Hydration of Reactive MgO as Partial Cement Replacement and Its Influence on the Macroperformance of Cementitious Mortars

Instituto Superior Técnico, Universidade de Lisboa, Av. Rovisco Pais, 1049-001 Lisboa, Portugal CERIS, Instituto Superior Técnico, Universidade de Lisboa, Av. Rovisco Pais, 1049-001 Lisboa, Portugal Department of Inorganic Chemistry and Chemical Engineering, Universidad de Córdoba, Campus de Rabanales, Edificio Marie Curie, Planta Baja y 1 Planta, 14071 Córdoba, Spain Department of Inorganic Chemistry and Chemical Engineering, Engineering School Science of Belmez, 14240 Córdoba, Spain


Introduction
Cement industry is one of the most polluting industries in the world, contributing around 5% to 7% of the total global CO 2 emission [1], thereby making it of fundamental importance to seek new solutions that can further decrease its environmental impact.Some of the solutions involve improving the calcination process energy efficiency, using alternative fuels or raw materials, and exploring the role of cement systems in carbon capture and storage technologies [2].Among the aforementioned, the one considered the most effective from an environmental impact reduction point of view is direct replacement of cement (i.e., supplementary cementitious materials) with constituents that partake in the hydration and strength development processes, especially if sourced from industrial wastes (e.g., fly ash, ground granulated blastfurnace slag, palm oil fuel ash, and electric-arc furnace dust) [3][4][5][6].is makes the use of reactive magnesium oxide (MgO) a unique opportunity to reduce the resulting composite's carbon footprint, as well as to improve some of its features [7].
e use of MgO-based constituents in the clinkerization process has been restricted to no more than 5% of MgO [8], as the high temperatures involved in this process would yield the production of dead-burned MgO with a very slow hydration rate [9].Consequently, the late expansion of MgO in a hardened and stable cementitious microstructure would jeopardize its soundness.However, this is not the case for MgO subjected to lower calcination temperatures (i.e., light burned MgO calcined at 700-1000 °C).Not only does it typically present a great reactivity and early age expansion [10][11][12], but its overall CO 2 emissions were reported to be up to 73% lower than those of cement [13], considering the higher carbon capture capacity of MgO-based cements throughout their life cycle [7,14,15].Hydration reaction of MgO, which leads to the formation of Mg(OH) 2 , is as follows [16]: Since the volume of hydration products is higher than that of the reagents, the hydration of MgO leads to an expansion, which, if properly controlled, may be used to compensate the shrinkage typically observed in cementitious composites [10,[17][18][19][20][21] and possibly their deformation due to creep as well.e compatible chemical nature of MgO with that of cement may produce beneficial expansive components capable of reducing the number of microcracks and contribute to strength gain under certain circumstances [22].Similar to that observed in the hydration of cement, after the formation of Mg(OH) 2 , ensuing pozzolanic reactions can lead to the formation of magnesium silicate hydrates (i.e., MgO-SiO 2 -H 2 O or M-S-H) [23] capable of presenting considerable strength gain [24], if in the presence of amorphous silica-bearing mineral additions, such as fly ash [25,26] or silica fume [27,28].Another practical feature of using reactive MgO-based cementitious composites is their ability to capture a considerable amount of CO 2 thereby further decreasing the material's environmental impact throughout its life cycle.Furthermore, the carbonation of brucite may result in phases (i.e., magnesium carbonate hydrates such as nesquehonite, hydromagnesite, dypingite, and artinite) capable of significantly increasing the strength of the matrix [29,30].
In the light of the recent positive findings regarding this component and its practical impact in the construction industry, this paper reports the interaction of the hydration products of commercially available MgO, from two different sources, with those of cement, from a cementitious microstructure viewpoint and its implications on the composite's macroperformance. is study is the first of upcoming statefunded research projects, involving various companies within the construction sector.Contrasting with the mix design of other research studies, which typically involve lower replacement levels of cement with MgO, relatively high contents of 10%, 15%, and 20%, by weight, of MgO samples were used as cement replacement thus constituting one of the novelties of this paper.ese relatively higher contents were chosen with the purpose of further reducing the environmental impacts of concrete, given the considerable carbon capture of MgO-containing cementitious systems, and also its costs since the manufacture of cement is an energy-intensive process unlike that of light burned MgO.ermogravimetric analysis (TGA), energy-dispersive X-ray (EDX) analysis, differential thermal analysis (DTA), and powder XRD analysis were carried out on cement, the two MgO samples, and selected mortars.e specimens' mechanical-and durability-related performances were evaluated by means of their flexural and compressive strength, carbonation, water absorption by capillary action, and shrinkage.

