Peracetic acid as a novel blowing agent in the direct foaming of alkali-activated materials

In this study, peracetic acid (PAA) was investigated as a novel blowing agent for the preparation of porous alkali-activated materials from metakaolin or blast furnace slag by the direct foaming method. PAA is an organic peroxide with lower stability of the O – O bond in comparison to H 2 O 2 . It also introduces acetate anions to the system, which can chelate cations to potentially increase the extent of precursor dissolution and decrease the surface tension of the gas-liquid interfaces prompting open porosity formation. PAA was compared with H 2 O 2 to prepare porous alkali-activated materials by characterizing setting time, compressive strength, reaction kinetics (by isothermal calorimetry), porosity (by mercury intrusion porosimetry and helium gas pycnometry), and chemical composition (by electron probe microanalyzer, EPMA). The most significant finding was that the use of PAA caused 92% – 94% and 72% – 80% lower volume increase compared to H 2 O 2 upon curing for metakaolin and blast furnace slag-based foams, respectively. The enhanced dissolution of the precursors when using PAA were observed from the EPMA analysis, which promoted higher compressive strength for PAA-based foams (451 – 537% compared to H 2 O 2 -based foams). There was also an indication that the use of PAA promoted the open porosity formation. As a practical implication, PAA could be well suited for applications in which the volume increase during the foam setting must be minimized.


Introduction
Alkali-activated materials (AAMs), and geopolymer as their subgroup, are actively investigated due to their mechanical and physicochemical properties as well as potentially low cost, simple, and green preparation process.The major application of AAMs is in the construction and building industry where they are utilized as low-carbon binders (Luukkonen et al., 2018).They are produced from the reaction of amorphous or semicrystalline aluminosilicate precursors with alkali activators at low temperatures, usually < 100 • C, resulting in lower energy consumption in comparison to many alternative materials (Provis and van Deventer, 2014;Provis and Bernal, 2014;Davidovits, 2017).The alkali activation of aluminosilicate precursors generates Si-O-Al structures where tetrahedral SiO 4 and AlO 4 are connected alternately by sharing oxygen atoms (Davidovits, 1994).The aluminosilicate precursors include natural minerals and industrial side-streams, such as calcined 1:1 or 2:1 clays (e.g., metakaolin), volcanic tuffs, fly ashes, metallurgical slags, red mud, rice husk ash, or a combination of these (Badanoiu et al., 2015;Ducman and Korat, 2016;Gado et al., 2020;Wang et al., 2020).The widely used alkali activators are sodium or potassium hydroxides and/or silicates (Abdullah et al., 2018).The structure of AAMs is greatly dependent on the amount of calcium (or other network-modifying cations) present in the system: a low calcium content results in three-dimensional structures and the obtained materials are called geopolymers (Provis and Van Deventer, 2009), while a high calcium content causes partially cross-linked chain structures resembling tobermorite to form (Provis and van Deventer, 2013).
In recent years, the fabrication of alkali-activated foams (AAFs) has received significant attention because of their promising properties and versatile preparation processes (Novais et al., 2020;Bai and Colombo, 2018).The AAFs have been studied for adsorption, catalyst, membrane support, thermal insulation, and sound absorption applications, to name just a few examples (Bai and Colombo, 2017;Novais et al., 2020;Zhang et al., 2020;Bhuyan et al., 2022).The pore network of the AAFs provides high surface area, improves the water permeability, and enhances the mass transfer of soluble species within the material (Papa et al., 2022a).Furthermore, AAFs have good mechanical properties and regenerable adsorption sites.To fabricate AAFs, different methods are used, such as direct foaming, additive manufacturing, sacrificial filler, emulsion templating, or suspension-solidification (Bai and Colombo, 2018;Wang et al., 2022).Among these approaches, the most used one is direct foaming, which enables the generation of various pore sizes and porosities (Bai and Colombo, 2018).There, the porosity is generated by using blowing agents, such as hydrogen peroxide, metallic aluminum, sodium perborate, elemental silicon, or sodium hypochlorite, which all decompose at high pH and produce gas bubbles (Masi et al., 2014;Abdollahnejad et al., 2015;Böke et al., 2015;Kamseu et al., 2015;Luukkonen et al., 2020;Bhuyan et al., 2022).The produced gas bubbles are trapped inside the material during the consolidation, creating the pore structure.However, all of the blowing agents cause a large volume increase before the materials hardens.For example, hydrogen peroxide decomposition can more than double the initial volume of the fresh-state paste (Bhuyan et al., 2022).The volume increase causes practical problems such as the need to remove the expanded parts for the final application.To overcome this problem, it would be beneficial to introduce an alternative blowing agent which will form the porosity with a minimal volume increase.
Peracetic acid (PAA, CH 3 COOOH), an organic peroxide, is commercially available as an aqueous equilibrium solution containing PAA, acetic acid (AA), and hydrogen peroxide (Zhao et al., 2007).PAA is commonly used as an oxidizer, disinfectant, or sterilizer in industries such as pharmaceutical, chemical, textile, pulp and paper, food processing, or wastewater treatment (Kim and Huang, 2021).The possibility of using organic peroxides as a blowing agent is briefly mentioned by Davidovits (2020) but no further information is available.To the authors' best knowledge, PAA has not been used as a blowing agent for AAFs before.It is known that PAA decomposes into oxygen similarly as H 2 O 2 when exposed to highly alkaline conditions (Yuan et al., 1997).However, PAA is less stable as reflected by the oxygen-oxygen bond energy of 38 kcal/mol vs. 51 kcal/mol in the case of H 2 O 2 (Bach et al., 1996;Bianchini et al., 2002).Therefore, the gas formation in alkaliactivated paste is expected to occur with faster kinetics when using PAA.Another potential benefit of using PAA is the introduction of acetate anions to the mix design, which can act as chelating agents to, for example, calcium, and thus improve the dissolution of certain raw materials leading to a higher alkali-activation reaction extent (Smith et al., 2005).PAA can also affect the surface tension at the gas-liquid interface (Minofar et al., 2006) resulting in potentially a higher level of open porosity in the foam similarly as other surfactants.Thus, in the present work, the objectives were to study the suitability of PAA as a multifunction blowing agent for the preparation of AAFs with high-and low-Ca mix designs and to characterize the AAFs for fresh and hardened properties, chemical composition, and pore morphology to observe the possible benefits resulting from the use of PAA in comparison to hydrogen peroxide.

