Process Route for Electric Arc Furnace Dust (EAFD) Rinse Wastewater Desalination

Abstract

This study introduces a two-step treatment method for synthetic and real electric arc furnace dust (EAFD) wastewater, integrating sorption with Mg–Al layered double hydroxides (LDHs) and electrodialysis (ED). The hydrotalcite (LDH), mainly Mg₆Al₂(CO₃)OH₁₆·4H₂O (hydrotalcite-2H), was characterized by XRD, FTIR, SEM, and EDX, confirming its layered structure and ion-exchange capacity. Calcination at 550 °C was identified as optimal, enhancing sorption efficiency while retaining rehydration potential. Sorption tests demonstrated high effectiveness in removing multivalent ions, achieving over 99% elimination of Ca²⁺, SO₄²⁻, and Pb²⁺ ions and Cr from both synthetic and real wastewater. In contrast, monovalent ions such as Na⁺ and K⁺ were not effectively removed, except for partial removal of Cl⁻. To overcome this limitation, electrodialysis was applied in the second step, successfully targeting the remaining monovalent ions and achieving more than 95% conductivity reduction. A key challenge of ED, salt precipitation caused by calcium and sulphate in the concentrate, was effectively mitigated by the prior LDH treatment. The combined process minimized scaling risks, improved overall ion removal (above 97% for Na⁺ and K⁺), and produced low-salinity effluents (0.84 mS cm⁻¹), suitable for reuse in hydrometallurgical operations. These findings demonstrate that coupling LDH sorption with electrodialysis provides a sustainable and efficient strategy for treating high-salinity industrial wastewaters, particularly those originating from EAFD processes.

1. Introduction

Crude steel production generates multiple waste streams, including dust, sludge, slag, wastewater, and exhaust gases, which require treatment to prevent the release of hazardous substances and to minimize material losses. Among these, electric arc furnace dust (EAFD) is particularly significant, accounting for approximately 1–2% of the furnace charge and containing up to 40 wt.% Zn and 50 wt.% Fe [1,2,3]. At the same time, EAFD contains toxic metals such as Pb, Cr, and Cd, and is therefore frequently classified as hazardous waste. Hydrometallurgy has emerged as a promising approach for EAFD valorization, targeting the recovery of zinc and iron. However, the presence of mineral impurities such as KCl, NaCl, SiO₂, and CaCO₃ reduces zinc extraction efficiency and should be removed prior to further processing. Water leaching as a pre-treatment step has been shown to effectively remove water-soluble halides [1,4]. Nevertheless, this process generates large volumes of rinse wastewater with high alkalinity, elevated concentrations of chlorides, calcium, sodium, potassium, and sulphates, and potentially trace amounts of toxic metals such as chromium and lead.

Several technologies are available for the treatment of such saline wastewater, including membrane-based processes (reverse osmosis (RO), electrodialysis (ED)), electrochemical methods (capacitive deionization), sorption using ion exchangers or layered double hydroxides (LDHs), and precipitation techniques [5,6,7,8,9,10,11,12,13,14,15,16]. LDHs, featuring [M_1−x²⁺M_x³⁺(OH)₂](Aⁿ⁻)x/n·yH₂O (M²⁺ contains Mg²⁺, Ca²⁺, Zn²⁺, etc.; M³⁺ contains Al³⁺, Fe³⁺, Cr³⁺ etc.; Aⁿ⁻ represents interlayer anions such as CO₃²⁻, NO₃⁻, PO₄³⁻, etc.; x typically ranges from 0.20 to 0.33), have shown high potential for the removal of both anions (Cl⁻, SO₄²⁻, CO₃²⁻, PO₄³⁻, CrO₄²⁻) and cations (Ca²⁺, Pb²⁺, etc.) [17]. These lamellar inorganic solids feature brucite-like structures and generate positively charged hydroxide layers balanced by interlayer anions and water molecules. Their highly specific surface area and tunable interlayer composition provide abundant active sites for selective ion exchange, including chloride and sulphate removal. Interlayer anions are readily exchangeable, with affinity generally ranked as CO₃²⁻ > SO₄²⁻ > OH⁻ > F⁻ > Cl⁻ > Br⁻ > NO₃⁻ [18,19]. Their interactions with coexisting ions are selective: nitrate minimally affects Cl⁻ uptake, whereas sulphate can hinder adsorption of chlorides by stabilizing monosulphate phases. Cations such as Ca²⁺ and Mg²⁺ further influence hydration and anion incorporation, modulating intermediate hydroxide formation. LDHs also interact with polyvalent oxyanions through combined anion and ligand exchange. These properties, alongside recyclability and structural flexibility, render hydrotalcites as highly effective sorbents for halogen anions, oxyanions, and anionic dyes, highlighting their broad potential in water treatment and pollutant removal. LDHs demonstrate reversible anion exchange, high interlayer adsorption capacity, and selective interactions with coexisting ions, making them versatile sorbents for environmental remediation [18]. ED is a membrane-based separation process that utilizes an applied electric field to selectively transport ions through anion and cation exchange membranes, thereby reducing water salinity. Unlike pressure-driven methods such as RO, ED offers distinct advantages in terms of energy efficiency for low to moderate salinities, selective ion removal, and operational flexibility. Recent advancements—particularly in the development of monovalent- or divalent-selective membranes and hybrid system configurations—have significantly enhanced ED’s applicability in brackish water desalination and resource recovery. Recent studies have highlighted the following key developments: Comparative assessments show that ED is economically favorable for feed salinities up to 3 g L⁻¹, but at higher salinities, rising membrane and electricity costs increase the levelized cost of water (LCOW), giving reverse osmosis (RO) a clear advantage. Strategies such as reducing ion exchange membrane costs and enhancing membrane properties could help close this economic gap, though at salinities ≥ 5 g L⁻¹, cost reductions alone are insufficient to make ED competitive with RO [20]. Innovative two-stage hydraulic ED systems with optimized stack architectures—featuring resin-filled electrode cells, asymmetric cell pair designs, and counterflow arrangements—demonstrated energy savings of up to ~40% compared to conventional ED systems during brackish water desalination to potable standards [20]. Integrating ED with RO (ED–RO hybrid) in a single operational unit has shown mutual performance enhancement: increased RO permeate flux and improved ion rejection, along with 9–19% energy consumption reduction for RO and 2–3% for ED at feed concentrations of 5–10 g L⁻¹ [17]. Commercial ion exchange membranes (IEMs) were modified via interfacial polymerization (IP) to form dense polyamide layers, aiming to enhance monovalent ion selectivity for electrodialysis applications. These IP-modified membranes were characterized and evaluated for their ability to concentrate sodium chloride (NaCl) from seawater reverse osmosis (SWRO) brine, demonstrating improved divalent ion rejection and potential for direct use in the chlor-alkali industry [20]. These advancements collectively underscore ED’s potential as a sustainable and versatile desalination technology. Ongoing research is focused on optimizing membrane performance, stack design, energy integration (e.g., coupling with renewable energy sources), and hybrid configurations to overcome remaining limitations and bring ED closer to widespread practical deployment [17,21,22]. Despite these advances, comprehensive studies on integrated treatment of complex industrial wastewater combining sorption and ED for simultaneous removal of chlorides, sulphates, calcium, sodium, potassium, and toxic metals remain limited. Most prior research has focused on individual contaminants. This study addresses this gap by investigating ion removal from synthetic and real EAFD wastewater (RW) via sorption and electrodialysis, both independently and sequentially. The objective is to optimize conditions for reducing salinity and eliminating primary impurities (Cl⁻, SO₄²⁻, Ca²⁺, Pb²⁺, Cr, Zn²⁺, Na⁺, K⁺), contributing to the development of an effective and integrated wastewater purification system suitable for hydrometallurgical EAFD recycling processes.

