Arsenic Removal from Water Using Mg-Based Adsorbents in the Presence of Silicic Acid

Abstract

Dissolved silicic acid (Si) in groundwater can reduce the As-removal performance of adsorbents used for treating contaminated water. However, its effects on Mg-based adsorbents remain largely unexplored. In this study, As-removal tests were conducted under various test conditions to evaluate the suitability of Mg-based adsorbents (MgO, Mg(OH)₂, and MgCO₃) for the purification of As-contaminated water in the presence of Si. As-removal performance varied significantly depending on the Mg-based adsorbent type and dosage (W_Ad0/V), As valence, and the initial As and Si (C_Si0) concentrations. In some cases, As removal improved at relatively low C_Si0; however, overall performance decreased with increasing C_Si0 for all Mg-based adsorbents. Moreover, compared with Mg(OH)_2, the performance of MgO and MgCO₃ was more strongly affected by Si. This inhibition is attributed to competition between Si and As for adsorption sites on the adsorbent surface. Furthermore, for MgO and MgCO₃, the amount of As removed by coprecipitation with secondarily generated Mg(OH)₂ aggregates was inferred to decrease with increasing C_Si0, because higher C_Si0 lowered the solution pH. Overall, MgO and Mg(OH)₂ can function effectively as adsorbents for As treatment when W_Ad0/V is appropriately selected, considering the range of Si concentrations typically found in groundwater.

1. Introduction

Arsenic (As) occurs in inorganic and organic forms and exhibits both metallic and non-metallic properties [1]. The prevalent inorganic forms of As are trivalent As (arsenous acid; arsenite; As(III)) and pentavalent As (arsenic acid; arsenate; As(V)). Inorganic As(III) is highly toxic, more toxic than As(V), and more mobile than other forms [2]. Symptoms associated with chronic As toxicity have been reported in regions where groundwater is used for drinking, [3,4,5]. Chronic exposure to As-contaminated drinking water can cause a wide range of health effects. Typical As-associated manifestations include skin lesions (e.g., melanosis, leucomelanosis, and keratosis), neurological effects, obstetric problems, high blood pressure, diabetes mellitus, respiratory and vascular diseases including cardiovascular disease, and cancers of the skin, lung, and bladder [5,6]. To mitigate these harmful effects of As, drinking-water quality guidelines of the World Health Organization (WHO) have set 0.01 mg/L as a provisional guideline value for As [6]. The As environmental and wastewater standards in Japan are 0.01 and 0.1 mg/L, respectively [7,8].

Many studies have reported that As contamination of groundwater occurs worldwide [9,10]. Global As-prediction models are being developed to estimate exposed populations in regions beyond those where As-related impacts are currently evident. Podgorsk and Berg proposed a global prediction map of groundwater As exceeding 0.01 mg/L using a random-forest machine-learning model trained on 11 geospatial environmental parameters and more than 50,000 aggregated measurements of groundwater As concentration [11]. By combining this global As-prediction model with household groundwater-usage statistics, they estimated that 94 million to 220 million people may be exposed to elevated As concentrations in groundwater, with the vast majority (94%) residing in Asia [11].

The development of cost-effective As-treatment methods is essential because many regions that rely on As-contaminated groundwater for drinking water are economically disadvantaged. Liu et al. [12] reviewed conventional adsorbents (e.g., activated carbon, zeolites, and waste-derived sorbents) and more recent functional adsorbents (e.g., graphene oxide (GO), carbon nanotubes (CNTs), and metal organic frameworks (MOFs)). After briefly introducing other purification technologies, including ion exchange, photocatalysis, and membrane technology, they concluded that adsorption is an attractive option because of its simple design, low cost, and ease of operation. In their review, they compared the As-removal capacities of functional adsorbents using the As distribution coefficient and discussed the regeneration and disposal of spent adsorbents. They further noted that advanced sorbents such as ZrO(OH)₂/CNTs can exhibit high As distribution coefficients and good regenerability [12]. However, the review did not address management of the As-containing wastewater that is likely to be generated during regeneration of spent adsorbents.

A key concern is that As removal by the adsorbent may be compromised by the presence of coexisting ions in the contaminated water. Studies on Fe-based adsorbents for As removal have frequently reported that silicic acid (Si) can markedly affect As uptake. Meng et al. investigated the effects of silicate (Si), sulfate (SO₄²⁻), and carbonate (CO₃²⁻) on the removal of As(III) and As(V) by coprecipitation with FeCl₃ [13]. Batch experiments were conducted using a KNO₃ solution (0.04 M) as a background electrolyte under either a N₂ atmosphere or ambient air. The solution pH was adjusted with KOH and HNO₃ over the range of 3–11, and initial As concentrations ranged from 0.05 to 1.4 mg/L. Coprecipitation was initiated by adding FeCl₃ and, where applicable, SO₄²⁻ or Si stock solutions to the electrolyte. As pH decreased from 10 to 6, As(V) removal increased from 0 to 95%. Conversely, As(III) removal increased from 20 to 80% as pH increased from 4 to 9.5. Removal of As(V) and As(III) was essentially identical under a N₂ atmosphere and in ambient air, indicating that carbonate species had a negligible influence on As removal over pH 4–9.5. Similarly, adding SO₄²⁻ did not affect As(V) and As(III) removal over pH 4–10; no apparent change in As(V) and As(III) removal was observed as sulfate concentration increased from 0 to 300 mg/L at an equilibrium pH of 6.8. Between pH 4 and 10, less than 30 mg/L Fe remained soluble in the KNO₃ solution (0.04 M). However, at 5 mg/L Si, increasing pH from 8.6 to 9.4 increased the soluble Fe concentration from 47 to 2040 mg/L and substantially reduced As removal only at pH > 8.6. Overall, As(III) and As(V) removal decreased from approximately 90 to 45% as Si concentration was increased from 1 to 10 mg/L. At pH > 7, As(V) removal decreased in the presence of 4.5 mg/L Si, whereas at pH < 6.8, As(V) removal was unaffected by Si. In contrast, As(III) removal was significantly reduced by Si across a broad pH range, and the magnitude of the Si effect increased as pH increased from approximately 4 to 9.5. The authors concluded that high Si concentrations render As(III) removal by coprecipitation with FeCl₃ ineffective [13]. Waltham and Eick investigated the adsorption behavior of As(V) and As(III) on goethite (FeO(OH)) in the presence of Si [14]. For the kinetic experiments, As was added only after Si adsorption had reached steady state (60 h). Adsorption of both As(III) and As(V) on goethite was rapid and essentially complete within 2 h. As pH decreased, the adsorption rates of As(III) and Si decreased, whereas the adsorption rate of As(V) increased. Furthermore, Si uptake on goethite decreased as pH was reduced, while As(III) and As(V) adsorption remained nearly constant over pH 4–8. Si adsorption exhibited an initial rapid phase followed by a considerably slower phase. Across all pH values and concentrations tested, Si decreased both the rate of As(III) adsorption and the total amount of As(III) adsorbed. Inhibition of As(III) adsorption ranged from ~4% at pH 6 and 0.10 mM Si to 40% at pH 8 and 1.0 mM Si. At 0.10 mM Si, As(III) adsorption was reduced the most at pH 4 followed by pH 8. At 1.0 mM Si, As(III) adsorption was reduced the most at pH 8, followed by similar reductions at pH 4 and 6. Si also reduced the rate of As(V) adsorption, with the rate decreasing further as pH and Si concentration increased; however, the total amount of As(V) adsorbed remained nearly constant from pH 4 to 8 at both Si concentrations. The authors concluded that Si at naturally occurring concentrations can reduce both the rate and total amount of As adsorbed on goethite [14].

Mg-based compounds have also been investigated for As removal. Park et al. [15] suggested a method to withdraw As(V) from a plant solution with approximately 470 mg/L of As(V) and 70 g/L of Mo(VI) using MgCl₂ or MgSO₄. They reported that As(V) can be removed by precipitation of Mg₃(AsO₄)₂ at pH 8–10; however, at pH 12, this process fails because Mg precipitates predominantly as Mg(OH)₂ [15]. Yu et al. [16] first prepared flower-like hydromagnesite (F-hydromagnesite) or nest-like hydromagnesite (N-hydromagnesite) by mixing Mg(NO₃)₂ and K₂CO₃ solutions, and subsequently calcined them to produce two types of MgO (F-MgO and N-MgO, respectively). The As-adsorption capacities of the MgO samples were substantially higher than those of the corresponding hydromagnesites, and N-MgO exhibited a higher adsorption capacity than F-MgO [16]. Tresintsi et al. [17] investigated the use of granulated MgO for regenerating Fe oxyhydroxide arsenic adsorbents. Their batch adsorption tests were conducted at initial As(III) and As(V) concentrations of 0.25–12.5 mg/L and pH 10–12. They reported maximum adsorption capacities of 59.4 mg/g for As(V) (at a residual concentration of approximately 5 mg/L) at pH 10 and approximately 50 mg/g for As(III) (at a residual concentration of 3 mg/L) at pH 11. Moreover, As K-edge EXAFS spectra indicated that As(V) and As(III) likely adsorbed onto Mg(OH)₂ formed by hydrolysis of MgO [17]. Lin et al. [18] prepared monodispersed, porous, pinecone-like Mg(OH)₂ (PLMH) for As(V) removal and investigated the effects of coexisting ions (PO₄³⁻, CO₃²⁻, SO₄²⁻, NO₃⁻, and Cl⁻) at a 1:1 concentration ratio relative to arsenate. They reported that SO₄²⁻, NO₃⁻, and Cl⁻ had a minimal effect on As(V) removal, whereas PO₄³⁻ and CO₃²⁻ significantly inhibited As(V) removal due to competitive adsorption [18]. However, they did not examine the Si effects on As(V) removal. The presence of Si in solutions has been reported to affect the leaching behavior of As from spent Mg-based adsorbents containing As [19]. This suggests that Si may also influence As adsorption onto Mg-based adsorbents. Therefore, the Si effects on As adsorption by Mg-based adsorbents should be investigated.

