Durability of Ternary Cements Based on New Supplementary Cementitious Materials from Industrial Waste

: Cement-based materials decay with exposure to aggressive agents, a development that raises infrastructure operation and maintenance costs substantially. This paper analyses the inclusion of ultraﬁne construction and demolition (U C&DW ) and biomass-fuelled power plant (BA) waste as pozzolanic additions to cement in pursuit of more sustainable and eco-respectful binders and assesses the durability of the end materials when exposed to seawater, chlorides (0.5 M NaCl) or sulphates (0.3 M Na 2 SO 4 ). The effect of adding silica fume (SF) at a replacement ratio of 5% was also analysed. Durability was determined using the methodology proposed by Koch and Steinegger, whilst microstructural changes were monitored with mercury intrusion porosimetry (MIP), X-ray diffraction (XRD) and scanning electron microscopy (SEM) for a fuller understanding of decay processes. According to the ﬁndings, the new blended cements containing 20%U C&DW + 10%BA or 20%U C&DW + 20%BA + 5%SF resist the attack by the aggressive media studied, with a 56-d corrosion index of over 0.7. The composition of the reaction products generated with the attack is essentially the same in OPC and the SCM-bearing materials. The results show that the optimal replacement ratio for SCM is 30%.


Introduction
The physical, chemical, mechanical and biological actions affecting concrete structures throughout their service life compromises their durability and compliance with the ultimate limit state and service standards to which they were designed. The result is substantial yearly repair and maintenance costs, which in 2016 came to~$123 × 10 9 worldwide according to ASCE 2017 data on infrastructures [1]. Islam et al. [2], in turn, reported that maintenance costs amount to 14% to 20% of design costs in the UK, Hong Kong and Australia.
The presence of physical-chemical pathological processes in such structures is associated with parameters (choice of materials, type of environmental exposure, w/c ratio in concrete, structural shapes, etc.) defined in project stages ranging from design and construction to operation [3]. When choosing materials, the type of binder (cement), as the concrete component that interacts most readily with aggressive media (sulphates, chlorides, seawater, etc.) in the environment is of particular relevance to durability. Such interaction directly and indirectly induces the formation of expansive products that prompt cracking and/or surface spalling to the detriment of the bearing section of the concrete. Given that cement susceptibility to such action depends on the chemical composition, crystalline or gelatinous nature and porosity of cement hydration products, those factors play a decisive role in the reactions with aggressive agents. The presence of calcium silicate or aluminate hydrates and especially portlandite is therefore a determinant in cement durability [4].
Along those lines, blending supplementary cementitious materials (SCMs) into cement or concrete is known to improve concrete structure durability [5]. Pozzolanic SCMs in particular favour durability due to their effect on pore system refinement and reduction of pore solution alkalinity and portlandite content in cement matrices [4,6]. In addition to those technical advantages, the use of SCMs carries social and environmental benefits associated with lower natural resource consumption and the valorisation of industrial byproducts, as well as a significant decline in CO 2 emissions. CEM II/A-LL or L manufacture generates 16.2% and CEM III/B manufacture 73.0% less of that greenhouse gas than CEM I [7].
Against that backdrop, the scientific community has shown a growing interest in the last 10 years in identifying SCMs that may serve as alternatives to the silica fume, fly ash, blast furnace slag, natural or calcined pozzolans and limestone envisaged in European standard EN 197-1 [8]. One of the most prominent lines of research underway involves the assessment of the mechanical viability of new SCMs sourced from construction and demolition (C&DW) and agroforestry waste in the formulation of more sustainable binary eco-cements.
