Microstructure and engineering properties of Fe 2 O 3 (FeO)-Al 2 O 3 -SiO 2 based geopolymer composites

The objective of this study is to develop low cost, eco-friendly and sustainable building materials by applying the technology of mineral polymerization (geopolymerization) process on naturally abundant iron-rich aluminosilicate (laterite) materials. Iron-rich aluminosilicates based-geopolymer composites containing 10 to 40wt% of rice husk ash (RHA) were cured at room temperature and at 90 °C. This paper examines the phase transformation, microstructural and mechanical changes that occur in the geopolymer composites when fine aggregates of quartz sand are added. Experimental results indicate good polycondensation and more cohesion resulting in high strength due to the better dissolution of RHA that provides soluble reactive silica to equilibrate the Si/Al and Si/Fe molar ratios. Ferro-sialates, Fe(


Introduction
Laterites and lateritic soils, described as Fe 2 O 3 (FeO)-Al 2 O 3 -SiO 2 -H 2 O matrices, are made from kaolinite in which a high proportion of Al 3+ is replaced by Fe 2+ or Fe 3+ (Lyon Associates, Inc., 1971;Obonyo et al. 2014). These materials are available in tropical and sub-tropical areas of the world, with nearly 67% found in the Cameroonian territory (Mbumbia et al. 2000). The replacement of aluminium by iron atoms in kaolinite structure affects their crystallinity with consequent increase of amorphous content and enhance their vulnerability to chemical attack Pignatelli et al. 2014;Kaze et al. 2017). Geopolymers produce from natural iron-rich aluminosilicate (laterites) without treatment should be eco-friendly and more greener in comparison of metakaolin based geopolymer which requires an energy for the thermal treatment in the range of 700-800 °C (Kamseu et al. 2011;Elimbi et al. 2011, Tchakoute et al. 2016. The use of laterites as raw materials for the development of geopolymers might be considered as the way for the reduction of amount of CO 2 emitted from cement industries and also the energy used to activate the clay materials. As reported by  and , it is easier in tropical areas to have laterites, generally at the surface, than struggle for clays for which exploitation will be detrimental for the environment because they are covered in most cases by laterites or various types of soils. Therefore a sustainable materials from activated laterites locally sourced, with negligible transport costs and environmental impact, will present thermal efficiency, financial viability and low energy required in the manufacturing process as mentioned by Duxson et al. (2007). This paper is the second in a series of articles describing the use of laterite for the development of geopolymer composites. It explores the engineering properties of two laterites, with similar degree of laterisation (~35wt% of FeO(Fe 2 O 3 ) used for the preparation of alkali activated materials, hereafter indicated as geopolymer composites, with addition of rice husk ash under different curing conditions. Physical and chemical properties of laterites based geopolymers were described in the first paper (Kaze et al. 2017). We observed that, the values of flexural strength of laterites based geopolymers are similar to those of standard metakaolin based geopolymers and decrease with the thermal activation of solid precursors above 500 °C (Kaze et al. 2017). The setting ACCEPTED MANUSCRIPT 3 time and the water absorption also decreased with thermal treatment of laterite. Their microstructure appeared coarse and more inhomogeneous with the increase of the calcined temperature above 500 °C. In recent times these soils were valorised as potential candidate for the development of geopolymer cements (Cristiane et al. 2010;Obonyo et al. 2014;Lemougna et al. 2014;Lassinantti et al. 2015;Nik Ab Aziz et al. 2015;Kaze et al. 2017) while others preferred mixtures with cement (Jayasinghe and Mallawaarachchi 2009).
The most used of traditional precursor materials for geopolymers are like metakaolinite (Fe 2 O 3 content of 2-5%), fly ash (Fe 2 O 3 content of 10%), volcanic scoria (10-15%) and blast furnace slag (Fe 2 O 3 content of 0.5%). Nevertheless, recent studies have shown that precursors with iron content higher than the usually found in fly ashes and volcanic scoria may be activated in alkaline environment (Kaze et al. 2017;Gomez et al. 2010;Obonyo et al. 2014;Lemougna et al. 2014) with potential applications in engineering. Gomes et al. (2007) observed that, the alkaline activation of raw material with high iron and low aluminium content produced the geopolymer with compressive strength ranging from 20 to 80 MPa. Obonyo et al. (2014) reported that the addition of calcined low iron content laterite (up to 30% mass) to poorly reactive natural laterite in alkaline medium, led to geopolymers with improved 28 days physical and mechanical properties. The detailed microstructure of products of the geopolymerization of laterites revealed low level of homogeneity and phase distribution although the relative good mechanical strength in comparison with standard geopolymer cements. The presence of some large capillary pores also affect the final properties of products of geopolymerization of laterites. In our point of view the poor homogeneity and larger and non-homogeneous porosity are due to alumina and iron-rich oligomers that remain unbounded to the geopolymer system after geopolymerization. This study aims to investigate the possibility of using rice husk ash (RHA) as SiO 2 source for the improvement of the strength and stability of Fe 2 O 3 (FeO)-Al 2 O 3 -SiO 2 systems, sustainable inorganic polymers materials (geopolymers). We plan the experimental design so that sufficient amount of soluble silica added to the laterite-based geopolymer composites will combine all the residual Al and Fe-oligomers that do not react effectively in laterites. The soluble silica provides Si-based oligomers to develop more bending phases. The three-point flexural strength was used as an indicator for the bending bonds development. The microstructural analysis of geopolymer composites was done through Environment Scanning Electron Microscope (ESEM) analysis to understand the features of the composites. The Mercury Intrusion Porosity (MIP) and water absorption were also studied to evaluate and ACCEPTED MANUSCRIPT 4 appreciate the pore sizes distribution within specimens. The main factors that govern our investigations include percentages of RHA (SiO 2 source) and curing process.