The above review of the literature suggests that although silicic acid can substantially affect the As-removal performance of adsorbents, its effects on Mg-based adsorbents have scarcely been studied. Therefore, in this study, batch As-removal tests were conducted using powdered reagents of common Mg-based compounds as adsorbents and synthetic As-contaminated waters containing Si. The test parameters were adsorbent type (MgO, Mg(OH)₂, and MgCO₃), adsorbent concentration (approximately 0.2 and 2 g/L), As valence (trivalent and pentavalent), initial concentration of As (approximately 1 and 10 mg/L), and initial concentration of Si (five concentration levels between 0 and 120 mg/L, including 0 mg/L). The differences in As-adsorption behavior associated with these parameters were evaluated, and the effects of Si on the As-removal performance of Mg-based adsorbents were investigated in detail.

2. Materials and Methods

2.1. Mg-Based Adsorbents

Commercial MgO, Mg(OH)₂, and MgCO₃ powdered reagents were used as Mg-based adsorbents. Unless otherwise specified, all reagents were procured from FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan). The measurement values of median particle size (D_p50; μm), BET surface area (S_BET; m²/g), Mg content (α_Mg; %), and reagent purity (based on α_Mg) (P; %) in

Table 1. Characteristics of the Mg-based adsorbents used in this study.

No.	Adsorbent	Dp50 (μm)	SBET (m2/g)	αMg (%)	P (%)
(1)	MgO	1.54	4.3	59.1	98.0
(2)	Mg(OH)2	4.13	22.0	40.6	97.3
(3)	MgCO3	15.0	26.0	24.8	86.1

Data obtained from Sugita et al. [19]. α_Mg: Mg content; D_p50: median particle size; P: reagent purity; S_BET: BET surface area.

were obtained from a previous study [19]. The main reason that the p values were less than 100% was the presence of adsorbed water.

2.2. As-Removal Tests with As-Contaminated Water Containing Si

2.2.1. Synthetic As-Si Contaminated Water

Powdered NaAsO₂ and Na₂HAsO₄·7H₂O were dissolved separately in deionized water to prepare As(III) and As(V) stock solutions (2000 mg/L), respectively. The Si stock solution was prepared by dissolving approximately 1 g of sodium silicate solution (21.2–23.8% with SiO₂/Na₂O = 3.0–3.6 molar ratio at the nominal value) in 1 L of deionized water. The Si concentration in the stock solution was measured using inductively coupled plasma–atomic emission spectrometry (ICP-AES). Synthetic As-contaminated water containing Si was prepared by mixing and diluting the stock solutions with deionized water at foreordained ratios. The synthetic contaminated water comprised two initial As concentrations (C_AS0) (approximately 1 and 10 mg/L) and five initial Si concentrations (C_Si0) ranging from 0 to approximately 120 mg/L (including 0 mg/L). The solution pH of the contaminated water was adjusted and maintained at approximately 7 using HNO₃ and NaOH solutions. The pH directly after adjustment is represented to as the initial pH (pH₀).

2.2.2. Experimental Procedure

Approximately 0.01 or 0.1 g of an adsorbent and 50 mL of synthetic contaminated water were added to a reaction tube (50 mL polypropylene centrifuge tube). These masses correspond to adsorbent concentrations (W_Ad0/V) of approximately 0.2 and 2 g/L, respectively, where W_Ad0 (g) and V (L) are the amounts of adsorbent and liquid, respectively. The tube was capped and shaken for 24 h in a thermostatic shaker at room temperature, after which it was centrifuged at 4500 rpm for 20 min. Incidentally, in this study, six shakers of the same type were used to conduct numerous tests simultaneously. Even with the shaking speed setting memory (rotary-dial type) of these six units set to the same position, the measured shaking speeds ranged from 150 to 180 rpm due to individual differences in the shakers. The supernatant was then filtered through a syringe filter (0.45 μm pore size), and the filtrate (treated water) was collected in a polypropylene bottle. The pH of the filtrate was measured using a pH meter (LAQUA F-72, HORIBA, Ltd., Kyoto, Japan). The As concentration in each filtrate was measured by inductively coupled plasma–mass spectrometry (ICP-MS; Agilent 7700X, Agilent Technologies Japan, Ltd., Hachioji, Japan), and the Mg and Si concentrations were measured using ICP-AES (SII SPS3500DD, Seiko Instruments Inc., Chiba, Japan; PlasmaQuant 9100 Elite, Analytik Jena Japan Co., Ltd., Yokohama, Japan).

2.2.3. Experimental Reproducibility

Experimental reproducibility was confirmed by conducting As-removal tests for each adsorbent in triplicate under specific conditions. The characteristics of the synthetic As(V)-Si contaminated water used in the reproducibility experiments were C_AS0 = 1.054 mg/L, C_Si0 = 56.6 mg/L, and pH₀ = 6.74, and those of the synthetic As(III)-Si contaminated water were C_AS0 = 1.069 mg/L, C_Si0 = 56.6 mg/L, and pH₀ = 7.26. The value of W_Ad0/V was approximately 2 g/L (

Table 2. Mean values and standard errors obtained from three replicated experiments.

Adsorbent	As Valence	WAd0/V (g/L)	CAS (mg/L)	CSi (mg/L)	CMg (mg/L)	pHf
MgO	As(V)	2.05 ± 0.00	0.194 ± 0.008	14.0 ± 0.8	14.0 ± 0.6	10.54 ± 0.36
Mg(OH)2	As(V)	2.04 ± 0.00	0.033 ± 0.001	33.7 ± 0.5	11.7 ± 0.2	9.81 ± 0.06
MgCO3	As(V)	2.03 ± 0.00	0.625 ± 0.009	25.4 ± 1.4	43.0 ± 0.6	9.95 ± 0.01
MgO	As(III)	2.04 ± 0.00	0.131 ± 0.002	5.26 ± 0.06	13.6 ± 0.5	10.65 ± 0.04
Mg(OH)2	As(III)	2.03 ± 0.00	0.191 ± 0.008	33.4 ± 0.2	10.6 ± 0.2	9.81 ± 0.01
MgCO3	As(III)	2.01 ± 0.00	0.601 ± 0.015	9.27 ± 0.40	46.3 ± 0.3	9.99 ± 0.00

C_AS: residual As concentration; C_Mg: leached Mg concentration; C_Si: residual Si concentration; pH_f: treated water pH; W_Ad0/V: adsorbent addition concentration.

). The measured values of the residual As concentration in the treated water (C_AS), residual Si concentration (C_Si), leached Mg concentration (C_Mg), treated water pH (final pH, pH_f), and standard errors are also listed in Table 2. These results indicate that the experimental reproducibility of this method was acceptable.

2.3. Sample Preparation for X-Ray Diffraction (XRD) Analysis

The forms of As and Si adsorbed by the Mg-based adsorbents were examined by preparing the following samples for XRD analysis.

(a)

Unused Mg-based adsorbent;

(b)

Solid sample withdrawn after adding 2 g/L of Mg-based adsorbent into deionized water;

(c)

Solid sample withdrawn after adding 2 g/L of Mg-based adsorbent into As(III) solution (C_AS0 = 10 mg/L and pH₀ = 7);

(d)

Solid sample withdrawn after adding 2 g/L of Mg-based adsorbent into As(V) solution (C_AS0 = 10 mg/L and pH₀ = 7);

(e)

Solid sample withdrawn after adding 2 g/L of Mg-based adsorbent into Si solution (C_Si0 = 50 mg/L and pH₀ = 7).

The solutions of As(III), As(V), and Si in (c)–(e) were made using the stock solutions in Section 2.2.1. In (b)–(e), the prepared solutions (250 mL of deionized water and As(III), As(V), or Si solution) were each placed into different TPX beakers. The Mg-based adsorbent (0.5 g) was added into each beaker and stirred with a magnetic stirrer. After shaking for 24 h, the samples were subjected to solid–liquid separation using suction filtration (0.45 μm of filter pore size). The concentrations of Mg, Si and As in the filtrates obtained in (b)–(e) were measured by ICP-AES and ICP-MS. The solid samples collected via solid–liquid separation operation were left overnight in a drying oven (40 °C). The crystalline phases for their solid samples were identified using a powder diffractometer (RINT-2500, Rigaku Co., Akishima, Japan) at the GSJ-Lab (AIST).

2.4. Sample Preparation for Scanning Electron Microscopy—Energy Dispersive X-Ray Spectroscopy (SEM-EDS) Analysis

To investigate the morphological and chemical changes in the Mg-based adsorbents before and after As or Si adsorption in detail, the morphologies of solid samples (a), (c), and (e) prepared for XRD analysis in Section 2.3 were observed using SEM. Additionally, elemental mapping was conducted using EDS. Powdered samples were dried overnight at 40 °C and then were attached on carbon tape. After osmium coating (Neoc-Pro, Meiwafosis Co., Ltd., Shinjuku, Japan), they were analyzed using SEM-EDS (JSM-6610LV, JEOL Ltd., Akishima, Japan) at the GSJ-Lab (AIST). Incidentally, elemental mapping was performed for Mg, O, and As or Si.

3. Results

3.1. As-Removal Tests Containing Silicic Acid

This section presents the results of As-removal tests using synthetic As-contaminated water with and without Si.

