Rheology of Alkali-Activated Mortars: Inﬂuence of Particle Size and Nature of Aggregates

: The e ﬀ ect of two precursors (slag and ﬂy ash), di ﬀ erent particle size distribution, and three types of aggregate (siliceous sand, limestone, and recycled concrete) on alkali-activated material (AAM) mortar rheology were studied and compared to their e ﬀ ect on an ordinary Portland Cement (OPC) mortar reference. Stress growth and ﬂow curve tests were conducted to determine plastic viscosity and static and dynamic yield stress of the AAM and OPC mortars. In both OPC and AAM mortars, a reduction of the aggregate size induces a rise of the liquid demand to preserve the plastic consistency of the mortar. In general terms, an increase of the particle size of the siliceous aggregates leads to a decrease of the measured rheological parameters. The AAM mortars require higher liquid / solid ratios than OPC mortars to attain plastic consistency. AAM mortars proved to be more sensitive than OPC mortars to changes in aggregate nature. The partial replacement of the siliceous aggregates with up to 20% of recycled concrete aggregates induced no change in mixing liquid uptake, in either AAM or OPC mortars. All the AAM and OPC mortars studied ﬁtted to the Bingham model.


Introduction
Alkali-activated materials (AAMs) are considered eco-efficient and sustainable alternatives to Portland cement (OPC). They are based on the activation of aluminosilicate precursors by highly alkaline solutions. The precursors may be sourced from either industrial waste (such as vitreous blast furnace slag (BFS), coal fly ash (FA)), or used cracking catalysts or natural materials, such as pozzolans or burnt clays [1][2][3][4]. AAMs have been shown to owe their mechanical strength and durability to the dense microstructures generated during activation [5][6][7][8]. Their performances in those respects vary, however, depending on the aluminosilicate and alkaline activator used and the activation process used to prepare mortars and concretes.
The rheological behavior of such alkali-activated materials differs from that of OPC systems. The key factors determining the rheology of AAM pastes are the precursor nature and the concentration and nature of the activator [9][10][11][12]. Although the factors that govern AAM paste rheology have been established, the wide variety of precursors and activation systems renders the accurate prediction of system rheology during on-site placement elusive. Prior optimization studies must be run for nearly every precursor-activator combination. Sodium silicate alkali-activated slag (AAS) system workability and rheology is particularly complex, given the setting and short hardening times involved.
Puertas et al. [9] observed, after only 7 min of reaction, a sharp increase of the shear stress in waterglass-activated slag pastes. However, a breakdown of the structure and decrease of the shear stress occurred when a constant shear rate was applied. The intensity of the signal was enhanced and its time of appearance shortened with the increase of the Na 2 O concentration and SiO 2 /Na 2 O ratio of the activating solution. These authors justified this phenomenon to the formation of the early reaction products. This effect has also been observed in mortars [13] and concretes [14], and in both cases, the use of longer mixing times led to a better workability. This described phenomenon has not been observed in alkali activated slag pastes with NaOH, KOH, or alkaline carbonates (or even mixtures of all of them). Thus, the determining role of sodium silicate in the rheology of these activated slag systems is clear.
AAM mortars and concrete rheology research is even scarcer where, in addition to the precursor and alkaline activator, consideration must be given to the presence of sand and aggregate morphology and inter-particle friction, as well as the volume fraction occupied by solids in such systems [13][14][15][16][17][18]. Alonso et al. [15], working with alkali-activated slag, fly ash, and Portland cement mortars, concluded that AAM mortar rheology was more sensitive to changes in the liquid/solid (l/s) and sand/binder ratios than OPC. As noted in the literature, particle size distribution and aggregate nature are determinants in OPC paste, mortar, and concrete rheology [19,20].
It is well known that the particle size of aggregates influences the rheology of mortars and concretes, due to the effect of the packing factor and on the different specific surface of the aggregates [19,21,22]. A smaller size of aggregates leads to a higher demands of mixing liquid in the preparation of their mortars and concretes while preserving their rheological properties. These effects are well known in OPC systems, but much less information is available for AAM systems.
