Pr21. Borcelo Et Al. Astm c1012 Mod. 2014

8/10/2019 Pr21. Borcelo Et Al. Astm c1012 Mod. 2014

1/14

A modi ed ASTM C1012 procedure for qualifying blended cementscontaining limestone and SCMs for use in sulfate-rich environments

Laurent Barce lo a ,b , , Ellis Gartner b , Rmi Barbarulo b , Ashlee Hossack c, Reza Ahani d , Michael Thomas c,Doug Hooton d , Eric Brouard b , Anik Delagrave a ,e , Bruce Blair ea Lafarge Canada Inc., 334 Avro, Pointe Claire, QC H9R 5W5, Canadab Lafarge Centre de Recherche, 95 rue du Montmurier, 38291 St Quentin Fallavier, Francec University of New Brunswick, Fredericton, New Brunswick E3B 5A3, Canadad University of Toronto, 35 St. George Street, Toronto, Ontario M5S 1A4, Canadae Lafarge North-America, 12018 Sunrise Valley Drive, Suite 500, Reston, VA 20191, USA

a b s t r a c ta r t i c l e i n f o

Article history:Received 19 December 2013Accepted 6 May 2014Available online 6 June 2014

Keywords:Sulfate attack (C)ThaumasiteDurability (C)Expansion (C)Thermodynamic calculations (B)

Blended Portland cements containing up to 15% limestone have recently been introduced into Canada and theUSA.These cements were initially not allowed for use in sulfate environments but this restriction has been liftedin the Canadian cement speci cation, provided that the limestone cement includes suf cient SCM and that itpassesa modi ed version of the CSAA3004-C8 (equivalentto ASTM C1012)testprocedure runat a low temper-ature (5 C). This new procedure is proposed as a means of predicting the risk of the thaumasite form of sulfateattack in concretes containing limestone cements. The goal of the present study was to better understand howthis approach works both in practice and in theory. Results from three different laboratories utilizing the CSAA3004-C8test procedure arecompared andanalyzed, while also taking into account theresults of thermodynamicmodeling and of thaumasite formation experiments conducted in dilute suspensions.

2014 Elsevier Ltd. All rights reserved.

1. Introduction

European standards for limestone-blended cements (up to 20%limestone for CEM II/A-L and up to 35% for CEM II/B-L) have existedfor several decades, and such cements are widely used in Europe.They were introduced more recently into North America, rst inCanada in 2008 [1] and then in the USA in 2012 [2]. In Canada theyare called Portland limestone cements (PLCs) and there are onlyslight differences between the CSA A3000 PLC and the EuropeanCEM II-A/L speci cations [3]. There are similar quality criteria forthe limestone permitted (based on minimum total calcium carbon-

ate and maximumtotal organic carbon contents as well as maximummethylene blue values), but the maximum permitted limestonecontent in North America is only 15%. The Canadian and US speci ca-tions, as does EN 197, also providefor more complex blendedhydrauliccements, combining limestone with supplementary cementitiousmaterials (SCMs).

In the Canadian speci cation for concrete [4], three sulfate exposurecategories are de ned: S1 (very severe), S2 (severe) and S3 (moder-ate). 1 Concrete exposed to S1 or S2 environments should be madewith highly sulfate-resistant (HS) cements, while moderately sulfate-resistant (MS) cements are permitted for concretes subject to S3 envi-ronments. They can be either pure Portland cements (HS and MS) orblended hydraulic cements (HSb and MSb). HS and MS cements havespeci c chemical requirements (maximum C 3A contents of 5% and 8%,respectively) and must not exceed a speci ed 14-day expansion limitwhen tested in standard mortars according to CSA A3004-C6 (equiv-alent to ASTM C452). HSb and MSb blended cements do not have dif-

ferent chemical requirements than other blended cements, but theymust meet stringent expansion limits (maximumexpansion 0.10% at6 months for MSb and 0.05% at 6 months or 0.10% at 12 months forHSb2) when tested in mortar according to CSA A3004-C8 (equivalentto ASTM C1012). The option of using blended cements in sulfate-rich

Cement and Concrete Research 63 (2014) 75 88

Corresponding author.E-mail address: [email protected] (L. Barcelo).

1 It should be noted that theminimum solublesulfate concentrationin groundwater forthe S1 class of exposure in Canada (10 g/l) is much higher than the maximum of the XA3category in Europe (6 g/l). This re ects the very high levelsof sulfates foundin thePrairieprovinces of Western Canada.

2 The 0.10% limit at 12 months is an option that can be used to qualify HSb cements if the 6-month expansion exceeds 0.05%

http://dx.doi.org/10.1016/j.cemconres.2014.05.007

0008-8846/ 2014 Elsevier Ltd. All rights reserved.

Contents lists available at ScienceDirect

Cement and Concrete Research

j o u rn al h o mep ag e : h t t p : / / ee s . e l sev i er . co m/ CE MC O N /d ef au l t . a sp
http://dx.doi.org/10.1016/j.cemconres.2014.05.007http://dx.doi.org/10.1016/j.cemconres.2014.05.007http://dx.doi.org/10.1016/j.cemconres.2014.05.007mailto:[email protected]://dx.doi.org/10.1016/j.cemconres.2014.05.007http://www.sciencedirect.com/science/journal/00088846http://www.sciencedirect.com/science/journal/00088846http://dx.doi.org/10.1016/j.cemconres.2014.05.007mailto:[email protected]://dx.doi.org/10.1016/j.cemconres.2014.05.007http://crossmark.crossref.org/dialog/?doi=10.1016/j.cemconres.2014.05.007&domain=pdf


2/14

environments, provided that the blend is tested by CSA A3004-C8 orASTM C1012 and meets acceptance criteria is included in the CanadianCSA A23.1 and US ACI 318 speci cations. Such blended cements havebeen used industrially for decades without any indication of problems(according to the scienti c literature).

To the knowledge of the authors, the CSA A3004-C8 (or ASTMC1012) test method for external sulfate attack does not have an equiv-alent in Europe, although country-speci c test methods exist. As thistest method has a central role in this article, its principle is brie y sum-marized here (see [5] for a detailed description). In the traditional ver-sion of the test, standard mortar is prepared by mixing 2.75 parts of standard sand with 1 part of cement. The water/cementing materialratio (W/CM)used forblended cements is 0.485 by mass. 3 Aftermixing,mortar bars of dimensions 25 mm 25 mm 285 mmare cast togetherwith companion mortar cubes for strength measurement. The samplesare cured in a water tank (the molds being watertight) at 35 3 Cfor the rst day and are then demolded and stored in lime-saturatedwater at 23 2 C until the strength of the mortar, measured on thecompanion cubes, reaches 20 1 MPa. At that time, the initial lengthof the mortar bars is recorded and the bars are immersed and storedin a highly concentrated sodium sulfate solution (50 g/l of Na 2SO4) at23 2 C. The length change of the bars is measured periodicallyand the sulfate solution is replaced at each measurement.

When PLC was rst introduced to the Canadian speci cations, in2008 [1], due to insuf cient data, it was not permitted for use in appli-cations requiring traditional sulfate-resistant cements. But in 2010 [5],the CSA A3001 cement speci cation was updated and two new cementcategories were created: HSLb and MSLb (for highly sulfate-resistantand moderately sulfate-resistant Portland-limestone-based blendedhydraulic cements). These cements must comply with all the require-ments of HSb and MSb cements, including the 6-month expansionlimit when tested in accordance with CSA A3004-C8, plus there arealso two additionalrequirements.Firstly, thereare prescribed minimum

SCM contents. For blends containing a single SCM, they are 25% for typeF y ash, 40% for slag and 15% for metakaolin. There are also two differ-ent possible SCM combinations, i.e. N5% silica fume plus N25% slag, orelse N5% silica fume plus N 20% type F y ash. Secondly, the standardmortars made with these cements must meet an 18-month expansionlimit of not more than 0.10% when tested according to a modi ed ver-sion of CSAA3004-C8 conducted at 5 C.Furthermore, if mortars expandby more than 0.03%duringthe last 6 months(i.e.from12 to 18 months)the test must be continued to 24 months and the 0.10% limit must bemet at this later age. Concurrently, the CSA A3004-C8 test method wasalso updated. The traditional procedure at 23 C (outlined above) wasrenamed Procedure A and a Procedure B was introduced, involvingtesting at 5 C. Procedures A and B are similar, with the only differencebeing that when the 20-MPa strength is reached on the companioncubes, the bars undergoing procedure B are stored in the same sodiumsulfate solution but at 5 2 C instead of 23 2 C. Procedure Bwas created to determineresistance to thepotential for the thaumasiteform of sulfate attack (TSA) [5] , based on the knowledge that TSAwould be accelerated by the use of lower temperatures.

To summarize the rationale of the Canadian approach, there is agradation of requirements regarding CSA A3004-C8 4:

HS or MS Portland cements do not have to be tested in accordance

with this procedure. They are considered sulfate-resistant if theirC3A contents are below speci ed limits and the cement does notexceed the 14 d expansion limit when tested in accordance withCSA A3004-C6;

HSb or MSb blended hydraulic cements have to pass procedure A(23 C) at 6 months (or 12 months) to demonstrate that the SCMquantity is adequate to reduce the traditional form of external sulfateattack to an acceptable level;

HSLb or MSLb blended hydraulic cements with a PLC base cementhaveto meeta speci ed minimum SCMadditionleveland to passpro-cedure A (23 C) at 6 months (or 12 months) as well as procedure B

3 In case the blended cement is not made at the cement plant but is reproduced byblending a PC or a PLC with SCM, the W/CM of 0.485 applies to a control mix done withthe corresponding PC or PLC. Then when the mix with SCM is produced, W/CM has to

be adjusted to the ow of the control mix 5%.