3.1.1. Concentration of Residual as in Treated Water and As-Removal Ratio

Figure 1 shows C_AS from the As-removal tests plotted against C_Si0. Figure 1a–d present results for (a) C_AS0 = 1 mg/L and W_Ad0/V = 0.2 g/L, (b) C_AS0 = 1 mg/L and W_Ad0/V = 2 g/L, (c) C_AS0 = 10 mg/L and W_Ad0/V = 0.2 g/L, and (d) C_AS0 = 10 mg/L and W_Ad0/V = 2 g/L, respectively. In each panel, the data are organized by adsorbent type and As valence. The measured C_AS0 values for each test are also provided in the corresponding figures.

As shown in Figure 1, the effect of C_Si0 on C_AS varies with adsorbent type, As valence, and W_Ad0/V. The dependence of C_AS on C_Si0 can be broadly classified into four trends.

(i)

C_AS increases with increasing C_Si0 (e.g., MgO for As(III) and Mg(OH)₂ for As(V) in Figure 1a).

(ii)

C_AS initially decreases and then increases with increasing C_Si0 (e.g., MgO for As(V) in Figure 1a and most cases for MgCO₃ in Figure 1a–d).

(iii)

C_AS increases with increasing C_Si0 up to a certain value and then shows no significant further change (e.g., Mg(OH)₂ for As(III) in Figure 1a,c).

(iv)

C_AS increases or decreases irregularly with increasing C_Si0.

In addition, the relative magnitude of C_AS among adsorbent types varied significantly with C_Si0. Under the present test conditions, at relatively low C_Si0, the overall ordering of C_AS was MgO, Mg(OH)₂ < MgCO₃. Conversely, at relatively high C_Si0, the overall ordering was Mg(OH)₂ < MgO, MgCO₃.

Interpretation of the effects of C_Si0 was facilitated by converting C_AS to the As-removal ratio, R_AS [%], and organizing the figures by the adsorbent type. R_AS was calculated as follows:

R_AS = (C_AS0 − C_AS)/C_AS0 × 100

(1)

Figure 2 shows R_AS obtained from Equation (1) plotted against C_Si0. Figure 2a–c present the results for MgO, Mg(OH)₂, and MgCO₃, respectively.

For MgO, for both As(III) and As(V), R_AS at W_Ad0/V = 2 g/L maintained high values up to C_Si0 ≈ 25 mg/L, but decreased rapidly when C_Si0 exceeded ~25 mg/L. R_AS increased with increasing C_Si0 up to ~25 mg/L under two conditions—C_AS0 = 1 mg/L and W_Ad0/V = 2 g/L for As(III), and C_AS0 = 1 mg/L and W_Ad0/V = 0.2 g/L for As(V)—and then decreased rapidly at C_Si0 > 25 mg/L. Under the other conditions, R_AS generally decreased with increasing C_Si0.

For Mg(OH)₂, the R_AS for As(V) reduced with increasing C_Si0 at C_AS0 = 1 and 10 mg/L with W_Ad0/V = 2 g/L, as well as at C_AS0 = 1 mg/L with W_Ad0/V = 0.2 g/L. For As(III) at W_Ad0/V = 2 g/L, R_AS decreased significantly at C_Si0 ≈ 50 mg/L but increased again at C_Si0 > 100 mg/L. In contrast, at C_AS0 = 10 mg/L and W_Ad0/V = 0.2 g/L, no clear effect of C_Si0 on R_AS was observed for either As(III) or As(V).

For MgCO₃, R_AS was higher at very low C_Si0 (~5 mg/L) than in the absence of dissolved Si, but then decreased rapidly with increasing C_Si0. Overall, at W_Ad0/V = 0.2 g/L, R_AS differed negligibly among As valence states; however, at W_Ad0/V = 2 g/L, the R_AS for As(III) was lower than that for As(V).

3.1.2. pH of Treated Water

In Figure 3, the final pH (pH_f) in the As-removal tests is plotted against C_Si0. The results for MgO, Mg(OH)₂, and MgCO₃ are shown in Figure 3a–c, respectively.

In all cases, pH_f decreased with increasing C_Si0. For MgO and Mg(OH)₂, regardless of As valence and C_AS0, the pH_f at W_Ad0/V = 2 g/L was higher than that at W_Ad0/V = 0.2 g/L. At W_Ad0/V = 2 g/L, pH_f showed no significant dependence on As valence or C_AS0; however, at W_Ad0/V = 0.2 g/L, the pH_f of As(V) was higher than that of As(III). For MgCO₃, the pH_f of As(III) was higher than that of As(V). In addition, for As(III) with MgCO₃, the pH_f at C_AS0 = 1 mg/L was higher than that at C_AS0 = 10 mg/L, and the pH_f at W_Ad0/V = 2 g/L was higher than that of W_Ad0/V = 0.2 g/L.

3.1.3. Concentration of Leached Mg in Treated Water

The C_Mg (mg/L) values obtained from the As-removal tests are shown in Figure 4. Figure 4a–c present the results for MgO, Mg(OH)₂, and MgCO₃, respectively.

As shown in Figure 4a, for MgO, the relationship between C_Mg and C_Si0 depended on the As valence and W_Ad0/V. Under the test conditions, in the relatively high C_Si0 range, C_Mg was higher at W_Ad0/V = 2 g/L than at 0.2 g/L and was higher for As(V) than for As(III). In contrast, in the relatively low C_Si0 range, C_Mg was, in some cases, higher at W_Ad0/V = 0.2 g/L than at 2 g/L, and the effect of As valence was not clear.

As shown in Figure 4b, for Mg(OH)₂, the relationship between C_Mg and C_Si0 depended on the As valence, W_Ad0/V, and C_AS0. For both As(III) and As(V), C_Mg was higher at W_Ad0/V = 2 g/L than at 0.2 g/L and was slightly higher at C_AS0 = 1 mg/L than at 10 mg/L. In addition, C_Mg was lower for As(III) than for As(V). Overall, C_Mg increased with increasing C_Si0 under most conditions, except at W_Ad0/V = 0.2 g/L for As(III). Under this condition, C_Mg appeared to increase slightly with increasing C_Si0 and then decreased slightly, for both C_AS0 = 1 and 10 mg/L.

For MgCO₃ (Figure 4c), C_Mg increased gradually with increasing C_Si0 under all test conditions. C_Mg was significantly higher at W_Ad0/V = 2 g/L than at 0.2 g/L, whereas no clear difference was observed as a function of As valence or C_AS0.

3.1.4. Concentration of Residual Si in Treated Water

The C_Si (mg/L) values in the As-removal tests with C_Si0 are plotted against C_Si0 in Figure 5. Figure 5a–c show the plots for MgO, Mg(OH)₂, and MgCO₃, respectively, including the 1:1 line (C_Si:C_Si0 = 1:1). Data points below the 1:1 line indicate that a fraction of the dissolved silicic acid species was removed from solution by adsorption and/or precipitation.

For MgO and Mg(OH)₂, C_Si was higher at W_Ad0/V = 0.2 g/L than at 2 g/L, as shown in Figure 5. At W_Ad0/V = 0.2 g/L, the C_Si data generally lay close to the 1:1 line, except in the relatively high C_Si0 range, where the points tended to deviate below the line as C_Si0 increased; the exception was MgO at C_AS0 = 1 mg/L for As(V). For MgO at W_Ad0/V = 0.2 g/L, C_Si at C_AS0 = 1 mg/L for As(V) was nearly zero in the relatively low C_Si0 range and increased rapidly with increasing C_Si0.

At W_Ad0/V = 2 g/L, the relationship between C_Si and C_Si0 varied significantly with adsorbent type. As shown in Figure 5a, for MgO, C_Si was nearly zero in the relatively low C_Si0 range and increased rapidly with increasing C_Si0, similar to the behavior observed at C_AS0 = 1 mg/L and W_Ad0/V = 0.2 g/L for As(V). As shown in Figure 5b, the plot of C_Si at W_Ad0/V = 2 g/L for Mg(OH)₂ appears to increase approximately in proportion to C_Si0, although it remained below the 1:1 line.

For MgCO₃, most data points are below the 1:1 line (Figure 5c). C_Si increased with increasing C_Si0, but the rate of increase in C_Si decreased as C_Si0 increased. The effects of As valence and C_AS0 on C_Si were unclear for MgO and Mg(OH)₂, but appeared to be present for MgCO₃. Specifically, at W_Ad0/V = 0.2 g/L, C_Si was lower at C_AS0 = 1 mg/L than at 10 mg/L, and at W_Ad0/V = 2 g/L, C_Si tended to be lower for As(III) than for As(V).

3.2. XRD Analysis

Figure 6a–e shows the XRD patterns for the original (unused), hydrated, As(III)-adsorbed, As(V)-adsorbed, and Si-adsorbed MgO, respectively. The XRD patterns in Figure 6a–d are similar to those reported in previous studies [20,21].

As reported previously [20,21], the original MgO contained almost no hydroxide phase (Figure 6a), whereas hydroxide formation was clearly observed in the hydrated sample (Figure 6b). Hydroxide formation was also confirmed in the As(III)-adsorbed MgO (Figure 6c), but was limited in the As(V)-adsorbed MgO (Figure 6d). In Figure 6c,d, peaks attributable to crystalline phases such as magnesium arsenite or magnesium arsenate species were not detected. Figure 6e shows that hydroxide formation was extremely low compared with that in the original MgO, and no other crystalline phases were observed.

Figure 7a–e show the XRD patterns for the original (unused), hydrated, As(III)-adsorbed, As(V)-adsorbed, and Si-adsorbed Mg(OH)₂, respectively. The XRD patterns in Figure 7a–d are similar to those reported previously [20,21].

For the original Mg(OH)₂ (Figure 7a), only diffraction peaks attributable to the hydroxide phase were observed, consistent with previous studies [20,21]. In Figure 7b–e, only hydroxide peaks were also observed; no other crystalline products were detected.