Preparation of the foam
The direct foaming method (Luukkonen et al., 2020;Bhuyan et al., 2022) was applied to prepare the foams by mixing precursor (BFS or MK), alkali activator solution, and water using a high shear mixer (IKA EUROSTAR 20) for 4 min (speed 3000 rpm).Next, PAA or H 2 O 2 , with or without Triton X-100, was added and mixing was continued for a further 2 min.After that, the sample was cast in a 50 × 50 × 50 mm 3 steel mold and cured at 60 • C for 4 h in an airtight bag.After curing, the samples were preserved in an airtight bag at room temperature.The following theoretical molar ratios were used for the BFS foam: SiO 2 /Al 2 O 3 = 7.00, Na 2 O/SiO 2 = 0.14, Na 2 O/Al 2 O 3 = 0.96, and H 2 O/Na 2 O = 18.13 and for the MK foam: SiO 2 /Al 2 O 3 = 3.24, Na 2 O/SiO 2 = 0.27, Na 2 O/Al 2 O 3 = 0.87, and H 2 O/Na 2 O = 11.76.The weight ratio of precursor and alkali activator solution in the mix design for BFS foam was 3.4 and for MK foam was 0.73.The Na 2 O weight-% of BFS and MK foam was ~5% and ~23%, respectively.The amounts of PAA and H 2 O 2 (Table 1) were selected based on earlier publications (Luukkonen et al., 2020;Bhuyan et al., 2022) to result in an equimolar amount of total peroxides (i.e., the sum of PAA and H 2 O 2 ), and thus also for an equimolar oxygen gas evolution (i.e., ½ mol of O 2 produced from 1 mol of PAA or H 2 O 2 ).

