1.1 Raw materials for the synthesis of CMC

The raw materials useful to prepare silicoaluminate CMC must comply with some specific requirements: (1) Chemical composition rich in SiO₂ and Al₂O₃; (2) An abundant fraction of glassy phase, fundamental for the susceptibility to the alkaline attack; (3) Sufficiently small particle size in order to promote its reactivity.

Clays are silicoaluminate raw materials, and are among the most abundant minerals in the earth crust. Calcined clays have been used as raw materials in the production of CMC (^{Barbosa et al, 2000}); most of the reports have been based on the use of metakaolin (Al₂Si₂O₇) (^{Davidovits, 1982}), which is obtained after firing kaolinite (2SiO₂·Al₂O₃·2H₂O). The latter is a clay of SiO₂:Al₂O₃ configuration of the type 1:1 (^{Rowles y O´connor, 2003}), constituted by one layer of Silicon atoms in tetrahedral coordination with Oxygen atoms and one layer of aluminum atoms in octahedral coordination with oxygen and OH^- ions (also known as gibbsite layer). The chemical composition of kaolinite is (wt %) 46.54%SiO₂; 39.50%Al₂O₃ 39.50%; H₂O 13.96%; it shows little reactivity towards alkali under normal conditions. However, after firing at 650-900ºC, the kaolinite losses the OH^- groups and the aluminum layer collapses (^{Shvarzman et al, 2003}) and the aluminum coordination changes to tetrahedral, while the layer of Silicon remains (^{Kakali et al, 2001}). The resulting product is metakaolin, which is amorphous to X-ray diffraction. This conversion increases the reactivity towards alkaline environments even at room temperature, which is the key for the CMC materials. The kaolinitic minerals may contain impurities such as quartz and other clays, as well as the substitution of iron and/or titanium for aluminum. Regarding the purity, previous work in Cinvestav Saltillo (^{Arellano-Aguilar et al., 2014}, ^{Burciaga-Diaz et al., 2012}, Burciaga, 2014) has shown that even low purity metakaolinitic minerals are useful to produce CMC.

1.2 Processing of CMC materials

Compared to the processing of conventional Portland cements, the CMC have lower environmental emissions and better technological properties; moreover, these can be prepared using widely available raw materials, processing fluid and moldable pastes. The CMC harden by virtue of chemical reactions that advance gradually with time, reaching properties of interest for a wide range of applications. The preparation of CMC depends on various factors, for which their effect has not been completely understood according to the literature, some of these are:

The chemical composition of the overall formulation plays a major role that defines the mechanical strength (^{Burciaga-Diaz et al, 2012}); it is defined on the basis of molar ratios such as: SiO₂/Al₂O₃, M₂O/Al₂O₃ y M₂O/H₂O (^{Burciaga-Diaz and Escalante-García, 2004}). The literature indicates variable compositional ranges for several raw materials; the ratios reported by the patents of Davidovits, do not always result in the best properties for any raw material. The chemical composition of the CMC can also be reported including the concentration of mineral raw materials and alkaline activators. The most common among the latter are the alkaline silicates of the type M₂O:xSiO_2, which promote high pH values (^{Palomo et al, 1999}; ^{Davidovits, 1984}), where M can be Na or K.

The chemical reactions during the synthesis of CMC have been proposed to take place in three stages: (1) Destruction of the atomic structure of the mineral. The alkaline media offers chemical stimuli such as the variation of the ionic strength caused by the presence of alkaline metals that perform as electron-donors; this causes the breaking of the bonds Me-O, Si-O-Si, Al-O-Al y Al-O-Si, caused by the alteration of the electronic density around the Si and Al atoms. Some of the species formed include silicic acid (Si(OH)₄), and anions such as Si-O^-, Al(OH)₄ ^-, Al(OH)₅ ^2- y Al(OH)₆ ^3-. (2) The mentioned species saturate the solution and a policondensation takes place, forming new chemical products, leading to the setting, and a reduction of the pH takes place due to the interaction of the hidrosilicates and hydroaluminates with the alkalis. (3) Precipitation of products as a result of the particles formed in the previous stage, which results in an enhancement of the mechanical strength; the reduction of pH leads also to de condensation of silica gel from the alkaline activator, which favors the strength.

