Synthesis of pincer-type extractants for selective extraction of palladium from PGMs: An improved liquid-liquid extraction approach to current refining processes

SCS pincer ligands 1–4 were synthesised, and their ability to extract Pd(II) from HCl and HNO3 media was studied. The Pd(II) extraction properties of 1–4 were compared with those of commercial extractants (DOS and LIX®84-I) in kerosene. 1 and 2 showed superior Pd(II) extractability (E% = 99.9) relative to DOS and LIX®84-I from 0.1–8.0 M HCl and to DOS from 0.1–8.0 M HNO3 and mixed HCl + HNO3 media. The Pd(II) extraction rate, acid durability, the most suitable organic/aqueous (O/A) phase ratio, and Pd(II) loading capacity of extractants 1, 2, and DOS were evaluated. 1 and 2 exhibited a greater Pd(II) extraction rate and Pd(II) loading capacity than DOS. 1 was very stable in acid media (HCl and HCl + HNO3), whereas 2 and DOS deteriorated in HCl + HNO3. Selective extraction of Pd(II) by 1 and 2 was achieved from a mixed solution containing Pd, Pt, Rh, rare metals, and base metal ions that simulated the leach liquors of automotive catalysts. The back extraction of Pd(II) and reusability of extractants 1 and 2 were studied. The Pd(II) extraction mechanism of 1–4 was investigated using FT-IR, UV-visible, and NMR spectroscopy.

Germany) carry out hydrometallurgical refining by leaching precious metals from concentrated primary and secondary sources using HCl + Cl 2 1, 11 . The liquors obtained from primary ore leaching contain Au and PGMs as chloro complexes. Secondary sources including automotive catalysts are leached with HCl/Cl 2 , HCl/H 2 O 2 , or aqua regia 2,5 . The corresponding liquors contain Pd, Pt, and Rh and other rare earth and base metals depending on the manufacturer. The Vale, Johnson Matthey, Anglo American Platinum, Lonrho, Sumitomo Metal Mining (Ehime, Japan) 12 , and Heraeus refineries use solvent extraction to recover PGMs from acid-leached liquors. Other refineries use precipitation and ion-exchange processes 1,10,11 . Commercial extractants such as di-n-hexyl sulphide (DHS), di-n-octyl sulphide (DOS), and 2-hydroxy-5-nonylacetophenone oxime (LIX ® 84-I) 1,11 are used in the solvent extraction process. Although these reagents are useful in separating Pd(II) and Pt(IV), they are not without issues regarding extraction rate, selectivity, and durability. The extractants can be oxidised by contact with a highly acidic aqueous phase, which renders Pd(II) separation ineffective. Johnson Matthey and Anglo Platinum refineries recover Pd(II) from acidic leach liquors with LIX ® 84-I. The kinetics of Pd(II) extraction using LIX ® 84-I are extremely slow, and it is necessary to use an organic amine as an accelerator to increase the extraction rate 1 .