The nature of the aggregates also has a relevant influence on the rheology of OPC systems, mainly due to differences in the morphology of the aggregates that play a key role in the packing factor and possible interactions with cement matrices [23]. Specifically, limestone aggregates generally improve the rheological properties of their OPC mortars and concretes, although their role in the rheology of AAM systems is still largely unknown. Moreover, the total or partial replacement of natural aggregate with recycled construction and demolition waste furthers environmental protection, shrinks the carbon footprint, and ultimately enhances construction industry sustainability. The effect of recycled aggregate on the rheology, mechanical strength, and durability of OPC mortars and concretes has been studied in depth [24][25][26]. Recycling aggregates contain remaining mortar on their surface that increases the water adsorption [27] and decreases the fluidity of mortars and concrete. Furthermore, recycling aggregates normally have more angular shapes that natural aggregates, leading to lower packing factors than the latter [22,28] The possibility of preparing AAM mortars and concretes with recycled aggregate is of particular interest given the beneficial effect on sustainability. Only a handful of papers have been published on AAM mortars and concretes bearing recycled aggregates, however. Previous research conducted by the authors [29] showed that replacing natural with recycled aggregate by up to 20% has no adverse effect on strength. At the same time, although obviously necessary to understand and use AAM-Recycled aggregate systems and studies on their rheology are practically non-existent.
The research described here consequently aimed to determine the effect of the nature of the sand (siliceous, limestone, or recycle aggregates) on the rheological properties of alkali-activated slag (AAS) and alkali-activated fly ash (AAFA) mortars. Furthermore, the impact of the granulometry of the siliceous sand and its total or partial replacement by limestone and recycle aggregates on the AAS and AAFA mortars rheology has been investigated comparing the findings with the results for OPC mortars.

Materials
The alkali-activated mortars studied were AAS, with blast furnace slag (BFS) as the precursor, and AAFA, prepared with coal fly ash (FA), whilst OPC mortars were used as the reference. Table 1 gives the chemical composition of these starting materials determined by XRF in a S8 Tiger Bruker X-ray fluorescence spectrometer. Table 2 shows the particle size distribution determined by laser granulometry and the Blaine specific surface of BFS, FA, and OPC.
The aggregates used were: (i) a siliceous aggregate (S), (ii) a limestone aggregate (L), and (iii) a recycled aggregate (R) from construction and demolition waste. A picture of the three aggregates obtained on a binocular loupe Nikon SMZ-2T with a light supplier Intralux 4000, is shown in Figure 1.
To study the influence of the aggregate granulometry on the rheological parameters of AAM mortars, the siliceous aggregate (S) was sieved to obtain maximum D max values of 2 mm, 1 mm, and 0.5 mm. The percentage distribution by weight of each range used is presented in Table 3. To study of the influence of the nature of the aggregate on the mortar rheology of AAMs, the three aggregates were sieved manually to a maximum diameter of 2 mm and their particle size distributions were equalized to elude possible grain-size effects on the results.     Table 4 lists the physical and morphological properties of the siliceous, limestone, and recycled aggregates used. The humidity, water absorption, and density values of S, L, and R aggregates were determined according to EN 1097-6-2014 [30], while the bulk density and percentage of voids were determined according to the EN 1097-3 standard [31] for the different particle size fractions of aggregate S, and for same grain size (as S D max 2 mm) for aggregates L and R. Morphological characterization (including the aspect, circularity, and convexity values) was determined according to the methodology proposed by Westerholm and Hafid [32,33] and Equations (1) to (3).
Moreover, Table 4 lists the mineralogical characterization of different aggregates determined from the Rietveld analysis [34] of the XRD patterns (Bruker AXS D8 Advance X-ray diffractometer).

Mortar Preparation and Test Conducted
The AAM and OPC mortars were prepared as recommended in EN 196-1 [36] in an Ibertest IB32-040 C2000 mixer. The waterglass solution used to activate the slag (AAS-WG) had a SiO 2 /Na 2 O ratio of 1.5 and a Na 2 O concentration of 4 wt.% (by weight of slag), whilst the AAFA mortars (AAFA-10M) were activated with a 10M NaOH solution and the OPC mortars (OPC) were hydrated with deionized water. The aggregate/binder ratio was 2:1 throughout. All the mortars prepared are described in Table 5.