The conventional synthesis of CMC involves mixing the mineral raw materials with the alkaline solutions of activators, as well as the molding and curing to attain solidification. The curing commonly takes place at temperatures from ambient to 120°C (^{Barbosa y McKenzie, 2003}; ^{de Vargas et al, 2011}; ^{Burciaga-Diaz et al, 2012}). Curing at high temperature allows gains of high early strength, some authors have pointed that higher strengths are reached with higher temperatures, e.g. 60MPa after curing 24 h at 85°C (^{Palomo et al, 1999}, ^{Rowles y O´connor, 2003}); however, it has also been reported that after curing at high temperature the strength gains are negligible (^{Arellano-Aguilar et al, 2014}). Information about the use of curing temperatures above 100°C is scarce in the literature.

1.3 This investigation

There exist in the literature a considerable number of reports related to the synthesis of geopolymeric materials based on high purity clay minerals and industrial by products such as metallurgical slags, fly ash, muds, etc. Nonetheless, there exists a limited number on reports based on kaolinitic minerals of low grade (^{Burciaga-Diaz et al., 2012}; ^{Arellano-Aguilar et al., 2014}). The latter are an abundant resource worldwide, so their potential is considerable and the opportunities must be explored (^{Davidovits, 2002}), such is the case of this investigation. On the other hand, there are only a few papers on the synthesis of CMC materials that combine the CMC processing under the conditions proposed in this research (^{Asbridge et al., 2002}, ^{Zivica et al., 2011}), i.e.: very low water/solid ratios and uniaxial compression.

This paper presents the results of an investigation to develop a strategy to synthesize CMC materials in the shape of small tiles, using those raw materials commonly consumed by the traditional ceramic industry of tile production, which are in turn of lower purity than those used in CMC synthesized as cementitious binders. In contrast to the traditional ceramic industry, this paper proposes a simpler route, with less processing steps, using less water, less energy and generating less CO₂ emissions, but with the potential of preparing ceramic-like materials of properties similar to those currently produced by conventional routes but in a more sustainable way.

2.1 Materias Primas

Some experimental details are omitted to protect authorship rights and industrial secrets. Three silicoaluminate minerals were employed, labelled as MA, MB y MC; the MA was fired at 700°C and was labelled as MAc. Table 1 shows the chemical composition obtained by X-ray fluorescence spectroscopy (FRX), from which the differences in chemical composition among the silicoaluminate minerals were evident; MA shows high content of iron. The mineralogy of the raw materials was characterized by X-ray diffraction, the results are shown with the results in Figure 4 for reasons of space and for a better comparison and understanding of the results. The difference among MA and MAc was the disappearance of the reflections of kaolinite, although the other crystalline phases persisted after the firing at 700°C; on the other hand, MB and MC showed an amorphous hump in the 2theta position typical for silicoaluminates.

2.2 Processing of materials

After a series of preliminary tests carried out to determine a set of experimental parameters, 4 mixtures were prepared using the raw materials from Table 1; here the mixtures are made of MAc and any of the other 3 raw minerals. Table 2 lists the formulations prepared. The alkaline activators were chemical compounds of Na₂O·SiO₂·H₂O with weight ratios SiO₂/Na₂O lower than 2. The amounts of Na₂O used were similar to those commonly reported for CMC from the literature using calcined clays. The amount of water used was the minimum necessary to provide the mixtures a consistency useful for the conformation of green bodies ready for the subsequent processing stage.

The powdered minerals and activator solutions were mixed using a high speed mixer. The semi-dry mixtures were placed in the cavities of molds of dimensions of 4x16cm, the thickness of the resulting specimens was of 0.8-1.0cm. The mixtures were subjected to uniaxial compression using pressures up to 30.4 MPa. The tile specimens were demolded and thermally treated at 200° for 120 minutes and then left to cool to room temperature before characterization.

2.3 Characterization of materials

The specimens were tested by flexural strength using an hydraulic machine following the ISO 10545-4; due to difficulties associated to the processing of the specimens, only the flexural strength was measured, which is more important than the compressive strength for tile samples. The density was measured by the Archimedes method using liquid mercury. After the mechanical testing, the sample relics were used for further characterization. A fraction of the samples was ground in a planetary mill with agate media in order to pass the 105(m mesh, before characterization by DRX (Phillips PW3040) under the following operating conditions: CuKα radiation (1.542 Å), range 10-70° 2θ, step of 0,03° 2θ and incidence time of 3s per step. Solid fragments of specimens were mounted in resin for characterization by scanning electron microscopy (MEB Philips XL30ESEM coupled with energy dispersive spectroscopy, EDS) at an acceleration voltage of 20 ekV. The specimens were polished and coated with carbon. EDS spot analysis were performed to determine semi-quantitatively the chemical composition of specific areas in the samples, the measuring time was of 30 seconds for each analysis.

