Influence of brick flour or concrete rubble on the mechanical properties of fly-ash-based geopolymers
In the work presented, we investigate what influence the increasing addition of clay brick scrap and concrete rubble has on the setting behaviour and material properties of fly-ash-based geopolymers. The prepared geopolymers are tested to ascertain their compressive strength, apparent density, thermal conductivity and other properties. To meaningfully correlate the material characteristics with the setting behaviour and structures being formed, both the starting materials and the resulting binders are analysed by means of infrared spectroscopy, X-ray diffraction analysis and scanning electron microscopy. In the analyses, it could be shown that both brick scrap and concrete rubble are sufficiently suitable as matrix-forming raw materials for the production of geopolymers. At the same time, different material properties could be determined and attributed to different setting mechanisms and strength-forming mechanisms.
Cement is one of the most commonly used construction materials in the world today, with 4.1 bill. tonnes of it being used globally in 2020 . From an ecological perspective, this is particularly problematic. Cement, which is an elementary component of concrete, consists mainly of primary raw materials like limestone, chalk and clay, which are generally extracted at surface mines.
The extracted primary raw materials must be processed, by means of crushing, grinding and mixing to raw meal, which is then fired to cement clinker at 1 450 °C . The CO2 emissions of the energy necessary for thermal and electric processes make up 40 %, another 60 % are caused by calcination, during which chemically bonded CO2 is released. In this way, for the production of cement internationally, around 900 kg CO2 is emitted per tonne of cement clinker . In highly modern German production plants, 791 kg CO2 per tonne are still released. The production of Portland cement alone, with 0.6 to 0.8 kg CO2 per kg cement clinker accounts for 8 % of global CO2 emissions . In Germany, the cement industry is responsible for 19.99 mill.t CO2-equivalent, which, relative to the total emissions of the German industry sector, corresponds to a share of 17 % .
To achieve the climate action goals, already during their production, the construction materials of the future should boast lower energy consumption, reduced CO2 emissions, a feasible percentage of recycled input materials and recyclability into the materials cycle. The high volumes of construction and demolition waste in the construction sector present a serious environmental problem. The global volume of construction waste makes up around 25 to 30 % of all solid waste . In Germany alone, over 228 mill. tonnes construction and demolition waste were produced in 2018 . It is in global interest to bring the construction industry in harmony with its environment by saving energy and raw material resources, ensuring less CO2 is emitted and recycling construction and demolition waste.
The construction materials industry can get closer to achieving these goals by promoting the circular economy based on the processing of suitable construction and demolition waste and their recycling to new types of substitute construction materials. At the same time, value landfill space can be saved. One possibility to increase the recycling of construction waste is the technology of alkali activation or geopolymerization of mineral secondary resources [8, 9]. J. Davidovits, who investigated geopolymers on the basis of metakaolin, is regarded as the founder of this novel class of materials [10, 11]. According to current knowledge, geopolymerization is a complex multiphase reaction process divided into three main steps:
1. The partial and full dissolution of silicates and aluminosilicates e.g. in metakaolin or fly ash caused by the break-up of Si-O-Si or Si-O-Al bonds in alkaline solution.
2. The accumulation phase in which the silicate and aluminate tetrahedra alternately bond with each other as a result of condensation reactions, forming sialate monomers .
3. The cross-linking phase in which the entire system is transformed into an inorganic, three-dimensional network, the geopolymer [13–15].
Geopolymers can differ widely in their chemical composition and structure. Depending on their calcium content, they can be assigned to two categories as this has a major influence on the structure of the alkali-activated binder being formed.
To activate low-calcium starting materials like fly ash or brick flour, relatively high pH values of the activator solutions are necessary to start the reaction. The gel structure that would form in a low-calcium geopolymer can be assumed to be a disordered zeolite-like aluminosilicate structure . Here, the oxidic-polyhedral network structures only form statistically in certain circumstances, but follow Pauling’s rules . For Si and Al cations with tetrahedrally arranged oxygen anions, this means:
- Every cation forms a coordination polyhedron from anions (Rule 1).
- Charge neutrality prevails over the entire network (Rule 2).