4 Note that, in CSA A3004-C2 (referenced by CSA A3004-C8), the requirements forblended cements (HSb, MSb, HSLb, MSLb) apply for both pre-blended cements and for

combinations of cementing materials

blended

at the concrete plant (during mixing).

Table 1Chemical analysis of cements and supplementary cementitious materials (SCMs) used by the different laboratories.

Cements chemical analysis (%)

Laboratory Cement SiO 2 A12O3 Fe2O3 CaO MgO Na2O K2O SO3 LOI % limest.

UNB GU (OPC) 19.9 4.9 3.1 63.2 1.9 0.23 0.90 3.5 NA 4GUL #1 (PLC) 18.1 3.8 2.2 61.5 1.9 0.21 0.91 3.5 NA 15GUL#2(PLC) 16.8 4.2 2.6 59.6 1.8 0.23 0.90 3.4 NA 22

UOT GU #1 (OPC) 20.6 5.5 2.2 63.4 2.4 0.23 1.22 4.2 0.6 0GU #2 (OPC) 19.7 5.4 2.1 62.7 2.4 0.24 1.18 4.7 1.4 2.4GUL #1 (PLC) 18.9 5.1 2.0 61.9 2.3 0.21 1.15 4.3 4.5 11GUL #2 (PLC) 18.5 5.0 2.0 61.3 2.4 0.22 1.12 4.4 5.2 13GUL #3 (PLC) 17.4 4.6 1.9 59.8 2.3 0.20 1.07 4.1 8.9 22

Lafarge GU (OPC) 20.2 4.3 2.7 61.5 4.5 0.22 0.60 2.8 2.9 3.7 a

GUL #1 (PLC) 18.8 4.0 2.4 61.2 4.3 0.19 0.57 2.7 5.7 10.5 a

GUL #2 (PLC) 19.1 4.5 2.6 60.2 2.5 0.12 0.74 3.8 6.0 14.7GUL #3 (PLC) 18.2 4.3 2.5 61.2 2.8 0.27 0.89 3.4 5.6 NA

SCMs chemical analysis (%)

Laboratory SCMs SiO2 Al2O3 Fe2O3 CaO MgO Na2O K2O SO3 LOI

UNB Type F y ash 57.8 18.0 9.4 3.0 2.0 1.32 2.03 1.1 NAGGBFS 34.1 9.4 0.3 40.1 11.6 0.37 0.43 2.7 NA

UOT GGBFS 38.1 7.2 0.7 40.0 10.6 0.33 0.46 2.7 0.3Lafarge Type F y ash #1 53.7 23.3 3.7 12.7 1.3 2.49 0.72 0.3 0.6

Type F y ash #2 43.4 23.1 20.7 5.1 1.2 0.64 1.67 0.6 1.8GGBFS 37.8 9.2 0.7 37.4 11.7 0.29 0.5 3.0 1.6b

a Values refer to calcium carbonate content instead of limestone.b Negative values: mass increase corresponding to oxidation of slag during LOI.

76 L. Barcelo et al. / Cement and Concrete Research 63 (2014) 75 88


3/14

(5 C) at 18 months (or 24 months) to demonstrate that the SCMquantity is also adequate to reduce the thaumasite form of sulfateattack to an acceptable level;

2. Testing and materials

2.1. CSA A3004-C8 procedure A testing (23 C) and procedure B testing (5 C)

For this analysis, several mortar compositionswere tested in univer-sity laboratories (University of New Brunswick, labeled UNB and Uni-versity of Toronto, labeled UOT) as well as in two Lafarge laboratories.Table 1 lists thechemical composition of thebinders used by the labora-tories. When blendsof cements andSCMs have been done, there wasnomodi cation of the SO 3 content. All mixes were tested following CSAA3004-C8 procedures A and B, however, small deviationsfrom the test-ing protocol should be noted: in the experiments from LG the binders were tested at different

water-to-binder ratio (W/CM): the W/CM ratio of the test mix withSCM was adapted to match the ow of the control mix without SCMwhen this control mix is tested at W/CM = 0.485. This is the

procedure speci ed when binder materials are blended at the con-crete plant;

in the UNB and UOT laboratories, the mixes were tested at a constantwater-to-binder ratio (W/CM = 0.485). This is the procedure speci-

ed when the blended cement is made at the cement plant.

Typical differences in water-to-binder ratio (for LG) and ow (forUOT) are provided in Fig. 1. It is apparent that following the procedurespeci ed when mixes are made at the concrete plant (constant ow LG case) leads to signi cant variations in the water-to-binder ratio.

2.2. Thermodynamic modeling

Thermodynamic calculations were run using the geochemical soft-

ware CHESS [6] to calculate theevolution in the equilibrium phase as-semblage of a fully hydrated cement paste as it reacts with an alkalisulfate solution. The thermodynamic database was composed of thestandard CHESS database for minerals and aqueous ionic species, towhich data for the relevant cement-related minerals was added. Theformation equations and related solubility products (expressed as log-K values) are given in Table 2 , expressed in CHESS basis species. Phasedensities and data sources are also included.

Fig. 1. Example of variation of W/CM ratio for different SCM combinations used in LG to obtain constant ow (A) and of variation of ow used in UOT at a xed W/CM of 0.485 (B).

Table 2Thermodynamic data used for the different minerals.

Mineral Formation equation log-K (temperature in C within brackets) Density

(kg/m3)

Data source

Portlandite 1 Ca [2+] , 2 H[+] , 2 H2O, 24.87(0) 22.81(25) 21.03(50) 2240 Cemdata 07.2 [7 12]CSH_16 1.6 Ca[2+] , 3.2 H[+] , 2.18 H2O, 1 Si(OH)4(aq) , 29.74(0) 27.76(25) 25.99(50) 1892 [13]CSH_08 0.8 Ca[2+] , 1.6 H[+] , 0.34 H2O, 1 Si(OH)4(aq) , 11.42(0) 10.8(25) 10.19(50) 1630 [13]Gypsum 1 Ca [2+] , 1 SO4[2

], 2 H2O 4.5331(0) 4.4823(25) 4.6094(60) 2305.16 [6]Thaumasite 6 Ca [2+] , 2 HCO3[

], 6 H[+] , 28 H2O, 2 Si(OH)4(aq) , 2 SO4[2 ] 17.94(0) 18.86(25) 18.34(50) 1768 Cemdata 07.2 [7 12]

Ettringite 6 Ca [2+] , 2 Al[3+] , 12 H[+] , 38 H2O, 3 SO4[2 ], 62.08(0) 56.85(25) 51.15(50) 1774 Cemdata 07.2 [7 12]

Calcite 1 H[+] , 1 Ca[2+] , 1 HCO3[ ] 2.2257(0) 1.8487(25) 1.333(60) 2709.89 [6]

Fe-monocarboaluminate 4 Ca [2+] , 1 HCO3[ ], 13 H [+] , 17 H2O, 2 Fe[3+] , 37.86(0) 33.36(25) 28.76(50) 2144 Cemdata 07.2 [7 12]

Strtlingite 2 Ca [2+] , 2 Al[3+] , 10 H[+] , 11 H2O, 1 Si(OH)4(aq) , 55.97(0) 49.86(25) 44.37(50) 1768 Cemdata 07.2 [7 12]Monocarboaluminate 4 Ca [2+] , 2 Al[3+] , 1 HCO3[

], 13 H [+] , 17 H2O, 88.79(0) 80.61(25) 73.05(50) 2169 Cemdata 07.2 [7 12]Monosulfoaluminate 4 Ca[2+], 2 Al[3+], 12 H[+], 18 H 2O, 1 SO4[2 ], 80.76(0), 72.49(25), 64.85(50),

57.8(75), 51.34(100)2014 Cemdata 07.2 [7 12]

Fe-monosulfoaluminate 4 Ca[2+], 12 H[+], 18 H 2O, 1 SO4[2 ], 2 Fe[3+], 76.27(0), 67.63(25), 59.53(50), 51.99(75), 45.03(100)

2118 Cemdata 07.2 [7 12]

C3AH6 3 Ca[2+], 2 Al[3+], 12 H[+], 12 H 2O, 90.42(0), 81.25(25), 73.11(50), 65.84(75), 59.35(100)

2521 Cemdata 07.2 [7 12]

77L. Barcelo et al. / Cement and Concrete Research 63 (2014) 75 88


4/14

The system composition was de ned as follows: a model OPC ,composed of:

C2.7 S C3A (1 to 9% wt. of OPC), C4AF so that C3A + C4AF = 10% wt. of OPC Gypsum (3% wt. of OPC).

The Ca/Si atomic ratio of 2.7 used here to represent the calciumsilicate phases is a typical value for standard OPC, simulating an OPCwith a C2S/C3S mass ratio of 0.42.

To this OPC was added variousamounts of limestone (5, 15 and 20%wt. of blend), treated as pure calcite. The impact of SCMs on the mineralassemblage was estimated by adding pure reactive silica to the systemat 10% by mass of the total blend.

