1. Introduction
The chemical environment of cement maintains a high pH pore solution (>12.5) which, in turn, keeps the protective oxide film on steel reinforcement in the passive state and protects it from corrosion. Once the passive layer is broken down, aggressive ions in the concrete including air, water, salt, etc., are free to attack the reinforcement which can, over time, lead to structural instability and failure unless preventative measures are taken. Atmospheric carbon dioxide (CO
2) readily dissolves in cement and the carbonic acid this produces partially neutralises the alkalis, reducing the overall pH, with carbonate and bicarbonate ions remaining in solution [
1].
In terms of solid hydrates, the dissolution of CO
2 during hydration causes the destabilisation of portlandite and ettringite while calcite, gypsum, and aluminum hydroxide precipitates [
1]. These changes in the solid hydrates are also accompanied by changes in the pore solutions which affect the sorptivity of reducing alkalis by the C-S-H as calcium is diminished in the system due to ongoing carbonation [
2,
3,
4,
5,
6]. The changes in the hydrated cement due to carbonation as reported by [
7] includes the preference of monosulfate and Friedel salt to monocarbonate which, in turn, forms stratlingite, high Ca;Si C-S-H gels to low Ca:Si C-S-H gels, calcite and water as ettringite dissolves and gibbsite precipitates, leaving sulphate ions in solution.
Sulfate ingress into concrete and mortar can lead to expansion, softening, cracking, and spalling which allow further aggressive agents to enter and cause severe durability issues [
8,
9,
10,
11,
12]. Sulfates such as Na
2SO
4 and MgSO
4 can be found in groundwater and wastewater that, unless properly accounted for in the design, will lead to costly shutdowns while the concrete is being repaired or replaced. The influence of external sulfates like Na
2SO
4 lead to additional or ‘secondary’ internal sulfate phases like ettringite and gypsum precipitation along with thaumasite formation that leads to the creation of cracks. Previous work in this area has led to mixed opinions on the influence of cement solid phases on the true cause of crack formation and expansion in cementitious materials subject to sulfate attack [
13,
14].
The ingress of seawater into cement leads to the conversion or replacement of monocarbonate with Friedel’s salt, along with the removal of typical solid hydrate phases, such as C-S-H, portlandite, ettringite, etc., with a corresponding decrease in solid volume [
15,
16,
17,
18,
19]. These phases are replaced with Friedel salt, Magnesium Silicate Hydrate (M-S-H), and the re-precipitation of brucite. These changes come with an initial increase in the total volume of ettringite primarily due to the presence of sulfate, carbonate, and magnesium in the seawater.
Thermodynamic modelling of cement hydration offers a reliable means of simulating changes in an anhydrous clinker during its hydration and providing predictions of phase assemblages and pore solution chemistries over time for many cement systems. Accurate predictions are possible once suitable input parameters (clinker oxide proportions, water/cement (w/c) ratio, curing temperature, relative humidity, and the Blaine fineness), an appropriately formatted thermodynamic database, and suitable geochemical software are used with a degree of expertise and a level of understanding to include other practical factors such as oxide components dissolved in the OPC clinker phases, oversaturation of specific phases during the first 12 h of hydration, and the release and uptake of alkali elements (K and Na) by the C-S-H. Most hydration simulations use these factors over time using empirical rate equations to describe the dissolution of OPC clinker phases. The authors have successfully used [
20,
21] the freely available geochemical software PHREEQC [
22] to predict the hydration behaviour of Portland cement over time. PHREEQC employs the law of mass action equations to perform complex geochemical simulations allowing the inclusion of kinetics and rates, and competing reactions between solids, liquids, and gases at equilibrium, details of which can be found in [
22]. However, at the time of writing, no models of CO
2-induced carbonation or changes in phase assemblage due to seawater, NaCl solution, or sulfate ingress into cement have been developed using the PHREEQC geochemical software.
Previous work on thermodynamic modelling of carbonation has been undertaken by De Weerdt and co-authors [
1] who used the Gibbs free energy minimization software GEMS 3.5 [
23] with a general thermodynamic database [
24] further expanded with the cemdata18 database [
25] containing solubility products of solids relevant for cementitious systems. They modelled mortars with two different types of cement (CEM I and CEM II/B-V according to EN 197-1 [
26]) exposed to accelerated carbonation conditions. Thermodynamic predictions of the changing phase assemblages and pore solution chemistries were predicted with increasing levels of CO
2. Thermodynamic modelling of cement-based mortars subject to increasing sulfate levels by [
27] demonstrated how the solid hydrates in partially hydrated systems were converted to higher volumes of ettringite, gypsum precipitates, and thaumasite formation as the amount of Na
2SO
4 solution increased.
In the current paper, the phase assemblages have been predicted for a CEM I commercial cement undergoing separate carbonation, sulfate, NaCl, and seawater-induced reactions using the geochemical code PHREEQC. The cemdata18 thermodynamic database was used with several Discrete Solid Phases (DSPs) used to represent the solid solution CSHQ and M-S-H models. Due to the amorphous and poorly crystalline nature of C-S-H in cement with a corresponding range of Ca/Si ratios (0.6–1.7), it displays a strongly incongruent dissolution behaviour [
28] where the release of calcium into solution is several orders of magnitude greater than that of silicon. A suitable C-S-H gel solubility model is therefore required to accurately describe its variable composition and solubility behaviour and predict stable phase assemblages, pH, and pore chemistries. While solid solution analysis is possible in PHREEQC, it has a long computational time. The use of DSPs to model stable phase assemblages etc. has been shown by the authors [
21,
29] to be a suitable replacement with no loss in accuracy with less computational effort.