- The coordination tetrahedra are linked via common corners. Share edges or surfaces destabilize the structure, as the objective must always be maximized distance between the positive-charged cations (Rule 3 and 4).
Paulings’ axioms are extended with the Loewenstein Rule, which says that Al-O-Al bonds are thermodynamically unfavourable in tetrahedra structures and therefore occur with low probability. As long the Si / Al ratio is larger than one, the Loewenstein rule is followed in geopolymers. This thermodynamic preference can be attributed to the large radius of the cation in the aluminate tetrahedra with the simultaneously low coordination number four. As a result, the oxygen bridge bonds between aluminate tetrahedra would have to bridge a relatively large distance, while the tetrahedra with their combined negative charge tend to repel each other. For this reason, cation spaces around an aluminate tetrahedron are mainly occupied by relatively small silicon atoms and their tetrahedra or polyhedra with central cations of a higher coordination number . This behaviour also explains why zeolites frequently occur as by-products in low-calcium geopolymers. It can therefore be speculated that the structure of the geopolymer gels is similar to that of zeolites, as already suggested by Davidovits . This speculation could be proven in various works by means of magic angle spinning nuclear magnetic resonance (MAS NMR) spectroscopy [19–22, 19, 23].
A further structural feature of low-calcium geopolymers are alkali cations outside the basic structure, which offset the negative net charges generated by the aluminate tetrahedra. If, as in the work described here, caustic soda is used to activate low-calcium precursors, accordingly a sodium alumino-silicate-hydrate gel (N-A-S-H) is formed with a very strongly cross-linked zeolite-like structure [24, 25]. In high-calcium geopolymers, with slight differences, a similar structure forms with somewhat other properties. A crucial difference is that high-calcium precursors, like, for example, granulated slag sand or concrete rubble can already be activated by activators with much lower pH value. Accordingly, high-calcium geopolymers can be produced with a much wider range of activator solutions [26, 27]. With the activation of high-calcium precursors by means of alkaline hydroxide solutions, an aluminium-substituted calcium silicate hydrate gel (C-A-S-H) is formed [24, 28]. This gel is less similar to the zeolite structures, but exhibits a more tobermorite-like structure, as also found in the gels of Portland cement hydration.
However, high-calcium geopolymers show a lower calcium concentration than the Portland cement gels and a stronger Al substitution of tetrahedra of the three-chain type . This leads to a much higher degree of polymerization and cross-linking between tobermorite chains, which was proven based on the occurrence of Q3 groups . As soon as the Al content of the C-A-S-H gel exceeds a certain limit value, which depending on the chain length lies between six and ten chain places, the degree of cross linking decreases with further Al incorporation. This can be attributed to the previously described Loewenstein rule, according to which Al-O-Al bonds are not formed . In the C-A-S-H gel, too, considerable quantities of sodium ions are embedded, which is why it is often described as C-(N)-A-S-H-gel [30, 31]. At the same time, more strongly cross-linked N-A-S-H gels are present as by-products in the forming binders. This applies particularly when mixes of low-calcium and high-calcium precursors are used, as is the case in this work [29, 32]. As these two aluminosilicates N-A-S-H and C-A-S-H can also coexist alongside each other, it is interesting to explore which material properties result from the various combinations of the low-calcium and high-calcium starting materials fly ash, brick flour and concrete rubble [33–35].
2. Materials and methods
2.1. Starting materials – fly ash, brick scrap and concrete rubble
The fly ash used here, “Microsit 10”, with a calcium content of around 5 % is classified as a class F fly ash. By means of laser diffraction, a particle size of x50 = 2.9 µm was determined. The unmixed brick scrap comes from a masonry brick plant and was ground in a ball mill to brick flour with a particle size of x50 = 28.90 µm. The unmixed concrete rubble comes from a Mid-Franconian construction waste landfill and was ground in a ball mill to a particle size of x50 = 90.67 µm. The secondary raw materials were characterized by means of X-ray fluorescence analysis (XRF) and X-ray diffraction (XRD). The XRF analysis shows the main components of the starting materials (» Table 1). The loss on ignition determined at 950 °C is 3.86 wt% for the fly ash, 0.92 wt% for the brick flour and 14.47 wt% for the concrete rubble.
