Substituent-dependent selective extraction of environmentally relevant metal ions by benzothioates: a combined experimental and DFT study

Abstract

This study presents a systematic investigation into the substituent-dependent extraction behavior of benzothioate derivatives, focusing on chloro-(BzT-Cl), methyl-(BzT-Me), and methoxy-(BzT-MeO) analogues. Competitive liquid–liquid ion-pair extraction experiments were conducted using six environmentally relevant metal ions (Zn²⁺, Mn²⁺, Cr³⁺, Fe³⁺, Co²⁺, and Cu²⁺), establishing a comprehensive structure–activity relationship (SAR) supported by both experimental data and density functional theory (DFT) calculations. The results indicate a pronounced preference for Fe³⁺ across all ligand systems; notably, BzT-Cl achieved an extraction efficiency of 87.74% and a selectivity factor (S_f) exceeding 1400 relative to Zn²⁺. DFT analysis (B3LYP/def2-TZVP) suggests that this selectivity arises from the interplay between the high charge density of Fe³⁺ and the substituent-induced electronic substituent-dependent electronic properties of the ligand framework. The performance of BzT-Cl correlates with its higher chemical hardness (η = 2.315 eV) and wider HOMO–LUMO energy gap (ΔE = 4.630 eV), whereas the broader extraction behavior of BzT-MeO toward divalent ions such as Mn²⁺ is associated with its increased nucleophilicity (ε > 2.9 eV). Extraction experiments in reservoir, river, and tap water further revealed pronounced matrix-dependent variations in metal removal efficiency, highlighting the influence of sample composition on ligand performance. Overall, the combined experimental and computational results demonstrate that electronic modification of the benzothioate framework can modulate metal-ion extraction and selectivity, providing a basis for the further development of substituent-engineered extraction systems.

Tailoring benzothioate subtituents enables precise metal recovery; chloro-induced electronic hardening drives Fe³⁺ selectivity, while methoxy-enhanced nucleophilicity ensures broad-spectrum remediation in real aquatic matrices.

1. Introduction

Organic ligands containing heteroatoms such as oxygen, sulfur, nitrogen have been widely investigated due to their strong coordination capabilities and their importance in separation science, environmental remediation, analytical chemistry and supramolecular host–guest systems.^1–5 The pioneering work of Pedersen established crown ethers as effective macrocyclic hosts for alkali and alkaline earth metal ions, and subsequent studies extended this concept to sulfur-containing analogues, including thia-crown ethers and benzothio derivatives.^6–11 These sulfur-based systems exhibit enhanced affinity toward transition and heavy metal ions due to the strong donor properties of sulfur atoms.^12,13

In recent years, benzothioate derivatives have attracted increasing attention because of the presence of a polarized thioester functional group (–C(O)–S–), which acts as a soft donor center and is highly compatible with hard–soft acid–base (HSAB) principles in selective metal binding.^8,14,15 This property makes benzothioate ligands particularly promising for applications involving selective metal ion recognition and separation.

The extraction of metal ions using thioester and thioether-containing ligands has become an important research area, especially in wastewater treatment, ion pair extraction, hydrometallurgy, and sensing technologies.^16–18 Among these approaches, liquid–liquid ion-pair extraction is a powerful technique for evaluating metal–ligand interactions, as it enables direct quantification of metal transfer between aqueous and organic phases. This method provides key thermodynamic and distribution parameters, including extraction constants (K_ext), distribution ratios (K_D), Gibbs free energy changes (–ΔG_ext), and selectivity factors (S_f), which collectively allow for a detailed understanding of extraction mechanisms and structure–property relationships.^15,19

Substituent effects play a critical role in determining the extraction behavior of benzothioate ligands. Electron-withdrawing substituents such as chlorine (–Cl) can alter the electron density distribution of the aromatic ring and the thioester moiety, often enhancing metal ion affinity. Conversely, electron-donating substituents such as methyl (–CH₃) and methoxy (–OCH₃) influence both electronic and steric properties, leading to distinct extraction behaviors depending on the target metal ion. Despite extensive studies on crown ethers and thia-crown systems, systematic investigations focusing on substituent-dependent extraction behavior in benzothioate ligands remain limited.

The synthesis and metal-ion extraction properties of chloro-substituted (BzT-Cl)^20,21 and methyl-substituted (BzT-Me)^14,22 benzothioate derivatives have been previously studied individually, including in our own work, whereas methoxy-substituted derivatives (BzT-MeO) have also been studied in terms of their electronic structures, including HOMO–LUMO characteristics.^14,23 These earlier studies established the extraction behaviour or electronic characteristics of individual ligand series, but the resulting datasets were evaluated separately and not compiled into a unified substituent-dependent comparison. In particular, a direct comparative assessment of the BzT-Cl, BzT-Me and BzT-MeO series under a common analytical framework, encompassing extraction efficiency, distribution behaviour, metal-ion selectivity, electronic descriptors and performance in real water matrices has not been reported.

Accordingly, the objective of the present study was not to re-establish the extraction behavior of the previously reported ligand systems individually but to integrate and comparatively evaluate the three substituent classes within a common experimental and theoretical framework. Chloro-(BzT-Cl), methyl-(BzT-Me), and methoxy-substituted (BzT-MeO) benzothioates were comparatively assessed toward Zn²⁺, Mn²⁺, Cr³⁺, Fe³⁺, Co²⁺, and Cu²⁺ under identical competitive liquid–liquid ion-pair extraction conditions. Extraction efficiencies (Ext. %), distribution ratios (K_D), extraction constants (K_ext), Gibbs free energy changes (–ΔG_ext), and selectivity factors (S_f) were evaluated together to identify substituent-dependent trends. These experimental observations were further considered alongside DFT-derived frontier molecular orbital energies and global reactivity descriptors to examine possible relationships between ligand electronic properties and extraction behaviour. In addition, extraction experiments using reservoir, river, and tap water were included to evaluate how the ligand systems behave in more complex aqueous matrices. Thus, the principal contribution of the present work is the integrated comparison of substituent effects across the three benzothioate families rather than the isolated characterization of an individual ligand series.

2. Experimental

2.1. Materials and reagents

All chemicals used in this study were of analytical grade or higher purity. Thiobenzoic acid derivatives, substituted anilines, thionyl chloride, acetonitrile, dichloromethane (DCM), chloroform, sodium hydroxide, and hydrochloric acid were supplied by Merck, Sigma-Aldrich and Carlo Erba. High-purity, commercially available nitrate salts—chromium(iii) nitrate nonahydrate [Cr(NO₃)₃·9H₂O, ≥99%, Sigma-Aldrich], cobalt(ii) nitrate hexahydrate [Co(NO₃)₂·6H₂O, ≥98%, Merck], copper(ii) nitrate trihydrate [Cu(NO₃)₂·3H₂O, ≥99%, Sigma-Aldrich], manganese(ii) nitrate tetrahydrate [Mn(NO₃)₂·4H₂O, ≥98%, Merck], iron(iii) nitrate nonahydrate [Fe(NO₃)₃·9H₂O, ≥98%, Sigma-Aldrich], and zinc(ii) nitrate hexahydrate [Zn(NO₃)₂·6H₂O, ≥99%, Merck]—were used as received without further purification as metal ion sources. All aqueous solutions were prepared using deionized water (18.2 MΩ). Natural water samples (reservoir, river, and city tap water) were filtered prior to extraction and were used without acidification or pH adjustment.

2.2. Instrumentation

All microwave-assisted synthesis reactions were performed using a Milestone START SYNTH Microwave Synthesis Labstation oven equipped with a control terminal. Metal ion concentrations before and after extraction were determined using ICP-OES and ICP-MS instruments with the following operational specifications: ICP-OES was performed using a PerkinElmer Optima 2100 DV instrument, and ICP-MS measurements were carried out using an Thermo Scientific iCAPQ system. FTIR spectra were recorded with a PerkinElmer Spectrum Two spectrometer. The accuracy and precision of all measurements were confirmed using standard reference solutions, and all determinations were performed in triplicate.