Mechanical strength
The compressive strength of the prepared foams, at 7 d after preparation, was studied by using ZwickRoell Z10 (10 kN) testing machine with a loading rate of 1 mm/min.The dimensions of the samples were 50 mm × 50 mm × 25 mm (area under compression was 50 × 50 mm 2 ) and three specimens were used for each test.The force to calculate compressive strength was considered at the onset of crushing point when the first pore walls are broken (Scheffler and Colombo, 2006).Eq. 1 was used to calculate the compressive strength (σ, [MPa]).
where F is the force (N), and A is the surface area (mm 2 ) under compression.

Specific surface area and nanoscale porosity
The Brunauer-Emmet-Teller (BET) isotherm and the Barrett-Joyner-Halenda (BJH) method were employed to determine the specific surface area and pore volume of the foam samples by adsorption/desorption of liquid N 2 at − 196 • C. The measurements were conducted using a Micromeritics ASAP 2020 instrument.For this characterization, freeze-dried (lyophilization) foam pieces with a particle size of 4-8 mm were used.The pieces were freeze-dried for 72 h.

Volume increase
The volume increase of the foam samples was carried out by pouring a fresh-state alkali-activated slurry (initial volume was recorded) into a 50 mL graduated plastic tube, and the volume after curing the samples at 60 • C for 4 h was recorded as the final volume.The volume increase was calculated from the difference between the initial and final volume.

Porosity of the foam samples analysis by He-gas pycnometer
To assess the total and open porosity of the foam samples, a helium gas pycnometer (AccuPyc II 1340, Micromeritics) was employed.The rectangular prismatic foam samples were dried at 60 • C until a constant mass was reached.After drying, the samples were stored in a desiccator to prevent moisture adsorption.Sample dimensions were measured using a caliper and geometric density was calculated (ρ g [g/cm 3 ]).Crushed foam pieces with dimensions of < 1 cm and pulverized samples were used to study the apparent densities [ρ a (g/cm 3 )] and true densities [ρ t (g/cm 3 )] by the helium gas pycnometer, respectively.Eqs. 2 and 3 were used to calculate the open and total porosities, respectively.

Porosity, pore size distribution, and pore volume analysis by mercury intrusion porosimetry
To better understand the pore size distribution of foam samples, Micromeritics AutoPore IV 9500 mercury porosimeter was used.The mercury intrusion porosimetry, MIP, is based on the Washburn equation and assumes cylindrical pore geometry.This technique is commonly employed to determine the pore structure of meso-and macropores.However, a limitation of this technique is that when using pressure required to penetrate the micropores, the accuracy of this technique is reduced and the internal structure of the foam sample may be damaged.The samples were initially dried in an oven for around 24 h at 75 • C and placed in a desiccator to prevent moisture re-absorption.There were two stages in the analysis: (1) low pressure (up to 50 psi) to examine pores ranging from 4 nm to 200 μm and (2) high pressure (up to 60 000 psi) to examine pores ranging from 4 nm to 10 μm.The Washburn equation (Eq.4) was employed to identify the relation between applied pressure, P, (psi), and pore diameter, D, (nm).The surface tension (σ) and contact angle (θ) were 0.485 N/m and 130 • , respectively, during the analysis.
Detailed information about the pore diameter and volume can be obtained by drawing graphs of the log differential pore volume (dV/dlog D, cm 3 /g, the first derivative of the cumulative pore volme curve) as a function of the pore diameter (nm).Hereby, V stands for pore volume (typically in cm 3 /g) and D indicates an average pore diameter (in nm) in that range.The pore size distribution data was determined by considering the pore size ratios of various pore area classes (i.e., micropores [0-2 nm], mesopores [2-50 nm], and macropores [> 50 nm].

