The relevance of laboratory studies on delayed ettringite formation to DEF in field concretes

Many laboratory studies of delayed ettringite formation (DEF) have been conducted on thin mortar bar specimens, heat treated, and then immersed in water. Under these conditions, rapid diffusion of alkali hydroxide into the surrounding water occurs and necessarily reduces the alkali hydroxide concentration of the mortar pore solution. Results reported recently by Famy indicate that the DEF process is triggered as a consequence of such leaching. When it is prevented by immersion into alkali hydroxide solution instead of water DEF expansion is delayed or prevented entirely. Results reported by Zhang indicate that 51-mm mortar cubes behave differently than more leaching-susceptible mortar bars when exposed to the same wet environment. Mortars that show severe DEF as mortar bars remain almost free of DEF symptoms if they are stored as cubes, even after 900 days. Attention is called to the fact that DEF in concrete is found commonly in thick concrete members where the possibility of leaching is remote. For such concrete, the reduction in internal alkali hydroxide concentration that occurs with ASR can substitute for the effect of leaching. It is postulated that without effective reduction of alkali hydroxide concentration by one or the other process, DEF remains latent.     

Studies on delayed ettringite formation in heat-cured mortars: II. Characteristics of cement that may be susceptible to DEF

Expansions of mortar bars, stored over (but not in) water after simulated steam curing to 85 °C, were related to certain cement compositional parameters. The relationship is expressed in the form of a “delayed ettringite formation (DEF) index.” The DEF index is computed as the joint product of the SO₃/Al₂O₃ molar ratio of the cement, the sum of its SO₃ and Bogue C₃A percentages divided by 10 and the square root of the alkali content expressed as equivalent % Na₂O. The mortars studied were made with 18 different cements, prepared from a set of six representative clinkers by incorporating Terra Alba gypsum to total SO₃ contents that were 1% below optimum, at optimum and 1% above optimum (as defined in ASTM C 563). Measurements of expansion were recorded at intervals for up to 1400 days. Severe cracking and prominent DEF-induced expansions were observed in mortar bars derived from four of the six ‘oversulfated’ cements and lesser expansions from three of the six cements prepared at optimum SO₃ contents. No expansion was found for cements of DEF index below a threshold value; above this value expansions were approximately proportional to the difference between DEF index and its threshold value. The relationship confirms the significance of all three compositional parameters making up the index, e.g., the SO₃/Al₂O₃ molar ratio, the joint contents of SO₃ and C₃A, and the alkali content, in influencing the extent of DEF-induced expansion. In these measurements, the apparent pessimum effect for SO₃ content previously reported by others was not found, although SO₃ contents examined spanned the supposed pessimum value of 4%. Rather, expansion increased with increasing SO₃ content for mortars made with all clinkers exhibiting expansion.     

Some factors affecting delayed ettringite formation in heat-cured mortars

Although more than 10 years of studies on delayed ettringite formation (DEF) have led to consensus in numerous areas of past disagreements, some questions remain experimental work is needed to complete the knowledge of this pathology. Following this objective, this paper studies the influence of pre-existing microcracking, wetting/drying cycles and the type of sulfated addition on DEF in steam cured mortars. The mortar specimens were prepared using an Ordinary Portland Cement and two types of sulfate were added to the mixtures: calcium sulfate (CaSO₄) or sodium sulfate (Na₂SO₄). The results confirm the well-known effect of temperature: no expansion was observed in any of the mixtures cured at room temperature. Moreover, no expansion was observed after 800 days for the reference mortar or for the mortar containing calcium sulfate but all the specimens of heat-cured mortars containing sodium sulfate expanded markedly after about 50 days whatever the supplementary treatments applied (thermal shrinkage or wetting/drying cycles). These results show the significant role played by alkalis in the occurrence of delayed ettringite. The supplementary treatments intended to cause prelimiray microcracking of the specimens did not promote expansion but contributed to a slight acceleration of the reaction. The ultimate values of expansion were similar to those obtained with sound mortars.     