2. Materials and Methods

2.1. Wastewater for Experiments

Samples of synthetic and real wastewater were used for sorption and electrodialysis studies. The synthetic solutions, which mimic real EAFD wastewater, were prepared by dissolving precise amounts of the following chemicals in deionized water: PbSO₄, ZnSO₄, FeSO₄·7H₂O, Na₂SO₄, K₂SO₄, Na₂CO₃, CrCl₃·6H₂O, CaCl₂, NaOH, and KOH. All chemicals used were of analytical grade and were purchased from Merck Life Science spol.s r.o., Bratislava, Slovakia. The real wastewater was obtained by leaching a sample of EAFD in deionized water with a liquid-to-solid ratio (L:S) of 10. After 60 min of stirring, the EAFD sample was filtered, and the resulting solution was directly used for experiments: (i) sorption using laboratory-prepared commercial Mg-Al LDH sorbent for both RW and synthetic solution (S1), prepared to match RW solution composition; (ii) ED for both RW and S1; and (iii) sorption followed by ED using the synthetic solution (S2) that imitated the composition of RW after sorption. A fresh solution was prepared for each experiment to prevent calcium carbonate precipitation. The composition and characteristics of the RW, S1, and S2 are summarized in

Table 1. Composition and characteristics of the input solutions synthetic (S1), EAFD (RW), and synthetic simulation of RW after sorption (S2) used in the experiments.

Parameter	Unit	S1	RW	S2
pH	-	11.75 ± 0.44	12.15 ± 0.48	12.10 ± 0.37
Electrical Conductivity (k)	mS cm−1	8.54 ± 0.05	11.88 ± 0.11	8.95 ± 0.16
Calcium	mg L−1	684 ± 0.09	664 ± 1.17	*
Chloride	mg L−1	1800 ± 1.91	1276 ± 1.61	*
Sulphate	mg L−1	1400 ± 1.98	800 ± 2.22	1250 ± 0.97
Sodium	mg L−1	825 ± 5.29	739 ± 5.77	730 ± 6.1
Potassium	mg L−1	2951 ± 6.03	1379 ± 5.58	1320 ± 6.44
Lead	mg L−1	8.95 ± 0.21	22.3 ± 0.22	2.52 ± 0.17
Chromium	mg L−1	0.17 ± 0.23	0.76 ± 0.34	*
Zinc	mg L−1	-	3.48 ± 0.12	0.145 ± 0.14
Carbonates	mg L−1	-	240 ± 0.14	360 ± 0.7

RW = real EAFD wastewater; S1 = synthetic solution imitating the composition of RW; S2 = synthetic solution imitating the composition of RW after sorption, *—not relevant.

. The values for potassium and sulphates fluctuated in the EAFD dust samples and, consequently, in the real wastewater obtained from the treatment of these samples. This explains why the values for K and SO₄²⁻ in RW differ from those in S1.

The concentrations of Ca²⁺, K⁺, Na⁺, Cr, and Pb ions in samples were measured using Atomic Absorption Spectrometry (AAS) (SpectrAA 20 PLUS, Varian, Prague, Czech Republic). Chloride and sulphate ions were quantified through the Mohr titration method and with a multiparameter photometer (Hanna Instrument HI 83 200, Woonsocket, RI, USA). Solution pH and electrical conductivity were monitored using a multimeter (WTW inoLab Multi 9420 IDS, Fisher Scientific, Leicestershire, UK). Chemical analyses of the solutions were conducted in triplicate to ensure accuracy and reproducibility.

2.2. Sorption Experiments

The sorbent used for sorption experiments was commercial LDH obtained from Sigma-Aldrich (Merck KGaA, Darmstadt, Germany). To identify its chemical and phase composition, atomic absorption spectroscopy using AAS (SpectrAA 20 PLUS, Varian, Prague, Czech Republic) and X-ray diffraction using the PANalytical X’Pert PRO with Co-Kα radiation (XRD, Malvern Panalytical Ltd., Almelo, The Netherlands) were employed.

The microstructure, morphology, and grain size of the LDH precursor sample were observed using the Scanning Electron Microscope (SEM), MIRA FE-SEM, TESCAN (Brno, Czech Republic). Finally, the LDH sample was analyzed using Fourier Transform Infrared Spectroscopy with Attenuated Total Reflectance (FTIR-ATR), Nicolet iS50 FTIR Spectrometer, Thermo Scientific (Waltham, MA, USA), to examine functional groups.

Uncalcined LDH was modified in an electric oven at different temperatures (350 °C, 500 °C, 550 °C, and 600 °C) to obtain its calcined products, which could rehydrate and reconstitute during sorption. To determine the optimal temperature for sorbent calcination, where LDH achieves the highest sorption efficiency and reconstruction ability, calcined products were analyzed by XRD. Additionally, the LDH sample calcined at 550 °C was analyzed by electron microscopy combined with energy-dispersive X-ray spectroscopy (SEM-EDX) (MIRA FE-SEM, TESCAN (Brno, Czech Republic)) to obtain information about the material’s chemical composition, grain size, and morphology for use in further experiments.

To determine the rate of sorption and select the appropriate dosage of material, experiments were carried out with S1. Samples of LDH calcined at the selected temperature of 550 °C for 24 h, with different dosages (50, 75, 100, and 150 g L⁻¹), were added to beakers containing 50 mL of S1 solution. After adding the precise amount of LDH to the solution, the beakers were stirred on a magnetic stirrer at 500 rpm. Once the necessary sorption time was achieved, the sample was drawn off and filtered using a vacuum filter. The collected samples, along with the primary S1 solution, were examined for Cl⁻, SO₄²⁻, Ca, K, Na, Cr, Pb, pH value, and conductivity.

Sorption experiments with RW were conducted using 100 and 150 g L⁻¹ of LDH calcined at 550 °C for 24 h. The remaining sorption steps were analogous to those for the S1 solution.

The rehydrated and reconstructed sorption material obtained from the experiments with both the S1 solution and RW, using HT dosage of 100 g L⁻¹, was dried at room temperature and analyzed by XRD and SEM-EDX. The content of Ca, K, Na, Cr, and Pb in the liquid/solid samples was determined by Atomic Absorption Spectrometry (AAS, Varian 240). Chlorides and sulphates were determined by the Mohr method and a multiparameter photometer, Hanna Instrument HI 83 200. The pH value and conductivity in the solutions were measured using a multimeter (WTW inoLab Multi 9420 IDS). All sorption experiments and chemical analyses of solutions after sorption were repeated three times. The efficiency of sorption was calculated using the following formula:

$$\%\mathrm{r}\mathrm{e}\mathrm{m}\mathrm{o}\mathrm{v}\mathrm{a}\mathrm{l}=\left[{\displaystyle \frac{100\times \left({\mathrm{C}}_0-{\mathrm{C}}_{\mathrm{e}}\right)}{{\mathrm{C}}_0}}\right]$$

(1)

where C₀ and C_e are the initial and equilibrium concentration of metals ions (mg L⁻¹), respectively.

