under the destructive action of the ionic source beam. A Ga beam was used in this
study at a current of 2 or 5 nA. The scanning mode (8 scans per image) was used to obtain
distribution maps for Li, Si and Ca over the selected scanned area of the two polished
concrete sections. Under the conditions used, the lateral resolution is much lower than
1 micron and the minimum limit of detection is much lower than 1 ppm.

2.3 Composite specimens of aggregate particles and cement paste immersed in


NaOH ± LiNOj solutions (test series no. 3)

2.3.1 Polished aggregate particles embedded in a LiNO^-bearing cement paste, with the
composite specimens immersed for 28 days in a NaOH+LiNOi solution at 80°C

Particles from one glassy and highly-reactive volcanic rock (obsidian), one highly-reactive
aggregate (rhyolite), and one moderately-reactive aggregate (granitic rock), 10/14 mm in
size, were sawed and polished on one face (using a diamond paste of 15 pm at the last
polishing step), then embedded in cement paste containing LiNÛ3 at a [Li]/[Na+K] of 0.74.
The composite specimens were then immersed at 80°C in a IN NaOH + 0.74N LiN03
solution ([Li]/[Na] = 0.74). After 28 days, the specimens were broken along the original
polished aggregate interface, which was treated with a 10 wt.% HC1 solution to remove the
residual cement paste adhering to the surface. Both the polished aggregate surface and the
211

cement paste surface in immediate contact were analyzed by X-ray diffraction (XRD)
(model Philips PW1080) at a voltage of 40kV, a current of 20 mA, using a Cu source
(1.54059A), a continuous scan of l°(20)/mn from 10° to 60° or 80°, and the software Jade
2.0 for data processing. Afterwards, all surfaces were examined under the SEM.

2.3.2 Polished sections taken from prisms made of aggregate particles and cement paste
and immersed for 28 days in NaOH and NaOH+LiN0 3 solutions at 80°C

Sections, 25 mm in thickness, were sawed perpendicularly to the length of prisms, 75 by 75


by 300 mm in size, made of cement paste and 10/14 mm particles of three reactive
aggregates (rhyolite, Spratt limestone, and greywacke). These sections were polished on
one (sawed) face, immersed in IN NaOH (control) and IN NaOH + 0.74N LiN03
([Li]/[Na] = 0.74) solutions at 80°C, periodically examined under the stereomicroscope,
and analyzed under the SEM after 28 days.

2.4 Aggregate particles immersed in Li-bearing solutions (test series nos. 4a to 4d)

2.4.1 Polished aggregate particles immersed in a 1N LiOH solution for J hour at 350°C
(in autoclave at 300 psi or 2070 kPa) or for 6 months at 80°C (only obsidian,
unpolished) (test series no. 4a)

To better visualize and to accelerate at the same time the chemical interactions taking place
between reactive aggregates and lithium-based admixtures, particles of a highly-reactive
volcanic glass (obsidian) and of 8 natural aggregates (see Table 7.1), were sawed, polished
on one (sawed) face, and immersed for 1 hour in a IN LiOH solution in the autoclave at
350°C (300 psi or 2070 kPa). In addition, one obsidian particle, unpolished but with a quite
even surface, was immersed for 6 months in a IN LiOH solution at 80°C. The polished
surfaces were analyzed by XRD before and after immersion, with the obsidian particle
immersed for 6 months in a IN LiOH at 80°C also observed under the SEM.
212

2.4.2 Polished aggregate particles immersed for 28 days in a IN NaOH + 0.74N LiNO;
solution at 80°C (test series no. 4b)

Polished particles of the same 8 natural aggregates (see Table 7.1) were also tested for
28 days in a IN NaOH + 0.74N LiN0 3 solution ([Li]/[Na] = 0.74) at 80°C. The polished
surfaces were analyzed by XRD before and after immersion.

2.4.3 Aggregate powders (150-300 pm) immersed for 28 days in various Li-bearing
solutions at 60°C (test series no. 4c)

The obsidian and 3 reactive aggregates (see Table 7.2) were finely ground. 20-g samples of
150-300 pm particles were immersed for 28 days in 40 ml ofthe following solutions at
60°C: (1), 0.38N NaOH + 0.30N KOH (control); (2), IN LiOH; (3), 0.8N (Na,K)OH +
0.28N LiN0 3 ([Li]/[Na] = 0.35), and (4), 0.8N (Na,K)OH + 0.56N LiN0 3 ([Li]/[Na] =
0.74). The test containers were shaken manually for 1 min every day for the first 5 days,
then every 3 days until 28 days. After 1, 3, 7, and 28 days, the solutions were sampled
(1 mL) and chemically analysed for Na, K, and Li, by ICP-OES.

2.4.4 Dissolution of particles of volcanic glass (obsidian), silica varieties, silicate


minerals, and natural aggregates immersed for 28 days in various Li-bearing
solutions at 80°C (test series no. 4d)

Particles of various reactive and non-reactive materials (see Table 7.3) were immersed for
28 days in the following solutions at 80°C: (1) IN NaOH (control); (2), IN LiOH; (3),
IN NaOH + 0.25N LiN0 3 ([Li]/[Na] = 0.25), and (4), IN NaOH + 0.50N LiN0 3 ([Li]/[Na]
= 0.50). Opal and chalcedony particles were also immersed in IN NaOH with other LiN03
concentrations (i.e. 0.1, 0.2, and 0.3N for chalcedony; 0.75N and 1.0N for opal), with a few
particles also transferred in the IN NaOH control solution after 28 days in a
IN NaOH + 0.5N (chalcedony) or 1.0N (opal) LiNÛ3. Particles of a number of materials
were also tested in Li-bearing solutions in the presence of portlandite. Only one particle,
weighing 1.0 g ± 0.1, was tested per container with 200 ml of test solution, supported on a
plastic grid. After deeming its surface with distilled water and drying for 3 h at 80°C (which
213

was always sufficient to reach mass equilibrium), its mass was periodically measured up to
a maximum of 28 days, depending on the test solution used and the kinetics of dissolution,
with a precision of 0.0001 g. XRD was also performed after the tests on reacted particles of
opal and chalcedony.

3 Results and discussion

This study involved a large number of very different tests series, each of them involving
very different materials (e.g. concrete, cement paste, pure aggregates), experimental
conditions, and analytical techniques. In order to prevent any confusion, the results
obtained for each series are presented and immediately discussed one series by one with
respect to the various mechanisms proposed above to explain the effectiveness of LiNÛ3
against ASR. Accordingly, for each test series (no. 1 to 4d), each mechanism (no. A to F)
will be considered "likely", "possible", "non-conclusive", "unlikely", or "impossible",
based on the results obtained. The conclusions drawn in this respect are summarized in
Table 7.4, along with the rationale behind.

3.1 Pore solution chemistry of cement paste and concrete specimens (test series
no. 1)

The main conclusions drawn from the chemical analyses of the pore solution of the cement
paste and concrete specimens (details given in Bérubé et al. [12] and Tremblay et al. [9],
respectively) are the following:

The amount of Si in solution was always low with no significant differences


between control and Li-bearing specimens. For instance, for the 8 sets of control
concretes and concretes containing LiN03 at a [Li]/[Na+K] of 0.74, incorporating
4 different reactive aggregates (two siliceous limestones, one polygenic gravel, and
one chloritic schist), stored at 38°C, and tested at different expansion levels (from
28 up to 365 days of storage at 38°C), the Si concentration averaged 0.0153 and
214

0.0135 mol/L for the control and Li-bearing concretes, respectively. On average, the
[Si] is only slightly lower in the presence of LiN03;

In the presence of LiN03 and after cement hydration has largely completed, i.e.
after 91 days and even earlier, the [Li]/[Na+K] in the pore solution of all specimens
investigated ranged between 0.30 and 0.45, which is about half of the original
mixture ratio of 0.74. In fact, Li ions are incorporated in greater proportion than Na
and K ions into the cement hydrates [12]. Consequently, the [Na+K] in the pore
solution was significantly higher in the presence of LiNÛ3 with respect to the
control mixtures, while the pH was just slightly lower, by about 0.1. Nevertheless,
the expansion due to ASR was largely reduced in the concretes incorporating
reactive aggregates [9];

In the presence of LiN0 3 at a [Li]/[Na+K] of 0.74, for all concretes tested and
incorporating a variety of aggregates (non-reactive pure limestone, moderately-
reactive Sudbury gravel for which LiN03 was very effective, moderately-reactive
chloritic schist for which LiNÛ3 was poorly effective (i.e. at the standard dosage of
0.74 and even more), and highly-reactive Spratt limestone for which LiN03 was
also poorly-effective), the [Li] in the pore solution decreased continuously with
time, much more rapidly in the presence of reactive aggregates, and even after
91 days, i.e. after cement hydration has largely completed [9].

3.1.1 Discussion

The results of test series no. 1 clearly indicate that, since the pH in the pore solution is not
significantly reduced in the presence of LiN03 and that the Si concentration in the pore
solution is not modified in the presence of LiN03, mechanisms A and F are not possible.
Considering that all other test series did not involve any measurement of pH and Si
concentrations in solution, these two mechanisms will not be considered in the following
until the general discussion of section 3.8. For its part, the significantly lower [Li] observed
in the presence of reactive aggregates suggests the formation of some Si-Li reaction
products, which could be crystalline or amorphous, expansive or not, and/or protective
against ASR. Consequently, the mechanisms C (protective product), D (non expansive
215

crystalline product) and E (non to less expansive gel) remain possible based on this set of
data.

3.2 Visual examination and microanalysis of concrete specimens (test series no. 2)

3.2. J Visual examination and SEM observations

Significant amounts of characteristic reaction products from ASR (siliceous gel and
microcrystalline products) were observed visually and under the SEM in the control
concretes showing high expansion, i.e. in the absence of LiN03. On the other hand, a gel
containing Si, Ca, Na, and K, and possibly also Li (lithium cannot be detected in SEM as its
characteristic X-ray wavelength is out of most EDS and WDS detectors' range), was the
only reaction product observed in the LiN03-bearing concretes, but always in much lower
amounts than in corresponding control concretes, without any noticeable difference in
morphology and (detectable) chemical composition with respect to the classical ASR gel
(observed in the control concretes). Moreover, no other type of reaction product was
observed at the surface of the reactive aggregate particles (as a potentially protective
coating) or elsewhere in the cement paste. These observations apply, whatever the reactive
aggregate used and the effectiveness of LiN03 in its respect.

3.2.2 Visual examination under ultraviolet light after uranyl acetate treatment

The amounts of alkali-silica-gel observed under ultraviolet light on the broken sections of
control and Li-bearing concretes after the application of uranyl acetate always seemed to be
proportional to concrete expansion (due to ASR), whatever the reactive aggregate in
concrete. This suggests that this gel is similarly expansive in both control and Li-bearing
concretes, but present in much lower amounts (by visual examination under ultraviolet
light) in the latter. For instance, for all concretes made with LiN03 and showing no or
limited expansion, no or only traces of reaction gel were observed under the ultraviolet
illumination after uranyl acetate treatment. These results disagree with Diamond and Ong
216

[20] who observed, after uranyl acetate treatment, that significant amounts of reaction gel
were present in non-expanding mortar bars containing lithium, however in the form of
LiOH.

3.2.3 SIMS microanalysis

The micro-SIMS analyses showed that the atomic element Li was regularly distributed
everywhere within the cement paste of the two concretes examined, containing reactive and
non-reactive aggregates, respectively, without any particular/noticeable local
concentrations in the cement paste, in the aggregates or along the aggregate/cement paste
interfaces (Figure 7.1).

3.2.4 Discussion

It is likely that most ofthe lithium detected with the micro-SIMS has precipitated from the
pore solution when the test specimens were dried for analysis. In other words, it was not
possible to detect the presence of Li-bearing reaction products in the concretes tested,
(which would correspond to areas with a higher brightness on the Li images of Figure 7.1,
if present), even in the concrete made with the highly-reactive rhyolite and for which
LiN03 was very effective in controlling expansion due to ASR [9].

The results of test series no. 2 from all visual and microscopic examinations on concrete
specimens are in total agreement with each other. They do not support mechanisms D and E
(which would call, in the presence of lithium admixtures, for the formation of significant
amounts of non to less expansive crystalline or amorphous Li-bearing reaction products),
and mechanism C as well (since the presence of a protective reaction product was not
observed/detected at the surface of the reactive silica grains or aggregate particles, even
when using the micro-SIMS technique). They rather support mechanism B. i.e. silica
dissolution significantly reduced/suppressed but not due to mechanism A (since the pH is
not affected by LiN03) or C (since a protective product was not observed). Indeed, in the
presence of lithium and reactive aggregates, the lower the concrete expansion due to ASR,
217

the lower the amount of (likely expansive) reaction gel observed in concrete, which
suggests in turn that the reactive silica is more stable in the presence of lithium.

3.3 Composite specimens of aggregate particles and cement paste immersed in


NaOH ± LiN03 solutions (test series no. 3)

3.3.1 Polished aggregate particles embedded in a LiNOybearing cement paste, with the
composite specimens immersed for 28 days in a NaOH+LiNOi solution at 80°C

Reaction products (crystalline and amorphous) were not detected by XRD neither observed
under the SEM on the polished surface ofthe reactive aggregate particles tested nor on the
surface of the cement paste in contact with the polished aggregate surfaces during
immersion.

3.3.2 Polished sections taken from prisms made of aggregate particles and cement paste
and immersed for 28 days in NaOH and NaOH+LiNÛ3 solutions at 80°C

Typical ASR gel was observed on the exposed polished surfaces of all reactive aggregate
particles after immersion in the NaOH control solution. However, reaction products
(crystalline or amorphous) were not observed after immersion in the LiN03-bearing
solution.

