Alkaline Activation of Ceramic Waste Materials

Ceramic materials represent around 45 % of construction and demolition waste, and originate not only from the building process, but also as rejected bricks and tiles from industry. Despite the fact that these wastes are mostly used as road sub-base or construction backfill materials, they can also be employed as supplementary cementitious materials, or even as raw material for alkali-activated binders. This research aimed to investigate the properties and microstructure of alkali-activated cement pastes and mortars produced from ceramic waste materials of various origins. Sodium hydroxide and sodium silicate were used to prepare the activating solution. The compressive strength of the developed mortars ranged between 22 and 41 MPa after 7 days of curing at 65 °C, depending on the sodium concentration in the solution and the water/binder ratio. These results demonstrate the possibility of using alkali-activated ceramic materials in building applications. Ceramic materials represent around 45 % of construction and demolition waste in Spain. In the present study, two different ceramic materials, red ceramic bricks and porcelain stoneware were alkali-activated in order to produce pastes and mortars. The figure shows the microstructure of pastes obtained using a sodium silicate solution and NaOH pellets as activators. Both materials presented differences related to the process and the optimum concentration of activator. Mortars with compressive strengths ranging from 22 to 41 MPa were obtained after 7 days of curing at 65 °C, which make them suitable for building applications.

Review survey [3] reported a world tile production of 9,515 million sq m in 2010, of which 96.0 % (9,170 million m 2 ) implied the 30 major manufacturing countries. Although the Spanish industry recorded a reduction in production of almost 40 % from 2006 to 2010 (608-366 million m 2 , respectively), it still ranked seventh in the list of manufacturers in 2010 with 3.8 % of total world production. Since consumption in the Spanish market has diminished by 54.0 % since 2006, two-thirds of manufactured tiles were exported.
Despite the vast majority of the ceramic wastes being used in landfills with a low added value, prior research has proved its suitability in concrete and as cementitious materials. In the studies by Medina et al. [4], up to 25.0 % of natural coarse aggregates were replaced with ceramic sanitaryware wastes to obtain concrete for structural purposes. Similarly, Pacheco-Torgal and Jalali [5] observed not only a slight increase in water absorption and permeability when replacing traditional coarse aggregates with ceramic wastes, but also superior durability when traditional sand was replaced. Furthermore, several studies have confirmed the potential of ceramic wastes to produce pozzolanic cements [1,[6][7][8]. Among them, Puertas et al. [1] not only successfully used up to 35.0 % of certain types of ceramic wastes as pozzolan admixtures, but also proved their suitability as raw materials for Portland cement clinker production.
Although ceramic materials can be used as cement admixtures and concrete aggregates, in these applications only a portion of cement is replaced (usually 10-35 %) so, it is interesting to develop binders that are made entirely, or almost entirely, from waste materials [9]. This can be achieved by the alkali activation process. A conceptual model of the reaction processes involved was proposed by van Deventer et al. [10] for the metakaolin system. An aluminosilicate-based material is dissolved by a highly concentrated alkali hydroxide and silicate solution, giving rise to sodium or potassium silicate and aluminate monomers. These precipitate and form an alkali-aluminosilicate gel (amorphous) and sites for the nucleation of zeolitic type phases (nanocrystalline), brought about by the transformation of the gel. As Deventer et al. [10] pointed out, the reaction mechanism is modified depending on the chemistry of the raw material.
The success of converting waste materials into useful products following this process has been extensively proved in materials such as silicoaluminous fly ash, metakaolin or blast furnace slag [11][12][13], and its suitability has also been confirmed for other waste materials, such as hydrated-carbonated cement [14], glass [15] or ceramic materials [16][17][18]. In the study by Puertas et al. [16], six different ceramic wastes were mixed with NaOH and sodium silicate solution to give a maximum compressive strength of 13 MPa for pastes cured for 8 days at 40°C.
Mortars with similar compressive strengths (14 MPa) were obtained by Reig et al. [17,18] by mixing red hollow bricks with a NaOH solution and by curing samples for 7 days at 65°C. However, further research must be done in order to understand the influence of the alkali activator on the alkali activation process of ceramic materials of different natures. For this purpose, two different ceramic products, with very different sintering temperatures and chemical compositions, were selected: red clay bricks and porcelain stoneware. While the former are sintered at temperatures ranging from 800 to 1,000°C [8], porcelain tiles are usually sintered between 1,190 and 1,220°C [19]. According to Baronio and Binda [20], powder from bricks is expected to present pozzolanic activity as the crystalline network is destroyed when the structural hydroxyl groups of clay minerals (phyllosilicates or sheet silicates) are lost (600-900°C). According to the study by Zanelli et al. [21], which analyzed 93 porcelain stoneware samples, porcelain tiles are also presumed to react. Their mineralogy is composed of some crystalline phases, such as quartz, mullite or feldspars, which are dispersed throughout a main vitreous phase whose proportion varies from 40 to 80 % depending on the sample.
This research aimed to develop binders by the alkali activation of two different ceramic waste materials (porous red clay brick and porcelain stoneware) and to analyze the influence of the alkali activator concentration on the mechanical strength and microstructure of the binders formed.