3. RESULTS AND DISCUSSION

Table 3 presents the results of compressive strength and density of the formulations investigated, which showed notable variations in flexural strength. The highest strength was for the mixture M1 (MAc-MB with low content of MB), this was close to the requirements for conventional ceramic tiles and 35% higher than the strength of the mixture M4 (MAc-MC with high content of MC). Higher contents of MB among mixtures M1 and M2 resulted in a reduction of 43% of the strength. The density values did not correlate to those of strength, i.e. M3 was denser and also weaker than mixtures M1 y M4; this could be due to differences in the intrinsic mechanical properties of the reaction products formed during the thermochemical activation.

It is noteworthy that the flexural strength values registered were superior to those reported for CMC processed in a conventional way, which are commonly much lower than 10 MPa after 28 days; this indicates that the proposed processing method is quite efficient to attain high strength in considerably short curing times. Considering that for concretes, the flexural strength is approximately 10% of the compressive strength of the specimen, the mixtures investigated would be expected to potentially show compressive strengths of 69 to 157 MPa after only 2 h of preparation. The obtained flexural strength results could be considered similar to those reported previously using uniaxial compression and low water/solid ratios (^{Zivica et al., 2011}); they reported compressive strengths of up to 146 MPa, which would correspond to an estimate of approximately 15 MPa of flexural strength. Nonetheless, the minerals used in this investigation are of far lower grade that those of the referred report, which makes our route more economical.

Figure 1 presents the microstructure of the mixture M1, obtained by backscattered electron imaging. The matrix of reaction products was dense, in agreement with the good flexural strength. Some particles of MB showed internal porosity. Some particles of MAc showed bright zones, which occasionally showed an elongated morphology; the brightness of such zones is attributed to a higher emission of backscattered electrons due to the presence of compounds of higher average atomic number, such zones could be unreacted particles that have not incorporated water, as happens in the reaction products that appear of a darker gray tone.

Figure 1 Microstructure obtained by scanning electron microscopy and backscattered electrons of mixture M1. 

Figure 2 shows the microstructure of mixture M3 (the lowest flexural strength). The densification of the matrix of reaction products was similar to that of mixture M1, so the reduced strength is attributed to the formation of reaction products of lower intrinsic strength. The microstructure shows some bright zones, which were rich in Fe y Ti, in agreement with the chemical composition of the MAc.

Figure 2 Microstructure obtained by scanning electron microscopy and backscattered electrons of mixture M3. 

Figure 3 shows the micrographs from the microstructure of mixture M4, which displayed a similar densification as mixtures M1 and M3. The bright zones showed high contents of Fe and Ti (high atomic number) in the matrix corresponding to MA, while the matrix of MC showed a chemical composition consistent with a silicoaluminate in most of the particles (similar to MB).

Figure 3 Microstructure obtained by scanning electron microscopy and backscattered electrons of mixture M4. 

The results show that the type and amount of Component 2 (see Table 2) influences directly on the resulting properties. The differentiation among reaction products and unreacted raw materials is complicated in the micrographs, as both are silicoaluminates, so the differentiation by means of the gray tone is not as easy as in the case of hydrated conventional Portland cement. However, due to the high flexural strength obtained, it can be inferred that the reaction cementitious products are intimately intermixed with the unreacted raw materials.

Figure 4 presents the XRD patterns of the investigated mixtures, the patterns of the unreacted raw materials are included for reference purposes. In general terms, it was noted that the crystalline fractions remained unreacted after the solidification, indicating that these are inert during the curing process at 200°C for 2h. The formation of zeolites was not observed as reported for some CMC materials. On the other hand, the reaction products of cementitious CMC are characterized by an amorphous nature. For the mixtures M1, M2 y M4 the amorphous hump persisted, although it was noted that such humps were widened and shifted slightly to the left, indicating the formation of new reaction products of amorphous nature but different to those of the starting raw materials. In contrast, for the mixture M3 the amorphous hump was not clear, although the consolidation and mechanical strength indicated the formation of reaction products, so it is possible that the high crystallinity of the specimens overlaps and overshadows the presence of amorphous products.

Figure 4 X-ray diffraction patterns of the mixtures and raw materials investigated. 

4. CONCLUSIONS