3.3.3 Discussion

The results of test series no. 3 do not support mechanisms C, D, and E, since reaction
products were not observed along the aggregate/cement paste interfaces (first sub-series of
tests) neither on the polished sections of aggregate particles and cement paste in direct
contact with the Li-bearing solution (second sub-series of tests). However, in this particular
test series, the absence of any Li-Si reaction product (which does not support mechanisms
C, D and E), rather supports mechanism B, i.e. increased stability of silica, particularly
when considering that the specimens from the first sub-series were real hydraulic mixtures
(i.e. incorporating cement and portlandite) for which it has been already proved that the pH
218

is not significantly reduced (mechanism A impossible) and the solubility of silica is not
affected by the presence of LiN03 (mechanism F impossible).
219

3.4 Aggregate particles immersed in a IN LiOH solution (test series no. 4a)

3.4.1 Polished aggregate particles immersed for 1 hour in a IN LiOH solution at 350°C
(in the autoclave)

Table 7.1 gives the mineral phases detected by XRD, before and after autoclaving, on the
polished surface of the 8 natural aggregates tested. After autoclaving, a crystalline
Li-silicate, Li2Si03, was detected by XRD on the polished surfaces of obsidian (Figure 7.2)
and the 3 reactive aggregates tested for which LiN03 proved to be effective in concrete, i.e.
rhyolite (see Figure 7.3), dolostone, and granitic rock, using a [Li]/[Na+K] < 0.74 [9]
(Table 7.1). L_2Si03 was also detected, however just in traces, on the surface of one (i.e.
clayey limestone) of the 4 other reactive aggregates tested, for which a greater LiN03
content was needed in concrete to control ASR expansion, i.e. a [Li]/[Na+K] > 0.93 or
more [9] (Table 7.1); it was not detected, however, for the three other aggregates (Spratt
limestone, greywacke, chloritic schist) (Table 7.1). Secondary portlandite was also detected
after autoclaving on the surface of all aggregates containing calcite in large amounts (i.e.
Spratt limestone, clayey limestone, chloritic schist, and the non-reactive pure limestone), a
Li-carbonate being even found on the pure limestone (Table 7.1). One could argue that a
relatively thick layer of portlandite could explain that Li-silicate was not detected by XRD
or just detected in traces in the case of 4 of the 7 reactive aggregates tested; however, one
of these aggregates (greywacke) does not contain significant amounts of calcite while not
showing portlandite after autoclaving.

3.4.2 Obsidian particle immersed for 6 months in a IN LiOH solution at 80°C

After immersion, the obsidian surface was covered with a layer of crystalline Li2Si03
(Figure 7.4), identified by XRD. The presence of a continuous void between the Li2S03
layer and the obsidian surface, which appears very smooth and microtexturally-unchanged
with respect to the original material (Figure 7.4), suggests that this "coating" product
precipitated from the soak solution after some silica was dissolved rather to be a residual
product left aside after the soak solution reacted with the obsidian. On the other hand, the
absence of any significant sign of corrosion on the obsidian surface suggests that the
dissolution of this material proceeded very regularly and at a very-fine scale all over the
220

surface exposed to the soak solution. A finely-dispersed white product (not analyzed) was
also observed at the bottom of the test containers, which may be crystalline or amorphous
in nature, and expansive or not.

3.4.3 Discussion

The results from test series no. 4a suggest that a crystalline Li-silicate (Li2S03) can form in
the presence of reactive aggregates for which LiN0 3 is effective in controlling concrete
prism expansion using a [Li]/[Na+K] of 0.74 or less, however not in the presence of
reactive aggregates for which LiN03 is needed in greater proportions for expansion control.
They thus support, to some extent, mechanisms C and D (crystalline Si-Li reaction product
observed on obsidian and 4 of the 7 reactive aggregates tested, which could be also
protective). However, with respect to LiN03, these mechanisms are considered possible but
uncertain considering that all tests involved special testing conditions (i.e performed in a
IN LiOH solution and at 350°C for most of them).

3.5 Polished aggregate particles immersed for 28 days in a IN NaOH + 0.74N


LiN03 solution at 80°C (test series no. 4b)

Polished particles ofthe 8 natural aggregates tested in the autoclave were also immersed for
28 days in a IN NaOH + 0.74N LiN0 3 solution at 80°C. For each aggregate, the XRD
results were exactly the same before and after immersion, no crystalline Li-bearing product
being detected on the polished surfaces.

3.5.7 Discussion

These results from test series no. 4b do not support at least mechanism D (since the
presence of a crystalline product was not detected by XRD). Moreover, no other types of
reaction product were observed in the test solutions, neither at the bottom of the test
containers. At this stage of the study, it is thus possible that the crystalline Li-silicate
observed in test series no. 4a just forms in LiOH (e.g. for obsidian in the long term at 80°C)
221

and under special conditions for some reactive aggregate particles (i.e. very high
temperature and pressure in the autoclave). In the present experiment, there is a possibility
that an amorphous Li-bearing gel, undetectable by XRD, formed at the surface of the
polished particles rather than a crystalline product. Since the particles tested were not
examined after the tests (e.g. using the SEM), one cannot thus conclude about mechanism
B (amorphous gel could be absent), mechanism C (amorphous gel could be present which
may act as a protective coating), and mechanism E (amorphous gel could be present, which
could be non to less expansive).

3.6 Aggregate powders (150-300 pm) immersed for 28 days in various Li-bearing
solutions at 60°C (test series no. 4c)

When aggregate powders (150-300 pm) were immersed for 28 days in (Na,K)OH control
solutions at 60°C, the K and Na concentrations in the (Na,K)OH control solution were quite
constant in the presence of all 4 materials tested, even the most reactive obsidian (Figure
7.5A). For the materials immersed in the LiOH solution, the [Li] in solution decreased for
each material, however more rapidly for obsidian (Figure 7.5B). As illustrated in
Figures 7.5C and 7.5D for rhyolite, but also for each of the materials tested, the [Na] and
[K] in the two (Na,K)OH + LiNÛ3 solutions were quite stable over time, while the [Li]
significantly decreased, more rapidly in the solution with higher concentration of LiN03
(Figures 7.5C vs. 7.5D). The consumption of Li after 28 days is given in Table 7.2. In the
presence of lithium, a finely-dispersed whitish product was always observed in the test
containers.

3.6.1 Discussion

These results are in good agreement with those from pore solution chemistry of concretes
(Section 3.1) where Li ions were consumed more rapidly than Na and K ions; the only
difference is that in the concretes tested, Na and K ions were also progressively consumed
with time (i.e. disappeared from the pore solution), due to alkali-leaching and to formation
of more Li-Na-K-bearing cement hydrates, and expansive gel due to ASR as well. In the
222

present case, cement was not present in the system and Li was still progressively consumed
with time.

The results concerning the Li consumption (Table 7.2) clearly suggest that a secondary
Li-bearing reaction product is formed. As mentioned above, a finely-dispersed whitish
product was effectively always observed in the presence of lithium. The largest quantity
was obtained for the obsidian immersed in the LiOH solution and also corresponds to the
largest Li decrease in solution (~ 83wt% after 28 days; Table 7.2). Unfortunately, this
product was not chemically analyzed neither analyzed with the SEM or by XRD; it could
thus be crystalline or amorphous but it likely contains, in addition to Li, some Si, Na,
and K (the system containing substantially these four elements). The important difference
observed between series 4b (where no reaction product were detected) and series 4c (where
a reaction product was always present in the presence of lithium solution) is likely
attributed to extreme differences in specific area (i.e. single polished aggregate particles vs.
aggregate powders).

It is not possible to conclude about a possible influence ofthe reactive aggregate tested; for
instance, the amounts of lithium consumed after 28 days are quite similar for the highly
reactive rhyolite and the moderately reactive chloritic schist (reactivity level based on
expansion in control concretes), despite the fact that LiN03 was very effective in
controlling concrete expansion with the rhyolite (effective [Li]/[Na+K] = 0.63) and less
effective with the schist (effective [Li]/[Na+K] > 0.93); Table 7.1).

The results from test series no. 4c suggest the formation of a Si-Li reaction product in the
form of a finely-dispersed whitish reaction product. In the absence of any XRD analysis
and SEM examination, this product could be crystalline or amorphous, expansive or not,
and could act as a protective coating. Mechanism C (protective reaction product),
mechanism D (crystalline and non expansive reaction product), and mechanism E (non to
less expansive amorphous gel), thus remain possible. For its part, mechanism B appears
doubtful (unlikely) based on the results obtained in this test series considering that a
reaction product is always formed, which most likely contains silicium, in addition to Li.
223

3.7 Dissolution of particles (~lg) of volcanic glass (obsidian), silica varieties,


silicate minerals, and natural aggregates immersed for 28 days in various Li-
bearing solutions at 80°C (test series no. 4d)

3.7.1 Immersion in IN NaOH control solution

All reactive materials immersed for 28 days in the control solution at 80°C presented
significant mass losses (Table 7.3). The extremely reactive opal (amorphous silica) and the
three highly reactive cryptocrystalline varieties of quartz (chalcedony, red chert, green
chert) were completely dissolved after 2, 14, 15, and 15 days, respectively. For all materials
tested, dissolution occurred without the formation of any reaction product, thus suggesting
that the dissolved silica remained in solution, likely because ofthe high quantity of solution
involved (not saturated with respect to silica) and/or the absence of calcium in the system.
Nevertheless, when the percentage of dissolution was important (in the absence of lithium),
a translucent colloidal material, likely a silica gel, was generally observed in the test
solution.

3.7.2 Immersion in Li-bearing solutions

On the other hand, the presence of lithium in the soak solution reduced the dissolution of all
reactive materials tested (except obsidian), often dramatically, in fact for most reactive
aggregates tested and the three varieties of cryptocrystalline quartz, i.e. chalcedony, red
chert, and green chert (Table 7.3). In the presence of lithium and reactive material, a finely-
dispersed whitish product, amorphous according to XRD and likely containing Li and Si
(the system containing substantially only these two elements), was always observed at the
bottom ofthe test containers, unless no dissolution (i.e. mass loss) was observed; the higher
the dissolution, the more abundant this product.

For chalcedony, a [Li]/[Na] of 0.2 in the soak solution (i.e. IN NaOH + 0.2N LiN0 3 )
decreased the dissolution from 100wt% (complete dissolution in the NaOH control
solution) to llwt% after 28 days, while a ratio of 0.3 was almost sufficient to largely
prevent the dissolution of the particle (Figure 7.6A). This value of 0.3 is close to the
224

[Li]/[Na+K] found at equilibrium in the pore solution of a concrete made with a


[Li]/[Na+K] mixture ratio of 0.74 [9]. Crystalline Li 2 Si0 3 was detected by XRD on the
chalcedony surface even when large amounts of lithium were present in the system
(Figure 7.7). However, it was not possible to observe this Li-silicate when using a 10X
magnification lens, likely because the coating was very thin. This assumption is supported
by the fact that the underlying chalcedony (cryptocrystalline quartz) is also well-detected
by XRD (Figure 7.7). Based on these results, i.e. no significant dissolution in the presence
of sufficient quantity of LiN03, and reaction products just in tiny amounts and likely too
thin to act as an effective physical barrier, it is very likely that the presence of lithium in
sufficient amounts was very effective in largely suppressing the dissolution of this highly
reactive phase. As observed in the presence of opal (next paragraph), even a thick reaction
product is even not sufficient to act as en effective physical barrier against ASR. After 28
days in the IN NaOH + 0.5N LiN03 solution, a particle of chalcedony, mostly unreacted
(no significant dissolution in this solution) and just covered with a very thin layer of
Li2Si03, was immersed in the IN NaOH control solution; the particle then started to
dissolve at a rate (~12wt% per day, with complete dissolution after 8 days; Figure 7.6C)
very similar to the dissolution rate of the chalcedony particle initially immersed in the
NaOH control solution (~10wt% per day for the first 8 days; Figure 7.6A). This confirms,
once again, that chalcedony does not significantly react or dissolve when the Li
concentration is sufficiently high in solution.

It was more difficult to prevent dissolution in the case of opal, by far the most reactive
material tested. In the IN NaOH + 0.25N LiN0 3 solution ([Li]/[Na+K] = 0.25), the opal
particle completely dissolved after 2 days (Figure 7.6B) and the finely-dispersed whitish
product mentioned above was observed in significant amounts at the bottom of the test
containers. In the IN NaOH + 0.5N LiN0 3 solution ([Li]/[Na+K] = 0.5), the mass ofthe
particle tested decreased by about 70wt% after 3 days, then remained constant afterwards; a
finely-dispersed whitish reaction product was also observed, however in relatively smaller
amounts. Interestingly, a naked-eye visible compact layer of secondary reaction product,
also observed by Yin and Wen [18], formed around the reacting particle. This product was
found to be amorphous based on XRD. In actual fact, the opal completely dissolved after
about 3 days, then leaving a particle essentially made of an empty shell of amorphous
225

reaction product weighing about 30% of the original mass (overall mass loss of 70wt%;
Figure 7.6B). The same observations were made for the particle immersed in the
IN NaOH + 0.75N LiN0 3 solution ([Li]/[Na+K] = 0.75) except that: (1), the opal
completely dissolved after about 7 days, (2), leaving a residual empty shell of reaction
product corresponding to about 55wt% of the initial mass (overall mass loss of 45wt%;
Figure 7.6B), and (3), the amount of finely-dispersed whitish product was again smaller. In
the IN NaOH + IN LiN0 3 solution ([Li]/[Na+K] = 1.0), a mass gain of over 60wt% was
rather observed. In this particular case, the finely-dispersed whitish product was absent; the
opal again dissolved completely but the totality of the dissolved silica precipitated as the
reaction shell. The 60wt% increase in mass is explained by the fact that this shell is
composed of a hydrous and hydrated amorphous material which, in addition to the totality
of the silica from the reacted opal, incorporated other substances from the solution (e.g.
lithium, OH radicals, water molecules, and possibly also some NO3). The two amorphous
products formed (residual shell and finely-dispersed whitish product) have not been
chemically analysed, but they could correspond to the same reaction product. It is thus
obvious that the layer of reaction product around the original opal particle did not protect
the underlying opal against dissolution; nevertheless, Figure 7.6B suggests that the
more important the thickness of this layer (which progressively increases from 0.5
to 1.0N L1NO3), the slower the rate of dissolution. After 28 days in the
IN NaOH + 1.0N LiN0 3 solution ([Li]/[Na+K] = 1.0), the residual opal particle, in fact
totally composed of an amorphous reaction product, was immersed in the IN NaOH control
solution (Figure 7.6C). The particle progressively dissolved, however at a much lower rate
(<10wt% per day) than the opal particle initially immersed in the control solution (100wt%
dissolution after two days). This demonstrates that the reaction product also remains highly
susceptible to dissolve in the IN NaOH solution (in the absence of lithium), however less
than the original opal. The opal thus behaved differently in many respects from chalcedony
and most other reactive materials tested (except obsidian). The presence of lithium did
reduce its dissolution, but the silica dissolved was largely and more or less rapidly
incorporated into amorphous Si-Li reaction products (surrounding shell and whitish product
finely-dispersed in the soak solution), depending upon the lithium concentration. The
higher this concentration, the lower the mobility of the silica dissolved, the higher the
226

amount of reaction product formed close to the reaction site (e.g. in the shell), and the
lower the amount of finely-dispersed whitish product in the test solution.

Obsidian, a totally amorphous volcanic glass, also performed differently from most other
reactive materials tested. As for opal, lithium does not seem to reduce dissolution (Table
7.3), which, however, never exceeded 10wt% (mass loss), and no crystalline product was
detected on its surface by XRD; the finely-dispersed whitish product mentioned above and
observed at the bottom ofthe test containers was found to be the only product formed in the
presence of lithium.

