Characterization of Fresh and Cured Properties of Polymer Concretes Based on Two Metallurgical Wastes

Polyester polymer concretes can substitute conventional concretes based on their usually good mechanical strength, adequate physical properties, and high resistance against aggressive chemical environments. They also show a high potential for using recycled targets in their manufacturing. This paper analyzes the fresh and cured properties of polyester polymer concretes containing two metallurgical wastes, namely: ladle slag and alumina filler. Both targets require a higher resin dosage than sand. The standard consistency test showed a low representativeness of the recycled fresh mixes’ workability. The ladle slag and alumina filler samples showed a higher length plastic shrinkage than those containing sand. All of the targets obtained cured density values in the range of 1.589–1.912 g/cm3. From a mechanical point of view, the sand and alumina filler containing polyester polymer concretes reached 11.02 and 10.93 kN, respectively, of flexural strength, while the ladle slag samples showed the best result with 19.31 kN. In the compressive strength test, the sand and alumina filler combinations reached 106.16 and 104.21 MPa, respectively, while the ladle slag achieved 160.48 MPa. The flexural and compressive elasticity modulus showed similar trends related to the resin content.


Introduction
Polyester polymer concretes (PPCs) are composite materials made up of organic resins as a binder, and aggregates or fillers as the target materials. These composite materials develop, by means of exothermic chemical reactions with hardeners and accelerators, a continuous resistant polymer matrix around the target particles. PPCs have been receiving increasing interest in recent years because of their good characteristics compared with traditional concrete, namely: improved mechanical strength, superior durability, good resistance to water, lower permeability, high chemical and corrosion resistances, excellent adhesion and a fast curing process, among others [1,2]. However, PPCs also show some disadvantages, like a lower workability, fresh resin odor, lower resistance to high temperatures, and higher cost compared with conventional concrete [3]. An effective way to decrease these shortcomings, in agreement with the actual environmentally friendly thinking trends, consists of using recycled targets instead of raw materials for PPC manufacturing. PPCs show a high potential for the incorporation of recycled materials because of the nature of their polymer matrix and their lack of bonds with the target particles [4]. In addition, the use of recycled targets for PPC manufacturing could suppose an effective method of valorization for large quantities of different wastes. This could not only improve the sustainability of the resulting PPC products, but also the natural resources preservation, in accordance with the circular economy principles [5]. Different applications could benefit from the use of PPC-containing recycled target materials, mainly those that are not subject to strict structural concrete regulations. Nevertheless, not all wastes have the same potential for PPC manufacturing, as their own characteristics can modify the fresh or final product properties. Extensive research works have been conducted for determining the properties of PPCs made up of different recycled targets, like aggregates, sand, clay, fillers, fly ash, fuel ashes, plastics, or recycled fiber reinforced polymers, among others [1,3,4,[6][7][8][9][10][11][12][13]. However, nowadays, huge amounts of different wastes, with potential as PPC targets, continue to be disposed in landfills around the world and pollute the environment. Based on their good properties as aggregates for conventional concrete manufacturing, one of the most promising is metallurgical wastes [14][15][16][17]. Nonetheless, there is a lack of knowledge about the ability of these materials as PPC components. Thus, this study analyzes the aptitude of two metallurgical wastes as target materials for PPC manufacturing-alumina filler (AF) and ladle slags (LS). Large quantities of both wastes are continuously generated, and because of the lack of effective valorization methods, high volumes are available around the world. Their use for PPC manufacturing would have important environmental and economic advantages for waste producers and PPC manufacturing companies. Under these environmental and economic considerations, an experimental laboratory investigation was carried out. Its aim was to develop the knowledge-which is currently not available-between these non-conventional target materials and the fresh and mechanical properties of polyester polymer concrete.

Materials
The resin used in this work was an unsaturated ortoftaic tixotropic non-accelerated polyester mixture known as CRYSTIC 406 NT. This product was specifically developed for PPC manufacturing based on its chemical and mechanical features, as well as its lower cost compared with other resins. This resin contains ultra-violet absorbents that provide good resistance against solar radiation. Its low viscosity and quick setting time allow for a high production efficiency. The polyester mixture comprises two parts that are supplied separately, namely: the resin itself and an accelerator based on cobalt salts and metiletilketone peroxide as the catalyzer. Table 1 shows the main liquid and cured characteristics of this resin. AF is generated as a secondary waste during the valorization of alumina salt slags that are produced during the alumina recycling process. Around 110,000 tonnes of AF waste is generated per year in Spain. LSs are also called secondary refining or white slags. They are produced when the steel is desulfurized in the transport ladle during iron or steel making from scrap-iron in electric arc furnaces. The yearly Spanish production of LS exceeds 340,000 tonnes. A siliceous sand from siliceous rock crushing was used as a reference target for the sample manufacturing. Figure 1 shows the granulometric curves of the different target materials used in this investigation. Table 2 shows their chemical composition, expressed as the main oxides, obtained by X-ray fluorescence (XRF), and the mineralogical composition, which is estimated by X-ray diffraction (XRD) analysis.