Studies on delayed ettringite formation in early-age, heat-cured mortars: I. Expansion measurements, changes in dynamic modulus of elasticity, and weight gains

Mortars were prepared from laboratory cements blended from a set of six representative ground clinkers and Terra Alba gypsum. The addition of gypsum was such that cements containing 1% SO₃ less than the optimum SO₃ content, the optimum SO₃ content, and 1% greater than the optimum SO₃ content were produced. Mortar bars and mortar cubes containing each of these cements were exposed to continuous room temperature (23 °C) curing, or to early-age curing cycles involving maximum temperatures of 55 and 85 °C, followed by long-term exposure at 100% RH over water, but not immersed in water. Measurements of expansion, dynamic elastic modulus, and weight gain were recorded at intervals of up to 900 days. Severe cracking and prominent delayed ettringite formation (DEF)-induced expansions were observed in 85 °C cured mortar bars derived from four of the six “oversulfated” cements. Much smaller expansions were observed in mortar bars from two cements with optimum SO₃ content cements also cured at 85 °C. No expansion or other visible indication of distress was observed for any of the 55 °C or continuously room-temperature-cured mortars. The dynamic elastic modulus increased progressively on prolonged exposure for the unaffected mortar bars, but it decreased precipitously after the onset of expansion in affected mortar bars. Significant weight increases also accompanied the processes of expansion. Mortars that showed severe cracking and deterioration when exposed as mortar bars suffered almost no visible damage when exposed as cubes.     

The formation and role of ettringite in Iowa highway concrete deterioration

Some Iowa highway concrete constructed of coarse carbonate aggregate exhibits premature deterioration, which is, in part, caused by the growth of secondary minerals, including ettringite. Petrographic scanning electron microscope (SEM) and energy-dispersive analytical X-ray (EDAX) studies were conducted to determine the abundance, spatial location, and morphology of ettringite and the spatial relationship of ettringite to the occurrence of oxidized pyrite and coarse/fine carbonate aggregate.In poorly performing concrete (<16-year service life), ettringite completely fills many small voids, occurs as rims lining the margin of larger air entrainment voids and as microscopic disseminations in the paste. Pyrite (FeS₂) is commonly present in coarse aggregate, and goethite [FeO(OH)], one of its oxidation products, is observed in many concrete samples. Sulfate ions derived from pyrite oxidation apparently contribute to ettringite formation. The direct precipitation of ettringite from solution was responsible for most of the observed ettringite in voids and cracks. Microscopic ettringite, which commonly occurred in the paste, most likely was formed by the replacement of calcium aluminate. Severe cracking of cement paste is often spatially associated with ettringite, which strongly suggests that ettringite contributed to cracking and resultant deterioration.     

The influence of potassium-sodium ratio in cement on concrete expansion due to alkali-aggregate reaction

In concrete containing potentially reactive aggregates, deleterious alkali-aggregate-reaction (AAR) can be prevented by the use of suitable mineral admixtures or by limiting cement content and alkalis (Na₂O-equivalent) of the cement. However, the Na₂O-equivalent of cement may not always accurately define the potential of cement to cause AAR. In this study, the potential reactivity of concrete produced with cements having similar Na₂O-equivalents but different K/Na-ratios has been measured and the composition of gel has been analyzed. Additionally, pastes and mortars have been produced to study the development of pore solution composition.The expansion of the concrete mixtures shows significant differences depending on the cement used. The different K/Na-ratio present in the cements is reflected in the pore solution of pastes and mortars and in the gel present in aggregates of the concrete mixtures. As the hydroxide concentration in the pore solutions of pastes and mortars produced with the different cements is nearly identical, the difference in K/Na-ratio has to be the reason for the observed differences in concrete expansion.     

Delayed ettringite formation in Swedish concrete railroad ties

A petrographic examination of cracked Swedish concrete railroad ties identified delayed ettringite formation (DEF) as the damaging mechanism. This was unexpected because the concrete railroad ties were steam-cured with a maximum concrete temperature below 60 °C.The consensus in the published literature is that DEF only occurs in concrete subjected to heat curing above 70 °C. However, DEF is not only influenced by the curing temperature, but also by various other factors, such as cement composition (alkalis, C₃S, C₃A, SO₃, and MgO), fineness, etc. If an unfavorable combination of these parameters exists, delayed ettringite may occur at lower temperatures than 70 °C.In this paper, the influence of various parameters on DEF is discussed with reference to the investigated concrete.     