Materials
Two different ceramic waste materials were used for the alkaline activation process: red clay brick (B) and porcelain stoneware (P). The materials were crushed in a jaw crusher to obtain a particle diameter of less than 4 mm. This granulated material was then ground in a laboratory-type ball mill (alumina medium, 40 min). Particle size distribution was measured using a laser analyzer (Mastersizer 2000, Malvern Instruments). Figure 1 presents both cumulative curves. As shown, both powders presented a similar particle distribution, with particles ranging from 0.2 to 100 lm, 90 % in volume under 50 lm, and almost 7 % had a diameter under 1 lm. Despite both ceramic materials having a mean particle diameter close to 20 microns, a slightly larger amount of thinner particles (under 10 micron) was observed in the red clay brick powder.
In Fig. 2, both ground materials were examined by scanning electron microscopy (JEOL JSM-6300). The irregular shape of the milled particles is observed and particles show no significant porosity.
The chemical composition of the milled samples was determined by an X-ray fluorescence analysis (XRF). As shown in Table 1, the amount of SiO 2 was larger in the porcelain waste (71 vs 51 %), while the presence of other compounds (CaO, K 2 O, MgO and Fe 2 O 3 ) was barely noticeable. In fact, 94 % of porcelain stoneware was composed of SiO 2 , Al 2 O 3 and Na 2 O. For red clay brick waste, the sum of silica, alumina and sodium amounted to less than 70 %. In both cases, loss on ignition (LOI) was below 2 %.

Preparation of Paste and Mortar Samples
To develop the alkali-activated binders, ceramic waste materials were mixed with an alkaline solution. The activating solution was prepared by dissolving sodium hydroxide pellets (Panreac, 98 % purity) with water and a sodium silicate solution (Merck, Waterglass SiO 2 = 28 %, Paste samples were obtained by mixing ground ceramic with the required alkaline solution for 4 min, and they were cast in plastic containers. Mortar samples were prepared by mixing the ceramic material with the activating solution for 2 min. Siliceous sand (4.36 modulus fineness and   maximum particle diameter of 2 mm) was then added and the mixing process was continued for a further 3.5-min period. The formed mortar samples were placed into a mould and were vibrated for 4 min. Both pastes and mortars were stored in a thermostatically controlled bath at 65°C for 7 days at 100 % relative humidity. Table 2 summarises the mix proportions used in this study. Mixes are coded as 'x/x/m/r-c', where x is the type of ceramic waste (binder: B = brick and P = porcelain), x is the amount of water per 100 g of binder, m is the molality (mol/kg) of Na ? in the activating solution, r is the SiO 2 /Na 2 O molar ratio in the activating solution and -c is the percentage of Ca(OH) 2 (93 % purity) used to replace ceramic waste.
A water/binder (w/b) ratio of 0.45 was used for the samples made with brick, which was lowered in samples 'B/40/8.0/1.60' and 'B/35/9.0/1.60' to analyze the influence of this parameter (w/b). Due to the reduced water absorption of the porcelain stoneware (less than 0.5 %) [23], the w/b ratio was lowered to 0.35 for all the porcelain waste mixes. A binder/sand (b/s) ratio of 1:3 was employed for all the mortars tested in this paper (see Table 2).