All the calculations were run at 5 C, in order to be representativeof CSA A3004-C8 procedure B. All phases present were allowed toreact completely so as to reach equilibrium. Thus, the initial systemstate was calculated as the equilibrium state for a fully reactedmixture of the OPC + limestone (+ silica fume) blend with waterat a water/solid ratio of 0.5, which corresponds to a fully-hydratedcement paste.

Then this initial system was ushed 333 times with a sodium sul-fate solution at 30 mmol/L. Every ushing step consists in removingthe totality of the interstitial solution (at equilibrium with the stablemineral assemblage) and replacing it by an equivalent volume of sodi-um sulfate solution. This approach simulates, without taking into ac-count the complex process of transport in the porous medium, whatmayhappenat equilibrium when a mineral assemblage is put in contactwith an aggressive medium, including leaching and ingress of externalions.

The phase assemblages are expressed graphically as phase volumes(in mL/100 g of binder) for minerals and porosity on the ordinate, as afunction of the number of ushing steps on the abscissa (333 steps onthe right hand side of the graphs).

2.3. Thaumasite growth rate experiments

Experiments were carried out in order to try to understand the fac-tors in uencing the rate of growth of thaumasite, using dilute suspen-sions of synthetic C-S-H (with a Ca/Si ratio of 1) mixed with gypsumand ultra ne calcium carbonate in proportions corresponding to thestoichiometry of thaumasite (a 1:1:1 molar mixture). During thisprocess, the addition of several potential catalysts was also tested.

The experimental procedure is detailed below.Laboratory-grade chemicals (typically N99% pure) were usedthroughout, and the water used was de-ionized water which had beenboiled to remove dissolved CO 2. For each experiment, 1.25 mol (2.653g) of Na2SiO3 5H 2O were dissolved in 500 ml of water. To this wasadded a solution of 1.25 mol (1.839 g) of CaCl 2 2H 2O in 500 ml of water, and the mixture (which gave a massive precipitation of C-S-Hwith a Ca/Si atom ratio very close to 1) was agitated for 30 min in aclosed polyethylene container. The slurry was vacuum- ltered on agrade 42 lter and washed four successive times with 50 ml aliquotsof water. The wet C-S-H was then scraped off the lter (using a Te onspatula) and placed into a ask, together with 1.25 mol (2.152 g) of CaSO4 2H 2O and 1.25 mol (1.251 g) of precipitated CaCO 3 . The askwas then partially lled with additional water to give a water/initialsolid mass ratio of 20, and small amounts of additional reagents wereadded if desired, (typically either readily water-soluble compounds or,if poorly-soluble solids, then as nely-ground powders). The ask wasimmersed in a temperature-controlled water bath and the mixture(suspension) in the ask was agitated continually by means of aTe on-coated magnetic stirrer. Experiments were run mostly at 20 Cand 8 C, but some were also run at 5 C. Small samples of the suspen-sion were removed every week, ltered and dried with acetone andether, and subject to X-ray diffraction (XRD) analysis. Rough estimatesof the degree of advancement of the reaction were made by comparingthepeakheights of thestrongestpeak for thaumasite withthe strongestpeaks for gypsum and calcite. In cases where gypsum disappearedcompletely, the thaumasite formation reaction was assumed to be

Fig. 2. Evolution of expansion with time at 23 C (CSA A3004-C8 procedure A) for the different systems tested by the different laboratories.

http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B2%80


5/14

complete; but there was always some residual calcite present, mostprobably due to atmospheric carbonation.

3. Results

3.1. Expansion testing

3.1.1. 23 C expansion results (CSA A3004-C8-procedure A)The evolution of expansion with time at 23 C for the different

cementitious systems tested in the different laboratories is presentedin Fig. 2. Results are consistent with expectations. Themixtures withoutSCM expand quite rapidly and eventually exceed the expansion limit of 0.05%. The addition of SCM (20%slag or higher and 15% Type F y ash orhigher) signi cantly reduces the expansion and none of the systemswith SCM exceed the 0.05% expansion limit at 6 months or the alterna-tive limit (for HSb or HSLb) of 0.10% at 12 months.

3.1.2. 5 C expansion results (CSA A3004-C8-procedure B)Expansionresults with time at 5 C (CSA A3004-C8-procedureB) are

presented in Fig. 3. Compared to the equivalent experiments at 23 C,the levels of expansion reached are generally much larger, especiallyin mixtures containing SCMs. When such large expansions occur, com-paring expansion values to try to assess the performance of differentsystems can be misleading as mortar bars gradually lose integrity,

undergo warping, etc. Therefore, in order to better compare the differ-ent systems, a 5 C failure index is de ned here for evaluating theCSA A3004-C8-procedure B test data. It is de ned as the time difference(in days) between the moment when the mix reached 0.10% expansion(estimated by linear interpolation of the data) and 18 months, which isthe duration of the test (see Fig. 4). Mixes that meet the expansioncriteria ( b 0.1% at 18 months) have a failure index of 0. Otherwise, thesooner the 0.1% threshold is reached, the higher the failure index.Note that this index is designed to compare the performance of mixesin the context of this article and has not been correlated with any kindof eld performance.

A plot of the 5 C failureindex for thedifferent mixes in the3 differ-ent laboratories is shown in Fig. 5. As expected, mixes without any SCMaddition have the highest failure indexes as they reach the expansionlimit very rapidly.

Performance with SCM additions shows a much-contrasted picturebetween the different laboratories:

For slag additions: InUNB,20%slagmixesfailed and 40% slag mixes passed the 5 C test

(for both OPC and PLC) In UOT, 30% slag mixes with OPC passed the5 C test while with PLC

they did not (the PLC with the lowest amount of limestone did notpass the 24 months limit, as the expansion between 12 and18 months was larger than 0.030%). All mixes passed at 50% slagaddition;

In LG, all mixes (PLC or OPC) failed at 50% slag addition. At 65% slagaddition, the slag-PLC combination tested passed, while the slag

OPC combination failed to meet the expansion limit. For y-ash:

In UNB, some mixes (both with OPC and PLC) met the 5 C expan-sion limit even for addition levels as low as 15%, while in morerecent test series (with a different clinker) mixes at 25% Type F y

ash addition failed (both with OPC and PLC).

Fig. 3. Evolution of expansion with time in the CSA A3004-C8 test procedure B (5 C) for different SCM combinations. See legend. Speci cation limit for HS cement: 0.1% expansion at18 months.

Fig. 4. The 5 C failure index.

http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B4%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B3%80


6/14

For LG, mixes with 25% Type F y ash addition failed (both with OPCand PLC). Some mixes at 35% addition met the expansion limit,while others with different cements and y ashes failed even at50% ash addition.

3.1.3. Expansion at 5 C vs. expansion at 23 C Results from Figs. 2 and 3 suggest that much larger expansions are

reached at 5 C compared to 23 C. However, as shown in Fig. 6-A to F,this does not appear to be true early on: at 56 days, 91 days and

Fig. 5. 5 C failure index for the different mixes from the different laboratories. Color key is similar to Figs. 2 or 3.

Fig. 6. Expansions at 5 C vs. 23 C fordifferenttimes.Red linecorresponds to orthogonalregressionline (variance ratio of 1) andthe green linerepresents the equality line. A: 56 days, B:

91 days, C: 105d, D: 182 days, E: 182 days considering expansions below 0.1% only and F: 364 days considering expansions below 0.1% only.



7/14

105 days, the levels of expansion at 5 C or 23 C are very similar. At182 days andlater expansions at 5 C start to become signi cantly larg-er than expansions at 23 C, even when the data sets are restricted tosamples that expanded less than 0.1%.

3.1.4. In uence of limestone addition (OPC vs. PLC)In this section, a comparison is made between the performance of

OPC (with 2 to 4% limestone addition) with PLC having limestonecontents between 10 and 15%. Figs. 7 and 8 show the results of matched-pair t-tests done on the 6 month expansion at 23 C and onthe 5 C failure index, respectively. At 23 C, when analyzing the dataglobally ( Fig. 7-A), the average matched-pair difference at 6 months is0.017% with a slightly higher expansion for the mixes with PLCcompared to OPC. However, this difference is not statistically signi cant( = 0.05). It also seems that most of thedifferencecanbe attributed tothe mixes with high expansion values (i.e. exceeding the 0.1% failurelimit). Fig. 7-B, which represents expansions below 0.1%, con rms thisand shows a different picture: at low expansion values (i.e. mixes thatare sulfate resistant), mixes with PLC tend to expand slightly less thantheir OPC counterpart.

At5 C(cf.Fig. 8-A), theaverage difference in failure index is 16 days(in favor of OPC) and this difference is not statistically signi cant ( =0.05). In Fig. 8-B only the higher failure indexes (mixes that reach the

expansion limit more rapidly) are considered and again, there is nostatistical difference between OPC and PLC.

3.2. Thermodynamic modeling

3.2.1. Typical chemical path when ushing a cement paste with a Na 2 SO4solution at 5 C

The examples below refer to an OPC with 5% wt. C 3A and 5% wt.C4AF.

Withoutlimestone (Cf. Fig. 9), theinitialmineral assemblageat 5 C iscomposed of C-S-H, portlandite, ettringite, C 4AH19 and C4FH13 . It shouldbe noted that from this modeling, monosulfate is not predicted asC4AH19 appears to be more stable. However, the level of undersaturationof monosulfate is very lowand this phasecould very well appear insteadof C4AH19 in reality.

