Confirming the Molecular Basis of the Solvent Extraction of Cadmium(II) Using 2-Pyridyl Oximes through a Synthetic Inorganic Chemistry Approach and a Proposal for More Efficient Extractants

The present work describes the reactions of CdI2 with 2-pyridyl aldoxime (2paoH), 3-pyridyl aldoxime (3paoH), 4-pyridyl aldoxime (4paoH), 2-6-diacetylpyridine dioxime (dapdoH2) and 2,6-pyridyl diamidoxime (LH4). The primary goal was to contribute to understanding the molecular basis of the very good liquid extraction ability of 2-pyridyl ketoximes with long aliphatic chains towards toxic Cd(II) and the inability of their 4-pyridyl isomers for this extraction. Our systematic investigation provided access to coordination complexes [CdI2(2paoH)2] (1), {[CdI2(3paoH)2]}n (2), {[CdI2(4paoH)2]}n (3) and [CdI2(dapdoH2)] (4). The reaction of CdI2 and LH4 in EtOH resulted in a Cd(II)-involving reaction of the bis(amidoxime) and isolation of [CdI2(L’H2)] (5), where L’H2 is the new ligand 2,6-bis(ethoxy)pyridine diimine. A mechanism of this transformation has been proposed. The structures of 1, 2, 3, 4·2EtOH and 5 were determined by single-crystal X-ray crystallography. The complexes have been characterized by FT-IR and FT-Raman spectra in the solid state and the data are discussed in terms of structural features. The stability of the complexes in DMSO was investigated by 1H NMR spectroscopy. Our studies confirm that the excellent extraction ability of 2-pyridyl ketoximes is due to the chelating nature of the extractants leading to thermodynamically stable Cd(II) complexes. The monodentate coordination of 4-pyridyl ketoximes (as confirmed in our model complexes with 4paoH and 3paoH) seems to be responsible for their poor performance as extractants.


Introduction
Organic matter and heavy toxic metals are the main pollutants of wastewaters, the threat from the latter being more serious [1][2][3][4][5][6]. This is due to the non-biodegradable and non-decomposable nature of the toxic metals, making the development of efficient approaches for their removal and uptake extremely important [7]. Methods in action involve chemical precipitation, microbial treatment, electrodeposition, reverse osmosis, physical/chemical adsorption and solvent extraction [1,[8][9][10][11][12]. Solvent extraction is very useful for base metal recovery; the desired metal ions are transferred selectively (after the organic phase are the neutral complexes [CdCl 2 (2PC12) 2 ] and [CdCl 2 (2PC14) 2 ]. Using a synthetic coordination chemistry approach [14], we proved that such complexes are capable of existence. The primary goal of the present study was to contribute in understanding the molecular basis of the experimental fact (a) mentioned above. Although the reason of the superior extraction capability of 2PC12, 2PC14 compared to those of 4PC12, 4PC14 might be obvious, i.e., the formation of stable 5-membered chelating rings with the N-donor atoms of the former extractants-which give a thermodynamic stability of the complexes in solution-and the inability of chelating behavior in the case of 4PC12, 4PC14, we were interested in providing synthetic and structural evidence for this working with model complexes. Thus, we used the three isomers of pyridyl aldoximes (2-pyridyl aldoxime, 2paoH; 3-pyridyl aldoxime, 3paoH; 4-pyridyl aldoxime, 4paoH), Figure 1, which gave crystalline complexes. The compounds 2paoH and 4paoH are somewhat satisfactory analogs (albeit not ideal ones) of the extractants 2PC12, 2PC14 and 4PC12, 4PC14, respectively. The main difference is the presence of a long aliphatic (hence hydrophobic) chain in the real extractants instead of the -CHNOH group in the ligands. Cadmium(II) complexes with the ligand 3paoH were also studied. This paper describes results from the synthetic investigation of the general reaction systems CdI 2 /2paoH, CdI 2 /3paoH and CdI 2 /4paoH. For consistency reasons, we used the same inorganic anion of the Cd(II) source; although we originally worked with CdCl 2 ·2H 2 O to better simulate the real extraction conditions (e.g., chloride solutions), CdI 2 gave better crystallinity of the products and their structures were solved through single crystal X-ray crystallography (vide infra). A secondary goal of this work was to propose types of ligands that might be better extractants than the 2-pyridyl ketoximes 2PC12 and 2PC14. For this reason, we performed reactions of CdI 2 with 2,6-diacetylpyridine dioxime (dapdoH 2 , Figure 1) and 2,6-diacetylpyridine diamidoxime (LH 4 , Figure 1), to investigate whether the potentially N(pyridyl), {N(oxime)} 2 tridentate chelating character of these ligands could be realized; in such a case, molecules analogous to dapdoH 2 and LH 4 might be better extractants for Cd(II). The ligand LH 4 possesses two amidoxime moieties which, when introduced in materials (e.g., polymers) give excellent adsorbents for the efficient recovery of Cd(II) from aqueous media [1,21,28].

