Vous êtes sur la page 1sur 33

Accepted Manuscript

Sustainable alkali activated materials: precursor and activator derived from

industrial wastes

Ana Fernández-Jiménez, Nuno Cristelo, Tiago Miranda, Ángel Palomo

PII: S0959-6526(17)31319-7

DOI: 10.1016/j.jclepro.2017.06.151

Reference: JCLP 9901

To appear in: Journal of Cleaner Production

Received Date: 02 February 2017

Revised Date: 31 May 2017

Accepted Date: 16 June 2017

Please cite this article as: Ana Fernández-Jiménez, Nuno Cristelo, Tiago Miranda, Ángel Palomo,
Sustainable alkali activated materials: precursor and activator derived from industrial wastes,
Journal of Cleaner Production (2017), doi: 10.1016/j.jclepro.2017.06.151

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form.
Please note that during the production process errors may be discovered which could affect the
content, and all legal disclaimers that apply to the journal pertain.

1 April 2017

3 Sustainable alkali activated materials:

4 precursor and activator derived from
5 industrial wastes

7 Ana Fernández-Jiménez a; Nuno Cristelo b,*;

8 Tiago Miranda c; Ángel Palomo d


11 a Eduardo Torroja Institute for Construction Science (IETcc- C.S.I.C.). E28033-Madrid, Spain
12 E-mail address: anafj@ietcc.csic.es
13 Telephone: + 34 91 302 0440; Telefax: + 34 91 302 0700
15 bCQVR, Department of Engineering, University of Trás-os-Montes e Alto Douro, 5000-
16 801 Vila Real, Portugal
17 E-mail address: ncristel@utad.pt
18 * Corresponding author
20 cISISE, Department of Civil Engineering, University of Minho, Campus de Azurém,
21 4800-058 Guimarães, Portugal
22 E-mail address: tmiranda@civil.uminho.pt
24 d Eduardo Torroja Institute for Construction Science (IETcc- C.S.I.C.). E28033-Madrid, Spain
25 E-mail address: palomo@ietcc.csic.es


28 Abstract
30 The aim of the present paper is to analyse the potential application of a solution used to
31 wash moulds from the aluminium casting industry as an alkali activator, for inclusion in
32 the production of alkaline cement. To evaluate the effectiveness of this cleaning
33 solution (CS), two different by-products / wastes were considered as a precursor: fly ash
34 type F (FA) and glass powder (GP) from the production of optical lenses. An 8 molar
35 sodium hydroxide commercial solution was used to establish threshold values for the
36 CS activator. Several mixtures were prepared and cured for 20h, at 85ºC, at high
37 humidity (≥ 90%). Mechanical behaviour was assessed by uniaxial compression
38 strength tests. The reaction rate was determined by calorimetry, and the resulting
39 reaction products were characterized by X-ray diffraction (XRD), electron microscopy
40 (SEM/EDX) and nuclear magnetic resonance (NMR-MAS). The results show that the
41 cleaning solution has a high potential to be used as an activator in the manufacturing of
42 alkali activated cement, especially due to its tetrahedral aluminium content, which is
43 particularly effective for the activation of low-Al raw materials, since it acts
44 simultaneously as a cheap aluminium source.
46 Keywords: Alkaline Activation; cleaning solution; fly ash; glass powder


47 1. Introduction
49 Environmental concerns are becoming increasingly more significant worldwide. The
50 construction industry, as a major contributor, is no exception, especially with the rapidly
51 growing demand of raw materials, making the development of alternative materials an
52 absolute priority. The biggest environmental impact caused by the industry comes from
53 the production of Ordinary Portland Cement (OPC) (Cristelo et al., 2015; Li et al., 2014,
54 2012; Provis and Deventer, 2014). For instance, the production of a ton of clinker
55 releases something between 0.66 and 0.86 tons of CO2 to the atmosphere (Li et al.,
56 2011; Turner and Collins, 2013). Additionally, OPC production demands high levels of
57 energy consumption, mainly due to the high temperatures needed for clinker production,
58 but also due to the milling of the raw materials and the clinker. Such factors make the
59 OPC industry one of the major contributors to the overall emission of greenhouse gases.
60 This is a growing concern that mobilises a rapidly increasing slice of the research
61 resources associated with the construction industry.
63 Furthermore, the also increasing production of industrial residues and its respective
64 storage is, independently of their classification (which in Europe is ruled by the Council
65 Directive 1999/31/EC, with limit values specified by the Council Decision
66 2003/33/EC), another major environmental concern. With a few exceptions (e.g. fly ash
67 incorporation in concrete) these waste materials have low commercial value, due to the
68 lack of possible applications, and in addition have strict requirements for its suitable
69 disposal, which creates a financial burden. Therefore, suitable solutions must be found
70 to reintroduce a significant proportion of these residues in the production chain, thus
71 decreasing environmental and financial costs and transforming these by-products in
72 materials as technically efficient as the current available commercial products (in some
73 cases, and taking advantage of some specific properties, these new materials can even
74 present improved technical performances than traditional materials).
76 The construction industry presents very interesting possibilities to apply, in a direct or
77 indirect approach, very significant percentages of several types of by-products and/or
78 waste. Some of the best examples include fly ash (Deschner et al., 2012; Donatello et
79 al., 2014, 2013; Lothenbach et al., 2011), construction and demolition waste (Arulrajah
80 et al., 2013; Avirneni et al., 2016; Hoy et al., 2016a, 2016b; Mohammadinia et al.,


81 2016) and blast furnace slags (Das et al., 2007; Maghool et al., 2016; Tsakiridis et al.,
82 2008), which have been used as a cement or aggregate replacement, embankment fill
83 and road and railway pavement foundation. It is believed that there is room for further
84 use of increased volumes of these residues and, also, to introduce new types of
85 industrial by-products in the production chain. Furthermore, the alkaline activation
86 technique constitutes a very promising vehicle to help establish, in a more consistent
87 and convincing manner, the paradigm of building with recycle materials.
89 Alkaline activation requires two main components, namely the solid precursor, rich in
90 silica and alumina (with the highest possible amorphisation degree); and the activator
91 solution, based on an alkali metal (usually sodium or potassium) or an alkali earth metal
92 (usually calcium). It constitutes an excellent solution to enhance the properties of most
93 industrial residues. The target of the ongoing research dealing with the application of
94 this technique to different engineering applications is twofold: maximise the mechanical
95 behaviour and to minimise the environmental impact of the final products. Regarding
96 the former, several studies have already proved the suitability of alkali activated
97 materials to substitute OPC, in a wide variety of applications (Fernández Jiménez et al.,
98 2013; Martin et al., 2015; Palomo et al., 2007; van Deventer et al., 2007). The second
99 target is at least as important as the first, since the original materials are, in essence,
100 wastes, with a high polluting capacity, which justifies the large number of related
101 research projects (Al-Zboon et al., 2011; Antunes Boca Santa et al., 2013; Kamseu et
102 al., 2016; Noor-ul-Amin et al., 2015; Pacheco-Torgal et al., 2008; Palomo et al., 2014;
103 Perná and Hanzlíček, 2014; Shi and Fernández-Jiménez, 2006; Zhuang et al., 2015).
104 Nevertheless, while the role of the precursor is commonly assumed by a residue, usually
105 fly ash (Cristelo et al., 2012; Fernández-Jiménez et al., 2006, 2005; Rios et al., 2016) or
106 blast furnace slag (Fernández-Jiménez et al., 2003; Fernández-Jiménez and Puertas,
107 2003; Haha et al., 2012; Ismail et al., 2014), with significant economic and
108 environmental benefits, the activator is usually prepared with first grade commercial
109 reagents (sodium or potassium hydroxide and sodium or potassium silicate are the most
110 common options), which significantly increases the financial costs and severely dilutes
111 the environmental initial advantages, due to the high CO2 release during the production
112 of the reagents (Cristelo et al., 2015). This severely hinders a possible wider spreading
113 of the technique, thus creating the need to develop low cost activators, based partially or
114 solely on industrial residues.