Upon ushing with a 30 mmol/L sodium sulfate solution, the phaseassemblage is modi ed in the following way: formation of ettringite atthe expense of C 4AH19 and portlandite, then formation of Fe-ettringiteat the expense of C 4FH13 once C4AH19 is depleted, and nally formationof gypsum when C 4FH13 is depleted, inducing further consumption of portlandite.

During this process, the solid volume increases by about 13%, of which 10% is due to ettringite. However, this does not necessarilyimply that the paste will undergo bulk expansion (see discussion).

Fig. 7. MatchedPairs t -test differencein the6 monthexpansionat 23 C (%)vs. mean 6 month expansion at 23 C.Red lineand dottedlines represent theaverage differenceand its95%con dence limit. A: All samples and B: combinations with expansions lower than 0.1%.

Fig. 8. MatchedPairs t -test difference in the5 Cfailure index(d) vs. Mean 5 Cfailure index. Red line anddottedlines representthe averagedifference and its 95% con dence limit. A:

All data and B: potential outlier removed.



8/14

In presence of 15% limestone (cf. Fig. 10), the initial assemblage at5 C is composed of C-S-H,portlandite, ettringite,monocarboaluminate,Fe-monocarboaluminate and calcite.

Upon ushing with 30 mmol/L sodium sulfate solution, the phaseassemblage is modi ed as follows: formation of ettringite at theexpense of monocarboaluminate and portlandite, then formation of thaumasite at the expense of C-S-H, portlandite and calcite, and nallyformation of gypsum at the expense of portlandite when calcite isdepleted.

The solid volume increases by about 10% due to ettringite formationand then a further 45%due to thaumasite formation. Thepossible impli-cations of these volume changes will be discussed later.

An interesting aspect of the calculation results is that gypsum pre-cipitates only once all possible thaumasite formation has occurred(which in this case is limited by the exhaustion of calcite). This is in

contradiction with some experimental observationsthat detect gypsumformation before thaumasite. We interpret this discrepancy as beingdue to the slow kinetics of thaumasite formation in reality, as opposedto the idealized instantaneous equilibrium of the calculations.

Fig. 11 gathers the results of the simulations for different limestone(5, 15 and 20%) and C 3A (1, 5 and 9%) contents. The sequences are thesame as for the example previously described: formation of ettringiteat the expense of monocarboaluminate and portlandite, then formation

of thaumasite at the expense of C-S-H and calcite, and nally formationof gypsum when thaumasite formation is complete. The amount of ettringite formed obviously increases with the C 3A content of the bind-er, while the amount of thaumasite formed in these calculations in-creases with the limestone content because there is always an excessof silica available from C-S-H and thus it is the amount of calcitewhich is limiting.

3.2.2. Role of pozzolanic SCMs: evaluation using silica fumeThe system considered here is anOPC with 5%C 3Aand5%C4AF,used

in a blend containing 15%limestone. Therole of pozzolanicSCMs on thissystem was estimated by adding 10% silica fume to the initial blend. Ascan be seen in Fig. 12, the nature of the minerals is unchanged in theinitial system: C-S-H, portlandite, ettringite, monocarboaluminate, Fe-monocarboaluminate andcalcite. Of coursethe amounts of themineralshave changed, essentially some portlandite has combined with silica toform more C-S-H.

Upon ushingwitha 30 mmol/L sodiumsulfatesolution,the miner-al assemblage is modi ed in the following way: formation of ettringiteat the expense of monocarboaluminate and portlandite, then formationof thaumasite at the expense of C-S-H and calcite until calcite is deplet-ed.At thesametime, some monocarboaluminate forms again at theex-pense of ettringite when portlandite is depleted. The calcium sulfatereleased by this ettringite also goes into forming thaumasite. As limesaturation is further reduced by the sodium sulfate solution there isthen formation of strtlingite and nally signi cant decalci cation of C-S-H. Contrary to what was observed without silica fume, gypsumdoes not appear, due to the lower calcium activity in this system.

Although thesequence of phase changes is subtly different from thecase without silicafume,it is believed that this difference cannot explainwhypozzolanic SCMs improve thesulfate resistance of OPC limestone

SCM systems. In both cases, similarly large volumes of ettringite andthaumasite are predicted to form after about the same degree of reac-tion with the sulfate solution.

Simulations fordifferent limestone(5,15 and20%) andC 3A(1,5and9%)contentshave alsobeen performed, in all cases with10% silica fume.Overall, the sequence of phase formation does not change, although theamount of ettringite formed increases with increasing C 3A content,while the amount of thaumasite formed increases with increasing lime-stone content.

Based on these calculations, the following conclusions can be drawnregarding the equilibrium phase assemblages produced by immersionin alkali sulfate solutions:

during contact witha sodium sulfate solution for an OPC + limestonebinder: the amount of ettringite formed depends primarily on the C 3A

content of the binder the amount of thaumasite formed depends mainly on the amount

of limestone available gypsum should not form before all thaumasite formation is

complete. The fact that gypsum is often observed to form beforethaumasite in real exposed concretes can be attributed to the lowrate of growth of thaumasite.

when additional reactive silica is brought in the system, as is the casewith pozzolanic SCMs: the sequence of ettringite and thaumasiteformation does not change signi cantly. Therefore the positive effect

of pozzolanic SCMs on sulfate resistance is not likely to be the result

0.00 100 200 300

0.2

0.4

0.6

0.8

1.0

1.2

flushing steps

v o l u m e o

f s o

l i d s

( m L / 1 g o

f b i n d e r )

CSH_16

ettringite

gypsum

C4AH19

portlandite

C4FH13

Fe-ettringite

Fig. 9. Evolution of the mineral assemblage of an OPC cement paste when ushed 333times with a 30 mmol/L sodium sulfate solution, at 5 C.

0.0

0.2

0.4

0.6

0.8

1.0

1.2

v o l u m e o

f s o

l i d s ( m

L / 1 g o

f b i n d e r )

CSH_16

thaumasite

ettringite

gypsum

calcite

portlandite

Fe-monocarboaluminate

monocarboaluminate

0 100 200 300

flushing steps

Fig. 10. Evolution of the mineral assemblage of an OPC + 15% limestone cement paste

when ushed 333 times with a 30 mmol/L sodium sulfate solution, at 5 C.

http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B9%80


9/14

of a differentequilibrium phase assemblage. It seems more reasonableto propose that it may be the result of a lower rate of sulfate ingress(dueto the lower permeability of the cement paste), perhaps coupledwith enhanced paste strength. This point probably needs furtherinvestigations as Kunther et al. [14] did not nd any impact of thetype of binder on the depth of sulfate ingress.

3.3. Thaumasite growth rate experiments

It was anticipated that some of the materials tested would act as cat-alysts for thaumasite formation in stirred C-S-H/gypsum/calcite suspen-sions. Small amounts (1% by mass of totalsolids) of nely divided solidswere tested as potential nucleating agents, including pure lab-grade

Mg(OH) 2 , Al(OH)3 , Ca(OH)2 , and also a sample of ettringite made by

agitating an aqueous suspension of lab-grade Ca(OH) 2 and Al2(SO4 )3in a 6:1 molar ratio. Obtaining pure thaumasite for use as a nucleatingagent was initially not possible, so the impure source used for the pre-liminary experiments was simply a pulverized sample of paste takenfrom a eld specimen of a severely TSA-degraded concrete containinga fairly high fraction of thaumasite as shown by XRD. Some water-soluble compounds were also tested as potential catalysts, includingNaOH, triethanolamine and sucrose.

The principal results of these screening tests can be summarized asfollows:

1. At both 8 C and 20 C, no thaumasiteformation could be detected inthe pure agitated suspensions after 12 weeks except for suspensions

to which 1% of either pure ettringite or impure thaumasite had

Fig. 11. Impact of the C3A and limestone contents on the evolution of the mineral assemblage of an OPC + limestone cement paste when ushed 333 times with a 30 mmol/L sodiumsulfate solution, at 5 C.



10/14

initially been added. But the suspensionswith ettringite or thaumasiteadded showed signi cant amountsof additional thaumasiteformationafter as little as 2 weeks, and the amount of thaumasite continuedto increase rapidly with time after that. Moreover, in a test runwith the combined initial addition of 1% impure thaumasite and1% portlandite, the suspension was apparently completely reacted(i.e. no gypsum remained) after 13 weeks at 8 C.

2. None of the water-soluble chemical additives tested individuallyresulted in any detectable thaumasite formation after 12 weeks.But the use of a 0.16 molar NaOH solution coupled with a very high(20%) aqueous concentration of sucrose was a very effective catalystdespite including no solid nuclei. It resulted in the complete con-sumption of gypsum by 7 weeks at 5 C and by 12 weeks at 20 C.This use of alkaline sucrose as a catalyst (already well known fromthe literature) is interesting in that it appears to be capable of accel-erating the nucleation process itself, and not just the growth of thaumasite; butit is only really effectivewhen very high sucrose con-centrations are used, and does not represent a situation that is likelyto occur under any natural eld conditions.

These results show that the nucleation of thaumasite is very slow inagitated mixtures of C-S-H, gypsum and precipitated (ultra ne) calcite;but that thaumasite, once nucleated, can grow at a signi cant rate. It isalso important to note that pure ettringite was very effective at nucleat-ing thaumasite growth, and this has important implications for betterunderstanding the sequence of events that may occur during the degra-

dation of concrete under sulfate attack, where ettringite inevitablyforms before the conditions are right for thaumasite to form.

