Preparation Conditions for the Synthesis of Alkali-Activated Binders Using Tungsten Mining Waste

The raw materials used in this investigation consisted of TMW, WG, sodium hydroxide (NaOH) (SH), and sodium silicate (${\mathrm{Na}}_2{\mathrm{SiO}}_3$) (SS). The TMW was derived in powder form from the Panasqueira mine in Castelo Branco, Portugal, and the WG was received from the local municipality of Covilhã, Portugal. The chemical composition of the TMW and WG was obtained by scanning electron microscopy and energy-dispersive X-ray (SEM-EDX) analysis (Supra 35VP/EDAX, Oberkochen, Germany). Due to the waste nature and microscopic inhomogeneity of the TMW, chemical analyses were made from different batches collected from the mine. Therefore, the results of TMW reported in Table 1 are the average values accompanied by the standard deviation (SD). According to Table 1, for TMW, the oxides ${\mathrm{SiO}}_2$, ${\mathrm{Al}}_2{\mathrm{O}}_3$, ${\mathrm{Fe}}_2{\mathrm{O}}_3$, and ${\mathrm{SO}}_3$ are the most abundant, while the oxides ${\mathrm{SiO}}_2$, MgO, ${\mathrm{SO}}_3$, and ${\mathrm{Fe}}_2{\mathrm{O}}_3$ showed the greatest variability. Grain size distribution analysis was performed for the precursor materials after mechanical sieving by laser diffraction analysis according to BS ISO 13320 :2009 (ISO 2009). The TMW has a mean particle size of 26 μm, while the waste glass has a mean particle size of 39.68 μm. The WG was intentionally used with a slightly larger mean particle size to reduce the energy consumption during the milling process. The bulk powder densities of TMW and WG were determined using a gas displacement pycnometer (AccuPyc II 1340, Micromeritics, Norcross, Georgia) and were determined as 3.08 and $2.53\mathrm{g}/{\mathrm{cm}}^3$, respectively. Sodium hydroxide solution was prepared by dissolving sodium hydroxide pellets (98% purity obtained from Fisher Scientific, Schwerte, Germany) in deionized water and allowed to cool before use. Sodium silicate (obtained from Solvay SA, Póvoa de Santa Iria, Portugal) had a ${\mathrm{SiO}}_2/{\mathrm{Na}}_2\mathrm{O}=3.23$ (8.60% by weight ${\mathrm{Na}}_2\mathrm{O}$, 27.79% by weight ${\mathrm{SiO}}_2$, 63.19% by weight ${\mathrm{H}}_2\mathrm{O}$, and 0.4% by weight ${\mathrm{Al}}_2{\mathrm{O}}_3$).

All sample preparation was carried out at 20°C. The TMW alkali-activated binders with up to 40% by mass replacement with WG were blended with an IKA (Staufen, Germany) Ultra-Turrax T50 mixer at 360 revolutions per minute (rpm) for 60 s. Based on the research on alkali activation available in the literature concerning mechanical strength and efflorescence formation potential in AABs combined with the experience gained from previous studies (Pacheco-Torgal et al. 2008; Silva et al. 2012b; Kastiukas et al. 2016), the following ranges were selected for the constituents of the AABs:

The following ratios produced an AAB with a flowability of $130\pm 5\mathrm{mm}$ determined using the method proposed by EN 1015-3:1999 (CEN 2006) and initial and final setting times of 90 and 110 min determined by EN 196-3 (CEN 2005). In the precursor:activator ratio, the precursor is the TMW and WG, and the activator is the solution containing the alkali, the silicate, and the water.

To produce the TMW-WG AAB, the TMW and WG were mixed in the dry state for 5 min, forming the precursor materials. The sodium hydroxide and sodium silicate solutions were mixed for a period ranging from 2.5 to 60 min at 700 rpm, depending on the type of condition being tested, forming the alkali activator. The alkali-activator solution was slowly added to the precursor materials, and the resulting paste was stirred for 2.5 min at 200 rpm, followed by 2.5 min at 400 rpm. The resulting AAB was then placed in $40\times 40\times 160\mathrm{mm}$ prismatic polystyrene foam molds. The mold was filled with the AAB in three stages and manually vibrated after each successive filling stage to release trapped air bubbles, and sealed with a film to avoid the loss of water and ingress of ${\mathrm{CO}}_2$. Samples that were made to test curing exposed to the atmosphere were cured unsealed. The specimens were placed in a temperature- and humidity-controlled environmental chamber at 50% relative humidity (RH) for curing between 20 and 80°C for 4–36 h, depending on the type of condition being tested. After curing, prisms were demolded and left in a laboratory condition of 20°C for curing until the test age. The specimens cured at 20°C were demolded after 48 h due to a slow setting. Table 2 summarizes synthesis conditions tested in this study, i.e., activator solution mixing time, curing temperature, and curing time.