In 2016, the European Union alone generated 923.9 Mt of C&DW, accounting for around 35% of all the solid waste produced by its 28 members [9]. Those by-products can be separated, crushed and sieved to yield recycled aggregate (RA), in turn classified under three categories depending on concrete (Rc) and masonry (Rb) content [10]: concrete RA (Rc ≥ 90%; Rb ≤ 10%), mixed RA (Rc < 90%; Rb ≤ 30%) and mixed masonry RA (Rc < 70%; Rb > 30%). Whilst the coarse (>4 mm), and especially the concrete, fraction have been extensively researched [11,12] for use as the granular skeleton in recycled concrete, little scientific-technical knowledge has been collected on the fine (<4 mm) and ultrafine (<0.125 µm) fractions. The valorisation of those materials, comprising 30% to 40% of all C&DW [13], is a challenge to be confronted if circular economy principles are to be instituted in the construction industry.
The use of C&DW ultrafines as SCMs in new cements is an innovative line of work undertaken in recent years by a number of authors, whose interest has focused primarily on the mechanical performance of pastes and/or mortars bearing those materials as partial (3% to 30%) cement replacements. The consensus conclusion reached [14][15][16][17][18] is that GRC contains SiO 2 , Al 2 O 3 and Fe 2 O 3 (44% ≤ SiO 2 + Al 2 O 3 + Fe 2 O 3 ≤ 59%), along with CaO (42% to 16%) and exhibits low pozzolanicity and hydraulicity. Its filler effect likewise-identified would favour hydration, providing nucleation sites that would densify the cement matrix microstructure. Masonry and mixed C&DW ultrafines [19][20][21][22] comprise essentially SiO 2 , Al 2 O 3 and Fe 2 O 3 (SiO 2 + Al 2 O 3 + Fe 2 O 3 ≥ 84%), with pozzolanicity at least as high as fly ash and similar to that of other masonry materials, such as masonry [4,23] and fired clay sanitary ware [24] rubble. Further to research on mechanical behaviour, GRC has no adverse effect on new eco-cement performance at low replacement ratios (≤7%), whereas masonry ultrafines induce no significant decline in compressive or flexural strength at percentages of up to 25%.
Another line of research that has evoked considerable interest, as noted earlier, is the valorisation of agroforestry waste as an SCM. Two types of such waste are distinguished, depending on the origin [25]: laboratory-calcined (500 • C ≤ T ≤ 700 • C) biomass agroforestry waste (primarily bagasse, rice husk and to a lesser extent bamboo ash) [26][27][28]; and biomass or bottom ash, BA, from power and/or heat plants [29][30][31][32], which generate around 10 Mt of BA yearly [33]. SiO 2 , Al 2 O 3 and Fe 2 O 3 (SiO 2 + Al 2 O 3 + Fe 2 O 3 ≥ 70%) are the majority compounds in such waste, which exhibits high early and late age lime fixation capacity (=pozzolanicity). No common pattern has been identified in the effect of including such SCMs on end product mechanical performance, however, which has on the contrary been observed to depend on the origin and nature of the waste and the replacement ratio [30,[34][35][36].
In that regard, no prior study has been published in the international literature assessing the durability of ternary or quaternary eco-cements, simultaneously bearing mixed C&DW ultrafines (U C&DW ) and BA or both with silica fume (U C&DW + BA + SF).
The research described hereunder used the Koch and Steinegger method [37] to analyse the seawater-, chloride-and sulphate-resistance of cement pastes bearing 30% or 45% SCMs (U C&DW + BA or U C&DW + BA + SF). Secondarily, the processes governing cement decay were determined on the grounds of XRD-identified mineralogical variations and the morphological changes observed in the exposed materials with scanning electron microscopy. The variations in pore size distribution detected with mercury porosimetry were also explored.

Materials and Methods
For ready and more convenient reference, the materials and methods deployed in this study are described schematically below.