2.3. Synthesis of ligands

Three families of benzothioate ligands were synthesized: chloro-substituted benzothioates (BzT-Cl), methyl-substituted benzothioates (BzT-Me), and methoxy-substituted benzothioates (BzT-MeO). All ligands were prepared following the same general esterification and thiocarbonyl activation strategy. Ortho-, meta-, and para-chlorobenzene-substituted benzothioate ligands were synthesized as described in ref. 20. Methyl-substituted benzothioates and methoxy-substituted benzothioates were synthesized following the methods reported in ref. 14. Reaction was monitored by TLC and purified products were obtained through column chromatography. The resulting compounds were characterized by NMR, FTIR, and elemental analysis as described previously.^14,20

2.4. LUMO–HOMO levels

Geometry optimizations and electronic structure calculations were performed within the framework of density functional theory (DFT). All quantum chemical calculations of the synthesized compounds were performed using ORCA (version 6.1.0, macOS ARM64 OpenMPI 4.1.1 build)²⁴ on a macOS Sequoia (15.7.4) platform with 16 GB RAM. Molecular structures were built and visualized using Avogadro 2 (version 1.103.0, Qt version 6.10.1), an open-source molecular editor, was obtained from https://avogadro.cc.²⁵ All quantum chemical calculations were carried out within the framework of density functional theory (DFT) using the B3LYP functional combined with the def2-TZVP basis set.^26,27 Dispersion interactions were accounted for using Grimme's D3 correction with Becke–Johnson damping.^28–30 The RIJCOSX approximation along with the corresponding auxiliary basis set was employed to improve computational efficiency.³¹ All quantum chemical calculations were performed in the gas phase (vacuum conditions).³² Frontier molecular orbital energies, including E_HOMO and E_LUMO, were obtained, and the corresponding global reactivity descriptors such as HOMO–LUMO energy gap (ΔE), chemical hardness (η), global softness (σ), electronegativity (χ), chemical potential (µ), electrophilicity index (ω), nucleophilicity index (ε), ionization energy (I), and electron affinity (A) were calculated for each molecule using well-established procedures reported in the literature.^33–35 For all compounds, the most stable structures corresponding to the lowest energy minima were considered in the analysis.^36,37 The calculated results, together with the optimized geometries and minimum-energy values.

2.5. Competitive liquid–liquid ion-pair extraction

2.5.1. Liquid–liquid ion-pair extraction

Liquid–liquid extraction experiments were carried out using a classical biphasic system consisting of an organic phase in which the ligand was dissolved in chloroform and an aqueous phase containing a mixed-metal solution of Zn²⁺, Mn²⁺, Cr³⁺, Fe³⁺, Co²⁺ and Cu²⁺ ions (counter anion ion: NO₃⁻).^21–23 For each experiment, 10 mL aliquots of 1.0 × 10⁻⁴ M metal solutions were transferred into polypropylene falcon tubes, after which an equal volume of the ligand-containing organic phase (1.0 × 10⁻⁴ M) was added to maintain a constant chloroform/water ratio of 50 : 50 (v/v). All extractions were performed at 25 °C and under natural pH conditions.

Following preparation, the biphasic mixtures were placed on an orbital shaker and agitated at 300 rpm for 2 hours to allow sufficient contact time between the two phases and promote equilibrium complexation. After shaking, the tubes were left undisturbed for 1 hour to enable complete phase separation. The upper aqueous layer was then carefully withdrawn using an automatic micropipette to prevent cross-phase contamination. Metal concentrations in the recovered aqueous phase were quantified using ICP-OES, and extraction efficiencies were calculated based on the difference between initial and final concentrations. A metal-free blank containing the ligand solution was also analyzed to account for potential background contributions originating from the ligand solution. All experiments were conducted in triplicate to ensure reproducibility and reliability of the extraction data across all chlorinated, methylated and methoxylated benzothioate ligand systems.

2.5.2. Formulation

The general ligand–metal ion-pair extraction equilibrium (eqn (1)) used in this study follows the classical thiacrown ether–metal extraction model:^{8,15,16,20,38}

1

A schematic diagram of the ion-pair extraction mechanism is given in Fig. 1.

Fig. 1. A schematic diagram of the ion-pair extraction mechanism.

The extraction percentage (Ext. %), distribution ratio (K_D), extraction constant (K_ext), Gibbs free energy of extraction (ΔG_ext), and selectivity factor (S_f) were calculated using eqn (2)–(6). The extraction percentage (Ext. %) was calculated with eqn (2).

2

The distribution ratio (K_D), defined as the ratio of the extracted metal-containing species in the organic phase to the residual metal ion in the aqueous phase at equilibrium, was calculated using eqn (3).

K_D = [MLA_n]_(aq)/[Mⁿ⁺]_(aq) 3

The extraction constant (K_ext) calculated with eqn (4). In this expression, [Mⁿ⁺]_(aq) represents the equilibrium concentration of the metal ion in the aqueous phase, [A⁻]_(aq) represents the equilibrium concentration of the counter anion (NO₃⁻), n corresponds to the charge of the metal ion (n = 2 for the divalent ions and n = 3 for the trivalent ions), [L]_(aq) represents the equilibrium concentration of the free ligand in the organic phase, and [MLA_n]_(aq) represents the extracted ion-pair species in the organic phase. The term [A⁻]_(aq) was therefore raised to the corresponding ionic-charge power n in the calculation of K_ext.

4

The Gibbs free energy parameter was calculated with eqn (5).

−ΔG_ext = RT ln(K_ext) 5

Selectivity factors (S_f) were calculated using eqn (6).

S_f = K_D,M1/K_D,M2 6

A higher S_f value indicates a stronger extraction preference of the ligand for first metal ion (M₁) than second metal ion (M₂), meaning that the ligand forms a more stable or more efficiently extractable complex with M₁, resulting in its preferential transfer into the organic phase compared to M₂.¹⁵

In these equations the abbreviations symbolize the following: [Mⁿ⁺]_(aq) metal ion at equilibrium, n ion charge, [A⁻]_(aq) anion ion at equilibrium, [L]_(aq) the ligand present in the aqueous phase at equilibrium, [MLA_n]_(aq) complex formed in the organic phase in equilibrium.

These equations were consistently applied across all ligand families (Cl–, Me–, and BzT-MeO), allowing direct comparison.

2.6. Extraction from real water samples

Real water samples (reservoir: water-1, river: water-2, tap water: water-3) were filtered and directly subjected to extraction without pH modification. Each 9 ligand (Cl–, Me–, and BzT-MeO) solution in chloroform (1.0 × 10⁻⁴ M) was contacted with the natural water sample as same volume (10 mL) in a 50 mL plastic falcon tube. All extractions were performed at 25 °C and under natural pH conditions. Following preparation, the biphasic mixtures were placed on an orbital shaker and agitated at 300 rpm for 2 hours to allow sufficient contact time between the two phases and promote equilibrium complexation. After shaking, the tubes were left undisturbed for 1 hour to enable complete phase separation. The upper aqueous layer was then carefully withdrawn using an automatic micropipette to prevent cross-phase contamination. Metal concentrations in the recovered aqueous phase were quantified using ICP-MS, and extraction efficiencies were calculated based on the difference between initial and final concentrations. All experiments were conducted in triplicate to ensure reproducibility and reliability of the extraction data across all chlorinated, methylated and methoxylated benzothioate ligand systems. The purpose of these experiments was specifically to evaluate changes in the concentrations of the target metal ions in the investigated water matrices. Accordingly, pH, conductivity, total organic carbon, and major-ion composition were not determined as part of the present study.

2.7. ICP-OES and ICP-MS analysis

ICP-OES analysis parameters: PerkinElmer Optima 2100 DV. Plasma aerosol type: wet, source equiliration delay (s): 5, nebuliser start-up: instant, plasma (L min⁻¹): 15, plasma conditions: same for all elements, Aux (L min⁻¹): 0.2, nebuliser (L min⁻¹): 0.80, power: 1300 W, view dist.: 15.0, plasma view: axial, sample flow rate (mL min⁻¹): 1.50. Standards: 1–2.5–5–10 ppm. Dalgaboyları (nm): Fe 238.204, Co 228.616, Cu 327.393, Zn 206.200, Cr 267.716, Mn 257.610. Correlation coefficients (R²): Cu 0.999438, Co 0.998807, Fe 0.999424, Zn 0.999776, Cr 0.997066, Mn 0.998447.

ICP-MS analysis parameters: Thermo Scientific iCAP Q. Isotope: 52Cr, 55Mn, 57Fe, 59Co, 63Cu, 66Zn. Mode: KED, Dwell time (s): 0.01, no. of channels: 1, spacing: 0.01, resolution: normal, number of sweeps: 10, time per sweep (s): 0.07, time per main run (s): 0.7, order of modes: KED. Correlation coefficients (R²), equation, LoD, and BEC: Cr: f(x) = 552.4862x + 33 408.0296 R² = 0.9249 BEC = 60.469 ppb LoD = 0.6192 ppb. Mn: f(x) = 1747.4782x + 4460.8086 R² = 0.9994 BEC = 2.553 ppb LoD = 0.3961 ppb. Fe: f(x) = 31.0680x + 4994.3462 R² = 0.9377 BEC = 160.755 ppb LoD = 30.9173 ppb. Co: f(x) = 947.5329x + 493.3449 R² = 0.9976 BEC = 0.521 ppb LoD = 0.2693 ppb. Cu: f(x) = 597.1653x + 4607.5279 R² = 0.9951 BEC = 7.716 ppb LoD = 1.0829 ppb. Zn: f(x) = 313.9315x + 16 457.5466 R² = 0.9851 BEC = 52.424 ppb LoD = 6.0111 ppb.