Surface tension
The surface tensions of PAA and corresponding acetic acid solutions were determined with the Du Noüy ring method by using a Pt ring and a Krüss tensiometer.The measurement was conducted at 20 • C ± 1.5 • C. The accuracy of the tensiometer was ±0.1 mN/m.The tensiometer was calibrated with distilled water at 20 • C ± 1.5 • C to determine the correction factor (Eq. 5) for measurement, in which a is the surface tension of water (72.74 mN/m (Rumble, 2021)) and b is the measured value (mN/m).Each acetic acid or PAA solution was measured at least three times and the standard deviation of replicates was 0.1 mN/m.q = a b (5)

Setting time of the foam samples
The hardening of the foams was assessed via setting time measurements by using a Vicat apparatus (Vicatronic Matest).The measurement of a foam sample with this method is not accurate due to the presence of pores, and thus the results should be interpreted as semi-quantitative.For this experiment, the foam sample was cast into a mold of 80 mm diameter and 40 mm height.The sample was placed in an oven of 60 • C and taken out every 3 min and placed in the Vicat apparatus.The penetration depth of a 1 mm diameter needle was recorded as a function of time.

Isothermal calorimetry
More quantitative information about the kinetics of the setting and blowing agent decomposition was obtained with isothermal calorimetry using a TAM-Air isothermal calorimeter (TA Instruments).For this characterization, the precursor and solid sodium metasilicate were placed in a 20 mL glass admix ampoule, and water and blowing agent (PAA or H 2 O 2 ) were loaded into the syringes of the admix ampoule.Then, the sample was placed in the calorimeter and when the thermal equilibrium was reached, water was injected into the vial, and mixing, as well as the data collection, was started.After 10 min, the blowing agent was injected, and mixing was continued for 5 min.The heat release rate values were normalized by the mass of MK or BFS and solid sodium silicate.Representative amount of deionized water was used as a reference sample in the calorimeter.

Electron probe microanalyzer and optical microscopy
Elemental maps of foam cross-sections were obtained by using a Jeol JXA-8530F Plus field emission electron probe microanalyzer (EPMA), which was equipped with wavelength or energy dispersive detectors.The specimen was prepared by embedding a foam piece into epoxy resin (Buehler, EpoxiCure 2 Epoxy Resin and Epoxy Hardener), polished by using silicon carbide grinding papers (P240, P600, and P1200), and coated with carbon.The setting parameters of the probe were as follows: focused spot, an accelerating voltage of 15 kV, a current of 20 nA, a step size of 0.5 μm, and a dwell time of 500 ms.Elemental mapping was done for the following elements: Na, Mg, Al, Si, Fe, Mn, Ca, K, S, and Ti.
Leica DFC420 CAMERA (Leica MZ6) was used to capture the optical microscopic images for pore morphology characterization.

Mechanical strength
The compressive strength resulting from using PAA as a blowing agent was higher than with H 2 O 2 for both MK and BFS precursors (Fig. 1).This result is in line with the observed open and total porosities, which were higher when using H 2 O 2 in comparison to PAA (Table 2).Moreover, Fig. 3 indicates that especially the volume of large pores (> 10 000 nm) is lower with PAA compared to H 2 O 2 (with both MK and BFS), also explaining the higher compressive strength for PAA-based foams.There could be one more factor contributing to the higher compressive strength of foams prepared with PAA: the acetate ion (CH 3 COO − ), which can chelate Ca and Mg (Smith et al., 2005;Jun et al., 2017) present in BFS, and thus potentially increase the dissolution of BFS and subsequently contribute to the increased mechanical strength.
When comparing the precursors with a similar blowing agent and surfactant combination, the mechanical strength of MK foams was always lower compared to the BFS foams.This is due to the differences in the structure of resulting gels, such as the SiO 2 /Al 2 O 3 ratios (Santana et al., 2021).The higher porosity of MK foams compared to BFS foams (Table 2) also contributed to the lower mechanical strength of the former.The effect of Triton X-100 surfactant on the mechanical strength of the foams appeared to be minor.