Materials and characterization
The iron-rich aluminosilicates (laterites), LATOD and LATEL, used in this study were The chemical composition of both laterites carried out by X-Ray Fluorescence analysis (Thermo ARL, ARL Advant XP and XP + X-Ray Fluorescence Spectrometer, Thermo Fisher Scientific, MA, USA) using argon-methane as inert gas is reported in Table 1. The concentration is given in wt% or ppm by the data Quanta As modelling software. The mineralogical composition of both laterites were determined by X-ray powder diffraction (XRD) analysis (PANAlytical X'Pert PRO diffractometer, Ni filtered Cu-Kα radiation,  = 1.5405 Å). Both laterites present the similar mineral phases (kaolinite, quartz, goethite, anatase, rutile, hematite, maghemite and ilmenite), except lepidocrocite contained in Eloumden's laterite (see Figure 1). The particle size and B.E.T. surface area were determined by Kaze et al. (2017). The average values of the particle sizes d 0.5 were 34.90 and 38.30 µm for Eloumden and Odza, respectively, those of d 0.1 and d 0.9 were: 2.48 and 3.39 µm for Eloumden; 106.20 and 170.70 µm for Odza. The B.E.T. surface are 21.92 ± 0.11 m 2 /g and 23.87 ± 0.13 m 2 /g for Eloumden and Odza, respectively (Table 2).
Rice husk was collected from Ndop, Department of Ngoketundjia, Region of North-West (Cameroon). The silica source (RHA) was obtained from calcination of dried rice husk at 600 °C for 2 h (heating/cooling rate of 5 °C/min). The chemical composition of RHA is presented in Table 1, and its XRD pattern ( Figure 2) shows a typical broad hump diffraction peak in the range between 2θ = 15° and 35° with maximum around 21°, which implies the high amount of amorphous phase of silica from calcined rice husk (Tchakouté et al. 2016). Some quartz peaks were also observed.

5
The NaOH solution was prepared by dissolving laboratory grade granules (98 wt%, Sigma Aldrich, Italy) into distilled water to have 8 M concentration. The commercial sodium silicate solution (SiO 2 /Na 2 O = 3.00; loss of ignition = 60 wt%) was provided by Ingessil S.r.l., Verona, Italy.