To observe for chemical changes in the activator solution during mixing, first it was decided to isolate the activating solution and monitor its temperature during the mixing process. The temperature evolution of the activator solution was measured using the setup shown in Fig. 1. The activator solution was prepared at the same SS:SH solution ratio as that used to make the TMW-WG ABB, i.e., 4.0. A polystyrene enclosure was used to contain the activator solution and provided a thermodynamically stable environment. Two K-type thermocouples with the tips wrapped in temperature-sensitive copper tape were connected to a multichannel data logger and used to measure the temperature of the activator solution and enclosure’s interior, respectively. Once the activator temperature was deemed constant, mixing was started and continued for 20 min at 700 rpm.

The demolded samples were left to rest at 20°C and a relative humidity of 75% until the specific age of testing. The compressive strength of the prismatic sample fractured counterparts was tested after 1, 3, 7, and 28 days in accordance with EN 196-1 using a universal testing machine (Instron 5960, Buckinghamshire, United Kingdom) at a constant loading rate of $144\mathrm{kN}/\min$. The compressive strength value was the average of values obtained from three specimens.

The mineralogical compositions of the TMW and WG were obtained by powder X-ray diffraction (XRD) (Bruker D8 Advance, Karlsruhe, Germany) with an automatic slit, monochromated CuKα radiation ($\lambda =1.5405\mathrm{\AA }$), 5–80° 2θ range, 0.600 s count time, Cu radiation, 40 kV, and 40 mA. Peak shapes were studied using the program DIFFRAC.SUITE.

Fourier transform infrared (FTIR) analyses spectra were recorded from 400 to $\mathrm{4,000}{\mathrm{cm}}^{-1}$ with a $2\text{-}{\mathrm{cm}}^{-1}$ resolution, 5-kHz scanning speed, and 25 scan count using a Shimadzu (Kyoto, Japan) IRAffinity-1 fitted with a Specac (Kent, United Kingdom) Quest attenuated total reflectance (ATR) accessory.

Microstructural studies utilized SEM (Zeiss Supra 35VP, Oberkochen, Germany) equipped with energy-dispersive X-ray (EDX) analyzer (EDAX, Mahwah, New Jersey). Backscattered and secondary electron images were collected from polished specimens to overcome the main limitation of fracture surfaces. To prepare the polished specimens, 5-mm-thick slices were cut using a low-speed saw. The samples were first impregnated with ultralow-viscosity resin and then polished.

Both the XRD and FTIR samples were tested in a state where the alkali activation process was stopped using the combined water and solvent extraction protocol developed by Chen et al. (2014). In summary, it involved stirring the AAB specimen in deionized water and then removing the liquid by centrifuging. Upon addition of methanol, soluble silicate species could be observed in the liquid layer. Thus, water extraction by centrifuging was used to remove the precipitates. Specimens were then ground to micrometer-sized particles using a mortar and pestle, and a solvent of methanol and acetone mixture was added, followed by further grinding. The solvent was removed using vacuum filtration; this latter procedure was repeated five times.

The recorded FTIR spectra of raw TMW and WG and blended as TMW-WG are shown in Fig. 2. For the WG spectra, the highest absorption coefficient is associated with the Si-O bending vibration near $453{\mathrm{cm}}^{-1}$. A weaker band due to the bending mode near $\mathrm{1,000}{\mathrm{cm}}^{-1}$ is accompanied by the still weaker feature near $775{\mathrm{cm}}^{-1}$. For the TMW, the highest absorption coefficient is associated with the bending vibration of the Si-O between 465 and $424{\mathrm{cm}}^{-1}$. Weaker features are associated with the bending vibration of Si-O at $984{\mathrm{cm}}^{-1}$ and its symmetric stretching vibration between 797 and $694{\mathrm{cm}}^{-1}$. The weakest bands at 827 and $\mathrm{1,163}{\mathrm{cm}}^{-1}$ can be associated with the bending of the Si-O bond of the original TMW. The absorbance spectrum of the TMW-WG blend displays the same spectral bands as the raw TMW, only at lower intensities, due to the combination of different intensities.