M1
Components: 70%OPC + 20%U C&DW + 10%BA, yielding a type II/B blended portland cement (21 % to 35 % total additions) or a type IV/A pozzolanic cement (11 % to 35 % total additions) in different proportions in keeping with European standard EBA, at replacement ratios of 10 % to 20 % according to the classification set out in standard EN 197-1 [8] Physical, mechanical characteristics: Table 2  Components: 55%OPC + 20%U C&DW + 20%BA + 5%SF, yielding a type IV/B pozzolanic cement (36% to 55% total additions) in different proportions in keeping with European standards, at replacement ratios of 10% to 20% according to the classification set out in standard EN 197-1 [8] Physical, mechanical characteristics:  [37,40]), deemed optimal for assessing blended cement resistance to such agents as it simultaneously accommodates the pozzolanic reaction and the assessment of its most prominent advantages [41] Soaking time; T: 21 d, 56 d or 90 d; 20 • C Procedure: specimens triple-washed in deionized water to remove any excess salt and dried in a laboratory kiln at 40 • C to a constant weight prior to characterisation at each exposure time Mechanical and microstructural analysis Parameters measured: flexural strength, chemical resistance and XRD-identification of new compounds at each age Determination of chemical resistance to aggressive media: Koch-Steineger corrosion index, calculated as (Equation (1)): where CI = corrosion index; F SX = flexural strength at exposure time 'i' to medium 'X'; (sw: seawater; ch: 0.3 M NaCl; s: 0.5 M Na 2 SO 4 ); F SW = flexural strength in water-soaked specimens, same exposure times Pore size distribution, SEM/EDX: in 56-d specimens  [42] Mean pore size calculation (Equation (2)): where V = median pore diameter (volume); A = median pore diameter (area) Test frame (mechanical strength) Instrument: IBERTEST AUTOTEST 200/10-SW test frame fitted with an adapter for 1 × 1 × 6 cm specimens.

Mechanical Properties
The decline in flexural strength values of the water-soaked cement and the materials exposed to seawater, sodium chloride (0.5 M NaCl) or sodium sulphate (0.3 M Na 2 SO 4 ), was an indication that the effect of the dilution resulting from the lower amount of clinker present in new blends M1 and M2 was greater than the effect of the pozzolanicity of this type of SCMs (U C&DW and BA). That finding was consistent with results reported by other authors who recorded declines in strength associated with smaller proportions of C-S-H gel and greater porosity in pastes prepared with SCMs processed from industrial by-products [36,43,44].
On the whole, strength rose with exposure time in the pastes exposed to seawater and 0.3 M Na 2 SO 4 . That pattern was associated with the formation of salts (see point 3.4) that would initially fill pores to generate a more compact matrix. Subsequent salt expansion, however, would raise inner stress and induce microcracking [41,45].
The rise in strength in the pastes exposed to 0.5 M NaCl may be attributed to: (i) the formation of more hydration product [46], favoured by the enhancement of Ca(OH) 2 solubility in NaCl solutions; (ii) greater C-S-H gel stability [47] with the uptake of some of the Na + into the C-S-H interlayer spaces [48]; or (iii) Friedel's salt precipitation in the pores [49].
A comparison of the behaviour of the blends exposed to the three types of aggressive media is provided in Table 3. OPC and M1 were observed to follow a similar pattern except when soaked in seawater, where the simultaneous inclusion of the new SCMs (BA and U C&DW ) lowered strength by~5 % relative to the water-soaked 90 d pastes. Of the pastes exposed to sodium chloride, performance declined the least in 90 d M1, to~1 % below its water-soaked counterpart, compared to the~9% decline observed for OPC. Strength rose in the same materials in the 90-d attack by Na 2 SO 4 , by 72% in OPC and 13% in M1. The M2 blends (OPC + U C&DW + BA + SF) exhibited lower performance than the other two pastes irrespective of the exposure medium. In the samples exposed to seawater, the Appl. Sci. 2021, 11, 5977 7 of 16 90 d values for M2 were~38% lower than for OPC~8% lower where the medium was 0.5 M NaCl and~2% lower when attacked by 0.3 M Na 2 SO 4 . Those findings infer that including SF with the Ba + U C&DW combination had no beneficial effect, possibly because the amount of portlandite present in the binder/SCM system was insufficient to support the pozzolanic reaction, which would be expected to generate a more compact and stronger matrix [50].