5. Discussion
Carbonation in concrete is caused by the diffusion of CO
2 which reacts with portlandite and the calcium-bearing structural units in CSH to lower the pH by the creation of Ca(HCO
3)
2 in the pores. Below a pH of approximately 9, the microscopic passive layer surrounding the embedded steel is destabilised and is exposed to the oxygen and water in the pore structure leading to its rusting and expansion that causes the concrete to spall [
59]. In the cement under analysis here, a pH of 9 exists when over 90 g of CO
2 exists in the system (
Figure 14), and all of the portlandite, hydrotalcite, monocarbonate, ettringite, and siliceous hydrogarnet is depleted with minor amounts of C-S-H, re-precipitated gypsum and M-S-H, Ferrihydrite, Quartz, and gibbsite remaining. The ingress of CO
2 has also led to a significant increase in calcite. This provides a useful tool for concrete suppliers and those who maintain structures as a guide to when carbonation may be occurring so appropriate repair works be planned and carried out. Both the individual calcium-bearing hydrates and the calcium structural units of the CSH gel are susceptible to carbonation in the presence of carbon dioxide and water (or water vapour). The order in which these reactions occur can be predicted by thermodynamic modelling, or as demonstrated here, the maximum extent of carbonation may be predicted for a specific quantity of carbon reacting with the cement. Thermodynamics does not, of course, predict chemical kinetics, although kinetic constraints may be imposed on a reaction scheme that is known to be rapid. This has been applied in other works to model (for example) cement hydration.
Carbonation of cement and concrete is often a self-limiting process, as the molar volume of the reaction products is greater than that of the reactants. The consequence of carbonation is to close the surface porosity and to limit further gas transfer to the interior of structural concrete, restricting further carbonation to a vanishingly slow rate. This metastable state may persist for many decades (often longer) in well-made structural concrete. Carbonation becomes a serious degradation mechanism when a transport route is established, which allows atmospheric gasses (CO2 and O2) to react with the cement and the steel reinforcement in structural concrete. The carbonation results in a local fall of pore solution pH to the point where corrosion may be initiated. The expansive corrosion products push the concrete apart, increasing the rate of gas transport and hence corrosion. The critical factor in carbonation-induced corrosion is the ease with which carbon dioxide can reach steel reinforcement. This may be along a crack, joint, or poorly compacted region of locally high porosity, or through a section where the depth of cover over the reinforcement was too low to provide adequate protection from carbonation.
Temperature influences the behaviour of mortars undergoing sulfate attack both in terms of the change in solid hydrate and the time at which expansion occurs. Delayed ettringite formation (DEF) occurs when concrete is cured at higher temperatures (>70 °C) during precast concrete production or steam curing, and re-equilibrates at lower temperatures. Lothenbach et al. [
37] discussed how DEF can lead to greater ettringite expansion and gypsum re-formation at later stages as sulfates continue to react with hydrates [
59]. In addition, monosulfate, now the dominant AFm phase replacing monocarbonate at the higher temperature, converts to ettringite adding to its increased volume. This is shown in
Figure 15 where the cement here is modelled at a curing temperature of 80 °C. As may be seen, ettringite is not stable above approximately 35 °C where it dissolves and monosulfate re-precipitates [
60]. The reaction is reversed on cooling and DEF may cause problems many years later.
The effect of seawater on the phase assemblage of a hydrating cement used a breakdown of the chemical elements from a sample of Atlantic Ocean water [
36]. The Atlantic has the highest levels of aggressive chemicals in its waters on average than all the Earth’s oceans and seas.
Table 10 shows the composition of seawater taken from the East coast of Scotland which lies along the North Sea. As may be seen, compared to the Atlantic (
Table 9), the levels of aggressive ions are much lower. For instance, in terms of chlorides, which are responsible for most mild steel reinforcement corrosion [
59], the Atlantic has almost twice the chloride content (15,704 ppm). Using the concentrations in
Table 10,
Figure 16 shows the change in phase assemblage with increasing North Sea water. There is little difference in behaviour between both except for slightly less Friedel salt and expansive ettringite in the reaction of the North Sea water. There is also additional M-S-H and brucite as well as less C-S-H and portlandite in the Atlantic Ocean analysis, which may be due to the slightly more acidic conditions at higher water contents, as shown in
Figure 17. As with the above environments, this analysis will provide concrete designers and suppliers with valuable insights as to the likelihood of possible durability issues in saline environments and ensure that monitoring in extreme environments [
61] can maintain performance-based standards (EN 206, [
62]).
6. Conclusions
Gaining an understanding of how cementitious-based materials behave in harsh environments is vital for continued performance in service. This work has shown how thermodynamic modelling of cements undergoing carbonation, sulfate attack, and exposure to seawater agrees with previous experimental and modelling work by pioneers in this area, which has been brought up to date using the most recent version of the cemdata database and the PHREEQC geochemical model. By using several Discrete Solid Phase (DSP) models to account for the solid solution nature of the C-S-H gel model used here (CSHQ), along with other thermodynamically derived phases, the modelling predictions show the depletion of the solid hydrates that contribute mostly to the volume (C-S-H, portlandite, and monocarbonate) and the initial increase then sudden decrease in ettringite as the concentration of aggressive agent increases over time. The ingress of CO2, Na2SO4, and seawater has been shown to decrease the volume of the main solid hydrates and the growth of minor phases is also presented. Furthermore, the influence of temperature on sulfate attack has also been presented that conforms to the later precipitation of ettringite along with the conversion of monocarbonate to monosulfate. It should be noted, however, that a simplified model of increasing the concentration of aggressive agents has been employed. A better understanding of the evolution of engineering components could be gained from coupling the chemistry simulated here, with reactive transport modelling. PHREEQC offers this facility, allowing users to simulate the spatial and temporal evolution of materials chemistry where geometrical and transport properties are known, and this will be the subject of a future study supported with experimental data.