The XRD patterns for the original (unused), hydrated, As(III)-adsorbed, As(V)-adsorbed, and Si-adsorbed MgCO₃ are shown in Figure 8a–e, respectively. The data in Figure 8a–d are consistent with those reported previously [20,21].

As described previously [20,21], the crystalline phase of the original MgCO₃ comprised Mg₅(CO₃)₄(OH)₂·4H₂O (Figure 8a). Hydrated MgCO₃ showed minimal change in its diffraction pattern; however, the formation of trace amounts of Mg(OH)₂ was confirmed (Figure 8b). Similarly, a weak peak corresponding to Mg(OH)₂ was detected for As(III)-adsorbed MgCO₃ (Figure 8c). In contrast, no distinct Mg(OH)₂ peak was observed for As(V)-adsorbed or Si-adsorbed MgCO₃ (Figure 8d,e). This trend in the extent of hydroxide formation was consistent with the XRD results from the MgO tests. In Figure 8b–e, no crystalline products were observed other than trace amounts of Mg(OH)₂.

3.3. Morphological Observation on Mg-Based Adsorbents

The SEM images of the original, As(III)-adsorbed, and Si-adsorbed MgO are shown in Figure 9a–c, respectively, revealing cuboid MgO crystals. Numerous fine particles are observed on the surfaces of the cuboid crystals (Figure 9b,c). Based on the XRD results in Section 3.2, these fine particles were considered to be primarily Mg(OH)₂.

The SEM images of the original, As(III)-adsorbed, and Si-adsorbed Mg(OH)₂ are shown in Figure 10a–c, respectively.

As shown in Figure 10a, the original Mg(OH)₂ comprised smaller particles than the original MgO (Figure 9a), with a more spherical than cuboid morphology. The As(III)-adsorbed Mg(OH)₂ and Si-adsorbed Mg(OH)₂ differed in appearance (Figure 10b,c) from the original Mg(OH)₂ (Figure 10a) and appeared to be coagulated rather than aggregated.

The SEM images of the original, As(III)-adsorbed, and Si-adsorbed MgCO₃ are shown in Figure 11a–c, respectively.

The original MgCO₃ had a slightly rounded cuboid morphology (Figure 11a). Both the As(III)-adsorbed and Si-adsorbed MgCO₃ exhibited an appearance similar to that of original MgCO₃ (Figure 11b,c).

3.4. Elemental Mappings (EM) on Mg-Based Adsorbents

The SEM image and elemental mappings (EM) of Mg, O, and As for As(III)-adsorbed MgO are shown in Figure 12a–d, respectively.

In the EDS analysis, As detection and mapping relied on the As Kα line (10.53 keV), because the As Lα line (1.282 keV) overlaps with the Mg Kα line (1.253 keV). The distributions of Mg, O, and As were nearly identical (Figure 12), confirming that As was uniformly distributed on the surfaces of the MgO particles without localized enrichment, although As was difficult to visualize because of its low adsorption amount.

The SEM image and EM of Mg, O, and As for As(III)-adsorbed Mg(OH)₂ are shown in Figure 13a–d, respectively.

Consistent with As(III)-adsorbed MgO, the distributions of Mg, O, and As were nearly identical, confirming uniform As distribution on the surface of Mg(OH)₂ (Figure 13).

The SEM image and EM of Mg, O, and As for As(III)-adsorbed MgCO₃ are shown in Figure 14a–d, respectively.

As shown in Figure 14, consistent with As(III)-adsorbed MgO and Mg(OH)₂, the distributions of Mg, O, and As were nearly identical, confirming uniform As distribution on the MgCO₃ surface.

The SEM image and EM of Mg, O, and Si for Si-adsorbed MgO are shown in Figure 15a–d, respectively.

As shown in Figure 15, the elemental distributions of Mg and O were consistent. In contrast, Si was more concentrated in aggregate-like particles than in cuboid particles (Figure 15d).

The SEM image and EM of Mg, O, and Si for Si-adsorbed Mg(OH)₂ are shown in Figure 16a–d, respectively.

As shown in Figure 16, the elemental distributions of Mg, O, and Si are nearly identical, confirming that Si is uniformly adsorbed on the Mg(OH)₂ surface.

The SEM image and EM maps of Mg, O, and Si for Si-adsorbed MgCO₃ are shown in Figure 17a–d, respectively.

As shown in Figure 17, Mg and O are uniformly distributed, whereas Si appears clustered. Similar to MgO (Figure 15), Si is preferentially distributed in aggregate-like particles rather than in cuboid particles.

4. Discussion

4.1. Verification of Correlation Between pH_f and C_AS, C_Mg, or C_Si

In this section, the correlation between pH_f and the dissolved components in solution is examined.

Plots of C_AS versus pH_f are shown in Figure 18. Figure 18a–d present the results for (a) C_AS0 = 1 mg/L and W_Ad0/V = 0.2 g/L, (b) C_AS0 = 1 mg/L and W_Ad0/V = 2 g/L, (c) C_AS0 = 10 mg/L and W_Ad0/V = 0.2 g/L, and (d) C_AS0 = 10 mg/L and W_Ad0/V = 2 g/L, respectively. Across Figure 18a–d, C_AS varies with adsorbent type, As valence, C_AS0, and W_Ad0/V; however, overall, C_AS tends to increase as pH_f decreases. This trend is more pronounced for MgO and MgCO₃ than for Mg(OH)₂ (Figure 18a–d). As described in Section 3.2 and Section 3.3, the XRD and SEM observations confirmed the presence of Mg(OH)₂ aggregates formed secondarily in hydrated, As-adsorbed, and Si-adsorbed MgO and MgCO₃. As shown in Section 3.4, the EM of As further confirmed that As is distributed uniformly on the adsorbent surface and within these aggregates. Therefore, it is concluded that the As-removal mechanism for MgO and MgCO₃ proceeds not only via adsorption on the adsorbent surface but also via coprecipitation with secondarily formed Mg(OH)₂ aggregates. Accordingly, the stronger increase in C_AS with decreasing pH for MgO and MgCO₃ than for Mg(OH)₂ is attributed to suppression of Mg(OH)₂ aggregate formation at lower pH.

No clear effects of As valence or C_AS0 were observed on C_Mg or C_Si. Therefore, C_Mg and C_Si versus pH_f are plotted only by adsorbent type and W_Ad0/V (Figure 19a,b, respectively). As shown in Figure 19a,b, both C_Mg and C_Si clearly increase with decreasing pH_f. For C_Mg, values at W_Ad0/V = 2 g/L are consistently higher than those at 0.2 g/L, whereas C_Si shows no clear dependence on W_Ad0/V. Therefore, C_Si is governed only by adsorbent type and pH_f and is independent of As valence and W_Ad0/V, whereas C_Mg depends significantly on W_Ad0/V in addition to adsorbent type and pH_f. Under the tested conditions (W_Ad0/V = 0.2 and 2 g/L), the adsorbents partially dissolved, but most of the material remained in the solid phase. Based on this, if C_Mg reflected only the intrinsic solubility of the Mg-based compounds comprising each adsorbent, C_Mg would be expected to be independent of W_Ad0/V. The results of this study suggest that C_Mg is strongly affected by silicic-acid adsorption or the generation of magnesium silicate species on the adsorbents. Conversely, as shown in Figure 19b, C_Si exhibits no clear dependence on W_Ad0/V; therefore, under the conditions tested in this study, C_Si is inferred to be determined only by adsorbent type and pH_f.

4.2. Estimation of Adsorbent Residual Ratio

Leaching of Mg, a base component of the adsorbent, reduces the amount of the adsorbent remaining as a solid phase in solution. Therefore, the residual ratio of the adsorbent was calculated from C_Mg using the following procedure.

The initial Mg amount, W_Mg0 (g), was determined using α_Mg (Table 1) as follows:

W_Mg0 = (α_Mg/100) W_Ad0

(2)

The amount of remaining Mg in the solid phase, W_Mgf (g), 24 h after the added adsorbent was expressed as follows:

W_Mgf = W_Mg0 − (C_Mg/1000) V

(3)

As the unit of C_Mg is mg/L, it was converted to g/L via dividing by 1000. Assuming that the Mg-residual ratio is equal to the adsorbent residual ratio γ (%), the value of γ can be calculated using following equation:

γ = W_Mgf/W_Mg0 × 100

(4)

The γ values obtained from the As-removal tests are plotted against C_Si0 in Figure 20. Figure 20a–c show the results for MgO, Mg(OH)₂, and MgCO₃, respectively.

As shown in Figure 20, for all adsorbents, γ at W_Ad0/V = 2 g/L remained high regardless of the As valence and C_AS0 but gradually decreased with increasing C_Si0. At W_Ad0/V = 0.2 g/L, the γ for As(V) tended to decrease with increasing C_Si0, whereas the γ for As(III) decreased with increasing C_Si0 in the relatively low C_Si0 range but then either showed minimal change or increased slightly with increasing C_Si0.

4.3. As- and Si-Adsorption Amounts per Unit Mass of Adsorbent

Most Mg-based adsorbents partially dissolve in solution. Nevertheless, the amount of adsorption per unit mass of adsorbent has been often calculated based on the initial amount of adsorbent added. In previous studies on Mg-based adsorbents [20,21], the adsorption amount was estimated considering the Mg leaching from the adsorbents itself. Similarly, the amounts of As and Si adsorbed per unit mass of adsorbent (Q_AS and Q_Si, respectively) were calculated using the following procedure.