Porosity
Macro-and mesopore size distribution in the 56 d OPC, M1 and M2 pastes is graphed in Figure 2. The refinement in water-soaked paste M1 and M2 pore systems relative to OPC was attributable to the pozzolanic reaction between the SCMs and the portlandite generated during cement hydration. C-S-H gel, the predominant reaction product, would fill the pores, reducing the volume of macro-and raising the volume of mesopores [51]. The M2 blends (OPC + UC&DW + BA + SF) exhibited lower performance than the other two pastes irrespective of the exposure medium. In the samples exposed to seawater, the 90 d values for M2 were ∼38% lower than for OPC ∼8% lower where the medium was 0.5 M NaCl and ∼2% lower when attacked by 0.3 M Na2SO4. Those findings infer that including SF with the Ba + UC&DW combination had no beneficial effect, possibly because the amount of portlandite present in the binder/SCM system was insufficient to support the pozzolanic reaction, which would be expected to generate a more compact and stronger matrix [50].

Porosity
Macro-and mesopore size distribution in the 56 d OPC, M1 and M2 pastes is graphed in Figure 2. The refinement in water-soaked paste M1 and M2 pore systems relative to OPC was attributable to the pozzolanic reaction between the SCMs and the portlandite generated during cement hydration. C-S-H gel, the predominant reaction product, would fill the pores, reducing the volume of macro-and raising the volume of mesopores [51]. Pore systems in 56-d pastes exposed to distilled water (w), seawater (sw), 0.5 M NaCl (ch) or 0.3 M Na2SO4 (s).
The pore system in the OPC exposed to aggressive media was refined more intensely than the systems in pastes M1 or M2, enhancing its durability and inducing higher Kock & Steinegger indices (see Section 3.3). That finding is consistent with prior accelerated testing for durability in cement-based materials bearing fly ash and either silica/alumina nanoparticles [52] or masonry industry sludge [4].
The total porosity (Pt) values for the 56-d water-or aggressive medium-soaked pastes listed in Table 4 show that porosity was higher in SCM-bearing pastes M1 and M2 than in OPC due to the intrinsic effect of the blend of SCMs on that parameter. The decline in total porosity observed in the pastes exposed to the aggressive media relative to the watersoaked specimens was attributed to the precipitation of new products such as Friedel's salt, ettringite, gypsum and brucite resulting from the chemical attack on the pore system. Their presence induced a rise in mesopore volume [53][54][55] (Figure 2), lowering permeability and retarding decay [56]. The common denominator in all the pastes analysed was Figure 2. Pore systems in 56-d pastes exposed to distilled water (w), seawater (sw), 0.5 M NaCl (ch) or 0.3 M Na 2 SO 4 (s).
The pore system in the OPC exposed to aggressive media was refined more intensely than the systems in pastes M1 or M2, enhancing its durability and inducing higher Kock & Steinegger indices (see Section 3.3). That finding is consistent with prior accelerated testing for durability in cement-based materials bearing fly ash and either silica/alumina nanoparticles [52] or masonry industry sludge [4].
The total porosity (Pt) values for the 56-d water-or aggressive medium-soaked pastes listed in Table 4 show that porosity was higher in SCM-bearing pastes M1 and M2 than in OPC due to the intrinsic effect of the blend of SCMs on that parameter. The decline in total porosity observed in the pastes exposed to the aggressive media relative to the watersoaked specimens was attributed to the precipitation of new products such as Friedel's salt, ettringite, gypsum and brucite resulting from the chemical attack on the pore system. Their presence induced a rise in mesopore volume [53][54][55] (Figure 2), lowering permeability and retarding decay [56]. The common denominator in all the pastes analysed was that the decline in Pt was steepest in 0.3 M Na 2 SO 4 (followed by seawater and 0.5 M NaCl pastes in that order) and steeper than in the distilled water-soaked samples. That finding was closely associated with the nature of the products of the chemical attack [6]. Those patterns were consistent with prior reports of the behaviour of cement pastes bearing pozzolans such as blast furnace slag, fly ash, silica fume, C&DW or biomass waste after exposure to aggressive agents [20,25,56,57].