2.3. Electrodialysis Experiments

A laboratory-scale electrodialysis cell with five compartments was used in the experiments. The compartments were separated by two pairs of heterogeneous cation exchange membranes (CEMs) and anion exchange membranes (AEMs) with 16 cm² of effective area, respectively, named as HDX100 (also known as IONSEP-HC-C) and HDX200 (also labelled as IONSEP-HC-A), both provided by Hidrodex^® (Bassersdorf, Switzerland). Commercial Ti/70TiO₂30RuO₂ electrodes were employed as the anode and cathode. The experimental ED setup is shown in Figure 1. More details about the ED setup can be found elsewhere [23]. The concentrated and diluted reservoirs of the cell were filled with 0.5 L of the working solution (S1, S2 or RW), whereas the electrodes’ reservoir was filled with 0.5 L of a 12 g L⁻¹ Na_sSO₄ solution, in order to maintain the electrical conductivity of the ED setup.

The ED experiments were carried out in triplicate, at room temperature and in galvanostatic mode. The applied current density corresponded to 80% of the value of limiting current density (i_lim), determined by current–voltage curves (CVCs) [24,25]. These experiments were conducted until the time necessary to achieve a conductivity value lower than 0.2 μS cm⁻¹ (usually reported by the local water supply) or until reaching the maximum potential value of the power supply (64 V).

The pH and electrical conductivity of the concentrated and diluted solution were reported hourly. Additionally, aliquots at pre-established times were collected and the concentrations of calcium, chloride, sodium, sulphate, potassium, and calcium ions were obtained by Ion Chromatography (Dionex^®, ICS 3000, Waltham, MA, USA), zinc and lead ions by simultaneous ICP-OES (Perkin Elmer^®, Optima 8300 DV, Waltham, MA, USA), and carbonates ions by IC methodology in a TOC analyzer (Shimadzu^®, TOC-LCPH, Kyoto, Japan)—all equipment was available at the Federal University of Rio Grande do Sul, Brazil. For the RW, prior to these analyses, the samples were subjected to acidification with 2 M H₂SO₄ (to avoid compounds precipitation) and filtration with an analytical 45 µm filter. The SO₄²⁻ amount added by H₂SO₄ acidification was deducted from the concentration value reported by the analytical method.

The energy consumptions, EC (kWh m⁻³), for the ED experiments of individual solutions (S1, RW, and S2) were determined using the following Equation (2), adapted from [26]. In this study, the applied current was constant throughout the experiment and, therefore, removed from the integral variation. Moreover, since a multicomponent solution was evaluated, it is not technically feasible to determine the EC for each ionic species; thus, a global EC to treat 1 m³ of solution was calculated.

$$\mathrm{EC} (\mathrm{kWh} {\mathrm{m}}^{-3})=\left({\displaystyle \frac{\mathrm{I} {\int }_{\mathrm{t}=0}^{\mathrm{t}}\mathrm{U}\left(\mathrm{t}\right)\mathrm{dt}}{\mathrm{V}}}\right)$$

(2)

where:

I = current applied during the experiments.

U(t) = electrodialysis cell voltage (V).

t = time (h).

V = volume of treated solution (m³), diluted compartment.

3. Results and Discussion

3.1. Characterization of Uncalcined LDH

AAS analysis has shown that uncalcined LDH contains 22.94% of Mg and 11.8% of Al. By XRD (Figure 2), dominant phase Mg₆Al₂(CO₃)OH₁₆·4H₂O (hydrotalcite-2H, JCPDS: 00-054-1030) was confirmed. Moreover, different polytype variations, which are the result of the metal oxide’s gradual transformation at a relatively low temperature (30 °C) [27], were detected, and were mainly ((Mg₄Al₂)(OH)₁₂CO₃(H₂O)₃)_0.5 (JCPDS: 01-070-2151), meixnerite (JCPDS 00-038-0478, 00-050-1684), quintinite-2H Mg₄Al₂(OH)₁₂(CO₃)(H₂O)₃ (JCPDS 01-087-1138), and Mg₂Al(OH)₇ (JCPDS 35-1275). They are a kind of hydrotalcite-like layered Mg/Al double hydroxide compound with a hexagonal or rhombohedral crystal structure derived from brucite Mg(OH)₂, where Al³⁺ cations replace some of the Mg²⁺ cations. In addition, the sample of LDH contained metals like Cu, Ca, and S as a by-product of synthesis-produced or thermal oxidation processes. A small portion of the thermal decomposition products of Cu from synthesis were detected in phases such as Cu₂SO₄ (JCPDS: 00-028-0401), CuO (JCPDS: 01-089-2530), and Cu_1.8S (JCPDS: 00-026-0476). The calcium in the LDH sample was in the form of calcite (JCPDS: 01-081-2027). In the structure of LDH were observed boehmite crystals (JCPDS 1-774) with octahedral plates. γ-AlO(OH) as boehmite was formed from gibbsite at approximately 250–300 °C. In the structure was detected γ-Al₂O₃ (JCPDS 10-425) formed by thermal dehydroxylation during the hydrothermal synthesis process (180–250 °C) of LDH production from boehmite. In addition to this, ɛ-Al₂O₃ (JCPDS 21-10) was detected in the structure and its existence is assumed to relate to the transformation of pseudoboehmite (gelatinous boehmite) to γ-Al₂O₃ and further to ɛ-Al₂O₃. XRD patterns also indicate the possible presence of metastable χ-Al₂O₃ (JCPDS 4-880). According to [28], gibbsite is the only aluminum hydroxide that produces chi-alumina crystals by thermal dehydroxylation.

FTIR analysis of uncalcined LDH (Figure 3) provided information about the main functional groups present in the structure of the sample. Region 3600–3200 cm⁻¹ presents a characteristic broad and robust peak centered at 3408 cm⁻¹, with an O-H stretching vibration at the surface and water molecules in the interlayer space [29]. This vibration peak could be attributed to either boehmite or Mg₂Al(OH)₇ [30]. Several signals were detected between 1800 and 1200 cm⁻¹, such as the bending vibration of water around 1600–1700 cm⁻¹, an infrared (IR) vibration associated with the Mg-Al band at 1571 cm⁻¹, and an intense absorption ν3 vibration of interlayer carbonate at 1362 cm⁻¹—with the lattice vibration of the main layers from 500 to 1000 cm⁻¹ [31]. The IR bands at 768 cm⁻¹ and 560.8 cm⁻¹ are assigned to quintinite-2H [32]. The absorption peak at 1560 cm⁻¹ is ascribed to κ-(Al₂O₃) [33]. The IR absorption bands from 1000 to 500 cm⁻¹ cover several overlapping bands, which may not be fully interpreted considering that the exact material procedure preparation is unknown. However, absorption peaks at 934 cm⁻¹ and 768 cm⁻¹ could be assigned to the Al-OH bending and stretching vibration, respectively [34].

SEM images of the uncalcined LDH sample (Figure 4) showed the crystals of Mg-Al hydroxide with regular mature shapes with an average grain size of 200 to 800 nm.