Geopolymer synthesis
Alkaline solution used in this study was prepared by a mixture of 8 M NaOH and the commercial sodium silicate solution in the volume proportion 1:1. For each 100g of laterite, 10, 20, 30 and 40 g of RHA were added. The mix composite was dried, homogenized for 5 min and received the alkaline solution in the proportion solid/liquid of ~ 1-6. Finally the river sand (sand was collected from Sanaga river with particles size between 2.0 -0.075 [AASHTO]; specific gravity and fineness modulus of 2.38) was added with a proportion paste: sand of 1:2 and homogeneous paste was obtained by ball-milling. The pastes were poured into plastic moulds with dimensions 1×1×14 cm 3 at room temperature (23±3 °C; RH 55%) for 3 h, then oven cured at temperatures of 90 °C for 24 h. Curing at room temperature with ambient conditions was also applied for comparison. The geopolymer samples were left for 28 days at ambient temperature prior to three-point flexural strength testing. The geopolymers specimens were labelled as GPEL, GPOD for the as received and ground laterites from Eloumden and Odza. Materials obtained with addition of RHA were labelled: GPEL10, GPEL20, GPEL30, GPEL40, GPOD10, GPOD20, GPOD30 and GPOD40 as in Table 3.

Physico -chemical tests
Three-point flexural strength of the samples was measured with an Instron® 1195 compression machine with a displacement of 5 mm/min. The results shown are an average of five to six replicate specimens. The strength is given by the equation (1): where σ is the maximum center tensile stress (MPa), F maximum load at fracture (N), L the distance between the supports (mm), b the width and h the thickness of the specimen (mm).

ACCEPTED MANUSCRIPT 6
The three-point bending strength tests were performed according to the standard test method for flexural strength of concretes ASTM C78-16.
The water absorption analysis was carry out by immersing the specimen in water at ambient temperature for 24 h and comparing the humid weight (mh) to the dry weight (md) according to equation (2):

Microstructural characterization and phase's analysis of geopolymer composites
The microstructure of the inorganic polymer cement specimens was studied using an Pieces collected from the mechanical test were used to prepare specimens of ~1 cm 3 of volume for the Mercury Intrusion Porosimeter (MIP) (Autopore IV 9500, 33000) tests using psia (228 MPa) MIP covering the pore diameter range from approximately 20-0.001 µm having two low-pressure ports and one high-pressure chamber.
The mineralogical phases of resulting geopolymers were identified via X-ray powder diffractometer, XRD, as described above.
Fourier Transformed Infrared Spectroscopy, FTIR, (VERTEX-70 FTIR, Bruker, Ettlingen, Germany) was performed on selected sample analyzing fine powder of ground specimens (ф ≤ 80 µm) collected from pieces of the mechanical test cured at 28 days. A minimum of 32 scans between 4000 and 500 cm -1 were averaged for each spectrum at intervals of 1 cm -1 .

Effect of amorphous silica (RHA)
The geopolymerization products of laterite (35% FeO ( Figure   6) showed that, at this level, the iron phases are distributed homogeneously into the matrix.
The surface fractures give the impression that the composite possess a high fraction of amorphous mass as described by Gallup (Gallup 1989 these "vitreous mass" and when the temperature increases (50-200 °C), the iron silicates (ferri and ferrosilicates) content is raised. According to Quong (1976), the iron silicate does not exhibit an X-ray diffraction pattern and therefore it has been considered as an amorphous mass. In general, the amorphous mass has a non-stoichiometric composition with Si/Fe ratio in the range of 0.1-1 (Gallup 1989). The amorphous mass could be ascribed to hydrated silica, with iron being bound to silica through oxygen bridges, like Fe-O-Si structure extended into the matrix of geopolymer composite. This is also confirmed by the large absorption bands observed on FTIR spectra (Figure 7). The micrographs of composites show in the inter pores spaces, a dense and homogeneous interconnected structure similar to that observed for volcanic scoria (Polacci et al. 2008). These observations prove that the present Fe atoms in iron silicate are chemically bounded to silica and cannot consist of a simple mixture of silica and iron oxide within the matrix of geopolymer. Gallup (1989)