Comments on the Syntheses of the Complexes
Since the pH of the aqueous phases during the real extraction experiments was~4.0 and the extractants were neutral, we avoided adding external bases (e.g., OH − , Et 3 N, etc.) which would deprotonate the ligands used. The optimized reaction systems used are conveniently summarized in Figure 2. Optimized experimental reaction conditions that lead to Cd(II) complexes (1)(2)(3)(4)(5); the lattice EtOH molecules in the dapdoH 2 complex are not shown. The structural formula of the ligand L H 2 and a possible mechanism of the LH 4 → L H 2 transformation are presented in Figure 3 (vide infra).
The reaction schemes illustrated in Figure 2 deserve synthetic comments. Thus: (i) The use of 1:1 reaction ratio in the case of the CdI 2 /(2,3,4)paoH systems (the complexes were isolated using 1:2 reaction ratios, which coincide with their stoichiometries) does not affect the product identity, leading again to compounds 1-3 (microanalytical and IR evidences). (ii) Complexes 2,3 can be precipitated from MeCN solutions (IR evidence), but their crystallinity is lower compared to that using EtOH and Me 2 CO as solvents, respectively. (iii) The use of excess dapdoH 2 in its reaction with CdI 2 (e.g., a 2:1 dapdoH 2 :CdI 2 ratio) does not affect the product identity (IR and Raman evidences); the complex was originally isolated using the 1:1 reaction which coincides with its stoichiometry. However, if the reaction solution is concentrated enough (to increase the low yield), the solid product is contaminated with free ligand; and (iv) The study of the 1:1 CdI 2 /LH 4 reaction system in EtOH provided us with surprising results. The initially precipitated solid (in a yield of 10%) was analyzed perfectly as [CdI 2 (LH 4 )] (5a), as expected. The identity of the product was confirmed by its IR and 1 H NMR spectra (vide infra). Despite our intense efforts, we could not grow crystals of 5a for detailed structural characterization. Somewhat to our surprise, an unexpected experimental result was observed during the storage of the filtrate at room temperature; within four days, its color turned slowly pale violet! The pale violet solution was kept in the refrigerator (~5 • C) and X-ray quality, colorless crystals of [CdI 2 (L H 2 )] (5) were grown in a~40% yield (based on the initially used metal source) within one week. Single-crystal X-ray crystallography revealed that 5 contained the transformed ligand L H 2 , i.e., the LH 4 → L H 2 transformation had occurred. The free compound L H 2 is not known in organic chemistry. We do believe that the observed transformation is CdI 2 -promoted/assisted, since ethanolic solutions of LH 4 are stable for months. The metal ion-involving reactions of the amidoxime group are well known [46] in the field of the reactivity of coordinated ligands. No analogous reaction has been observed to date in the chemistry of amidoximes. Without any detailed mechanistic studies, we propose the simplified mechanism shown in Figure 3 for the LH 4 → L H 2 transformation. The formation of the L H 2 probably involves nucleophilic attack of the solvent to the oxime C atom (Cd(II) coordination to the oxime nitrogen might activate the amidoxime groups towards nucleophilic attack) followed by the reduction in the oxime group by the generated HI, with the simultaneous production of I 2 ; the latter justifies the observation of the pale violet color of the reaction solution. In accordance with our proposal, the transformation does not take place in MeCN from which a very small quantity of 5a is precipitated.