Cement and MgO.
e cement used was CEM I 42.5R, and its chemical composition is presented in Table 1.e cement presents initial and final setting times of 161 min and 232 min, respectively.e water used in the production of all specimens was potable tap water.
Two types of MgO were used: MgO A was supplied by Styromag, an Austrian company, and MgO G from Grecian Magnesite S.A., a Greek company.Both are light burned MgO due to their relatively low calcination temperature of about 800 °C [11].Both MgO samples' chemical compositions are presented in Table 1.e morphology of both MgO samples can be observed by means of the scanning electron microscopy (SEM) micrographs presented in Figure 1.
e particle-size distributions of cement and both MgO samples are presented in Figure 2(a).MgO A and cement present similar distributions, as most of their particles have a diameter between 10 μm and 60 μm.MgO G, which is similar to MgO A, contains slightly smaller particles (mostly between 10 μm and 30 μm) and presents a less extensive particle-size distribution when compared to the other two.About 52.1% and 65.3% of the particles of MgO A and MgO G, respectively, were found in the range of 3 µm-32 µm.Above 32 µm (20.2% of MgO A and 3.5% of MgO G), the particles may be too large to hydrate rapidly, and below 3 µm (27.7% of MgO A and 31.1% of MgO G), they may have a smaller contribution to the mechanical strength and simultaneously demand more water [42][43][44].e bulk densities of MgO A, MgO B, and cement were 653, 618, and 1040 kg/m 3 , respectively.
e BET specific surface areas of MgO A and MgO G were 20.12 m 2 /g and 18.01 m 2 /g, respectively.e pore size distributions, obtained from the adsorption-desorption isotherms of N 2 , are shown Figure 2(b).Both additions show a range of pore diameter of 4-200 nm, according to the BJH method.Although the distribution of the pore volume is comparable in both materials, for MgO A, the highest quantity of the pore volume is centred at a slightly lower pore diameter than MgO G (23 nm vs. 33 nm, respectively).
e results of the XRD analysis are presented in Figure 3. e pattern of MgO A shows that the main phase was periclase (MgO) (International Centre for Diffraction Data-ICDD #45-0946), suggesting its crystallinity due to the narrow peak.For MgO A, peaks corresponding to the presence of calcite (CaCO 3 ) (ICDD #05-0586), dolomite (CaMg(CO 3 ) 2 ) (ICDD #36-0426), and talc (ICDD #13-0558) are also visible.e presence of the first two could also be 2 Advances in Materials Science and Engineering observed in the TGA in Figure 4(a).Unlike the XRD pattern of MgO A, the one of MgO G shows just a periclase phase.e absence of other smaller phases indicates a lower degree of impurity when compared to MgO A.
Figure 4 presents the results of TGA and DTA of MgO A and MgO G. e initial weight loss up to 150 °C, slightly higher in MgO G when compared to that in MgO A, is possibly due to the OH group residual in the surface of the material or to the presence of the hydromagnesite compound.At temperatures between 300 °C and 400 °C, a weight loss was observed in MgO G (4.67%), attributable to the dehydroxylation of Mg(OH) 2 .is suggests greater partial hydration of MgO G in comparison with MgO A. e increase in weight loss and decrease in heat flow up to around 750 °C may have been due to the decarbonation of calcite (CaCO 3 ) and dolomite (CaMg(CO 3 ) 2 ) as suggested by the XRD patterns of MgO A.