3.2. Characterization of Calcined Products of LDH

The uncalcined LDH sample was subjected to calcination at temperatures of 350 °C, 500 °C, 550 °C, and 600 °C to analyze the changes in phase composition during thermal treatment. This analysis is crucial for identifying the temperature at which the LDH structure irreversibly transforms into a stable spinel-type phase. At this point, it loses its ability to rehydrate and restore its original layered structure—an essential property for ion exchange applications.

The calcined products were characterized using XRD for qualitative phase analysis, and the results were compared to those of the uncalcined LDH and a reference MgO sample (Figure 5). The XRD patterns of the samples calcined at increasing temperatures reveal a progressive transformation of the original LDH structure into a mixed oxide phase with the composition Mg_0.8Al_0.2O_1.1. This phase exhibits diffraction patterns similar to periclase (MgO), indicating the formation of a defective cubic oxide structure characteristic of thermally decomposed LDH materials containing both Mg and Al.

From Figure 5, a gradual decomposition and transformation of the well-crystallized LDH phases in the original uncalcined sample is observed as the temperature increases from 350 °C to 600 °C. This transformation leads to the formation of more amorphous phases with smaller grain sizes and crystallites. Upon heating to 350 °C, the basal reflection (003) of the hydrotalcite 2H phase diminishes at the peak position 13.49° (2θ) [28]. The phases from the original LDH gradually fade, including hydrotalcite-2H, meixnerite, and magnesium aluminum carbonate hydroxide. A new phase, quintinite-2H, with the chemical formula Mg₄Al₂(OH)₁₂(CO₃)·3H₂O (JCPDS: 01-087-1138), appears instead.

At the same time, the compound Mg₂Al(OH)₇ begins to decompose, leading to the formation of MgO (JCPDS: 01-075-0447), which shows a distinct peak at 50.16° (2θ) in Figure 5. The phase qualitative analysis also reveals minor phases such as nesquehonite (MgCO₃·3H₂O, JCPDS: 01-070-1433) and its dehydration product, MgCO₃ (JCPDS: 01-086-2348). As reported by [35], nesquehonite loses its water below 300–350 °C, while CO₂ is released above 350 °C, typically between 500 and 550 °C, though the exact thermal decomposition mechanism is influenced by experimental conditions.

Additionally, the Cu_1.8S phase decreases due to its thermal decomposition into CuO and Cu₂SO₄ [36]. The compound γ-Al₂O₃ (JCPDS: 10-425) is also identified as an amorphous phase in the sample.

Analyzing the structure of the LDH sample calcined at 500 °C, it can be observed that the original diffraction peaks from hydrotalcite-2H, quintinite, and nesquehonite are no longer present. By-products that decomposed below 450 °C led to the formation of α-Al₂O₃ and MgO, while those that decomposed above 500 °C formed MgAl₂O₄, accompanied by a sharp increase in MgO dispersity. Between 350 °C and 500 °C, crystallization of MgO (periclase, JCPDS 96-900-6817) was observed. This behavior is likely attributed to the dehydroxylation of brucite-like layers and the removal of interlayer anions (CO₃²⁻). The abundant hydrotalcite phases were transformed into crystalline, non-stoichiometric MgO (JCPDS: 96-900-6817). The compound Mg₂Al(OH)₇ underwent dehydroxylation to form smaller portions of spinels, such as Mg_0.99Al_2.01O₄ (JCPDS: 96-900-1373) and MgAl₂O₄. The presence of CuO and CaCO₃ is still evident, while new phases of Cu₂O (JCPDS: 01-077-0199) and Cu₂SO₄ (JCPDS: 28-0401) form as a result of CuS decomposition. Amorphous χ-Al₂O₃ and CaCO₃ also remain present in the structure [37].

At 550 °C, the calcined LDH sample contains dominant unstable MgO, along with small proportions of stoichiometric and non-stoichiometric spinels. Other identified phases include α-Al₂O₃ (JCPDS 96-210-1053), χ-Al₂O₃, CuO, Cu₂O, and CaCO₃.

By 600 °C, dehydroxylation and decarbonization of the LDH are essentially complete. The majority of phases in the LDH sample calcined at 600 °C are periclase MgO (JCPDS: 96-900-6804) and spinels. Minor phases such as α-Al₂O₃ and χ-Al₂O₃, along with impurities like Cu₂O and CaCO₃, remain. Based on these observations, it can be concluded that temperatures above 550 °C are not recommended for LDH calcination, as at 600 °C the LDH structure transforms into stable spinel phases, thus losing its ability to rehydrate and reconstruct its structure when in contact with water.

The sample of LDH calcinated at 550 °C was subjected to EDX analysis to obtain its chemical composition, Figure 6.

EDX analysis revealed the of presence of the main elements (marked with white colour in the Figure 6) Mg (26.6%) and Al (14%) in the LDH sample. Additionally, the sample contains minor or trace amounts (marked with red colour in the Figure 6) of other elements such as N (2.2%), Cu (0.4%), Ca (0.3%), and S (0.1%). The presence of these elements is likely due to unwashed impurities or ions incorporated into the structure during the coprecipitation and synthesis of the LDH. As mentioned previously, the exact procedure used to prepare the LDH is unknown. Element C was excluded from the spectrum map in Figure 6 because carbon was added to improve the sample’s electrical conductivity. Despite the presence of nitrogen in the EDX analysis, XRD did not detect any inorganic phase containing nitrogen. Therefore, it is highly probable that N originated from organic substances added during the coprecipitation process.

After calcination at 550 °C, the sample experienced a weight loss of ~39%, which is consistent with the decomposition and removal of water and other volatile components.

The grain size in the LDH sample after calcination at 550 °C (Figure 7) decreases to a range of 100–400 nm, compared to the original uncalcined LDH sample (200–800 nm, Figure 4). This reduction in grain size is attributed to the processes of dehydration, dehydroxylation, and decarbonization that occur during calcination. Despite this size reduction, the plate-like morphology of the LDH sorbent is largely preserved, with the plates maintaining an almost regular shape even after calcination.

3.3. Sorption Tests

Sorption tests were conducted using the S1 solution, a synthetic solution that mimics EAFD wastewater. The primary goal was to optimize the dosage of sorbents and the duration of the sorption process to maximize ion removal efficiency. Based on previously determined optimal calcination at 550 °C, LDH dosages of 50, 75, 100, and 150 g L⁻¹ were tested. Concentrations of Cl⁻, SO₄²⁻, and Ca²⁺ ions in the S1 solution during the sorption process were measured at specific intervals alongside pH and conductivity changes. The concentrations of other ions, such as K⁺, Na⁺, Cr, and Pb, were analyzed after 24 h of sorption. Furthermore, the impact of LDH dosage on removing chloride, sulphates, and calcium ions over time was investigated.

The time-dependent sorption of chlorides by LDH at various dosages from the S1 solution is shown in Figure 8.