Spectroscopic Characterization in Brief
The complexes were characterized in the solid state by IR and Raman spectroscopies. Representative spectra are shown in Figure 4, Figure 5, Figure S1, Figure S2, Figure S3, Figure S4 and Figure S5. The spectra do not exhibit bands that are present in the free ligands 2paoH, 3paoH, 4paoH, dapdoH 2 and LH 4 , suggesting their purity; if such bands were present, the complexes would have been contaminated with the free ligands used as starting materials. The presence of neutral oxime groups in 1, 2, 3 and 4 (well-dried unsolvated sample) is manifested by broad bands (with 2-4 submaxima in the spectra of 1 and 4) at~3400 cm −1 assigned to ν(OH) [38,39]. The bands at 3444 and 3292 cm −1 in the IR spectrum of 5 reflect the existence of the imino (=NH) groups in the complex [47]. In 5a, the bands at 3480 [ν as (NH 2 )], 3432 [ν s (NH 2 )] and 3372 [ν(OH)] reflect the existence of -NH 2 and -OH groups supporting the incorporation of coordinated LH 4 in the complex. The corresponding bands in the free LH 4 appear at 3484, 3420 and~3380 cm −1 . The broadness of the ν(OH), ν(NH), ν as (NH 2 ) and ν s (NH 2 ) bands, combined with their relatively low wavenumber, are both indicative of hydrogen bonding [38,39]. As expected, the O-H and N-H stretching vibrations are hardly seen in the Raman spectra of the complexes. The peaks at 3076-2922 cm −1 are assigned to ν(C-H) vibrations [14,48,49]. The in-plane, δ(py), and out-of-plane, γ(py), deformation vibrations of the 2-pyridyl ring of free paoH (at 627 and 404 cm −1 , respectively) shift upwards (at 646 and 466 cm −1 , respectively) in 1 suggesting coordination of the ring-N atom [38]. The δ(py) vibration appears as a weak peak in its Raman spectrum [48]. The same trend is observed in the vibrational spectra of the other complexes. For example, the δ(py) and γ(py) bands are at 658 and 486 cm −1 , respectively, in the IR spectrum of 3, while the corresponding vibrations in the spectrum of free 4paoH are located at 640 and 452 cm −1 , respectively. The ν(C=N) vibration of the oxime group(s) in the IR spectra of 1, 2, 3, 4 and 5a appear as medium to weak bands at 1640, 1624, 1638, 1594 (overlapping with an aromatic stretch) and 1654 cm −1 , respectively [38]; the corresponding Raman vibrations are assigned [14,48] to the peaks at 1632 (1), 1624 (2), 1622 (3) and 1589 (4) cm −1 . The wavenumbers for 2 and 3 are approximately the same with those of the free ligands 3paoH and 4paoH, respectively. This is strong evidence that the oxime nitrogen does not participate in coordination to Cd(II), a fact that has been confirmed through single-crystal X-ray crystallography (vide infra). The ν(C=N) band/peak (IR at 1594 cm −1 and Raman at 1638 cm −1 ) for 4 is located at lower wavenumbers compared with those of free dapdoH 2 , suggesting oxime-N coordination. Somewhat to our surprise, the 1626 cm −1 band of 2paoH is shifted to a higher wavenumber in the IR spectrum of 1 (1640 cm −1 ) for which the oxime-N coordination has been confirmed (vide infra). This fact, which is not unusual [14], has been interpreted [50] on the basis that some ligands containing a C=N bond (with the carbon atom attached to an aromatic ring) have been shown to undergo a change in the s character of the N lone pair upon coordination; the s character of the N orbital in the C=N bond increases resulting in a greater C=N stretching force constant relative to the free neutral ligands, and this in turn shifts the ν(C=N) band in the spectra of the