116 One of the most significant by-products from the aluminium foundry industry is an
117 alkaline aqueous solution, which is used to wash the extrusion sand moulds and other
118 tools used in the moulding process. After a few cycles, the solution eventually becomes
119 saturated with the dissolved aluminium and is no longer sufficiently reactive, being
120 discarded as a residue. The Portuguese Environment Agency does not reveal the
121 volumes of each individual residue produced in the country, but it is unofficially
122 estimated to be around 500 m3. If used as an alkali activator, this solution could increase
123 the pH of the mixture to initiate the reactions, while also providing some soluble
124 aluminium. The use of a similar solution in the role of the alkali activator was
125 previously reported by Van Riessen et al. (2013). However, in this case the residue was
126 the ‘Bayer liquor’, resulting from the well-known Bayer process to refine bauxite for the
127 production of alumina. These authors used two versions of the Bayer liquor – a
128 synthetic (clean) version, i.e. a clean sodium aluminate solution; and a processed plant
129 liquor (with all the associated impurities) – as a primary source of caustic sodium
130 aluminate, to activate Class F fly ash, with good results. The amount of Bayer liquor
131 used to replace sodium hydroxide (NaOH) did not have any significant impact on the
132 compressive strength, and the processed liquor yielded a higher compressive strength
133 than the clean liquor. In essence, the authors demonstrated that the Bayer processed
134 liquor could be used for the production of geopolymer products. Although the solution
135 concentration used in the above cited research was slightly higher than the solution
136 concentration in the present work, it still seems that this cleaning solution might
137 constitute a promising possibility, and its use could result in financial and
138 environmental cost reductions, by enabling the partial substitution of OPC at a
139 reasonable cost and by creating new applications to a residue which otherwise would
140 have to be landfilled.
142 Based on the scarce research papers on the subject, the possibility of fabricating alkali
143 activated cement composites using only industrial by-products can be considered as
144 absolutely innovative. Such alternative would certainly boast the number of live
145 applications of sustainable binders made from alkali activation of industrial by-
146 products. The most interesting applications could be those where the maximum strength
147 is not the main target (e.g. the stabilisation of road pavement foundations, which
148 requires large material volumes and, as a consequence, is significantly dependent of the


149 cost/effectiveness ratio); or those where the activator does not require high levels of
150 alkalinity (e.g. when the precursor possesses significant percentages of calcium, and
151 thus a low pH activator is favoured). The aim of the present paper is thus to analyse the
152 potential application of an aluminium mould cleaning solution (CS) as an alkali
153 activator used in the preparation of alkaline cement. In order to evaluate the
154 effectiveness of the CS, a threshold was created using a common 8M commercial
155 solution. Two distinct types of waste were considered as precursors in this paper, a
156 Portuguese Class F fly ash (FA) and a glass powder (GP) resulting from the production
157 of optical lenses. The inclusion of two different precursors – with significantly different
158 chemical compositions – was related with the need to better understand the role of the
159 soluble aluminium present in the CS. The performance assessment was based on
160 flexural and compressive strength, which was complemented by a thorough
161 microstructural analysis, including scanning electron microscopy, energy dispersive X-
162 ray spectroscopy, X-ray diffraction and nuclear magnetic resonance.
165 2. Materials and methods
167 2.1 Material identification and analysis
169 2.1.1 Precursors
171 The FA was provided by a Portuguese thermo-electric power plant and, based on
172 ASTM C 618 (2012), was classified as Class F. To better control the synthesis process
173 and facilitate the alkali reactions, the FA was previously milled (MFA), in a RETSCH
174 RS200 mill, for 1 minute, at 1200 rpm. The GP results from the Portuguese industrial
175 production of optical lenses, thus constituting a viable and low-cost source of soluble
176 silica. It was received in slurry form and, after dried, it was hand-milled in a porcelain
177 mortar and pestle, for de-flocculation.
179 The chemical composition of both the FA and the GP, obtained by X-ray fluorescence
180 (XRF) on a PHILIPS PW-1004 X-ray spectrometer, is presented in Table 1. The mass
181 loss on ignition (L.O.I.), determined between 100ºC and 1000°C for 1 h on a platinum
182 crucible, is also presented. The sum of silica and alumina in the FA amounts to


183 approximately 75%, while the calcium (Ca) represents only 1.68%. Regarding the GP,
184 its most relevant feature is the silica content, up to almost 60%. The aluminium (Al)
185 content represents the most significant difference between the chemical composition of
186 both precursors, with 21.5% and 3.9%, in the FA and GP, respectively.
188 Also included in Table 1 are the results obtained from a selective chemical attack on
189 both precursors, using a 1% (v/v) hydrofluoric acid solution. This method was
190 described, in detail, by Ruiz-Santaquiteria et al. (2013), and indicates the reactive SiO2
191 and Al2O3 wt% present in the precursor. Based on these values, the corresponding
192 reactive SiO2/Al2O3 ratio was calculated. The FA showed a potentially total reactive
193 phase of approximately 63%, including the SiO2 (42.60%) and the Al2O3 (20.57%).
194 This means that approximately 77,8% and 95.0% of the total SiO2 and Al2O3 present in
195 the FA are reactive. Therefore, the SiO2+Al2O3 reactive content determined by chemical
196 attack (63%), is lower than the total value determined by XRF (75%). The GP showed a
197 lower potentially reactive phase (45.77%), mainly due to the low initial Al content.
199 Table 1: Chemical composition of the fly ash and the glass powder (% wt)
Type Element FA GP
Na2O 1.180 8.750
SiO2 54.77 58.37
Al2O3 21.57 3.940
MgO 1.870 0.492
Chemical composition

K2O 2.559 4.659

CaO 1.680 6.128
TiO2 1.010 2.730

Fe2O3 7.325 0.148

ZnO 0.026 2.572
ZrO2 0.045 2.155
BaO 0.121 2.300
PbO 0.004 5.072
Others 1.368 0.833
aL.o.I. 6.470 1.849
Si/Al 2.53 14.81
Reactive SiO2 42.60±0.80 42.25±0.27

1% HF

Reactive Al2O3 20.57±0.50 3.52±0.07


Reactive SiO2 + Al2O3 63.17 45.77

Reactive SiO2 / Al2O3 2.07±0.01 12.00±0.17
201 Figure 1 shows the particle size distribution curves of the glass powder and the fly ash,
202 before (FA) and after (MFA) milling. These curves were obtained by laser diffraction,
203 using a Sympatec Helos BF particle-size analyser, with a measuring range between 0.9
204 mm and 175 mm. The analyses were conducted after a 5-minute ultrasonic bath, in a


205 water-ethanol suspension, for enhanced particle dispersion, and the results were
206 interpreted using Fraunhofer diffraction theory. The particle size below 45µm was 89.8,
207 98.8 and 99.2 wt% for the FA, MFA and GP, respectively. A substantial percentage
208 (10.2 wt%) of particles with a diameter higher than 45 µm were found in the original
209 FA, which justified the milling, to increase its original reactivity. On the contrary, all
210 the GP particles are finer than 45µm, and so no milling was though necessary.

