Effects of Iron Minerals on the Compressive Strengths and Microstructural Properties of Metakaolin-Based Geopolymer Materials

The current study aims to investigate the influence of iron minerals on the amorphous phase content, compressive strengths and the microstructural properties of the geopolymer materials. Geopolymer materials were prepared by the substitution of metakaolin by 10 and 20 wt.% of each iron mineral sample. Sodium waterglass from rice husk ash was used as a hardener, and metakaolin was used as an aluminosilicate source. The X-ray patterns show that the iron minerals denoted FR and FB are associated with hematite and magnetite, respectively. FY contains goethite together with a significant content of kaolinite and quartz. It is observed in the XRD patterns and FTIR absorption spectra that the additions of hematite, magnetite and goethite remain largely unreacted in the geopolymer binder. The compressive strengths of the related geopolymer composites show some significant variations indicating certain effects for mechanical stability obtained: 10 wt.% replacement of metakaolin by hematite increased the compressive strength from 51.1 to 55.5 MPa, while 20 wt.% hematite caused a decrease to 44.9 MPa. Furthermore, 10 and 20 wt.% replacement with FB revealed decreased values 47.0 and 40.3 MPa, respectively. It was also found that 10 and 20 wt.% of FY caused lower values of 30.9 and 39.1 MPa, respectively. The micrographs of geopolymer materials present some voids and cracks. The denser matrix is related to a superior gel formation producing a better glue between the crystalline additions. The unsubstituted geopolymer sample provides with about 50% the highest X-ray-amorphous content, whereas the substituted samples range between 35 and 45%, indicating systematically smaller gel contents without any clear trend with the compressive strength variation, however. The strength dependencies reveal more complex interaction between the gel and crystalline additions.


Introduction
Iron is found mainly in the Earth's crust in minerals of iron oxide and iron oxyhydroxide such as hematite, magnetite, maghemite, goethite, limonite, etc. Amongst these minerals, hematite (α-Fe 2 O 3 ) and magnetite (Fe 3 O 4 ) are mostly used as an ore for making metal iron. Iron minerals exist to some extent in aluminosilicate sources such as kaolin and more significantly in laterite. Most iron minerals generally present in these aluminosilicate sources are hematite and goethite. The iron mineral content in the aluminosilicate sources depends on the geological formation. Goethite contained in the raw aluminosilicates are transformed to hematite during the calcination process, which is performed at about 700 • C using these sources for the preparation of geopolymers. Several researchers have used hematite and magnetite nanoparticles as additives to increase the hydration process of the Portland cement mortar and concrete. For example, Kani et al. [1] investigated the the compressive strength values measurement were used to monitor some changes in the geopolymer materials as well as the amorphous and crystalline phase contents.

X-ray Patterns
The X-ray patterns of different iron minerals (FR, FB and FY) are depicted in Figure 1, whereas those of rice husk ash (RHA) and metakaolin (MK-MY3) are shown in Figure 2. The X-ray patterns of different iron minerals such as FR (red powder) and FB depict the peaks of hematite (α-Fe2O3) and magnetite (Fe3O4), respectively. The sample FY (yellow powder) consists of goethite or oxyhydroxide (α-FeO(OH)) (26 wt.%), kaolinite (35 wt.%), quartz (33 wt.%) and muscovite (5 wt.%), according to Rietveld phase analysis. This shows that FR, FB and FY are associated with hematite, magnetite and goethite, respectively. Using the TOPAS package, the broadening of the peak profiles reveal average crystal sizes of 120 nm, 80 nm and 40 nm for hematite, magnetite and goethite, respectively (Lorentzian peaks, strain not considered). The oxidation state of iron in the structure of hematite and goethite is iron(III) ,whereas the one in the structure of magnetite is the combination of iron(III) and iron(II). The X-ray pattern of rice husk (RHA) indicates the peaks of cristobalite together with some minor tridymite. Metakaolin (MK-MY3) shows the peaks of illite, quartz, anatase and traces of hematite ( Figure 2). Besides these minerals, both diffractograms show the broad hump structure from 18 to 38° (2θ) for RHA and 15 to 35° (2θ) for metakaolin. This halo diffraction is attributed to the amorphous silica and amorphous aluminosilicate phases in the structure of rice husk ash and metakaolin, respectively.