Sample Testing
Compressive strength was determined on the alkali-activated mortars following the UNE EN 196-1 standard. The microstructure was examined on paste samples using SEM-EDX (JEOL JSM-6300) and mineralogical phases were identified by X-ray diffraction XRD. Thermogravimetry (TG, TGA-850 Mettler-Toledo thermobalance) was done to determine the mass loss of samples in an N 2 atmosphere using sealed pinholed aluminium crucibles at a heating rate of 10°C min -1 . Tests were run from 35 to 600°C to assess the mass loss related to bonded water molecules or hydroxyl groups in the pastes. Samples were taken on day 7 of curing at 65°C and with a relative humidity of 95-100 %.

Results and Discussion
Alkali Activation of Porcelain Stoneware Waste: Preliminary Study Alkali-activated cement pastes and mortars using NaOH and sodium silicate solutions as activators were prepared by employing porcelain stoneware powder and after considering that the water demand for this residue was lesser than that for red clay brick. Thus, the water/binder ratio was fixed at 0.35. Sodium molality and the SiO 2 /Na 2 O molar ratio varied and fell within the range of 6.5-8.5 and 1.12-2.13, respectively. Table 3 summarises the behaviour of the fresh pastes developed with the porcelain waste. As regards the setting of pastes, a strong dependence on the amount of Ca(OH) 2 was observed so that only those mortars with 2 % calcium hydroxide (replacing the porcelain stoneware  powder) actually set during the first 24 h. While mortars without calcium hydroxide did not harden, the setting time for those whose calcium hydroxide contents were higher than 5 % was so short that it was not possible to place samples into the mould. The presence of soluble calcium in the binding material was required to produce a stable matrix after curing, which is in agreement with [24,25], who observed that deficiencies in the CaO content of the raw material can be compensated by adding mineral additives.

Compressive Strength Development
Effect of Na ? Concentration on the Alkali Activation of Ceramic Waste The compressive strength of the mortars mixed at a constant 'SiO 2 /Na 2 O' ratio (1.60) and different Na ? concentrations (6.0-9.0 mol/kg) is depicted in Fig. 4. For the activation of the porcelain stoneware-based mortars, and according to ''Effect of Na ? Concentration on the Alkali Activation of Ceramic Waste'' section, 2 % of the waste material was replaced with Ca(OH) 2 . Furthermore, the presence of Ca(OH) 2 was not necessary for the alkaliactivated red clay brick based mixtures. The mortars containing B waste exhibited the highest compressive strength for a sodium concentration within the 6-7 mol/kg range. However, the mechanical strength of porcelain mortars increased with alkali concentration. These results agree with the studies reported by Provis et al. [26], who observed that the optimum activator concentration depends on the precursor, and that this must be the case to balance the charges of tetrahedral Si and Al, thus avoiding the presence of unreacted sodium or silica. Duxson et al. [27]. and Tashima et al. [28] also found an optimal concentration for the alkali activation of metakaolin and the fluid catalytic cracking catalyst residue, respectively.

Influence of Na ? Concentration and the Water/Binder (w/b) Ratio on Alkali-Activated Red Clay Brick Waste
In order to assess the effect of the amount of water on the activating solution, clay brick-containing mortars were prepared with a w/b ratio that fell within the 0.45-0.35 range. Although low workability was observed when reducing the quantity of water, it was sufficient to allow the mortars to be cast into the mould. As observed in Fig. 5, compressive strength increased by more than 40 % when the w/b ratio lowered from 0.45 to 0.35, meaning that the amount of water was still sufficient to effectively wet brick particles. According to Komnitsas et al. [29], enhanced compressive strength may be motivated by the lesser available amount of free water after the hydration process, which is expected to diffuse or evaporate, leading to pores and cracks. The good compressive strength behaviour observed when lowering the w/b ratio also implies an improvement from the environmental point of view, since better mechanical properties are obtained while maintaining the 'Na ? /binder' and 'SiO 2 /binder' ratios constant (3.15 mol of Na ? and 2.53 mol of SiO 2 per kg of ceramic waste).