The concentration of the adsorbent remaining in the solid phase after 24 h, W_Adf/V (g/L), was determined as follows:

W_Adf/V = (γ/100) W_Ad0/V

(5)

Q_AS (mg/g) and Q_Si (mg/g) were calculated using the following equations:

Q_AS = (C_AS0 − C_AS)/(W_Adf/V)

(6)

Q_Si = (C_Si0 − C_Si)/(W_Adf/V)

(7)

Plots of Q_AS against C_Si0 are shown in Figure 21. Figure 21a–d show the results for (a) C_AS0 = 1 mg/L and W_Ad0/V = 0.2 g/L, (b) C_AS0 = 1 mg/L and W_Ad0/V = 2 g/L, (c) C_AS0 = 10 mg/L and W_Ad0/V = 0.2 g/L, and (d) C_AS0 = 10 mg/L and W_Ad0/V = 2 g/L, respectively, organized by adsorbent type and As valence.

In Figure 21a–d (note the different y-axis scales), Q_AS is generally higher at W_Ad0/V = 0.2 g/L than at 2 g/L for the same C_AS0 and is higher at C_AS0 = 10 mg/L than at 1 mg/L for the same W_Ad0/V. These results are consistent with the expectation that the amount adsorbed per unit mass of adsorbent depends primarily on the relative ratio of adsorbent to the adsorbing species (As or Si in this study). Although competitive adsorption between As and Si would be expected to decrease Q_AS with increasing C_Si0, this trend was not universal. For most MgO and MgCO₃ conditions, Q_AS decreased with increasing C_Si0; however, in some cases Q_AS increased with increasing C_Si0. Notably, for MgCO₃, the Q_AS at C_Si0 ≈ 120 mg/L was higher than that at C_Si0 = 0 mg/L for both As(III) and As(V) (Figure 21c). The effects of C_Si0 on the Q_AS of Mg(OH)₂ were smaller than those for MgO and MgCO₃ and were significantly smaller for As(III). As mentioned in Section 3.2, the XRD results for MgO and MgCO₃ confirmed the formation of Mg(OH)₂ (Figure 6 and Figure 8). However, pH_f decreased with increasing C_Si0 (Figure 3), suggesting that the amount of Mg(OH)₂ formed after the addition of MgO and MgCO₃ to the solution decreased with increasing C_Si0. Because As removal by MgO and MgCO₃ is inferred to occur partly via coprecipitation during Mg(OH)₂ formation [20,21], the decrease in Q_AS with increasing C_Si0 is presumably attributable to reduced Mg(OH)₂ production. Although no distinct XRD peaks of magnesium silicate species were detected, the formation of low-crystallinity magnesium silicate phases cannot be excluded. Therefore, particularly at high C_Si0, the apparent As adsorption amount may increase due to the incorporation of As into such low-crystallinity magnesium silicate phases during their formation.

Q_Si as a function of C_Si0 is plotted by adsorbent type, C_AS0, and W_Ad0/V in Figure 22. Figure 22a,b show the results for W_Ad0/V = 0.2 g/L and 2 g/L, respectively, because no clear effect of As valence on Q_Si was observed.

The Mg(OH)₂ data in Figure 22a fit approximately on a single fitted curve regardless of C_AS0, whereas the fitted curves for the MgO and MgCO₃ data differ depending on C_AS0. In contrast, in Figure 22b, the data for each Mg-based adsorbent fit approximately on a single approximation curve regardless of C_AS0. These results suggest that competitive adsorption of As affects Si adsorption at lower W_Ad0/V (0.2 g/L) but is almost negligible at higher W_Ad0/V (2 g/L). The higher Q_Si of MgO and MgCO₃ at C_AS0 = 1 mg/L than at 10 mg/L at W_Ad0/V = 0.2 g/L also supports this inference. As an exception, the Q_Si of MgO in Figure 22b tends to be slightly lower at C_Si0 = 100 mg/L than at 50 mg/L; however, this trend may be related to the sudden drop in pH_f (Figure 3a).

4.4. Langmuir Isotherm Model

The suitability of the data for adsorption of As(III), As(V), and Si in the As-removal experiments conducted in the presence of Si to the Langmuir isotherm model, which is a general adsorption model, was evaluated.

The Langmuir isotherm plots of As(III) adsorption on MgO, Mg(OH)₂, and MgCO₃ are shown in Figure 23a–c, respectively. For convenience, C_Si0 in Figure 23 was grouped into five divisions (0, 5, 25, 50, and 100 mg/L); the corresponding ranges of actual C_Si0 values were 0, 4.25–6.06, 22.8–29.5, 44.9–57.6, and 89.5–111 mg/L, respectively. In addition, the plots in Figure 23 are further separated by W_Ad0/V. However, when both C_Si0 and W_Ad0/V are fixed, the remaining variable, C_AS0, takes only two values (approximately 1 and 10 mg/L), resulting in only two data points for each condition (or fewer if no adsorption occurred). Under these conditions, reliably assessing whether the data follow the Langmuir model is challenging. Therefore, the analysis was also performed using four data points (or fewer if no adsorption occurred), assuming that W_Ad0/V had no effect, and the goodness of fit to the Langmuir model was evaluated using the correlation coefficient r.

The Langmuir equation for As-adsorption is described by Equation (8):

1/Q_AS = 1/(K_L Q_AS-MAX)(1/C_AS) + 1/Q_AS-MAX

(8)

where Q_AS-MAX is the maximum adsorption amount of As [mg/g], and K_L is the Langmuir adsorption coefficient.

Table 3. Values of Q_AS-MAX, K_L, and r for the approximate straight lines obtained by applying Equation (8) to As(III)-adsorption data.

		WAd0/V = 0.2 and 2 g/L				WAd0/V = 0.2 g/L *		WAd0/V = 2 g/L *
CSi0 (mg/L)	Adsorbent	Number of Data	r	QAS-MAX (mg/g)	KL	QAS-MAX (mg/g)	KL	QAS-MAX (mg/g)	KL
0	MgO	4	0.542	3.96	1.715	−92.81 **	−0.017 **	−0.05 *	−6.191 **
5	MgO	4	0.947	5.21	1.286	−72.42 **	−0.021 **	−9.46 *	−0.608 **
25	MgO	4	0.884	2.14	13.711	3.56	0.457	10.99	2.074
50	MgO	4	0.809	1.70	2.762	5.98	0.119	6.91	0.557
100	MgO	4	0.971	8.75	0.051	7.05	0.087	2.53	1.74 × 10−5
0	Mg(OH)2	4	1.000	6.85	0.801	8.35	0.627	5.46	1.025
5	Mg(OH)2	4	0.991	4.04	2.460	5.97	0.606	6.60	1.436
25	Mg(OH)2	4	0.992	4.07	3.428	7.99	0.488	4.67	2.947
50	Mg(OH)2	4	0.991	8.38	0.299	5.94	0.724	4.33	0.598
100	Mg(OH)2	4	0.993	4.27	3.054	10.86	0.345	4.16	3.155
0	MgCO3	3	0.868	0.21	0.245	- ***	- ***	0.09	0.884
5	MgCO3	4	0.753	15.47	0.123	4.96 **	−27.9 **	3.09	0.504
25	MgCO3	4	0.774	3.19	0.251	9.04	0.264	0.90	0.964
50	MgCO3	4	0.667	2.79	0.135	1.86	5.157	1.18	0.225
100	MgCO3	4	0.689	17.01	0.017	25.49	0.054	1.32	0.156

* The correlation coefficient r is 1 for each case, because the number of data for each condition is only two. ** These values do not fit the Langmuir model but are listed as a reference. *** Cannot be calculated as there was only one data plot.

shows the Q_AS-MAX and K_L values obtained by fitting the data at W_Ad0/V = 0.2 and 2 g/L together and separately to the Langmuir model. The r values obtained when the W_Ad0/V = 0.2 and 2 g/L data are fitted together are also listed in Table 3. Incidentally, the r values obtained when the data at W_Ad0/V = 0.2 and 2 g/L are fitted separately are 1 for all cases, because only two data points are available for each condition. As shown by the fitted lines in Figure 23 and the values in Table 3, the As(III)-adsorption behavior for MgO and MgCO₃ is influenced not only by C_Si0 but also by W_Ad0/V. Conversely, for Mg(OH)₂, the As(III)-adsorption behavior is affected by C_Si0 but not by W_Ad0/V. As shown in Table 3, the Q_AS-MAX values for MgO are negative at C_Si0 = 0 and 5 mg/L, which is not physically meaningful.

Considering Q_AS-MAX at C_Si0 = 25–100 mg/L, it increases with increasing C_Si0 at W_Ad0/V = 0.2 g/L but decreases at W_Ad0/V = 2 g/L. For Mg(OH)₂, the effect of C_Si0 on Q_AS-MAX is relatively small, although Q_AS-MAX shows a slight overall decrease. For MgCO₃, As(III) is barely removed at W_Ad0/V = 0.2–2 g/L in the absence of Si, whereas its removal improves in the presence of Si. However, because Q_AS-MAX at 2 g/L tends to be significantly lower than that at 0.2 g/L, increasing W_Ad0/V beyond 2 g/L is unlikely to improve As(III)-removal performance.

Focusing on the value of r for the approximate straight line obtained using all data points at W_Ad0/V = 0.2 and 2 g/L, the As(III)-adsorption behavior on MgO and MgCO₃ was judged not to follow the Langmuir model because most r values are significantly low. Conversely, As(III) adsorption on Mg(OH)₂ appears to follow the Langmuir model, as all r values are high (>0.99) regardless of C_Si0.

The Langmuir isotherm plots of As(V) adsorption on MgO, Mg(OH)₂, and MgCO₃ are shown in Figure 24a–c, respectively. The ranges of the actual C_Si0 values corresponding to these divisions were 0, 4.99–6.72, 25.7–33.1, 49.1–64.7, and 100–120 mg/L, respectively. In addition,

Table 4. Values of Q_AS-MAX, K_L, and r for the approximate straight lines obtained by applying Equation (8) to the As(V)-adsorption data.