213 Figure 1: Cumulative volume (a) and Size distribution (b) of the FA, MFA and GP
215 The specific gravity of each precursor was determined using a Le Chatelier
216 volumenometer, based on UNE 80103 (2013). The values of 2.26, 2.39 and 2.63 g/cm3
217 were obtained for the FA, MFA and GP, respectively.
219 2.1.2 Activators
221 Two different activators were used: a commercial 8M NaOH solution (N8), used to
222 establish a reference; and an alkali cleaning solution (CS) from the Portuguese
223 aluminium processing industry. The N8 solution was prepared by dissolving NaOH
224 pellets, from Sigma-Aldrich Co. USA – with a specific gravity of 2.13 and 99 wt.%
225 purity – in distilled water, while the CS was used as received in the laboratory.
227 The elemental composition of both solutions, presented in Table 2, was determined by
228 means of inductively coupled plasma atomic emission spectrometry (ICP-AES), on a
229 Variant 725-ES ICP atomic emission spectrometer, with the following characteristics:


230 plasma power of 1.40 kW, plasma gas flow of 15.00 l/min, nebulizer gas flow of 0.85
231 l/min and read time of 5s. The CS alkali concentration is approximately 5M, which is
232 slightly lower than the threshold accepted as ideal for this type of materials – between
233 8M to 12M (Cristelo et al., 2016; Fletcher et al., 2005; Granizo et al., 2014;
234 Panagiotopoulou et al., 2007; Xu and van Deventer, 2000). Also worth mention is the
235 presence of a small percentage of sulphates (1.18%) on the CS. The densities of both
236 solutions are very similar (approximately 1.3 g/cm3).
238 Table 2: Chemical composition of the alkali solutions used as activators (% wt)
Element NaOH Cleaning
8M solution
(N8) (CS)
Al2O3 - 7.14
Na2O 19.5 12.13
SO3 - 1.18
H2O 80.5 79.5
pH ≤ 14 ≤ 14
a [OH]- 7.9 5.3
Density (g/cm )
3 1.27 1.30
239 a Acid-base reaction with 5N HCL (Panreac S.A.)

241 Figure 2 shows the 27Al MAS Nuclear magnetic Resonance (NMR) spectrum acquired
242 for the CS (see Section 2.4 for details). Results show a principal resonance at -80 ppm,
243 which is characteristic of the tetrahedrally coordinated Al – δ Al(IV) ≈ -80 ppm –
244 present in [Al(OH)4-] specimens (Engelhardt and Michel, 1987).

247 Figure 2: 27Al MAS NMR spectra of the cleaning solution
249 2.2 Compressive and flexural strength tests


251 To assess the mechanical behaviour, different pastes were prepared with the precursors
252 (MFA and GP) and the activators (N8 and CS, both in solution form). Pastes were cast
253 into prismatic moulds (1 cm x 1 cm x 6 cm), following the procedure described by Koch
254 and Steinegger (1960). The content in alkali solution was defined based on the
255 minimum liquid phase necessary to obtain a viscous consistency, enough to avoid the
256 need to compact the pastes, and was thus different for each precursor. Therefore,
257 activator/precursor weight ratios of 0.35 and 0.48 were obtained for the MFA and GP
258 mixtures, respectively. The specimens were cured for 20 h, at 85 ºC and > 90% relative
259 humidity. This procedure, developed in previous studies (Criado et al., 2005; Torres-
260 Carrasco and Puertas, 2015), was defined to prevent an excessively fast setting and,
261 especially, carbonation, by ensuring that the humidity of the mixtures is kept at a
262 constantly high value. After the 20h period, the specimens were removed from the
263 climatic chamber and tested, 24 h after fabrication. Other selected samples were
264 maintained in the same conditions during 28 days, after which they were also tested.
265 The mechanical strength of the cured pastes was determined on an IBERTEST
266 AUTOTEST-200/10 SW test frame, as the arithmetic mean of six specimens (for
267 flexural strength) and twelve specimens (for compressive strength values).
269 2.3 Calorimetry tests
271 To determine the heat evolution rate through calorimetric tests, MFA and GP mixtures
272 with both solutions were prepared, using the activator / precursor weight ratios
273 previously mentioned (i.e. 0.35 for the MFA and 0.48 for the GP). The pastes were
274 manually mixed for a period of 3 minutes, and then sealed. The containers were then
275 immediately placed inside the chamber, and their heat evolution, under a constant
276 external temperature of 85 ±1°C, was recorded. The equipment used was a conduction
277 isotherm calorimeter TAM Air Thermometric.
279 2.4 Characterization techniques
281 The mineralogical and microstructural characterization of the two precursors and the
282 resulting reaction products was also addressed, to enhance the understanding of the
283 composition-microstructure-strength relationship. After mechanical testing, some
284 fragments of a specimen from each batch were submerged in an acetone / ethanol mix to


285 detain the chemical reactions, ground to a fine powder (< 45 μm), and analysed by X-
286 ray diffraction (XRD) and Nuclear Magnetic Resonance (NMR). Changes in
287 microstructure and paste reaction products were analysed by Scanning Electron
288 Microscopy (SEM) and Energy Dispersive X-ray (EDX), although in this case the
289 analyses were made on the surface of broken fragments, after covered with carbon.
291 The mineralogical diffractograms were obtained with a Bruker-AXS D8 Advance
292 equipment. The scans covered a 2 range of 5 to 60º, with a nominal step size of
293 0.0119736º and 0.5 s/step. A Cu anode, with release energies of K-alpha1 (1.54060 Å)
294 and K-alpha2 (1.54443 Å), was used.
296 The morphology of the pastes was examined through SEM, using a JEOL JSPM-5400
297 microscope. The equipment is coupled with an EDX detector, model Oxford-Link-Isis.
299 In addition, a Bruker Advance III 400 MHz Nuclear Magnetic Resonance (NMR) mass
300 spectrometer was also used to analyze the pastes, under the following conditions: 29Si

301 resonance frequency of 79.5 MHz, spinning rate of 10 kHz, pulse sequence with a
302 single pulse of 5 μs and a recycle delay of 10 s, number of transients of 4912 and
303 external standard TMS (tetramethylsilane); 27Al resonance frequency of 104.3 MHz,
304 spinning rate of 10 kHz, pulse sequence with a single pulse of 2 μs and a recycle delay
305 of 5 s, number of transients of 400 and external standard Al(H2O)63+.
308 3. Results
310 3.1 Compressive and flexural strength
312 The flexural and compressive strengths of all the mixtures are shown in Figure 3. MFA-
313 based specimens presented an overall higher strength than the specimens made with the
314 GP. However, some differences were observed related with the different activator-
315 precursor combinations tested, namely the fact that the higher strength values in the
316 MFA pastes were obtained with the commercial N8 solution, while in the GP pastes the
317 higher strength values were produced with the cleaning solution. This was observed
318 both in terms of flexural and compressive strength.



321 Figure 3: Flexural (a) and compressive (b) strength of the pastes after 1 and 28 days note the different
322 vertical scales)
324 3.2 Calorimetry
326 Figure 4 shows the heat flow and cumulative heat from the activation of the MFA and
327 the GP, with the N8 and CS solutions. The initial reaction is very fast in all cases, so
328 much so that the common initial peak associated with the dissolution process and the
329 subsequent dormant period were not detected in this case (Palomo and Grutzeck, 1999).
330 Thus, the first peak detected in the heat flow curves was associated with the
331 acceleration/deceleration steps i.e., with the main precipitations of the reaction products,
332 which in this case was gel type N-A-S-H (Fernández-Jiménez et al., 2006, 2005). The
333 intensity of this first peak is higher in MFA than in GP pastes, which in accordance with
334 the mechanical results obtained. Also, in the MFA pastes the peak obtained with the N8
335 solution is higher than the peak obtained with the cleaning solution, and was formed
336 approximately 15 minutes before. In the GP pastes, the differences between N8 and CS
337 solutions are smaller, with the peaks showing very similar intensities and appearing
338 with an interval of just 9 minutes (after 30 and 39 minutes, respectively).

340 The cumulative heat released by all systems is also shown in Figure 4. The total heat
341 release, after 30 h, is in the range 70-80 J/g (MFA) and 60-70 J/g (GP). Also, the total
342 heat release is higher with the N8 than with the CS, which could be related to a higher
343 reaction degree, of both precursors, when activated with the N8 solution.