The results indicate chloride removal efficiency increases with higher LDH dosages and longer sorption times. Equilibrium sorption of chlorides is achieved after 24 h at both 50 g L⁻¹ and 150 g L⁻¹ LDH dosages. The maximum removal efficiency of Cl⁻ (100%) is attained with a 150 g L⁻¹ dosage, while 92% of chlorides are removed with 100 g L⁻¹ after the same 24 h period. A significant increase in the rate of chloride sorption was observed after 16 h at an LDH dosage higher than 150 g L⁻¹. Lower dosages, specifically 50 and 75 g L⁻¹, achieved only 17% and 54% removal efficiency after 24 h, and at dosages of 75 and 100 g L⁻¹, sorption equilibrium was not reached within 24 h. This indicates that these amounts are insufficient for complete chloride removal, and a longer sorption time would be required to achieve equilibrium. At the 150 g L⁻¹ dosage, the solution density increased, initially slowing the sorption rate for approximately 1000 min. During this period, partial settling of the sorbent at the bottom of the reactor was observed; however, once the sorbent became effectively dispersed in the solution, mass transfer was facilitated, resulting in a sudden increase in chloride removal efficiency.

The sorption of sulphates was also investigated using the S1 solution, as shown in Figure 9. The data reveal that the rate of sulphate sorption increases as the LDH dosage increases. Sorption equilibrium is reached within 3 h, with 100% sulphate removal achieved at 100 g L⁻¹ and 150 g L⁻¹ dosages. To achieve complete sulphate removal with 50 g L⁻¹ and 75 g L⁻¹ of LDH, it takes 12 h, although more than 95% of SO₄²⁻ ions are removed within the first 2 h.

The sorption of calcium ions from the S1 solution is remarkably swift, as shown in Figure 10. Sorption equilibrium is reached in just 1 min when the LDH concentration is 150 g L⁻¹. In contrast, a lower dosage of LDH takes up to 120 min to reach equilibrium. Interestingly, at 50 g L⁻¹ and 75 g L⁻¹ dosages, calcium release begins after 6 h of sorption. This behavior may result from the solution becoming enriched with CO₃²⁻ ions, which could be captured from the air during rehydration, leading to ion exchange with SO₄²⁻ or an increase in OH⁻ groups. Once equilibrium for chloride sorption is reached, calcium readily attaches to the LDH, illustrating the dynamic interplay between these ions.

The sorption efficiency for Cr and Pb ions exceeded 99.9% across all dosages tested. However, the removal of sodium (Na⁺) and potassium (K⁺) was not detected, as hydrotalcite exhibits a low affinity for monovalent ions. Therefore, no removal of Na⁺ or K⁺ was observed in the synthetic solutions. The initial concentrations of chromium (0.17 mg L⁻¹) and lead (8.95 mg L⁻¹) decreased below the detection limits at every LDH dosage. In contrast, the removal of sodium (Na⁺) and potassium (K⁺) was negligible, as hydrotalcite exhibits low affinity for monovalent ions. No significant decrease in Na and K concentrations was observed within 24 h. For instance, at an LDH dosage of 150 g L⁻¹, only 5.3% of Na and 9.3% of K were removed.

Changes in pH and conductivity during the sorption process in the S1 solution were closely monitored, as shown in Figure 11a,b.

As shown in Figure 11a,b, the dynamics of pH and conductivity were meticulously observed during the sorption process in solution S1. Initially, the pH fluctuated between 9 and 12. At the onset of the process, the pH value slightly decreased before gradually increasing again, eventually reaching equilibrium. The fundamental nature of LDH plays a key role here, as it releases OH⁻ ions during sorption. This release occurs almost immediately after a significant amount of calcium is removed from the solution, which typically leads to a reduction in OH⁻ concentration.

Figure 11b presents the changes in conductivity during the same sorption process. A consistent trend was observed across all four LDH dosages tested. Within the first minute after LDH was introduced into the S1 solution, conductivity dropped by approximately 30%, likely due to the rapid removal of calcium ions. After this initial drop, conductivity gradually increased again. This increase is mainly attributed to the release of OH⁻ ions into the solution, resulting from the rehydration of Mg-Al oxides. The following reaction can describe this transformation:

Mg_0.80Al_0.20O_1.10 + 1.10 H₂O → Mg_0.80Al_0.20OH₂(OH)_0.20

(3)

During this process, rehydrated Mg-Al oxides interact with anions present in the solution, which is accompanied by the release of OH⁻ ions [38].

3.4. Sorption from EAFD Wastewater

The sorption rates of chlorides (Cl⁻), sulphates (SO₄²⁻), calcium (Ca²⁺), chromium (Cr), lead (Pb), sodium (Na⁺), and potassium (K⁺) were tested using RW with an LDH dosage of 150 g.L⁻¹ over a period of 24 h. Figure 12 illustrates the sorption dynamics for Ca²⁺, Cl⁻, and SO₄²⁻ over the first 240 min of the process.

The sorption experiments revealed distinct behaviors for different ions. Calcium (Ca²⁺) and sulphate (SO₄²⁻) were removed very rapidly, with more than 99% of these ions eliminated within the first five minutes, indicating a highly efficient ion-exchange process with the layered double hydroxide. In contrast, chloride (Cl⁻) ions were removed more slowly, with only 36% sorption after five minutes and equilibrium reached after 240 min. When all three anions were present, calcium and sulphate were preferentially and rapidly removed, followed by the slower sorption of chloride. This behavior can be explained by the rehydration and reconstruction of Mg-Al oxides in LDH, which releases hydroxide ions (OH⁻) while enabling the incorporation of anions into the interlayer structure, as described with Reaction (3) in Section 3.3 [38].

During sorption in real EAFD wastewater, visible turbidity was observed, likely due to partial dissolution of Al³⁺ ions from the hydrotalcite, which subsequently precipitated with OH⁻. This phenomenon contributes to the maintenance of a relatively stable OH⁻ concentration in the solution, in accordance with the general reaction (4) of anion incorporation into LDH, as reported in previous studies [39]:

$${\mathrm{M}\mathrm{g}}_{1-1}{\mathrm{A}\mathrm{l}}_{\mathrm{x}}{\mathrm{O}}_{1+\mathrm{x}/2}+{\displaystyle \frac{\mathrm{x}}{\mathrm{n}}}{\mathrm{A}}^{\mathrm{n}-+\left(1+{\displaystyle \frac{\mathrm{x}}{2}}\right){\mathrm{H}}_2\mathrm{O}\to {\mathrm{M}\mathrm{g}}_{1-1}{\mathrm{A}\mathrm{l}}_{\mathrm{x}}{\left(\mathrm{O}\mathrm{H}\right)}_2{\mathrm{A}}_{\mathrm{x}/\mathrm{n}}+\mathrm{x}{\mathrm{O}\mathrm{H}}^{-}}$$

(4)

where x is a molar ratio Al/(Mg + Al) and An is an n-valent anion. This phenomenon of the precipitation of Al with OH⁻ helps to more or less maintain a constant concentration of OH⁻ groups in solution [7].

The sorption study [40] reports the precipitation of Ca²⁺ ions with hydrotalcite from a CaCl₂ solution, forming Ca(OH)₂ through OH⁻ ions released from hydrotalcite. The authors suggest that Cl⁻ ions are sorbed only when calcium precipitates with the released OH⁻ ions as Ca(OH)₂. Furthermore, chloride sorption is more likely in systems where the solution contains an excess of OH⁻ groups.