It should be noted that the calibration coefficients obtained for Cr (R² = 0.9249) and Fe (R² = 0.9377) were lower than those obtained for the other investigated elements. Therefore, the quantitative ICP-MS results for Cr and Fe in the real-water samples should be interpreted with appropriate caution.

3. Results and discussion

The extraction performances of the Cl–, Me– and MeO–substituted benzothioate ligands were systematically examined toward six metal ions (Zn²⁺, Mn²⁺, Cr³⁺, Fe³⁺, Co²⁺ and Cu²⁺). The data obtained from Cl–, Me– and MeO–substituted benzothioate ligands were harmonized under the same interpretation framework, enabling a direct comparison of substituent effects on extraction efficiency, selectivity factors, and thermodynamic parameters.

BzT-Cl-1, BzT-Cl-2, and BzT-Cl-3 ligands served as the electron-withdrawing substituent reference group. BzT-Me-1, BzT-Me-2, and BzT-Me-3 ligands represent weak electron-donating substituent effects. BzT-MeO-1, BzT-MeO-2, and BzT-MeO-3 ligands have methoxy groups, which provide both electron-donating inductive (+I) and resonance (+M) effects, offering a more nucleophilic thiocarbonyl environment than Me- or Cl-substituents.

3.1. Extraction efficiencies of Cl–, Me– and MeO–substituted benzothioates

The data obtained within the scope of this study are given in Tables 1–3. To facilitate interpretation of the extraction data, the values of K_ext, K_D, Ext. %, and obtained for the BzT-Cl, BzT-Me, and BzT-MeO ligand series toward the mixed-metal system (Zn²⁺, Mn²⁺, Cr³⁺, Fe³⁺, Co²⁺, Cu²⁺) are summarized in Table 1. The numerical distribution of these parameters clearly reflects substituent-dependent extraction trends and provides a quantitative basis for comparing the extraction performance of each ligand family. Consistent with the thermodynamic parameters presented in the table, the Cl–, Me–, and MeO–substituted benzothioates exhibit markedly different extraction profiles, each shaped by the electronic nature of the substituent attached to the aromatic ring.

Table 1. Extraction constants (K_ext), distribution ratios (K_D), extraction efficiencies (Ext. %), and calculated extraction free-energy parameters (−ΔG_ext) for the competitive liquid–liquid extraction of Zn²⁺, Mn²⁺, Cr³⁺, Fe³⁺, Co²⁺, and Cu²⁺ with the BzT-Cl, BzT-Me, and BzT-MeO ligand series.

Ligand	Ion	K ext	K D	Ext. %	ln(Kext)	(−)ΔGext
BzT-Cl-1	Co2+	4.15 × 1010	1.98 × 10−2	1.94	2.44 × 101	1.45 × 104
	Cr3+	3.66 × 1015	1.11 × 10−1	10.02	3.58 × 101	2.12 × 104
	Cu2+	1.45 × 1012	3.12 × 10−1	23.81	2.80 × 101	1.66 × 104
	Fe3+	8.18 × 1020	7.16	87.74	4.82 × 101	2.85 × 104
	Mn2+	1.27 × 1012	1.92 × 10−1	16.09	2.79 × 101	1.65 × 104
	Zn2+	6.70 × 109	4.91 × 10−3	0.49	2.26 × 101	1.34 × 104
BzT-Cl-2	Co2+	9.53 × 109	4.75 × 10−3	0.47	2.30 × 101	1.36 × 104
	Cr3+	1.33 × 1015	5.08 × 10−2	4.83	3.48 × 101	2.06 × 104
	Cu2+	1.19 × 1012	2.77 × 10−1	21.66	2.78 × 101	1.65 × 104
	Fe3+	5.68 × 1018	2.34	70.04	4.32 × 101	2.56 × 104
	Mn2+	1.06 × 1012	1.68 × 10−1	14.41	2.77 × 101	1.64 × 104
	Zn2+	5.23 × 1010	3.49 × 10−2	3.37	2.47 × 101	1.46 × 104
BzT-Cl-3	Co2+	1.09 × 1010	5.43 × 10−3	0.54	2.31 × 101	1.37 × 104
	Cr3+	1.45 × 1015	5.43 × 10−2	5.15	3.49 × 101	2.07 × 104
	Cu2+	1.30 × 1012	2.93 × 10−1	22.64	2.79 × 101	1.65 × 104
	Fe3+	6.06 × 1018	2.38	70.38	4.32 × 101	2.56 × 104
	Mn2+	1.23 × 1012	1.87 × 10−1	15.75	2.78 × 101	1.65 × 104
	Zn2+	8.09 × 1010	5.13 × 10−2	4.88	2.51 × 101	1.49 × 104
BzT-Me-1	Co2+	2.05 × 1011	8.18 × 10−2	7.56	2.60 × 101	1.54 × 104
	Cr3+	4.81 × 1015	1.35 × 10−1	11.88	3.61 × 101	2.14 × 104
	Cu2+	1.61 × 1012	3.31 × 10−1	24.86	2.81 × 101	1.66 × 104
	Fe3+	2.98 × 1019	3.50	77.77	4.48 × 101	2.66 × 104
	Mn2+	2.09 × 1012	2.67 × 10−1	21.10	2.84 × 101	1.68 × 104
	Zn2+	6.53 × 1011	2.41 × 10−1	19.41	2.72 × 101	1.61 × 104
BzT-Me-2	Co2+	1.50 × 1011	6.30 × 10−2	5.92	2.57 × 101	1.52 × 104
	Cr3+	6.37 × 1015	1.62 × 10−1	13.97	3.64 × 101	2.15 × 104
	Cu2+	1.50 × 1012	3.18 × 10−1	24.14	2.80 × 101	1.66 × 104
	Fe3+	3.15 × 1019	3.54	77.99	4.49 × 101	2.66 × 104
	Mn2+	1.93 × 1012	2.54 × 10−1	20.24	2.83 × 101	1.68 × 104
	Zn2+	3.26 × 1011	1.53 × 10−1	13.27	2.65 × 101	1.57 × 104
BzT-Me-3	Co2+	8.73 × 1010	3.93 × 10−2	3.78	2.52 × 101	1.49 × 104
	Cr3+	4.69 × 1015	1.32 × 10−1	11.70	3.61 × 101	2.14 × 104
	Cu2+	1.53 × 1012	3.22 × 10−1	24.35	2.81 × 101	1.66 × 104
	Fe3+	3.24 × 1019	3.57	78.11	4.49 × 101	2.66 × 104
	Mn2+	1.56 × 1012	2.20 × 10−1	18.03	2.81 × 101	1.66 × 104
	Zn2+	2.82 × 1011	1.38 × 10−1	12.14	2.64 × 101	1.56 × 104
BzT-MeO-1	Co2+	5.00 × 1011	1.61 × 10−1	13.89	2.69 × 101	1.60 × 104
	Cr3+	1.04 × 1016	2.19 × 10−1	17.99	3.69 × 101	2.18 × 104
	Cu2+	1.92 × 1012	3.66 × 10−1	26.79	2.83 × 101	1.67 × 104
	Fe3+	4.78 × 1017	1.17	53.87	4.07 × 101	2.41 × 104
	Mn2+	3.47 × 1012	3.65 × 10−1	26.72	2.89 × 101	1.71 × 104
	Zn2+	2.26 × 1012	4.72 × 10−1	32.07	2.84 × 101	1.68 × 104
BzT-MeO-2	Co2+	3.60 × 1011	1.27 × 10−1	11.27	2.66 × 101	1.58 × 104
	Cr3+	6.89 × 1015	1.71 × 10−1	14.58	3.65 × 101	2.16 × 104
	Cu2+	1.77 × 1012	3.50 × 10−1	25.94	2.82 × 101	1.67 × 104
	Fe3+	3.37 × 1017	1.05	51.11	4.04 × 101	2.39 × 104
	Mn2+	3.00 × 1012	3.35 × 10−1	25.07	2.87 × 101	1.70 × 104
	Zn2+	1.74 × 1012	4.15 × 10−1	29.31	2.82 × 101	1.67 × 104
BzT-MeO-3	Co2+	3.01 × 1011	1.11 × 10−1	9.99	2.64 × 101	1.57 × 104
	Cr3+	8.36 × 1015	1.92 × 10−1	16.14	3.67 × 101	2.17 × 104
	Cu2+	1.43 × 1012	3.09 × 10−1	23.63	2.80 × 101	1.66 × 104
	Fe3+	3.63 × 1019	3.66	78.56	4.50 × 101	2.67 × 104
	Mn2+	2.00 × 1012	2.60 × 10−1	20.60	2.83 × 101	1.68 × 104
	Zn2+	7.63 × 1011	2.65 × 10−1	20.92	2.74 × 101	1.62 × 104

Table 2. Relative selectivity factors (S_f) obtained from the liquid–liquid extraction of the mixture of Zn²⁺, Mn²⁺, Cr³⁺, Fe³⁺, Co²⁺, Cu²⁺ ions with the compound series BzT-Cl, BzT-Me, and BzT-MeO.