Effect of temperature and residual humidity
The three-point flexural strength of geopolymer composite cured at room temperature are in the range of 10-12 MPa, with water absorption between 10-13% (Figures 8 and 9). The to Greenwood and Gibbs (1971), at relatively high temperature (200 °C), iron silicate is present with iron in the ferric (+3) oxidation state with only trace of ferrous (+2) iron. The low temperature (100 °C) is shown to consist of both ferric (46%) and ferrous iron (54%).
In both materials the mineral phases such as lepidocrocite, ilmenite, goethite and hematite are potential mineral sources of ferric and ferrous irons. Ferric irons were detected in the iron-rich This scale-forming reaction is believed to occur quite rapidly because ferric iron is known to have a strong affinity for silica (Gallup 1989). In the same time a fraction of kaolinite in raw laterite, under oven curing, is converted to hydrosodalite or sodalite. The kaolinite present in higly corroded laterites is amorphous or metastable (disordered) making it easier to be transformed into polysialates or ferrosialates in alkaline medium (Equation 2).
Si 2 O 5 , Al 2 (OH) 4 + NaOH → Na(Si-O-Al-O)n (2) The action of the temperature affects the porosity and pore-size distribution while the residual humidity is significant for the effectiveness of the dissolution and polycondensation.

Phases evolution
The infrared spectra of resultant products are presented in vibrations (Farrell, 1972;Sidhu, 1988). Those around 1300-1450 cm -1 are linked to stretching C-O bonds in all spectra due to the formation of sodium/iron carbonate from reaction of residual alkali metal with CO 2 . The last one with lower frequency band at 3240-3280 cm -1 appear to be an overtone of the vibration of the silicate framework as already pointed by White and Keester (1965). The bands of Si−O−T (T = Al, Fe or Si) which were in the range of 1015-980 cm −1 in previous work (Kaze et al. 2017) are shifted to the high values of wave numbers about 1020 and 1030 cm -1 with increase of reactive silica from rice husk ash. This behavior is due to the formation of (Si, Fe and Al) -rich gel. Similar trend was reported by Rees et al. (2007) and Kamseu et al. (2017), which attributed the band positions greater than 995 cm -1 to a very high Si content. The high values of wavenumbers also described the presence of fayalite mineral as reported by Hofmeister (1987).

10
The XRD patterns of raw laterites (LATOD and LATEL) and selected geopolymer composites samples (GPEL20; GPOD20; GPEL40 and GPOD40) treated at about 90 °C are shown in Figures 8a-b. The mineral phases identified into the geopolymer composites are: hematite, quartz, maghemite, anatase, ilmenite, which are also present in raw laterites, suggesting that they were not involved in geopolymerization reaction. It should be noted the decrease of kaolinite reflections after geopolymerization: the decrease confirms the poorly crystallized nature of the kaolinite, the principal mineral of laterite (Kaze et al. 2017). The phase's formation that includes zeolite, hydrosodalite and iron silicate minerals (fayalite, and others) has evolution depending on the curing cycle adopted. At room temperature the products of geopolymerization of laterite is essentially hydrosodalite, while the curing cycle including the treatment between 60-100 °C and controlled humidity produces hydrosodalite, zeolite and fayalite. Peaks characteristics of zeolite A, most notably at ~ 7-8° (2θ), are observed in geopolymer diffractograms particularly most evident when the curing includes a treatment above 50 °C. Sore et al. (2016) studied the mixture of metakaolinite with rice husk ash and collected similar results when considering the geopolymerization: in fact while the reaction at room temperature gave essentially amorphous phases (geopolymer gel) they found that the curing between 60-90 °C promoted zeolite as the predominant crystalline phase.

11
The three-point flexural strength increases with the rice ash husk content and curing cycles as summarized in Figure 9. The geopolymer products of laterite without addition gave an average of ~4MPa (Kaze et al. 2017). The addition of 10, 20, 30 and 40 wt% gave respectively 6, 10, 24 and 39 MPa. This behavior is in line with the increase of chemical bonds due to the formation of new phases. According to Obonyo et al. (2014) the strength is function of the amount of geopolymeric gel developed capable to embed the matrix.

Porosity, Pore Size Distribution and Water Absorption
The values of water absorption of geopolymer specimens cured at room temperature and at 90 °C are given in Figure 10. It is seen that the values of water absorption decrease from 14.03 to 8.37% and 16.03 to 9.76% at room temperature respectively for GPOD and GPOL ( Figure   10a). When the curing implies treatment at 90 °C, the values of water absorption decrease from 15.89 to 7.40% and 15.03 to 7.89% (Figure 10b). This decrease of water absorption is due to the action of RHA, which favors the connectivity and polycondensation of raw materials, then improves the cohesion between particles and leads to smaller pores occurred into the matrix network. These values are relatively low in comparison with those obtained from metakaolin based geopolymers (Elimbi et al. 2014).