Diagnosing delayed ettringite formation in concrete structures

There has been a number of cases involving deteriorated concrete structures in North America where there has been considerable controversy surrounding the respective contributions of alkali–silica reaction (ASR) and delayed ettringite formation (DEF) to the observed damage. The problem arises because the macroscopic symptoms of distress are not unequivocal and microscopical examinations of field samples often reveal evidence of both processes making it difficult to separate the individual contributions. This paper presents the results of an investigation of a number of concrete columns carrying a raised expressway in North America; prior studies had implicated both DEF and ASR as possible causes of deterioration. Although the columns were not deliberately heat-cured, it is estimated that the peak internal temperature would have exceeded 70 °C and perhaps even 80 °C, in some cases. The forensic investigation included scanning electron microscopy with energy-dispersive X-ray analysis and expansion testing of cores extracted from the structure. Small-diameter cores stored in limewater expanded significantly (0.3 to 1.3%) and on the basis of supplementary tests on laboratory-produced concrete specimens it was concluded that expansion under such conditions is caused by DEF as the conditions of the test will not sustain ASR. In at least one column, DEF was diagnosed as the sole contributory cause of damage with no evidence of any contribution from ASR or any other deterioration process. In other cases, both ASR and DEF were observed to have contributed to the apparent damage. Of the columns examined, only concrete containing fly ash appeared to be undamaged. The results of this study confirm that, under certain conditions, the process of DEF (acting in isolation of other processes) can result in significant deterioration of cast-in-place reinforced concrete structures.     

The effects of chemical environment on the nucleation, growth, and stability of ettringite [Ca3Al(OH)6]2(SO4)3·26H2O

Ettringite is responsible for both the initial set of Portland cement and for premature concrete deterioration. A new method of ettringite crystal growth by combining calcium hydroxide and aluminum sulfate solutions was devised to reliably produce crystals that could be seen with a light microscope (45×-320×). The nucleation, growth, morphology, and stability of ettringite in the presence of over 300 chemicals and admixtures, many of which are present in the concrete environment, was then investigated. The plasticizers sorbitol, citrate, and tartrate were found to inhibit ettringite nucleation and growth, as did certain lignosulfonate air-entraining admixtures. The Type B set retarder borax inhibited ettringite formation at <44 ppm. The consequences and implications of this are discussed.     

Dual effectiveness of lithium salt in controlling both delayed ettringite formation and ASR in concretes

The influence of lithium nitrate on expansions due to delayed ettringite formation (DEF) and alkali-silica reaction (ASR) has been investigated. Effects of the lithium salt were examined in heat-cured mortars and concretes containing one or both damage mechanisms. The mortars and concretes made using reactive and/or non-reactive aggregates were subjected to heat treatment consisting of a hydration delay period of 4 h at 23 °C followed by steam-curing at 95 °C and then stored in limewater. Results showed that the lithium salt admixture was able to reduce the occurrence of deleterious expansion due to delayed ettringite formation in addition to controlling alkali-silica reaction in cementitious systems containing one or both mechanisms. In concretes made using non-reactive limestone aggregates, incorporation of lithium nitrate in a proportion of 0.74 M ratio of Li to (Na + K) was found to control delayed ettringite formation during the one-year period of this study.By analyzing the leaching properties of lithium and other alkalis from mortars during storage, it was found that a substantial amount of lithium was retained in the cementitious system in a slightly soluble form, and is expected to be responsible for reducing DEF.     

Alkali mass balance during the accelerated concrete prism test for alkali-aggregate reactivity

The alkali mass balance was calculated in concrete specimens submitted to the storage conditions of the Canadian standard CSA A23.2-14A concrete prism test for expansion due to alkali-aggregate reaction (AAR). The alkali concentration of both the concrete pore solution expressed under high pressure and the water below specimens in storage pails (bottom water) was measured. Measurements were conducted over a 1-year period, which corresponds to the length of the above test. Two reactive aggregates were tested [Potsdam sandstone (PO) and Spratt limestone (SP)]. Each aggregate was incorporated in two concrete mixtures (mass concrete and structural concrete), for a total of four batches. Significant alkali leaching occurred at 38 °C while performing tests in high moisture storage conditions even though prisms were covered with plastic sleeves. After 52 weeks, the alkali loss ranged from 12% to 25% of the original Na₂O_e content of the concrete, depending on the mixture proportioning and the aggregate type. After estimation of the proportion of alkalis fixed in cement hydrates, it appears that about 23% to 39% of the original alkalis released by the cement are quickly sorbed on aggregate surfaces or have rapidly migrated inside aggregate particles, which may have been incorporated with time in the AAR product. After 52 weeks at 38 °C, the pore solution alkalinity expressed from mass concrete made with PO was 250 mmol/l, whereas the alkalinity was 270 mmol/l in mass concrete incorporating SP. Since prisms of both mixtures were still expanding at 1 year, these alkalinity values are above the thresholds required for sustaining AAR in these concrete mixtures.     