		WAd0/V = 0.2 and 2 g/L				WAd0/V = 0.2 g/L *		WAd0/V = 2 g/L *
CSi0 (mg/L)	Adsorbent	Number of Data	r	QAS-MAX (mg/g)	KL	QAS-MAX (mg/g)	KL	QAS-MAX (mg/g)	KL
0	MgO	4	0.994	6.76	11.0	12.6	0.780	16.0	4.45
5	MgO	4	0.998	7.00	9.56	6.88	4.37	15.5	4.13
25	MgO	4	0.998	7.30	11.8	4.74	1303	19.7	4.17
50	MgO	4	0.835	1.23	2.53	0.86	3.35	15.4	0.141
100	MgO	4	0.682	17.69	0.00807	2.44	0.253	20.7	4.20 × 10−3
0	Mg(OH)2	4	1.000	6.30	28.664	6.99	15.9	6.17	29.3
5	Mg(OH)2	4	0.998	5.11	33.540	7.21	2.94	5.55	30.7
25	Mg(OH)2	4	0.998	4.46	15.900	6.73	1.94	4.29	16.6
50	Mg(OH)2	4	0.999	3.70	4.627	4.71	1.77	3.60	4.79
100	Mg(OH)2	4	0.996	5.52	0.449	7.45	0.356	3.32	0.784
0	MgCO3	4	0.989	20.3	0.143	11.7	0.504	4.89	0.639
5	MgCO3	4	0.991	9.65	0.636	8.96	1.62	4.03	1.63
25	MgCO3	4	0.898	−6.29 **	−0.111 **	7.94	0.405	6.11	0.109
50	MgCO3	4	0.876	−8.07 **	−0.0509 **	21.1	0.0487	3.12	0.119
100	MgCO3	4	0.758	−5.37 **	−0.0424 **	−18.3 **	−0.0388 **	2.84	0.0596

* The correlation coefficient r is 1 for each, because the number of data for each condition is only two. ** These values do not fit the Langmuir model but are listed as a reference.

shows the Q_AS-MAX and K_L values obtained by fitting the data at W_Ad0/V = 0.2 and 2 g/L, either jointly or separately, to the Langmuir model.

The r values of the fitted lines obtained by applying the data at W_Ad0/V = 0.2–2 g/L were high for Mg(OH)₂ across all C_Si0 values, and were also high for MgO and MgCO₃ in the lower C_Si0 range (Table 4). Focusing on the Q_AS-MAX values obtained from the four data points at W_Ad0/V = 0.2–2 g/L, MgO showed a minimum at C_Si0 = 50 mg/L, followed by a sharp increase at C_Si0 = 100 mg/L. For Mg(OH)₂, Q_AS-MAX appeared largely insensitive to C_Si0. In contrast, for MgCO₃, Q_AS-MAX was negative at C_Si0 = 25–100 mg/L (Table 4), indicating that these data do not fit to the Langmuir model.

Similarly to the Langmuir equation for As adsorption (Equation (8)), the Langmuir equation for Si adsorption is described by Equation (9)

1/Q_Si = 1/(K_L Q_Si-MAX)(1/C_Si) + 1/Q_Si-MAX

(9)

where Q_Si-MAX is the maximum adsorption amount of Si [mg/g], and K_L is the Langmuir adsorption coefficient, as in Equation (8).

The Langmuir isotherms for Si adsorption on MgO, Mg(OH)₂, and MgCO₃ are shown in Figure 25a–c, respectively.

Unlike Figure 23 and Figure 24, the plots in Figure 25 are grouped by combinations of As valence, C_AS0, and W_Ad0/V. Each grouped data set generally comprises four data points obtained at different C_Si0 values, excluding C_Si0 = 0 mg/L. However, for Mg(OH)₂, some data sets include fewer than four points because Si adsorption did not occur under certain conditions (C_AS0 = 1 mg/L and W_Ad0/V = 0.2 g/L, for both As(III) and As(V)).

Table 5. Values of Q_Si-MAX, K_L, and r for the approximate straight lines obtained by applying Equation (9) to the Si-adsorption data.

Adsorbent	As Valent	CAS0 (mg/L)	WAd0/V (g/L)	Number of Data	r	QSi-MAX (mg/g)	KL
MgO	As(III)	1	0.2	4	0.994	149	0.00817
MgO	As(III)	1	2.0	4	0.996	27.3	0.446
MgO	As(III)	10	0.2	4	0.439	17.0	0.254
MgO	As(III)	10	2.0	4	0.990	23.5	0.374
MgO	As(V)	1	0.2	4	0.759	275	0.0647
MgO	As(V)	1	2.0	4	0.998	15.7	1438
MgO	As(V)	10	0.2	4	0.992	−68.2 **	−9.56 × 10−3 **
MgO	As(V)	10	2.0	4	0.384	11.1	1.10
Mg(OH)2	As(III)	1	0.2	2 *	1.000	−12.8 **	−7.61 × 10−3 **
Mg(OH)2	As(III)	1	2.0	4	0.987	9.80	0.762
Mg(OH)2	As(III)	10	0.2	4	0.637	12.5	0.143
Mg(OH)2	As(III)	10	2.0	4	0.989	95.5	0.0613
Mg(OH)2	As(V)	1	0.2	3 *	0.045	27.8 **	−329 **
Mg(OH)2	As(V)	1	2.0	4	0.980	10.1	0.526
Mg(OH)2	As(V)	10	0.2	4	0.994	207	1.38 × 10−3
Mg(OH)2	As(V)	10	2.0	4	0.992	10.2	0.0559
MgCO3	As(III)	1	0.2	4	0.993	357	0.0404
MgCO3	As(III)	1	2.0	4	0.998	122	0.0133
MgCO3	As(III)	10	0.2	4	0.997	−189 **	−0.0104 **
MgCO3	As(III)	10	2.0	4	0.982	20.5	0.0988
MgCO3	As(V)	1	0.2	4	0.999	1097	9.89 × 10−3
MgCO3	As(V)	1	2.0	4	0.977	16.2	0.0807
MgCO3	As(V)	10	0.2	4	0.999	−33.9 **	−0.0133 **
MgCO3	As(V)	10	2.0	4	0.988	−5.49 **	−0.0295 **

* The number of data is smaller than 4, because there were cases where Si adsorption did not occur. ** These values do not fit the Langmuir model but are listed as a reference.

lists Q_Si-MAX, K_L, and r obtained by fitting each data set to the Langmuir model.

As shown in Table 5, MgCO₃, exhibited r values > 0.97 across all data sets; however, the values of Q_Si-MAX and K_L were negative in multiple cases. Therefore, the Si adsorption on MgCO₃ in the presence of As is not adequately described by the Langmuir model. For MgO and Mg(OH)₂ at W_Ad0/V = 0.2 g/L, the r values were generally low and the Q_Si-MAX and/or K_L values were negative, again indicating poor agreement with the Langmuir model. By contrast, for both MgO and Mg(OH)₂ at W_Ad0/V = 2 g/L, r exceeded 0.98 and both Q_Si-MAX and K_L were positive, suggesting that the Langmuir model provides an adequate fit under these conditions.

4.5. Dissolved Forms of Arsenous Acid and Arsenic Acid in Solution

This section considers As speciation in liquid phase. The forms of dissolved As(III) are described by the following equations for dissociation reactions of arsenous acid

H₃AsO₃ ⇌ H₂AsO₃⁻ + H⁺

(10)

H₂AsO₃⁻ ⇌ HAsO₃²⁻ + H⁺

(11)

HAsO₃²⁻ ⇌ AsO₃³⁻ + H⁺

(12)

where pKa₁ = 9.1, pKa₂ = 12.1, and pKa₃ = 13.4 (25 °C) [22]. In the synthetic contaminated water before adsorbent addition in the As(III) tests, the dominant species is presumed to be H₃AsO₃ because the maximum pH₀ was 7.62, which is well below pK_a1. In the treated water from the As(III) tests, AsO₃³⁻ is expected to be negligible because pH_f << pKa₃. The pKa₁ and pKa₂ values are used in Equations (13) and (14), corresponding to Equations (10) and (11), respectively.

[H₂AsO₃⁻]/[H₃AsO₃] = 10 exp (pH − pKa₁)

(13)

[HAsO₃²⁻]/[H₂AsO₃⁻] = 10 exp (pH − pKa₂)

(14)

Using the pH_f values of 9.06–11.025 in the As(III) tests, (H₂AsO₃⁻)/(H₃AsO₃) and (HAsO₃²⁻)/(H₂AsO₃⁻) were calculated as 0.902–84.1 and 0.001–0.084, respectively. These results indicate that the predominant forms of As(III) dissolving in the treated water were H₃AsO₃ and H₂AsO₃⁻.

Subsequently, the forms of As(V) dissolved in water are described by the following equations for dissociation reactions of arsenic acid

H₃AsO₄ ⇌ H₂AsO₄⁻ + H⁺

(15)

H₂AsO₄⁻ ⇌ HAsO₄²⁻ + H⁺

(16)

HAsO₄²⁻ ⇌ AsO₄³⁻ + H⁺

(17)

where pK_a1 = 2.24, pK_a2 = 6.96, and pK_a3 = 11.5 (25 °C) [23], and the amount of each arsenic acid species dissolved in water is estimated by the following reactions:

[H₂AsO₄⁻]/[H₃AsO₄] = 10 exp (pH − pKa₁)

(18)

[HAsO₄²⁻]/[H₂AsO₄⁻] = 10 exp (pH − pKa₂)

(19)

[AsO₄³⁻]/[HAsO₄²⁻] = 10 exp (pH − pKa₃)

(20)

For the synthetic contaminated water before adding the adsorbent in the tests with As(V), the abundances of H₃AsO₄ and AsO₄³⁻ were presumed to be extremely small and therefore negligible because pKa₁ << pH₀ << pKa₃. Substituting the pH₀ values (6.70–7.50) yields [HAsO₄²⁻]/[H₂AsO₄⁻] values of 0.55–3.47. Therefore, the dominant As(V) species dissolved in the synthetic contaminated water were presumed to be H₂AsO₄⁻ and HAsO₄²⁻. For the treated water, the abundances of H₃AsO₄ and H₂AsO₄⁻ were presumed to be extremely small and therefore negligible because pKa₂ << pH₀. Substituting pH_f affords [AsO₄³⁻]/[HAsO₄²⁻] values of 0.014–0.349 for MgO, 0.005–0.104 for Mg(OH)₂, and 0.010–0.175 for MgCO₃. Therefore, the dominant As(V) species dissolved in the treated water were estimated to be HAsO₄²⁻ and AsO₄³⁻. However, pH_f tended to decrease as C_Si0 increased; accordingly, the fraction of AsO₄³⁻ decreased with increasing C_Si0.