Infrared Spectra
The infrared spectra of the iron mineral samples (FR, FB and FY) are shown in Figure 3. The infrared spectrum of FR presents three absorption bands at 460, 543 and 3433 cm −1 . According to Schwertmann and Cornell [11], Andrate et al. [12] and Deligöz et al. [13], the bands observed at 460 and 543 cm −1 are ascribed to the vibration modes of the Fe-O characteristics of hematite. The one of FB also shows three bands at 420, 571 cm −1 , and the one with the lowest intensity is at about 3433 cm −1 . The vibration modes of Fe-O were observed at 420 and 571 cm −1 , which are characteristic of magnetite [11,12]. The absorption band at 3433 cm −1 on the infrared spectra of FR, FB and FY is assigned to the OH group stretching modes [14]. The absorption band at 3138 cm −1 presents only on the spectrum of FY is attributed to the Fe-OH stretching modes characteristic of goethite, and the one at 1624 cm −1 is assigned to the bending vibration modes of water molecules [15]. The latter band that appears at 1643 cm −1 on the spectra of metakaolin and rice husk ash indicates the presence adsorbed water. The absorption bands at 3617 and 3690 cm −1 on the spectrum of FY have been ascribed to the inner and external hydroxyl groups of kaolinite [16]. A rather close agreement is observed between the spectrum of sample FY and that of kaolinite. Therefore, the contribution of goethite is hard to see here in the IR-spectra.

Infrared Spectra
The infrared spectra of the iron mineral samples (FR, FB and FY) are shown in Figure 3. The infrared spectrum of FR presents three absorption bands at 460, 543 and 3433 cm −1 . According to Schwertmann and Cornell [11], Andrate et al. [12] and Deligöz et al. [13], the bands observed at 460 and 543 cm −1 are ascribed to the vibration modes of the Fe-O characteristics of hematite. The one of FB also shows three bands at 420, 571 cm −1 , and the one with the lowest intensity is at about 3433 cm −1 . The vibration modes of Fe-O were observed at 420 and 571 cm −1 , which are characteristic of magnetite [11,12]. The absorption band at 3433 cm −1 on the infrared spectra of FR, FB and FY is assigned to the OH group stretching modes [14]. The absorption band at 3138 cm −1 presents only on the spectrum of FY is attributed to the Fe-OH stretching modes characteristic of goethite, and the one at 1624 cm −1 is assigned to the bending vibration modes of water molecules [15]. The latter band that appears at 1643 cm −1 on the spectra of metakaolin and rice husk ash indicates the presence adsorbed water. The absorption bands at 3617 and 3690 cm −1 on the spectrum of FY have been ascribed to the inner and external hydroxyl groups of kaolinite [16]. A rather close agreement is observed between the spectrum of sample FY and that of kaolinite. Therefore, the contribution of goethite is hard to see here in the IR-spectra. Figure 4 shows the spectra of rice husk ash (RHA) and metakaolin (MK-MY3). Those peaks that appear at 464 and 470 cm −1 on the infrared spectra of metakaolin and rice husk ash, respectively, are associated with the vibration modes of Si-O. The one at 539 cm −1 on the infrared spectrum of metakaolin is related to the vibration modes of Si-O-Al and Fe-O bonds. The interpretation of the Fe-O peak here is supported by the X-ray pattern, which shows a trace of hematite in the metakaolin sample. The broad absorption peak with a maximum at 1063 cm −1 is the typical density of state effect of asymmetric Si-O vibrations of calcined kaolinite. The weaker peaks at 693 and 799 cm −1 also appear typically in calcined kaolinite. The spectrum of rice husk ash typically composes features of SiO 2 -glass and cristobalite [17,18]. The main peak at 1098 cm −1 is ascribed to the asymmetric vibration modes of Si-O bonds.  Figure 4 shows the spectra of rice husk ash (RHA) and metakaolin (MK-MY3). Those peaks that appear at 464 and 470 cm −1 on the infrared spectra of metakaolin and rice husk ash, respectively, are associated with the vibration modes of Si-O. The one at 539 cm −1 on the infrared spectrum of metakaolin is related to the vibration modes of Si-O-Al and Fe-O bonds. The interpretation of the Fe-O peak here is supported by the X-ray pattern, which shows a trace of hematite in the metakaolin sample. The broad absorption peak with a maximum at 1063 cm −1 is the typical density of state effect of asymmetric Si-O vibrations of calcined kaolinite. The weaker peaks at 693 and 799 cm −1 also appear typically in calcined kaolinite. The spectrum of rice husk ash typically composes features of SiO2-glass and cristobalite [17,18]. The main peak at 1098 cm −1 is ascribed to the asymmetric vibration modes of Si-O bonds.   