complexes to higher frequencies. An analogous trend is observed in the IR spectrum of 5a for which coordination of both oxime N atoms is proposed. The medium-intensity IR band at 942 cm −1 for free 2paoH is assigned to the ν(NO) oxime mode; the corresponding weak Raman peak appears at 950 cm −1 [49]. The 942 cm −1 band is shifted to a lower wavenumber (932 cm −1 ) in 1; this shift is due [14,39] to the coordination of the neutral oxime nitrogen. The same trend is observed in the Raman spectrum of the complex where this mode is located at 925 cm −1 . The assignment of the ν(NO) oxime mode for 3paoH, 4paoH, dapdoH 2 and LH 4 is not an easy task and any discussion about coordination (or non-coordination) shifts in the spectra of the complexes would be risky.  The complexes were characterized in solution by 1 H NMR spectroscopy in d 6 -DMSO ( Figure S6, Figure S7 and Figure S8). The spectrum of free 2paoH displays singlet signals at δ 11.66 and 8.08 ppm assigned to the hydroxyl proton and to the proton attached to the oxime carbon, respectively, and a doublet signal at δ 8.57 ppm assigned to the α aromatic proton (i.e., the proton bonded to the carbon next to the pyridyl nitrogen atom). The corresponding signals in the spectrum of 1 appear at δ 10.97, 7.42 and 9.13 ppm. The upfield shift of the oxime carbon-bonded proton and the downfield shift of the α proton in the spectrum of the complex both indicate coordination of the two 2paoH N atoms in solution [51], suggesting that the structure of 1 is retained in solution. This is corroborated by the almost zero value of the molar conductivity of the complex in DMSO (Λ M = 4 S cm 2 mol −1 for a 10 −3 M solution at 25 • C) [52]. Given the stability of 1 in DMSO, we were rather surprised to realize that 4 (which possesses two chelating rings per ligand dapdoH 2 ) decomposes in solution! Its 1 H NMR spectrum in d 6 -DMSO is identical to that of free dapdoH 2 displaying a simple set of signals at δ 11.51 (-OH, singlet), 7.80 (aromatic protons, multiplet) and 2.25 (-CH 3 , singlet) ppm in the expected 2:3:6 integration ratio. This result, together with the negligible Λ M value in DMSO, indicates a decomposition probably through Equation (1), where x ≥ 4. The Cd II -I bond is very stable as suggested by the absence of crystal structures of Cd(II)-DMSO complexes possessing ionic iodides [53]. A rather poor-quality (due to solubility reasons) 1 H NMR spectrum of 4 in CDCl 3 is complicated indicating two different solution species, both of which seem to contain coordinated dapdoH 2 . The spectrum of 2 in d 6 -DMSO is identical to that of free 3paoH displaying signals at δ 11.57 (-OH,

Description of Structures
The structures of 1-3, 4·2EtOH and 5 were determined by single-crystal X-ray crystallography. Crystallographic data are gathered in Table 1. Structural plots are shown in Figures 6-11 and S9-S12. Selected interatomic distances and angles are listed in Tables 2-6.       Neighboring molecules of 1 interact through π-π stacking interactions involving the 2-pyridyl rings (symmetry operation: −x −1/2, −y + 1/2, −z) forming chains parallel to the [101] crystallographic direction ( Figure S9), which are further connected through weak hydrogen bonding interactions creating the 3D architecture of the crystal structure. The distance between the neighboring centrosymmetric 2-pyridyl rings within the chain is 3.83(1) Å.   a Symmetry code: ( ) = −x, y, −z + 1/2. Atoms C1 and C1 , not labelled in Figure 6, are the oxime carbon atoms of coordinated 2paoH.