12
The results of MIP analysis of selected geopolymers (GPEL40 and GPOD40) samples are shown in Figure 11. The values of the cumulative pore volume are 92.58 and 120.21 mm 3 /g, respectively for specimens GPEL40 and GPOD40 (Figure 11a). The larger pore size was observed to be 9 and 60 m, respectively for GPEL40 and GPOD40. The high value of pore volume observed on GPOD40 samples is linked to the presence of larger porosity with respect to GPEL40. This could be explained by a detailed analysis of ESEM micrographs collected for the different samples at higher magnification (Figures 3-6). These results are in accordance with the trend of mechanical properties. From the difference mentioned above, it is seen that the geopolymers GPEL40 exhibited a more compact structure than GPOD40, which allows a low value of water absorption due to smaller pores radius after an immersion in water. From Figure 11b, the large pores that are randomly dispersed in both geopolymer matrices, and in some cases larger than the range of measurement of the mercury intrusion porosimeter as reported by , are linked to bubbles formed during the iron corrosion in alkaline media and water evaporation. These larger pores were not taken in account from MIP evaluation. Then pores analyzed using MIP in this current work are those from the bulk interpores spaces as observed and described with ESEM.

Discussion
The action of RHA in laterites based geopolymer composites yields better reactivity and more polycondensation together with the formation of more dense interpores spaces that in this case substitute the matrix. The laterites based geopolymer composites with high volume of RHA evidenced a fundamental difference for the standard curing at room temperature and the ACCEPTED MANUSCRIPT 13 curing at temperatures around 90 °C. While the curing at room temperature gave the flexural strength that remained between 8 and 12 MPa, as generally obtained with sand or fine aggregates added metakaolin based geopolymer composites , the curing around 90 °C resulted in a more complex microstructure in which geopolymer gel and iron silicates cover completely the fine aggregates with homogeneous distributed micropores (5-53µm). The combined action of geopolymer gel (polysialates and ferrosialates) and the iron silicates produced a particular dense interpores space that justifies the high flexural strength obtained: 30-40 MPa. The cumulative pore volume of the composites with the highest flexural strength was 93 and 120 mm 3 /g (respectively for GPEL40 and GPOD40) relatively low with respect to the values of cumulative pore volume of silica-rich geopolymer composites Kamseu et al. 2014;Kamseu et al. 2016) This could be considered just as the fraction of nanometric and fines micrometric pores since the final products of laterites based geopolymer composites present a porous nature.
Appropriate investigations including the optimal microscope might permit to evaluate and quantify those of porosity that cannot, in this case, being investigate with the Mercury Intrusion porosimetry (MIP) . However there was no evident correlation between the pore volume and the mechanical strength due to the formation of good compact interpores spaces as from the polysialates, ferrosialates and iron silicates interconnection and densification.
It is important to note that the presence of amorphous/reactive silica from RHA firstly reequilibrates the Si/Al and Si/Fe ratios in the laterites based geopolymer composites. In the context of optimum activation all these elements are dissolved into the alkaline solution and participate actively to the formation of polysialates/ferrosialates and iron silicates. The curing in a controlled humidity gave optimum reactive conditions for the better densification and strengthening mechanism. The presence of iron mineral was confirmed by FTIR spectra and XRD analysis, which revealed the insertion of iron atom within 3D network amorphous phase. The characteristics of end products obtained in this study help to validate our approach with objective to produce a high strength geopolymers from the N-A(Fe)-Si

Conclusions
Two laterites with similar degree of laterisation and iron content (~ 35 wt%) used in the present study as solid precursors were mixed with quartz sand, rice husk ash and alkaline solution for the production geopolymer composites under different curing conditions. The laterites based geopolymer composites appear more sustainable, low cost and environmentally-friendly. The solid precursors that do not need pretreatment appear ideal raw materials for the cleaner production of eco-friendly cements and composites.
The following conclusions can be derived: Further investigations will be carried out to strictly control the solid to water ratio in the geopolymeric pastes either during room temperature curing or in 60-100 °C range.

ACCEPTED MANUSCRIPT
 High strength matrices is designed used natural iron-rich aluminosilicates  Polymer matrices need less energy and lower global warming potential  Polysialates, ferrosialates and ferrosilicates are combined into an optimal matrix