Characterization of Aggregates.
e sand's particle-size distribution was made in accordance with EN 1015-1 [45].
e fine particles content classification was determined in accordance to EN 13139 [48].e physical properties and particle-size distributions of both fine and coarse sand (0/2 and 0/4 mm, respectively) are presented in Table 2.
An XRD analysis was also carried out on the sand used in this study (Figure 5).Both fine and coarse sand showed that the main phase was quartz (crystalline SiO 2 ) (ICDD #33-1161) and were thus expected to have marginal reactivity with the cement's products of hydration and Mg(OH) 2 .

Mix Design.
e mortar's composition was based on the method proposed by Nepomuceno et al. [49], and the contents are presented in Table 3. e method adopted for the mix design allows the production of mortars exhibiting characteristics and performance equivalent to those of a concrete counterpart and not mortars typically used in rendering and masonry.All mortars were produced with a volumetric ratio of 1/3 (binder/aggregate).e mixes' spread was fixed at 200 ± 15 mm, which was achieved by varying the water/binder (w/b) ratio, starting from 0.50 for the control mixes (CMs).e cement was partially replaced by each type of MgO with the ratios of 10%, 15%, and 20%, by weight.Mortars incorporating MgO A and MgO G, henceforth, are MA and MG, respectively.

Test Methods.
e unbound constituents and the resulting hardened specimens were analyzed by means of XRD patterns using a Bruker D8 Discover A25 instrument with Cu-Kα radiation.All diffraction patterns were obtained by scanning using a goniometer from 10 °to 80 °(2θ) at a rate of 0.05 °•s −1 .TGA was performed using a Setaram SETSYS     Advances in Materials Science and Engineering diffraction particle size analyzer (Malvern Instruments) with ethanol as dispersant.

Physicochemical Characterization of Hardened Mortars.
Figure 4 presents the TGA and DTA of control mortars and those incorporating 20% MgO A as partial cement replacement with a curing age of 28 days.e initial mass loss for temperatures up to 200 °C can be attributed to the dehydration of C-S-H and AFt phases.
e thermal decomposition of M-S-H usually occurs in the same range of temperatures [24] and thus cannot be clearly observed, but it is likely that this phase did not form due to lack in the presence of amorphous SiO 2 capable of reacting with Mg(OH) 2 [35].Two drops are readily noticeable for MA20 between 300 °C and 500 °C.e first one, between 300 °C and 400 °C, can be attributed to the dehydroxylation of Mg(OH) 2 .
e second drop, observed in both specimens, between 400 °C and 500 °C, is due to the dehydroxylation of Ca(OH) 2 , as a result of the cement's hydration.e final endothermic peak at around 700 °C may be attributed to the decarbonation of CaCO 3 for control mixes and both CaCO 3 and MgCO 3 for MA20 specimens.
Figure 6 presents the results of the XRD analysis of the CM and MA20 specimens.It is clear from both patterns that the main mineral phase was quartz (ICDD #33-1161), being a result of the incorporation of sand.