Corrosion Index
The OPC, M1 and M2 corrosion index/exposure time (t exp ) curves in Figure 3 show that in all the pastes, with the exception of seawater-soaked M2, an initial rise in the index (t exp ≤ 56 d) was followed by a decline after 90 d of exposure. Further to Koch-Steinegger criteria, pastes OPC, M1 and M2 could be deemed sulphate-resistant, for their 56-d corrosion index was greater than or equal to 0.70. that the decline in Pt was steepest in 0.3 M Na2SO4 (followed by seawater and 0.5 M NaCl pastes in that order) and steeper than in the distilled water-soaked samples. That finding was closely associated with the nature of the products of the chemical attack [6]. Those patterns were consistent with prior reports of the behaviour of cement pastes bearing pozzolans such as blast furnace slag, fly ash, silica fume, C&DW or biomass waste after exposure to aggressive agents [20,25,56,57].

Corrosion Index
The OPC, M1 and M2 corrosion index/exposure time (texp) curves in Figure 3 show that in all the pastes, with the exception of seawater-soaked M2, an initial rise in the index (texp ≤ 56 d) was followed by a decline after 90 d of exposure. Further to Koch-Steinegger criteria, pastes OPC, M1 and M2 could be deemed sulphate-resistant, for their 56-d corrosion index was greater than or equal to 0.70.  In this respect also, the inclusion of SF to the BA + U C&DW blend was observed to have no beneficial effect on paste durability where the M1 corrosion index was consistently lower than the OPC value, irrespective of test age and aggressive medium.

X-ray Diffration Findings
The XRD patterns for pastes OPC, M1 (70%OPC + 20%U C&DW + 10%BA) and M2 (55%OPC + 20%U C&DW + 20%BA + 5%SF) soaked in water or exposed to the aggressive media analysed are reproduced in Figures 4-7. The crystalline compounds identified in the water-soaked OPC specimens (Figure 3) included primarily calcium monocarboaluminate (CH 22 Al 2 Ca 4 O 20 ), ettringite (CaO) 6 (Al 2 O 3 )(SO 3 ) 3 ·32H 2 O), portlandite (Ca(OH) 2 ), calcite (CaCO 3 ) and anhydrous dicalcium silicate (C 2 S) in the form of larnite, a mineral present in portland cement detected by XRD after soaking in water because it hydrates more slowly than other mineralogical phases in the cement [58]. no beneficial effect on paste durability where the M1 corrosion index was consistently lower than the OPC value, irrespective of test age and aggressive medium.
The same phases as observed in OPC were identified in phases M1 and M2, along with quartz and the feldspars albite and orthoclase present in the SCMs used, as discussed in Section 2.1. The intensity of the portlandite reflections declined as a result not only of lower cement content, but also of the pozzolanic reaction induced by the UC&DW, BA and SF added. The calcium monocarboaluminate observed in the water-soaked specimens was absent in the pastes exposed to seawater ( Figure 5). The diffractograms for the latter also contained lines for other crystalline compounds resulting from the reaction between hydration products and the sulphate and chloride ions in seawater. The Cl − ions diffusing through paste pores were partially retained by the phases bearing Al2O3, giving rise to calcium monochloraluminate hydrate or Friedel's salt (3CaO⋅AI2O3⋅CaCl2⋅AI2O3⋅10H2O) [59,60], from which ettringite may have formed via reaction with SO4 2− ions [4]. Brucite (Mg(OH)2) also formed as the product of the reaction between the Mg 2+ in the seawater and the OHions present in portlandite [61]. The brucite diffraction lines were more intense in the OPC than in M1 or M2 (see Section 3.4). The lines for portlandite were significantly less intense inn the seawater-than in the water-soaked specimens, from which seawater may be inferred to have attacked that hydration product directly. In contrast, the ettringite and calcite lines were more intense due to the interaction between the phases in the paste and the sulphates and carbonates present in seawater, resulting in the formation of secondary ettringite [62] and calcite [61]. The pastes exposed to 0.5 M NaCl ( Figure 6) were observed to form Friedel's salt as a product of the reaction between phases C3A and C4AF, thereby limiting the secondary formation of ettringite, an expansive mineral with a consequently adverse effect on cement-based materials. The intensity of the respective diffraction lines was greater in pastes M1 and M2 than in OPC. The intensity of the lines for ettringite remained unchanged relative to the water-soaked pastes, a finding consistent with Ekolu et al. [63] reports to the effect that ettringite remains stable at NaCl concentrations of 0.5 M or lower. The pastes exposed to 0.5 M NaCl ( Figure 6) were observed to form Friedel's salt as a product of the reaction between phases C3A and C4AF, thereby limiting the secondary formation of ettringite, an expansive mineral with a consequently adverse effect on cement-based materials. The intensity of the respective diffraction lines was greater in pastes M1 and M2 than in OPC. The intensity of the lines for ettringite remained unchanged relative to the water-soaked pastes, a finding consistent with Ekolu et al. [63] reports to the effect that ettringite remains stable at NaCl concentrations of 0.5 M or lower. Specimen exposure to the action of Na2SO4 (Figure 7) translated primarily into a rise in the intensity of the diffraction line for ettringite. The gypsum (Ca2SO4⋅2H2O) formation identified with SEM/EDX (see Section 3.4), could not be corroborated via XRD because the respective diffraction line overlapped with the reflections for other phases present in the pastes, such as the monocarboaluminates (2θ = 11.7°) and quartz (2θ = 20.8°) [64].