The same whitish secondary product was also observed with the three feldspars tested
(microcline, orthoclase, and anorthite) after immersion in the LiOH solution without,
however, any significant dissolution reduction with respect to the control solution
(Table 7.3). Evenmore, obsidian, anorthite, and microcline suffered more dissolution in the
IN LiOH solution than in the IN NaOH control solution (Table 7.3), which cannot be
explained for the time being.

As mentioned before, particles of a number of materials were also subjected to Li-bearing


solutions in the presence of portlandite. The results obtained were similar, then are not
presented here.

3.7.3 Discussion

Globally, the results from the test series no. 4d support mechanism B relating to
reduction/suppression of silica dissolution due to another mechanism than pH reduction
(i.e. mechanism A) or, to some extent, the early formation of a protective reaction product
on the reactive silica grains or aggregate particles (i.e. mechanism C). Indeed, most reactive
materials presented dramatic dissolution reductions in the presence of Li, except the
extremely reactive opal (lower reductions) and the highly reactive volcanic glass (obsidian),
for which the dissolution was almost unchanged in the NaOH+LiN03 solution (-7.0wt% vs.
-7.8wt% in the control solution; Table 7.3) and even increased in the LiOH solution
227

(-10.2wt%; Table 7.3). For its part, mechanism E remains possible: when silica dissolution
was not completely stopped in the presence of lithium, amorphous Si-Li reaction products
in the form of a finely-dispersed whitish product was observed, in amounts proportional to
dissolution (except for opal where two amorphous products were found (i.e. the finely-
dispersed material and the surrounding shell); however, the non to less expansive character
of this amorphous reaction product is unknown. On the other hand, mechanisms C and D
are unlikely. In the first case, when a significant coating of reaction product was observed
on the reactive particles, in fact just for opal, this coating was not protecting the underlying
reactive phase from further dissolution. In the second case, a well-crystallized Li silicate
was only observed at the surface of chalcedony, with no significant dissolution, but this
deposit was very thin.

It is not easy to draw conclusive statements about the effect of the lithium on the different
types of reactive aggregate/material tested. For instances: (1), two ofthe highest dissolution
(mass loss) values after 28 days in the control solution correspond to aggregates with one of
the highest (rhyolite: 7.8wt%; Table 7.3) and the lowest (granitic rock: 7.8wt%; Table 7.3)
expansions in concrete [9]; (2), concrete expansion due to ASR in the presence ofthe above
two aggregates is easily controlled with LiNÛ3 (effective [Li]/[Na+K] < 0.74; [9]);
nevertheless, these two aggregates also present the two highest dissolution values after
28 days in the IN NaOH + 0.25N LiN0 3 solution (1.3wt% for both; Table 7.3); (4), the
lowest dissolution value (0.1 wt%) in the same solution corresponds to an aggregate
(chloritic schist) for which LiNÛ3 is not highly effective in concrete (effective [Li]/[Na+K]
> 0.93; [9]) and so on. However, it may be of interest for further research to note that: (1),
all 5 reactive silicate aggregates tested (rhyolite, greywacke, granitic rock, chloritic schist,
and quartzitic sandstone) dissolved less or not more in the IN NaOH + 0.25N LiN03
solution than in the IN LiOH solution, despite a lower Li content, and (2), at the opposite,
the two calcite-rich reactive limestones tested dissolved less in the IN LiOH solution than
in the two NaOH+LiN03 solutions tested.

According to Feng et al. [28], the differences in the LiNÛ3 effectiveness observed for
concretes incorporating the rhyolite (effective [Li]/[Na+K] = 0.63 according to Tremblay et
al. [9], and 0.74 according to Feng et al. [28]) and the greywacke (effective [Li]/[Na+K] >
228

1.11 according to Tremblay et al. [9], and 1.48 according to Feng et al. [28]), could be
attributed to textural differences between the two types of aggregates; this would cause the
reactive silica to be more rapidly accessible from the rhyolite (which contains 85vol%
devitrified glass) than from the greywacke (which contains about 8vol% microcrystalline
quartz). However, in the present study, the two aggregates showed very similar dissolution
values in all test solutions, including the NaOH control solution (see Table 7.3 for 28-day
results).

3.8 General discussion

The most realistic and conclusive experiments carried out in this study and summarized in
Table 7.4 are those performed on concrete specimens stored in air at high humidity and
relatively low temperatures (38°C and 60°C), and incorporating natural reactive aggregates
(i.e. test series nos. 1 and 2), followed by the experiments on composite specimens made of
reactive aggregate particles and cement paste immersed in a NaOH+LiN03 solution at 80°C
(test series no. 3). Conclusions based on these tests, which strongly support or disprove one
or more of the mechanisms proposed, call for greater consideration. For their part, the
immersion test series nos. 4a to 4d were all performed in the absence of cement and at
relatively high temperatures (60, 80, or 350°C). The corresponding results must thus be
used with circumspection, particularly those obtained at 350°C (autoclave), in LiOH
solution, and/or for extremely to highly reactive materials such as opal, fused silica,
chalcedony, chert, and obsidian. Such materials are uncommon in real concretes, at least in
significant amounts, while they often performed very differently in some test series with
respect to the natural reactive aggregates tested in parallel.

Up to now, for each test series taken individually, each mechanism proposed has been
considered "likely", "possible", "non-conclusive", "unlikely", or "impossible", based on
the results obtained and the rationale summarized in Table 7.4. It remains to draw the
global conclusion described hereafter regarding each ofthe mechanisms considered.

Mechanism B, i.e. chemical stability of reactive silica increased for another reason than pH
reduction (i.e. mechanism A) or early formation of a protective coating on the reactive
229

silica grains or aggregate particles (i.e. mechanism C), is the mechanism among all those
considered in this study that better explains the effectiveness of LiN03 against ASR. This
mechanism is supported (i.e. considered "likely") by test series nos. 2, 3, and 4d, however
disproved (i.e. considered "unlikely") by the test series no. 4c (Table 7.4). This mechanism
is strongly supported by the fact that all reactive materials tested in test series 4d (including
7 reactive aggregates) dissolved much less in LiOH and NaOH + LiNÛ3 solutions than in a
NaOH control solution, with only one exception (obsidian). Mechanism A. i.e. chemical
stability of reactive silica increased due to pH reduction, and mechanism F. i.e. expansion
reduced/suppressed due to higher solubility of silica but that remains in solution, are
considered impossible, based on pore solution chemistry (test series no. 1). Mechanism C,
i.e. chemical stability of reactive silica increased due to the early formation of a protective
coating over the reactive silica grains or aggregate particles, disproved by 3 test series
while not supported by any other, is unlikely. Mechanism D. i.e. expansion
reduced/suppressed due to formation of a non expansive Si-Li crystalline reaction product,
disproved by 4 test series while not supported by any other, is also unlikely. Mechanism E,
i.e. expansion reduced/suppressed due to the formation of a non or less expansive Si-Li
amorphous gel, disproved by two test series while not supported by any other, remains
possible, at the most. The three mechanisms C, D, and E, are particularly disproved
following the examination of the concrete specimens (test series no. 2), which rather
suggests that the only reaction product formed, and this only when lithium is not highly
effective, contains lithium, looks like classical ASR gel, and is expansive as well.

It was already well established that the effectiveness of LiN03 against ASR varies from one
reactive aggregate to another, irrespective of their inherent degree of reactivity (i.e.
expansivity in concrete prism tests) [9]. An important subject of concern in this study is
how the nature of the particular reactive aggregate to counteract can affect the effectiveness
of LiNÛ3 in increasing the chemical stability of reactive silica. Unfortunately, none of the
experiments performed in this study led to conclusive results in this respect. At the very
least, the dissolution tests (series 4d) proved that the capacity of lithium in reducing the rate
of silica dissolution greatly varies with the type of reactive silica involved.
230

In numerous studies relating to ASR, extremely to highly reactive natural or synthetic


materials, which are rarely used as concrete aggregates, are often tested, such as opal,
chalcedony, chert, volcanic glass (e.g. obsidian), fused silica, pyrex, and so on. As clearly
demonstrated in the present study, such rather unusual materials may behave very
differently from natural reactive aggregates (e.g. opal and obsidian in this study),
particularly when they are tested as very fine particles. This thus recalls the importance of
testing real reactive aggregates under usual particle size distributions, to be more realistic.

4 Conclusions

Numerous sets of experiments were carried out with the objective of determining which
mechanisms(s) better explain the effectiveness of LiN03 against ASR, and variations in this
effectiveness as well with the type of reactive aggregate to counteract. These experiments
included: (1), pore solution chemistry of cement pastes and concretes incorporating LiN03
and various reactive and non reactive aggregates; (2), visual examination and microanalysis
ofthe above concretes; (3), immersion of composite specimens made of reactive aggregates
and cement paste in a NaOH + LiN03 solution, and (4), immersion of particles of various
reactive aggregates, silica phases, silicates, and volcanic glass, in various Li-bearing
solutions (LiOH, NaOH + LiN0 3 ) at various temperatures (60°C, 80°C, and 350°C in the
autoclave).

Six different mechanisms proposed to explain the beneficial effect of L-NO3 against ASR
were considered and critically examined against the results of the above experiments,
which can be grouped in two main categories:

1. Chemical stability of reactive silica increased (i.e. compared to solution without


lithium) due to: pH decrease in the concrete pore solution (mechanism A), other
change(s) in the chemistry ofthe pore solution (mechanism B), or early formation,
over the reactive silica grains or aggregate particles, of a Si-Li reaction product
which acts as a physical barrier against further reaction (mechanism C);
231

2. Chemical stability of silica almost unchanged or remaining significant, but concrete


expansion significantly reduced or suppressed due to: formation of a crystalline and
non expansive Si-Li reaction product (mechanism D), formation of an amorphous
Si-Li reaction product (i.e. a gel) that is non expansive or less expansive than the
classical ASR gel (mechanism E), or increased solubility of silica, which thus
mostly remains in solution without forming an expansive gel (mechanism F).

The following conclusions can be drawn from the results obtained in our studies:

Mechanisms A (pH reduction) and F (increased solubility of silica) are not


supported based on the pore solution chemistry results which showed that the pH
was not significantly reduced in the concrete pore solution in the presence of LiN03
and that the silica concentration in solution was always low and not affected by the
presence of LiN03;

• Mechanism C (protective coating) is considered unlikely. In the presence of


LiN03, reaction products were not observed at the surface of or somehow
surrounding the reactive phases in all investigated concretes and composite
specimens made of cement paste and reactive particles. A layer of well-crystallized
Li2SiÛ3 has been observed on the surface of obsidian and a number of reactive
aggregate particles but following immersion in a LiOH solution at 350°C in the
autoclave (also at 80°C for obsidian). However, in the presence of LiNÛ3, a (very
thin) layer of such a lithium-bearing silicate has been only observed at the surface of
chalcedony particles immersed in NaOH + LiN03 solutions at 80°C. The protective
character of this layer is doubtful, while a clearly non-protective layer of an
amorphous reaction product has been only observed at the surface of opal particles
immersed in the same solutions;

• Mechanism D (non expansive crystalline product) is also considered unlikely. In


the presence of LiN03, crystalline reaction products were not observed in any ofthe
investigated concretes and composite specimens made of cement paste and reactive
particles. In actual fact, in the presence of LiN03, as just mentioned, a likely non
232

expansive crystalline reaction product (i.e. Li2Si03) was only detected in


tiny amounts at the surface of unreacted chalcedony particles immersed in
NaOH + LiN0 3 solutions at 80°C;

Mechanism E (non to less expansive amorphous gel) is also unlikely. This


mechanism is strongly disproved by the direct examination of concrete specimens
which rather suggest that the only reaction product formed, when LiN03 is present
in insufficient amounts/dosage, contains lithium, looks like classical ASR gel, and is
expansive as well. In actual fact, an amorphous reaction product in the form of a
finely-dispersed whitish product was observed in some immersion tests, but its non
to less expansive character is unknown;

Mechanism B (increased chemical stability of reactive silica due to another


reason than pH reduction or early formation of a protective coating) is the
mechanism among all those investigated in this study that seems to better explains
the effectiveness of L-NO3 against ASR, and the variations in this effectiveness (i.e.
extent of expansion reduction) from one reactive aggregate to another. This
mechanism remains unexplained for the time being but is strongly supported by the
facts that: (1), an amorphous gel containing Na, K, and Si, and likely also Li
(undetectable under the SEM), was the only type of reaction product observed in
concrete specimens incorporating LiN03 and a variety of natural reactive
aggregates; (2), this gel was looking exactly like the classical ASR gel and its
abundance was somewhat proportional to the concrete expansion attained; for
instance, for all concretes made with LiNÛ3 and showing no or limited expansion,
no or only traces of reaction gel were observed; and (3), most reactive materials
tested, including all the 8 natural reactive aggregates tested, dissolved much less in
NaOH + LiN03 (and LiOH) solutions than in a NaOH control solution;

Some of the most reactive materials tested in this study, (e.g. opal and obsidian),
which are quite unusual in concrete aggregates, may behave very differently from
natural reactive aggregates, particularly when they are tested as very fine particles.
This arises the importance of testing natural reactive aggregates under realistic
particle size distributions;
233

It is still unknown how the presence of LiNÛ3 in the concrete pore solution can
increase the stability of reactive silica and how this phenomenon can be affected by
the particular reactive aggregate to counteract, irrespective of its pétrographie nature
and inherent reactivity (based on expansivity in control concretes).

5 Acknowledgements

Euclid Admixtures Canada Inc., the Fonds Québécois de la Recherche sur la Nature et les
Technologies of Québec (FQRNT), and the Natural Science and Engineering Research
Council of Canada (NSERC), are greatly acknowledged for their financial support.

6 References

[1] S. Diamond, ASR - Another look at mechanisms, in: K. Okada, S. Nishibayashi, M.


Kawamura (Eds.), Proceedings of the 8th International Conference on AAR, Society
of Materials Sciences, Kyoto, 1989, pp. 83-94.

[2] P. Longuet, L. Burglen, A. Zelmer, La phase liquide du ciment hydrate, Revue des
Matériaux (676) (1973) 35-41.

[3] T. Ichikawa, M. Miura, Modified model of alkali-silica reaction, Cem. Concr.


Res. 37(2007)1291-1297.

[4] L.S. Dent-Glasser, N. Kataoka, The chemistry of alkali-aggregate reaction, in:


Proceedings of the 5th International Conference on AAR, National Building
Research Institute, Cape-Town, South-Africa, 1981, paper S252 (23 pp.).

[5] M.A. Bérubé, B. Fournier, Les produits de la reaction alkalis-silice dans le béton -
Étude de cas de la région de Québec, Canadian Mineralogist 24 (1986) 271-288.