4.6. Dissolved Forms of Silicic Acid in Solution

Consistent with a previous study [19], the forms of silicic acid dissolved in solution were evaluated based on the following equations for dissociation reactions of silicic acid:

H₄SiO₄ ⇌ H₃SiO₄⁻ + H⁺

(21)

H₃SiO₄⁻ ⇌ H₂SiO₄²⁻ + H⁺

(22)

Denoting the equilibrium constants of the dissociation formulas of silicic acid (Equations (21) and (22)) as Ka₁ and Ka₂, respectively, affords the following equations

[H₃SiO₄⁻]/[H₄SiO₄] = 10 exp (pH − pKa₁)

(23)

[H₂SiO₄²⁻]/[H₃SiO₄⁻] = 10 exp (pH − pKa₂)

(24)

where the units of the applicable molecular formula are mol/L, pKa₁ = 9.86, and pKa₂ = 13.1 (25 °C) [22].

The main form of Si dissolved in the synthetic contaminated water before adding the adsorbent was presumed to be H₄SiO₄ because pH₀ << pK_a1. The pH_f values obtained in this study ranged from 9.06 to 11.04. Substituting these pH_f values into Equation (23) yields 0.157–15.2. Additionally, substituting the pH_f values into Equation (24) yields values < 0.009, indicating that the dominant dissolved forms were H₃SiO₄⁻ (a monovalent ion) and H₄SiO₄ (a non-valent molecule), whereas H₂SiO₄²⁻ did not exist.

4.7. Dissolution Reactions of Mg-Based Adsorbents and Adsorption Reaction of Silicic Acid

The dissolution mechanism of MgO is considered to proceed via both direct dissolution (Equation (25)) and a pathway involving initial hydration to transform into Mg(OH)₂ followed by dissolution (Equations (26) and (27)).

MgO + H₂O ⇌ Mg²⁺ + 2OH⁻

(25)

MgO + H₂O ⇌ Mg(OH)₂

(26)

Mg(OH)₂ ⇌ Mg²⁺ + 2OH⁻

(27)

Reaction (27) is also the dissolution reaction for Mg(OH)₂.

The dissolution reaction of MgCO₃ is simply expressed by Equation (26); however, the carbonate species released differ depending on the solution pH.

MgCO₃ + 2H₂O ⇌ Mg²⁺ + H₂CO₃ + 2OH⁻

(28)

H₂CO₃ ⇌ H⁺ + HCO₃⁻

(29)

HCO₃⁻ ⇌ H⁺ + CO₃²⁻

(30)

When the equilibrium constants for the dissociation reactions of carbonic acid (Equations (27) and (28)) are denoted as Ka₁ and Ka₂, respectively, the following equations are applicable

[HCO₃⁻]/[H₂CO₃] = 10 exp (pH − pKa₁)

(31)

[CO₃²⁻]/[HCO₃⁻] = 10 exp (pH − pKa₂)

(32)

where the units of the applicable molecular formula are mol/L, pKa₁ = 6.35, and pKa₂ = 10.33 (25 °C) [22].

The main forms of carbonic acid dissolved in the synthetic contaminated water before adding the adsorbent were presumed to be H₂CO₃ and HCO₃⁻ because pH₀ ≈ pK_a1 << pK_a2. However, in this study, the initial amount of carbonic acid was not considered in the calculations of carbonic acid species concentrations because the amount of dissolved carbonic acid in the synthetic contaminated water before added adsorbent was unknown. The abundance of H₂CO₃ in the treated water can be considered negligible because pH_f << pKa₃. The pH_f values obtained in this study ranged from 9.06 to 11.04. Substituting these pH_f values into Equation (32) yields 0.151–3.36. Therefore, the dominant forms of carbonic acid dissolved in the treated water were presumed to be HCO₃⁻ and CO₃²⁻.

As described above, the form of silicic acid varies with solution pH. Below, adsorption of silicic acid on Mg-based adsorbents is considered using H₄SiO₄ as an example.

The adsorption mechanism of silicic acid species on MgO is also considered to proceed via both direct adsorption (Equation (33)) and a pathway involving initial hydration to form solid-Mg-OH followed by adsorption (Equations (34) and (35)).

Solid-Mg-O-Mg-Solid + Si(OH)₄ ⇌ Solid-Mg-O-Si (OH)₃ + HO-Mg-Solid

(33)

Solid-Mg-O-Mg-Solid + H₂O ⇌ Solid-Mg-OH + HO-Mg-Solid

(34)

Solid-Mg-OH + Si(OH)₄ ⇌ Solid-Mg-O-Si(OH)₃ + H₂O

(35)

Equation (31) is also the adsorption reaction of silicic acid on Mg(OH)₂.

The adsorption mechanism of silicic acid species on MgCO₃ is also considered to proceed via both direct adsorption (Equation (36)) and a pathway involving initial hydration to form solid-Mg-OH followed by adsorption (Equation (35) via Equation (37)).

Solid-Mg-O-C(O)OH + Si(OH)₄ ⇌ Solid-Mg-O-Si (OH)₃ + H₂CO₃

(36)

Solid-Mg-O-C(O)OH + H₂O ⇌ Solid-Mg-OH + H₂CO₃

(37)

In addition, the adsorption mechanisms of arsenous and arsenic acid species on Mg-based adsorbents can be expressed by replacing H₄SiO₄ in the reaction equations in this section. This implies that silicic acid species competitively adsorb with arsenous and arsenic acid species.

4.8. Mass Balance for OH⁻

Solution pH reflects the OH⁻ mass balance of the reaction system. As explained in Section 4.7, during dissolution of MgO, Mg(OH)₂, and MgCO₃, two OH⁻ ions are released for each Mg ion released (Equations (25), (27), and (28)). However, for MgCO₃, the acid dissociation reaction of H₂CO₃ (i.e., OH⁻ consumption) must also be considered. Specifically, one OH⁻ is consumed to convert H₂CO₃ to HCO₃⁻, and two OH⁻ are consumed to convert H₂CO₃ to CO₃²⁻. In addition, when silicic acid is present, OH⁻ is consumed during its ionization. As mentioned in Section 4.6, H₄SiO₄ predominates in the neutral pH solution before adsorbent addition, whereas substantial H₃SiO₄⁻ is produced under the alkaline conditions after adding Mg-based adsorbents. OH⁻ is not released during adsorption of silicic acid onto Mg-based adsorbents, as described in Section 4.7. If, instead, adsorption of H₃SiO₄⁻ (rather than H₄SiO₄) was assumed, OH⁻ would be released; however, the net OH⁻ balance across the overall system remains zero because OH⁻ is consumed during the preceding ionization of H₄SiO₄. In contrast, during adsorption of silicic acid onto MgCO₃, OH⁻ is consumed via dissociation of the H₂CO₃ produced. In principle, OH⁻ is also consumed during ionization of arsenous acid and arsenic acid; however, their concentrations were relatively low in this study and can be ignored in this section to simplify the following calculations.

The values of C_Mg, C_Si0, and C_Si (converted to mM) are denoted as [Mg²⁺], [Si]₀, and [Si]_f, respectively. The OH⁻ concentrations (mM) calculated from pH₀ and pH_f are represented as [OH⁻]₀ and [OH⁻]_f, respectively. The concentrations (mM) of H₄SiO₄, H₃SiO₄⁻, HCO₃⁻, and CO₃²⁻ in the treated water are denoted as [H₄SiO₄], [H₃SiO₄⁻], [HCO₃⁻], and [CO₃²⁻], respectively.

Because silicic acid in the treated water exists as H₄SiO₄ and H₃SiO₄⁻, as mentioned in Section 4.6, the following equation holds:

[Si]₀–[Si]_f = [H₄SiO₄] + [H₃SiO₄⁻]

(38)

In addition, because [H₂CO₃] in the treated water can be neglected, as mentioned in Section 4.7, the following equation holds for MgCO₃ based on Equations (26) and (34):

[Mg²⁺] + [Si]₀–[Si]_f = [HCO₃⁻] + [CO₃²⁻]

(39)

Furthermore, using the chemical equilibrium constants given in Section 4.6 and Section 4.7, [H₄SiO₄], [H₃SiO₄⁻], [HCO₃⁻], and [CO₃²⁻] can be determined.

In summary, the OH⁻ mass balance is as follows:

The mass balance of OH⁻ for both MgO and Mg(OH)₂:

2[Mg²⁺] = [OH⁻]_f − [OH]₀ + [H₃SiO₄⁻]

(40)

The mass balance of OH⁻ for MgCO₃:

2[Mg²⁺] = [OH⁻]_f − [OH]₀ + [H₃SiO₄⁻] + [HCO₃⁻] + 2[CO₃²⁻]

(41)

The corresponding plots for Equations (40) and (41) are shown in Figure 26. Figure 26a–c correspond to MgO, Mg(OH)₂, and MgCO₃, respectively.