However, use of an organic amine decreases the selectivity of Pd(II) extraction. LIX ® 84-I extracts Pd(II) from pH 1-3 HCl media with properties that depend greatly on the acidity of the aqueous phase 13,14 . The raffinate contains < 5 mg L −1 Pd(II) after extraction 1 . The Vale, Lonrho and MINTEK refineries use DHS or DOS to separate Pd(II) from acid leach liquors 1, 10-12 . Pd(II) extraction kinetics with DHS or DOS are slow, but < 1 mg L −1 Pd is retained in the raffinate after extraction 1 . DHS and DOS are oxidised to di-n-hexyl sulphoxide (DHSO) and di-n-octyl sulphoxide (DOSO), respectively, during extraction by contact with oxidising agents in the acidic aqueous phase 1,[15][16][17] . Oxidised DHS and DOS extract Fe(III), Rh(III) and Pd(II) considerably, which decreases the selectively of Pd(II) extraction 1,15,16 . Many researchers have employed thiodiglycolamide [18][19][20] and acyclic dithioethers 21,22 to selectively extract Pd(II) from PGM-containing solutions. Commercial hydrocarbon fluids (kerosene and ISOPAR M) have been used as diluents in PGMs refineries. Many synthesised and commercial extractants [23][24][25][26][27][28][29][30][31][32][33][34][35][36] exhibit slow extraction rates and extract other PGMs and base metals in minor amounts in addition to Pd ions, which affects the selectivity of extraction. In these cases, extraction removes only Pd(II) into the organic diluent, chlorinated diluent, or mixture of diluents [18][19][20][21][22][28][29][30][31][32][33] , and oxidation in acidic media causes Pd(II) extraction to be highly dependent on the acidity and type of leach liquors. Pd(II) extraction efficiency is good in weakly acidic media, but decreases dramatically in strong acid media 22,31,[33][34][35] . Synthesis of new, robust extractants faces several major issues including use of expensive reagents, multiple synthetic steps 22,31 , advanced purification methods, durability, and difficulties in back extraction and reusability. To date, no effective extractants applicable to industrial operations have been developed for complete recovery of Pd(II) from primary and secondary PGM resources. Development of novel, inexpensive extractants possessing a high extraction rate, selective and efficient Pd(II) separation, and superior durability in acidic media is required. Our goal is to develop new extracting agents for the selective and efficient recovery of Pd(II) from leach liquors of automotive catalysts. We previously used the 1,3-bis(dimethylthiocarbamoyloxy)benzene pincer ligand in various organic solvents to selectively extract Pd(II) from the leach liquors of automotive catalysts 30 . Unfortunately, the solubility of 1,3-bis(dimethylthiocarbamoyloxy)benzene in hydrocarbon diluents is low, which limits its industrial application 30 .
The objective of our research is to synthesise new SCS pincer-ligand extractants that are soluble in commercial hydrocarbon diluents and capable of effectively separating Pd(II) in the presence of other PGMs from the acid leach liquors of automotive catalysts. We designed the 1,3-bis(2-(octylthio)propan-2-yl)benzene ligand (1) having two n-octyl moieties to increase hydrophobicity as a pincer-type extractant to achieve this goal. We also synthesised the SCS pincer-type extractants 1,3-bis(octylthio)methyl)benzene (2), dimethyl[1,3-phenylenebis (1-methylethylidenethio)]diacetate (3) and dimethyl[1,3-phenylenebis(methylenethio)]diacetate (4) and compared their properties with 1. Extractant 1 can be synthesised at the gram to kilogram level in high yield at room temperature from commercially available reagents via a one-step reaction with simple workup procedures. 1 and 2 exhibit large extraction capacities and Pd(II) selectivities compared with 3, 4, and commercial LIX ® 84-I and DOS extractants. Extractant 2 exhibits behaviour similar to 1 and is suitable for Pd(II) extraction from HCl media, but is not recommended for HCl + HNO 3 media. The Pd ion extraction mechanisms have been confirmed using various spectroscopic methods. Pd(II) extractability of extractant 1 is higher than commercial DOS and it can be used for the successful separation Pd(II) in PGM refineries.

Results and Discussion
Extraction of Pd(II) from HCl media with 1-4, DOS, and LIX ® 84-I. The structures of the SCS-type pincer extractants (1-4) are shown in Fig. 1. Extractants 1-4 contain two sulphur atoms, whereas DOS has only one. To establish sulphur atom equivalence in comparative studies, the concentration of DOS was maintained at twice that of 1-4. Pd(II) extraction was studied as a function of HCl concentration from 0.1-8.0 M HCl using 1.0 mM 1-4 and 2.0 mM DOS in kerosene for 60 min. The effect of HCl concentration on Pd(II) extraction is shown in Fig. 2(a). The Pd(II) extraction percentage (E%) of 1, 2, and DOS was > 99.1% in 0.1 M HCl; extractants 4 and 3 exhibited 91.4 and 35.7% efficiency, respectively. The E% of all extractants decreased with increasing HCl concentration. The E% of 1, 2, and 4 in 1.0 M HCl equalled 68.4, 84.4, and 84.7%, respectively, but that of 3 and DOS was only 11.9 and 28.5%, respectively. In 8.0 M HCl, the Pd(II) extractability of 2, 4, 1, and DOS decreased to 73.9, 55.2, 37.3, and 19.6%, respectively. The Pd(II) extraction capacity of 3 was nil from 3.0-8.0 M HCl. The experimental results show a distinct dependence of E% on HCl concentration and extractant structure.