Ligand	Ion	Co2+	Cr3+	Cu2+	Fe3+	Mn2+	Zn2+
BzT-Cl-1	Co2+	—	0.18	0.06	0.00	0.10	4.03
	Cr3+	5.63	—	0.36	0.02	0.58	22.68
	Cu2+	15.79	2.81	—	0.04	1.63	63.61
	Fe3+	361.71	64.28	22.91	—	37.34	1457.59
	Mn2+	9.69	1.72	0.61	0.03	—	39.04
	Zn2+	0.25	0.04	0.02	0.00	0.03	—
BzT-Cl-2	Co2+	—	0.09	0.02	0.00	0.03	0.14
	Cr3+	10.70	—	0.18	0.02	0.30	1.46
	Cu2+	58.24	5.44	—	0.12	1.64	7.93
	Fe3+	492.30	46.03	8.45	—	13.88	67.00
	Mn2+	35.46	3.31	0.61	0.07	—	4.83
	Zn2+	7.35	0.69	0.13	0.01	0.21	—
BzT-Cl-3	Co2+	—	0.10	0.02	0.00	0.03	0.11
	Cr3+	10.00	—	0.19	0.02	0.29	1.06
	Cu2+	53.89	5.39	—	0.12	1.57	5.71
	Fe3+	437.58	43.77	8.12	—	12.71	46.36
	Mn2+	34.42	3.44	0.64	0.08	—	3.65
	Zn2+	9.44	0.94	0.18	0.02	0.27	—
BzT-Me-1	Co2+	—	0.61	0.25	0.02	0.31	0.34
	Cr3+	1.65	—	0.41	0.04	0.50	0.56
	Cu2+	4.05	2.46	—	0.09	1.24	1.37
	Fe3+	42.77	25.96	10.57	—	13.08	14.52
	Mn2+	3.27	1.98	0.81	0.08	—	1.11
	Zn2+	2.95	1.79	0.73	0.07	0.90	—
BzT-Me-2	Co2+	—	0.39	0.20	0.02	0.25	0.41
	Cr3+	2.58	—	0.51	0.05	0.64	1.06
	Cu2+	5.05	1.96	—	0.09	1.25	2.08
	Fe3+	56.27	21.83	11.13	—	13.97	23.16
	Mn2+	4.03	1.56	0.80	0.07	—	1.66
	Zn2+	2.43	0.94	0.48	0.04	0.60	—
BzT-Me-3	Co2+	—	0.30	0.12	0.01	0.18	0.28
	Cr3+	3.37	—	0.41	0.04	0.60	0.96
	Cu2+	8.19	2.43	—	0.09	1.46	2.33
	Fe3+	90.82	26.94	11.09	—	16.22	25.82
	Mn2+	5.60	1.66	0.68	0.06	—	1.59
	Zn2+	3.52	1.04	0.43	0.04	0.63	—
BzT-MeO-1	Co2+	—	0.74	0.44	0.14	0.44	0.34
	Cr3+	1.36	—	0.60	0.19	0.60	0.46
	Cu2+	2.27	1.67	—	0.31	1.00	0.77
	Fe3+	7.24	5.32	3.19	—	3.20	2.47
	Mn2+	2.26	1.66	1.00	0.31	—	0.77
	Zn2+	2.93	2.15	1.29	0.40	1.29	—
BzT-MeO-2	Co2+	—	0.74	0.36	0.12	0.38	0.31
	Cr3+	1.34	—	0.49	0.16	0.51	0.41
	Cu2+	2.76	2.05	—	0.33	1.05	0.84
	Fe3+	8.23	6.13	2.99	—	3.13	2.52
	Mn2+	2.63	1.96	0.96	0.32	—	0.81
	Zn2+	3.26	2.43	1.18	0.40	1.24	—
BzT-MeO-3	Co2+	—	0.58	0.36	0.03	0.43	0.42
	Cr3+	1.73	—	0.62	0.05	0.74	0.73
	Cu2+	2.79	1.61	—	0.08	1.19	1.17
	Fe3+	33.00	19.04	11.84	—	14.12	13.85
	Mn2+	2.34	1.35	0.84	0.07	—	0.98
	Zn2+	2.38	1.37	0.85	0.07	1.02	—

Table 3. Metal removal efficiency (Ext. %) across different water matrices^a.

Ligand no	Water matrixa	Metal removal efficiency (Ext. %)
		Cr3+	Mn2+	Fe3+	Co2+	Cu2+	Zn2+
BzT-Cl-1	W1	—	1.2	32.8	43.0	—	27.3
	W2	8.0	1.8	16.6	25.4	6.5	12.4
	W3	2.7	93.2	13.7	20.1	17.6	13.0
BzT-Cl-2	W1	—	2.1	27.4	31.0	—	97.1
	W2	13.3	0.2	64.1	83.3	0.4	96.7
	W3	2.1	89.7	19.3	1.9	5.0	1.2
BzT-Cl-3	W1	—	3.3	10.5	17.3	—	99.7
	W2	9.1	2.8	3.6	18.1	1.4	17.2
	W3	5.9	78.7	7.1	5.1	1.8	7.4
BzT-Me-1	W1	—	2.9	16.4	17.9	—	99.7
	W2	4.5	5.1	31.4	21.0	3.7	0.9
	W3	4.2	66.6	13.4	4.7	3.5	17.2
BzT-Me-2	W1	—	20.9	62.6	67.0	—	99.5
	W2	4.7	5.8	29.1	1.4	3.3	31.9
	W3	2.8	81.2	15.9	8.4	5.3	8.8
BzT-Me-3	W1	—	4.8	17.0	23.1	—	25.3
	W2	2.4	5.3	5.5	15.0	5.2	51.0
	W3	2.4	53.5	29.3	4.5	4.4	36.8
BzT-MeO-1	W1	—	3.2	5.4	23.1	—	99.9
	W2	2.3	6.5	13.2	2.3	1.8	31.3
	W3	3.6	48.6	11.2	13.0	3.3	6.9
BzT-MeO-2	W1	—	3.0	56.4	52.4	—	99.6
	W2	3.2	2.9	30.0	7.7	3.2	98.2
	W3	2.0	79.7	9.6	14.7	8.8	58.4
BzT-MeO-3	W1	—	4.6	47.4	49.3	—	99.6
	W2	4.9	5.7	34.4	10.4	3.6	72.1
	W3	2.4	82.5	2.9	7.2	5.3	15.7

^aWater samples: reservoir (water-1, W1), river (water-2, W2), and tap water (water-3, W3). “–” indicates that the removal efficiency was not calculated because the initial metal concentration was below the limit of detection (LoD).

According to the hard and soft acids and bases (HSAB) concept, Fe³⁺ and Cr³⁺ are generally classified as hard Lewis acids, whereas Co²⁺, Cu²⁺, and Zn²⁺ exhibit intermediate or borderline character. The classification of Mn²⁺ is less straightforward and may depend on the coordination environment; therefore, its extraction behavior is discussed here primarily on the basis of the experimental results rather than assigning it to a strict HSAB category.^39,40

Cl–substituted ligands exhibited extraction efficiencies following the order: Fe³⁺ > Cu²⁺ > Mn²⁺ > Cr³⁺ > Zn²⁺ > Co²⁺. Fe³⁺ extraction reached 87.74%, showing strong preference due to electrostatic attraction and favorable hydration/desolvation balance. Notably, Zn²⁺ extraction was relatively low for BzT-Cl, suggesting that the electron-withdrawing character of the –Cl substituent may contribute to the observed differences in extraction behavior. The data confirm that BzT-Cl ligands generate the most selective yet narrow extraction profile—excellent for Fe³⁺, but moderate to weak for divalent cations. These results indicate the potential of the BzT-Cl series for selective Fe³⁺ extraction in multi-ion systems.