Figure 4 shows the spectra of rice husk ash (RHA) and metakaolin (MK-MY3 peaks that appear at 464 and 470 cm −1 on the infrared spectra of metakaolin and r ash, respectively, are associated with the vibration modes of Si-O. The one at 539 the infrared spectrum of metakaolin is related to the vibration modes of Si-O-Al a bonds. The interpretation of the Fe-O peak here is supported by the X-ray patter shows a trace of hematite in the metakaolin sample. The broad absorption peak wit imum at 1063 cm −1 is the typical density of state effect of asymmetric Si-O vibration cined kaolinite. The weaker peaks at 693 and 799 cm −1 also appear typically in calci linite. The spectrum of rice husk ash typically composes features of SiO2-glass and c ite [17,18]. The main peak at 1098 cm −1 is ascribed to the asymmetric vibration mod O bonds.   5-7 depict the X-ray patterns of geopolymer materials from metakaolin-FR (GR10 and GR20), metakaolin-FB (GB10 and GB20) and metakaolin-FY (GY10 and GY20), respectively. The control geopolymer material has been denoted G0, and metakaolin has been substituted by 10 and 20 wt.% of different iron mineral samples. The peaks of illite, quartz, anatase, cristobalite and the residual kaolinite are observed on the X-ray patterns of all geopolymer materials. In addition to these crystalline minerals, the X-ray patterns show the broad hump structure between 18 and 40 • (2θ), attributed to the amorphous aluminosilicate phase or gels in the network of geopolymer materials. It is worth mentioning that this amorphous phase appears between 15 and 38 • (2θ) on the X-ray pattern of metakaolin ( Figure 2). The shift of this broad peak is characteristic of the depolymerization process of metakaolin or iron-metakaolin followed by the polycondensation reaction, entailing the formation of the geopolymer binders. Besides these phases, the X-ray patterns of geopolymer materials from metakaolin-FR and metakaolin-FB ( Figures 5 and 6) indicate rather clearly the peaks of hematite and magnetite, respectively. For the geopolymer samples GY10 and GY20 (Figure 7), the goethite related peaks are observed, albeit with relatively lower intensities. Rietfeld refinements reveal in all cases the same average crystal sizes as obtained for the raw iron minerals.   5-7 depict the X-ray patterns of geopolymer materials from metakaolin-FR (GR10 and GR20), metakaolin-FB (GB10 and GB20) and metakaolin-FY (GY10 and GY20) respectively. The control geopolymer material has been denoted G0, and metakaolin has been substituted by 10 and 20 wt.% of different iron mineral samples. The peaks of illite quartz, anatase, cristobalite and the residual kaolinite are observed on the X-ray patterns of all geopolymer materials. In addition to these crystalline minerals, the X-ray patterns show the broad hump structure between 18 and 40° (2θ), attributed to the amorphous aluminosilicate phase or gels in the network of geopolymer materials. It is worth mentioning that this amorphous phase appears between 15 and 38° (2θ) on the X-ray pattern of metakaolin ( Figure 2). The shift of this broad peak is characteristic of the depolymerization process of metakaolin or iron-metakaolin followed by the polycondensation reaction, entailing the formation of the geopolymer binders. Besides these phases, the X-ray patterns of geopolymer materials from metakaolin-FR and metakaolin-FB ( Figures 5 and 6) indicate rather clearly the peaks of hematite and magnetite, respectively. For the geopolymer samples GY10 and GY20 (Figure 7), the goethite related peaks are observed, albeit with relatively lower intensities. Rietfeld refinements reveal in all cases the same average crystal sizes as obtained for the raw iron minerals.    The crystalline and amorphous phase (gels) contents in the geopolymer materials G0, GR10, GR20, GB10, GB20, GY10 and GY20, as obtained in the first step by integration, are depicted in Table 1. These data show that the X-ray amorphous contribution in G0, GR10, GR20, GB10, GB20, GY10 and GY20 are 49.4, 38.6, 35.4, 43.5, 39.9, 45.1 and 42.0%, respectively. It can be observed that the X-ray amorphous contribution in the geopolymer materials systematically decreases when the iron minerals increase from 0 to 20 wt.%. This is obvious because the iron minerals used for the substitution of metakaolin remain crystalline minerals ( Figure 1). As the liquid-to-solid ratio remains almost constant, this does not lead to higher contents of gel. The reduction in the amorphous phase content when the iron minerals content increases in the system is confirmed by the reduction in the intensity of the broad hump network that appears from 18 to 40 • (2θ) on the X-ray patterns of geopolymer materials ( Figures 5-7). The decrease in the amorphous-phase content seemed to be more pronounced when metakaolin was substituted by 10 and 20 wt.% of hematite (FR). This could be related to the higher degree of crystallinity of hematite or to the formation of the crystalline network in the structure of GR10 and GR20, which is not detected on their X-ray patterns compared to those of GB10, GB20, GY10 and GY20. On the other hand, the integration of the X-ray scattered intensities ignores any element-specific contributions. Therefore, using structural and element specific contributions, as included in the Rietveld method of phase quantification using an internal standard, could reveal more realistic quantifications, which were not further considered here.