Methods
For the sample manufacturing, the resin and target were carefully mixed using an electrical mixing drill for five minutes. The catalyzer was then added, and the mix was mixed again for an additional five minutes in order to guarantee complete homogenization. After that, the mixture was filled in prismatic 40 × 40 × 160 mm molds coated with a mold-release agent. The fresh samples were vibrated to eliminate air, as well as for the correct filling of the molds. The samples were prepared by changing the resin content by 5% in the mass of the total sample, to state the optimum resin dosage range for each target. After preparation, the samples were maintained at room conditions for 1 h before demolding. A 24 h cure in an oven at 40 ºC was carried out to guarantee the complete polymerization of the resin and the curing homogeneity of all of the samples. Finally, the samples were maintained at room temperature for seven days before testing. Each combination was encoded by letters that identified the aggregate type (S for sand, LS for the ladle slag, and AF for alumina filler), followed by the resin content, expressed as a mass percentage of the total mix. Samples with no target materials were manufactured as a reference and were identified as resin.
Because of the lack of specific standardized tests for PPC characterization, and considering the target material particles' size, conventional mortars tests, in accordance with European standards, were considered for the PPC characterization. Thus, the following properties were considered: (i) fresh characteristics, (ii) cured physical properties, and (iii) mechanical strength. Table 3 shows the tests carried out and the reference standards. The highest resin content limit was defined based on the observed excess of resin on the upper surface of the samples after vibrating. On the other hand, the workability needs for the fresh samples' mixing, molding, and vibration were considered to show the lowest limit of the resin optimum content. A fresh consistency test was carried out in a flow table in accordance with the EN 1015-3 [19] standard. The plastic shrinkage and density were stated in accordance with the EN 12390-1 [20] and EN 12390-7 [21], respectively. The flexural strength and flexural modulus of the elasticity were obtained in accordance with EN 12390-5 [22]. Finally, the compressive strength and compressive modulus of the elasticity tests were carried out, based on the procedure of standards EN 12390-13 [23] and EN 12390-3 [24], respectively. Figure 2 shows the fresh consistency variations of the fresh mixes for each target material in its optimum resin dosage range. Considering the workability criterion, the sand samples required a dosage above 20% resin, while the LS and the AF samples' workability started with a minimum resin content of 35%. The excess of resin was observed for dosages above 30% in the case of sand, and 45% for the LS and AF combinations. The higher resin content required for the LS and AF samples is probably related to the high content of these materials in the finest particles, which are 66.23% and 31.40% under 80 µ, respectively. The fresh consistency test expresses the combinations consistency as the "cake diameter" of the fresh PPC sample after a shaking process, defined in the European Standard EN 1015-3 [19]. For the three target materials, the highest consistencies, corresponding to the lowest cake diameters, were observed at the lowest resin dosages. Thus, the S-20 combination reached a consistency of 8.5 cm, LS-35 reached 17.6 cm, and AF-35 reached 17.7 cm. In all of the cases, the mixes' workability improved as the resin content increased. Thus, the S-25 combination diameter increased until 12.4 cm, LS-40 until 23.3 cm, and AF-40 until 22.2 cm. As expected, the lowest consistencies were obtained for the highest resin dosages, namely: 21.1 cm for the S-30 combination, 26.6 cm for the LS-40, and 27.0 cm for the AF-40. It is noticeable that although sand requires a lower resin content, it showed a lower workability than LS and AF, which obtained very close consistencies. These results demonstrated that, as well as the resin dosage range, the fresh PPC consistency depends on the existence of a minimum content of finest particles, which strongly control the fresh mix rheology.  The pure resin results for all of the following tests are depicted as a continuous line as a reference. In this case, the resin plastic shrinkage was 2.263%. The sample plastic shrinkages were lower than the control resin value, and showed a direct relationship with the resin content. The lowest length plastic shrinkage values were obtained in the sand combinations within the 0.067%-0.604% range, probably due to the lower resin dosages needed for this target. Sand also showed the lowest result variability, with standard deviations (σ) between 0.026 and 0.044. The highest length plastic shrinkages and variability values were obtained by the LS combinations with ranges of 0.917%-1.235% and 0.118-0.191, respectively. Finally, the AF samples reached plastic shrinkages between 0.640% and 0.898% with σ values between 0.031 and 0.088. This test shows the importance of the use of target materials for the PPC length plastic shrinkage decrease, because of the reduction of resin content in the PPC. The length plastic shrinkage differences and the results' variability highlights the influence of the target granulometry in this PPC characteristic.

Cured Physical Properties
The cured PPC densities are shown in Figure 4. An increase in the sample densities was observed from 1.220 g/cm 3 , corresponding to the resin, until 1.870-1.912 g/cm 3 for the sand, until 1.749-1.902 g/cm 3 for the LS, and until 1.589-1.749 g/cm 3 in the case of the AF combinations. The sand samples showed very low-density differences for the different resin dosages, with a maximum observed density of 1.912 g/cm 3 for the 25% resin dosage. On the other hand, AF and LS showed an indirect relationship between the samples' cured densities and the resin content. Thus, the LS densities varied between 1.749 g/cm 3 for the LS-45 combination and 1.902 g/cm 3 for LS-35, and the AF samples varied between 1.589 g/cm 3 for AF-45 and 1.749 g/cm 3 for AF-35. The sample densities showed, in general, low standard deviations (σ) in the range of 0.007-0.016 for the sand, 0.004-0.013 for the LS, and 0.008-0.030 for the AF. Different density relationship trends were observed between the sand, as well as between the LS and AF resin content combinations. This suggests that the resin does not only fill the interparticle pores, but also produces a partial substitution of target particles, thus decreasing the sample density. This highlights the complexity of the relationship between the bulk target material densities, their particle sizes, the optimum mixes resin dosage ranges, and the PPC densities.