Results of the XRD analyses shown in Fig. 3 reveal that the TMW precursor material predominantly consists of muscovite and quartz with traces of albite and pyrite and is similar to the TMW chemical composition identified by Pacheco-Torgal et al. (2009a). The WG is revealed to consist of quartz, lime, and sodium oxide with traces of potassium and iron oxide. Using a general nonlinear least squares system software (DIFFRAC.SUITE), the TMW and WG were determined to be 97 and 15% crystalline, respectively. Compared with other materials commonly used as AAB precursors such as fly ash (van Jaarsveld and van Deventer 1999) and metakaolin (Provis et al. 2005), the TMW is of a far less amorphous nature. In this study, a sustainable approach was chosen to increase the amorphicity of the TMW through the addition of WG. The addition of 40% by weight WG led to an increase in the amorphicity, qualitatively indicated by the more intense amorphous background from 15 to 45° in the TMW-WG blend XRD spectrum and also by a 21% calculated reduction in crystallinity. The compressive strength of TMW-WG AAB was used to evaluate the strength contribution potential as a function of the degree of amorphicity. Thus Fig. 4 shows the evolution of compressive strength in TMW-WG AAB with 20, 30, and 40% by weight WG replacement over 28 days. The results obtained for pure TMW AAB are also included. Each reported result corresponds to the average measurement in three specimens per each WG replacement value and age; the deviation of results fluctuated between 0.22 and 1.4%. In Fig. 4 it can be observed that the compressive strength increased with an increase in the WG content at all ages. The highest 28-day strength was obtained by the 60TMW40WG sample at 41 MPa, which is 127% higher than the control sample 100TMW. The compressive strength would be influenced primarily by the additional release of reactive silica. However, it also expected that the CaO content in the WG would contribute to the strengthening of the reaction products, most likely in the form of a (C, N)-A-S-H gel.

The compressive strength results with the highest replacement level of WG, i.e., 40% by weight, is consistent with the compressive strength results previously obtained by Pacheco-Torgal et al. (2009b), specifically 39.6 MPa at 28 days, for mortar prepared with TMW. However, this was achieved only after an energy-intensive calcination treatment of the TMW at 950°C for 2 h.

Finally, the reactive silica-containing WG combined with the highly alkaline activator solution may create the potential for the deleterious process of alkali-silica reaction (ASR) and required validation. The TMW-WG AAB with the highest replacement of WG, i.e., 40% by weight, was stored at a RH of 80% at 38°C to accelerate the ASR reaction; a thin section of this sample is shown in Fig. 5. Observation of the section, which is representative of the WG as a whole, revealed that there were no signs of ASR gel formation around the WG particles or in the open voids. Data reported in the literature established that if the waste glass is ground to be less than 75 μm, the ASR effect does not occur, and binder durability is guaranteed (Shao et al. 2000). Water is also a necessary condition for ASR; considering the TMW-WG AAB only required a water or precursor demand of 0.179, this may also be the reason for the absence of ASR.