The following tests were conducted in the different mortars: • determination of the liquid/solid ratio required for plastic consistency as per the flow table test, determined on the grounds of a 150 ± 10 mm diameter spread pursuant to European standard EN 1015-3:2000 [37]; • determination of fresh state packing density (φ) as per Equation (4), and voids content as per Equations (5) and (6) for each mortar [38][39][40]; where: V is the total volume occupied by the mortar (Equation (7)); vs. is the solids volume; M is the mortar mass; ρ W is the density of the alkaline solution or water; u W is the liquid/solid ratio by volume; ρ α is the cementitious material density; R α is the volume ratio of the cementitious material to the total solids content; ρ β is aggregate density; and R β is the volume ratio of the aggregate to the total solids content; • stress growth test (SGT) conducted on a Viskomat NT mortar rheometer with a blade impeller operating at 1.5 rpm (0.025 s −1 ) as described in earlier studies [15]; • flow curve test (FCT) run on the same mortars as used for the stress growth test, subsequently subjected to preliminary shear at 30 rpm (0.5 s −1 ) for 25 s, and then ramped down by 4.5 rpm in 10 s steps, graphing the mean torque at each step against rotational speed and fitting to the respective rheological model to determine the dynamic yield stress and plastic viscosity.
Viskomat NT rheometer was calibrated as described in the literature [41] to express the readings in fundamental units.

Results and Discussion
The variables explored in the preparation of AAM mortars for this study of their rheological behavior were as follows: (a) The nature of the precursor (BFS, FA) or binder (OPC) activated or hydrated in conditions specific to each as described above; (b) The particle size distribution of siliceous aggregate (S); (c) The nature of the aggregates: 100% siliceous (S), 100% limestone (L), or 100% recycled, referred to total aggregate mass. A constant and equal particle size distribution with a maximum diameter of 2 mm and a constant sand/binder ratio of 2/1 were used throughout (d) S replacement ratio by: 10 wt.%, 20 wt.%, or 50 wt.% of R, likewise referred to total aggregate mass.
The results observed after performing the operations described in the preceding section are discussed below.

Mortar Liquid/Solid Ratio Required for Plastic Consistency
The liquid/solid ratios required for each mortar to reach European standard EN-1015-3 plastic consistency at 150 ± 10 mm diameter spread are shown in Table 6. In all mortars, and regardless of the precursor or binder used, the smaller the maximum particle diameter of the aggregate, the greater the amount of liquid required to obtain a plastic consistency and, therefore, a certain workability. This effect is due to the increase in the specific surface area with the decrease in the aggregate particle size, which leads to greater liquid demands to moisten the aggregates, as previously described in the literature [42][43][44][45]. In particular, the liquid/solid ratio increases 3% when going from D max of 2 mm to 1 mm, while an increase up to 14% has been measured for aggregates of a D max of 0.5 mm.
AAS-WG and AAFA-10M mortars require a higher demand of liquid compared to OPC for the same aggregate granulometries.  Table 6 shows that where siliceous or limestone aggregate was used the liquid/solid ratios were practically the same for the AAS-WG, AAFA-10, and OPC mortars. A closer analysis revealed that the liquid/solid ratio for both aggregate types was 0.56 in AAS-WG mortars, 0.60 for AAFA-10M mortars, and 0.52 in the OPC materials. In other words, the AAM mortars required more liquid to attain the same spread as OPC. That may be attributable to the greater density of the waterglass and 10M NaOH activators (around 1.27 g/cm 3 ). An earlier paper [15] reported the need for a higher liquid/solid ratio in AAFA-10M mortars and the generally greater AAM mortar sensitivity to change in liquid/solid ratios.
As expected, when 100% recycled aggregate was used, the liquid/solid ratio needed, was 25% to 30% higher. At 4.81%, water absorption was much higher in R (Table 4) than in aggregates S (0.08%) or L (0.13%), because the mortar adhered to the recycled aggregate particles rendered a higher specific surface area, consequently raising mixing water uptake [46]. Replacing S with up to 20% R (Table 6) led to no change in water demand, with values nearly identical to those found with 100% siliceous aggregate in both AAM and OPC mortars. At a replacement ratio of 50%, however, the liquid/solid ratio rose by a substantial 16% to 23% in all three mortars, with the greatest increase in AAFA-10M due to the added effect of the higher density of the activating solution, as noted earlier. In short, the AAM mortars, particularly AAFA-10M, required somewhat higher liquid/solid ratios than OPC mortars to attain a plastic consistency, whereas the partial replacement of S with up to 20% R induced no change in mixing liquid uptake in either the AAM or the OPC mortars.
Packing density (ϕ) calculated with Equation (4) and voids content with Equation (6) are given in Table 7 for each mortar. The effect of aggregate type on packing density in OPC mortars and concretes has been firmly established [47][48][49][50]. The analogous effect on AAM mortars has not been studied to date, however. Since in this study plastic consistency was defined as one of the parameters for mortar analysis, the l/s ratio was explored, given its relevance to the subsequent analysis of packing density and voids content.