Infrared Spectra
The geopolymer material both with and without iron minerals show some changes detected by the Fourier Transform Infrared spectroscopy in Figures 8-10. The infrared spectra of geopolymer materials after iron mineral addition (10 and 20 wt.% of FR, FB and FY) show absorption bands at 529 cm −1 , 571 cm −1 and 527 cm −1 , as denoted in Figures 8-10, respectively. Peaks between 571 and 529 cm −1 can be attributed to vibration modes of Fe-O-Si, as discussed in the literature [19][20][21][22][23][24][25]. According to Essaidi et al. [4], the iron mineral modifies the formation of the geopolymer material, leading to the formation of different specific networks with the binder. It is also clearly observed that the infrared spectrum of the geopolymer material without the addition of iron mineral does just show an absorption wing between 520 cm −1 up to 600 cm −1 , but not a peak, as observed for the 10 and more intensive values for 20% of the samples. This could indicate that iron minerals added to the metakaolin could react with sodium waterglass during the geopolymerization process. However, we may also give a different explanation. A reaction of Fe 3+ from hematite for the formation of the poly(ferro-silico-aluminate) network cannot be easily supported by the present results. It is observed that the samples GR10 and GR20 show an additional peak at 529 cm −1 (Figure 8) compared to G0. This peak can be related to the presence of hematite as having shifted from 543 cm −1 due to superimposition. We checked this by adding up the appropriate spectra of hematite and G0, reproducing the peak position as observed experimentally. This showed us that the observed peak shift cannot be taken as an indication of incorporation of Fe-O units in the geopolymer network. The absorption bands of magnetite and the FY sample at 571 and 527 cm −1 (Figure 3) remain unchanged on the spectra of GB10 and GB20 ( Figure 9) and those of GY10 and GY20  (Figure 3) remain unchanged on the spectra of GB10 and GB20 ( Figure 9) and those of GY10 and GY20 (Figure 10), respectively. This suggests that magnetite and hematite do not participate in the geopolymerization process. For GY10, and more clearly for GY20, there are additional peaks that can be related to the unreacted kaolinite, but there is no clear indication of the effect of goethite observed as for the FY sample ( Figure 3).   (Figure 3) remain unchanged on the spectra of GB10 and GB20 ( Figure 9) and those of GY10 and GY20 (Figure 10), respectively. This suggests that magnetite and hematite do not participate in the geopolymerization process. For GY10, and more clearly for GY20, there are additional peaks that can be related to the unreacted kaolinite, but there is no clear indication of the effect of goethite observed as for the FY sample ( Figure 3).    All geopolymer spectra show silanol groups (Si-OH) identified at around 869 cm −1 . A significant broad band centred at the wavenumber of around 1006 cm −1 on the infrared spectra of geopolymer materials, which corresponds to the density of states of the Si-O asymmetric stretching of the various [SiO4]-units of the geopolymer network. This band appears at the wavenumber of 1063 cm −1 on the spectrum of metakaolin ( Figure 4). It was shifted to a lower wavenumber on the spectra of geopolymer materials owing to the depolymerization of metakaolin in the presence of sodium waterglass. Due to the consumption of NaOH from the waterglass, poly-siloxo chains are formed, which become crosslinked via sialate and siloxo units in the following. A significant incorporation of Fe in the network may also be ruled out here due to rather precisely the same position and shape in the density of Si-O states for the G0-sample and the G10 and G20 samples.
The absorption bands with lower intensity at 3617 and 3690 cm −1 on the infrared spectrum of geopolymer material after 20 wt.%, GY20 and somewhat weaker for GY10 ( Figure 10) are related to kaolinite in the FY sample, which remains unreacted during geopolymerization. The quartz is identified at 682-691 cm −1 and 782-788 cm −1 on the infrared spectra of geopolymer materials. These bands were also attributed to the vibration modes of Si-O-Fe and Fe-O-Al, respectively [19][20][21][22][23][24][25][26][27][28], which cannot be supported by the present results.

Micrography Image Investigations
The morphologies of geopolymer materials G0, GR10, GR20, GB10, GB20, GY10 and GY20 examined at magnifications of 100×, 1000× and 5000× are shown in Figures 11-13. The micrographs of the prepared geopolymer materials show a relatively compact consistency, indicating that the crystalline phases present in the geopolymer materials are embedded in their networks. This could affect the compressive strength of the geopolymer materials, as discussed below. Some samples observed at a magnification of 100× present some voids and air bubbles, which could lead to substantially smaller compressive strengths of the geopolymer materials. The cracks appear on the micrographs observed at a magnification of 1000×  (Figure 4). It was shifted to a lower wavenumber on the spectra of geopolymer materials owing to the depolymerization of metakaolin in the presence of sodium waterglass. Due to the consumption of NaOH from the waterglass, poly-siloxo chains are formed, which become crosslinked via sialate and siloxo units in the following. A significant incorporation of Fe in the network may also be ruled out here due to rather precisely the same position and shape in the density of Si-O states for the G0-sample and the G10 and G20 samples.