Flexural and Compressive Strength. Figures 7(a) and 7(b)
present the flexural and compressive strength, respectively, of all mixes at 3, 7, 14, 28, and 56 days.A general decrease in strength can be observed with increasing MgO content, which was also observed by other researchers assessing lower replacement levels [57].e magnitude of this decrease in strength is more noticeable for MG specimens.is decline is a result of two factors, i.e., the need to increase the w/b ratio and the cement's dilution with increasing MgO content.In the former, the additional water led to a more porous microstructure for specimens containing 15% and 20% MgO.In the latter, the addition of MgO as cement replacement, in spite of the formation of Mg(OH) 2 , leads to a decreasing quantity of available C 2 S and C 3 S and consequently of C-S-H phases [35,58,59].Nevertheless, if in the presence of an addition with amorphous SiO 2 , a greater rate of strength development could be observed as a result of pozzolanic reactions with Mg(OH) 2 [25].Although the formation of Mg(OH) 2 is vital for strength gain at early curing ages, the ensuing formation of M-S-H will be the main contributing factor of the improvement of cementitious materials incorporating MgO [23].Moreover, the existence of alumina from aluminosilicate mineral additions may also offer an advantage concerning the material's mechanical behaviour.It may interact with Mg(OH) 2 and form a hydrotalcite-like phase [60] that is capable of enhancing the overall mechanical performance through improved bond with the cementitious matrix [23] and due to the hydrate's ability of decreasing the porosity of the resulting composite [61].
Concerning the strength development of MA and MG, the former presented improved performance when compared to the latter.One of the reasons for this may be due to the fact that MgO A presented a more extensive particle-size distribution similar to that of cement (Figure 2), which may have improved the packing of the resulting mortar specimens.Furthermore, as a result of the smaller particle size and greater surface area of MgO G, more water was required (Table 3) to obtain mixes with comparable workability levels.Naturally, this led to a more porous and less resistant microstructure.Still, in cases wherein an additional amount of water was not required (i.e., MA10 and MG10), specimens containing MgO A exhibited lower loss in performance.e TGA on both MgO samples showed that MgO G presented a greater mass loss probably due to the dihydroxylation of Mg(OH) 2 (mass losses of 1.76% and 4.67% within the 300-400 °C range for MgO A and MgO G, respectively), which suggests that it was already partially hydrated and did not contribute to the mortars' strength gain as effectively as MgO A. 4 presents the carbonation depth of the MgO-containing specimens after being exposed for 28 and 91 days in a CO 2 chamber.

Carbonation. Table
e results show greater carbonation depths with increasing MgO content, and there is no significant difference between MA and MG specimens.Even though less porous microstructures incorporating MgO have been observed elsewhere [25,62] suggesting reduced carbonation, the opposite trend was observed in the present study, and in others [62,63], they were substantiated by the results in Figure 8. e higher porosity in mixes containing increasing MgO content may have led to a greater surface area within the cementitious microstructure available to carbonate in the presence of CO 2 .Nevertheless, Advances in Materials Science and Engineering this higher porosity may exist only at an initial stage as ensuing carbonation yields the formation of further products of hydration reducing the volume of the porous microstructure [62].
Another factor must also be taken into account, which is the suitability of the phenolphthalein test to ascertain the amount of CO 2 uptake in the specific case of mortars containing MgO.Since cement was partially replaced with MgO, there was a lower quantity of the former available to produce Ca(OH) 2 , which exhibits a higher pH level than that of its counterpart from the hydration of MgO (i.e., Mg(OH) 2 ) (∼12.5 and ∼10.5, respectively) thereby  Advances in Materials Science and Engineering reducing the specimens' overall pH level.Nevertheless, the slightly greater mass loss of MA20 specimens in comparison with that of CM specimens between 500 °C and 850 °C in Figure 4 (3.71% vs. 3.34%, respectively) suggests increased CO 2 uptake and thus increased rate of carbonation.