SEM/EDX Microstructural Analysis
Micrographs of the C-S-H gels in water-soaked pastes are reproduced in Figure 8ac. The EDX analysis (ten analyses/gel) showed that the Ca/Si ratio in the OPC gel was 2.0, compared to 1.7 in M1 and 1.9 in M2. The findings also showed that adding SCMs induced aluminium uptake in the gel, as reported earlier by other authors who used pozzolanic additions to partially replace clinker [22,24,36,57]. Inasmuch as the aluminium may have replaced Si in the gel structure, those findings would explain the lower Ca/Si ratio in the pastes bearing SCMs. Compact portlandite plates were observed in all the pastes analysed, although less frequently and in smaller mean sizes in paste M1 and M2 than in OPC, due to the pozzolanicity of the former. Those findings were consistent with the XRD data. The same phases as observed in OPC were identified in phases M1 and M2, along with quartz and the feldspars albite and orthoclase present in the SCMs used, as discussed in Section 2.1. The intensity of the portlandite reflections declined as a result not only of lower cement content, but also of the pozzolanic reaction induced by the U C&DW , BA and SF added.
The calcium monocarboaluminate observed in the water-soaked specimens was absent in the pastes exposed to seawater ( Figure 5). The diffractograms for the latter also contained lines for other crystalline compounds resulting from the reaction between hydration products and the sulphate and chloride ions in seawater. The Cl − ions diffusing through paste pores were partially retained by the phases bearing Al 2 O 3 , giving rise to calcium monochloraluminate hydrate or Friedel's salt (3CaO·AI 2 O 3 ·CaCl 2 ·AI 2 O 3 ·10H 2 O) [59,60], from which ettringite may have formed via reaction with SO 4 2− ions [4]. Brucite (Mg(OH) 2 ) also formed as the product of the reaction between the Mg 2+ in the seawater and the OH -ions present in portlandite [61]. The brucite diffraction lines were more intense in the OPC than in M1 or M2 (see Section 3.4). The lines for portlandite were significantly less intense inn the seawater-than in the water-soaked specimens, from which seawater may be inferred to have attacked that hydration product directly. In contrast, the ettringite and calcite lines were more intense due to the interaction between the phases in the paste and the sulphates and carbonates present in seawater, resulting in the formation of secondary ettringite [62] and calcite [61]. The pastes exposed to 0.5 M NaCl ( Figure 6) were observed to form Friedel's salt as a product of the reaction between phases C 3 A and C 4 AF, thereby limiting the secondary formation of ettringite, an expansive mineral with a consequently adverse effect on cementbased materials. The intensity of the respective diffraction lines was greater in pastes M1 and M2 than in OPC. The intensity of the lines for ettringite remained unchanged relative to the water-soaked pastes, a finding consistent with Ekolu et al. [63] reports to the effect that ettringite remains stable at NaCl concentrations of 0.5 M or lower.
Specimen exposure to the action of Na 2 SO 4 ( Figure 7) translated primarily into a rise in the intensity of the diffraction line for ettringite. The gypsum (Ca 2 SO 4 ·2H 2 O) formation identified with SEM/EDX (see Section 3.4), could not be corroborated via XRD because the respective diffraction line overlapped with the reflections for other phases present in the pastes, such as the monocarboaluminates (2θ = 11.7 • ) and quartz (2θ = 20.8 • ) [64].