Pore structures and specific surface areas of the foam materials
The open and total porosities and the specific surface areas are shown in Table 2.In all cases, the open and total porosities were approximately equal, which inferred that the pores were interconnected.This result also implied that PAA was able to promote the formation of open porosity similarly as Triton X-100.For both precursors, PAA generated systematically lower porosity compared to the H 2 O 2 when analyzed by He gas pycnometry.However, MIP indicated a contradicting result for MK-based foams, which might be due to the inaccessibility of mercury to micropores.Thus, it appears that the combination of MK as precursor, H 2 O 2 , and Triton X-100 produced very small pores (accessible to He but not to Hg), which is also evident from Fig. 2.However, with PAA, Triton X-100 did not cause a similar effect.
The explanation for the overall lower porosity with PAA was likely related to the lower stability of PAA at alkaline conditions (i.e., the lower energy related to the O -O bond in PAA in comparison to H 2 O 2 (Bach et al., 1996;Bianchini et al., 2002)), and thus a faster decomposition into oxygen bubbles.As PAA accelerated the hardening of the foam (see Figs. 5 and 6) at the same time, the bubbles were likely trapped in the matrix immediately upon their formation without the possibility for coalescence, and they remained relatively small.
When comparing the MK and BFS-based foams, the setting of MKbased foams was slower (Fig. 6), enabling the paste to remain in a fluid state for longer, and the oxygen gas bubbles have more time to grow resulting in more porosity.Thus, the MK-based foams had higher porosities compared to BFS-based foams.Another explaining factor could be the difference in the viscosities of the pastes.
The cumulative pore volumes of the foams as a function of pore diameter determined from the N 2 desorption data are illustrated in Fig. 2. Most of the pore volume for all the foams was in the range of 2-30 nm pore diameter (i.e., mesopores).The cumulative pore volumes for MK with H 2 O 2 were higher compared to the cumulative pore volumes for MK with PAA.On the other hand, the cumulative pore volumes for BFS with H 2 O 2 and PAA were very similar.The cumulative pore volume for MK foam with H 2 O 2 is approximately similar to the previously reported value for MK foams with the same surfactant (Luukkonen et al., 2020).The cumulative pore volumes for BFS foams with either PAA or H 2 O 2 were very small.The MIP method was used to identify the pore structures of the samples mainly in diameters larger than 50 nm.Figs.3A and B depict the pore size distributions of the various foams.H 2 O 2 formed mesopores in both MK and BFS-containing foams and the pore size distribution ranged from 5 nm to 500,000 nm.MK-and BFS-based foams have more pores smaller than 100 nm and larger than 10 000 nm diameter when using H 2 O 2 .However, with both MK and BFS, the mid-range, 100-10 000 nm, is formed to a greater extent when using PAA.The BFS with PAA is an exception, as it contains a small amount of all pore sizes, and the cumulative graph (Fig. 3D) is approximately linear.The addition of Triton X-100 surfactant to mixtures containing PAA enhanced the cumulative pore volume of mixtures containing BFS, while the cumulative pore

Table 3
Effect of PAA or acetic acid (AA) on the surface tension of the water-air interface.The measurements were conducted at 20  volume of samples containing MK was reduced.
The alkali activation itself was mainly responsible for the formation of micropores as a resulf of the expulsion of water in polycondensation reactions (Zhang et al., 2012;Papa et al., 2022b).The larger pores were formed due to the foaming reaction.The MIP results regarding the micro-and mesopores are in agreement with the BJH data (Fig. 2): those pore sizes are largely missing when PAA is used instead of H 2 O 2 .
The cumulative pore volumes in the meso-and macropore regions for MK-and BFS-based foams are shown in Figs.3C and D. There is one major zone in the mesopores region with pore diameters of 2-50 nm and two major zones in the macropores region with pore diameters of 50-1000 nm (smaller macropores) and above 1000 nm (larger macropores).The MK-containing samples had the following cumulative pore volumes in the mesopores region: 0.49, 0.46, and 0.44 cm 3 /g for MK-PAA, MK-H 2 O 2 -TX100, and MK-PAA-TX100, respectively.As evident also from Fig. 3A, samples MK-PAA, MK-PAA-TX100, and MK-H 2 O 2 -TX100 had the highest cumulative pore volume in the 50-500 nm range (i.e., 0.45, 0.45, and 0.27 cm 3 /g, respectively).