[6] W.J. McCoy, A.G. Caldwell, New approach in inhibiting alkali-aggregate


expansion, ACI Materials Journal, 22 (9) (1951) 693-706.
234

[7] X. Feng, M.D.A. Thomas, T.W. Bremner, B.J. Balcom, K.J. Folliard, Studies on
lithium salts to mitigate ASR-induced expansion in new concrete: a critical review,
Cem. Concr. Res. 35 (9) (2005) 1789-1796.

[8] C. Tremblay, M.A. Bérubé, B. Fournier, M.D.A. Thomas, D.B. Stokes,


Performance of lithium-based products against ASR: application to Canadian
aggregates, reaction mechanisms, and testing, in: M. Tang, M. Deng (Eds.),
Proceedings ofthe 12th International Conference on AAR, International Academic
Publishers, Beijing, 2004, pp. 668-677.

[9] C. Tremblay, M.A. Bérubé, B. Fournier, M.D.A. Thomas, K.J. Folliard,


Effectiveness of lithium-based products in concrete made with Canadian aggregates
susceptible to ASR, ACI Material Journal 104 (2) (2007) 195-205.

[10] D.B. Stokes, H.H. Wang, S. Diamond, A lithium-based admixture for ASR control
that does not increase the pore solution pH, in: V.M. Malhotra (Ed.), Proceedings of
the 5 CANMET/ACI International Conference on Superplastisizers and Other
Chemical Admixtures in Concrete, American Concrete Institute, ACI SP 173, 1997,
pp. 855-868.

[11] S. Diamond, Unique response of LiN03 as an alkali silica reaction-preventive


admixture, Cem. Concr. Res. 29 (8) (1999) 1271-1275.

[12] M.A. Bérubé, C. Tremblay, B. Fournier, M.D. Thomas, D.B. Stokes, Influence of
lithium products proposed for counteracting ASR on the chemistry of cement
hydrates and pore solution, Cem. Concr. Res. 34 (2004) 1645-1660.

[13] M. Lawrence, H.F. Vivian, The reactions of various alkalies with silica, Aust. J.
Appl. Sci. 12(1961)96.

[14] P.W.J.G. Wijnen, T.P.M. Beelen, J.W. De Haan, C.P.J. Rummens, L.J.M. Van De
Ven, R.A. Van Santen, Silica gel dissolution in aqueous alkali metal hydroxides
studies by 29Si NMR, Journal of Non-Crystalline Solids 109 (1989)
85-94.
235

[15] CL. Collins, J.H. Ideker, G.S. Willis, K.E. Kurtis, Examination ofthe effects of
LiOH, LiCL, and LiNÛ3 on alkali-silica reaction, Cem. Concr. Res. 34 (2004)
1403-1415.

[16] X. Feng, M.D.A. Thomas, T.W. Bremner, B.J. Balcom, K.J. Folliard, Effect of
L1N03 on ASR-expansion and dissolution of silica, in: B. Fournier (Ed.), Marc-
André Bérubé Symposium on Alkali-Aggregate Reactivity in Concrete, Montréal,
Canada, 2006, pp. 139-152.

[17] L.D. Mitchell, J.J. Beaudoin, P.E. Grattan-Bellew, The effects of lithium hydroxyde
solution on alkali silica reaction gels created with opal, Cem. Concr. Res. 34 (4)
(2004)641-649.

[18] Q. Yin, Z. Wen, Effects of lithium hydroxide on alkali silica reaction gels, in:
M. Tang, M. Deng (Eds.), Proceedings of the 121 International Conference on
AAR, International Academic Publishers, Beijing, China, 2004, pp. 801-804.

[19] Y. Sakaguchi, M. Takakura, A. Kitagawa, T. Hori, F. Tomosawa, M. Abe, The


inhibiting effect of lithium compounds on ASR, in: K. Okada, S. Nishibayashi,
M. Kawamura (Eds.), Proceedings of the 8l International Conference on AAR,
Society of Materials Sciences, Kyoto, 1989, pp. 229-234.

[20] S. Diamond, S. Ong, The mechanisms of lithium effects on ASR, in: Proceedings of
the 9th International Conference on AAR, London, UK, 1992,
pp. 269-278.

[21] X. Mo, C. Yu, Z. Xu, Long-term effectiveness and mechanism of LiOH in


inhibiting alkali silica reaction, Cem. Concr. Res. 33 (1) (2003) 115-119.

[22] K.E. Kurtis, P.J.M. Monteiro, Chemical additives to control expansion of alkali-
silica reaction gel: proposed mechanisms of control, Journal of Materials Sciences
38 (2003) 2027-2036.
236

[23] M. Prezzi, P.J.M. Monteiro, G. Sposito, Alkali-silica reaction - Part 1 : Use of the
double-layer theory to explain the behavior of the reaction product gels, ACI
Materials Journal 94 (1) (1997) 10-17.

[24] M. Prezzi, P.J.M. Monteiro, G. Sposito, Alkali-Silica Reaction - Part 2: The effect
of chemical admixtures, ACI Materials Journal 95 (1) (1998) 3-10.

[25] Canadian Standards Association (CSA), Potential expansivity of aggregates,


procedure for length change due to alkali-aggregate reaction in concrete prisms,
CAN/CSA standard A23.2-14A (2004), CSA International, Mississauga, ON,
Canada.

[26] American Society for Testing and Materials (ASTM), Standard test method for
determination of length change of concrete due to alkali-silica reaction, ASTM
Standard CI293 (2006), ASTM International, West Conshohocken, PA, USA.

[27] American Association of State and Highway Transportation Officials (AASHTO),


Standard method of test for rapid identification of alkali-silica reaction products in
concrete, AASHTO standard T 299 (1993), Washington, DC, USA.

[28] X. Feng, M.D.A. Thomas, T.W. Bremner, B.J. Balcom, K.J. Folliard, B. Fournier,
Summary of research on the effect of LiN03 on alkali-silica reaction in new
concrete, in: Proceedings ofthe 13th International Conference on AAR, Trondheim,
Norway, 2008 (9 pp).
237

Table 7.1 - Mineral phases detected by XRD on the polished surface of various materials
before and after immersion for one hour in a IN LiOH solution at 350°C
(300 psi or 2070 kPa in the autoclave)
Aggregates Mineral phases observed on the polished surface
([Li]/lNa+K] (Listed from the most to the least abundant)
effective ratio1) Before autoclaving After autoclaving
Highly reactive aggregates or materials
Amorphous glass (see ¥, „. , „. _ _.
Obsidian F- _ ». ° Li-sihcate (see Figure 7.2)
Rhyolite Quartz, Na-feldspar, Quartz, Li-silicate, Na-feldspar,
(0.63) biotite (traces) (see Figure 7.3) biotite (traces) (see Figure 7.3)
Spratt limestone
Calcite, quartz Portlandite, calcite, quartz
(1.04)
Greywacke Quartz, Na-feldspar, calcite,
Quartz, Na-feldspar
(>1.11) chlorite, muscovite (traces)
Clayey limestone Calcite, quartz, Na-feldspar, Portlandite, calcite, quartz, chlorite,
(>1.11) chlorite (traces) Na-feldspar (traces), Li-silicate (traces)
Moderately reactive siggregates
Dolostone Dolomite, Li-silicate, calcite,
Dolomite, quartz, calcite
(0.61) quartz (traces)
Chloritic schist Quartz, calcite, chlorite, Quartz, portlandite, chlorite,
(>0.93) Na-feldspar, muscovite (traces) muscovite (traces), calcite (traces)
Granitic rock Quartz, Na-feldspar, Quartz, Li-silicate, Na-feldspar,
(0.56) K-feldspar K-feldspar
Non reactive aggregates
Pure limestone Calcite | Calcite, portlandite, Li-carbonate

The mineral phases in bold are those of greater importance in the present study.

Table 7.2 - Li consumption (in wt%) after 28 days in the test solution when aggregate
powders (150-300 pm) were immersed in various Li solutions at 60°C

Aggregates Li consumed (wt%) ± 1%


0.8N (Na,K)OH + 0.28N LiNQ3 0.8N (Na,K)OH + 0.56NLiNO3 IN LiOH
Spratt limestone 36.4 74.9 56.3
Rhyolite 26.3 48.9 53.8
Chloritic schist 31.7 47.7 54.3
Obsidian 30.5 29.2 82.5
238

Table 7.3 - Mass variation (in wt%) of various aggregates and mineral phases (varieties of
silica, volcanic glass, and feldspars) after 28 days in various Li solutions at
80°C
Materials IN NaOH IN NaOH + IN NaOH + IN LiOH
([Li]/[Na+Kl effective ratio1) (no Li) 0.25N LiN0 3 0.5N LiNQ 3
Natural aggregates (all reactive except two)
Pure limestone (non reactive) -0.1 not tested not tested 0.0
Spratt limestone (1.04) -4.4 -0.9 -0.6 -0.1
Rhyolite (0.63) -7.8 -1.3 -0.9 -2.3
Clayey limestone (>1.11 ) -2.2 -0.9 -0.4 -0.1
Greywacke (>1.11) -8.5 -0.9 -0.4 -1.0
Granitic rock (0.56) -7.8 -1.3 -0.8 -1.8
Chloritic schist (>0.93) -4.7 -0.1 +0.8 -0.1
Dolostone (0.61) -1.1 -0.8 -0.4 -0.8
Quartz sandstone -4.8 -0.7 +0.3 -1.2
Quartzite (non reactive) -0.4 not tested not tested -0.4
Varieties of silica
'y

Quartz (non reactive) -0.1 not tested -0.1 0.0


Opal (amorphous silica) -100.0 -100.0 -71.5 +9.44
Chalcedony3 -100.0 -6.3 -0.1 5 o.o5
Red chert3 -100.0 not tested +1.9 not tested
Green chert3 -100.0 not tested -1.9 not tested
Fused silica (synthetic) -13.7 not tested +4.44 -3.6
Volcanic glass and varieties of feldspars
Obsidian (volcanic glass) -7.8 not tested -7.04 -10.2
Anorthosite (non reactive)2 -0.7 not tested not tested -1.2
Microcline (non reactive) -0.4 not tested not tested -1.3
Orthose (non reactive)2 -1.8 not tested not tested -1.4
Based on concrete prism tests;
Fragment from a coarse crystal;
Cryptocrystalline quartz;
Li-silicate not observed on the particle tested;
Li-silicate observed on the particle tested.
239

Table 7.4 - Summary ofthe degree of agreement between the results of test series nos. 1 to 4d
and the mechanisms A to F proposed to explain the effectiveness of lithium against
ASR
Agreement between the test results and the following mechanism proposed :
chemical stability of silica chemical stability of silica unchanged but
Experiments decreased due to mechanism: expansion reduced due to mechanism:
(series #) B (Other C (Si-Li D (Non-exp. F (Increased
A(pH E (Non/less-
decrease) chemical protective Si-Li cryst. solubility of
exp. Si-Li gel)
reason) coating) product) silica)
Possible ([Li] decreases more rapidly with time in Impossible
1. Pore solution Impossible
presence of reactive aggregates -» possible ([Si] always
chemistry of (pHnot Not
formation of Si-Li products, which could be low and non
cement pastes affected by conclusive
crystalline or amorphous, non to less expansive, affected by
and concretes LiN0 3 )
and/or protective) LiNQ3)
Not Likely (just Unlikely (protective coating not observed, Si-Li
Not conclusive
2. Examination/ conclusive traces of Si-Li crystalline product not observed, amount of typical
(Si not
microanalysis products when ASR gel proportional to concrete expansion, and
(pH not measured in
of concretes L1NO3 is very absence or just traces of reaction products, limited solution)
measured)
effective) to typical ASR gel, when L.NO3 is very effective)
3. Polished aggr. Not Likely (Si-Li Not conclusive
+cement paste in conclusive reaction (Si not
Unlikely (Si-Li reaction products not observed)
NaOH±LiN0 3 at (pH not products not measured in
80°C measured) observed) solution)
Not conclusive
Not conclusive
4a. Polished (Si-Li product Possible (crystalline Li2Si03
(gel not observed
aggregates in Not absent on 3 of covering obsidian and all 3 Not conclusive
but tests in
LiOH at 350°C conclusive the 4 reactive aggregates for which L1NO3 is (Si not
LiOH, in
(autoclave) or (pHnot aggr., but for very effective in concrete, but measured in
absence of lime
80°C (obsidian measured) which L1NO3 is tests in LiOH and with most of solution)
and with most of
only) poorly effective them performed at 350°C)
them at 350°C)
in concrete)
Not conclusive
Not conclusive Not conclusive
4b. Polished Not (non to less Not conclusive
(silica gel, (protective gel,
Unlikely ( no expansive gel,
aggregates in conclusive undetectable by undetectable by (Si not
crystalline undetectable by measured in
NaOH+LiN0 3 at (pH not XRD, could be XRD, could be
product) XRD, could be
80°C measured) absent; SEM not present; SEM not solution)
present; SEM
done) done)
not done)
Possible (Li- Possible (cryst. or amorphous (no
4c. Ground aggr. Non Not conclusive
Unlikely product always XRD nor SEM) finely-dispersed
in LiOH or conclusive (Si not
(Li product formed, which Li-product always formed, pro-
(Na,K)OH + (pHnot measured in
always formed) could be portionally to Li consumption,
LiNO, at 60°C measured) solution)
protective) but expansivity unknown)
Likely for most Possible for some
Unlikely
4d. Aggregates reactive Unlikely materials (finely-
Not (crystalline Not conclusive
particles in materials (Si (significant disp. amorphous
conclusive product only on (Si not
NaOH+LiNOj or dissolution coating only on product ac to
(pH not undissolved measured in
LiOH at 80°C significantly opal, which is dissolution but
measured) chalcedony and solution)
(dissolution test) reduced or non-protective) expansivity
very thin)
suppressed) unknown)
Overall Likely Impossible
Impossible Unlikely Unlikely Unlikely
conclusion
Mechanism considered: likely, possible, not conclusive, unlikely, or impossible, based on the test results.
240

£
_t

245 um
IWTTTTTÏÏ ilTTnT!

Figure 7.1 -Micro-SIMS elemental scans on a polished section of a concrete made


with a highly reactive rhyolite and a [Li]/[Na+K] of 0.74, after 2 years at
38°C and >95% R.H. (0.01vol% expansion).
241

U2SI03 • Lithium SilicaU


J L

Figure 7.2-XRD analysis of a polished section of obsidian before (A) and after
(B) one hour of autoclaving in a IN LiOH solution at 300 psi or 2070
kPa (350°C). (A) The obsidian is mostly composed of amorphous
glass. (B) The obsidian surface is covered by a layer of crystalline
lithium silicate Li 2 Si0 3 (ICDD card file 29-0829).
242

r
ti^^^Kf^MnMnf^^,
20 30 to 50 «O 70 BO
Quartz, iyn ­ Si02
■ ■ ­ ■ ■

Albil», o r t e r M . W l S . 3 < ­ _
î­i­i L—__J—■ " . . . ■*"
BioBl» ­ KF.Mg2(«Si3O10XOH)2
'■■'■'■ '—'*■ i I um ­ ­.■...■ .. J . . . . . ■ _ — . . . . i , , m m
2­Hwk

■JAJL» K^AjAJiUjUyj .­Ur.1A_yiLr*. IV.. ­ i » ,


20 30 40 50 60
QE_nC tyn ' 10­

llSattSiUhiumSStcl."