The extractant concentration was increased ten-fold to increase E% in concentrated HCl (cf. Fig. 2(a)   Extraction by 1 and 2 reached saturation within 30-60 min, whereas that by DOS required 180 min 1,19 . Many researchers have reported that slow Pd(II) extraction kinetics by DOS is a major disadvantage in PGM refineries 1,11 . The superior extractability and extraction kinetics of 1 and 2 relative to DOS may derive from Pd(II) coordination by adjacent sulphur atoms in the pincer ligand. Thus, the SCS pincer-type reagents 1 and 2 are more potent Pd(II) extractants than DOS. The Pd(II) extraction mechanism of 1 and 2, which differs from that of DOS will be discussed in later section. We previously reported the rapid and selective extraction of Pd(II) using the SCS pincer ligand 1,3-bis(dimethylthiocarbamoyloxy)benzene and described the extraction mechanism in terms of intramolecular coordination 30 . Extraction of Pd(II) from HNO 3 and mixed HCl-HNO 3 media. In many cases, PGM-leached liquors are prepared from spent automotive and industrial catalysts using HNO 3 and aqua regia [37][38][39] . To check the Pd(II) extraction ability of 1, 2, and DOS in HNO 3 or mixed HCl + HNO 3 solution, studies were conducted using 10 mM 1 or 2 or 20 mM DOS with 1 mM Pd(II) from 0.  40 and other reagents have been reported to deteriorate upon repeated extraction of PGMs from leach liquors 1,15,19 . Extractant degradation has been attributed to oxidation or other reactions upon contact with acid-leached solutions, which decreases extraction capacity and reusability. Deteriorated reagents also co-extract undesired metals ions 15 . To study the acid durability of 1, 2, and DOS, 1 mL of each extractant was diluted in 10 mL of CHCl 3   The FT-IR spectra of acid-treated DOS showed new sharp peaks at 1016 and 1088 cm −1 corresponding to protonated -S=O and free -S=O, respectively 19 . The HCl + HNO 3 medium oxidises the sulphide group of DOS to sulphoxide 19,41 , consistent with reports that DHS and DOS are prone to oxidation upon acid contact 1,15,19,41 . The 1 HNMR spectra of native and acid-treated DOS are shown in Fig. S3. The 1 H NMR spectrum of acid-treated DOS differs from that of native DOS due to oxidation, and corresponds exactly to the standard 1 H NMR spectrum of di-n-octyl-sulphoxide 42  The FT-IR and 1 HNMR spectra of 1 treated with 2.0 M HCl + 1.0 M HNO 3 are shown in Figs. S6 and S7. In these spectra, the acid-treated extractant shows peaks corresponding exactly to its native form. This result confirms that extractant 1 is stable in 2.0 M HCl + 1.0 M HNO 3 and that its sulphur atoms are not oxidised by acid treatment. The acid resistance of the sulphur atoms in 1 may be due to the steric hindrance of the neighbouring methyl groups, which would impede direct approach of acidic components to the sulphur atoms.

Durability of 1, 2, and DOS in various acid solutions. Commercial DOS and DHS
The FT-IR and 1 H NMR spectra of extractants 1, 2, and DOS treated with 12.0 M HCl did not change and exactly matched the spectra of their native forms. Thus, 1, 2, and DOS are not oxidised or otherwise degraded in HCl. Extractant 1 was very stable and displayed high oxidation resistance in HCl and HNO 3 , whereas DOS and 2 were stable only in HCl. The oxidation of DHS, DOS, and other sulphur-based extractants in acid media decreases the efficiency and selectivity of Pd(II) extraction 1,15,19,21 .  PGMs generally are leached from primary and secondary resources using HCl + Cl 2 or aqua regia 1, 36-39 . The acid leach liquors contain Cl − ions and have the potential to oxidise extractants over time. To examine the effect of such deterioration on 1, 2, and DOS, Pd(II) extractability from HCl + HNO 3 was measured at various intervals. Results are presented in Fig. 6. The Pd(II) E% of 1 was 99.9% from 1 h to 7 days. On the other hand, the E% of DOS and 2 equalled 99.9% from 1 h to 24 h, whereafter efficiency declined gradually with increasing contact time to 80.2% for 2 and 39.5% for DOS after 7 days. The oxidised product of DOS (DOSO) is sparingly soluble in kerosene, which may adversely affect extraction efficiency.