In contrast, in real EAFD wastewater, where carbonates are present, Ca(OH)₂ formation is unlikely. Partially, the OH⁻ groups released from hydrotalcite appear to inhibit chloride sorption because they have a higher affinity for the LDH than chloride ions. Consequently, chloride ions begin to adsorb only after Al³⁺ ions dissolve from hydrotalcite to form Al(OH)₄⁻, and only when sulphate ions are absent from the solution. Moreover, divalent anions such as SO₄²⁻ or CO₃²⁻ exhibit higher selectivity than monovalent ions. The selectivity of Mg-Al oxides increases with increased anion charge and decreased anion size. Therefore, hydrotalcite does not readily combine with Cl⁻ ions, as it preferentially reacts with OH⁻ formed during Reaction (4) [11]. Sorption of heavy metals was highly efficient, with chromium (Cr) and lead (Pb) completely removed within the first minute, largely unaffected by the presence of other anions. Conversely, monovalent cations such as sodium (Na⁺) and potassium (K⁺) were poorly removed, with less than 10% sorption achieved after 240 min. This indicates a low affinity of LDH for monovalent cations, which are primarily sorbed onto the surface, whereas anions are incorporated into the interlayer space.

During the LDH sorption process, the pH of the RW solution decreased from the initial value of 12.15. Initially, the pH dropped to 8.56 within the first minute of sorption. After this, the pH gradually increased, reaching 12.0 by the end of the process. This increase in pH is attributed to the release of OH⁻ ions during the sorption process as the LDH interacts with the ions in the wastewater. A similar trend, but with a more intensive change, was observed in the conductivity of the solution during sorption. In the first three minutes, conductivity dropped sharply from 11.88 mS cm⁻¹ to a minimum of 2.8 mS cm⁻¹. This decrease corresponds to the removal of almost all calcium and sulphate ions from the solution. After 240 min, conductivity increased to 4.23 mS cm⁻¹, reflecting the rehydration and ion exchange activities occurring within the system. Over the next 24 h of sorption, the conductivity rapidly increased to 9.00 mS cm⁻¹. These changes in pH and conductivity during the sorption process closely mirror the behavior observed in the S1 solution, indicating similar sorption dynamics for both the real wastewater and the synthetic solution.

3.5. SEM Observation and Structural Characterization of the LHD After Sorption

After the sorption experiments were conducted with a dosage of 100 g L⁻¹ for both the S1 and RW solutions, the suspension was filtered, and the solid residue was air-dried for further analysis. The dried samples were subjected to SEM, EDX, and XRD analyses to investigate the structural changes in the adsorbent.

The XRD analysis observed that the adsorbent largely retained its original hydrotalcite structure (Figure 13). The amount of adsorbed ions (Cl⁻, SO₄²⁻) was insufficient to form any new, visible crystalline phases. Notably, the phase of meixnerite disappeared in both the S1 and RW samples.

Further examination of the XRD patterns from the S1 and RW samples revealed a decrease in the intensity of the hydrotalcite peaks. This reduction in peak intensity suggests ion exchange between the Cl⁻ and SO₄²⁻ ions and the hydrotalcite structure. A noticeable increase in the peak intensity at 34° 2θ was observed, indicating the calcium sorption into the structure. The peak broadening reveals the change in lattice parameters of the input material due to intercalation of ions in the lattice of LDH, tension, and the occurrence of defects, as well as a decrease in the crystallite dimension.

Figure 14 and Figure 15 present the SEM images and EDX analysis of LDH after 24 h of sorption with the S1 and RW solutions, respectively. The SEM images show that the morphology of the LDH is similar after sorption from both solutions. By comparing the initial LDH morphology (before sorption, as shown in Figure 6 and Figure 7) with the structure after sorption (Figure 15), it is evident that the microstructure and chemical composition of the LDH have changed. The initial calcined LDH sample consists of aggregates sized around 100–300 nm, with distinct individual layers. In contrast, the LDH, after sorption, loses its organized aggregate structure and instead exhibits a swollen network of seemingly mismatched clusters, with a larger size ranging from 1 to 2 µm. These clusters also contain visible whisker-like particles. Hydrotalcites can absorb water into their interlayer spaces. Water molecules help expand the interlayer and participate in the ion-exchange process, as the LDH structure becomes more hydrated and stable. Initial ion absorption (including water) typically involves hydration followed by ion-exchange processes of cations (such as Ca²⁺, Pb²⁺, Zn²⁺, and Cr³⁺) facilitated by the Mg²⁺ and Al³⁺ ions in the hydrotalcite (HT) structure. Subsequently, anions (such as Cl⁻ and SO₄²⁻) are bound through intercalation into the interlayer space of hydrotalcite [41,42]. The nanometric scale of the particles and LDH layers, the presence of OH⁻ groups, the high internal surface area due to the fluffy, multi-layered, and three-dimensional structure, and the combination of active, exchangeable cations (like Mg and Al) with their amphoteric character work synergistically during the sorption process. These characteristics enable the LDH material to achieve remarkable efficiency in the removal of both cations and anions from wastewater.

EDX analysis revealed comparable compositions in the adsorbents from the S1 and RW solutions after the sorption process. The main LDH elements (Mg, Al and O, marked with white in Figure 14) were present in similar amounts, as were the levels of Ca, Cl and S. However, the percentage of Ca, Cl and S in the LDH from the RW solution is slightly lower. The presence of sulphur and potassium (marked with red in Figure 14) cannot be ruled out, but is difficult to confirm due to the low content and detection limits of the method used. Silicon was only detected in the LDH sample after sorption from the RW solution, which suggests the possible presence of silicon in the RW solution and the incorporation of silica (silicate) species into the LDH structure during sorption from the RW solution. Conversely, the Si content is quite low (0.2%) (marked with red in Figure 15), which requires further investigation. As in Figure 6, carbon (C) was excluded from the EDX analysis in Figure 14 and Figure 15 due to the improvement in the electrical conductivity of the sample.

The results obtained from sorption experiments using the S1 and RW solutions with an HT dosage of 100 g L⁻¹ are summarized and compared in

Table 2. Ion removal with hydrotalcite-2H (LDH) calcined at 500 °C. The summary of ionic species in S1 and RW solutions after sorption.

Ionic Species	Symbol	Ion Removal (%)
		S1 Solution	RW Solution
Chloride	Cl−	98 ± 0.9	100 ± 0.91
Sulphate	SO42−	100 ± 0.7	100 ± 0.12
Sodium	Na+	0 ± 9.7	0 ± 0.11
Potassium	K+	0 ± 14.2	0 ± 15.1
Calcium	Ca2+	100 ± 0.21	100 ± 2.1
Carbonates	CO32−	-	0 ± 0.01
Zinc	Zn2+	100 ± 0.09	95.6 ± 3.2
Lead	Pb2+	100 ± 0.01	88.7 ± 11.8

, which shows the total ion removal efficiency (%) after 24 h of sorption.

The comparison of the results in Table 2 shows that sorption from the synthetic solution is 100% for most elements, except for chlorides. In contrast, the sorption from the RW source is also 100%, although certain cations, such as zinc and lead, are not fully sorbed. Additionally, sodium (Na) and potassium (K) are only sorbed in minimal amounts for both solutions, and hydrotalcite is generally not utilized for their removal.