SEM/EDX Microstructural Analysis
Micrographs of the C-S-H gels in water-soaked pastes are reproduced in Figure 8a-c. The EDX analysis (ten analyses/gel) showed that the Ca/Si ratio in the OPC gel was 2.0, compared to 1.7 in M1 and 1.9 in M2. The findings also showed that adding SCMs induced aluminium uptake in the gel, as reported earlier by other authors who used pozzolanic additions to partially replace clinker [22,24,36,57]. Inasmuch as the aluminium may have replaced Si in the gel structure, those findings would explain the lower Ca/Si ratio in the pastes bearing SCMs. Compact portlandite plates were observed in all the pastes analysed, although less frequently and in smaller mean sizes in paste M1 and M2 than in OPC, due to the pozzolanicity of the former. Those findings were consistent with the XRD data.

SEM/EDX Microstructural Analysis
Micrographs of the C-S-H gels in water-soaked pastes are reproduced in Figure 8ac. The EDX analysis (ten analyses/gel) showed that the Ca/Si ratio in the OPC gel was 2.0, compared to 1.7 in M1 and 1.9 in M2. The findings also showed that adding SCMs induced aluminium uptake in the gel, as reported earlier by other authors who used pozzolanic additions to partially replace clinker [22,24,36,57]. Inasmuch as the aluminium may have replaced Si in the gel structure, those findings would explain the lower Ca/Si ratio in the pastes bearing SCMs. Compact portlandite plates were observed in all the pastes analysed, although less frequently and in smaller mean sizes in paste M1 and M2 than in OPC, due to the pozzolanicity of the former. Those findings were consistent with the XRD data. Magnesium in the form of brucite predominated in the surface layer observed to form in all the 56-d pastes exposed to seawater (Figure 9a-c), corroborating the XRD findings. Brucite, the result of the reaction between portlandite and the MgCl2 in seawater, was present on specimen surfaces only. The layer was thicker in the OPC (e = 107.1 µm) than Magnesium in the form of brucite predominated in the surface layer observed to form in all the 56-d pastes exposed to seawater (Figure 9a-c), corroborating the XRD findings. Brucite, the result of the reaction between portlandite and the MgCl 2 in seawater, was present on specimen surfaces only. The layer was thicker in the OPC (e = 107.1 µm) than in blend M1 (e = 82.4 µm) or M2 (e = 83.5 µm). Friedel's salt formation was also observed in seawater-soaked pastes OPC, M1 and M2, corroborating the XRD findings (Section 3.3).  Plates identified as Friedel's salt on the grounds of composition (Al, Cl and Ca) and morphology [60,65] were observed in the 56-d pastes exposed to 0.5 M NaCl ( Figure 10). The pastes exposed to the 0.3 M Na2SO4 solution were observed to form gypsum deposits clustering primarily in pores (Figure 11a), where the conditions for nucleation were most favourable [66,67]. Typically elongated needle-like ettringite (Figure 11b) was also identified [68,69], primarily inside pores [70], favoured by pressure conditions and the presence of the necessary ions [71,72]. For those very reasons, pores, together with cracks, are the sites where ettringite normally crystallises.  