By means of XRD analysis, the phase components of the input materials used, i.e. fly ash, brick flour and concrete rubble were analysed. It was established that the crystalline components of the fly ash consist mainly of mullite and quartz, while haematite could be detected as an auxiliary component. It should be noted that the largest mass percentage of fly ash is made up of X-ray amorphous phases. These are particularly important in respect of geopolymerization as they make a higher reaction potential available than the crystalline components.
In the XRD analysis of the brick flour, quartzite, muscovite, albite and sodium-aluminium-oxide-silicate are detected as the main crystalline phases, with haematite as an auxiliary phase. The brick flour also shows an amorphous phase coexisting with the crystalline phase.
In the analysis of the concrete rubble, quartz calcite, portlandite, dolomite and orthoclase could be detected as the main crystalline phases. The high SiO2 and low CaO content in the XRF analyses is confirmed by the high quartz content detected in the XRD analysis. This is added to the cement in the form of sand and gravel to make concrete. It can be assumed that the quartz crystals introduced constitute an inert filler as these react only in the surface region during the geopolymerization. Hydration products like, for example, ettringite, could not be detected in the XRS analyses. This can be attributed to the fact that ettringite is transformed into X-ray amorphous phases in the grinding process.
2.1.1 Geopolymer synthesis
The geopolymers were synthesized at a constant activator – solids ratio of 1:3. The exact compositions are shown in » Tabelle 2. Here the solid content of fly ash (FA) and brick flour (BS) or fly ash (FA) and concrete rubble (CR) was varied, which, with uniform activator composition, resulted in different SiO2/Al2O3 and Na/Al ratios. With increasing content of hygroscopic brick flour, the water requirement of the mix increases, which is offset with additional water, to guarantee acceptable workability. With increasing concrete rubble content, the water / solids ratio (l/s) could be kept constant.
The test formulations are homogenized in a laboratory mixer for a time of t = 15 min in each case and at a speed of u = 250 rpm-1. The homogeneous paste is filled in silicon moulds and air bubbles introduced expelled over a period of t = 5 min on a vibrating plate. For hardening, the samples are covered with PE film and hardened for a period of 48 h at 85 °C in a drying cabinet. After the hardening process, the geopolymer test specimens are demoulded and machined to suitable test specimens. Compressive stress specimens with edge lengths of 40 mm x 40 mm x 40 mm; three-point bending test specimens with edge lengths of 160 mm x 40 mm x 40 mm and specimen plates for analysis of the thermal conductivity with the dimensions 100 mm x 100 mm x 25 mm were prepared. All specimens that come into contact with water during sawing and polishing are dried to a constant weight in the drying cabinet prior to measurement of the material characteristics and then stored in the desiccator until the measurement. For the IR, XRD and SEM analyses, fragments from the compressive strength tests were used.
2.1.2 Preparation of the activator solution
The alkaline activator solution is produced from a liquid sodium silicate water glass of the type “Betol 39 T” supplied by the Woellner-Werke, Ludwigshafen and NaOH pellets. The solution is composed of 85.46 wt% liquid sodium silicate and 14.54 wt% sodium hydroxide. To ensure an always constant quality of the activator solution, each batch is homogenized for 24 hours by means of a magnetic agitator at a speed of u = 500 rpm-1 (r•min-¹)
2.2 Determination of the material characteristics
The maximum compressive stresses σd are determined in accordance with the standard DIN EN 196 Part 1. During the compressive strength testing, the pressure on the test surface is generated with a constant increase in force from 1.5 N (mm2 s)-1. The maximum compressive stress σd is calculated according to Equation (1) from the maximum applied force F on failure of the test specimen and the cross-sectional area A of the test specimen. From each batch, 15 test specimens were tested to obtain a reliable average value.