A fewadditional tests were conducted to determine the rate of reac-tion of hydrated Portland cement mortars with gypsum and limestone,using a similar methodology. The mortars were crushed to completelypass a 3-mm sieve, from which only the material passing a 200 msieve was taken, this being strongly enriched in cement paste. Thismaterial was mixed with 11% gypsum and 7% nely ground limestone(calculated to be more than suf cient quantities to convert the pasteinto a mixture of ettringite and thaumasite) and stirred strongly inexcess water at room temperature (22 2 C) for two months. XRDanalyses of the products clearly showed the presence of ettringite butwere inconclusive for thaumasite. This supports the hypothesis thatettringite forms before thaumasite during sulfate attack on Portland

cement pastes.

4. Discussion

In the recent years, there has been a renewed interest among thescienti c community in the sulfate resistance of cements containinglimestone (see for instance [9,15 34] for relevant research or reviewsin the last 10 years). A signi cant portion of this recent work was fo-cused on the conditions of thaumasite formation in the case of externalsulfate attack and was probably mainly fueled by the occurrence of a

few high-pro le cases of TSA in the UK at the end of the 90s (in con-cretesmade with limestoneaggregates)and subsequentlyby theindus-trial development of limestone-based blended cements.

4.1. Thaumasite formation in the sequence of sulfate attack reactions

As shown in the thermodynamic modeling section, in equilibratedcement pastes containing carbonates, it is clear that thaumasite forma-tion follows ettringite formation as the content of sulfate is increased,and can only occur after all of the AFm phases have been convertedinto ettringite. As expected, thaumasite grows at the expense of C-S-H,portlandite and calcium carbonates (the sulfate ions being suppliedexternally). This indicates that thaumasite formation does not occurconcurrently with traditional sulfate attack but rather represents alater stage of the process of concrete deterioration. This point is also invery good agreement with the conclusions of the thermodynamicmodeling study of Schmidt et al. [9].

A growing number of microstructural analyses made on samplesthat contain carbonates and are subject to sulfate attack also recognizethat ettringite appears rst and is followed by thaumasite formation[9,18,32,34] . Based on the work of Gollop and Taylor [35], Irassar et al.[18] summarize the sequence of the processes that occur in cementpastes immersed in fairly concentrated alkali sulfate solutions:

1. Diffusion of sulfate ions and leaching of portlandite;

2. Ettringite formation;3. Gypsum formation and depletion of portlandite;4. Decalci cation of C-S-H5. Thaumasite formation.

According to this sequence of events, thaumasite formation shouldonly occur in already severely degraded portions of the sample. In theCSA A3004-C8-procedureB testing, this corresponds well with observa-tions made by Ramezanianpour and Hooton [32], who detected signi -cant amounts of thaumasite in thesoft deteriorated piecesof spalled-off surfaces but did not detect thaumasite in the solid mortar pieces.

The main difference between microstructural analyses and thermo-dynamic modeling lies in the apparition of gypsum [9,32]. While gyp-sum is predicted to precipitate only after all possible thaumasite hasformed (if limited by depletion of carbonates), it is frequently identi edin laboratory studies prior to theformation of thaumasite. This hasbeenattributed to the very high alkali sulfate concentrations used in thestandard laboratory experiments, coupled with the very slow rate of

thaumasite formation [9]. Schmidt et al. [34] and Yu et al. [36] alsostate that this gypsumprecipitation does not make an important contri-bution to the damage process, as it forms in pre-existing cracks at lowsupersaturation levels. We believe that the relatively high mobility of both calcium and sulfate ions in the systems in question also facilitatesgypsum formation in relatively unstressed locations where it createslittle if any damage. Note that precipitation of gypsum is directlycoupled with consumption of portlandite and ultimately also the decal-ci cation of C-S-H. It only occursif the sulfate solution usedto attack thespecimen contains sulfate ions at a concentration high enough to formgypsum by reaction with portlandite or C-S-H. This is generally thecase in accelerated lab tests using sodium sulfate solutions, but notnecessarily the case in all sulfate eld exposure conditions. However,this consumption of calcium hydroxide per se is not required, thermo-

dynamically, before thaumasite formation.

0.0

0.2

0.4

0.6

0.8

1.0

1.2

v o l u m e o

f s o

l i d s

( m L / 1 g o

f b i n d e r )

thaumasitecalcite

ettringite

CSH_08

CSH_16

strtlingite

monocarboaluminate

portlandite

Fe-monocarboaluminate

0 100 200 300

flushing steps

Fig. 12. Evolution of themineral assemblageof an OPC+ 15%limestone+ 10%silicafumepaste when ushed 333 times with a 30 mmol/L sodium sulfate solution, at 5 C.

http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2http://localhost/var/www/apps/conversion/tmp/scratch_6/image%20of%20Fig.%E0%B1%B2


11/14

4.2. Mechanism of thaumasite formation

In classical sulfate attack, (for a description of the mechanism, see[37] ) crystallization pressure arising from the conversion of very neAFmcrystals into ettringite is considered to be thesourceof thedamage[14,36] . This mechanism requires con nement of the crystals in verysmall pores ( b 0.1 m) and also requires thesolution to be highly super-saturated with respect to ettringite. This mechanism also explains why,

although the increasing volume fraction of ettringite is the drivingforcefor the expansion, there is no good direct correlation between theamountof ettringite formed andthe observed expansionwhen compar-ing different systems [14].

In the thaumasite form of sulfate attack, the mechanism of degrada-tion appears to be very different. While there is controversy regardingwhether or not there is any expansion uniquely associated withthaumasite growth, the most damaging result appears to be relatedto the disappearance of C-S-H and its replacement with a form of thaumasite that appears to have no binding properties [27,31] . Thus,the end product of the attack is a paste without cohesion thathas been widely described as a mush . Therefore, as the growth of thaumasite is not clearly associated with expansion, but the CSAA3004-C8-procedure B speci cation nevertheless involves lengthchange measurements, which are explicitly intended to determineresistance to the potential for the thaumasite form of sulfate attack,

we can legitimately wonder if this procedure is appropriate and likelyto meet its objectives.

Several reaction pathways or mechanisms for thaumasite formationhave been proposed in the literature. Bensted [38] proposed two differentreaction pathways:

1. A direct route involving the reaction of C-S-H with gypsum andcalcite;

2. From ettringite, C-S-H and calcite via the woodfordite route,woodfordite being the name sometimes used for a partial range of solidsolutions between ettringiteand thaumasite:Bensted proposeda reaction between ettringite, C-S-H and calcite to give thaumasite( rst passing through woodfordite) plus (apparently) calcium and

aluminum hydroxides.Crammond [15] proposed three possible mechanisms:

3. A through solution mechanism;4. A topochemical replacement of ettringite by thaumasite;5. A heterogeneous nucleation process, ettringite acting as a template

for thaumasite nucleation.

These ve propositionsarenot directly comparable because they arenot truly independent; some are mechanisms and others are pathways.It is therefore necessary to clarify the main mechanistic issues in thelight of the data presented in this paper, which we choose to do byposing three key questions:

1. Is there a direct reaction between C-S-H, gypsum and calcite?: This

pathway was clearly demonstrated in our own thaumasite growthrate experiments as discussed above. Such a reaction also clearly in-volvesthe passage of many (ifnot all) of thecomponent ions throughsolution, because gypsum, calcite and C-S-H are all consumed duringthe reaction, while no intermediate solid phases are observed toform. One must therefore assume that at least two (and probablyall three) of the starting materials must fully dissolve and all of their component ions must pass through solution to the site of thaumasite formation. So one cannot avoid the conclusion that a through solution mechanismmust be involved in this simpledirectpathway.

2. Does ettringite act as a template for thaumasite nucleation?: Ourwork clearly indicates that the direct reaction requires an initialnucleation step, which is extremely slow in the presence of just

th e normal starting materials plus some other common potential

concrete impurities. Theresults indicate that C-S-H, gypsum and cal-cite, as well as portlandite, brucite and gibbsite, do not act as goodnucleationsitesfor thaumasite.Whenthaumasite is added,however,the nucleation problem is overcome; and the use of ettringite ap-pears to be similarly effective. It is concluded, therefore, that nucle-ation of thaumasite can be a rate-determining step in the reaction,and that ettringite surfaces can de nitely act as a site for thaumasitenucleation.

3. Is topochemical replacement of ettringite by thaumasite possibleunder TSA conditions?: Isomorphic replacement reactions are wellknown in geology, but tend to be very slow. Crammond suggests thepossibility of such a replacement mechanism during TSA but did notgive a very plausible pathway. But we can consider the woodforditeroute proposed by Bensted to be a pathway involving replacementof ettringite by thaumasite,so thispathway, if it is thermodynamicallypermitted, could be a mechanism for the topochemical process. How-ever, our thermodynamic analysis shows that it cannot occur inthe way described by Bensted (i.e. as a reaction between ettringite,C-S-H and calcite to give thaumasite plus gypsum, portlandite andgibbsite). So, under what conditions might a topochemical replace-ment reaction actually occur? In order to help answer this question,it is helpful to treat the reaction as the simplest conceivable exchangeof ions or dissolved neutral species with the solution; for example:

C3A 3CD 32H 2 Si OH 4 2 HCO3

2 CS C^

C CD 15Hg 2 Al OH 3 SO4 :n

The above equation 5 shows clearly that sources of silica and carbondioxideare needed for the reaction to progress, and that the reaction it-self produces alumina and sulfate in addition to thaumasite. The factthat this reaction actually releases sulfate ions clearly shows that it isunlikely to be important in sulfate attack, because increasing sulfateconcentrations will tend to drive the reaction in reverse. On the otherhand, the fact that this reaction consumes carbon dioxide suggeststhat it might occur, for instance, in concretes subject to attack by highlycarbonated waters. Matschei and Glasser [32] showed clearly thatthaumasite should form at the expense of ettringite (and that their

solid solution should form rst) during concrete carbonation, providedthat the relative humidity remains high enough. So it is now clear thatthe woodfordite route should not occur during simple sulfate attackconditions, although it is theoretically possible under carbonationconditions.