MM>, t r i m * ­ NiAisooe

. . . . . _.__ ■ KF«Mg2<AISi3O10XOH)2
Bkilito _...
■ ' " ' ' ' ' ' . J ■ I 4 — ■•. ■ ■ ■ ­ ­ ■ . ........ . J ­ ­ . ..­. ­ ­ . .
2­Thah

Figure 7.3 ­ XRD analysis of a polished section of a highly reactive rhyolite


before (A) and after (B) one hour of autoclaving in a IN LiOH
solution at 300 psi or 2070 kPa (350°C). After autoclaving (B), a
thin crystalline lithium silicate Li 2 Si0 3 (ICDD card file 29­0828) is
detected on the rhyolite surface.
243

Figure 7.4 - SEM views at different scales of the Li-silicate layer (Li2Si03) (analysed
by XRD) observed on the surface of an obsidian particle immersed for 6
months in a IN LiOH solution at 80°C.
244

A
­rn—IIIII—rn—r­1—r­!—|—rn—m—|—i—pn—i—l i l
1.0­
Aggregates in control solution
0.9­

0.8 ­ Legend
­ ■ Spratt Limestone
0.7­ ­ • Rhyolite
0.6­
♦ ­ ­ ♦ ­ ­ ♦ Obsidian
0.5­
Potassium
0.4 ­

0.3 ­

0.2

0.1 ­
i i |—i—i—rn—i—i—|—IIIIII—[iiiiii ­l—!—l—TT—I—T­1—I—I—I—I—l­
0.0
7 14 21 28 7 14 21 28
Age (days) Age (days)
C D
1.0 T—I—I—I—I ­
1.0
Rhyolite in .35 [Li]/[Na+k] Rhyolite in .74 [Li]/[Na+K]
0.9
0.9 solution solution
0.8
0.8
Legend Legend
' 0.7 • — • — • Potassium • — • — • Potassium
■—■—■ Sodium ■—■—■ Sodium
• 0.6
X X X Lithium X X X Lithium
0.5

0.4

0.3 1 * = J ^
0.2
0.2
0.1
0.1
­i—i—i—i—i—i—i—i—i—i—r~\—i—i—r 0.0 ~i i i i i r~ 1
■ : '

0.0
14 21 28 14 21 28
Age (days) Age (days)

Figure 7.5 ­Alkali concentrations in the soak solution over time when aggregate powders
(150­300 pm) were immersed in various Li solutions at 60°C. (A) All
aggregates in 0.68N (Na,K)OH control solution. (B) All aggregates in IN
LiOH. (C) Rhyolite in 0.8N (Na,K)OH + 0.28N LiN0 3 ([Li]/[Na+K]) = 0.35).
(D) Rhyolite in 0.8N (Na,K)OH + 0.59N LiN0 3 ([Li]/[Na+K]) = 0.74).
245

>
w
w
ra
E
o

­100
14
A g e (days)

1N NaOH + Li Return INNaOH noLi


80

60

40

20

­20

­40

­60

­80

­100 4 ! ' I ■I | : I I I i i j i i ! i ! i 4i i ; n¥l [TTTTTTTTT

7 14 21 28 35 42 49 56
A g e (days)

Figure 7.6 ­Mass variation of opal and chalcedony particles immersed at 80°C in solutions
of IN NaOH (control) and IN NaOH + LiN0 3 in various amounts ([Li]/[Na]
from 0 to 1.0). (A) Chalcedony. (B) Opal. (C) Opal and chalcedony particles
immersed for 28 days in a solution of IN NaOH + 0.5N LiN03 (chalcedony)
or 1 .ON LiNÛ3 (opal), then immersed in the IN NaOH control solution.
246

150-
C Chalcedony surface
1 IN NaOH + [Lij/[Na] = 0.5

100

o
O

50

**^vJUvJ
0 20
Viw^.vn_W
30 40 50 6(
1 > 46-1045: Quartz, syn ■ Si02

I i n i i
2> 29-0828: Li2Si03 - LM lum Silicate
I I

I I Il ■ II u II

2-Theta

Figure 7.7 - XRD analysis of the crystalline lithium silicate Li 2 Si0 3 (ICDD
card file 29-0828) detected on the surface of a chalcedony particle
after immersion for 28 days in a IN NaOH + 0.5N LiN03 solution
at 80°C.
247

8 Conclusion générale

Les travaux de recherche complétés dans le cadre de ce projet de doctorat ont conduit aux
principales conclusions suivantes:

1. Les résultats des essais sur prismes de béton (CSA A23.2-28A) ont démontré que le
L1NO3 à un ratio [Li]/[Na+K] de 0.74, recommandé par le fabricant, permet de
réduire l'expansion sous la limite spécifiée de 0.040% après deux ans à 38°C avec
six des douze granulats réactifs testés, trois granulats exigeant un dosage compris
entre 0.74 et 1.11, alors qu'un dosage de 1.11 n'a pas été suffisant avec les trois
autres granulats testés;

2. Le dosage requis pour contrer la réaction alcalis-silice varie d'un granulat réactif à
l'autre et il n'a pas été possible d'obtenir une corrélation entre ce dosage et la
composition des granulats et/ou leur degré de réactivité;

3. Le verre de lithium (Li-glass) en poudre ne s'est pas avéré efficace, même broyé
plus finement que le produit disponible commercialement. Il semble que la faible
dissolution de celui-ci ne permet pas de mettre en solution une quantité de lithium
suffisante pour être efficace, du moins pour la période couverte par l'essai sur
prismes de béton;

4. Tel que prévu, les cendres volantes, le laitier et le ciment ternaire testés sont
efficaces pour contrer la RAS en présence des granulats réactifs testés dans la
mesure où les ajouts cimentaires en cause ont une composition appropriée et sont
utilisés à un dosage adéquat;

5. La combinaison d'ajouts cimentaires et de LiN03 n'a pas montré d'efficacité


supplémentaire systématique;

6. À l'essai du prisme de béton CSA A23.2-14A (adapté selon la pratique normalisée


CSA A23.2-28A) ou ASTM C 1293, l'expansion à 60°C après six mois se compare
bien à l'expansion à 38°C après deux ans pour les mélanges témoins mais est plutôt
faible pour les mélanges incorporant le LiN03;
248

7. Une méthode sécuritaire et accélérée (d'une durée de 28 jours) est proposée pour
évaluer le ratio [Li]/[Na+K] efficace à utiliser dans le béton incorporant des
granulats réactifs. Cette méthode utilise deux essais AMBT, dont un mélange ou le
nitrate de lithium est incorporé à la fois dans les barres de mortier (100% du dosage
standard) et dans la solution de trempage (20% du dosage standard), et un mélange
témoin sans lithium. Cette méthode permet de prédire un ratio [Li]/[Na+K] efficace
pour les granulats qui répondent relativement bien au lithium. La méthode permet
aussi d'identifier les granulats qui répondent moins bien et pour lesquels l'essai sur
prismes de béton est recommandé pour évaluer le dosage minimum en LiN03 à
utiliser pour contrôler la RAS;

8. L'analyse des solutions interstitielles de pâtes de ciment hydraté et de bétons


indiquent que le ratio [Li]/[Na+K] en solution est environ la moitié de celui
initialement incorporé dans les mélanges. Ainsi, une bonne partie du lithium est
incorporée aux hydrates de ciment lors de l'hydratation;

9. Des analyses au microscope électronique à balayage (MEB), par spectrométrie de


masse à ionisation secondaire (SIMS), par diffraction X et sous lumière ultra-
violette, ont été réalisées sur divers échantillons de bétons, de mortiers, de granulats
et de particules de silice réactive. Ces analyses ont permis d'établir qu'il est possible
de former un silicate de lithium cristallin seulement sous certaines conditions bien
précises et avec certains types de silice. Aucun silicate de lithium n'a pu toutefois
être observé avec des granulats naturels dans des bétons âgés jusqu'à deux ans et
analysés au MEB et par SIMS;

10. Il a été démontré que la présence de lithium réduit significativement ou empêche la


dissolution de plusieurs phases de silice réactive immergées dans des solutions
alcalines. Ces résultats, ainsi que les nombreux autres essais et observations réalisés
tout au long de ce projet de recherche, indiquent que la réduction ou la suppression
de la dissolution de la silice par un autre mécanisme que la réduction du pH ou la
formation d'un produit protecteur sur les phases réactives, qu'il soit cristallin ou
249

amorphe, est le mécanisme le plus probable pour expliquer l'efficacité du LiN03


contre la RAS.

De façon générale, le nitrate de lithium a le potentiel de réduire et parfois même


pratiquement annuler l'expansion dans le béton causée par la réaction alcalis-silice.
Toutefois, les dosages requis varient énormément d'un granulat réactif à un autre, sans
toutefois que l'on puisse relier l'efficacité du lithium à certains types de granulats ou à
certains faciès. De plus, certains granulats répondent moins bien et il semble que même des
dosages très élevés ne permettent pas de réduire l'expansion sous la limite imposée lors de
l'essai sur prismes de béton.

Le nitrate de lithium peut donc efficacement contrer la RAS, mais le dosage requis doit être
évalué pour chaque granulat réactif. Il est de notre avis que l'essai sur prismes de béton
d'une durée de 2 ans demeure présentement l'essai le plus fiable pour évaluer
adéquatement la quantité de lithium à utiliser dans le béton. Toutefois, il a été démontré que
l'essai sur barres de mortier modifié peut-être utilisé dans certains cas pour évaluer la
quantité de lithium nécessaire et ce, de façon conservatrice.

9 Recommandations / Perspectives

Malgré l'avancement de la recherche sur les adjuvants à base de lithium utilisés dans le
béton, plusieurs aspects pourraient être étudiés davantage et ce, afin d'utiliser encore plus
judicieusement ces produits. Voici les recommandations que cette étude propose:

1. L'essai accéléré sur barres de mortier modifié qui a été développé dans le cadre de
cette étude afin d'évaluer le ratio [Li]/[Na+K] efficace à utiliser dans le béton
devrait être validé avec un plus grand nombre de granulats, malgré le fait qu'il a été
testé sur plusieurs granulats (14) de natures et compositions diversifiées dans ce
projet. De plus, une étude de la variabilité de cette méthode devra aussi être
effectuée;
250

2. Le phénomène par lequel l'ajout de LiNÛ3, en petite quantité dans la solution de


trempage, augmente l'expansion pour certains granulats réactifs devrait être étudié
et mieux compris.

3. Il a été démontré qu'une grande proportion du lithium est incorporée lors de


l'hydratation du ciment et n'est donc plus disponible pour contrer la RAS, d'où
l'idée d'utiliser un verre silicate contenant du lithium qui relâcherait du lithium une
fois les hydrates formés. Toutefois, le verre de lithium testé ne s'est pas avéré
efficace du fait qu'il demeure peu soluble et qu'il ne relâche pas suffisamment de
lithium. Le principe est toutefois valable et vaut la peine d'être approfondi. Peut-
être qu'un autre produit à base de lithium plus soluble pourrait s'avérer plus
efficace;

4. Les mécanismes par lesquels le lithium réduit et empêche la dissolution de certaines


phases de silice réactive ne sont pas encore bien connus. Il est primordial que les
études ultérieures expliquent adéquatement les mécanismes en cours et ce, afin
d'utiliser ces adjuvants en toute connaissance de cause; il demeure en effet difficile
de recommander un moyen de prévention de la RAS en ne connaissant pas toutes
les raisons à la base de son efficacité;

5. Il est de notre avis qu'afin de parvenir à expliquer plus complètement l'efficacité du


lithium, l'étude de modèles simples avec différentes variétés de silice réactive pure
est une avenue à suivre, tout en gardant à l'esprit que certaines variétés de silice
sont très différentes de celles que l'on retrouve dans les granulats naturels. Jusqu'à
maintenant, il a été impossible, dans cette étude et dans les études antérieures,
d'expliquer clairement les mécanismes à partir de spécimens de béton, de mortier et
même de granulats naturels. Il demeure que la réaction alcalis-silice est très
complexe, que les divers composants (solution interstitielle et pâte de ciment
hydraté) entrant dans la réaction sont aussi fort complexes et que la composition des
granulats réactifs à contrer est très variable;

6. Le choix du type de silice réactive (ou silicate) à étudier doit être considéré très
sérieusement dans les études ultérieures. Plusieurs variétés de silice réactive
251

utilisées couramment dans les recherches antérieures (opale, cristobalite, pyrex,


calcédoine, etc.) ont des compositions, des structures atomiques et des
comportements souvent bien différents des variétés de silice réative retrouvées plus
communément dans les granulats naturels et ce, malgré une expansion similaire lors
d'essais accélérés. Ces types de silice ne devraient pas être retenus en priorité dans
le cas d'études mécanistiques.
252

ANNEXE A

Analyses chimiques des granulats et descriptions


Pétrographiques détaillées
253

Analyses chimiques des granulats

NSl NS2 NB1 NB2 NB3 QCl QC2 ONI ON2 ALI AL2 BC NF

Oxydes % en masse
Si02 71.4 74.3 30.4 63.4 72.7 14.4 51.6 11.5 70.9 81.5 93.6 74.9 0.53

CaO 1.21 1.34 27.2 4.23 2.63 24.9 10.3 46.7 2.17 5.20 1.15 3.26 54.6

A1203 13.5 12.5 8.20 15.8 11.8 3.20 13.4 0.57 13.0 3.46 1.94 11.1 0.20

Fe203 4.56 3.38 3.40 6.76 4.81 1.30 8.17 0.28 4.54 1.79 1.48 4.21 0.11

MgO 1.89 0.64 3.28 3.08 1.92 16.9 4.65 1.71 2.54 1.44 0.15 1.84 0.40

S0 3 0.10 0.15 0.44 0.16 0.10 0.80 0.11 0.24 0.10 0.10 0.10 0.13 0.10

Na 2 0 2.65 4.35 0.88 2.08 2.28 0.14 2.18 0.07 3.36 0.58 0.31 2.90 0.07

K20 2.39 3.82 1.66 2.83 1.83 1.68 1.06 0.05 1.97 1.05 0.81 1.07 0.03

LOI 2.18 1.47 24.5 1.16 2.91 37.0 9.38 38.4 1.49 5.32 1.35 1.29 43.2

Na2Oeq 4.22 6.86 1.97 3.94 3.49 1.25 2.87 0.07 4.65 1.27 0.84 3.60 0.05
254

Descriptions pétrographiques détaillées

NSl

2 faciès : grès lithique (77%), grès schisteux (23%).