346 Figure 4: Heat flow and cumulative heat flow from the activation of the MFA (a) and the GP (b), with N8
347 and CS solutions
349 3.3 Mineralogy
351 Figure 5 shows the XRD diffractograms of the original materials (MFA and GP) and the
352 respective alkali activated pastes, prepared with both alkali solutions. These original
353 materials exhibited a hump in the 20º to 30º (2θ) angle range, which is characteristic of
354 structurally disordered compounds, with a high amorphisation level. As expected, the
355 fly ash presented several sets of peaks associate with different crystalline phases,
356 namely quartz, mullite and hematite. On the other hand, the diffractogram obtained
357 from the glass residue did not revealed any common mineral. After alkaline activation,
358 the mentioned hump in the diffractograms shifted to the right, towards 2θ angle
359 positions approximately between 25º and 35º, while simultaneously decreasing in
360 intensity (due to the scales used, this effect is easier to observe in the GP pastes (Figure
361 5b). This shift is related to the formation of a reaction compound of an amorphous
362 nature, namely a sodium aluminosilicate gel (a zeolitic precursor), which constitutes the
363 major source of the material’s binding properties. Regarding the type of activator, when
364 CS was used some small peaks were detected, regardless of the precursor, which were
365 associated with the formation of new crystalline phases. The presence of these
366 crystalline phases is clearer in the diffractograms obtained after 28 days of curing
367 (under 85 ºC and 99% relative humidity).


370 Figure 5: XRD patterns of the MFA (a) and GP (b) pastes (Legend: m - mullite PDF-015-0776; Q - quartz
371 PDF-046-1045; S - hydroxy-sodalite (PDF-076-1639); H-herschelite, PDF-19-1178; P-zeolite P(Na),
372 PDF-40-1464)
374 Although the chemical composition of the two precursors showed some significant
375 differences, it interesting to observe that when the CS was used, the same types of
376 crystalline phases were formed, after 28 days, as secondary reaction products: hidroxy-
377 sodalite (already detected after 2 days), herschelite zeolite (NaAlSi2O6.3H2O) and Na-P
378 zeolite (Na3.6Al3.6Si12.4 O3214H2O). If the zeolites are originated by a zeolitic precursor –
379 which in this case would be the N-A-S-H gel – the formation of two types of zeolites,
380 with different Si/Al ratios (2 and 3.4), partially support the hypothesis of the formation
381 of gels with different Al contents when the CS is used as activator.
383 3.4 Nuclear magnetic resonance
385 Samples activated with CS were selected for a more thorough study of the respective
386 gel, using nuclear magnetic resonance (NMR). The 27Al MAS NMR spectra (Figure 6)
387 showed that both the original materials (MFA and GP), as well as the products
388 emerging from the reaction with the activators (N8 and CS), presented the Al in
389 tetrahedral coordination. In the original materials, the tetrahedral aluminium is part of
390 the glass structure, acting as network forming species, which was confirmed by the wide
391 signal centred at 51/53 ppm, associated with a high degree of structural disorder. During
392 the analysis of the reaction products, a significantly more acute signal was obtained,


393 centred at around 59 ppm, thus indicating a certain degree of short range structural order
394 in these newly formed gels. Regardless of the very different Al content in both
395 precursors, the Al signal from all the reaction products is very similar and identical to
396 those detected by other authors, when studying N-A-S-H gels (Benavent et al., 2016;
397 Criado et al., 2008; Fernández-Jiménez et al., 2006, 2005; García-Lodeiro et al., 2015;
398 Torres-Carrasco and Puertas, 2015) and zeolites (Granizo et al., 2014; Querol et al.,
399 2002; Takeda et al., 2013). Also relevant is the fact that the signal at 80 ppm, detected
400 during the Al MAS NMR analysis of the cleaning solution (Figure 2), was not present
401 in the reaction products, indicating that the aluminium from the CS has indeed reacted
402 with the precursors.

405 Figure 6: 27Al NMR-MAS spectra of the MFA (a) and GP (b) and their respective reaction products
407 Regarding the 29Si MAS NMR spectra Figure 7, some differences can be observed, as a
408 function of the precursor used and the reaction time. These differences illustrate the
409 chemical and microstructural changes that happened throughout the alkaline activation
410 of the original materials. Both the original MFA and GP spectra present fairly wide,
411 asymmetrical and poorly defined shapes, which is due to the overlapping of a variety of
412 different signals. MFA spectrum shows the centre of gravity at -108 ppm, which
413 indicates overlapping of Q4(mAl) units. However, the GP pastes present the center of
414 gravity at -93ppm, which, together with the low aluminum content of this precursor
415 (3.94%, see Table 1), indicates that the GP is formed mainly by Q3(mAl) units and,


416 maybe, some additional Q4(mAl) and Q2(mAl) units. These results suggest a lower
417 degree of polymerization of the glass, when compared with the fly ash, due to the lower
418 Al content of the former and, simultaneously, to its higher Na and Ca contents, which
419 act as depolymerisation agents (Garcia-Lodeiro et al., 2016; Ruiz-Santaquiteria et al.,
420 2012).
422 The one-day curing spectrum of the MFA-CS paste shows the appearance of an intense
423 signal at -88 ppm, together with a less intense peak at -93 ppm. The former is associated
424 with the formation of Q4(4Al) silicon units, and the latter with the formation of Q4(3Al).
425 The presence of a peak at -108 ppm, prevailing from the original MFA spectrum,
426 reveals a significant volume of unreacted fly ash after 1 day curing. The spectrum
427 obtained after 28 days shows some relevant changes, namely the abrupt decrease in
428 unreacted material (the peak at -108 ppm is less intense) and the shift in the previous -
429 88 and -93 signals, now peaking at -98.5 ppm and -104 ppm, respectively. These peaks
430 are associated with Q4(2Al) and Q4(1Al) silicon units, which, together with the
431 significant intensity reduction of the peak registered at -108 [Q4(0Al) unit], reveal the
432 development of a N-A-S-H type gel, a precursor of the herschelite (or Chabazite-Na)
433 zeolite (Engelhardt and Michel, 1987; Fernández-Jiménez et al., 2006). As the curing
434 time increased (from 1 to 28 days), the width of the Q4(2Al) and Q4(1Al) respective
435 peaks also increased, thus reducing the intensity of the Q4(4Al) and Q4(3Al) units, and
436 indicating that the silicon content in the gel increased with curing time. A similar
437 behaviour was previously observed when activating fly ash with a 8M sodium
438 hydroxide solution (Fernández-Jiménez et al., 2006).
440 Regarding the glass residue, it was also possible to detect, after 1 day curing, the
441 presence of a newly formed peak, at -89 ppm, near a high unreacted glass content. Just
442 like in the case of the fly ash, significant changes were observed in the glass structure
443 after 28 days. The unreacted glass content is now much lower, and a new, very intense
444 peak has formed at -93 ppm, with shoulders at -89 and -104 ppm. The -93 ppm signal
445 could be associated with Q4(3Al) units, like in the case of the fly ash. However,
446 considering the low aluminium content, as well as the presence of some calcium, it is
447 wiser to assume that it is associated with Q3(0Al) silicon units (Engelhardt and Michel,
448 1987). Signal overlapping is a strong possibility, which would be a consequence of the
449 lower polymerization of the activated GP, when compared with the activated MFA. The


450 N-A-S-H gel developed in the MFA-based pastes is formed by Q4(mAl) silicon units,
451 thus possessing a three-dimensional structure. The gel developed in the GP-based pastes
452 is mainly formed by Q3 units (which can be overlapped with some Q4 units), possibly
453 representing a mixture of sodium silicate gels with different calcium and aluminium
454 contents, thus possessing a two-dimensional structure.

457 Figure 7: 29Si NMR-MAS spectra of the MFA (a) and GP (b) and their respective reaction products
459 3.5 SEM-EDX
461 Figure 8 shows some micrographs of the MFA and GP-based pastes, activated with the
462 CS and cured for 1 day at 85ºC. In the MFA paste (Figure 8a), the presence of a large
463 number of unreacted fly ash particles was observed, together with the formation of a N-
464 A-S-H gel. The variation in fly ash reactivity ranges from no appreciable reaction (point
465 1), through surface dimpling (point 2) to partial or complete reaction. The main reaction
466 product is sodium silicoaluminate gel (point 3). This picture also suggests that, after 1
467 day, a low to moderate reaction degree was obtained, resulting in the presence of a
468 significant number of pores in the microstructure.