The data in Table 7 attest to a number of effects: (a) With the maximum particle size (D max 2 mm), the packing density is slightly higher due to the greater particle size distribution, allowing particles of different sizes to present better compaction. This effect appears to be somewhat more pronounced in AAFA mortars. Therefore, the higher the packing density, the lower the void content and the greater the amount of available paste to lubricate the mortar [51][52][53]. This effect has been observed in both OPC and AAM mortars; (b) Generally speaking, and irrespective of the size and nature of the aggregate, AAM mortars had slightly higher packing densities than the OPC reference. A possible explanation for those findings may lie in the physical characteristics of the activator solutions (higher density and viscosity) and to a lesser extent in the surface characteristics of the precursors (the vitreous slag in particular); (c) Replacing siliceous with limestone aggregate always entailed a decline in packing density and a rise in the voids content in the mortars studied. The same observation was made by authors researching OPC concretes [47,49,54] and has been attributed to the fact that limestone aggregate is crushed, generating uneven particles with sharp edges (see morphological factors Table 4) that hinder compact packing and favor the formation of inter-particle voids; (d) The partial replacement of siliceous with recycled aggregate in AAM and OPC mortars apparently slightly lowered packing density and raised the voids content. Further to Table 7, at 20% replacement the voids values were quite similar to those observed for 100% siliceous aggregate-bearing mortar in AAM and OPC mortars both.
The foregoing confirmed that the size and nature of the aggregate affects mortar packing density and voids content, slightly more in Waterglass-AAS than in NaOH-AAFA and OPC mortars. Mortars bearing L aggregates showed a slightly lower packing density than mortars prepared with S aggregates. The same albeit less intense effect was observed in the mortars with recycled concrete aggregate.  Table 8 show the static (τ static ) and dynamic (τ dynamic ) yield stress and plastic viscosity (µ) for all the mortars studied. The τ static values calculated from the stress growth test (SGT) findings were converted to the International System of Units. Plastic viscosity (µ) was found as the slope of the linear regression line and dynamic shear as the y-intercept for the Bingham equation (Equation (8)) calculated from the flow curve test (FCT) data (see Figures 2-4).

Mortar Rheological Parameters
The conclusion to be drawn from the FCT data in Figures 2-4 is that all the mortars studied here behaved as Bingham fluids.
The values for the three rheological parameters listed in Table 8 for the AAM and OPC mortars prepared with siliceous aggregate with different particle sizes and the various aggregates are indicative of the following: (a) Irrespective of particle size or nature aggregate used, the AAS-WG mortars had consistently lower τ static and τ dynamic values than the AAFA-10M and OPC materials. The inference is that such mortars call for less energy to begin to flow. Similar findings reported for AAS paste [15] have been associated essentially with the fluidizing characteristics of the waterglass activator. The AAS-WG mortars also exhibited the greatest plastic viscosity as a result of the high viscosity of the activator; (b) The highest τ static and τ dynamic values were found for the OPC mortars followed by the AAFA-10M materials, with the former exhibiting the lowest plastic viscosity. These observations are attributable to inter-particle interaction, especially in OPC mortars, and low initial binder reactivity in the AAFA-10M mortars; (c) Regardless of the precursor or binder used, rheological parameters increase when the maximum particle size decreases. This effect is mainly due to several reasons: i) the lower the D max , the higher surface area of the aggregates, which leads to an increase in the friction between the particles and the yield stress [32,53]. ii) As the surface area of the aggregates increases, the adsorption of liquid rises, leaving less water available for the paste to lubricate the aggregates, and iii) the smaller the maximum particle diameter of the aggregate has a lower grain size distribution that leads to a packing density [44,47,54] and higher volume of intergranular voids (See Table 7); (d) In both the AAM and OPC mortars replacing siliceous with limestone and recycled aggregate entailed rises in τ static and τ dynamic , a logical outcome inasmuch as siliceous aggregate particles are more spherical and limestone and recycled aggregates are more uneven (see the sphericity and convexity factor data in Table 3); (e) Incrementally replacing siliceous by recycled aggregates from 10% to 100% also induced (practically linear) rises in τ static and τ dynamic .   The effect of the nature of the aggregate on τ static, τ dynamic and µ in these mortars was analyzed more exhaustively by comparing (in Figure 5) the values found for the AAM and OPC mortars with L and R aggregates to the values for the mortars bearing S aggregate.
