MgO content of slag controls phase evolution and structural changes induced by accelerated carbonation in alkali-activated binders

Abstract

The structural development and carbonation resistance of three silicate-activated slags (AAS) with varying MgO contents (< 7.5 wt.%) are reported. AAS with lower MgO content reacts faster at early age, forming gismondine and C-A-S-H type gels, while in slags with higher MgO content (> 5%), hydrotalcite is identified as the main secondary product in addition to C–A–S–H. Higher extent of reaction and reduced Al incorporation in the C–S–H product are observed with higher MgO content in the slag. These gel chemistry effects, and particularly the formation of hydrotalcite, seem to reduce the susceptibility to carbonation of AAS produced with higher MgO contents, as hydrotalcite appears to act as an internal CO₂ sorbent. This is evidenced by an inverse relationship between natural carbonation depth and slag MgO content, for paste samples formulated at constant water/binder ratio. Thus, the carbonation performance of AAS can be enhanced by controlling the chemistry of the precursors.

Introduction

The production of Portland cement is currently associated with the generation of 5–8% of global anthropogenic carbon emissions [1], and the increased demand for civil infrastructure in the developing world will lead to the need for production of higher amounts of cement in the near future. To reduce the environmental footprint of the construction industry, there is the imminent need to adopt a ‘tool-kit’ of environmentally friendly alternative binders, including alkali-activated cements, that has the potential to be more sustainable in production and during service life when used in fit-for-purpose applications [2].

Alkali-activated slag binders are produced through the chemical reaction between granulated blast furnace slag (GBFS) and an alkaline activator [3]. The microstructural development of these binders is strongly dependent on the chemistry and mineralogy of the slag [4], [5], including its particle size [6], [7] and degree of amorphicity, along with the nature and concentration of the alkaline activator [8], [9], [10], [11], and the curing conditions including relative humidity, temperature and duration [12]. Although hundreds of studies have been published in this area over the past decades, there are discrepancies in the literature regarding the kinetics of reaction, optimal activation conditions, type and characteristics of the reaction products formed, and performance development of alkali-activated slag binders from different sources. The variability in results is associated with the large number of physicochemical parameters which can affect the alkali-activation of slags, and is a major obstacle to gaining a true understanding of the mechanisms of activation [13], [14]. This then contributes to the lack of globally accepted standardised protocols for formulation of binders with desired mechanical and durability performance based on generic mix designs [14].

Within the quaternary oxide system CaO–MgO–SiO₂–Al₂O₃ that describes much of the chemistry of GBFS [15], when it is necessary to omit one component to enable the system to be plotted on a ternary phase diagram for discussion of slag chemistry in alkali activation (or in blending with Portland cements), the omitted component is almost always MgO. Relatively little attention has been given to the effect of MgO content of the slag in alkali activated binders, as the majority of European and North American slags which have been subjected to alkali activation have had a fairly consistent 7–10% MgO content. Ben Haha et al. [4] identified in three different slags with MgO contents between 7.7% and 13.2% that the effect of the MgO content in slag on the structural evolution and mechanical strength is strongly dependent on the nature of the activator. In alkali metasilicate-activated slags, higher contents of MgO favour larger extent of reaction and increased mechanical strengths at early times of curing. This is consistent with the observations of Douglas et al. [16] who assessed three silicate-activated slags with MgO contents between 9.62% and 18.60%, obtaining mechanical strength up to 65 MPa after 28 days of curing for the slag with the highest MgO content. Ben Haha et al. [4] attributed their results to the increased formation of layered double hydroxides of the hydrotalcite group (similar to Mg₆Al₂(CO₃)(OH)₁₆·4H₂O, although a carbonate-free composition was proposed by those authors) over the time of curing, and decreased Al uptake by the C–S–H type gel. Conversely, no marked differences were observed among slags when NaOH was used as activator [4]. In silicate-activated slag with increased contents of Al₂O₃ (~ 16%) and reduced contents of MgO (< 5%), formation of hydrotalcite is not identifiable by X-ray diffraction (XRD), and instead the formation of zeolites such as gismondine and garronite is favoured [17], [18]. This indicates that MgO has a much more significant influence on the binder microstructure in this compositional range than has been identified in activated slag with MgO contents higher than 7 wt.%.

The interaction between a cementitious material and atmospheric CO₂ results in carbonation of the binder, and its rate is influenced by the chemical interaction between the CO₂, the reaction products, pore solution and the mass transport of the CO₂ into the pore structure [19]. Carbonation leads to reduction of the alkalinity of a cement, accompanied by an increased susceptibility to corrosion of steel rebar [20]. Natural carbonation can be a slow process as the concentration of CO₂ in the air is low, and this has motivated the adoption of accelerated testing methods using high CO₂ concentrations to predict how carbonation will affect cementitious materials over the time of service. However, it has been identified in both Portland cements [21] and alkali-activated materials [22], [23], [24] that the CO₂ concentration strongly affects the type of reaction products forming upon accelerated carbonation, and therefore different accelerated carbonation conditions do not always replicate well what is identified in specimens subjected to natural carbonation. Recent results show that variables such as the nature of the alkaline activator [25], the paste volume in a concrete [26], and the testing conditions [27] seem to have a strong influence on the carbonation rate of alkali-activated materials, when tested under accelerated carbonated conditions.

In order to complement the existing results available in the literature about the potential effect of MgO content in the activation of slags, this study analyses the structural evolution of three slags with MgO contents of 1.2%, 5.2% and 7.4%, activated with sodium metasilicate. This is achieved through the use of isothermal calorimetry, X-ray diffraction, ²⁹Si and ²⁷Al magic angle spinning nuclear magnetic resonance (MAS NMR) spectroscopy, and environmental scanning electron microscopy (ESEM) coupled with energy dispersive X-ray (EDX) spectroscopy. Early age specimens (14 days) are exposed to different CO₂ concentrations (between natural conditions and 1% CO₂), and the carbonation products forming are analysed.