471 Figure 8: SEM images and EDXS spectra of the MFA-CS (a) and GP-CS (b) pastes after 1 day
473 Similar results were obtained in GP-CS paste. However, in the morphology of the
474 resulting matrix is more difficult to distinguee between unreacted particles and reaction
475 product. The same thing happened with the nature of the reaction products, which
476 presented a morphology not very well defined. Nonetheless, it can be described as a
477 light compact granular media. EDX data can help to distinguish between unreacted GP
478 particles (point 4) constituted by silica with relatively significant contents of Na, Ca, Pb
479 and Ba, and two possible reaction products: a sodium silicate with a small amount of Al
480 (point 5) – possibly a N-A-S-H gel with low Al content; and a sodium-calcium-silicate
481 (point 6), obtained by the solubilisation of the original silica and calcium from the GP,
482 and their subsequent reaction with the Na from the solution or from the glass itself.
484 Figure 9 shows the microstructure of the same pastes analysed in Figure 8, although
485 after 28 days curing. A more compacted microstructure can clearly be observed,
486 meaning that in both cases a higher reaction degree was obtained, relatively to the same
487 pastes after 1 day curing. In both materials (MFA-CS and GP-CS) the presence of the
488 zeolites Na-P and herschelite was detected (through XRD) as secondary reaction
489 products. The presence of such zeolites might be related with the alkaline activator used


490 (cleaning solution), especially because zeolite ‘P’ is not usually detected during
491 activation of fly ash with an 8M NaOH solution. Instead, the formation of this particular
492 zeolite is favoured by lower alkalinity solutions (Criado et al., 2008, 2007; Murayama et
493 al., 2002; Querol et al., 2002). The formation of a more compact matrix, in both types of
494 paste (MFA and GP) is associated with a higher reaction degree, which resulted also in
495 an increase in mechanical strength between 1 and 28 days curing.

498 Figure 9: SEM images and EDXS spectra of the MFA-CS pastes (a) and GP-CS pastes (b) cured 28 day
501 4. Discussion
503 Alkaline cements consist essentially of two components: a cementitious material and an
504 alkaline activator. In this paper, two industrial by-products were used as a precursor (fly
505 ash type F and a glass power rich in Silica); and two alkaline solutions were used as an
506 activator (an 8M caustic solution common in this type of applications, as a reference, and a


507 cleaning solution resulting from the aluminium foundry industry). The cleaning solution
508 presented an OH- ion content of 5.3, slightly lower than the threshold recommend in the
509 literature for the efficient activation of fly ash. Furthermore, it contains soluble aluminium,
510 which further enhances the differences relatively to a more traditional activator, especially
511 if low-Al precursors are to be activated.
513 It is well-known that, in the alkali activation process, the OH- ions catalyse the hydrolysis
514 of the Si-O-Si and Si-O-Al bonds present in the original aluminoslicates. For pH values
515 between 3 and 11, the solubility of the silicon is low, although it will increase suddenly
516 when the pH reaches a value below 3 or above 11. Nevertheless, in acidic pH the hydrates
517 formed are not stable and do not form consistent structures. On the contrary, basic pH not
518 only favours the dissolution of the silicon, but it also contributes to a more efficient
519 coagulation and polymerization process, resulting in the formation of stable hydrates and,
520 consequently, in better binding properties. The aluminium follows a similar pattern, i.e. the
521 solubility increases with low (< 3) and high (> 6) pH. The cation Al3+ remains intact in
522 acidic media, although coexisting with other cations, like Al(OH)2+ and Al(OH)+2; while it
523 is the predominant species in basic pH, forming Al(OH)-4 anions (Benavent et al., 2016;
524 Fernández-Jiménez et al., 2006; García-Lodeiro et al., 2015; Silva et al., 2007). However,
525 although the solubility of both silicon and aluminium-based species is favoured by a high
526 pH, that is not the case with calcium, which becomes less soluble with the increase in
527 alkalinity. Such behaviour results in the fact that the precursors formed by more
528 polymerized vitreous or amorphous structures, of the type Q4(mAl), like fly ash, need more
529 alkali content than aluminosilicates with less polymerized structures, of the type Q1 or Q2,
530 like blast furnace slag (Fernández-Jiménez et al., 2003; Fernández-Jiménez and Puertas,
531 2003; Torres-Carrasco and Puertas, 2015).
533 All the previous comments justify, in great measure, the reason why the activation of fly
534 ash with 8M NaOH solution developed a higher mechanical strength than with the
535 cleaning solution. Due to its vitreous structure – units Q4(mAl) – typical fly ash requires
536 energetic reaction conditions. Therefore, it is possible to assume that the early age
537 higher mechanical strength of the FA pastes prepared with N8 is a consequence of its
538 capacity to stimulate a higher reaction degree than the CS. This lower intensity of the
539 short-term reaction of the fly ash when activated with the cleaning solution is probably
540 a consequence of the lower alkalinity, when compared with the N8 solution. However,


541 with time, a stable material, with good mechanical strength, will form from the MFA-
542 CS combination. As the curing time increases (28 days), the strength values obtained by
543 the MFA-N8 and MFA-CS pastes tend to a similar plateau. The results obtained with
544 calorimetry corroborate this observation. The MFA-N8 paste reacted approximately 15
545 minutes before the MFA-CS paste, although the intensity of the reaction with the N8
546 solution is clearly higher than with the CS (see Figure 4), due to its higher alkali
547 content.
549 Regarding the GP, the opposite behaviour was detected, i.e. a higher strength was
550 registered with the CS than with the N8 solution. The GP residue used is a glass with a
551 lower polymerization level than the fly ash. It is formed by Q3 units, due to the lower Al
552 content and, simultaneously, to the higher Na and Ca contents, which act as
553 depolymerising agents. This results in a lower influence of the alkalinity (when ranging
554 between 5M and 8M) on its solubility, which is visible on the heat release data (Figure
555 4b), which showed very similar dissolution patterns for the GP-N8 and GP-CS. These
556 calorimetry results could be interpreted as similar reaction levels of the GP, with both
557 activating solutions. However, if the reaction levels were in fact similar, the question
558 would have to be “why then was the strength of the GP-CS paste higher than the GP-N8
559 paste”. At this point it is important to highlight the possible role of the soluble
560 aluminium present in the cleaning solution, since its presence can indeed affect the
561 structure and composition of the reaction products. During the alkaline activation of
562 aluminosilicates, with low Ca content, the main resulting product is a gel type N-A-S-H,
563 with a 3D structure, which is responsible for the performance of such materials in terms
564 of mechanical strength. When synthetizing the N-A-S-H gel, the SiO2/Al2O3 ratios of
565 the reaction matrix (glass), and not the mass relations, are preponderant. According to
566 the literature, the best binders (exhibiting the highest mechanical strengths) are obtained
567 with materials with starting SiO2/Al2O3 reactive ratios ranging from 2 to 4 (or
568 Al2O3/SiO2 reactive ratios between 0.25 and 0.5). Such ratios were obtained in the
569 present work, either with the N8 or with the CS. The aluminium plays a significant role
570 in the process of polymerization and formation of the N-A-S-H gel. During a previous
571 work regarding the activation of fly ash with NaOH, it was demonstrated how the
572 aluminium favours the early formation of Si-O-Al bonds and, hence, the formation of a
573 metastable, high aluminium gel. This gel was denominated ‘Gel 1’, which evolves with
574 time into a high silicon gel (‘Gel 2’), responsible for the long-term mechanical


575 performance of the product (Fernández-Jiménez et al., 2006). This exact same
576 behaviour was observed in the MFA-CS and GP-CS pastes, through NMR. After curing
577 for 1 day, the gel obtained is richer in Al (concluded after the higher percentage of
578 Q4(4Al) and Q3(3Al) units, respectively), while after 28 days the gel is predominantly
579 richer in silicon than aluminium (concluded after observing Q4(2Al) and Q4(1Al) units
580 in the MFA-CS and Q3 units in the GP-CS). Such behaviour can also be observed in
581 Figure 10, representing the EDX molar distribution of the composition of the gels (both
582 for MFA and GP), activated with CS and after 1 and 28 days curing. These results
583 clearly highlight how the different composition of each precursor affects the
584 composition of the reaction products and its evolution with time. The MFA-CS pastes
585 showed a composition very similar to those reported by other authors, when activating
586 FA with a NaOH solution. After 1 day, a N-A-S-H type gel is formed, which evolves
587 with time and promotes a more homogeneous distribution of the Al/Si and Ca/Si ratios.