Water
Absorption by Capillary Action.Figure 8 and Table 4 present the results of the water absorption by capillary action of MA and MG.As expected, there was a slight increase in water absorption with increasing MgO content, though not as obvious for MG10 and MG15.Even though the hydration of MgO may result in the formation of a phase that exhibits higher volume than that of the initial constituents, replacing a part of the cement will directly decrease the amount of C-S-H that is capable of producing a more tortuous and less interconnected porous network [37], especially after carbonation of the specimen [62,64].However, lower total pore volume and water absorption would likely be observed over time with ensuing carbonation reactions and formation of additional magnesium carbonate hydrates [30,37,65].
3.5.Shrinkage.Figure 9 presents the shrinkage strains of MA and MG.Up to a replacement level of 15%, both MA and MG exhibited lower shrinkage strain when compared to the control.is trend was also observed for MA-20, but not for MG-20, which presented significantly higher shrinkage strain.ese findings are in agreement with those of others [19,66] and can be explained by the aforementioned reason that Mg(OH) 2 presents a greater volume than that of its initial reagents.e specimens' expansion, prompted by the MgO particles' rapid reaction (typical of light-burned MgO [67]), occurred 2-3 days after casting.A similar behaviour was observed by Polat et al. [68] for mixes containing up to 7.5% MgO.After that, a distinctive behaviour of drying shrinkage could be observed, nonetheless significantly offset by the initial expansion; MA-15 and MA-20 exhibited about 80% lower shrinkage strain after 90 days.Others observed  8 Advances in Materials Science and Engineering similar decreases in shrinkage for MgO contents of 7.5% [68,69].However, for MgO G, the lower shrinkage strain could be observed only for replacement levels of 10% and 15%, with values of about 40% and 20%, respectively, lower than those of the control.is can be explained by the partial hydration of MgO G before its inclusion and thus did not contribute as effectively in the specimens' expansion as in MgO A-bearing mortars.Furthermore, for higher replacement levels of MgO G, the water requirement also increased, which increased the specimens' open porosity and decreased their stiffness.is facilitated the evaporation of free water and decreased the material's ability to restrain shrinkage, which outweighed the expansion of MgO G.

Conclusions
An experimental investigation on the role of MgO, as partial cement replacement, in the hydration reactions was performed.e following conclusions were drawn: (i) Commercially available reactive MgO exhibits a particle-size distribution that is comparable to that of cement, which avoids further size-reduction processing stages after the manufacture of light burned MgO before its inclusion in cementitious composites.(ii) Although the formation of Mg(OH) 2 was apparent in MgO-bearing mortars, TGA results did not present visible M-S-H-like patterns or those of hydrotalcite-like compounds, suggesting marginal interaction with the cement's products of hydration and siliceous aggregates.Nevertheless, additional analysis should be carried out on specimens without the incorporation of sand to further understand the hydration reactions.
(iii) e dilution of cement with progressively higher replacement levels of MgO, and thus lower quantity of C-S-H, probably led to the formation of a less resistant microstructure (lower flexural and compressive strength).Additionally, greater water requirement was observed in mixes including MgO since the mortars in this study were produced with equivalent consistence and without the use of waterreducing admixtures.e resulting increased porosity, apart from causing a decline in mechanical performance, also led to increased carbonation and water absorption by capillary action.(iv) Mortars incorporating MgO are likely to present lower shrinkage strain in comparison with corresponding control specimens made solely with cement.Nevertheless, this enhanced performance (∼80% less shrinkage strain for mixes containing 15% and 20% MgO A), prompted by the expansion of Mg(OH) 2 , heavily depends on the quantity of hydrated compounds and reactivity of MgO.From a structural point of view, the MgO-containing elements are likely to exhibit reduced shrinkageinduced cracking thereby improving the material's overall structural and durability-related performances.

Data Availability
e data used to support the results of this study are included in the manuscript.

Conflicts of Interest
e authors declare that they have no conflicts of interest.Advances in Materials Science and Engineering

Figure 3 :
Figure 3: XRD patterns for MgO A (a) and MgO G (b).

Figure 4 :
Figure 4: TGA (a) and DTA (b) of the two MgO samples and of the control mortar (CM) and mortar with 20% MgO A (MA20).

Figure 8 :
Figure 8: Water absorption by capillary action of mortars with (a) MgO A (MA) and (b) MgO G (MG).

Table 1 :
Chemical composition of cement and MgO (% by mass).

Table 2 :
Characteristics of fine and coarse sand.
CM: control mortar; MA: mortars with MgO A; MG: mortars with MgO G.

Table 4 :
Carbonation depth and sorptivity of specimens with increasing MgO content.