The maximum bending stresses σB of the test specimens are determined in the 3-point flexural strength test in compliance with the standard DIN EN 196 Part 1. Here, the bottom supports are arranged with a distance l = 100 mm, the upper pressure point is located in the centre. The actual test is conducted with constant force increase of 0.05 kN s-1. Detected is the maximum applied force F on failure of the test specimen. The maximum bending stresses σB in each case are calculated according to Equation (2) with consideration of the width b and the height h of the test specimen. From each batch, 20 test specimens were tested to obtained a reliable average value.
For testing of the apparent density, the dry test specimens are weighed at room temperature and the outer specimen dimensions determined. In accordance with Equation (3), with the mass m and the volume V, the apparent density ρb of the test specimens is calculated.
The thermal conductivity λ10,dr. is determined in compliance with the testing standard DIN EN ISO 8302:1991 with the so-called plate method. The specimen to be measured is fixed between a hot and a cool plate and the thermal conductivity measured at 15 °C, 20 °C and 35 °C test specimen temperatures. Here several sensors on the warm and cold sides of the specimens constantly determine the actual temperatures on the specimen surfaces. In addition, the electric power needed to maintain the temperatures is determined. The resulting thermal conductivity λ10,dr. is calculated as a function of the respective test specimen centre temperatures with the Equation (4), whereby the surface area A of the specimen and the layer thickness d are taken into consideration. Then, by means of linear regression, the thermal conductivity λ10,dr. at the specimen mean temperature of 10 °C is read off. To obtain reliable mean values, from all types of all the geopolymer, three test specimens are tested and from this the mean values for the thermal conductivity λ10,dr. calculated.
2.3.1. FT-IR Spectroscopy
To elucidate the structures of the geopolymers products, Fourier-Transformation Infrared Spectroscopy (FTIR) was performed in the ATR mode. The measurements are performed on a device from Bruker (Tensor II) with KBr beam splitter in the spectral range 5 000 – 400 cm-1. The influencing atmosphere is filtered out based on automatic background measurements.
2.3.2. X-ray diffraction analysis
The XRD analyses are conducted with an X’Pert PRO X-ray diffractometer from PANalytical, the cathode tubes of which are operated with a copper anode. The generator settings 40 mA at 45 kV are selected. Sample material ground in a mortar grinder is analysed.
2.4. Scanning electron microscopy
To obtain a deeper insight in the morphology of the geopolymers obtained, images were prepared by means of scanning electron microscopy. Here always the fracture surfaces of the geopolymers were analysed with 2 000x magnification.
3. Results and discussion
3.1. Compressive strength
First the compressive strength σd of the different geopolymer test specimens is analysed as a correlation between the mechanical and chemical stability can be assumed . » Figure 2 shows the obtained compressive strength σd of the geopolymer specimens with different composition. Here, the geopolymer batch FA_100 with a fly ash content of 100 wt% shows a compressive strength of σd = 83.2 MPa. If, starting from the geopolymer batch FA 100, the brick flour content is increased, the measured compressive strength initially increases, batch BS_1 with 33 wt% brick flour addition reaching a compressive strength of σd = 87.6 MPa. With a further increase of the brick flour addition above 33 wt%, the compressive strength falls steadily. Geopolymer batch BS_2 with a brick flour content of 50 wt% reaches a compressive strength of σd = 73.3 MPa; batch BS_3 with 66 wt% brick flour achieves compressive strength of σd = 47.7 MPa and batch BS_4 with a brick flour content of 100 wt% achieved only σd = 24.3 MPa.
If, starting from geopolymer FA_100 prepared from pure fly ash, the content of concrete rubble is increased, the compressive strength values obtained increase much more considerably in comparison with the addition of brick flour. While geopolymer FA_100 exhibits a compressive strength of σd = 83.2 MPa, the compressive strength of geopolymer CR_1 increases to σd = 102.3 MPa with the addition of 33 wt%. Geopolymer CR_2, in which fly ash and concrete rubble are present in the ratio of 1/1, reaches the highest compressive strength of this measurement series with σd = 113.2 MPa. If the addition of concrete rubble exceeds this, the compressive strength decreases slightly, so that batch CR_3 reaches a compressive strength of σd = 102.7 MPa similar to CR_1. Geopolymer CR_4, which consists of pure concrete rubble without fly ash, exhibits a similarly low compressive strength of σd = 25.65 MPa as that of batch BS_4 consisting of pure brick flour.