Plates identified as Friedel's salt on the grounds of composition (Al, Cl and Ca) and morphology [60,65] were observed in the 56-d pastes exposed to 0.5 M NaCl (Figure 10).  Plates identified as Friedel's salt on the grounds of composition (Al, Cl and Ca) and morphology [60,65] were observed in the 56-d pastes exposed to 0.5 M NaCl ( Figure 10). The pastes exposed to the 0.3 M Na2SO4 solution were observed to form gypsum deposits clustering primarily in pores (Figure 11a), where the conditions for nucleation were most favourable [66,67]. Typically elongated needle-like ettringite (Figure 11b) was also identified [68,69], primarily inside pores [70], favoured by pressure conditions and the presence of the necessary ions [71,72]. For those very reasons, pores, together with cracks, are the sites where ettringite normally crystallises.  The pastes exposed to the 0.3 M Na 2 SO 4 solution were observed to form gypsum deposits clustering primarily in pores (Figure 11a), where the conditions for nucleation were most favourable [66,67]. Typically elongated needle-like ettringite (Figure 11b) was also identified [68,69], primarily inside pores [70], favoured by pressure conditions and the presence of the necessary ions [71,72]. For those very reasons, pores, together with cracks, are the sites where ettringite normally crystallises.
Appl. Sci. 2021, 11, x FOR PEER REVIEW 12 of 16 in blend M1 (e = 82.4 µm) or M2 (e = 83.5 µm). Friedel's salt formation was also observed in seawater-soaked pastes OPC, M1 and M2, corroborating the XRD findings (Section 3.3). Plates identified as Friedel's salt on the grounds of composition (Al, Cl and Ca) and morphology [60,65] were observed in the 56-d pastes exposed to 0.5 M NaCl ( Figure 10). The pastes exposed to the 0.3 M Na2SO4 solution were observed to form gypsum deposits clustering primarily in pores (Figure 11a), where the conditions for nucleation were most favourable [66,67]. Typically elongated needle-like ettringite (Figure 11b) was also identified [68,69], primarily inside pores [70], favoured by pressure conditions and the presence of the necessary ions [71,72]. For those very reasons, pores, together with cracks, are the sites where ettringite normally crystallises.

Conclusions
The following conclusions can be drawn from this study.

•
As the pozzolanic reaction between SCMs and portlandite does not suffice to offset the dilution induced by a lower proportion of cement, the new pastes exhibit a larger mean pore size and greater porosity than OPC. • Including U C&DW + BA or U C&DW + BA + SF yields cements able to resist the chemical attacks analysed, with a 56-d corrosion index >0.7 in all cases. • Resistance to aggressive media attack is not enhanced by raising the percentage of SCM added, for the cements bearing the binary blend U C&DW + BA proved to be more resistant than those to which silica fume was added. • Paste M1 bearing 30% SCMs had a higher 56-d corrosion index, at 1.06, for the 0.5 M NaCl solution than both OPC (CI = 1.04) and M2 (CI = 1.02). • OPC paste resists seawater and 0.3 M Na 2 SO 4 better than M1, which in turn had a higher corrosion index than M2 irrespective of the medium.
The addition of U C&DW + BA or U C&DW + BA + SF prompts no significant change in cement paste morphology or in the compounds generated during the chemical attack.