The role of alkali content of Portland cement on the expansion of concrete prisms containing reactive aggregates and supplementary cementing materials

This paper presents results covering the effects of alkali content of Portland cement (PC) on expansion of concrete containing reactive aggregates and supplementary cementing materials (SCM). The results showed that the alkali content of PC has a significant effect on expansion of concrete prisms with no SCM. When SCM is used, the expansion was found to be related to both the chemical composition of the SCM and, to a lesser extent, the alkali content of the PC. The concrete expansions were explained, at least partly, on the basis of the alkalinity of a pore solution extracted from hardened cement paste samples containing the same cementing blends. An empirical relation was developed correlating the chemical composition (Ca, Si and total Na₂O_e) of the cementing blend (PC + SCM) and the alkalinity of the pore solution. Results from accelerated mortar bar test (ASTM C 1260) and a modified version thereof are also presented.     

Laboratory and field investigations of the influence of sodium chloride on alkali-silica reactivity

Concrete cylinders, 255 mm in diameter, were made with high- and low-alkali cements, a highly alkali-silica-reactive coarse aggregate, and subjected to various conditions at 38 °C: (1) immersion in 3% NaCl solution; (2) immersion in 6% NaCl solution; (3) humid air at 100% RH, and (4) 14-day cycles including 12 days in humid air, 2 days of drying, and 3 h in 6% NaCl solution. After 1 year, a number of cylinders were drilled to obtain dry powder samples from different depths, which were analyzed for total and soluble chloride and for soluble sodium and potassium. Concrete cores were also taken in a number of parapets and abutments, either exposed to deicing salts or not, on which chemical analyses were also performed on slices taken at different depths from the exposed surface. The results obtained suggest that making concrete with a low-alkali content is an effective way to prevent expansion due to alkali-silica reaction even for concretes exposed to seawater or deicing salts; this is attributed to the fact that the OH⁻ ion concentration in the pore solution, and then the pH, is decreased in the near-surface layers of concrete exposed to sodium chloride, which does not penetrate at depth in concrete.     

Thermal stability of ettringite in alkaline solutions at 80 °C

The thermal stability of synthetic ettringite was examined in NaOH solutions up to 1 M after 12 h of heat treatment at 80 °C, with or without the coexistence of C₃S in the system. Ettringite was found to convert to the U phase, a sodium-substituted AFm phase, over the heat treatment in the absence of C₃S. The presence of C₃S, leading to C-S-H formation, prevents the U phase formation and results in the conversion of ettringite to monosulfate. Sulfate ions generated from ettringite decomposition mostly remain in the solution, but some is incorporated into C-S-H. During subsequent storage at room temperature, the majority of monosulfate slowly converts back to secondary ettringite under moist conditions, using the supply of sulfate ions from the solution and C-S-H. The observations support the current mechanism of delayed ettringite formation (DEF).     

Implications of pre-formed microcracking in relation to the theories of DEF mechanism

Microcracks were induced in cementitious systems by freeze-thaw action and by alkali-silica reaction. These mechanisms often co-exist with delayed ettringite formation in concretes. Mortars and concretes were subjected to a heat treatment cycle consisting of a pre-set period of 4 h at 23 °C followed by accelerated curing at 95 °C. To isolate the mechanical effects of induced microcracking, heat-cured specimens were subjected to varied prescribed damage induced by freeze-thaw or alkali-silica reaction prior to the onset of delayed ettringite formation. It was found that inducement of pre-formed microcracks led to an earlier onset of expansion due to delayed ettringite formation. Initially, microcracks enhanced ultimate expansion until a certain relatively high extent of microcracking was reached. Thereafter, ultimate expansion decreased with any further increase in microcracking. This report gives support to the paste expansion theory.     

Effects of aggregate size and alkali content on ASR expansion

Attempts to model ASR expansion are usually limited by the difficulty of taking into account the heterogeneous nature and size range of reactive aggregates. This work is a part of an overall project aimed at developing models to predict the potential expansion of concrete containing alkali-reactive aggregates. The paper gives measurements in order to provide experimental data concerning the effect of particle size of an alkali-reactive siliceous limestone on mortar expansion. Results show that no expansion was measured on the mortars using small particles (under 80 µm) while the coarse particles (0.63-1.25 mm) gave the largest expansions (0.33%). When two sizes of aggregate were used, ASR-expansions decreased with the proportion of small particles. Models are proposed to study correlations between the measured expansions and parameters such as the size of aggregates and the alkali and reactive silica contents. The pessimum effect of reactive aggregate size is assessed and the consequences on accelerated laboratory tests are discussed.     