590 Figure 10: Chemical composition (EDX) of the different gels (qualitative data)
592 The reactive Al content in the glass powder is very low (approximately 3.52%, see
593 Table 1), which is clearly below the threshold values presented in the literature.
594 Therefore, it must be assumed that the formation of N-A-S-H gels, when using the N8
595 solution as activator, is highly improbable. Instead, what is favoured is the formation of
596 sodium calcium silicate gels (Cyr et al., 2012; Redden and Neithalath, 2014). This
597 means that the reaction that takes place when the glass pastes are activated with 8M


598 NaOH cannot be considered as a typical geopolymerisation process. The strength

599 obtained with this combination is low, and the use of material with a higher Al content,
600 like the cleaning solution (7.14% Al), might partially compensate such deficiency. In
601 Figure 10 it is possible to observe how indeed the gels formed with the GP-CS pastes
602 presented a decrease in the SiO2 / Al2O3 ratio, between 1 and 28 days curing. There is
603 an initial reaction between the aluminium, from the CS, and the silicon, from the GP,
604 producing N-A-S-H gels with low Al content (since the soluble Al from the CS is
605 limited). As the curing time increases, less and less aluminium is available, and so the
606 Si/Al ratio of the N-A-S-H gel increases. This shortness in aluminium is also
607 responsible for the lower polymerization of the gel, or mixture of gels, resulting from
608 the GP-CS pastes, which is confirmed by the predominance of Q3(mAl) units (Figure 7).
610 It was previously mentioned that the presence of Al favours the condensation of the
611 silicates, but it also increases the crystallinity of the reaction products (more zeolites and
612 less gel) (Criado et al., 2008, 2007; García-Lodeiro et al., 2015; Silva et al., 2007). The
613 reactive SiO2 / Al2O3 ratio is also very important during zeolite synthesis, but different
614 types of zeolites can be obtained with the same SiO2 / Al2O3 ratio, just by altering the
615 curing temperature or time, the activator concentration or the liquid / solid ratio (Breck,
616 1973; Querol et al., 2002). During the present work, the formation of zeolites was
617 detected, but only as secondary reaction products, in pastes (MFA or GP) activated with
618 the cleaning solution. In both types of pastes the same zeolitic products were observed,
619 namely herschelite and zeolite P(Na), even though the precursors were different.
620 Additionally, these zeolites were not detected with the 8M NaOH activator, at least
621 during the maximum curing period of 28 days considered. Therefore, the presence of
622 such zeolites should be attributed to the combined effect of the lower alkalinity and
623 soluble aluminium present in the cleaning solution.
625 In short, although the cleaning solution presents a slightly lower alkali concentration
626 than the value recommended for fly ash activation, it does present a significant potential
627 for alkaline activation applications, which is translated by the similar mechanical
628 strength obtained by both pastes (MFA-N8 and MFA-CS) after 28 days. Furthermore,
629 increasing the alkali concentration of the cleaning solution would be relatively easy, just
630 by adding alkalis from an external source or by reducing its water content. The Al


631 content presented by this solution can play a major role in the activation of precursors
632 with Al deficiency, like glass powder.
635 5. Conclusions
637 The aim of the presented research was to determine if a sodium hydroxide solution,
638 previously used in the process of cleaning foundry sand moulds, could later be used
639 (after rejected as a cleaning agent) as an alkaline activator for fabricating geopolymers.
640 This could constitute and interesting application since it would not only allow the
641 recycling of a hazardous waste, but it would also help mitigate one of the current major
642 drawbacks for the spreading of the alkaline activation technique, which is the financial
643 and environmental cost of the activating solutions. The use of the CS delivered some
644 very promising results, even if the alkalinity of the CS was lower than that of the 8M
645 NaOH solution used as reference. The presence of tetrahedral aluminium in the CS can
646 have a positive effect during the activation of low-Al, Si-rich raw materials, since it
647 would act as an aluminium correction source, as well as an activator.
650 Acknowledgments
652 This research was funded by the Spanish Ministry of the Economy and Competitiveness
653 and funds FEDER under research projects BIA2013-43293-R and BIA2016-76466-R.
654 The authors would also like to acknowledge the contribution of the Portuguese
655 companies W2V - Waste to Value for the supply of the cleaning solution; and POLO -
656 Produtos Óticos, SA for the supply of the glass residue.
659 References
661 Al-Zboon, K., Al-Harahsheh, M.S., Hani, F.B., 2011. Fly ash-based geopolymer for Pb
662 removal from aqueous solution. J. Hazard. Mater. 188, 414–421.
663 doi:10.1016/j.jhazmat.2011.01.133
664 Antunes Boca Santa, R.A., Bernardin, A.M., Riella, H.G., Kuhnen, N.C., 2013.


665 Geopolymer synthetized from bottom coal ash and calcined paper sludge. J. Clean.
666 Prod. 57, 302–307. doi:10.1016/j.jclepro.2013.05.017
667 Arulrajah, A., Piratheepan, J., Disfani, M.M., Bo, M.W., 2013. Geotechnical and
668 Geoenvironmental Properties of Recycled Construction and Demolition Materials
669 in Pavement Subbase Applications. J. Mater. Civ. Eng. 25, 1077–1088.
670 doi:10.1061/(ASCE)MT.1943-5533.0000652
671 ASTM C618, 2012. Standard specification for coal fly ash and raw or calcined natural
672 pozzolan for use in concrete. ASTM Int. Annu. B. Stand. 1–5.
673 Avirneni, D., Peddinti, P.R.T., Saride, S., 2016. Durability and long term performance
674 of geopolymer stabilized reclaimed asphalt pavement base courses. Constr. Build.
675 Mater. 121, 198–209. doi:10.1016/j.conbuildmat.2016.05.162
676 Benavent, V., Steins, P., Sobrados, I., Sanz, J., Lambertin, D., Frizon, F., Rossignol, S.,
677 Poulesquen, A., 2016. Impact of aluminum on the structure of geopolymers from
678 the early stages to consolidated material. Cem. Concr. Res. 90, 27–35.
679 doi:10.1016/j.cemconres.2016.09.009
680 Breck, D.W., 1973. Zeolite molecular sieves: structure, chemistry, and use. Wiley.
681 Criado, M., Fernández-Jiménez, A., de la Torre, A.G., Aranda, M.A.G., Palomo, A.,
682 2007. An XRD study of the effect of the SiO2/Na2O ratio on the alkali activation
683 of fly ash. Cem. Concr. Res. 37, 671–679. doi:10.1016/j.cemconres.2007.01.013
684 Criado, M., Fernández-Jiménez, a., Palomo, a., Sobrados, I., Sanz, J., 2008. Effect of
685 the SiO2/Na2O ratio on the alkali activation of fly ash. Part II: 29Si MAS-NMR
686 Survey. Microporous Mesoporous Mater. 109, 525–534.
687 doi:10.1016/j.micromeso.2007.05.062
688 Criado, M., Palomo, A., Fernández-Jiménez, A., 2005. Alkali activation of fly ashes.
689 Part 1: Effect of curing conditions on the carbonation of the reaction products. Fuel
690 84, 2048–2054. doi:10.1016/j.fuel.2005.03.030
691 Cristelo, N., Glendinning, S., Miranda, T., Oliveira, D., Silva, R., 2012. Soil
692 stabilisation using alkaline activation of fly ash for self compacting rammed earth
693 construction. Constr. Build. Mater. 36, 727–735.
694 doi:10.1016/j.conbuildmat.2012.06.037
695 Cristelo, N., Miranda, T., Oliveira, D.V., Rosa, I., Soares, E., Coelho, P., Fernandes, L.,
696 2015. Assessing the production of jet mix columns using alkali activated waste
697 based on mechanical and financial performance and CO2(eq) emissions. J. Clean.
698 Prod. 102, 447–460. doi:10.1016/j.jclepro.2015.04.102