• GRÈS LITHIQUE: chlorite verte (clinochlore) (15%), muscovite (5-10%),


quartz cristallin (5%), quartz microcristallin (10%), feldspath (dont albite) (30-
40%) (DRX : présence de calcite) ;
• GRÈS SCHISTEUX : séricite (85%), quartz microcristallin (2-3%), chlorite
fibreuse verte (10%).

NS2

4 faciès : granite rose (82%), granite altéré (9%), grés (8%), granite gris altéré (1%).

• GRANITE ROSE : quartz microcristallin (10%), mica (1%), microcline (2-3%),


plagioclase (60%);
• GRANITE GRIS ALTÉRÉ : plagioclase dont microcline (80%), quartz
microcristallin (5-10%), chlorite verte (5-10%);
• GRÉS : feldspath très fin (80%), quartz micro- à cryptocristallin (10%), quartz à
extinction roulante (2%), chlorite (2%), altération rouille (2%);
• GRANITE ALTÉRÉ : quartz microcristallin (5-10%), plagioclase (65%), mica
(5-10%), chlorite (10%), matrice altérée (milonite).

NB1

2 faciès : 3 sur lame mince : 2 calcilutites (98%), 1 calcaire plus ou moins pur (2%)

• CALCILUTITE (plus foncée) : boue calcareuse (85%) contenant minéraux


opaques (10%), quartz microcristallin (5%), (DRX : présence de chlorite);
• CALCILUTITE (grise) : boue calcareuse (85%), quartz microcristallin (5-8%),
K-feldspath (5%);
• CALCAIRE PUR : gros cristaux de calcite (95%), matériel détritique (5%) avec
possibilité de quartz cryptocristallin.

NB2

2 faciès : rhyolite (99%), granite (1%).

• RHYOLITE : matrice de verre dévitrifiée (85%), quartz cristallin (5%), K-


feldspath (dont albite) (8%), calcite (3%);
• GRANITE : feldspath (dont microcline) (35%), quartz cristallin (50%), quartz
microcristallin (5%), biotite verte (2-3%).
255
NB3

6 faciès : 7 sur lame mince : grés carbonate rouge (23%), gabbro ou basalte (23%),
schiste à séricite (12%), pélite argileuse (18%), roche magmatique felsique altérée (9%),
quartzite métamorphique (8%), granite (7%).

• GRÈS CARBONATE (rouge) : carbonate ferrugineux de type ankérite ou


dolomite (60%), quartz microcristallin (10%), plagioclase (10%), séricite,
carbonate, calcite, quartz cristallin;
• GABBRO/BASALTE : feldspath altéré (20%), épidote (25%), trace de quartz
cristallin (3%), minéraux opaques (5%), chlorite (%);
• SCHISTE À SÉRICITE : séricite (60%), chlorite (35%), quartz microcristallin
(1%), minéraux opaques (3%);
• PÉLITE ARGILEUSE : micrite (60%), minéraux opaques (2%), trace de quartz
microcristallin, argilite, fragments de feldspath, (bande gréseuse, bande
argileuse plus ou moins métamorphique);
• ROCHE MAGMATIQUE FELSIQUE ALTÉRÉE : séricite (beaucoup), un peu
de biotite, nodules (probablement surtout d'anciens cristaux de feldspaths
altéré), trace de quartz microcristallin;
• QUARTZITE MÉTAMORPHIQUE : quartz microcristallin (5%), quartz à
extinction roulante (majorité de gros grains de quartz fracturés ou plissés)
(95%);
• GRANITE : k-feldspath (40%), microcline (30%), biotite (10%), muscovite
(8%), quartz cristallin (5%).

OC1

2 faciès : dolomie (97%), dolomie argileuse (3%).

• DOLOMIE : cristaux rhomboédriques de dolomite (98%), quartz microcristallin


(1-2%), trace de minéraux opaques (1%), trace de calcite (moins de 1%);
• DOLOMIE ARGILEUSE : dolomite (60%), plage micritique (30%), quartz
microcristallin (2-3%), minéraux opaques (3-5%).

OC2

3 faciès : schiste à chlorite (63%), schiste à séricite (24%), quartz (13%).

• SCHISTE À CHLORITE : chlorite verte (30%), épidote (10%), calcite (15%),


quartz microcristallin (0,01 à 0,1 mm) (40%), K-feldspath (dont albite) (5%);
• SCHISTE À SÉRICITE : quartz microcristallin (0,01 à 0,1 mm) (70%), calcite
(25%), séricite (5%), trace de minéraux opaques;
• QUARTZ: quartz mierocristallin (0,01 à 0,1 mm) et cryptocristallin (60%),
quartz cristallin (30%), calcite (5%), chlorite (2-3%).
256

ONI (SPRATT)

3 faciès : calcarenite fine (65%), calcarenite grossière (34%), calcilutite (1%).

• CALCARENITE FINE : calcite cristalline (10%), calcite très fine avec quartz
très fin dans la matrice (90%) (DRX : présence de quartz);
• CALCARENITE GROSSIÈRE : calcite (100%), riche en fossiles
• CALCILUTITE : calcite fine (100%).

ON2

5 faciès : grès lithique (60%), grès quartzeux (28%), granite (10%), quartz rouge (2%).
• GRÈS LITHIQUE: quartz microcristallin (15%) à plus grossier (5%),
plagioclase et microcline (10%), calcite (5%), grains détritiques contenant
quartz, feldspath et chlorite (15%), chlorite (5%), minéraux opaques (1%),
carbonate (15-20%);
• GRÈS QUARTZEUX avec ciment épitaxique : quartz cristallin (75%) (quelques
grains de quartz à extinction roulante), ciment argileux et quartzeux avec
séricite, un peu de chlorite et trace de calcédoine, trace de minéraux opaques;
• GRÈS QUARTZEUX : quartz cristallin (65%), ciment argileux (possiblement
un peu quartzeux) (30%), minéraux opaques (5%), biotite (2%);
• GRANITE (ou granodiorite): feldspath (68%), minéraux opaques (2%),
trémolite (dans veine fileuse aciculaire) (30%), quartz micro- à cryptocristallin +
quartz à extinction roulante;
• QUARTZ ROUGE : quartz rouge très fissuré (95%) de taille variable de 2 mm à
cryptocristallin, carbonate (5%), minéraux opaques (1-2%), (faciès très altéré).

ALI

7 faciès : grès quartzeux (35%), grès lithique (23%), grès rose (14%), calcilutite (10%),
grès vert (8%), basalte vésiculaire altéré (8%), granite (2%).

GRÉS QUARTZEUX (type Postdam) : quartz microcristallin (5%), quartz avec


ciment épitaxique (95%) et souvent à extinction roulante (60%);
GRÉS LITHIQUE : plagioclase (30%), mica (5%), quartz microcristallin à plus
grossier (5-10%), phosphates;
GRÈS ROSE : quartz cristallin (90%), K-feldspath (peu), un peu de séricite, de
carbonate et de chlorite;
CALCILUTITE : boue calcareuse/matériel détritique (98%), quartz
cryptocristallin (2%);
GRÉS VERT : feldspath en bâtonnets (50%), séricite (30%), minéraux opaques
(10%), altération (10%);
BASALTE VÉSICULAIRE ALTÉRÉ : chlorite cristallisée (à l'origine ferro-
ma.gnésiens) (70%), chlorite fine remplissant des vésicules (15%), aussi en
remplissage des vésicules: quartz fin (1%) + calcédoine (moins de 1%),
carbonate (5-10%), feldspath (en baguettes) (10%), présence probable d'ilménite
et d'hématite en baguette;
• GRANITE : feldspath dont microcline (85%), mica (10%), quartz cristallin
(5%).
257

AL2

7 faciès: grès quartzeux (67%), grès cherteux (14%), pélite siliceuse (5%), particule
argileuse (5%), chert (6%), quartz cristallin (2%).

• GRÈS QUARTZEUX: feldspath (70%), quartz cristallin (7%), quartz


microcristallin (2-3%), carbonate (20%), (roche poreuse très métamorphisée);
• GRÈS CHERTEUX : particules très fines et siliceuses riches en quartz (80%),
calcédoine fibreuse (2 %);
• CHERT : quartz sédimentaire (95%), altération (5%) (roche très fracturée et
métamorphisée);
• PÉLITE SILICEUSE (oolithique) : quartz micro et cryptocristallin (45%),
quartz cristallin (5%), argile (30%), mica brun (1-2%), minéraux opaques (5%),
calcédoine (1%);
• PARTICULES ARGILEUSES : micrite dont phosphates très fins (95%), quartz
cristallin (3%), feldspath (2%);
• QUARTZ CRISTALLIN : quartz cristallin (90%), quartz à extinction roulante
(5%), altération (5%).

BC

7 faciès : grès lithique (35%), grés quartzeux (17%), gneiss (14%), chert quartzitique
(14%), basalte (12%), quartzite (8%).

GRÉS LITHIQUE : matrice composée de feldspath microcristallin (80%),


plagioclase (10%), minéraux opaques (5-10%);
GRÈS LITHIQUE VERT : feldspath (80%), quartz cristallin (5%), chlorite (5-
10%) (roche très métamorphisée);
GRÉS QUARTZITIQUE : feldspath (90%), quartz cristallin (10%);
GNEISS: feldspath (50%), hornblende (20%), chlorite verte (20%), minéraux
opaques (3%), quartz à extinction roulante (3%);
CHERT QUARTZITIQUE : quartz cristallin (80%), quartz microcristallin
(10%), altération en chlorite (5%), minéraux opaques (2%);
BASALTE : feldspath magnésien très fin (90%), quartz cristallin (2%), chlorite
(5%), minéraux opaques (5%);
QUARTZITE : quartz à joints suturés et à extinction roulante (potentiellement
réactif) (100%).

NF

2 faciès : calcilutite (98%), calcarenite (2%)

• CALCILUTITE : petits et moyens cristaux de calcite (10%), pélitoïdes riches en


calcite et trace de minéraux argileux (85%);
• CALCARENITE : calcite (95%), micro-inclusions de minéraux opaques, quartz
microcristallin (2%) (roche poreuse).
258

ANNEXE B

Mélanges de pâte de ciment


259

Melange 1 (Ciment 1,2S%Na20eq)


Date: 17 mars 2003 Volume voulue en m3:0,0015
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment haute teneur alcalis 1300 kg/m3 0,4127 1300,0 1,9500
Rapport E/C 0,42 0,5460 546,0 0,8190
Teneur en air 4% 0,0400 0,0
Ajout NaOH 0,0 0,0000
Adjuvant LiN03 0 litres/kg de Na20éq 0,0000
Total 0,9987 1846,0 2,7690

Melange 2 (Ciment 1,25%Nà20eq + 400%LtN©3 )


Date: 17 mars 2003 Volume voulue en m3:0,0015
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment haute teneur alcalis 1300 kg/m3 0,4127 1300,0 1,9500
Rapport E/C 0,42 0,4698 469,8 0,7046
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 0,0 0,0000
Adjuvant LiN03 4,6 litres/kg de Na20éq 0,0748 89,7 0,1346
Total 0,9972 1846,0 2,7892

Mélartge3 (Cfroent 1,25%Na20eq + 1Ô0%yGla«s)


Date: 17 mars 2003 Volume voulue en m3:0,0025
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment haute teneur alcalis 1300 kg/m 3 0,4127 1300,0 3,2500
Rapport E/C 0,42 0,5460 546,0 1,3650
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 0,0 0,0000
LiGlass 1,9 kg/kg de Na20éq 0,0124 30,9 0,0772
Total 1,0110 1846,0 4,6922

Mélange 4 (Ciment 1 r 25%Na20eq + 100%crushed LiGlass)


Date: 17 mars 2003 Volume voulue en m3:0,0020
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment haute teneur alcalis 1300 kg/m3 0,4127 1300,0 2,6000
Rapport E/C 0,42 0,5460 546,0 1,0920
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 0,0 0,0000
LiGlass broyé 1,9 kg/kg de Na20éq 0,0124 30,9 0,0618
Total 1,0110 1846,0 3,7538

Melange s (Ciment i,œmnZQm * %S%<&<$w«mœ}


Date: 20 mars 2003 Volume voulue en m3:0,0015
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment haute teneur alcalis 930 kg/m3 0,2952 930,0 1,3950
Masse de cendres volantes 310 kg/m3 0,1469 310,0 0,4650
Rapport E/C 0,42 0,5208 520,8 0,7812
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 0,0 0,0000
Adjuvant LiN03 0 litres/kg de Na20éq 0,0000
Total 1,0030 1760,8 2,6412
260

Melange 6 (Ciment 1,25%Na20eq + 25% CV Thunder Bay )


Date: 20 mars 2003 Volume voulue en m3: 3,0015
Volume 1m3 Masse 1m 3 Masse labo
kg kg
Masse de ciment haute teneur alcalis 960 kg/m3 0,3048 960,0 1,4400
Masse de cendres volantes 320 kg/m3 0,1203 320,0 0,4800
Rapport E/C 0,42 0,5376 537,6 0,8064
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 0,0 0,0000
Adjuvant LiN03 0 litres/kg de Na20éq 0,0000
Total 1,0027 1817,6 2,7264

Melange 7 (Ciment 1,25%Na20eq + 25% CV Trenton )


Date: 20 mars 2003 Volume voulue en m3: 0,0015
Volume 1m3 Masse 1m 3 Masse labo
kg kg
Masse de ciment haute teneur alcalis 935 kg/m3 0,2968 935,0 1,4025
Masse de cendres volantes 312 kg/m3 0,1393 312,0 0,4680
Rapport E/C 0,42 0,5237 523,7 0,7856
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 0,0 0,0000
Adjuvant LiN03 0 litres/kg de Na20éq 0,0000
Total 0,9999 1770,7 2,6561

Melange 8 (Ciment 1,25%Na20eq + 25% Laitier )


Date: 20 mars 2003 Volume voulue en m3: 3,0015
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment haute teneur alcalis 975 kg/m3 0,3095 975,0 1,4625
Masse de laitier de haut fourneau 325 kg/m3 0,1117 325,0 0,4875
Rapport E/C 0,42 0,5460 546,0 0,8190
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 0,0 0,0000
Adjuvant LiN03 0 litres/kg de Na20éq 0,0000
Total 1,0072 1846,0 2,7690

Melange 9 (Ciment 1,25%Na20eq + 25% filler calcaire )