3.6. Electrodialysis Tests

Ion removal from S1 and RW solutions was also evaluated by means of electrodialysis. Before the electrodialysis experiments, it is inherent to determine the limiting current density (i_lim) for both working solutions, which can be obtained using current–voltage curves. It can be seen in Figure 16 that the current–voltage curves showed a conventional shape, with three regions (except for the CEM, due to power supply limiting) and a limiting state [43]. The first region presents a quasi-ohmic behavior, with a linear dependence between the current density and membrane potential. The second region is associated with a limiting state, with a significant increase in the resistance of the membrane (an inclined “plateau”) and where the i_lim is obtained. The i_lim value for AEM and CEM in contact with the S1 solution was 6.4 mA cm⁻¹ and 8.6 mA cm⁻¹, respectively, while for the RW solution the i_lim values were 20.0 mA cm⁻¹ and 30.0 mA cm⁻¹. The third region is linked to the overlimiting current density condition, which can be attributed to water splitting and electroconvection mechanisms [44].

Based on the current–voltage curve results, the current density value to be applied in the electrodialysis experiments with both solutions was limited by the AEM. According to the literature [45], when operating under galvanostatic mode and underlimiting conditions, it is convenient to apply a current density value that corresponds to 70–80% of the limiting one, in order to mitigate the concentration polarization effects. In this sense, for the ED tests with the S1 solution, it was applied at 5.3 mA cm⁻² for approximately 840 min, while for the RW solution, the experiments were carried out for 240 min with an imposed current density of 15 mA cm⁻². The electric conductivity (k) of the solution in the diluted and concentrated reservoirs was monitored each hour, and the values are presented in Figure 17. Comparing both ED setups for the S1 solution (Figure 17a) and for the RW solution (Figure 17b), it can be noted that this parameter presented a similar behavior, with k values decreasing in the diluted reservoir along the treatment time, while increasing in the concentrated reservoir. This is an expected performance in the electrodialysis process [46] since the ionic species are transferred from the diluted to the concentrated reservoir. Moreover, it indicates a successful demineralization of the working solution, corresponding to a reduction of approximately 98% of the electric conductivity for the S1 solution and 95% for the RW solution.

The diluted and concentrated solution pH values for both ED setups were also recorded hourly. As can be observed, the pH of the concentrated reservoir solution remained nearly unchanged during the electrodialysis experiments at a value around 12 and 12.5 for the S1 (Figure 18a) and RW (Figure 18b) solutions, respectively. However, a different behavior was reported for the pH values of the diluted reservoir solution, which presented a decreasing tendency, especially after half of the treatment time. This behavior was also reported in [23] and may be associated with coupled effects of concentration polarization at the interface between the solution and the anion exchange membrane. Since ion concentration is one of the directly proportional parameters in the determination of the i_lim value [47], the demineralization of the diluted solution may lead to the system reaching or exceeding the i_lim value, and the generation of OH^- and H⁺ due to water splitting in the membrane/solution region may appear. Thus, the OH^- may pass through the anion exchange membrane, while the H⁺ ion may influence the pH value of the solution in the diluted reservoir [48].

The removal of chloride, sulphate, and calcium ions from the diluted solution is shown in Figure 19. With the S1 solution, it can be noted in Figure 19a that electrodialysis performed better than sorption in removing chlorides, reaching more than 99% extraction of these ions in 800 min, whereas at sorption this value was achieved after more than 1400 min with 150 g L⁻¹ of LDH. However, this behavior is not observed when it comes to the removal of SO₄²⁻ and Ca⁺ ions, since the removal rate of these ions was slower than sorption, despite reporting a percent extraction of 99.35% for sulphate and 98.91% for calcium ions. For the RW solution, none of the specified ions reached more than 90% of removal at the end of the ED tests, as can be seen in Figure 19b. The percentage removal of sulphate ions presented a singular behavior, where a considerable increase in its values can be seen after only 150 min. This can be explained by the competition of coexisting anions to pass through the AEM, where the transport of monovalent ions and those with a smaller ionic radius, such as chloride, is favored over divalent ions, such as sulphate [49]. Moreover, sulphate ions species had the lowest value of percentage removal, with less than 70%, compared to calcium and chlorides (approximately 73% and 83%, respectively).

From the comparison of the kinetic curves in Figure 19, it is evident that the sulphate removal rates differ significantly between S1 and RW, indicating that the mechanisms of sulphate removal are not the same. Although the initial concentrations of calcium, sulphate, and chloride were similar in S1 and RW (with RW having slightly higher conductivity), it can be concluded that sulphate removal from the RW solution is influenced by other ions and species present. These species may include carbonates, hydroxides, and heavy metal cations, which can significantly affect the rate of sulphate removal. This is also reflected in the higher removal efficiencies observed for Pb and Zn compared to other studied ions.

Interestingly, carbonates showed the lowest removal efficiency in the RW solution (56.46%). Conversely, the removal rates of Ca²⁺ and Cl⁻ within the 0–240 min range are higher in RW than in S1, despite the overall lower removal efficiency of RW at 240 min.

This behavior can be explained by the fact that the synthetic solution (S1) is more sensitive to polarization phenomena. In contrast, real water (RW), which contains a wider variety of ionic species, provides a greater number of charge carriers across the membranes. As a result, polarization is suppressed (or at least delayed), leading to improved process efficiency and a higher ion removal rate. It is therefore believed that the difference in transport rates arises from the ability of the RW system to operate at higher current densities, as illustrated by the polarization curve in Figure 16. Higher current density increases the amount of energy available to drive ion migration toward the concentrate compartment, which accounts for the enhanced performance observed in RW compared to S1. On the other hand, extending the duration of the experiment could further improve the overall ion removal efficiency in the RW system, allowing the process to take fuller advantage of its higher transport capacity.

The values of ion removal for all ionic species contained in S1 and RW solutions are summarized in

Table 3. Ion removal summary of ionic species in S1 and RW solutions by electrodialysis.

Ionic Species	Symbol	Ion Removal (%)
		S1 Solution	RW Solution
Chloride	Cl−	99.60 ± 0.11	83.49 ± 0.20
Sulphate	SO42−	99.36 ± 0.21	69.84 ± 3.12
Sodium	Na+	98.28 ± 0.28	75.18 ± 12.50
Potassium	K+	99.47 ± 0.08	82.77 ± 7.78
Calcium	Ca2+	98.91 ± 0.04	72.83 ± 3.31
Carbonates	CO32−	-	56.46 ± 5.35
Zinc	Zn2+	-	83.10 ± 5.20
Lead	Pb2+	90.32 ± 3.82	95.12 ± 1.10

. Overall, electrodialysis performed well in the desalination of both solutions, emphasizing the removal of sodium, potassium, and carbonate ions, which were not affected by the sorption process. However, it is important to highlight the occurrence of salt precipitation, especially in the concentrated reservoir, probably due to the high concentration of calcium and sulphate during the experiments. This phenomenon may cause severe damage to the integrity of the ion exchange membrane, reducing its mechanical properties and lifetime, as well as the efficiency of the electrodialysis process [50]. To overcome this situation, it is recommended to use a pre-treatment process that removes precipitable components.