Me–substituted ligands displayed extraction in the order: Fe³⁺ > Cu²⁺ > Mn²⁺ > Zn²⁺ > Cr³⁺ > Co²⁺. Fe³⁺ extraction remained high (≈78%), but was lower than that observed for the most efficient BzT-Cl derivative, indicating that methyl substitution modifies the extraction behavior while preserving the preference for Fe³⁺. Methyl substitution slightly increases electron density but lacks resonance contribution; therefore, changes in extraction performance are modest. Overall, the BzT-Me ligands exhibited a relatively balanced but less selective extraction profile than the most Fe³⁺-selective BzT-Cl systems.

MeO-substituted ligands displayed a distinctive extraction profile: Fe³⁺ > Mn²⁺ > Zn²⁺ > Cu²⁺ > Cr³⁺ > Co²⁺. Fe³⁺ extraction varied from 51.11 to 78.56% among the BzT-MeO derivatives, while appreciable extraction of several divalent ions, including Mn²⁺ and Zn²⁺, was also observed. This broader extraction behavior may be associated with the electron-donating character of the methoxy substituent and the resulting changes in the electronic properties of the benzothioate framework. However, the observed extraction behavior cannot be attributed solely to the electronic effect of the methoxy group, since metal–ligand interactions, ion-pair formation, solvation/desolvation, and phase partitioning may all contribute to the overall extraction process.

All ligand families displayed the highest extraction affinity toward Fe³⁺, consistent with its strong Lewis acidity and high charge density. Favorable interactions between Fe³⁺ and the benzothioate framework may contribute to this preference. However, in the present biphasic extraction system, the observed selectivity is likely governed by a combination of metal–ligand interactions, ion-pair formation, solvation/desolvation, and phase partitioning rather than by HSAB considerations alone.^41–43

For divalent metals, the extraction trends showed clear substituent dependence. The MeO–substituted ligands, bearing strong electron-donating resonance (+M), displayed significantly higher extraction toward Zn²⁺ and Mn²⁺ compared to Me– and Cl–analogues. This trend may be associated with the electron-donating character of the methoxy substituent and the resulting changes in the electronic properties of the benzothioate framework, which may influence its interactions with divalent metal ions.

As shown in Table 1, the extraction constants (K_ext), distribution ratios (K_D), extraction percentages (Ext. %), and calculated extraction free-energy parameters (−ΔG_ext) vary considerably among the investigated metal ions. Within the BzT-Cl series, Fe³⁺ exhibited the most pronounced extraction behavior, reaching an extraction efficiency of 87.74% and a K_D value of 7.16 for BzT-Cl-1. The corresponding conditional K_ext value was 8.18 × 10²⁰, substantially higher than those obtained for the other metal ions with this ligand. The calculated free-energy parameter was also consistent with thermodynamically favorable extraction of Fe³⁺ under the adopted equilibrium model. In contrast, Zn²⁺ and Co²⁺ generally showed much lower extraction efficiencies within the BzT-Cl series, contributing to the pronounced selectivity toward Fe³⁺. These differences indicate that the chloro-substituted benzothioates discriminate strongly among the competing metal ions under the investigated extraction conditions.

In contrast, the BzT-Me ligands produced a more balanced extraction profile. Fe³⁺ remained the most strongly extracted ion, with conditional K_ext values ranging from 2.98 × 10¹⁹ to 3.24 × 10¹⁹, K_D values of 3.50–3.57, and extraction efficiencies of 77.77–78.11%. The corresponding calculated extraction free-energy parameters were also consistent with favorable Fe³⁺ extraction. Compared with the BzT-Cl series, appreciable extraction of Zn²⁺, Mn²⁺, and Cu²⁺ was observed, although the extent of this difference varied among the individual ligands. These results suggest that methyl substitution modifies the distribution of extraction efficiencies among the competing metal ions while maintaining a pronounced preference for Fe³⁺.

The BzT-MeO series exhibited a comparatively broad extraction profile, with substantial variations in the calculated extraction parameters among both the metal ions and the individual ligands. For Fe³⁺, the conditional K_ext values ranged from 3.37 × 10¹⁷ to 3.63 × 10¹⁹, while the extraction efficiencies varied from 51.11 to 78.56%. Cr³⁺ also exhibited relatively high conditional K_ext values (6.89 × 10¹⁵ to 1.04 × 10¹⁶), although its extraction efficiencies remained considerably lower (14.58–17.99%). In addition, appreciable extraction of the divalent ions was observed, particularly for Zn²⁺ and Mn²⁺, with maximum extraction efficiencies of 32.07% and 26.72%, respectively. The broader extraction behavior of the BzT-MeO series may be associated with the electron-donating character of the methoxy substituent and the resulting modification of the electronic properties of the benzothioate framework. Nevertheless, the differences in conditional K_ext values should be interpreted within the adopted ion-pair extraction model and do not, by themselves, demonstrate a fundamentally different extraction mechanism for the BzT-MeO derivatives.

Collectively, these findings reveal a clear substituent-dependent variation in extraction behavior: the BzT-Cl series showed the most pronounced Fe³⁺ selectivity, the BzT-Me series retained a strong preference for Fe³⁺ while displaying a more balanced extraction profile, and the BzT-MeO series exhibited a broader distribution of extraction efficiencies among the investigated metal ions. These differences indicate that modification of the aromatic substituent influences both metal-ion discrimination and the overall extraction profile of the benzothioate framework.

It should be noted that the K_ext values reported here are conditional extraction constants derived from the adopted ion-pair equilibrium model and the corresponding equilibrium concentration terms. Consequently, their relatively large magnitudes should not be interpreted as intrinsic metal–ligand formation constants or as direct evidence of kinetically irreversible extraction. Because K_ext depends on the equilibrium concentrations and their stoichiometric powers, large values may coexist with moderate extraction efficiencies, as observed for several metal–ligand combinations in Table 1. The same equilibrium model and stoichiometric assumptions were applied consistently to all three ligand series, allowing their K_ext values to be used primarily for comparative purposes under the present experimental conditions. Although the substituents clearly influence the extraction profiles, the present data do not provide evidence for a fundamentally different extraction mechanism among the BzT-Cl, BzT-Me, and BzT-MeO series.

To clarify the relationship between the present work and our previous studies, representative numerical results from the previously reported benzothioate systems are compared with the corresponding results of the present study in Table S3 (SI). Some extraction data for the individual BzT-Cl, BzT-Me, and BzT-MeO series were previously reported separately, including in full conference proceedings. The present study does not treat these previously reported extraction values as entirely new experimental observations. Rather, the three ligand families are consolidated and comparatively evaluated within a common framework, in which their extraction behavior and selectivity are interpreted together with DFT-derived electronic descriptors to examine substituent-dependent structure–property–selectivity relationships.

3.2. Selectivity factors (S_f)

The selectivity factors (S_f) obtained for the BzT-Cl, BzT-Me, and BzT-MeO ligand series revealed clear substituent-dependent differences in metal-ion discrimination (Table 2). Across all BzT-Cl derivatives, Fe³⁺ consistently exhibited exceptionally high selectivity over the remaining metal ions, with S_f values exceeding 300 relative to Co²⁺ and surpassing 1400 relative to Zn²⁺ for BzT-Cl-1. This pronounced selectivity may reflect the combined effects of the high charge density of Fe³⁺ and the electronic properties introduced by the chloro substituent. The particularly low extraction of Zn²⁺ and Co²⁺ relative to Fe³⁺ resulted in the large S_f values observed for these ion pairs. Overall, the BzT-Cl series showed pronounced Fe³⁺ discrimination under the investigated competitive extraction conditions.

For the BzT-Me ligands, the magnitude of the S_f values decreased considerably compared with the most selective BzT-Cl system, indicating reduced discrimination between Fe³⁺ and the competing ions. Although Fe³⁺ remained the preferred ion, for example, S_{f(Fe³⁺/Co²⁺)} decreased from 361.71 for BzT-Cl-1 to 42.77 for BzT-Me-1, while S_{f(Fe³⁺/Zn²⁺)} decreased from 1457.59 to 14.52. This reduction in selectivity was accompanied by increased relative extraction of several divalent ions in the BzT-Me series. The observed change may be associated with the electron-donating character of the methyl substituent and the resulting modification of the electronic properties of the ligand framework.

The BzT-MeO ligands generally exhibited lower Fe³⁺ selectivity factors than the BzT-Cl and BzT-Me series, consistent with their broader extraction profiles. For BzT-MeO-1 and BzT-MeO-2, S_{f(Fe³⁺/Co²⁺)} values were 7.24 and 8.23, respectively, while the corresponding S_{f(Fe³⁺/Zn²⁺)} values were 2.47 and 2.52. BzT-MeO-3 showed greater Fe³⁺ discrimination, with S_{f(Fe³⁺/Co²⁺)} = 33.00 and S_{f(Fe³⁺/Zn²⁺)} = 13.85. In addition, several selectivity ratios between the divalent ions were close to unity, indicating comparatively weak discrimination among these ions. These results show that methoxy substitution is associated with a broader distribution of extraction selectivity, although the magnitude of this effect varies among the individual BzT-MeO derivatives.