Okuda et al. 15 studied the extraction of Pd(II) and Pt(IV) from HCl and HNO 3 and reported the degradation of DHS to DHSO. Degradation of DHS in HCl increased slowly upon repeated operations, but DHS was oxidised immediately in HNO 3 . Presence of DHSO and DHS in the organic phase also induced formation of an insoluble third phase between the organic and aqueous phases in mixed-metal solution extractions. Sato et al. 41 reported that the Pd(II) extraction efficiency of DHS was greater than that of DHSO in HCl. Pd(II) is a soft acceptor and tends to bind more strongly to soft donor atoms. Oxidation of the "soft" sulphur donor(s) in DOS and 2 to "hard" sulphoxide group donor(s) in the degraded products inhibits Pd(II) extraction. A further disadvantage of sulphoxide and protonated sulphoxide groups in oxidised extractants is the uptake of undesired metal ions. Among the compounds studied, 1 is highly resistant to oxidation in acidic media and is suitable for long-term use in industrial applications.   Fig. 7(a). Extractants 1 and 2 removed 99.9% of the Pd(II) from the simulated solution, which demonstrates their obvious selectivity for this metal. Extraction of other metals was negligible. The E% of DOS was 82.2%, and minor amounts of Zr(IV), Fe(III), and other base metals were co-extracted 15 . Extractants 1 and 2 are therefore very suitable for extracting Pd(II) from leach liquors containing various platinum group, rare earth, and base metals.    given in Table 2. The extraction of Pd(II) was conducted by shaking 10 mL of the diluted leach liquors and 10 mL of kerosene containing 10 mM 1 or 2 for 60 min. The extraction results are illustrated in Fig. 7(b). Extractants 1 and 2 displayed selective and nearly quantitative removal of Pd(II) (E% = 99.9%). No other metals were extracted from the leach liquors by 1 and 2. Because of the ability of the pincer-like structure to capture metal ions via soft-base sulphur atoms, extractants 1 and 2 can selectively extract Pd(II) from leach liquors by intramolecular coordination 30 . The Pd(II) loading capacity of 1, 2, or DOS (10 mM) from leach liquors of automotive catalysts with O/A = 1 for 60 min were carried out and results are given in Fig. S10. The metal concentrations in the leach liquors are those referred in Table 2. The capacity of 1 and 2 was found to be ~1.057 g L −1 from the leach liquors. On the other hand, DOS loaded ~ 0.528 g L −1 of Pd(II). In fact, the Pd(II) loading capacity of 1 and 2 is 2-fold higher than that of DOS (cf. Fig. S9).

Back extraction of Pd(II) and reusability of extractants 1 and 2. A successful process includes efficient
back extraction to facilitate the reuse of extractants in subsequent operations. Following selective extraction of Pd(II) from the leach liquors of automotive catalysts with 1 and 2, back extraction of Pd(II) from the organic phase was carried out using five stripping agents: 1 M HCl, 1 M HNO 3 , 1 M H 2 SO 4 , 5% aqueous NH 3 , and a mixture of 0.1 M thiourea and 1 M HCl. Stripping was conducted with equal volumes of the Pd(II)-loaded organic phase and stripping solution for 60 min. The 1 M HCl, 1 M HNO 3 , 1 M H 2 SO 4 , and 5% aqueous NH 3 stripping solutions performed poorly due to strong coordination between the extractants and Pd(II). Our group 30 and other researchers 20,21 have reported acidic thiourea to be a powerful stripping agent for back extraction of Pd(II). A solution containing 0.1 M thiourea and 1 M HCl was used to achieve the optimum stripping percentage (S%) of Pd(II). Complete back extraction of Pd(II) from organic phases of 1 and 2 by thiourea + HCl was confirmed by S% values greater than 99.9%. After back extraction, 10 mL portions of the organic phase were washed with 20 mL of water and reused over four additional cycles of extraction and back extraction under the same experimental conditions. The efficiency of recycled 1 and 2 was > 99.6% after five extraction/stripping cycles. This reusability is demonstrated in Fig. 8. The colour changes of the kerosene phases of 1 and 2 during stripping experiments are shown in Fig. S11.