3.7. Sorption and Electrodialysis as a Two-Stage Treatment

Considering the non-removal of monovalent ions like Na⁺ and K⁺ from sorption and the salt precipitation occurring in electrodialysis, the integration of both techniques in a two-stage treatment process with wastewater simulating the composition of RW, named S2, was studied. The solution S2 was prepared synthetically, reflecting the output of the sorption process with the RW solution. The composition of S2 is presented in Table 1. As a first step, it is suggested that LDH sorption be used since this technique presents a high capacity for the removal of calcium, chlorides, and sulphate ions, suppressing the formation of precipitates in the ED process. As a second step, the use of electrodialysis may reduce the salinity of the solution, removing monovalent ions, such as Na⁺ and K⁺, together with CO₃²⁻.

The current–voltage curves are shown in Figure 20, where it can be seen that the AEM limited the current density value to be applied, with an i_lim = 29.1 mA cm⁻².

At this process stage, electrodialysis experiments were carried out with an applied current density of 21.8 mA cm⁻² for 210 min. As in previous ED experiments, conductivity and pH values were monitored, and a similar behavior was reported in previous experiments. The conductivity value of the solution in the diluted reservoir decreased from (8.98 ± 0.42) mS cm⁻¹ to (0.84 ± 0.06) mS cm⁻¹, representing a reduction higher than 90% at the end of the experiments. The pH value of the concentrated solution remained nearly unchanged (pH = 12.8) throughout the experimental time. At the same time, a sharp decrease was observed in the diluted reservoir solution between 100 and 150 min, from around 12.5 to 2.8, being kept virtually constant after this period. As mentioned before, this is probably due to the coupled effects of concentration polarization that occur according to the working solution’s demineralization. Considering that the solution resulted in the ED diluted compartment having a low pH value, such produced water would be suitable for usage in the acid route in the hydrometallurgical process of Zn recovery. The percentage of ion removal from the S2 solution is shown in

Table 4. Ion removal efficiency in S2 solution.

Ionic Species	Symbol	Ion Removal (%) S2 Solution
Sulphate	SO42−	76.98 ± 2.95
Sodium	Na+	97.85 ± 0.21
Potassium	K+	98.86 ± 0.27
Carbonates	CO32−	98.31 ± 0.45
Zinc	Zn+	>90%
Lead	Pb2+	>93%

. Except for sulphate, which was added only to meet the concentrations of potassium and sodium ions, it can be seen that this step presented an ion removal higher than 90% for all ionic species in the working solution. The Na⁺, K⁺, and CO₃²⁻ ions, unable to be removed by sorption, reported a final concentration of (14.72 ± 2.21) mg L⁻¹, (12.10 ± 4.00) mg L⁻¹, and (5.36 ± 1.19) mg L⁻¹ in the diluted reservoir. Additionally, after the ED experiments the concentration of Pb⁺ and Zn⁺ ions, partially removed by sorption, presented values lower than the limit detection of the method, that is, less than 0.02 mg L⁻¹; in this sense, sorption–electrodialysis as a two-stage treatment may be suitable to reduce the high salinity of the real EAFD wastewater. With this experimental setup, obtaining water with a low salinity of the final EAF wastewater solution was possible, and it may be reused in hydrometallurgical operations as process water. In the absence of calcium and chloride (after their removal), the results of electrodialysis (ED) showed significantly higher removal efficiencies for K⁺, Na⁺, and CO₃²⁻, and also for sulphates. The calcium content appears to be one of the key parameters in the treatment of real EAFD wastewater, not only for ED but also for sorption using LDH.

From the comparison of the calculated EC values for S1 (38.55 kWh·m⁻³), RW (52.13 kWh·m⁻³), and S2 (61.36 kWh·m⁻³), it is evident that purifying the S1 solution requires less energy than treating the real RW solution. This difference is primarily attributed to the higher applied current density for RW and the higher impurity content in the RW solution. Integrating a sorption step before electrodialysis in the two-stage process slightly increases the EC, as observed for S2. On the other hand, the difference is insignificant, and the proposed two-stage process could be proven sustainable, in addition to preventing calcium and sulphate precipitation inside the electrodialysis cell.

Although the specific energy consumption values remain relatively high, several strategies can be adopted to mitigate this drawback. Operating under potentiostatic control constrains the applied potential, thereby limiting energy consumption even when treatment times are extended. In addition, the use of advanced low-resistivity membranes can significantly reduce ohmic losses, enhancing overall process efficiency and lowering the energy demand per cubic meter of the treated solution. Finally, electrodialysis and related electrochemical separation technologies are inherently compatible with decentralized renewable energy sources, such as solar and wind power. Their integration not only reduces dependence on conventional electricity grids but also enhances process sustainability and supports global decarbonization objectives [51].

4. Conclusions

This study demonstrated the successful application of a two-stage treatment system for electric arc furnace dust (EAFD) wastewater and synthetic solutions (S1, S2), combining sorption with Mg-Al layered double hydroxides (LDHs) and subsequent electrodialysis (ED). Sorption tests with synthetic solution (S1) showed high LDH efficiency in removing multivalent ions and heavy metals: Cr and Pb were removed almost immediately with >99.9% efficiency, while Ca²⁺ and SO₄²⁻ achieved >95% removal within 1–12 min at doses of 50–150 g L⁻¹. Cl⁻ required up to 24 h to reach equilibrium at 150 g L⁻¹. Monovalent ions (Na⁺, K⁺) were minimally removed, confirming the low LDH affinity. Similar trends were observed in real EAFD wastewater (RW), where Ca²⁺ and SO₄²⁻ were removed >99% within 5 min, Cl⁻ reached equilibrium after 240 min, and toxic metals were removed almost immediately. Electrodialysis achieved significant demineralization of the synthetic (S1) and real EAFD wastewater (RW) solutions. The electric conductivity of synthetic wastewater (S1) decreased by 98% after 840 min, while real wastewater (RW) decreased by 95% after 240 min. Ion removal efficiency in S1 reached >99% for Cl⁻, 99.3% for SO₄²⁻, and 98.9% for Ca²⁺, whereas removal efficiency in RW was lower (Cl⁻ ≈ 83%, Ca²⁺ ≈ 73%, SO₄²⁻ < 70%) due to competing ions. The worst removable ion in ED for RW was CO₃²⁻; however, by implementing the sorption process before ED, the efficiency of carbonate removal increased substantially in the S2 solution from 56% to 98%. As a result of the combination of sorption–ED treatment from S2 (simulated wastewater after sorption from RW), >90% removal of Na⁺, K⁺, and CO₃²⁻ and reduced Pb and Zn below detection limits were observed. Despite challenges such as pH decrease in the dilute compartment and salt precipitation in the concentrate, the integrated approach was highly effective in producing low-salinity effluents suitable for reuse in hydrometallurgical processes. Overall, the combination of LDH sorption and electrodialysis is an effective strategy for removing heavy metals, Ca²⁺, SO₄²⁻, and, partially, Cl⁻ ions from EAFD wastewater. An optimized LDH dosage and contact time significantly affect treatment efficiency.

Future work should focus on the study of the role of specific species, especially Ca or carbonates and sulphate interactions, for such systems and the mechanism of ion removal in ED, optimizing LDH regeneration and reusability, long-term ED membrane performance using real wastewater and proving the proposed two-stage process from a long-term perspective, and strategies to mitigate scaling in concentrates. Extending the system to pilot or industrial scale will be essential for assessing energy efficiency, cost-effectiveness, and environmental impact. Additionally, exploring LDH modifications for improved affinity toward monovalent ions could enhance treatment performance and broaden applicability across diverse wastewater streams.