Interatomic Distances (Å) Interatomic Angles ( • )
Cd1-N2/N2 2.353 (     The crystal structure of 4·2EtOH consists of mononuclear molecules [CdI 2 (dapdoH 2 )] and lattice EtOH molecules in an 1:2 ratio. The asymmetric unit of the cell contains the full complex molecule and the two solvent molecules. Each of the latter interacts (as acceptor) with one "free" (i.e., uncoordinated) oxime oxygen atom (donor) of the dapdoH 2 ligand (Figure 9, Table S1). The Cd II atom forms coordination bonds with two terminal iodo ligands (I1, I2), the two oxime nitrogen atoms (N1, N3) and the pyridyl nitrogen atom (N2) of dapdoH 2 . Thus, the organic molecule behaves as a 1.00111 ligand and participates in two 5-membered chelating rings with the metal ion. The Cd II -N(pyridyl) bond [2.333(2) Å] is slightly stronger than the Cd II -N(oxime) bonds [2.421(2), 2.443(2) Å]. The terminal Cd II -I bonds [2.722(1), 2.733(1) Å] are stronger than the corresponding bonds in 1 [2.829(1) Å], and this is due to the lower coordination number of Cd II in 4·2EtOH (five) compared with that in 1 (six). The coordination geometry of Cd II in the complex is extremely distorted, a fact that is primarily attributed to the small bite angles of the two 5-membered N(oxime)CCN(pyridyl)Cd chelating rings; both N(oxime)-Cd-N(pyridyl) coordination angles are 67.5(1) • . The geometry can be either described as a very distorted trigonal bipyramidal one with atoms N1 and N3 defining the axial positions, or as a very distorted square pyramidal one with atoms I1, I2, N1, N3 occupying the basal plane and N2 being at the apical position.
Compound 5 crystallizes in the orthorhombic space group Pcnb. As the complex possesses a 2-fold axis of symmetry passing through Cd1, N2 and C6 atoms (Wyckoff position 4c: 0, 1 4 , z), the asymmetric unit of the cell contains 1 2 of the [CdI 2 (L H 2 )] molecule. The 5-coordinate Cd II atom forms coordination bonds with the two terminal iodo groups (I1, I1 ), the two imino nitrogen atoms (N1, N1 ) and the pyridyl nitrogen atom (N2) of the transformed ligand L'H 2 . Thus, the organic molecule behaves as a 1.00111 ligand participating in two 5-membered chelating rings with the metal center. The terminal Cd II -I bonds [2.738(1) Å] are almost identical to those of 4·2EtOH [average 2.727(1) Å], a consequence of the 5-coordination of Cd II in the two complexes. As in 4·2EtOH, the coordination polyhedron of the metal ion in 5 is extremely distorted, a fact primarily arising from the small N1-Cd1-N2 and N1 -Cd1-N2 [68.2(1) • ] coordination angles of the chelating "parts" of L H 2 . The polyhedron can be better described as a very distorted trigonal bipyramid with atoms N1 and N1 occupying the axial positions. Neighboring molecules of 5 interact through pairs of C(methyl)-H . . . π interactions forming chains parallel to the a axis; neighboring chains interact through C(methyl)-H . . . I hydrogen bonds creating layers parallel to the (001) crystallographic plane ( Figure S12, Table S2). Compound 5 is the first structurally characterized complex of any metal containing the new ligand L H 2 .

Materials and Spectrocopic-Physical Measurements
Experimental manipulations were carried out under aerobic conditions. Deionized water was received from the in-house facility. Solvents and reagents were purchased from Sigma-Aldrich (Tanfrichen, Germany) and Alfa Aesar (Karlsruhe, Germany), and used as received without extra purification. The free ligands 2,6-diacetylpyridine dioxime (dapdoH 2 , Figure 1) and 2,6-pyridyl-diamidoxime (LH 4 , Figure 1) were synthesized by following the procedures published in the literature [61,62]. The products were recrystallized from refluxing EtOH, and their yields were >70%. The purity of the free organic ligands was checked by microanalyses and 1 H NMR spectroscopy.