Overall, the S_f data demonstrate that substituent modification is associated with substantial changes in metal-ion discrimination within the benzothioate series. The chloro-substituted derivatives exhibited the strongest Fe³⁺ discrimination, whereas methyl and particularly methoxy substitution resulted in a broader distribution of selectivity among the competing ions. These trends should be interpreted as comparative behavior under the present competitive extraction conditions rather than as evidence of a single substituent-controlled coordination mechanism.

3.3. Extraction from natural water samples

Extraction experiments conducted using reservoir, river and tap water samples demonstrated that the extraction behavior of the ligand systems varied among the investigated water matrices (Table 3). The detailed experimental data, including initial metal concentrations, post-extraction concentrations, and calculated removal efficiencies (%) for all investigated water matrices, are provided in Table S1 of the SI. The BzT-Cl derivatives exhibited high removal efficiencies for some metal ions in the real-water matrices, including Zn²⁺ (up to 99.7%) and Mn²⁺ (approximately 93.9%), although the extraction efficiencies varied considerably among the different water samples. In contrast, the BzT-Me ligands showed variable Fe³⁺ extraction efficiencies and generally lower removal efficiencies for several of the remaining ions, depending on the investigated water matrix. The BzT-MeO ligands demonstrated a relatively broad extraction behavior toward Fe³⁺, Mn²⁺, Cu²⁺, and Zn²⁺, although the removal efficiencies varied among the investigated water samples. This relatively broad extraction behavior suggests that methoxy-substituted benzothioate derivatives may warrant further investigation for the simultaneous extraction of multiple metal ions from complex aqueous matrices. Although similar trends were observed in some real-water matrices, the removal efficiencies varied considerably among the investigated samples, indicating that matrix composition may influence the extraction behavior. Because pH and detailed matrix characteristics, including conductivity, total organic carbon, and major-ion composition, were not determined for the individual water samples, the observed matrix-dependent differences cannot be attributed to a specific water-quality parameter. A ligand-containing metal-free blank was included to account for potential background contributions originating from the ligand solution. However, a ligand-free extraction control containing the metal-ion solution and organic phase alone was not included in the original experimental design. Therefore, the possible contribution of passive metal partitioning or non-specific losses cannot be independently quantified from the present dataset, and this should be considered a limitation when interpreting the extraction efficiencies. In addition, the comparatively lower calibration linearity obtained for Cr and Fe should be considered when interpreting the quantitative removal results for these elements.

3.4. LUMO–HOMO levels and Lewis acid–base analysis

All quantum chemical calculations for the synthesized benzothioate derivatives were performed using the ORCA (version 6.1.0) software package.²⁴ Geometry optimizations and electronic structure calculations were carried out within the framework of Density Functional Theory (DFT) at the B3LYP/def2-TZVP level of theory, incorporating Grimme's D3 dispersion corrections with Becke–Johnson damping to account for non-covalent interactions.^26–30 Molecular structures were initially built and visualized using Avogadro 2.²⁵ The frontier molecular orbital (FMO) energies (E_HOMO and E_LUMO) were obtained from the optimized gas-phase geometries (vacuum conditions) corresponding to the lowest energy minima.^32,36,37 Subsequently, global reactivity descriptors, including the HOMO–LUMO energy gap (ΔE), chemical hardness (η), global softness (σ), electronegativity (χ), chemical potential (0), nucleophilicity (ε), and electrophilicity index (ω), were calculated according to well-established procedures reported in the literature.^33–35

The HOMO–LUMO characteristics of similar BzT-Me and BzT-MeO derivatives have been previously reported in the literature.¹⁴ The ΔE values obtained in the present study are consistent with these earlier findings, confirming the reliability of the computational approach. However, unlike previous studies, which mainly focused on electronic structure and biological activity, the current study examines the relationship between electronic descriptors and extraction selectivity (S_f), providing a more application-oriented perspective.

A detailed examination of Fig. 2 reveals that the spatial distributions of the Frontier Molecular Orbitals (FMOs) vary with the nature of the chloro-, methyl-, and methoxy-substituents on the benzothioate skeleton. Across all three series (BzT-Cl, BzT-Me, and BzT-MeO), the HOMO density is primarily localized on the electron-rich regions, specifically the sulfur and oxygen atoms of the thioate linkage, as well as the adjacent aromatic ring. This localization, particularly on the sulfur atom within the C(O)–S group, suggests that these regions may represent potential nucleophilic sites for interaction with metal ions such as Fe³⁺. On the other hand, the LUMO distributions are mostly delocalized over the thioate group and the substituted aromatic ring, indicating regions potentially involved in electron-accepting interactions.^44,45

Fig. 2. HOMO–LUMO map of the examined compounds (BzT-Cl, BzT-Me, and BzT-MeO series) in the gas phase, utilizing the B3LYP functional and the def-2 TZVP basis set.

The calculated global reactivity parameters, such as chemical hardness (η), softness (σ), chemical potential (µ), electronegativity (χ), and electrophilicity index (ω), as well as frontier molecular orbital energies (HOMO and LUMO) and the HOMO–LUMO gaps (ΔE), for all benzothioate derivatives are summarized in summarized in Table S2 (SI) and Fig. 2.

The ΔE values of the ligands fall within a range of 3.976–4.630 eV. Traditionally, a smaller ΔE is associated with higher molecular reactivity and increased polarizability.⁴⁶ However, a detailed comparison between the computational descriptors in Table S2 and the experimental extraction data in Table 1 reveals a more complex relationship that deviates from simple FMO predictions.

Frontier orbital visualizations in Fig. 2 provide further insight into the potential reactive sites of the molecules. The HOMO distributions indicate regions with greater electron-donating character, while the LUMO distributions highlight regions with greater electron-accepting character. Additionally, ionization energies and electron affinities estimated from the negative HOMO and LUMO energies, respectively, provide complementary information on the electronic properties of the ligands.

These computational descriptors complement the experimental findings and suggest that substituent-dependent electronic effects may contribute to the observed differences in ligand reactivity, Fe³⁺ selectivity, and overall extraction behavior.

Within the framework of density functional theory, global reactivity descriptors such as chemical hardness (η) and softness (σ) provide valuable insight into the stability and reactivity of the investigated compounds. According to the Pearson hard–soft acid–base principle, hardness is defined as the resistance of a chemical species to charge transfer, while softness represents its inverse. In this context, molecules with lower hardness values are considered softer and generally more reactive.^47–50

The chemical hardness (η) of the synthesized compounds was calculated using the energies of the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO), based on Koopmans' approximation. The hardness parameter is expressed as:^15,51

η = 0.5 × (energy_LUMO − energy_HOMO) 7

The calculated η values for all compounds are presented in Table S2. As can be seen, variations in HOMO–LUMO energy gaps directly influence the hardness values. Compounds exhibiting larger HOMO–LUMO gaps possess higher hardness, indicating greater kinetic stability and lower chemical reactivity. Conversely, smaller energy gaps correspond to lower hardness values and higher softness, suggesting enhanced reactivity.^34,50,52 Furthermore, the calculated descriptors are consistent with the general HSAB framework, in which softer molecules are more polarizable and can undergo electron-density redistribution more readily than harder molecules. This general relationship, however, does not imply that global softness alone determines the observed metal-ion extraction or selectivity.^34,53 Overall, the hardness analysis complements the electronic structure findings and provides a quantitative descriptor for comparing the electronic characteristics of the studied compounds.

The numerical values presented in Fig. 2 demonstrate variations in the orbital energies and the resulting HOMO–LUMO gap (ΔE) associated with the nature and position of the substituents. Within each series, changes in the substituent position are accompanied by shifts in the HOMO and LUMO energy levels. For instance, the transition from BzT-Cl-1 to BzT-Cl-3, corresponding to a change in the position of the Cl substituent, shifts the HOMO energy from −6.422 eV to −6.174 eV, accompanied by a narrowing of the energy gap from 4.630 eV to 3.976 eV. Overall, the calculated orbital energies reflect both substituent type and positional effects, with the electron-withdrawing and electron-donating characteristics of the substituents contributing to the observed variations. However, the high Fe³⁺ extraction efficiency and selectivity observed for BzT-Cl-1, despite its large HOMO–LUMO energy gap and corresponding high hardness, indicate that global softness alone does not determine Fe³⁺ extraction behavior.^45,46

The relationship between the electronic properties of the ligands and their extraction performance was evaluated by correlating the HOMO–LUMO energy gap (ΔE) and chemical hardness (η) values with experimentally obtained selectivity factors (S_f). The calculated ΔE values for the BzT-Cl, BzT-Me, and BzT-MeO series range from 3.976 to 4.630 eV, corresponding to η values between 1.988 and 2.315 eV (Table S2), indicating moderate variations in electronic softness among the ligands.