The results with regard to the substitution of fly ash with brick flour show that with the substitution of 33 wt%, the compressive strengths initially increase. If the brick flour content is increased beyond this, the resulting compressive strengths steadily decrease. From this, it can be derived that the increase in compressive strength obtained with a substitution of 33 wt% can probably be attributed to a particle reinforcement in the microstructure with non-reacted material, as described by Bernal  and Gharzouni . The continuous decrease in the compressive strength with further increase of the brick flour content, the obvious assumption is that the brick flour used in this research is less reactive than the fly ash used. However, the brick flour must also contain amorphous and therefore reactive components, as the specimens of batch BS_4 consisting of 100 wt% brick flour develops good strengths with the alkaline activation.
If the fly ash is substituted with ground concrete rubble instead of with brick flour, considerably different strength developments were observed. In this case, a clear maximum compressive strength is achieved with the substitution of 50 wt% fly ash with concrete rubble in specimen batch CR_2. Both with the substitution of 33 wt% and 66 wt% fly ash with concrete rubble, almost the same increases in compressive strength are obtained compared to the pure compositions of batches FA_100 and CR_4. As the fly ash is a low-calcium and the concrete rubble a high-calcium raw material, the high strength developments in this case can probably be attributed to the mix of N-A-S-H And C-A-S-H gel phases in the materials with the different compositions. It can be assumed that the increasing substitution of fly ash with ground concrete rubble leads to a decrease in the reactive aluminates necessary for the geopolymer setting reactions, while at the same time the availability of the calcium hydrates is increased. From literature data, it is known that in this way the formation of C-A-S-H gel is favoured over N-A-S-H gel formation. The two phases can, however, coexist alongside each other, as already described under 1. Introduction. The results in this work suggest that an optimum mix of N-A-S-H and C-A-S-H gel structures effects the development of maximum strength values. With the concrete rubble, at the same time, a considerable content of crystalline SiO2, which is contained in the sand and gravel fraction of the concrete, is introduced as inert filler into the geopolymers. It can be assumed that these quartz crystals also effect a particle reinforcement as described by Bernal  and Gharzouni , which in combination with the previously mentioned effect leads to much higher compressive strengths of the concrete rubble specimen compared with the brick flour batches.
A comparison of the mechanical parameters of the brick-flour-, fly-ash-, and concrete-rubble-fly ash geopolymers determined in this work with the literature data could only be performed to a limited extent because of the different chemical compositions of the input materials and products. So for geopolymers consisting of “red clay brick waste (RCBW)“ similarly high compressive strengths of σd = 22 MPa were achieved by Reig , as in the work presented here for 0geopolymer type BS_4. By optimizing the activator solution, Reig was able to increase the compressive strength up to σd = 50 MPa. In the work of Fort , too, “brick powder waste” was processed to geopolymers with activator solutions of different compositions. The compressive strengths obtained ranged between σd = 10 MPa and σd = 43 MPa. If the values for the geopolymer type consisting of pure brick flour (BS_4) with the compressive strengths from the two previously mentioned works, these are in a similarly low range with regard to the compressive strengths. The compressive strengths of all geopolymer compositions containing fly ash lie significantly above these.
The most comparable results were obtained by Zawrah . Here, geopolymers were produced from “waste of fired clay bricks” and “ground granulated blast furnace slag” and tested. For instance, for geopolymers made of pure “clay brick waste”, compressive strengths up σd = 15 MPa were obtained and in a mix with 60 wt% “ground granulated blast furnace slag” a compressive strength up to σd = 83 MPa was obtained after a hardening time of 90 days. In the work here, for the geopolymer batch BS_4 produced from pure brick flour, compressive strengths of 24 MPa were obtained and for the geopolymer charge BS_1 with 66 wt% fly ash, after a hardening time of 48 hours, a compressive strength of σd = 87 MPa was achieved.