The absorption bands with lower intensity at 3617 and 3690 cm −1 on the infrared spectrum of geopolymer material after 20 wt.%, GY20 and somewhat weaker for GY10 ( Figure 10) are related to kaolinite in the FY sample, which remains unreacted during geopolymerization. The quartz is identified at 682-691 cm −1 and 782-788 cm −1 on the infrared spectra of geopolymer materials. These bands were also attributed to the vibration modes of Si-O-Fe and Fe-O-Al, respectively [19][20][21][22][23][24][25][26][27][28], which cannot be supported by the present results.

Micrography Image Investigations
The morphologies of geopolymer materials G0, GR10, GR20, GB10, GB20, GY10 and GY20 examined at magnifications of 100×, 1000× and 5000× are shown in Figures 11-13. The micrographs of the prepared geopolymer materials show a relatively compact consistency, indicating that the crystalline phases present in the geopolymer materials are embedded in their networks. This could affect the compressive strength of the geopolymer materials, as discussed below. Some samples observed at a magnification of 100× present some voids and air bubbles, which could lead to substantially smaller compressive strengths of the geopolymer materials. The cracks appear on the micrographs observed at a magnification of 1000× and 5000×. These cracks generally appeared at the time the compressive strengths were measured. Although most regions of the geopolymer materials GY10 and GY20 observed at a magnification of 5000× are a denser matrix, their micrograph images show some sheet shapes associated with the kaolinite also observed in the X-ray patterns of these geopolymer materials (Figure 7). The presence of the kaolinite in the samples of GY10 and GY20 could limit the binding system and therefore decrease their compressive strength values, as discussed below. and 5000×. These cracks generally appeared at the time the compressive strengths were measured. Although most regions of the geopolymer materials GY10 and GY20 observed at a magnification of 5000× are a denser matrix, their micrograph images show some sheet shapes associated with the kaolinite also observed in the X-ray patterns of these geopolymer materials (Figure 7). The presence of the kaolinite in the samples of GY10 and GY20 could limit the binding system and therefore decrease their compressive strength values, as discussed below. Figure 11. Micrograph images of geopolymer materials without (G0) and with the addition of hematite (GR10, GR20) with magnifications of x100, x1000, x 5000 (from left to right).     Typical results of the EDS analysis of the geopolymer materials G0, GR10 and GR20 are given in Figure 14. The spectra of G0 indicate that this geopolymer material is mainly composed of Si, Al and Na, which is ascribed to the geopolymer gel. For the geopolymer materials GR10, GR20, the addition of hematite shows increased Fe peak intensities. However, it is also possible to find areas without Fe, with increased Si due to quartz content and increased Fe due to the presence of hematite. For example, it is shown in Figure 14d for the sample GR20 with higher magnification and focused on an area of higher Fe content. Similar results were obtained also for samples GB10, GB20 and GY10 and GY20. This means that the iron minerals are not solved to a large extent by the waterglass. It shows that the geopolymer materials from the addition of iron minerals are heterogeneous materials. However, a systematic EDX analysis using the same areas at a magnification of 5000× reveal rather reliable Fe contents, which show similarly increased values for GR10 to GR20, GB10 to GB20 and GY10 to GY20 ( Table 2). For GY10 and GY20, the Fe contents are smaller due to the lower content of Goethite in the raw material added. Typical results of the EDS analysis of the geopolymer materials G0, GR10 and GR20 are given in Figure 14. The spectra of G0 indicate that this geopolymer material is mainly composed of Si, Al and Na, which is ascribed to the geopolymer gel. For the geopolymer materials GR10, GR20, the addition of hematite shows increased Fe peak intensities. However, it is also possible to find areas without Fe, with increased Si due to quartz content and increased Fe due to the presence of hematite. For example, it is shown in Figure 14d for the sample GR20 with higher magnification and focused on an area of higher Fe content. Similar results were obtained also for samples GB10, GB20 and GY10 and GY20. This means that the iron minerals are not solved to a large extent by the waterglass. It shows that the geopolymer materials from the addition of iron minerals are heterogeneous materials. However, a systematic EDX analysis using the same areas at a magnification of 5000× reveal rather reliable Fe contents, which show similarly increased values for GR10 to GR20, GB10 to GB20 and GY10 to GY20 ( Table 2). For GY10 and GY20, the Fe contents are smaller due to the lower content of Goethite in the raw material added.  Table 3. (d) EDS spectrum of GR20 taken at a magnification of 100× and focused on a smaller area with a higher iron content, suggesting a localization of hematite.   Table 3. (d) EDS spectrum of GR20 taken at a magnification of 100× and focused on a smaller area with a higher iron content, suggesting a localization of hematite.