Volume increase
The volume increase resulting from the decomposition of PAA or H 2 O 2 into gas bubbles is presented in Fig. 4. A noticeably lower volume increase was observed for PAA in comparison to H 2 O 2 for both MK and BFS even though the evolution of oxygen gas should be theoretically similar.As already mentioned, PAA is less stable compared to H 2 O 2 and as a result, PAA decomposes faster (Bach et al., 1996).At the same time, PAA accelerated the setting of foam (see section 3.5, Fig. 6), and thus the gas bubbles are trapped at the location of their formation.These two combined effects likely contributed to the lower volume increase with PAA.On the other hand, the longer setting time for H 2 O 2 (shown in Fig. 6) created the possibility of volume expansion.For both precursors, the Triton X-100 surfactant also affected the volume increase.The volume increase was slightly higher (~7-9%) for both MK and BFS with surfactant than without surfactant.The surfactant changes the freshstate paste rheology, for example, due to a decrease in surface tension of the liquid-gas interface that would possibly contribute to a slightly higher volume increase for both the precursors.

Surface tension of the water-air interface
The effect of PAA or acetic acid on the surface tension of the waterair interface is presented in Table 3.The concentration range of the PAA or acetic acid (20 000-80 000 ppm) was selected to cover the range applied in the present study.There was a clear decrease in the surface tension upon using PAA or acetic acid.This result implies that PAA/ acetic acid has surfactant properties, and it could promote the formation of connected pores (open porosity), which is also supported by the fact that the total porosity equaled the open porosity when using PAA without surfactant similarly to the sample containing Triton X-100 surfactant (Table 3).

Isothermal calorimetry
The normalized heat flow curves for MK and BFS foams based on H 2 O 2 or PAA are shown in Fig. 5.The intensity of the initial peak (from the reaction of sodium metasilicate with H 2 O) for the MK sample was higher compared to the BFS sample due to a larger amount of sodium metasilicate.For both MK and BFS, the initial reaction resulted in a higher extent of heat flow when using PAA in comparison to H 2 O 2 possibly due to the neutralization reaction of alkali activator with PAA or acetic acid.The second peak after the addition of H 2 O 2 or PAA is related to nucleation, growth, and precipitation of (C,N)-(A)-S-H gels (Torres-Carrasco and Puertas, 2017; Criado et al., 2018).The intensity of the second peak for both precursors was again clearly higher (especially for BFS as a high Ca-and Mg-content precursor) for PAA compared to H 2 O 2 , which likely reflects a higher dissolution of the precursor.As discussed above, the acetate anion can chelate, for example, Mg and Ca.In terms of reaction kinetics, both the first and second peaks appeared more quickly when using PAA in comparison to H 2 O 2 , which was also confirmed by the setting characterization shown in Fig. 6.
It has been shown that at a pH of 10.5 or higher, which were the conditions of the present work, the predominant mechanisms for PAA decomposition are hydrolysis and metal (M n+ ) catalyzed decomposition according to the reactions 6 and 7, respectively (Yuan et al., 1997;Zhao et al., 2007).
On the other hand, it has been reported that increasing the amount of suspended solids increases the PAA decay rate (Domínguez Henao et al., 2018).In the present study, the amount of suspended solids was much higher than in the studies addressing wastewater treatment with PAA.Likely, PAA decomposed predominantly directly in the oxygen without first forming H 2 O 2 since the foam morphology, volume increase behavior, and the heat evolution were completely different when comparing PAA-based foams to the foams prepared with H 2 O 2 .