699 Cristelo, N., Tavares, P., Lucas, E., Miranda, T., Oliveira, D., 2016. Quantitative and
700 qualitative assessment of the amorphous phase of a Class F fly ash dissolved
701 during alkali activation reactions - Effect of mechanical activation, solution
702 concentration and temperature. Compos. Part B Eng. 103, 1–14.
703 doi:10.1016/j.compositesb.2016.08.001
704 Cyr, M., Idir, R., Poinot, T., 2012. Properties of inorganic polymer (geopolymer)
705 mortars made of glass cullet. J. Mater. Sci. 47, 2782–2797. doi:10.1007/s10853-
706 011-6107-2
707 Das, B., Prakash, S., Reddy, P.S.R., Misra, V.N., 2007. An overview of utilization of
708 slag and sludge from steel industries. Resour. Conserv. Recycl. 50, 40–57.
709 doi:10.1016/j.resconrec.2006.05.008
710 Deschner, F., Winnefeld, F., Lothenbach, B., Seufert, S., Schwesig, P., Dittrich, S.,
711 Goetz-Neunhoeffer, F., Neubauer, J., 2012. Hydration of Portland cement with
712 high replacement by siliceous fly ash. Cem. Concr. Res. 42, 1389–1400.
713 doi:10.1016/j.cemconres.2012.06.009
714 Donatello, S., Fernández-Jimenez, A., Palomo, A., 2013. Very high volume fly ash
715 cements. Early age hydration study using Na 2SO4 as an activator. J. Am. Ceram.
716 Soc. 96, 900–906. doi:10.1111/jace.12178
717 Donatello, S., Maltseva, O., Fernandez-Jimenez, A., Palomo, A., 2014. The early age
718 hydration reactions of a hybrid cement containing a very high content of coal
719 bottom ash. J. Am. Ceram. Soc. 97, 929–937. doi:10.1111/jace.12751
720 Engelhardt, G., Michel, D., 1987. High-resolution solid-state NMR of silicates and
721 zeolites. Wiley.
722 Fernández-Jiménez, A., Palomo, A., Criado, M., 2005. Microstructure development of
723 alkali-activated fly ash cement: a descriptive model. Cem. Concr. Res. 35, 1204–
724 1209. doi:10.1016/j.cemconres.2004.08.021
725 Fernández-Jiménez, A., Palomo, A., Sobrados, I., Sanz, J., 2006. The role played by the
726 reactive alumina content in the alkaline activation of fly ashes. Microporous
727 Mesoporous Mater. 91, 111–119. doi:10.1016/j.micromeso.2005.11.015
728 Fernández-Jiménez, A., Puertas, F., 2003. Effect of activator mix on the hydration and
729 strength behaviour of alkali-activated slag cements. Adv. Cem. Res. 15, 129–136.
730 doi:10.1680/adcr.2003.15.3.129
731 Fernández-Jiménez, A., Puertas, F., Sobrados, I., Sanz, J., 2003. Structure of Calcium
732 Silicate Hydrates Formed in Alkaline-Activated Slag: Influence of the Type of


733 Alkaline Activator. J. Am. Ceram. Soc. 86, 1389–1394. doi:10.1111/j.1151-

734 2916.2003.tb03481.x
735 Fernández Jiménez, A., Flores, E., Maltseva, O., García-lodeiro, I., Palomo, Á., 2013.
736 Hybrid alkaline cements. Part III. Durability and industrial application. Rom. J.
737 Mater. 43, 195–200.
738 Fletcher, R.A., MacKenzie, K.J.D., Nicholson, C.L., Shimada, S., 2005. The
739 composition range of aluminosilicate geopolymers. J. Eur. Ceram. Soc. 25, 1471–
740 1477. doi:10.1016/j.jeurceramsoc.2004.06.001
741 Garcia-Lodeiro, I., Aparicio-Rebollo, E., Fernández-Jiménez, A., Palomo, A., 2016.
742 Effect of calcium on the alkaline activation of aluminosilicate glass. Ceram. Int.
743 42, 7697–7707. doi:10.1016/j.ceramint.2016.01.184
744 García-Lodeiro, I., Cherfa, N., Zibouche, F., Fernández-Jimenez, A., Palomo, A., 2015.
745 The role of aluminium in alkali-activated bentonites. Mater. Struct. 48, 585–597.
746 doi:10.1617/s11527-014-0447-8
747 Granizo, N., Palomo, A., Fernández-Jiménez, A., 2014. Effect of temperature and
748 alkaline concentration on metakaolin leaching kinetics. Ceram. Int. 40, 8975–8985.
749 doi:10.1016/j.ceramint.2014.02.071
750 Haha, M. Ben, Lothenbach, B., Le Saout, G., Winnefeld, F., 2012. Influence of slag
751 chemistry on the hydration of alkali-activated blast-furnace slag - Part II: Effect of
752 Al2O3. Cem. Concr. Res. 42, 74–83. doi:10.1016/j.cemconres.2011.08.005
753 Hoy, M., Horpibulsuk, S., Arulrajah, A., 2016a. Strength development of Recycled
754 Asphalt Pavement – Fly ash geopolymer as a road construction material. Constr.
755 Build. Mater. 117, 209–219. doi:10.1016/j.conbuildmat.2016.04.136
756 Hoy, M., Horpibulsuk, S., Rachan, R., Chinkulkijniwat, A., Arulrajah, A., 2016b.
757 Recycled asphalt pavement – fly ash geopolymers as a sustainable pavement base
758 material: Strength and toxic leaching investigations. Sci. Total Environ. 573, 19–
759 26. doi:10.1016/j.scitotenv.2016.08.078
760 Ismail, I., Bernal, S.A., Provis, J.L., San Nicolas, R., Hamdan, S., Van Deventer, J.S.J.,
761 2014. Modification of phase evolution in alkali-activated blast furnace slag by the
762 incorporation of fly ash. Cem. Concr. Compos. 45, 125–135.
763 doi:10.1016/j.cemconcomp.2013.09.006
764 Kamseu, E., Beleuk à Moungam, L.M., Cannio, M., Billong, N., Chaysuwan, D., Melo,
765 U.C., Leonelli, C., 2016. Substitution of sodium silicate with rice husk ash-NaOH
766 solution in metakaolin based geopolymer cement concerning reduction in global