As further evidence for our own mechanistic conclusions, we canalso cite the extensive study of the conditions of thaumasite nucleationin cement pastes and dilute cement suspensions by Kohler et al. [19].They testeddifferent sulfate carriers (sodiumsulfate or gypsum), differ-ent temperatures (5 C and 10 C), different base cement composi-tions (pure C 3S, alite, mixtures of alite and C 3A, mixtures or alite andC4AF),different seeding conditions (no seeding, seeding with ettringite)and the addition (or not) of very ne silica. They concluded that: Woodfordite was rarely observed and there was no evidence to sup-

port the idea of it being an intermediate step in the reaction to formthaumasite. Mixtures containing ettringite and calcite but no addi-tional source of sulfate did not form thaumasite.

Thaumasite wasalmostnever observed in mixes containing no alumi-na, even if the stoichiometric conditions for its formation were met.Without prior formation of ettringite, the formation of thaumasite incement pastes is extremely slow.

The heterogeneous nucleation mechanism involving ettringite is themost likely, with thaumasite growing as an epitaxial overgrowth onthe surface of ettringite.

The observations of our own study are certainly consistent with theconclusions of Kohler et al. [19].

5

$ represents SO 3 and represents CO 2



12/14

4.3. Kinetics of thaumasite formation

Although the thermodynamic analysis shows clearly that ettringiteformation should precede thaumasite formation in the normal se-quence of sulfate attack reactions on Portland cement pastes, the ques-tion remains as to why thaumasite formation, when it does occur, is somuch slower than ettringite formation. This is clearly the case in thelaboratory and also appears to be true inthe eld. While some have pro-

posed theunusual silicon in 6-coordination with hydroxyl as a potentialexplanation for thelow formationrate [38], we here propose an alterna-tive explanation based on our own observations.

Ettringite can be made easily and rapidly in the laboratory by themethod used here, involving the reaction between two relatively solu-ble precursor phases: portlandite (moderately soluble) and aluminumsulfate (highly soluble), via the overall reaction shown below (in anexcess of water):

AD3H18 6CH C3A 3CD 32H:

Because of the relatively high solubility of both initial reactants, theinitial supersaturation can locally reach very high values, which leadsto rapid nucleation. And, for similar reasons, diffusion of ions through

solution is unlikely to be rate limiting at least until thedegree of reactionis quite high.

In the more practically relevant case of sulfate attack on hardenedconcrete, ettringite typically forms by the reaction of one very poorlysoluble phase(e.g.monosulfoaluminate-AFm) withone moderatelysol-uble phase (e.g. gypsum) in the presence of excess water, as shownbelow:

C3A CD 12H 2CDH2 C3A 3CD 32H:

In this case, the calcium and sulfate ions can diffuse quite rapidly to-ward the poorly soluble precursor AFm phase, which means that theettringite formed is likely to occur very close to the site of the AFmphase, rather than close to the source(s) of calcium and sulfate. In thecase where sulfate ions are provided by an alkali sulfate solution theirconcentrations can be even higher. The calcium in that case will usuallycome from portlandite or C-S-H in the surrounding cement pastematrix, at moderate concentrations. But, either way, there is no needfor ions to diffuse long distances under very low concentrations gradi-ents, so rates of reaction can still be fairly rapid.

On the other hand, there are no such easy routes to thaumasite for-mation, because in all practical cases the reaction involves at least threeseparate precursor phases, of which at least two (e.g. C-S-H and calcite)have extremelylow solubilities. Thereaction(in excesswater) occursasshown below:

C S H C^

C CDH2 CS C^

C CD 15H:

Since there is no known stable intermediate phase containingboth silicate and carbonate ions, the reaction must involve counter-diffusion of carbonate and silicate ions under conditions where bothare present at extremely low source concentrations. Thus, it is unlikelythat a high degreeof supersaturation will occur initially (which explainsthe very low nucleation rates observed) and the rate of growth evenafter nucleation is also likely to be severely limited by the low averageconcentrations of silicate and carbonate ions (which also implies lowconcentration gradients). Additional support for this conclusion comesfrom qualitative observations that (a) strong agitation of the suspen-sions signi cantly enhances the reaction rate, and (b) the use of groundlimestone as the source of calcite, instead of the ultra ne precipitatedcalcite used in most of our suspension experiments, results in a much

lower reaction rate.

4.4. Relevance of CSA A3004-C8 procedure B

Regarding therelevanceof theCSAA3004-C8 procedureB to deter-mining resistance to the potential for the thaumasite form of sulfateattack , we rst summarize its main features as follows:

1. Likeprocedure A designed for assessing resistanceto traditional sul-fate attack, procedure B is a length-change measurement procedure;

2. Samples are kept immersed in a highly concentrated sulfate solutionat low temperature (5 C);

3. Immersion starts when samples have reached a compressivestrength of 20 1 MPa.

4.4.1. The relevance of length-change measurement Since both thermodynamic analysis and microstructural examina-

tions have shown that thaumasite formation can occur only after all po-tential ettringite formation is nished; and since it has also been shownthat, even once it becomes thermodynamically stable, thaumasitenucleation is very slow when ettringite is absent, it is reasonable touse a method that is sensitive to the formation of ettringite to evaluatethe potential for thaumasite formation. Moreover, direct evaluation of thaumasite formation using the currently available methods (XRD forinstance) would be dif cult and impractical in the context of anindustrial quality control procedure. Length change measurementappears adequate to ful ll this objective. An alternative approachusing surface-sensitive leaky Rayleigh waves was examined recently[34] , but the conclusion of the study was that this technique did notbring anyadvantage in terms of sensitivity compared to a lengthchangemeasurement. However, a qualitativerating of thesurface/edge damageof the test specimens could be added to the current procedure at littlecost.

One might reasonably still ask whether or not there might be a pessimum scenario, (e.g. for low C 3A cements), where the amount of ettringite formed would be small enough to generate only modestexpansions but yet still high enough to initiate subsequent thaumasiteformation that might severely weaken the samples without necessarilygiving much linear expansion. However, although this risk cannot becompletely discounted, it now appears quite unlikely, because the evi-dence shows that, if the microstructure is not opened-up by cracking,thaumasite formation tends to remain purely a surface phenomenon.As Irassar concludes in his review article [27]: Regardless of limestone

ller content, in pastes, mortars and concretes made with low-C 3Acements ( b 5%) and using a low effective w/c ratio ( b 0.50), whichcomplies with the ACI 201 recommendation for moderate sulfateenvironments, no early damage at low temperatures resulted. In thiscase, thaumasite is also found on the specimen surface on long timeexposure, but it lacks relevance in the damage process .

4.4.2. Comments on the use of low temperature and the 20 MPa strengthcriterion

As shown in this article, longer-term expansions aregenerally great-

er at 5 C than at 20 C, independently of the presence or absence of limestone. This is due to an earlier take-off of the expansion (therate during the initial slow expansion phase being apparently indepen-dent of temperature).This take off is generally attributed to theperco-lation of the crack network in the sample [36]. Thaumasite formation isalso generally reported to be thermodynamically favored by lowertemperatures.

Thereason forthe more rapid expansions observed at 5 C is dif cultto elucidate from the research conducted for this study. From the ther-modynamic model of Schmidt et al. [9] it does not seem that this tem-perature change should have a signi cant impact on the driving forcefor ettringite formation. One hypothesis is related to the fact that,when the mortar is put in the sodium sulfate solution it has a fairlylow degree of hydration (noting that 20 MPa is well below its ultimate

fully-cured strength). A low temperature (5 C) applied at this early



13/14

stage will then signi cantly decrease the subsequent hydration ratecompared to the use of 20 C curing, so the maturity of the sample atany subsequent time is likely to be lower in the 5 C test than in the20 C test, and its permeability is therefore likely to be higher. There-fore, sulfate ingress is likely to be faster in the test run at lower temper-atures, resulting in the same or higher crystallization pressures appliedto a weaker microstructure. This hypothesis can also explain why thetendency for greater and more rapid expansions at low temperatures

is not seen in other studies like the one of Schmidt et al. [34] wherethe samples are allowed to cure for 28 days at 20 C before immersionin the sulfate solution at low temperature.