Compressive Strengths
The influence of different iron minerals on the compressive strengths of the metakaolin-based geopolymer materials is visualized in Figure 15. This figure shows that compressive strengths of the geopolymer materials G0, GR10, GR20, GB10, GB20, GY10 and GY20 are 51. 11  It can be observed that the compressive strengths of geopolymer materials from the substitution of metakaolin by 0, 10 and 20 wt.% of FR (hematite) increase moderately from 51.11 to 55.53 MPa, followed by a decrease from 55.53 to 44.86 MPa. Those containing 0, 10 and 20 wt.% of FB (magnetite) decreased from 51.11 to 40.31 MPa with increasing iron mineral content. On the other hand, the compressive strengths of the geopolymer materials containing 10 wt.% of goethite decreased from 55.11 to 30.85 MPa and increased from 30.85 to 39.12 MPa when 20 wt.% of goethite was added to the system. The lower compressive-strength values of geopolymer materials containing goethite could be ascribed to the presence of unreacted kaolinite in their structure, which inhibits the geopolymerization process. The presence of kaolinite is supported by the XRD and SEM results of GY10 and GY20 (Figures 7 and 13). It can be seen that the compressive strengths of geopolymer materials containing hematite and magnetite are higher compared to the ones containing goethite. SEM results showed, for hematite-and magnetite-containing geopolymers, a more compact and denser matrix compared to the samples with goethite added, which supports the better mechanical properties. This finding also agrees to some extent with Davidovits and Davidovits [10], who reported that iron oxides such as hematite and magnetite are more favourable for geopolymer materials synthesis than that of goethite. However, in our case, the compressive strengths of geopolymer materials containing magnetite decrease as well. Only when metakaolin was replaced by 10 wt.% of hematite was the binder reduced in the system, but the compressive strengths increase. The increase in the compressive strengths could be related to the fact that the particles of hematite do not react during the geopolymerization reaction but may promote a better geopolymerization It can be observed that the compressive strengths of geopolymer materials from the substitution of metakaolin by 0, 10 and 20 wt.% of FR (hematite) increase moderately from 51.11 to 55.53 MPa, followed by a decrease from 55.53 to 44.86 MPa. Those containing 0, 10 and 20 wt.% of FB (magnetite) decreased from 51.11 to 40.31 MPa with increasing iron mineral content. On the other hand, the compressive strengths of the geopolymer materials containing 10 wt.% of goethite decreased from 55.11 to 30.85 MPa and increased from 30.85 to 39.12 MPa when 20 wt.% of goethite was added to the system. The lower compressive-strength values of geopolymer materials containing goethite could be ascribed to the presence of unreacted kaolinite in their structure, which inhibits the geopolymerization process. The presence of kaolinite is supported by the XRD and SEM results of GY10 and GY20 (Figures 7 and 13). It can be seen that the compressive strengths of geopolymer materials containing hematite and magnetite are higher compared to the ones containing goethite. SEM results showed, for hematite-and magnetite-containing geopolymers, a more compact and denser matrix compared to the samples with goethite added, which supports the better mechanical properties. This finding also agrees to some extent with Davidovits and Davidovits [10], who reported that iron oxides such as hematite and magnetite are more favourable for geopolymer materials synthesis than that of goethite. However, in our case, the compressive strengths of geopolymer materials containing magnetite decrease as well. Only when metakaolin was replaced by 10 wt.% of hematite was the binder reduced in the system, but the compressive strengths increase. The increase in the compressive strengths could be related to the fact that the particles of hematite do not react during the geopolymerization reaction but may promote a better geopolymerization process into a compact, integrated morphology. This could hinder the crack propagation, similarly to a toughening mechanism proposed by Ghazanlou et al. [3] for the gel of cement-micro/nano Fe 3 O 4 composites. Our results are contrary to the work of Choi and Lee [29] and Essaidi et al. [4], who reported that the addition of 5 wt.% of hematite inhibits better mechanical properties owing to the interaction between the specific network and the binder. On the other hand, the work here confirms the results obtained by Davidovits and Davidovits [10], insofar as metakaolin containing hematite to some extent causes higher compressive strengths compared to metakaolin-based geopolymers without any hematite. According to Zailani et al. [5], an iron silicate binder could be formed by a chemical reaction between silicate and iron(III) in addition to the geopolymer networks. Meanwhile, Kaze et al. [23] and Davidovits and Davidovits [10] identified the ferro-sialate in the structure of geopolymer materials and reported that this compound leads to an increase in the compressive strength and also contributes to the densification of geopolymer networks. To our understanding, the presence of hematite to a limited extent improves the gel-formation and interaction between crystal additions as a glue and therefore increases the compressive strength of the geopolymer materials. This assertion has been supported by the amorphous and crystalline phases content reported in Table 1. It appears in this table that GR10 has only 38.6 wt.