Similar tests were also conducted for geopolymers prepared with concrete waste, according to Lampris , for instance, for geopolymers made of concrete waste with 20 wt% metakaolin, a compressive strength of σd = 33 MPa was achieved. With a mix of concrete waste and 10 wt% metakaolin, Vásquez  determined compressive strengths of 46 MPa. In the work presented here, with the geopolymer batch CR_4 prepared from pure concrete rubble, similar compressive strengths could be obtained. With geopolymers that were produced from pure concrete waste Robayo-Salazar , Zaharaki  and Komnitsas  achieved relatively low compressive strengths of σd = 7 MPa, σd = 8 MPa and σd = 13 MPa, respectively. Only with the addition of 30 wt% Portland cement did Robayo-Salazar  manage to obtain σd = 34 MPa and therefore a compressive strength comparable with that of specimen batch CR_4 here. All the geopolymers presented in this work that are composed of mixes of concrete rubble and fly ash achieve compressive strengths higher by a factor of 3 than those geopolymer mixes known from the literature.
3.2. Thermal conductivity and apparent density
» Figure 3 shows the results for the thermal conductivity λ10,dr. and » Figure 4 the those for the associated apparent density ρb in graphic form. Geopolymer batch FA_100 consisting of pure fly ash exhibits with ρb = 1.69 gcm-3 the highest apparent density, with a mean thermal conductivity of λ10,dr. = 0.354 W m-1K-1. Geopolymer batch BS_1 in which 33 wt% fly ash was substituted with brick flour exhibits with ρb = 1.66 gcm-3 a lower apparent density, demonstrates, however, with λ10,dr. = 0.409 W m-1K-1 a higher thermal conductivity than the batch FA_100 geopolymer.
For the geopolymer batches in which fly ash was further substituted with brick flour (BS_2, BS_3 and BS_4), a continuous decrease in the apparent density was established, starting from ρb = 1.59 gcm-3 for batch BS_2, through ρb = 1.55 gcm-3 for batch BS_3 to ρb = 1.49 gcm-3 for batch BS_4. Corresponding to the decreasing apparent density of the geopolymer batches with increasing substitution of the fly ash with brick flour, there is a proportional decrease in the associated thermal conductivity λ10,dr.. Here, geopolymer batch BS_2 with a thermal conductivity of λ10,dr. = 0.356 W m-1K-1 exhibits a slightly higher thermal conductivity λ10,dr. than geopolymer batch FA_100. Geopolymer batch BS_3 has a lower thermal conductivity of λ10,dr. = 0.321 W m-1K-1 and geopolymer batch BS_4, with λ10,dr. = 0.293 W m-1K-1, demonstrates the lowest thermal conductivity of the compositions tested in this research.
If fly ash content is substituted with concrete rubble instead of with brick flour, the apparent density and the thermal conductivity initially increase considerably compared with geopolymer batch FA_100. Geopolymer batch CR_1 exhibits an apparent density of ρb = 1.75 gcm-3 and thermal conductivity of λ10,dr. = 0.439 W m-1K-1. Geopolymer batch CR_2 exhibits, with ρb= 1.80 gcm-3, the highest apparent density of the concrete rubble batches with a thermal conductivity of λ10,dr. = 0.526 W m-1K-1. If the content of concrete rubble is increased further, the apparent density remains on a constant saturation level. At the same time, the thermal conductivity increased constantly up to geopolymer batch CR_3 before falling slightly for geopolymer batch CR_4 consisting of pure concrete rubble. Here, geopolymer batch CR_3 with an apparent density of ρb = 1.79 gcm-3 achieves the highest thermal conductivity of λ10,dr. = 0,608 W m-1K-1 measured in this research. For batch CR_4, a somewhat lower thermal conductivity of λ10,dr. = 0.558 W m-1K-1 is registered at the same apparent density.
From the analysis, it is concluded that the substitution of fly ash with brick flour in the geopolymer lowers the apparent density and the associated thermal conductivity while the substitution of fly ash with concrete rubble results in the opposite effect.
(A second article in ZI 02/2023 will explore the influence of brick flour or concrete rubble on the setting behaviour of fly-ash-based geopolymers.)