Similarly, this maturity hypothesis can explain the unexpected fail-ure of some mixes containing Type F y ash or slag. Because the reac-tions of these SCMs are especially temperature-sensitive, the effect of reduced sample maturity at the lower curing temperature is likely tobe more severe with mixes containing such SCMs. This puts the mixeswith slag or y ash at a signi cant disadvantage, whereas in most realconcrete structures they would have much more time to hydrate beforebeing subject to signi cant sulfate exposure. This hypothesis is furtherreinforced by the fact that additional results have shown that silicafume additions are much more effective than the other SCMs in reduc-ing the expansion at 5 C, presumably because of its ability to reducethe permeability of the sample much more rapidly. As a result, we be-lieve that further research should be conducted to evaluate alternativecuring procedures foruse (in a modi ed test procedure) prior to sulfateexposure at 5 C.

5. Conclusions

The purpose of this article was to investigate the test method CSAA3004-C8 procedureB and its relevance to assess the risk of thaumasiteform of sulfate attack. Through comparisons of results obtained indifferent laboratories on different cementitious systems as well as ther-modynamicmodelingand kinetic experiments,we candraw thefollow-ing conclusions:

Thelengthchange measurement procedure proposed in CSAA3004-C8appears to be relevant as an indirect evaluation of the risk of thethaumasite form of sulfate attack:

1. Thermodynamic modeling con rms that the thaumasite form of sul-fate attack is the nal step of the sulfate attack and happens after theregular form of sulfate attack;

2. Kinetic experiments con rm that thaumasite formation is alwaysex-tremely slow compared to ettringite formation (and gypsum forma-tion) but can nevertheless be catalyzed to some extent by priorformation of ettringite;

3. Alternative, usually more complex, techniques investigated by otherauthors have notshownany bene t compared to thestandardsimple

length change measurement;4. Visual qualitative rating of corners and edges of the sample could be

added to thetest procedureat littlecost andcould provide additionaluseful information.

The testing condition of 5 C coupled with a high concentration of Na2SO4 (50 g/l) also appear to be relevant.

1. The temperature is representative of the temperature in soils inCanada;

2. High sodium sulfate concentrations have been questioned but theyare representative of sulfate concentrations in certain soils in NorthAmerica;

3. This high sodium sulfate concentration has been used in CSA A3004-

C8 procedure A for a long time and cementitious systems that have

been quali ed by this procedure seem to have performed satisfacto-rily in the eld.

In its current version, the CSA A3004 procedure B appears overlysevere:

1. Many SCM combinations fail the test, independently of the presenceof limestone. This includes some SCMcombinations that have a goodhistory of eld use in sulfate environments;

2. Mixes with limestone do not clearly appear to perform worse thanmixes without limestone, although thisshould be investigated furtheras some datasets are con icting. The fact that PLCs in North Americaare ground so as to give equivalent performance to Portland cementsin terms of strength development may explain the observation.

3. The severity of the test is probably due to the low maturity of thesamples when they are put in the solution at 5 C. At low tempera-tures, hydration rates, in particular for SCM, are expected to bemuch slower, and sulfate ingress is expected to be faster.

4. We suggest modifying CSA A3004-C8 procedure B to ensure suf -cient hydration maturity before the samples are subjected to sulfateattack at low temperatures.

References

[1] CSA3000-08:CementitiousMaterialsCompendium, Canadian StandardsAssociation,Mississauga, ON, Canada, 2008 .[2] ASTM C595-12: Standard Speci cation for Blended Hydraulic Cements, ASTM

International, 2012 .[3] EN 197-1: Composition, Speci cations and Conformity for Common Cements,

European Committee for Standardization, Brussels, Belgium, 2000.[4] CSA A23.1-09: Test Methods and Standard Practices for Concrete, Canadian

Standards Association, Mississauga, ON, Canada, 2009 .[5] CSA 3000-08: Cementitious Materials Compendium Update No. 2, Canadian

Standards Association, Mississauga, ON, Canada, 2010 .[6] van der Lee, Thermodynamic and mathematical concepts of CHESS, Technical

Report No LHM/RD/98/39, Fontainebleau, France, 1998 .[7] B. Lothenbach, F. Winnefeld,Thermodynamicmodelling of the hydration of Portland

cement, Cem. Concr. Res.36 (2006)209 226, http://dx.doi.org/ 10.1016/j.cemconres.2005.03.001 .

[8] T. Matschei, B. Lothenbach, F.P. Glasser, Thermodynamic properties of Portlandcement hydrates in the system CaO Al2O3 SiO2 CaSO4 CaCO3 H2O, Cem. Concr.Res. 37 (2007) 1379 1410, http://dx.doi.org /10.1016/j.cemconres.2007.06.002 .

[9] T. Schmidt, B. Lothenbach, M. Romer, K. Scrivener, D. Rentsch, R. Figi, et al., A ther-modynamic and experimental study of the conditions of thaumasite formation,Cem. Concr. Res. 38 (2008) 337 349, http://dx.doi.org/ 10.1016/j.cemconres.2007.11.003 .

[10] B. Lothenbach, L. Pelletier-Chaignat, F. Winnefeld, Stability in the systemCaO Al2 O3 H2 O, Cem. Concr. Res. 42 (2012) 1621 1634, http://dx.doi.org/10.1016/j.cemconres.2012.09.002 .

[11] B.Z. Dilnesa, B. Lothenbach, G. Renaudin, A. Wichser, E. Wieland, Stability of monosulfate in the presence of iron, J. Am. Ceram. Soc. 95 (2012) 3305 3316,http://dx.doi.org/ 10.1111/j.1551-2916.2012.05335.x .

[12] B.Z. Dilnesa, B. Lothenbach, G. Le Saout, G. Renaudin, A. Mesbah, Y. Filinchuk, et al.,Iron in carbonate containing AFm phases, Cem. Concr. Res. 41 (2011) 311 323,http://dx.doi.org/ 10.1016/j.cemconres.2010.11.017 .

[13] P. Blanc, X. Bourbon, A. Lassin, E.C. Gaucher, Chemical model for cement-based ma-terials: temperature dependence of thermodynamic functions for nanocrystallineand crystalline C-S-H phases, Cem. Concr. Res. 40 (2010) 851 866, http://dx.doi.org/ 10.1016/j.cemconres.2009.12.004 .

[14] W. Kunther, B. Lothenbach, K.L. Scrivener, On the relevance of volume increase forthe length changes of mortar bars in sulfate solutions, Cem. Concr. Res. 46 (2013)23 29, http://dx.doi.org/ 10.1016/j.cemconres.2013.01.002 .

[15] N. Crammond, The thaumasiteform of sulfateattackin the UK, Cem. Concr. Compos.25 (2003) 809 818, http://dx.doi.org/ 10.1016/S0958-9465(03)00106-9 .

[16] P. Nobst, J. Stark, Investigations on the in uence of cement type on thaumasiteformation, Cem. Concr. Compos. 25 (2003) 899 906, http://dx.doi.org/ 10.1016/S0958-9465(03)00118-5 .

[17] S. Tsivilis, G. Kakali, A. Skaropoulou, J.H. Sharp, R.N. Swamy, Use of mineral admix-tures to prevent thaumasite formation in limestone cement mortar, Cem. Concr.Compos. 25 (2003) 969 976, http://dx.doi.org/ 10.1016/S0958-9465(03)00153-7 .

[18] E.F.Irassar, V.L.Bonavetti, M.a.Trezza, M.a.Gonzlez, Thaumasite formation in lime-stone ller cements exposed to sodium sulphate solution at 20 C, Cem. Concr.Compos. 27 (2005) 77 84.

[19] S. Khler, D. Heinz, L. Urbonas, Effect of ettringite on thaumasite formation, Cem.Concr. Res. 36 (2006) 697 706, http://dx.doi.org/ 10.1016/j.cemconres.2005.11.006 .

[20] A. Skaropoulou, G. Kakali, S. Tsivilis, A study on thaumasite form of sulfate attack(TSA) using XRD, TG and SEM, J. Therm. Anal. Calorim. 84 (2006) 135 139, http://dx.doi.org/ 10.1007/s10973-005-7198-2 .

[21] F. Bellmann, J. Stark, Prevention of thaumasite formation in concrete exposed tosulphate attack, Cem. Concr. Res. 37 (2007) 1215 1222, http://dx.doi.org/ 10.1016/

j.cemconres.2007.04.007 .