In general, lower ΔE (or η) values are associated with higher chemical reactivity and enhanced charge-transfer capability, as predicted by frontier molecular orbital (FMO) theory and conceptual density functional theory (DFT).^44,45,54 However, the present results reveal that extraction selectivity does not follow a simple or direct correlation with these global reactivity descriptors. Comparison of the ΔE and η values with the S_f values indicates the absence of a clear linear relationship, suggesting that global electronic parameters alone are insufficient to explain the observed extraction behavior.

A distinct trend is observed among the ligand series. The BzT-Cl derivatives, which exhibit relatively higher ΔE values (up to 4.630 eV), show high selectivity factors, particularly toward Fe³⁺ ions, with S_f values exceeding 1000 in some cases. This behavior suggests that selectivity is not governed solely by global reactivity descriptors and may also depend on substituent effects, orbital localization, and metal–ligand compatibility.

The high Fe³⁺ selectivity observed for the BzT-Cl series may therefore be associated, at least in part, with the electron-withdrawing character of the chloro substituent and the resulting changes in the electronic structure of the ligand. In contrast, the BzT-Me and particularly BzT-MeO derivatives, which contain electron-donating substituents, generally exhibit lower Fe³⁺ selectivity factors and a broader distribution of selectivity among the competing metal ions. The electron-donating methoxy group, in particular, is associated with changes in the electronic properties of the ligand framework that may contribute to the broader extraction and selectivity profile observed for the BzT-MeO series.

These observations can also be considered within the framework of the HSAB concept. Fe³⁺ is generally classified as a hard Lewis acid and therefore tends to favor interactions with relatively hard donor environments; however, the observed extraction selectivity cannot be attributed to HSAB considerations alone. In the present biphasic system, substituent-dependent electronic effects, ion-pair formation, solvation/desolvation, and phase partitioning may collectively contribute to the experimentally observed selectivity.^48,49

In addition to HOMO–LUMO analysis, global reactivity descriptors such as chemical potential (µ), electronegativity (χ), nucleophilicity (ε), and electrophilicity index (ω) were evaluated to further understand ligand behavior.

Electronegativity (χ) and electrophilicity index (ω) values were calculated according to the literature.^48,55 The electronic chemical potential (µ), which is related to electronegativity, can be obtained from the ionization energy and electron affinity.⁴⁷ Parr introduced the global electrophilicity index (ω), which is computed using chemical hardness (η) and electronic chemical potential (µ).⁵⁵

Nucleophilicity (ε) values have been calculated according to the literature³² and are given in Table S2. According to the nucleophilicity scale proposed by Luis R. Domingo, species with nucleophilicity values (ε) greater than 3.00 eV are classified as strong nucleophiles, whereas those with values below 2.00 eV are considered weak nucleophiles; compounds falling between these limits are categorized as moderate nucleophiles.^56,57 An examination of Table S2 reveals that BzT-Me-3 has the highest calculated nucleophilicity value among the investigated ligands, indicating a comparatively greater electron-donating tendency within this descriptor. The calculated nucleophilicity values for BzT-Me-3 and BzT-MeO-1 were found to be 3.106 and 3.079 eV, respectively, indicating their strong nucleophilic character according to the Domingo classification. The nucleophilicity indices of the remaining compounds range between 2.678 and 2.943 eV, placing them in the moderate nucleophilicity category. The observed variations in nucleophilicity are consistent with substituent-dependent changes in the HOMO energies, including the effects of electron-donating groups on the ligand framework.⁵⁸ Overall, these values indicate moderate-to-strong nucleophilic character among the investigated ligands; however, this global descriptor alone does not establish their coordination strength or selectivity toward Fe³⁺.

The calculated ω values indicate that all compounds possess moderate electrophilic character, providing additional information on their electron-accepting tendencies within the conceptual DFT framework.^44,59 Nevertheless, no direct correlation was observed between electrophilicity index and extraction selectivity, further supporting the conclusion that global descriptors alone are insufficient to explain the observed trends.

Local reactivity considerations provide additional insight into the extraction mechanism. Although numerical Fukui functions were not calculated in the present study, the frontier molecular orbital distributions provide a qualitative indication of the electron-rich and electron-deficient regions of the ligands.^34,59 Regions with high HOMO density may represent potential electron-donating sites, whereas regions with pronounced LUMO density may represent potential electron-accepting sites. In this context, the sulfur and oxygen atoms of the thioate moiety may represent potential interaction sites, consistent with the observed HOMO distributions.

Substituent effects may contribute to variations in the electron density distribution at these reactive centers. The chloro and methoxy substituents produce different electronic environments within the benzothioate framework, which may contribute to the distinct extraction and selectivity profiles observed experimentally.

Overall, the extraction efficiency and selectivity appear to reflect the combined contributions of global electronic properties, potential local interaction sites, substituent effects, and HSAB considerations, rather than being determined by a single descriptor such as ΔE or η.

It should also be noted that the present DFT calculations were performed in the gas phase to provide a consistent reference for comparing the intrinsic electronic effects of the different substituents. Because the extraction experiments were conducted in a biphasic CHCl₃/H₂O system, solvation may influence orbital energies, HOMO–LUMO gaps, dipole moments, and the resulting reactivity descriptors. Therefore, the calculated gas-phase descriptors should be interpreted primarily as comparative electronic parameters rather than as direct representations of the solution-phase extraction process. Inclusion of implicit solvent models, such as CPCM or COSMO, and/or explicit solvation in future calculations would allow the robustness of the observed descriptor trends under solution-phase conditions to be further evaluated.

3.5. Influence of ionic radii and electrostatic interactions on extraction selectivity

To gain a more comprehensive understanding of the extraction selectivity and the competitive behavior observed among the investigated metal ions, the role of ionic radii and charge density must be considered alongside electronic descriptors. According to literature values used here,⁶⁰ the ionic radii of the investigated ions follow the order: Cr³⁺ (61.5 pm) < Cu²⁺ (73 pm) ≈ Zn²⁺ (74 pm) < Co²⁺ (74.5 pm) < Fe³⁺ (78 pm) < Mn²⁺ (83 pm). The experimental extraction data presented in Table 1 indicate that Fe³⁺, despite having a larger ionic radius than Cr³⁺, Cu²⁺, or Zn²⁺ according to these literature values, consistently exhibits the highest extraction efficiency across all ligand series. This behavior may be associated, at least in part, with the relatively high charge density of the trivalent Fe³⁺ ion. As a hard Lewis acid with high charge density, Fe³⁺ may undergo favorable interactions with potential donor sites within the benzothioate framework. These observations can be considered alongside the calculated chemical hardness (η) and electronegativity (χ) values in Table S2; however, these global descriptors alone do not establish the strength or nature of specific metal–ligand interactions. Furthermore, ionic size may also contribute to the differences observed among the divalent metal ions. Mn²⁺ has a larger reported ionic radius (83 pm) than Cu²⁺ (73 pm), corresponding to a lower charge density when only ionic charge and size are considered. However, the appreciable extraction of Mn²⁺ by the BzT-MeO series occurs alongside the relatively high nucleophilicity values calculated for these ligands, suggesting that ligand electronic properties may also contribute to the observed behavior. The electron-donating methoxy groups compensate for the lower charge density of Mn²⁺ by providing a more electron-rich donor environment, thus stabilizing the metal–ligand complex through enhanced covalent contributions. Ultimately, the extraction selectivity is not governed by a single physical parameter but by a synergistic effect between the ionic radius of the metal and the conceptual DFT descriptors of the ligand. The high selectivity for Fe³⁺ is a result of the optimal match between the ion's high charge density and the ligand's specific orbital localization, while the broader spectrum of the BzT-MeO series demonstrates how substituent-induced electronic shifts can overcome the limitations imposed by ionic size.

3.6. SAR (structure–activity relationship)

An extended structure–activity relationship (SAR) analysis reveals that the electronic nature of substituents (Cl, Me, MeO) contributes to the extraction performance and metal-binding preferences of benzothioate ligands. The chlorine substituent, characterized by a strong –I electron-withdrawing effect, is found to decrease the electron density on the thioester moiety. Consistent with its distinct electronic characteristics, BzT-Cl-1 exhibits the highest chemical hardness (η = 2.315 eV) and the largest HOMO–LUMO gap (ΔE = 4.630 eV) among the investigated ligands. These descriptors indicate a relatively ‘harder’ electronic character for BzT-Cl-1, which may contribute to its pronounced Fe³⁺ selectivity within the investigated ligand series.