As the Figure 5 shows, the AAM mortars were more sensitive to replacing siliceous with limestone or recycled aggregates than the OPC materials, for the three rheological parameters analyzed rose more steeply in the former, in particular in AAS-WG.
The extent of the effect of changing the type of aggregate varied depending on the parameter. The greatest variation in τ static was observed in the recycled aggregates followed by the limestone aggregates-bearing mortars. The rise in τ static in the limestone aggregate mortars was attributable to the high convexity factor found for L (f-convexity = 3.51; see Table 4), which induced greater friction between the particles, hindering initial mortar flow. In aggregate R, the effect on τ static was the result of the combination of two factors, the sharper edges in R than in S (see Table 4) and its rougher texture due to the mortar adhered to the surface, which increased inter-particle friction.
The widest variation in the τ dynamic values was induced by recycled aggregate in all the mortars. The inference, again, is that aggregate texture was the most significant factor once the mortar started to flow, i.e., in a state not beginning at rest. That finding was especially visible in the AAS-WG mortars, where the effect was heightened by the angular morphology of slag particles that have been shown to flocculate at early reaction times [9].
Irrespective of the precursor, the mortars with siliceous aggregate exhibited the lowest plastic viscosity (µ) values, for the greater particle sphericity induces greater packing, lower inter-particle friction, and lesser plastic viscosity [33,55]. The µ values barely varied with changes in aggregate in the OPC mortars. The effect of the change in aggregate type on AAFA-10M mortar viscosity was of scant significance, for FA activation reactions do not take place at ambient temperature [11], whereas wider differences were observed in µ in the AAS-WG mortars due to flocculation, as reported in earlier studies [9,13]. Longer mixing times in waterglass-activated slag have been shown [14,56] to improve mortar and concrete workability, lessening the adverse effect of such initial flocculation. The effect of replacing siliceous aggregate with recycled aggregate on the rheological parameters intensified with rising replacement ratios, as expected ( Figure 5). Rates of up to 10% induced no significant change in τ static or µ in any of the mortars except AAS-WG.
This study showed, in short, that the rheological parameters of AAM and particularly AAS-WG mortars undergo very significant rises when limestone aggregates or especially recycled aggregates are used instead of siliceous aggregate. These effects are much less accentuated in AAFA-10M mortars.

Conclusions
The conclusions drawn from the present findings are discussed below: (a) In both OPC and AAM mortars, the smaller the aggregate size, the higher the liquid/solid ratio required to achieve a similar plastic consistency. Plastic viscosity as well as static and dynamic yield stress increase with the decrease in maximum particle size. This effect is due to the increase of the interparticle friction and the decrease of the packing density of the mortars due to the narrower size distribution of the aggregate with the smaller size (D max = 0.5); (b) The AAM mortars, particularly the AAFA-10M materials, required higher liquid/solid ratios than OPC mortars to attain plastic consistency, whereas the partial replacement of the siliceous aggregate with up to 20% of recycled aggregate induced no change in in mixing liquid uptake in either AAM or OPC mortars; (c) Aggregate nature affected mortar packing density and voids content slightly more intensely in waterglass-AAS mortars than in NaOH-AAFA and OPC mortars. Possibly the most visible effect was found for mortars prepared with limestone aggregate, which, under all the conditions studied, exhibited lower packing density and a higher voids content than siliceous aggregate-containing mortars, due to the uneven shape of limestone aggregate particles. Although less accentuated, the same effect was observed in the mortars with recycled concrete aggregate; (d) All of the AAS-WG, AAFA-10M, and OPC mortars studied conformed well to the Bingham plastic model; (e) AAM mortars were found to be more sensitive than OPC mortars to changes in aggregate nature.
For as a rule, the values of the three rheological parameters studied (τ static, τ dynamic and µ) rose more substantially in the former and, consistently, more steeply in AAS-WG mortars when siliceous aggregate was replaced with recycled concrete or limestone aggregates. Aggregate texture and surface properties played an instrumental role in parameters τ static and τ dynamic in all the mortars, an effect especially visible when the mortars with limestone and recycled concrete aggregates were compared to the siliceous aggregate-bearing materials. The values of the three rheological parameters rose with rising replacement ratio (when recycled aggregates was substituted for siliceous one) in the AAM mortars, particularly AAS-WG. Rates of up to 10% induced no significant change in τ static or µ in any of the mortars except AAS-WG.