http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0165http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0165http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0170http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0170http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0170http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0170http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0175http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0175http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0175http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0175http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0180http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0180http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0185http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0185http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0185http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0185http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0190http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0190http://dx.doi.org/10.1016/j.cemconres.2005.03.001http://dx.doi.org/10.1016/j.cemconres.2005.03.001http://dx.doi.org/10.1016/j.cemconres.2007.06.002http://dx.doi.org/10.1016/j.cemconres.2007.11.003http://dx.doi.org/10.1016/j.cemconres.2007.11.003http://dx.doi.org/10.1016/j.cemconres.2007.11.003http://dx.doi.org/10.1016/j.cemconres.2012.09.002http://dx.doi.org/10.1111/j.1551-2916.2012.05335.xhttp://dx.doi.org/10.1016/j.cemconres.2010.11.017http://dx.doi.org/10.1016/j.cemconres.2009.12.004http://dx.doi.org/10.1016/j.cemconres.2009.12.004http://dx.doi.org/10.1016/j.cemconres.2013.01.002http://dx.doi.org/10.1016/S0958-9465(03)00106-9http://dx.doi.org/10.1016/S0958-9465(03)00118-5http://dx.doi.org/10.1016/S0958-9465(03)00118-5http://dx.doi.org/10.1016/S0958-9465(03)00118-5http://dx.doi.org/10.1016/S0958-9465(03)00153-7http://dx.doi.org/10.1016/S0958-9465(03)00153-7http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0060http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0060http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0060http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0060http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0060http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0060http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0060http://dx.doi.org/10.1016/j.cemconres.2005.11.006http://dx.doi.org/10.1016/j.cemconres.2005.11.006http://dx.doi.org/10.1007/s10973-005-7198-2http://dx.doi.org/10.1016/j.cemconres.2007.04.007http://dx.doi.org/10.1016/j.cemconres.2007.04.007http://dx.doi.org/10.1016/j.cemconres.2007.04.007http://dx.doi.org/10.1016/j.cemconres.2007.04.007http://dx.doi.org/10.1016/j.cemconres.2007.04.007http://dx.doi.org/10.1007/s10973-005-7198-2http://dx.doi.org/10.1016/j.cemconres.2005.11.006http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0060http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0060http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0060http://dx.doi.org/10.1016/S0958-9465(03)00153-7http://dx.doi.org/10.1016/S0958-9465(03)00118-5http://dx.doi.org/10.1016/S0958-9465(03)00118-5http://dx.doi.org/10.1016/S0958-9465(03)00106-9http://dx.doi.org/10.1016/j.cemconres.2013.01.002http://dx.doi.org/10.1016/j.cemconres.2009.12.004http://dx.doi.org/10.1016/j.cemconres.2010.11.017http://dx.doi.org/10.1111/j.1551-2916.2012.05335.xhttp://dx.doi.org/10.1016/j.cemconres.2012.09.002http://dx.doi.org/10.1016/j.cemconres.2007.11.003http://dx.doi.org/10.1016/j.cemconres.2007.11.003http://dx.doi.org/10.1016/j.cemconres.2007.06.002http://dx.doi.org/10.1016/j.cemconres.2005.03.001http://dx.doi.org/10.1016/j.cemconres.2005.03.001http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0190http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0190http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0185http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0185http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0180http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0180http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0175http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0175http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0170http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0170http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0165http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0165


14/14

[22] X.J. Gao, B.G. Ma, H.J. Ba, Inuence of cement types and mineral admixtures on theresistance of mortar to thaumasite form of sulfate attack, Int. Congr. Chem. Cem.,Montreal, Canada, 2007 .

[23] T. Schmidt, B. Lothenbach, K.L. Scrivener, M. Romer, D. Rentsch, R. Figi, Conditionsfor thaumasite formation, Int. Congr. Chem. Cem., Montreal, Canada, 2007 .

[24] A. Skaropoulou, K. Sotiriadis, G. Kakali, S. Tsivilis, A long term study on thaumasiteform of sulfate attack (TSA) in limestone cement pastes, Int. Congr. Chem. Cem.,Montreal, Canada, 2007 .

[25] S. Tsivilis, K. Sotiriadis, A. Skaropoulou, Thaumasite form of sulfate attack (TSA) inlimestone cement pastes, J. Eur. Ceram. Soc. 27 (2007) 1711 1714, http://dx.doi.org/ 10.1016/j.jeurceramsoc.2006.05.048 .

[26] Q. Zhou, E.W. Byars, C.J. Lynsdale, J.C. Cripps, J. Hill, J.H. Sharp, Relative resistance of Portlandand Pozzolanic cements to the thaumasite form of sulfate attack (TSA), Int.Congr. Chem. Cem., Montreal, Canada, 2007 .

[27] E.F. Irassar, Sulfate attack on cementitious materials containing limestone ller areview,Cem. Concr. Res.39 (2009)241 254, http://dx.doi.org/ 10.1016/j.cemconres.2008.11.007 .

[28] M.D.A. Thomas, R.D. Hooton, The Durability of Concrete Produced with Portland-Limestone Cement: Canadian Studies, 2010 .

[29] W. Kunther, Investigation of sulfate attack by experimental and thermodynamicmeans, Ecole Polytechnique Fdrale de Lausanne, 2012 .

[30] A.M. Ramezanianpour, R.D. Hooton, Sulfate resistance of Portland-limestone ce-ments in combination with supplementary cementitious materials, Mater. Struct.46 (2013) 1061 1073, http://dx.doi.org/ 10.1617/s11527-012-9953-8 .

[31] C. Shi, D. Wang, A. Behnood, Review of thaumasite sulfate attack on cement mortarand concrete, J. Mater. Civ. Eng. 24 (2012) 1450 1460, http://dx.doi.org/ 10.1061/(ASCE)MT.1943-5533.0000530 .

[32] A.M. Ramezanianpour, R.D. Hooton, Thaumasite sulfate attack in Portland andPortland-limestone cement mortars exposed to sulfate solution, Constr. Build.Mater. 40 (2013) 162 173, http://dx.doi.org/ 10.1016/j.conbuildmat.2012.09.104 .

[33] A. Skaropoulou, K. Sotiriadis, G. Kakali, S. Tsivilis, Use of mineral admixtures toimprove the resistance of limestone cement concrete against thaumasite form of sulfate attack, Cem. Concr. Compos. 37 (2013) 267 275, http://dx.doi.org/ 10.1016/j.cemconcomp.2013.01.007 .

[34] T. Schmidt, B. Lothenbach, M. Romer, J. Neuenschwander, K. Scrivener, Physical and

microstructural aspects of sulfate attack on ordinary and limestone blendedPortland cements, Cem. Concr. Res. 39 (2009) 1111 1121, http://dx.doi.org/ 10.1016/j.cemconres.2009.08.005 .

[35] R.S. Gollop, H.F.W. Taylor, Microstructural and microanalytical studies of sulfateattack. I. Ordinary Portland cement paste, Cem. Concr. Res. 22 (1992) 1027 1038 .

[36] C. Yu,W. Sun,K. Scrivener, Mechanism of expansion of mortars immersed in sodiumsulfate solutions, Cem. Concr. Res. 43 (2013) 105 111, http://dx.doi.org/ 10.1016/j.cemconres.2012.10.001 .

[37] G.W. Scherer, Stress from crystallization of salt, Cem. Concr. Res. 34 (2004)1613 1624, http://dx.doi.org/ 10.1016/j.cemconres.2003.12.034 .

[38] J. Bensted, Thaumasite direct, woodfordite and other possible formationroutes, Cem. Concr. Compos. 25 (2003) 873 877, http://dx.doi.org/ 10.1016/S0958-9465(03)00115-X .

http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0080http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0080http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0080http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0080http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0080http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0085http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0085http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0090http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0090http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0090http://dx.doi.org/10.1016/j.jeurceramsoc.2006.05.048http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0100http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0100http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0100http://dx.doi.org/10.1016/j.cemconres.2008.11.007http://dx.doi.org/10.1016/j.cemconres.2008.11.007http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0110http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0110http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0115http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0115http://dx.doi.org/10.1617/s11527-012-9953-8http://dx.doi.org/10.1061/(ASCE)MT.1943-5533.0000530http://dx.doi.org/10.1061/(ASCE)MT.1943-5533.0000530http://dx.doi.org/10.1016/j.conbuildmat.2012.09.104http://dx.doi.org/10.1016/j.cemconcomp.2013.01.007http://dx.doi.org/10.1016/j.cemconcomp.2013.01.007http://dx.doi.org/10.1016/j.cemconcomp.2013.01.007http://dx.doi.org/10.1016/j.cemconres.2009.08.005http://dx.doi.org/10.1016/j.cemconres.2009.08.005http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0145http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0145http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0145http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0145http://dx.doi.org/10.1016/j.cemconres.2012.10.001http://dx.doi.org/10.1016/j.cemconres.2012.10.001http://dx.doi.org/10.1016/j.cemconres.2003.12.034http://dx.doi.org/10.1016/S0958-9465(03)00115-Xhttp://dx.doi.org/10.1016/S0958-9465(03)00115-Xhttp://dx.doi.org/10.1016/S0958-9465(03)00115-Xhttp://dx.doi.org/10.1016/S0958-9465(03)00115-Xhttp://dx.doi.org/10.1016/j.cemconres.2003.12.034http://dx.doi.org/10.1016/j.cemconres.2012.10.001http://dx.doi.org/10.1016/j.cemconres.2012.10.001http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0145http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0145http://dx.doi.org/10.1016/j.cemconres.2009.08.005http://dx.doi.org/10.1016/j.cemconres.2009.08.005http://dx.doi.org/10.1016/j.cemconcomp.2013.01.007http://dx.doi.org/10.1016/j.cemconcomp.2013.01.007http://dx.doi.org/10.1016/j.conbuildmat.2012.09.104http://dx.doi.org/10.1061/(ASCE)MT.1943-5533.0000530http://dx.doi.org/10.1061/(ASCE)MT.1943-5533.0000530http://dx.doi.org/10.1617/s11527-012-9953-8http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0115http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0115http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0110http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0110http://dx.doi.org/10.1016/j.cemconres.2008.11.007http://dx.doi.org/10.1016/j.cemconres.2008.11.007http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0100http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0100http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0100http://dx.doi.org/10.1016/j.jeurceramsoc.2006.05.048http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0090http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0090http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0090http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0085http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0085http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0080http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0080http://refhub.elsevier.com/S0008-8846(14)00113-6/rf0080

Pr21. Borcelo Et Al. Astm c1012 Mod. 2014

Documents

Transcript of Pr21. Borcelo Et Al. Astm c1012 Mod. 2014