A 1:1 line is also shown in each panel. If Equations (40) and (41) are valid, the data should fall on the 1:1 line. In all panels, most points lie close to the 1:1 line; however, some plots deviate from these lines.

As shown in Figure 26a,b, most points—except those for As(III) at W_Ad0/V = 0.2 g/L—fall below the 1:1 line. In practice, OH⁻ in solution is also consumed by ionization of arsenite and arsenate and by dissolution of atmospheric CO₂ into the solution during pH measurement; however, these factors were not included in the calculations. Some points for As(III) at W_Ad0/V = 0.2 g/L lie above the 1:1 line, but the reason for this deviation is unclear.

As shown in Figure 26c, for MgCO₃ the points increasingly shift upward from the 1:1 line with increasing [Mg²⁺]. This behavior reflects that the MgCO₃ adsorbent used in this study is basic magnesium carbonate (magnesium carbonate hydroxide). Accordingly, adsorption of silicic acid onto the MgCO₃ adsorbent includes the reaction in Equation (35) in addition to Equation (36). However, because the amount of H₂CO₃ generated during Si adsorption onto the MgCO₃ adsorbent was calculated from the total decrease in aqueous Si, the generated H₂CO₃ was likely overestimated. This overestimation would cause [HCO₃⁻] and [CO₃²⁻] in Equation (41) to be higher than their actual values.

4.9. Suitability Evaluation in the Presence of Si

As described in Section 3.4, EM analysis of Si on Si-adsorbed adsorbents confirmed that Si was uniformly distributed on the Mg(OH)₂ surface, whereas for MgO and MgCO₃, Si was concentrated on aggregates rather than uniformly distributed on the adsorbent surface. For any Mg-based adsorbents, in the presence of Si, the amount of As adsorbed decreases due to competitive adsorption with Si. In addition, based on the above observations, for MgO and MgCO₃ the R_AS likely decreased further because Si co-precipitated instead of As, which would otherwise have been removed by coprecipitation with the secondarily generated Mg(OH)₂ aggregates.

The SiO₂ concentration in groundwater is usually <45 mg/L [24], with a typical range of 1–30 mg/L and averaging ~17 mg/L [25]. Converting these SiO₂ values to Si, the corresponding Si concentration is <21 mg/L, with a range of 0.4–14 mg/L and a typical value of ~8 mg/L. Accordingly, assuming that C_Si0 in groundwater is <21 mg/L, the suitability of each Mg-based adsorbent in the presence of silicic acid was evaluated.

As shown in Figure 2a, for MgO, when W_Ad0/V is sufficiently high (e.g., 2 g/L) and C_AS0 is 1–10 mg/L, high R_AS can be maintained for both As(III) and As(V) even in the presence of Si. For Mg(OH)₂, for both As(III) and As(V), R_AS at W_Ad0/V = 2 g/L does not decrease at C_AS0 = 1 mg/L but clearly decreases at C_AS0 = 10 mg/L (Figure 2b). However, when C_Si0 was ~25 mg/L, R_AS was ~10% at W_Ad0/V = 0.2 g/L but increased to ~80% at 2 g/L. Therefore, if W_Ad0/V is increased beyond 2 g/L, R_AS is expected to improve further. In contrast, MgCO₃ is unsuitable for treating As-contaminated water when C_Si0 substantially exceeds 5 mg/L because its As-removal performance is strongly affected by Si (Figure 2c). Nevertheless, because the R_AS for MgCO₃ improves at C_Si0 ~5 mg/L, it may be feasible to propose a treatment strategy in which MgCO₃ is used as a primary treatment for As-contaminated water with extremely low C_Si0, followed by secondary treatment using adsorbents that are more efficient (and more expensive) than MgCO₃.

As mentioned in the introduction, Waltham and Eick have reported on the effects of Si on the adsorption behavior of As(V) and As(III) on goethite (FeO(OH)) [14]. Their test conditions were C_AS0 = 1.34 mg/L (0.1 mM), W_Ad0/V = 1 g/L, and pH_f = 4, 6, and 8. The lowest As(III) adsorption inhibition ratio was 4% at C_Si0 = 3.56 mg/L (0.1 mM) in pH 6, and the highest ratio was 40% at C_Si0 = 35.6 mg/L (1.0 mM) in pH 8. For As(V), they reported there was no difference in the total amount of As adsorbed between C_Si0 = 3.56 and 35.6 mg/L. The test conditions most similar to theirs in this study are C_Si0 = 1 mg/L and W_Ad0/V = 2 g/L. The effects of C_Si0 on Q_AS under these conditions are shown in Figure 21b. For MgO and Mg(OH)₂, the Q_AS for both As(III) and As(V) hardly changed in the C_Si0 range of 0–50 mg/L, indicating no clear inhibition of As adsorption by Si. For MgCO₃, the Q_AS for both As(III) and As(V) decreased significantly with increasing C_Si0. Therefore, compared to Fe-based adsorbents (goethite), MgO and Mg(OH)₂ may be superior in that the Q_AS do not decrease even in the presence of Si, not only for As(V) but also for As(III), in the C_Si0 range of 0–50 mg/L.

Overall, MgO and Mg(OH)₂ appear to be suitable adsorbents for As removal within the typical C_Si0 range of natural groundwater. However, special care is required for alkaline groundwater or groundwater mixed with geothermal water, because the C_Si0 can be higher than in typical groundwater. The results of this study suggest that a major contributor to the decreased As-removal performance of Mg-based adsorbents in the presence of Si is the reduction in solution pH caused by OH⁻ consumption during the ionization of silicic acid. Therefore, achieving high R_AS with favorable cost performance, and improving sustainability, requires setting W_Ad0/V appropriately by considering not only C_AS0 but also C_Si0 in As-contaminated groundwater.

4.10. Limitations of This Study and Future Issues

In this study, various evaluations were based on the results obtained from tests using synthetic contaminated water. However, actual groundwater and industrial wastewater must contain not only Si but also multiple ions and organic substances that could compete for adsorption with As. Therefore, it will be necessary to conduct tests using actual contaminated groundwater and factory wastewater, and then compare the results with those in this study. The reaction time in this study was set to 24 h, the same as in many common studies on adsorption reactions [16,17,20,21], but it has not been confirmed whether a strict equilibrium state was reached. Therefore, in the future, to conduct kinetic studies, including confirming the equilibrium time, it is essential to perform adsorption tests with reaction time as a parameter. In this study, the matrix components leach from the adsorbents, making it difficult to control the ionic strength to a constant value. Therefore, tests with ionic strength as a parameter were not conducted. However, a previous study [14] suggests that the effects of ionic strength on adsorption behavior are not significant. pH control using pH buffering solution was not performed in this study. The reason for this is that there were concerns that the components contained in the pH buffer solution would affect the adsorption behavior of As and Si. Particularly, phosphoric acid contained in neutral phosphate solutions, which are often used as pH buffer solutions, is known to compete with As for adsorption [18]. Therefore, to conduct adsorption tests with controlled pH, it is necessary to select a pH buffer solution that does not contain competing adsorption components or to investigate the competitive adsorption effects of components contained in the pH buffer solution in advance. In addition to conducting adsorption tests with reaction time and pH as parameters, using chemical equilibrium calculation tools to consider the pH effects on the solubility of Mg compounds and to verify the chemical species formed could be an effective means of constructing advanced adsorption models.

5. Conclusions

Adsorption is a promising approach for purifying As-contaminated water; however, natural groundwater may contain Si, which has been known to reduce the As-removal performance of adsorbents. The effects of Si on the As-removal performance of adsorbents have been studied in detail for Fe-based adsorbents but have been scarcely investigated for Mg-based adsorbents. Therefore, in this study, we focused on MgO, Mg(OH)₂, and MgCO₃ and conducted As-removal tests using W_Ad0/V, As valence, C_AS0, and C_Si0 as test parameters to assess the effects of Si on As-removal performance. It was revealed that the effect of C_Si0 on As removal depends strongly on adsorbent type, As valence, and W_Ad0/V. At W_Ad0/V = 2 g/L, MgO maintained high R_AS values up to C_Si0 ≈ 25 mg/L but decreased rapidly when C_Si0 exceeded ~25 mg/L. For Mg(OH)₂, the R_AS for As(V) reduced with increasing C_Si0, whereas for As(III) it decreased significantly at C_Si0 ≈ 50 mg/L but increased at C_Si0 > 100 mg/L. For MgCO₃, R_AS was higher at very low C_Si0 (~5 mg/L) than under Si-free conditions, but then decreased rapidly with increasing C_Si0. Under all test conditions, pH_f decreased with increasing C_Si0. In particular, for MgO and MgCO₃, the impact of decreasing pH was greater than for Mg(OH)₂, because lower pH inhibits the formation of secondary Mg(OH)₂ aggregates and reduces As removal by coprecipitation with these aggregates. As the Si concentration in typical groundwater is reported to be <21 mg/L [24], when the Si concentration in As-contaminated water to be treated approaches this upper limit, MgCO₃ is not recommended for As removal. By contrast, MgO and Mg(OH)₂ should be able to achieve high R_AS, provided that W_Ad0/V is set appropriately.

Functional adsorbents are often assumed to be recycled because they cannot be readily discarded given their relatively high cost and/or the presence of valuable metals [12]. Consequently, conventional adsorbents that are readily available, inexpensive, and easier to handle are likely to be used to treat As-containing wastewater generated during their recycling. A potential option for spent adsorbents produced during secondary-wastewater treatment is to repurpose them as building materials (e.g., concrete); however, it is reasonable to anticipate psychological resistance to the use of arsenic-containing waste in such applications. Therefore, an important consideration is that adsorbents used to treat this secondary wastewater should be suitable for As treatment when landfill disposal is assumed. The Mg-based adsorbents evaluated in this study meet these requirements.