% of amorphous phase content and 61.4 wt.% of crystalline phase content, whereas GB10 and GY10 have 43.5 and 45.1 wt.% of amorphous phase content, respectively. This could indicate that the particles of hematite that do not participate in the geopolymerization process act as a filler and reinforce the microstructure of the geopolymer materials despite their low amorphous-phase content. The higher compressive strength value of GR10 corroborates the findings of Hawa et al. [30], who reported that the higher crystallinity content entails a higher compressive strength. Alvarez-Ayuso et al. [31] and Al Kakni et al. [32] concluded that fly-ash-based geopolymers with increased crystallinity exhibited increased compressive strength. The reduction in the strength after the incorporation of 20 wt.% of hematite could be ascribed to the excessive hematite in the system, which prevents the geopolymerization reaction. The compressive strength values of geopolymer materials containing 10 and 20 wt.% of magnetite (GB10 and GB20) and FY-sample (GY10 and GY20) decrease despite their higher amorphous phase content (Table 1). This could be related to the fact that the incorporation of the magnetite or FY-composition in the geopolymer structures hinders the geopolymerization process. Nevertheless, the compressive strength values of the geopolymer materials containing FB and FY remain acceptable ( Figure 15). Hematite (α-Fe 2 O 3 ) and goethite (α-FeO (OH)) contain Fe 3+ , whereas magnetite (Fe 3 O 4 ) contains Fe 2+ and Fe 3+ . Dan et al. [33] and Lemougna et al. [34] concluded that the Fe 3+ react with the geopolymer binder. A fine-dosed substitution of Al 3+ by Fe 3+ during the polycondensation reaction may increase the number of nucleation sites and therefore increases the compressive strength of the geopolymer materials from 51.11 to 55.53 MPa when 10 wt.% of hematite is added to the geopolymer materials. The decrease in the compressive strength of geopolymer materials when magnetite was used as an additive could be ascribed to the fact that Fe 2+ of magnetite did not participate in the geopolymerization process [34] and Fe 3+ ions in magnetite were affected by Fe 2+ and cannot participate in the reaction [28]. The decrease in the compressive strengths from 51.11 to 30.85 MPa of the geopolymer materials when the sample FY is added to 10 wt.% could well be related to the combined effect of adding uncalcined kaolinite, goethite, quartz and muscovite. The kaolinite cannot be activated by the solution and may weaken the composite. Otherwise, the increase in the compressive strength from 30.85 to 39.12 MPa with 10 wt% to 20 wt.% of goethite is in agreement with the findings of Liao et al. [35], Heikal and Ibrahim, [36] and Saleghi-Nik et al. [37], who reported that the inclusion of the nano clay in the structure of cementitious materials positively influences the mechanical properties.

Conclusions
Three iron minerals were used as additives to investigate their influences on the compressive strengths, the amorphous or crystalline phases content and the microstructural properties of metakaolin-based geopolymer materials. The X-ray patterns and the infrared spectra of red, black and yellow minerals showed that they are hematite (α- replaced by 0, 10 and 20 wt.% of each additive, and the sodium waterglass from rice husk ash was used as a hardener. The compressive strength of the geopolymer material without additive was around 51 MPa. There was an increase from 51 to 55 MPa and a decrease from 55 to 45 MPa when metakaolin was substituted by 10 and 20 wt% of hematite, respectively. The crystalline phase content is higher in the structure of geopolymer materials using hematite as an additive compared to the ones using the magnetite and FY sample. The X-ray patterns of geopolymer materials indicate that the intensity of the broad hump structure that appears between 18 and 40 • (2θ) decreases with increasing iron mineral content. The micrograph images of geopolymer materials show that the crystalline phases are embedded in the matrix. It is concluded that amongst the iron minerals used in this work, hematite of sizes in the range of 120 nm with contents of about 10 wt% could be used to improve the compressive strength of the geopolymer materials. Intensified investigations including the crystal size dependencies of the mineral additions similar to that carried out for cement-micro/nano Fe 3 O 4 composites by [3] could be worthwhile. It is without any doubt that calcined iron oxide containing clays and as-prepared sodium silicate solutions using rice ask ash are suitable to displace the use of ordinary Portland cement for any usual building construction.