To prepare the TMW-WG AAB, it is necessary that the alkali activator is in a homogeneous state upon mixing with the powder precursors. Fig. 6 shows the effect of activator mixing time on the 80TMW20WG AAB compressive strength. It can be seen that the 28-day strength increases when using the activator that has been mixed between 2 and 5 min only (i.e., M2.5 and M5), reaching the maximum 28-day strength of 17 MPa at an activator mixing time of 5 min (i.e., M5). As the activator mixing time is extended, an immediate drop in compressive strength is observed. The activator solution mixing time has a strong effect on the 28-day strength; a 26% drop in compressive strength is recorded for the TMW-WG AAB when prepared using an activator solution stirred for 20 min, i.e., M20. Fig. 7 presents the results of the activator solution temperature during mixing. The dashed and solid curves represent the activator and enclosure air temperature, respectively. Over the course of 20 min of mixing, an average reduction of 3.13°C in activator temperature was recorded from three identical tests, while the enclosure temperature was recorded to remain stable at $23\pm 0.1°\mathrm{C}$. This drop activator solution temperature is an endothermic process resulting from the reorientation of the water molecules, leading to a disruption of the hydration shells surrounding the ions. The positive metal ions, in this case, ${\mathrm{Na}}^{+}$, are particularly at risk because they inherently possess weaker attractions to the negative oxygen end of the water molecule. The prolonged mixing can be thought to cause a net stripping effect of the water molecules from the ions. The latter would effect the dissolution and subsequent mobility of the siliceous material present in TMW and WG, leading to a less intense attack on the silicon-oxygen bonds and thus a reduction in mechanical performance, as verified by the results in Fig. 6.

Further interpretation of this is shown by the ATR-FTIR spectra of the activator mixed for 5 and 20 min in Fig. 8. The $\mathrm{3,270}{\mathrm{cm}}^{-1}$ band (which is a sensitive and well-defined band corresponding to the O-H vibrations in water) is revealed to increase in intensity by 21.4%, which based on literature concerning FTIR spectra of water at different temperature (Praprotnik et al. 2004) may be considered as a significant amount. It is an indication that the activator solution mixed for 20 min possesses a higher unbound water content with fewer solvated ions and more available as free molecules. Also, visually observed after 20 min of mixing was the partial gelation of the soluble silicate anions detected by the loss of uniform fluid flow and the adherence of solid gel to the glass wall. Polymerization of silicates commonly occurs at pH close to neutral, but can also be triggered by an increased water content (Hu et al. 1993). Gelation would also contribute to reducing the effectiveness of the activator to balance the charge of the aluminate groups in the phyllosilicate, resulting in a negative effect on the kinetics of the reaction and therefore the development of mechanical strength. From a practical outlook, it must be emphasized that the preparation of the alkali activator is independent of the mixing of the final binder. Therefore, the alkali activators dependence on mixing time should not be considered to interfere with the upscaling potential of AABs.

Samples T20–T80 in Table 2 were made so that the effect of different curing temperatures on compressive strength of TMW-WG AAB could be studied. The prisms cast were cured under sealed conditions at 20, 40, 60, and 80°C for 24 h, then tested for compressive strength after 1, 3, 7, and 28 days. The compression strength results in Fig. 9 present details on the role of temperature on the properties of TMW-WG AAB. The compressive strength of all the samples at all ages increased with increasing curing temperature. Samples cured at 20°C (i.e., T20) did not develop appreciable compressive strength for the first 7 days of curing and were only able to attain 2.6 MPa after 28 days. The highest compressive strengths were obtained by curing at 80°C (i.e., T80), allowing the TMW-WG AAB to attain 22 MPa at 28 days. From this, it can be concluded that the reaction that took place was a temperature-driven process. van Jaarsveld et al. (2002) and Bakharev (2005) reported comparable compressive strength results for fly ash–based AABs. Curing the samples above 80°C was not attempted due to the sufficient strength gained from curing at 80°C. Higher temperatures would also require a greater energy input, a factor this study wanted to avoid by keeping the production of the binder as little energy intensive as practically possible.