767 warming. J. Clean. Prod. 142, 1–11. doi:10.1016/j.jclepro.2016.10.164

768 Koch, A., Steinegger, H., 1960. Ein Schnellprufverfahren fur Zemente auf ihr Verhalten
769 bei Sulfatangriff. Zement-Kalk-Gips 13, 317–324.
770 Li, C., Gong, X.Z., Cui, S.P., Wang, Z.H., Zheng, Y., Chi, B.C., 2011. CO2 Emissions
771 due to Cement Manufacture. Mater. Sci. Forum 685, 181–187.
772 doi:10.4028/www.scientific.net/MSF.685.181
773 Li, H., Xu, W., Yang, X., Wu, J., 2014. Preparation of Portland cement with sugar filter
774 mud as lime-based raw material. J. Clean. Prod. 66, 107–112.
775 doi:10.1016/j.jclepro.2013.11.003
776 Li, X.-G., Lv, Y., Ma, B.-G., Chen, Q.-B., Yin, X.-B., Jian, S.-W., 2012. Utilization of
777 municipal solid waste incineration bottom ash in blended cement. J. Clean. Prod.
778 32, 96–100. doi:10.1016/j.jclepro.2012.03.038
779 Lothenbach, B., Scrivener, K., Hooton, R.D., 2011. Supplementary cementitious
780 materials. Cem. Concr. Res. 41, 1244–1256. doi:10.1016/j.cemconres.2010.12.001
781 Maghool, F., Arulrajah, A., Horpibulsuk, S., Du, Y.-J., 2016. Laboratory Evaluation of
782 Ladle Furnace Slag in Unbound Pavement-Base/Subbase Applications. J. Mater.
783 Civ. Eng. 4016197. doi:10.1061/(ASCE)MT.1943-5533.0001724
784 Martin, A., Pastor, J.Y., Palomo, A., Fernández Jiménez, A., 2015. Mechanical
785 behaviour at high temperature of alkali-activated aluminosilicates (geopolymers).
786 Constr. Build. Mater. 93, 1188–1196. doi:10.1016/j.conbuildmat.2015.04.044
787 Mohammadinia, A., Arulrajah, A., Sanjayan, J., Disfani, M.M., Win Bo, M.,
788 Darmawan, S., 2016. Stabilization of Demolition Materials for Pavement
789 Base/Subbase Applications Using Fly Ash and Slag Geopolymers: Laboratory
790 Investigation. J. Mater. Civ. Eng. 28, 4016033. doi:10.1061/(ASCE)MT.1943-
791 5533.0001526
792 Murayama, N., Yamamoto, H., Shibata, J., Shibata, J., 2002. Mechanism of zeolite
793 synthesis from coal fly ash by alkali hydrothermal reaction. Int. J. Miner. Process.
794 J. Miner. Process 64, 1–17.
795 Noor-ul-Amin, Faisal, M., Muhammad, K., Gul, S., 2015. Synthesis and
796 characterization of geopolymer from bagasse bottom ash, waste of sugar industries
797 and naturally available china clay. J. Clean. Prod. 129, 491–495.
798 doi:10.1016/j.jclepro.2016.04.024
799 Pacheco-Torgal, F., Castrogomes, J., Jalali, S., 2008. Alkali-activated binders: A
800 review. Part 2. About materials and binders manufacture. Constr. Build. Mater. 22,


801 1315–1322. doi:10.1016/j.conbuildmat.2007.03.019

802 Palomo, A., Fernández Jiménez, A., López Hombrados, C., Lleyda, J.L., 2007. Railway
803 sleepers made of alkali activated fly ash concrete. Rev. Ing. construcción 22, 75–
804 80.
805 Palomo, A., Grutzeck, M., 1999. Alkali-activated fly ashes: a cement for the future.
806 Cem. Concr. Res. 29, 1323–1329.
807 Palomo, A., Krivenko, P., Garcia-Lodeiro, I., Kavalerova, E., Maltseva, O., Fernandez-
808 Jimenez, A., 2014. A review on alkaline activation : new analytical perspectives.
809 Mater. Construcción 64, 1–23. doi:e022 http:// dx.doi.org/10.3989/mc.2014.00314.
810 Panagiotopoulou, C., Kontori, E., Perraki, T., Kakali, G., 2007. Dissolution of
811 aluminosilicate minerals and by-products in alkaline media. J. Mater. Sci. 42,
812 2967–2973. doi:10.1007/s10853-006-0531-8
813 Perná, I., Hanzlíček, T., 2014. The solidification of aluminum production waste in
814 geopolymer matrix. J. Clean. Prod. 84, 657–662. doi:10.1016/j.jclepro.2014.04.043
815 Provis, J., Deventer, J. van (Eds.), 2014. Alkali Activated Materials: State-of-the-art
816 Report, RILEM TC 224-AAM. Springer.
817 Querol, X., Moreno, N., Umaa, J.C., Alastuey, A., Hernández, E., López-Soler, A.,
818 Plana, F., 2002. Synthesis of zeolites from coal fly ash: an overview. Int. J. Coal
819 Geol. 50, 413–423. doi:10.1016/S0166-5162(02)00124-6
820 Redden, R., Neithalath, N., 2014. Microstructure, strength, and moisture stability of
821 alkali activated glass powder-based binders. Cem. Concr. Compos. 45, 46–56.
822 doi:10.1016/j.cemconcomp.2013.09.011
823 Rios, S., Cristelo, N., da Fonseca, A.V., Ferreira, C., 2016. Structural performance of
824 alkali-activated soil ash versus soil cement. J. Mater. Civ. Eng. 28.
825 doi:10.1061/(ASCE)MT.1943-5533.0001398
826 Ruiz-Santaquiteria, C., Fernández-Jiménez, A., Skibsted, J., Palomo, A., 2013. Clay
827 reactivity: Production of alkali activated cements. Appl. Clay Sci. 73, 11–16.
828 doi:10.1016/j.clay.2012.10.012
829 Ruiz-Santaquiteria, C., Skibsted, J., Fernández-Jiménez, A., Palomo, A., 2012. Alkaline
830 solution/binder ratio as a determining factor in the alkaline activation of
831 aluminosilicates. Cem. Concr. Res. 42, 1242–1251.
832 doi:10.1016/j.cemconres.2012.05.019
833 Shi, C., Fernández-Jiménez, A., 2006. Stabilization/solidification of hazardous and
834 radioactive wastes with alkali-activated cements. J. Hazard. Mater. 137, 1656–63.


835 doi:10.1016/j.jhazmat.2006.05.008
836 Silva, P. De, Sagoe-Crenstil, K., Sirivivatnanon, V., 2007. Kinetics of
837 geopolymerization: Role of Al2O3 and SiO2. Cem. Concr. Res. 37, 512–518.
838 doi:10.1016/j.cemconres.2007.01.003
839 Takeda, H., Hashimoto, S., Yokoyama, H., Honda, S., Iwamoto, Y., 2013.
840 Characterization of Zeolite in Zeolite-Geopolymer Hybrid Bulk Materials Derived
841 from Kaolinitic Clays. Materials (Basel). 6, 1767–1778. doi:10.3390/ma6051767
842 Torres-Carrasco, M., Puertas, F., 2015. Waste glass in the geopolymer preparation.
843 Mechanical and microstructural characterisation. J. Clean. Prod. 90, 397–408.
844 doi:10.1016/j.jclepro.2014.11.074
845 Tsakiridis, P.E., Papadimitriou, G.D., Tsivilis, S., Koroneos, C., 2008. Utilization of
846 steel slag for Portland cement clinker production. J. Hazard. Mater. 152, 805–811.
847 doi:10.1016/j.jhazmat.2007.07.093
848 Turner, L.K., Collins, F.G., 2013. Carbon dioxide equivalent (CO2-e) emissions: A
849 comparison between geopolymer and OPC cement concrete. Constr. Build. Mater.
850 43, 125–130. doi:10.1016/j.conbuildmat.2013.01.023
851 UNE 80103, 2013. UNE 80103:2013 - Metodos de ensayo de cementos. Ensayos
852 físicos. Determinación de la densidade real. AENOR - Asoc. Española Norm. y
853 Certificación.
854 van Deventer, J.S.J., Provis, J.L., Duxson, P., Lukey, G.C., 2007. Reaction mechanisms
855 in the geopolymeric conversion of inorganic waste to useful products. J. Hazard.
856 Mater. 139, 506–513. doi:10.1016/j.jhazmat.2006.02.044
857 van Riessen, A., Jamieson, E., Kealley, C.S., Hart, R.D., Williams, R.P., 2013. Bayer-
858 geopolymers: An exploration of synergy between the alumina and geopolymer
859 industries. Cem. Concr. Compos. 41, 29–33.
860 doi:10.1016/j.cemconcomp.2013.04.010
861 Xu, H., van Deventer, J., 2000. The geopolymerisation of alumino-silicate minerals. Int.
862 J. Miner. Process.
863 Zhuang, X.Y., Chen, L., Komarneni, S., Zhou, C.H., Tong, D.S., Yang, H.M., Yu,
864 W.H., Wang, H., 2015. Fly ash-based geopolymer: Clean production, properties
865 and applications. J. Clean. Prod. 125, 253–267. doi:10.1016/j.jclepro.2016.03.019





 Potential of an industrial residue (cleaning solution, CS) as an alkaline activator of fly

ash (FA) or glass powder (GP).

 Lower alkalinity of the CS (5M), relatively to the reference 8M NaOH, produces at

early age’s lower strength (with FA).

 Strength values of the FA activated with CS and 8M NaOH tend to be very similar at

28 days.

 The Al (IV) content of the CS allows it to act also as an Al corrector, which is

especially significant with GP.