Carbon, hydrogen and nitrogen microanalyses were performed by the Instrumental Analysis Center of the University of Patras. FT-IR spectra were recorded using a Perkin-Elmer spectrometer, model 16PC, manufactured by Perkin-Elmer (Waltham, MA, USA); the samples were in the form of KBr pellets prepared under pressure. FT-Raman spectra were obtained in an EQUINOX spectrometer to which a Bruker (D) FRA-106/S component had been attached (Bruker, Karlsruhe, Germany); an R510 diode-pumped Nd:YAG laser at 1064 nm was used for Raman excitation with a maximum laser power of 500 mW on the sample, utilizing an average of 100 scans at 4 cm −1 resolution. 1 H and 13 C NMR spectra were recorded on a Bruker Avance DPX spectrometer (Bruker AVANCE, Billerica, MA, USA) at resonance frequencies of 400.13 MHz ( 1 H) and 100.62 MHz ( 13 C); the signals of the solvent (d 6 -DMSO) were used as a reference. Conductivity measurements were performed at room temperature (23-25 • C) in DMSO with a Metrohm-Herisau E-527 bridge (Herisau, Switzerland) and a cell of standard constant; the concentration of the solution was 10 −3 M.

Conclusions in Brief
According to our opinion, the important chemical messages of this work are: (a) The reported complexes enrich the coordination chemistry of 2-pyridyl oximes, especially that with Cd(II). (b) The molecular structures and supramolecular features of the complexes are interesting; and (c) The interesting Cd(II)-assisted/promoted LH 4 → L'H 2 transformation has been observed for the first time and this is a new, welcome example in the area of the reactivity of coordinated amidoximes.
From the viewpoint of the solvent extraction of toxic Cd(II) using 2-pyridyl oximes as extractants (which stimulated the present efforts) our inorganic chemistry approach has firmly confirmed the molecular basis of the excellent extraction ability of 2PC12 and 2PC14, and the poor one for 4PC12, 4PC14. With the drawbacks mentioned in Introduction, the very good extraction ability of the 2-pyridyl ketoximes can be attributed to the chelating nature of the extractants as structurally proven by the 2paoH-containing complex 1; this chelating behavior results in thermodynamically stable complexes of Cd(II) with the extractant, favoring this process. The monodentate coordination of 4-pyridyl ketoximes (as structurally proven in the 4paoH-containg compound 3) seems to be responsible for the poor performance of 4PC12 and 4PC14. In an analogous manner (as proven in the 3paoH-conataining compound 2), extractants such as 3PC12 and 3PC14 (i.e., those containing the oxime group at position 3 of the pyridyl ring; Figure 1) are predicted to disfavor the extraction process; such "real" experiments are not available.
We tentatively propose that the structurally established tridentate chelating character of dapdoH 2 towards Cd(II) in complex 4·2EtOH and the presumable such behavior of LH 4 in complex 5a could be exploited to develop extractants consisting of one pyridyl ring and two oxime groups that contain long alkyl components at the 2-and 6-positions, or even to carry out experiments with the bis(amidoxime) compound LH 4 (or better with derivatives containing aliphatic substituents on the pyridyl ring to ensure good solubility in non-polar organic solvents). The decomposition of 4 in DMSO (as evidenced by its 1 H NMR spectrum in d 6 -DMSO, see Section 2.2) should not be disappointing, since this solvent is an efficient donor forming complexes with Cd(II) and favoring decomposition; the presence of coordinated dapdoH 2 -containing species in CDCl 3 (albeit evidenced by poor quality 1 H NMR spectra) is a good sign towards the use of 2,6-pyridyl dioximes as extractants for toxic Cd(II). The scientific community dealing with solvent extraction experiments might obtain good results working with tridentate dioximes.
Author Contributions: A.R., Z.G.L. and V.A. contributed towards the synthesis, crystallization, and conventional characterization of the complexes. Z.G.L. also contributed to the interpretation of the results and performed the Raman experiments. C.P.R. and V.P. collected single-crystal X-ray crystallographic data, solved the structures, and performed their refinements; the latter also investigated the supramolecular characteristics of the crystal structures and wrote the relevant part of the paper. K.F.K., C.T.C. and S.P.P. coordinated the research and wrote the paper based on the reports of their collaborators. S.P.P. coordinated the cooperation between the teams and submitted the manuscript. All the authors exchanged opinions concerning the progress of the experiments and commented on the various drafts of the paper. All authors have read and agreed to the published version of the manuscript.