To better understand this selectivity, the ionic radii of the investigated metal ions should be considered. According to literature values used here, although the Fe³⁺ ion has a larger ionic radius than Cr³⁺, Cu²⁺, and Zn²⁺ ions, it consistently exhibits the highest extraction efficiency. This behavior may be associated, at least in part, with the relatively high charge density of the trivalent Fe³⁺ ion. This property may favor electrostatic interactions with potential donor sites within the BzT-Cl framework.

In contrast, the methyl group acts as a relatively weak electron-donating substituent, and the BzT-Me series exhibits an intermediate extraction profile between the more selective BzT-Cl and broader BzT-MeO systems. The methoxy substituent (–OMe), on the other hand, increases electron density via resonance (+M) interaction with the aromatic ring. The BzT-MeO derivatives also exhibit relatively high nucleophilicity values and comparatively smaller HOMO–LUMO energy gaps, consistent with their distinct electron-donating characteristics. Accordingly, the BzT-MeO derivatives exhibit appreciable extraction of divalent ions such as Mn²⁺ and Zn²⁺ while retaining substantial Fe³⁺ extraction, resulting in a comparatively broad extraction profile. This is particularly evident when considering the larger radius of Mn²⁺ (83 pm) compared with Cu²⁺ (73 pm); the appreciable extraction of Mn²⁺ by the BzT-MeO series occurs alongside the relatively high nucleophilicity of these ligands, suggesting that ligand electronic properties may contribute to the observed extraction behavior despite differences in ionic size.

The observed trends can also be considered within the HSAB framework, with Fe³⁺ representing a hard Lewis acid; however, the differences among the BzT-Cl, BzT-Me, and BzT-MeO series indicate that HSAB considerations alone are insufficient to explain the observed extraction selectivity. Methyl-substituted ligands are observed to display intermediate behavior, characterized by moderate selectivity and balanced extraction efficiency. Overall, a clear substituent-dependent trend is observed: BzT-Cl exhibits the highest Fe³⁺ selectivity and is characterized by relatively high hardness and pronounced orbital localization, whereas BzT-MeO shows a broader extraction profile together with comparatively higher nucleophilicity and polarizability.

To place the present results in a broader context, Table 4 compares the metal-ion recognition, complexation, and extraction behavior of representative crown and thia-crown systems reported in the literature with that of the benzothioate systems investigated in the present study. Although direct quantitative comparison is limited by differences in ligand structures, experimental media, metal-ion sets, concentrations, and analytical methods, the comparison highlights the strong dependence of metal-ion recognition and extraction selectivity on donor-atom composition and ligand architecture.

Table 4. Comparison of the metal-ion recognition, complexation, and extraction behavior of representative crown/thia-crown systems with the benzothioate systems investigated in the present studya.

Ligand system	Method/medium	Metal ions investigated	Quantitative indicator/main reported behavior	Ref.
Tetra-aza-crown ethers	Liquid–liquid extraction; CH2Cl2 and CHCl3, 25 °C	Transition-metal cations	Overall extraction constants (log Kex) and distribution ratios (KD) demonstrated ligand- and solvent-dependent extraction and selectivity	61
Thia-oxa-crown ethers (B1–B6)	Liquid–liquid ion-pair extraction; CHCl3/CH2Cl2	Mg2+, Ca2+, Cr3+, Zn2+, Pb2+, Fe3+, Co2+, Ag+	Ca2+ was the most extractable ion in CHCl3; competitive extraction showed pronounced selectivity toward Fe and Zn ions	15
Benzo-thio/oxa crown ethers (U1–U7)	Liquid–liquid ion-pair extraction; CHCl3/H2O (1 : 1), 25 °C	Li+, Na+, K+, Ni2+, Ca2+, Mg2+, Zn2+, Fe2+, Fe3+, Co2+, Pb2+, Cr3+, Ag+, Cd2+	KD, Ext. %, ΔG and log  Kext evaluated; Fe3+ was the most selective ion for all ligands in competitive extraction	8
Bismithiol crown ethers (Z1–Z4)	Liquid–liquid ion-pair extraction; organic/water, 10−3 and 10−4 M	Ag+, Ca2+, Zn2+, Fe3+, Cr3+, Co2+, Cd2+, Mg2+, Pb2+, Ni2+, Cu2+, Mn2+, Na+, K+	Competitive extraction showed pronounced selectivity toward Cr3+, Fe3+, Cd2+ and Ag+	7
Benzo-thia-aza/thia-triazole crown ethers (HA1–HA8)	UV-vis Job's plot; 50% 1,4-dioxane/H2O, 25 °C	Na+, K+, Mg2+, Pb2+, Cu2+, Co2+, Ca2+, Fe3+, Ag+, Zn2+	Ligand-dependent complexation; HA4–Zn2+, HA6–Mg2+, HA7–Pb2+ and HA8–Cu2+ were among the preferred ligand–ion combinations	62
Aromatic thiadiazole crown ethers (M1–M4)	Conductometric Job's plot	Na+, K+, Mg2+, Ca2+, Zn2+, Fe2+, Ag+, Co2+	M1 and M4 showed pronounced complexation with Ca2+, M2 with Fe2+, and M3 with Co2+	63
BzT-Cl/BzT-Me/BzT-MeO	Competitive liquid–liquid ion-pair extraction; CHCl3/H2O (1 : 1), 25 °C	Zn2+, Mn2+, Cr3+, Fe3+, Co2+, Cu2+	Fe3+ preference; BzT-Cl-1: Ext. = 87.74%, KD = 7.16 and Sf(Fe3+/Zn2+) = 1457.59; extraction/selectivity integrated with DFT descriptors and real-water behavior	This study

^aThe studies summarized above employed different ligand structures, metal-ion sets, solvents, concentrations, and analytical approaches. Therefore, the reported quantitative parameters should be considered for contextual comparison rather than as directly equivalent measures of extraction performance.

4. Conclusions

In this study, a systematic and comprehensive evaluation of chloro-(BzT-Cl), methyl-(BzT-Me), and methoxy-(BzT-MeO) substituted benzothioate ligands was achieved to address the challenge of selective metal removal from complex aquatic environments. The results demonstrate that extraction performance varies systematically with electronic modification of the ligand framework. This structure–activity relationship is associated with substituent-dependent changes in the electronic properties of the benzothioate framework.

Among the investigated ligands, BzT-Cl-1 exhibited the highest chemical hardness (η = 2.315 eV) and the widest HOMO–LUMO energy gap (ΔE = 4.630 eV), together with pronounced Fe³⁺ extraction (87.74%) and selectivity. Notably, this selectivity was observed despite the relatively larger reported ionic radius of Fe³⁺ (78 pm) compared with Cr³⁺ (61.5 pm), suggesting that ionic radius alone does not determine the observed extraction behavior. In contrast, In contrast, the BzT-MeO derivatives exhibited electronic characteristics associated with the electron-donating methoxy substituent, including relatively high nucleophilicity and elevated HOMO energy levels. Experimentally, the BzT-MeO ligands exhibited a broader extraction profile, including appreciable extraction of divalent ions such as Mn²⁺ and Zn²⁺. DFT calculations revealed frontier-orbital localization around the C(O)–S-containing region of the ligands, suggesting its possible involvement in metal–ligand interactions. Real-water experiments further demonstrated that the extraction efficiencies varied considerably among reservoir, river, and tap water samples, highlighting the influence of matrix composition on ligand performance. Overall, the results indicate that metal-ion extraction and selectivity reflect the combined influence of substituent-dependent ligand properties, metal-ion characteristics, and the extraction environment. The insights obtained provide a rational framework for the design of substituent-controlled benzothioate ligands for targeted metal removal and environmental remediation applications.

Author contributions

Ümit Çalışır: conceptualization, data curation, formal analysis, investigation, methodology, validation, visualization, writing (original draft and review & editing), resources, software, project administration, supervision, and funding acquisition. Baki Çiçek: conceptualization, data curation, formal analysis, investigation, methodology, validation, visualization, writing (original draft and review & editing), resources, software, project administration, supervision, and funding acquisition.

Conflicts of interest

There are no conflicts to declare.

Data availability

All data supporting the findings of this study are included within the manuscript tables and the supporting information (SI). Supplementary information: metal removal data for real water samples, DFT-calculated molecular descriptors, and a numerical comparison of the present results with previously reported extraction studies. Additional raw experimental data, including ICP-OES and ICP-MS results, as well as the comprehensive ORCA computational output files, can be made available upon reasonable request to the corresponding author. See DOI: https://doi.org/10.1039/d6ra04929j.