Date: 20 mars 2003 Volume voulue en m3: D.0014
Volume 1m3 Masse 1m 3 Masse labo
kg kg
Masse de ciment haute teneur alcalis 955 kg/m3 0,3032 955,0 1,4325
Masse de filler calcaire 318 kg/m3 0,1218 318,0 0,4770
Rapport E/C 0,42 0,5-347 534,7 0,8020
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 0,0 0,0000
Adjuvant LiN03 0 litres/kg de Na20éq 0,0000
Total 0,9997 1807,7 2,7115

Melange 10 (Ciment 0,51 %Na20eq + NaÔH )


Date: 20 mars 2003 Volume voulue en m3: D.0025
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment basse teneur alcalis 1300 kg/m3 0,4127 1300,0 3,2500
0 kg/m3 0,0000 0,0 0,0000
Rapport E/C 0,42 0,5460 546,0 1,3650
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 12,43 kg/m3 0,0062 12,432 0,0311
Adjuvant LiN03 0 litres/kg de Na20éq 0,0000
Total 1,0049 1858,4 4,6461
261

Melange 11 (Ciment 0,51 %Na20eq + NaOH + LiN03 )


Date: 20 mars 2003 Volume voulue en m3: 0,0025
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment basse teneur alcalis 1300 kg/m3 0,4127 1300,0 3,2500
0 kg/m3 0,0000 0,0 0,0000
Rapport E/C 0,42 0,4825 482,5 1,2062
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 12,43 kg/m3 0,0062 12,432 0,0311
Adjuvant LiN03 4,6 litres/kg de Na20éq 0,0748 89,7 0,2243
Total 1,0161 1884,6 4,7115

Melange 12 (Ciment 0,51 %Nà20eq + NaOH + LiGlass)


Date: 28 mars 2003 Volume voulue en m3: 3,0020
Volume 1m3 Masse 1m 3 Masse labo
kg kg
Masse de ciment basse teneur alcalis 1300 kg/m3 0,4127 1300,0 2,6000
0 kg/m3 0,0000 0,0 0,0000
Rapport E/C 0,42 0,5460 546,0 1,0920
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 12,43 kg/m3 0,0062 12,432 0,0249
Adjuvant LiGlass 1,9 kg/kg de Na20éq 0,0154 30,9 0,0618
Total 1,0204 1889,3 3,7786

Melange 13 (Ciment 0,51%Na20eq + NaOH * LiGlass broyé)


Date: 28 mars 2003 Volume voulue en m3: 3,0015
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment basse teneur alcalis 1300 kg/m3 0,4127 1300,0 1,9500
0 kg/m3 0,0000 0,0 0,0000
Rapport E/C 0,42 0,5460 546,0 0,8190
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 12,43 kg/m3 0,0062 12,432 0,0186
Adjuvant LiGlass broyé 1,9 kg/kg de Na20éq 0,0154 30,9 0,0463
Total 1,0204 1889,3 2,8340

Melange 14 (Ciment 0,51%Na2Oeq )


Date: 28 mars 2003 Volume voulue en m3: 3,0025
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment basse teneur alcalis 1300 kg/m3 0,4127 1300,0 3,2500
0 kg/m3 0,0000 0,0 0,0000
Rapport E/C 0,42 0,5460 546,0 1,3650
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 0,00 kg/m3 0,0000 0,000 0,0000
0 kg/kg de Na20éq 0,0000 0,0 0,0000
Total 0,9987 1846,0 4,6150

Melange 15 (Ciment 0,51%Na2Oeq + LiN03 )


Date: 28 mars 2003 Volume voulue en m3: 0,0025
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment basse teneur alcalis 1300 kg/m3 0,4127 1300,0 3,2500
0 kg/m3 0,0000 0,0 0,0000
Rapport E/C 0,42 0,5201 520,1 1,3003
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 0,00 kg/m3 0,0000 0,000 0,0000
Adjuvant LiN03 4,6 litres/kg de Na20éq 0,0305 36,5 0,0914
Total 1,0033 1856,7 4,6416
262

Melange 16 (Ciment 0,51 %Na20eq + LiGlass)


Date: 28 mars 2003 Volume voulue en m3: 3,0025
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment basse teneur alcalis 1300 kg/m3 0,4127 1300,0 3,2500
0 kg/m3 0,0000 0,0 0,0000
Rapport E/C 0,42 0,5460 546,0 1,3650
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 0,00 kg/m3 0,0000 0,000 0,0000
Adjuvant LiGlass 1,9 kg/kg de Na20éq 0,0063 12,6 0,0314
Total 1,0050 1858,6 4,6464

Melange 17 (Ciment 0,51%Na2Oeq + LiGlass broyé-)


Date: 28 mars 2003 Volume voulue en m3: 3,0015
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment basse teneur alcalis 1300 kg/m3 0,4127 1300,0 1,9500
0 kg/m3 0,0000 0,0 0,0000
Rapport E/C 0,42 0,5460 546,0 0,8190
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 0,00 kg/m3 0,0000 0,000 0,0000
Adjuvant LiGlass broyé 1,9 kg/kg de Na20éq 0,0063 12,6 0,0189
Total 1,0050 1858,6 2,7879

Melange 18 (Ciment 0,51%Na2Oeq + KOH )


Date: 14 avril 2003 Volume voulue en m 3 : 3,0015
Volume 1m3 Masse 1m 3 Masse labo
kg kg
Masse de ciment basse teneur alcalis 1300 kg/m3 0,4127 1300,0 1,9500
0 kg/m3 0,0000 0,0 0,0000
Rapport E/C 0,42 0,5460 546,0 0,8190
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout KOH 17,43 kg/m3 0,0087 17,430 0,0261
Adjuvant L i N 0 3 0 litres/kg de Na20éq 0,0000
Total 1,0074 1863,4 2,7951

Melange 19 (Ciment 0,51%Na20eq + KOH + L i N 0 3 )


Date: 14 avril 2003 Volume voulue en m3: 3,0015
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment basse teneur alcalis 1300 kg/m3 0,4127 1300,0 1,9500
0 kg/m3 0,0000 0,0 0,0000
Rapport E/C 0,42 0,4825 482,5 0,7237
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 17,43 kg/m3 0,0087 17,430 0,0261
Adjuvant L i N 0 3 4,6 litres/kg de Na20éq 0,0748 89,7 0,1346
Total 1,0186 1889,6 2,8344

,,,M
Melange 20 (Ciment 0,51 %Na20eq + kô*. + Nâô-.)
Date: 14 avril 2003 Volume voulue en m 3 : 3,0015
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment basse teneur alcalis 1300 kg/m3 0,4127 1300,0 1,9500
0 kg/m3 0,0000 0,0 0,0000
Rapport E/C 0,42 0,5460 546,0 0,8190
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout KOH 11,79 kg/m3 0,0059 11,79 0,0177
Ajout NaOH 4,03 kg/m3 0,0020 4,03 0,0060
Adjuvant L i N 0 3 0 litres/kg de Na20éq 0,0000
Total 1,0066 1861,8 2,7927
263

Melange 21 (Ciment 0,51%Na2Oeq + KOH ♦ NaOH 4 L i N 0 3 )


Date: 14 avril 2003 Volume voulue en m3: 3,0015
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment basse teneur alcalis 1300 kg/m3 0,4127 1300,0 1,9500
0 kg/m3 0,0000 0,0 0,0000
Rapport E/C 0,42 0,4825 482,5 0,7237
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout KOH 11,79 kg/m3 0,0059 11,79 0,0177
Ajout NaOH 4,03 kg/m3 0,0020 4,03 0,0060
Adjuvant L i N 0 3 4,6 litres/kg de Na20éq 0,0748 89,7 0,1346
Total 1,0178 1888,0 2,8320|

Melange 22 (Ciment 1,25%Na20eq + 25%CV­Sundance + LÏN03 )


Date: 14 avril 2003 Volume voulue en m3: 3,0015
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment haute teneur alcalis 930 kg/m3 0,2952 930,0 1,3950
Masse de cendres volantes 310 kg/m3 0,1469 310,0 0,4650
Rapport E/C 0,42 0,4754 475,4 0,7131
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 0,0 0,0000
Adjuvant L i N 0 3 4,6 litres/kg de Na20éq 0,0534 64,1 0,0962
Total 1,0110 1715,4 2,6693

Melange 23 (Ciment 1,25%Na20eq ••■25% laitier + LiN03)


Date: 14 avril 2003 Volume voulue en m 3 : 3,0015
Volume 1m3 Masse 1m3 Masse labo
kg kg
Masse de ciment haute teneur alcalis 968 kg/m3 0,3073 968,0 1,4520
Masse de laitier 322 kg/m3 0,1107 322,0 0,4830
Rapport E/C 0,42 0,4945 494,5 0,7418
Teneur en air 4 % 0,0400 0,0 0,0000
Ajout NaOH 0,0 0,0000
Adjuvant L i N 0 3 4,6 litres/kg de Na20éq 0,0556 66,8 0,1001
Total 1,0081 1784,5 2,7769
264

ANNEXE C

Essais sur prismes de béton


05
­Q)
en 265
w oo o
o O
o
OJ
a ra m CM CO D CN
oo
"*
CN

OJ o
J 5 CM OJ_ CD co"
w OJ" co" CM m
re

O
­3
o

24,155
re CO o
3,881
CSI
O
CD O) o
CN OJ
.__ CNI CD
o t
co" 1^
■«t
O
(fl en (A
0J m
o" re 3
S cr
_.
re
E E
o o JQ O. ■f a>
o CD o ■ * O EN CM
CD
176,
1097

U) CO O)
6 (fl o o CN
E CM m
w CN CO o CM
■«a­ (D "*­ CD
re co
3 CM O o
O
3 S
> r>
0>
E § Il
E ro CD 00 CM O o E © CM
3 OJ ■n
co h ­ •O r­ o 0J '0J in
O fc
« ­ *!■
CM o o (0 S­ co
> E
­
■ <

3 o O O o" o CN
O
>
CO
OO T ­ 00 CM ■OJ
co m
CM CM _f o^ CN Tl"
CD
co CD
CM CM r^ O
CN
O o OO
o o
T" O o■ ■«t
o" CM
O

CM CO CO C J 00 CO m (fl
C
OJ CM re
g S JQ S S co o
T— Q. jfl
o
m co CN i< Jii <*
o CM
i2 ^ o

CD co
co CO CO co N n n °. S CM
E I E
_D)
__
­r­ x­ O °­
O
o o

O
CM
CN
■*
s* __:
O OJ m r­ 4^­
•* .­T
o oo o m __­ co
CM ° CM © ™ CD N .
co m
O T­
OJ o" o"

X co
Û. 0 . m in
ca ­<t CM

0
ë S3 Z<
H Zre <
"5>
re
c
o
CD o
m ■o
>«­
o"

i
c
«M
O
O
C
S OJ D
= fc
.2
0 'ô.
*­ ­ t CO ^
_2
A
(0
" ' _ _ _ « ­ . — E ­OJ
w C3 •£.
+
;__,
3 S
a «S w S (O
re ^­
3
C
£2
a
o E
E r«e »
O) 3 W ro •a

CM
.2p » .£ jj g re 8 rë X O c S W (A
•*— re
<_ ­
© ' 0
g c » .? c fc g w z
3 2
ef S "■so tn° ­ — fl) row
3 0
(n
®
0
E S, COOM CM

o mE ■s ■■§«
2 a * JS J S = •s »
!s w
re
M
•c oo i­
O" o "
•­■5 22 = "S a. a « SE as 2 re ro w 0 m a a CM

■m. ÔJ
« il!
w rt 3 2 8 £ > 8 o g f o "3 O) +
< re a
o re
o"Si
OJ
3
(fl
re it £ c « E 'Ex
o ai o
a (A

(0 2
S 2 to f*
re re
0
(0
(A
a
re
3
'rë
(A
(fl
£ m
m
o
3
0 3
H
2S re ai o"
ii c re
« LU
Ul

Qj
E
c
re
O
O
O
o
co
0)

"> ,­< o 00
266
° £Is­ °io
S O) JQ o in in o
g» ^ CM CO _ H CM
o m 00
(fl co co co­ o" co­
(A CM in
re
S
O
JO
o CO o
24,245
o re
3,881
ro «C M* CO
m oo
CM
CM 0 o in
(A CO" o" co­ (A
O
(A oi" in fl)
o" re 3
5 E
» re
E o §8 OJ CM
E
fl o CO o T ^~ E o
0) E3
176,
1102

o co CO 0)
£ (A ro o" ■*"
CO
(A CM CO o" cg" 12 o"
3 re CM
3 S
O
> «
0 E
E T­ CO co oo CN o E û CD
3
■fl) r> CO f«­ o o CD ­Q) OJ
O ■>­ • * o o o
E E *~ CM
> 3
O
o" o" o" o" o" 4o o"
re ­­s
> fc UJ
< ■«J­
? . " . ioo Sco CO
CN
oo m
m
r^­
m
Ujin • t o o CM CD o ­­
o CM o" o"
<«■
CD
c
CM
■t
o
00 m CN "55
UJ CM CO CO O J CO O c
^ ro" o" m" g g OJ
CD
CM
o o re
CQ m co CM § g ■t 00 CD
f^ co co O O CN
Q. (fl
CO CO CO CO
* ­ ■*­ O o" o" CD" 5?
E o o
sm I ro
___
CN

O CM o m o OJ m
CM • *
"* o
go"
CN
o m ,_­
CM o "
CO
CM"
i. ;i o **
o2
O) o" o"

ro m co

c
£
co
O O)
"3T o o
•+■« o m o" o"
re «
CM 3 _Q
C
0
CM 2 ■o
(A
O)
(A

fl)
E
(A
£, «­ o
o co
re
00
CM
o o
a­ '_a.
_ o o"
v L­
>© 3
ro, (A
(A

°­ g(A e OJ
o
3­ Ul •<* o
0

T—
re
3
re
UJ
60°C Canmet
38°C Laval
0
OJ
(A o o
267
0 o
O O <­ o m CO
°­ 3)jn fe■r ■*§ CM
o co"
CD"
Of ^ CM OJ *■ in o"
«A Oj" CO" CM
In
re
o
CO o
CD
in CD
CD o co"
o re
«2 °> °° 3 m

O O I_ OJ o
m ° "* ­ co
«*"
O Tf
O CO
O CM
CM
N-
CO
is &
m
co
CM

go»» CO
CN o3­ co

o * CN I*­ ° 1^ o
>
0)
OJ
E CO CD 00 CM o o E CÙ
OJ o co
» n s o CM o o in
­CÙ
CO
3 E E *­. *­. <*. o" .<_2
1 _3
O
o o o
I
£> O
LU
CM

>
LU r-. CD *—
t—
CM m ■«1- 00 i^ O CN OO
tr O) CM CN o o
■<!-
i ­ OJ
0. CM co
UJ in CO CO ■4