Setting characterization
The setting characterization for MK and BFS foams with PAA or H 2 O 2 is presented in Fig. 6.The PAA shortened the setting times for both MK and BFS compared to H 2 O 2 .Here again, the reasons are likely similar as outlined earlier: the acetate ion can chelate Mg and Ca resulting in a faster setting (Smith et al., 2005).In the case of BFS, the decrease of setting time in the presence of PAA was more noticeable, since BFS contains higher amount of Ca and Mg.The addition of surfactant did not affect the setting significantly when PAA was used.

Microstructure of the foam materials
The optical microscopy images of MK and BFS foams with H 2 O 2 or PAA are presented in Fig. 7.The differences in the porosity and μmto mm-scale pore size distributions between the samples are noticeable: PAA has formed smaller pores, especially in the case of MK as a precursor.
The elemental maps from EPMA are presented in Fig. 8.The impact of PAA on MK-based foams was not dramatically different when compared to H 2 O 2 .However, it appeared that the concentration of Al when using H 2 O 2 was possibly more evenly distributed in comparison to PAA.In contrast, PAA had a drastic impact on the distribution of elements for BFS-based foams (Fig. 9).The concentrations of Al, Si, Ca, and Mg are much more evenly distributed across the sample cross-section when using PAA, thus indicating a higher extent of dissolution of BFS.On the other hand, the concentration of Na in the bulk binder phase is very low in comparison to using H 2 O 2 .That might indicate that Na + has been possibly replaced by H + from PAA or acetic acid as the chargebalancing cation.

Conclusion
In this study, the suitability of PAA as a blowing agent for alkaliactivated materials and geopolymers was evaluated for the first time.When compared with H 2 O 2 at an equimolar total peroxide dose (i.e., the dose of H 2 O 2 being the same as the combined H 2 O 2 and PAA), several differences were observed.The most interesting observed property was that the volume increase of the foam during hardening decreased significantly (up to 91-94% and 72-78% when using MK and BFS as the precursor, respectively) when compared to H 2 O 2 .The PAA also accelerated the setting of the foam, which is in line with the earlier reported capability of acetate ions to complexate Ca and Mg, for instance, and thus increase the dissolution rate.The resulting total and open porosity with PAA was smaller in comparison to H 2 O 2 , but PAA was able to promote the open porosity similarly to the studied reference surfactant (Triton X-100).This was due to the surfactant properties of PAA, which were confirmed with air-water surface tension measurements.The specific surface area of foams was lower with PAA.The analysis of pore size distributions by mercury intrusion porosimetry revealed that when PAA is used instead of H 2 O 2 , the pores smaller than 100 nm essentially disappear, while porosity in the area at 100-10 000 nm is formed.The number of pores larger than that also decreases.In line with the lower porosity but also due to the higher dissolution of the precursor, PAA generated higher compressive strength both for MK and BFS precursors with or without additional surfactant.
This study was a preliminary work reporting a novel blowing agent for the direct foaming method to produce porous alkali-activated materials.PAA could be a useful blowing agent, especially in the applications requiring as low volume increase as possible during the foam formation, such as with many construction materials, acoustic and thermal insulation materials, catalyst supports, or filter media in separation processes.Also, from the cost point of view, the introduction of PAA (~1200 €/t) instead of H 2 O 2 (~450 €/t) does not significantly affect the overall cost of the foamed material since the blowing agent constitutes only a small fraction of the total mass.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1.Comparison of mechanical strength of BFS and MK foams with PAA or H 2 O 2 as blowing agents.

Fig. 3 .Fig. 4 .
Fig. 3.The log differential pore volume (dV/dlog D) as a function of pore diameter for (A) MK and (B) BFS-containing samples and pore diameter vs. cumulative pore volume for (C) MK and (D) BFS-containing samples from MIP analysis.

Fig. 5 .
Fig. 5. Comparison of heat generation with PAA or H 2 O 2 and MK-or BFSbased pastes.

Table 1
Alkali-activated foams prepared by using two different blowing agents with or without additional surfactant.
M.A.H.Bhuyan et al.

Table 2
Porosities and specific surface areas of different foams.
Fig. 2. Cumulative pore volumes of the foams as a function of pore diameter determined by the BJH method.M.A.H.Bhuyan et al.