Impact of unrestrained Delayed Ettringite Formation-induced expansion on concrete mechanical properties

The consequences of Delayed Ettringite Formation (DEF) on the mechanical properties of concrete still remain imperfectly known. It is generally recognised that this pathology can decrease the strength of the material and its Young modulus.The comparative study between the expansion of concrete and the evolution of its dynamic modulus carried out in this study on a great number of samples, demonstrates that two types of swelling behaviours can be observed: a linear and a sigmoidal.A relationship between modulus and expansion rate is highlighted but it only remains in the case of linear swelling, which does not generate any significant damage.In the case of sigmoidal swelling, the damage process starts for expansions greater than 0.1% that can consequently reduce by 60% the dynamic modulus and by 65% the compressive strength. The relationship between modulus and expansion rate established for linear swelling cases is temporary verified for sigmoid swelling ones at the inflection point. At this point, the supersaturation is assumed to be mostly consumed and cannot further damage the matrix. Thereafter, the phase of stabilization of the expansion begins, and can be concomitant with a rehealing of the matrix induced by the continuation of cement hydration.     

Rapid evaluation of alkali-silica reactivity of aggregates using a nonlinear resonance spectroscopy technique

A new nonlinear acoustic technique — Nonlinear Impact Resonance Acoustic Spectroscopy (NIRAS) — is developed and used to characterize the alkali-reactivity of different aggregates. Cementitious materials such as mortar and concrete exhibit a hysteretic and nonlinear elastic behavior in their constitutive relations. This hysteretic nonlinearity is associated with interfacial debonding between the different constituents, and it changes with the progress of damage such as that induced by the alkali-silica reaction (ASR). One of the consequences of the hysteretic nonlinear property of these materials is the decrease in resonance frequencies, with increased excitation amplitude. This shift in the resonance frequency as a function of the material nonlinearity parameter can be used to directly characterize the damage state of the material. This research tracks the variation of the nonlinearity parameter during a standard accelerated mortar bar test (AMBT) to assess the potential for alkali-reactivity of aggregates. The results show that the NIRAS technique is more sensitive than conventional linear acoustic methods and is capable of accurately characterizing the reactivity of the aggregates examined. Furthermore, the results show advantages over standard expansion measurements for differentiating various aggregates having similar levels of reactivity, particularly at early test ages. These changes in the nonlinearity parameter are benchmarked against results from a petrographic analysis. Thus, the proposed NIRAS is a promising technique for the rapid identification of alkali-reactive aggregates.     

ASR expansion behavior of recycled glass fine aggregates in concrete

The alkali-silica-reaction (ASR) expanding behavior of different types of glass, all derived from cullet with different chemical composition, has been investigated. The glass reactivity was determined in different alkaline solutions based on sodium and/or calcium hydroxide to simulate concrete environment. The expansion of mortar containing different amounts of the investigated glass as fine aggregate has been carried out in different conditions: data collected underline a different response of glass towards the alkaline environment. Soda-lime glass shows negligible expansion, lead-silicate glass always generates expanding trends while boro-silicate glass has different behaviors depending on its colour. An attempt to link the behavior to the solubility and chemical reactivity of the glass is proposed.     

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

The effect of the type of alkalis on the expansion behavior of concrete microbars containing typical aggregate with alkali-silica reactivity and alkali-carbonate reactivity was studied. The results verified that: (1) at the same molar concentration, sodium has the strongest contribution to expansion due to both ASR and ACR, followed by potassium and lithium; (2) sufficient LiOH can completely suppress expansion due to ASR whereas it can induce expansion due to ACR. It is possible to use the duplex effect of LiOH on ASR and ACR to clarify the ACR contribution when ASR and ACR may coexist. It has been shown that a small amount of dolomite in the fine-grained siliceous Spratt limestone, which has always been used as a reference aggregate for high alkali-silica reactivity, might dedolomitize in alkaline environment and contribute to the expansion. That is to say, Spratt limestone may exhibit both alkali-silica and alkali-carbonate reactivity, although alkali-silica reactivity is predominant. Microstructural study suggested that the mechanism in which lithium controls ASR expansion is mainly due to the favorable formation of lithium-containing less-expansive product around aggregate particles and the protection of the reactive aggregate from further attack by alkalis by the lithium-containing product layer.