Materials
The kaolin used in this work was denoted MY3. It was collected from the locality of Mayouom in the West region of Cameroon. This kaolin has been studied and described by Njoya et al. [38]. They reported that this aluminosilicate source is mainly constituted of kaolinite associated with a minor content of illite, quartz, anatase and hematite. The raw kaolin, once harvested, was broken into small sizes using a hammer. The obtained material was dried in air for 24 h and pulverized in the ball mill (MGS, Srl) for 30 min. The powder of kaolin was calcined at 700 • C using a heating and cooling rate of 5 • C/min in the programmable electric furnace for 4 h. The obtained metakaolin has already been described by Tchakouté et al. [39,40] and is denoted as MK-MY3. Three iron minerals with different colours such as red, yellow and black, called FR, FY and FB, respectively, were purchased by the Serraciments company located in Barcelona, Spain. The company denoted UNVDA (Upper Nyoung Valley Development Association) situated in Ndop in the North-West region of Cameroon provided white rice husk ash, which is already calcined in an open air. This white silica source was ground in the same condition as kaolin and the powder obtained was denoted RHA. The chemical compositions of the kaolin and rice husk ash are presented in Table 3. Rice husk ash has already been used and is described by Melele et al. [41], Mabah et al. [42] and Tchakouté et al. [40]. NaOH pellet was provided by the laboratory-grade granules (96 wt.%, Sigma Aldrich, Milano, Italy).

Preparation of Sodium Waterglass from Rice Husk Ash and Geopolymer Materials
Sodium waterglass containing the molar ratio SiO 2 /Na 2 O equal to 1.6 has been prepared using rice husk ash as the silica source and sodium hydroxide solution as the Na 2 O source. Sodium hydroxide solution with a molar concentration of 5 M was obtained by dissolving a sodium hydroxide pellet in distilled water. The prepared sodium hydroxide solution was mixed with rice husk ash (100 g), and the mixture was heated for about 1 h at 100 • C using a magnetic stirrer to obtain a solution called sodium waterglass.
For the preparation of geopolymer materials, metakaolin was firstly substituted by 0, 10 and 20 wt.% of each iron oxide in order to have iron-metakaolin. Each powder obtained was mixed with the prepared sodium waterglass. The mass ratio of the liquid/solid of the geopolymer material without iron mineral addition and those containing iron minerals were 0.97 and 1.02, respectively. The obtained mixture was blended for about 5 min, and the obtained pastes of each composition were moulded in the cube moulds with dimensions of 40 mm × 40 mm × 40 mm. The moulded samples were cured for 24 h at the ambient temperature of the laboratory before demoulding. After demoulding, the compositions were sealed in the plastics and left in the laboratory for 28 days before their compressive strength measurements. The control geopolymer materials (without iron mineral addition) are denoted G0. The geopolymer materials from the incorporation of 10 and 20 wt.% of red iron mineral are named GR10 and GR20, respectively. Those from the inclusion of 10 and 20 wt.% of black iron mineral are designated as GB10 and GB20, respectively, whereas the ones of yellow iron mineral are called GY10 and GY20, respectively.

Methods of Characterization of Raw Materials and Metakaolin-Based Geopolymer Materials
The influence of the red, yellow and black iron minerals on the properties of geopolymer gels was monitored by measuring their compressive strengths and investigating their microstructural properties (infrared spectroscopy analysis, the powder X-ray diffractometry and the scanning electron microscope).
The compressive strength values of the geopolymer materials were determined after 28 days on the samples sealed in the plastics and maintained at room temperature in the laboratory according to the DIN 1164 standard. A loading rate was kept constant at 0.500 MPa/s. They were measured using an automatic hydraulic press with a 250 kN capacity (Impact Test Equipment Limited, UK KA20 3LR, Stevenston, United Kingdom).
After the compressive strengths were tested, the fragments of all samples were collected, and one part was finely crushed in the porcelain mortar. The obtained powders were used to record the X-ray diffractograms and infrared spectra. Others fragments were used to observe the morphologies of the samples.
The X-ray diffractograms were registered on the Bruker D8 (Karlruhe, Germany) Advance equipped with LynXeye XE T detector detecting CuKα1,2 in Bragg Brentano geometry. The 2θ range is from 5 to 80 • for 1 h with 0.01 2θ steps. The crystalline phases present in the structure of different samples were identified using X'Pert HighScore Plus software (version 3.0.0.123). The crystalline and amorphous phase contents were estimated by correcting for the experimental background obtaining the total intensity, S(tot) and then integrating only the crystalline part, S(Cryst) and calculating S(amorphous) = S(tot) − S(cryst). For this purpose, Opus software was used. Phase contents of the sample FY were refined using the Rietveld method (Topas 6.0, Bruker). Average crystals sizes were obtained using refinements of the raw iron minerals and these phases obtained in the geopolymer.
The infrared spectra were measured by FTIR (Bruker Vertex 80 v, Bremen, Germany) using the KBr method. In this method, about 200 mg of KBr was mixed with around 1 mg of the sample, and the whole was blended in an agate mortar. The mixture was pressed at 100 kN using a hydraulic press (ENERPAC P392, San Francisco, CA, USA) in order to obtain the pellets. Each pellet was used to record a spectrum with the resolution of 2 cm −1 and 16 scans, and the data were collected using OPUS software.