Fig. 10 shows the FTIR spectra of the TMW-WG blend of raw materials and the TMW-WG AAB cured at different temperatures (i.e., T20–T80). The TMW-WG AAB cured at 20°C (i.e., T20), as expected, displays the highest absorbance, which can be inferred as reduced hardening activity and thus slow strength development. The spectra of samples cured at 40 and 60°C (i.e., T40 and T60) match each other closely, indicating that similar molecular structures are present in both samples when curing at the respective temperatures. However, the change in absorbance for the sample cured at 80°C (i.e., T80) is more evident. The latter spectrum shows a great reduction in absorbance and broadening between 850 and $\mathrm{1,100}{\mathrm{cm}}^{-1}$ associated with the Si-O-Si symmetric stretching vibrations for the gel product and is an indication of increased activation. A similar feature can be observed at 775 and $694{\mathrm{cm}}^{-1}$, the region of the spectrum also representing symmetric stretching vibration of the raw material Si-O bonds. The development of a more amorphous gel phase with the increase in curing temperature can be inferred from the spectrum of the TMW-WG AAB sample at 80°C matching that of the raw WG, which is confirmed to be inherently amorphous from the XRD results (Fig. 3). Nonetheless, the presence of asymmetric stretching vibration (Si-O-Si) related to nonsolubilized particles at $\sim \mathrm{1,000}$ and $\sim 450{\mathrm{cm}}^{-1}$ in the TMW-WG AAB indicates that unreacted precursor materials are still present, supporting the same result found in the XRD analysis. Milling the precursor materials to a fine particle size has been shown to improve reactivity and dissolution in an alkali activator solution. In the case of fly ash, Temuujin et al. (2009) showed that vibration milling could reduce the median particle size by more than 50% and improve the compressive strength by 80%. This method of mechanical activation could potentially be used to improve the dissolution properties of TMW further, as long as the additional energy input did not compromise the low-energy potential of the TMW-WG AAB.

Just as important as curing temperature is the AAB heat curing duration. Samples D4–D36 in Table 2 were prepared to study the effect of curing time on the compressive strength of TMW-WG AAB. Samples were cured in sealed conditions for 4, 12, 24, and 36 h at 80°C and tested for compressive strength after 1, 3, 7, and 28 days. Fig. 11 shows that the compressive strength improves with an increase in curing time from 4 to 24 h. The improvement in compressive strength continues at an even higher rate when the curing time increases from 12 to 24 h. However, after 36 h of curing a depreciation of the compressive strength at 36 h (identified by the square on the 1-day curve in Fig. 11), 3 and 7 days is observed, leaving the 28-day strength unchanged. Thus, it can be concluded that the activation reaction that took place was time-dependent. Fig. 12 shows the SEM images of TMW-WG AAB samples cured at 80°C for 24 and 36 h. The microstructure of the sample cured for 24 h consists of close-packed quasi-unreacted WG particles embedded in a continuous matrix of gel products, while the sample cured for 36 h exhibits visible contraction, particularly around the WG particles. Although samples were kept in sealed conditions during curing, it was observed during demolding that some water was still able to evaporate into the surrounding air within the curing bag. Previous work by Mo et al. (2014) suggested that the contraction of AAB samples cured under sealed conditions at 80°C occurs after 7 days. However, the results of this study suggest that contraction can initiate as early as 36 h. It is possible that prolonged exposure to the elevated temperature may have led to the water evaporation rate being greater than that of resaturation, thus triggering the accumulation of internal stresses and subsequent contraction of the AAB matrix.

The absorbance spectra for the raw TMW-WG blend and TMW-WG AAB cured for varying periods of time (i.e., D4–D36) are shown in Fig. 13. A reduction in the absorption intensity of the main bands at $\sim \mathrm{1,000}$, $\sim 775$, and $\sim 446{\mathrm{cm}}^{-1}$ indicates that longer curing times led to the further dissolution of Si-O from the raw materials. Also, the position of Si-O bending vibration peaks shifted from 999 and $463{\mathrm{cm}}^{-1}$ in the raw TMW-WG to 989 and $446{\mathrm{cm}}^{-1}$, respectively, for the TMW-WG AAB specimen cured for 24 h, a type of shift associated with a greater extent of polymerization in aluminosilicates (Sarkar et al. 2015). Also, the absorbance spectrum for the TMW-WG AAB sample cured for 24 h (i.e., D24) displays lower absorbance intensities than the TMW-WG AAB sample cured for 36 h. It can be inferred from this latter result that curing times above 24 h can lead a reduction in Si-O dissolution and complements the mechanical strength results in Fig. 11, which clearly show reductions in compressive strength for samples cured for 36 h. Previous research regarding the curing time of AABs made from fly ash (Li et al. 2013) and metakaolin (Heah et al. 2011) have uncovered similar results. Also, the broadening of the characteristic bands between $\mathrm{1,100}–850{\mathrm{cm}}^{-1}$ and $800–750{\mathrm{cm}}^{-1}$ implies the overlap of more bands with a higher intensity, which in this case is attributed to the asymmetric stretching vibrations of T-O-Si (where T = Si or Al) (Khater 2013).