I-- 1 — o" o O ■<*- o o


Q. o i CM oo
o o
CM CO co OJ CO o '55
UJ 00 in c
-1 OJ o in in oo OJ CM re
o oo CD o Cl (A
m in CO CM o o t T—
K .b, O) o
< r^ CJ CJ o
o" o CM
UJ co CD 00
co CO CO CO CO o" o CO O T­
CM
o" o"
E E
fo I
"*o J__

O CM
o oo m OJ TJ­
CM t o m " CO
" • O"
£ °" CM"
co m
CD 00
CM o *­
o o" o"

O CN
_ ­ CN
C CD O *­
o m o" o"
'OJ

_­>
0)
s11 * g s TJ
(fl (fl * ­ CO
0)
f 81 S 2 3 1 E
(A
. t , o 1^
00 o o
•c
a_.
CM O O

3
(fl
re et »
8 e t
rê CM
UJ o
o

*-*
a)
ro E
mi c
CD
f)
o
U
X)
(
0
)
co o
CD
0

I O 00 o o
o
OJ
268

8.
oj m
Se CM_
ro m
oq
CM
co­
in
o s &
o
co
(A OJ" co" CN o" m
re
S

o
o n CO oo
CM JS o r­ £] CD
(D
m m
CM „, o> ■* °o c­ T— co
O
o" $ .* «H ® £
(A OJ co JQ
CD o
o
­fr"
m (A
­OJ
3
S o"

E
re
x: o
o
o
t
OJ E E O)
CD O "fr OJ
0
(A
ro o co" T— "■*­ O CO CM
0­ E m
m co o" co
(A .JC CM •t CM" m o" CD
CN oo CO
re
5
o
> «
fl) E
E T" CO CD co CM o E ©
3 OJ n co 1­^. o r­­ o eu ­QJ
«■" t o o
O E
3
t
o
*~
o o
CM
o"
«j
.<_> Q

> ë_u
S <
co m
CD co CD ­<fr
1—

11
■<*■
00 O
m CM CM
CD
o
"fr
CM
m"
ara o " O *­
o"
o" o
o „­ CM l­~
o
T— "S
UJ CM CO CO OJ co co m o c
■ O J ­ Q ­ i n _t_ °° OJ CM re
(D o o'
CD m co " o o oo O . (fl O ­fr
< ►­ co
~ co *­t o o CM X . t i j CD O
i/î T" T­ O
o" CD"
UJ t o O i­
co co co co CM o" o"
E
I E
O)
JC
O CM o m o OJm
CM t CN m o m _r co
00 CM"
* O" o" CO CD
CN ° e CJ OJ
O O
OJ O ' O "

X (O LU

Ë
CL ­a H ra
II
JE ^
ra i CO
_­ oo
O
N
O

f?
CD
m o " o"
0) Jr o'»©' re + re
a . ­ o _= 3
CM £ oS. 0 0 '­'•­ o (O
JS <~ C
a c >,
S
OJ 3 fc JS fc 2 <n ­
co
■= = >©
ra s w C X 2 «5 3 S 2
0
E 9 .2 o fl) ■a
"5 .® â ­ 3 0 O O (O ra (0 ro (A
"O o o Z ­ fc * ­ JS 2 co ©
OJ S" C ~j re re re
0
3 « °­ 0 0(0 (A
E £, tOO CM
CO
■o

2' I 11 "S 3
s 2 5o 12 |ra
_i «­■ u.
re
(A
(A 3 ©
■D (A
©
re
(A oo o " o"
O

3 fe t 0 2 a re
Ï3
111 3
>© .=
o "3 ra + <
0 E » i a
CM

re
3
<§§° A
(0
S a) S
re
3 re
a
* t £
re
0 t X
o <
ra c
"Ex
1 ©
o
(A
(A
O (2 a
o (fl„ S 4S », a.. III
Q
re \­
8 fu a
re Î
(A
S I­­
CM
a) 0 3 0 ï UJ o
(fl « IS
re fl (A ­C
fl
o
re 2 re «—■
S E
•t—■

ro a)
> E
rni i_

r i O
ro
0 O
00 0
CO
CD o
©
­©
269
2 ° Is­ S o T­ | ^ m
in co m
CM
CM
■3­ ra m OJ 2 o co i«~
o> j j CM «* ^ co co­
CA roco £ o ' o" rn
(A
re

o
4­3
o a o OJ 2
co
co T­ 1^ m
co CD
CM m co f­­_
CN „ O) " * CM JS o co co­
O co (A "fr
o" $* «H t 2­ o" o" rn ©
3
m
<A OJ CO J *
CD E
re
£ O O
co OJ CM E
O
co o "fr ­" © F
O Tfr CN 1
"fr" o co co
co E CM in
S j? Tof co'

O co
r­­
o" CM" "fr c­, «*. CD
CO
gre CNl N . CM

©
E
ë u
co co oo CM o E "Qj oo
3 ï « n s o o © ­CD —'|OJ
E E *­_ ­r. **. CM o o
(fl «­
3 O O O o" o"
O
> <

CM OJ
oo co co m oo O o
O m CD
oo o
"fr
m o oc o o "i­
■ * OJ OJ o o 00 CM O T­
O
CT
■QJ
c o OJ OJ ­r o o" "fr" o
oo o" o"
CM
OJ
O N­
CM CD io
ro CO CD CM OJ m CM O c
CM re
"fr oo" io g 5g CN d
© m 3­ <fl, o ■*
■o CN C
CM M g
CM ­ ­ Ô§ O x t i j CD O
D) o" _
co CO CO co =* o o" co" iu ^ O T­

Sùl
E
1o ^m I 5E <"
©
S _.__
CM o" o"

O CM o OJ m "fr co
CM "fr go o m _­
cw
CM CO CD
CM CO OJ
o o
O) o" o"

X V) LU Ul <A CO I*.
" ­ oo
^33
II
« <
ra
i
■JC

re
C
O

CD o
m o" o"
o

Q­ XJ _= re +
© 3 6Ï 2 ©
fc __3 _■ .3 re
JA rëX co ris
V)
IL 3 Zr 3 S 3
C
2
©
■o
E re © ­a, .2 re 0 re
O
C S c ^ 2 tn ro
Q . <A 3 <A ­35 ro V) (A
._ o> " E T J S c fc re Z — ra s »
O (O 2 co (0 © "fr CM
0) © 3 JS E 3 —: re ra"> 3 »
s
© © — I 4­> '•ë °­ v (0 E O CO

■S s . © . ? ' f •© 3
"S C «< JS — © « <o (A oo O o
■a. .­s
*© ts • ­ E _­ <­>
° E
if.* 0 S
3 .2,
O
SE JS IS 3 g* re w 0 re
a ro >© c
re © E ra E
■O­
©
re, CM o" o"

P
i= Ê 3

O a,
CA re
•3 3
3 S
> go
t _5<
lil
o 2 ra
<
©
c
© f*
fc X
Sw

C
o
3
3
(A
(0
'rë
a (A < O »(fl 5
<

re "3 ra + u re ,_
3 0 + Q­'
re (A £ r».
(A
i re K
ISS
rem4­1
©
C
©
3 £5
fl
(0 C
re •—■
fl
3
(A
UJ
CM
o

si
**
m
<A
©
S

w 2 ©
E
c
ro
O
O
o
o
co
©
•©
(A
o
270
© o CO i n m
a ro i n ? c5 S oo
© JC CM "fr "fr" o o
CD o"
(A
(A
oj­
co" CN o"
re
2
o
O ore o O) im
n CO
ro "fr CM
CM CD
CM
O ©
JC CM
"fr o
o"
(A o j ­ co" "fr" CD
o o in
CO (A o"
(A CM ©
3
re ro
S re
« E
E o o m OJ "fr ©
E o
o OJ o o oo
© U) o
CD "fr
CO OJ E CN m
E (A JC CN co" OJ "fr" o
o CM CM o "fr o
3
(A "fr o co "fr T—
o o"
CM *™
re
3 5
O
>
© E CT
© II
E T­ CO cn oo CN o E © m
3 © r) CO ■>­ o o © ■Cl) m
O
b b *" ­— "fr CM o o w I— o
> 3 o o o o" o" <£ o
O (»
fc LU
> <
co
g co ­­ «. 2 Ï OJ
o "fr "­
CO CM
ïï ID m ô m OJ O o
CT ­ ■^ N* S M o
y 8 o
o o" o "
o"
­a. "fr" 00
O CM O) o
CN co "(Â
LU co co CM OJ m CM
ro .­J "fr" oo" m" g <g CM o
Z CD "­ CM CM g S csi m o' re N­ "­
© rf N­ CM CM ° . S o a m CM
O) o"
T
l2 O O
CO CO CO CO w " o o CD"
o" o "
E E.
IN I
©
S S) .__
O) £ ;
_ * : ■ _ =

O CM o oo OJ m
o m "fr CO
CM " f r OJ "fr o ­_­
"* o" CM ° N­ CN
CM OJ m
O T­
OJ o o
o" o "

ro co m
JC O T­
C CD
«T O o o
m o" o"
*­_
re ©
3 ■o
C (A
re ©
r^­
E
(A
£ CM
o o
o o
re •c oo
o" o"
a. a CM

« 3

t (A
o (A
a
o.
re
S e■ i j m
o
(A
3 UJ <*1 o
o.
fl
©
ro E
>
ro c
_i co
O O
O
o
oo o
co o
co
©
•«
(A 271
© o o ° o OJ
QL ro in in îr o
© JC CM CO ­ o
(A CD CO
OJ­0O­.M 1^
o"
(A
re
S
o
JD
CO
o JS ocoS CD oo
CM
© ro CO
CM CO ' r _ ­ I—
CN
(A JC CD "fr o
o (A oi" co'S m o"
o" re
«
E o o g OJ

E © O "fr ° l
CD
"fr"
£ o
co
m o
(A XL ° " CD m CJ
(A
r­ CD o"
3 re co
3 2
O
>
©
«
b
F
CO CD CO CM in
3 © f)
CJ s o r­~ CM
O F b T­ " ­ "fr CM CO o
> 3 o o o o" CM
O
>

r­­ "­
T­ CM
O O
CT o" o "
­©
o
CM
c
o
ro (fl
Z c
re
© a. uj OJ co
■D x JE,
O) UJ ©■ OO
CM
O
420 kg/m3

CO CO CO =* CM
o" o"
E5 E E <"
0,42 E/C

"ro _ p
l ^ ___
o OJ i q

..c ■_=
"fr i n
2690
0,48

o in co r­
i~­
CM o"
" CM" m"
"fr CD
O " ­
o o
OJ o" o "

X tf) UJ ___.
CL a fr 3) ro <fl CO f~
11 ra JC,
O "­
Ê ^33 «* S3 co o o
G
< f?
(A in

z c
©

©
A » a­S £ F«
tf)
*­»
re •©
JQ
o" o"
CM
O © 3 ËJS
©
a2 s —> • ­ o a. re
CL + JS T"
c ^s.
o |
l_
ro © a> © .o co P tf) 3 S 2 ©
" a r e — 3 IS ■a
„• 5. "3 a re
_j CM

're 'G (A © re w
IS o ra 3 co
c S ro tf)
tf)
2 co ro iii

I 3­ w in ©
© "O g) 0) © c 0
flf «= S
O tf) ©
© ro co V
CM OJ
E o ■o ? 3
• fc E S 2
■o ­05 ­o
E3
= SE E 5 © 3 _­ (0
(A 3 0
il
OO
o •© fc *­ JS re ra iii o o
EÎ I
CM
CM a™•»■*« ° i; t_ 0 2 o JS JS 3 a . re o" o"
c _. > S o
o(A c © £ Q.I < < © 2 ra < re
©
vg> 0 re ©

re
re 5 3 3 JO re
3 <0 5 O L "3 ra + © X ■©
ra
g
c
>© 3
(A
3 O r> *~* ■fc
Q
o
a © < a <0
(A
(A
CO
^ re
Q£ IIS a
o
a < Ex
©
Ul
III in
+
o
a.
a.
(A

ffl m
re
« £~ 3 F (A
____ re '— J
re (A
LU
o
o
re o"
S 2 co LU fl
o 2
ii ©
E
c
ro
O
O
o
o
CD
©
­©
272
2 o m P o OJ
3. o> m CM ^ ^ O co oo
g) ^ CM OJ _0 "~ N.
1 « co" S CD O Of co
m
ro
2
o
o .o CJ ­ ra » oo
CM JS ra «fr oo
P­ § co r—
CM
XL CM_ 0O_
CN _o ■"­ S
O (fl
(A OJ co" "fr co O (N N
­ (fl
d re CN
©
odS 3
E
re
£E Eo
CM
i o o OJ T— 12
P« CD "fr
CD O "fr ©
•=^co
o CJ CO m
"fr" o
g 5 o" co" CJ
O CM
"fr
<M
So CD
CO
re ^ '•"
O
>
©
ë n
E CO CD CO CM o E © OJ
3 © " co rT~­­ o o © .© in
b b T "fr CM O o w >­ co
O 3 o o o o" o" .!2 O CM
> o S Lu
> <

" ­ CM CM
■<­ c o oo i­­ co O)
co" LO O * CN CM
m
O)
i«­ "fr "fr
CM CM O
~ O o O o
JT » «=■ ^ O o" "fr" o o o
CT CM o c
•© o o
O "fr '55
CM HI co co CN OJ m CM 1^ o C
ro J "j­" oo" m" g S r.
CN
in re
Z « ­ CN CM
CM
i^ a. (/J CM
©
h ­ CM CM
"­ «­ O P­ O x .s_­| T ­ OJ
«fr
■o co" iu io
co CJ CJ co o"
E E ra
CM O o
E
lrae ­~î.
JC LU c O)
XL ra ra
O CM
__.
O 00 o­ o ^ JC JC
O
OJ m "­ "fr OJ
CM «fr o O

m ,_­ co __­
"• O CM o" CM"
CN
(0 O "­
oo co
O «­
OJ o" d"

X tf) UJ ___ ^
(L a ro (fl O) CM
II ro ra ..__, CM CM
co O « ­
ra ra p~ < (A

re
m o" o"
t" "© >©" Q. re + re + 3
•©
CM Q. T3 C ■ 3 Z, C

CM
ris
ro 3 (fl O
tf) C
■a 2 co
tf) o,tf)
^

tf)
tf) 2
ro
©
■o
(fl
tf) ra co
■©■=
fc *­ JS
«
« °­ _= "> ©

3
© ■o
©
(A
(0
(A
©
©

E £ oo
m
o>
o
re » © re (A
oo
CM JS JS 3 o. re
­© re ■c o" o"
0 2 ra 0 E ra a a CM
E
"3 ra +
Sx
fc X
o _—"4k
Ia.
©

o
(0

I*
o LU (A
1 2 JS © +
Q, 0
(A
(0
a
re 8 "fr
(A m
re H re '—■■
3 Ul o
3 22
re o
Ul