Scanning Electron Microscopy Studies
The microstructure of the pastes obtained by the alkali activation of bricks and porcelain stoneware powders is represented in Fig. 6. Both matrices look denser for larger amounts of sodium, and porcelain is more compact for a given sodium molality. However, brick pastes presented areas with a high sodium concentration (detected by EDS), which became more frequent in sample 'B/45/9/1.60' (9 mol/kg). These results relate with the mechanical properties of brick mortars, suggesting that excess sodium is provided for concentrations higher than 7 mol/kg. Conversely, porcelain pastes looked denser with increasing sodium contents, which is consistent with the compressive strength presented by mortars. What this suggests is that even higher sodium concentrations may develop a denser microstructure, leading to higher compressive strength values. Furthermore, all the pastes presented unreacted particles surrounded by hydration products. According to the results obtained by Kourti et al. [15], porcelain particles are expected to influence the mechanical properties of the matrix more positively, as they are stronger and better bonded to the matrix than red clay brick particles.

X-ray Diffraction Studies
The diffraction patterns of the pastes obtained by the alkali activation of brick and porcelain wastes are presented in Fig. 7a and b, respectively. The XRD data of the ground material were plotted to make comparisons. As observed, the identifiable crystalline phases in the raw material were not completely dissolved by the alkaline solution, and most remained in the activated binder. Only wollastonite and gehlenite, found in small quantities in ground brick, were not identified in the activated pastes. Moreover, the amount of amorphous phases slightly increased in the porcelain binder when compared to the raw material, whose content increased with the highest sodium concentration (9 mol/kg). Small quantities of sodium carbonate (natrite, Na 2 CO 3 ) were identified only in brick paste 'B/45/9/1.60'. This corroborates that a concentration of 9 mol/kg provides brick samples with excess sodium which, according to  Provis et al. [26], can lead to the formation of carbonate salts from atmospheric carbonation.

Thermogravimetric Analysis of Pastes
The differential thermogravimetric curves and the total mass loss of alkali-activated brick and porcelain pastes are represented in Fig. 8. As observed, all the pastes had a similar total mass loss, which increased with a higher sodium concentration in the solution. According to Bernal et al. [30], this behaviour can be attributed to a more marked presence of OHgroups provided by NaOH in the activating solution.
All the pastes showed a single peak which centred at approximately 130°C and was attributed to the loss of the bound water of the pastes in [28,31]. Although several authors [10,12,30,32] have identified zeolitic reaction products in alkali-activated binders, no zeolitic signals were recognised by thermogravimetry as, according to Bernal et al. [30], they tend to overlap with the peak that centres at 130°C. Likewise, it was not possible to identify sodium carbonates with the TG analyses because their decomposition occurs at temperatures above 600°C [30]. Additionally for the porcelain stoneware containing pastes, no Ca(OH) 2 decomposition was detected at a high temperature (500-600°C), which implies that this reagent was totally consumed during matrix formation.

Conclusions
Alkali-activated binders have been obtained through the alkali activation of two different ceramic waste materials. During the alkali activation process of porcelain stoneware, the addition of Ca(OH) 2 proved an essential constituent, and only those samples containing 2 % of this reagent were obtained. The optimum mix for the alkali-activated porcelain stoneware was 'P/35/9/1.60-2' (9 mol/kg), which gives mortars whose compressive strengths come close to 30 MPa after 7 curing days at 65°C. Conversely for the red clay brick powder, the best compressive strength was obtained for mortar 'B/45/7/1.60' (7 mol/kg), which diminished for  higher sodium concentrations. The w/b ratio was seen to markedly influence the compressive strength for brick mortars as it increased by 40 % (29-41 MPa) when the w/b ratio lowered (from 0.45 to 0.35) and when the 'activator/ binder' ratio remained constant. Further research must be conducted in order to completely understand the influence of calcium on the properties of the binders developed by the alkali activation of porcelain waste and to obtain mixtures with settings at room temperature.