Pd(II) extraction mechanism of extractants 1-4. The extraction mechanism of ligands 1-4 was studied
using the corresponding extractant-Pd complexes. The complexes were prepared by extraction of 1.0 mM Pd(II) from HCl with 1.0 mM 1-4 in CHCl 3 . After extraction, the aqueous and organic phases were separated, and the CHCl 3 phase was evaporated to dryness to obtain the extractant-Pd complex. The FT-IR spectra of 1 and 2 and their Pd complexes are shown in Fig. S12. The FT-IR spectra of 1 and 2 change significantly upon complexation  of Pd(II). The C−S peak of 1 and 2 shifts from 703 to 727 cm −1 and from 708 to 720 cm −1 , respectively. The FT-IR spectra of 3 and 4 show similar changes after Pd(II) extraction (Fig. S13). The shifts of the C−S and C−H peaks to higher frequency and the appearance of new peaks suggest that Pd(II) extraction occurs by coordination to the sulphur atoms. The mechanism of Pd(II) extraction by 1-4 also was studied using UV-visible spectroscopy. UV-visible spectra of 1-4, an aqueous Pd(II) solution, and the extractant-Pd complexes are shown in Fig. S14.  1-Pd, 2-Pd, 3-Pd, and 4-Pd complexes differ following extraction. Ligand 1 possesses four methyl groups and produces a yellow-coloured product upon extraction. In contrast, extractant 2, which possesses no methyl groups, yields an orange product in the organic phase. The colour difference between 1 and 2 after Pd(II) extraction suggests a difference in extraction mechanism. A similar difference in colour upon extraction by 3 (with methyl groups) and 4 (without methyl groups) suggests a similar mechanistic difference.
The difference in Pd(II) extraction mechanism among the SCS pincer-type extractants was confirmed by 1 HNMR. Carbon numbering in the aromatic ring for extractant 1-4 was given in the Fig. S15. The 1 H NMR spectra of 1 and 1-Pd are shown in Fig. S16. After Pd(II) extraction, the aromatic C(2) proton peak is absent, and the integrated aromatic proton intensity is reduced by 1 H 30 . The C(4,6) proton peak shifts from 7.33 to 6.74 ppm, and the C(5) proton peak shifts from 7.26 to 7.03 ppm. The other proton peaks also shift. The results of the 1 HNMR analysis of 1 and 1-Pd are collected in Table S1. The 1 H NMR of 1-Pd shows that Pd ion is directly bonded to C (2) and that the C(2) hydrogen is removed by Pd complexation. Pd ion therefore bonds to the aromatic C(2) carbon and sulphur atoms of 1 to form a Pd-pincer complex upon extraction 30 . Palladation involving the benzene ring and sulphur moieties is promoted by formation of two fused five-membered chelate rings. The 1 H NMR spectra of 2 and 2-Pd are shown in Fig. S17. The 1 H-NMR spectrum of 2-Pd contains the aromatic proton peaks corresponding to 2 45 , which shift upon complexation by Pd(II). The results of the 1 HNMR analysis of 2 and 2-Pd are shown in Table S2. The 1 H NMR of 2-Pd indicates that Pd is not directly bonded to the C(2) carbon upon complexation and does not form an SCS Pd-pincer complex.
The 1 H NMR spectra of 3 and 3-Pd 46 are shown in Fig. S18. After Pd(II) extraction, the aromatic C(2) proton peak is absent, and the aromatic proton intensity is reduced by 1 H. The C(4,6) proton peak shifts from 7.39 to 6.79 ppm, the C(5) proton peak shifts from 7.28 to 7.09 ppm, and the other proton peaks also shift significantly. The results of the 1 HNMR analysis of 3 and 3-Pd are presented in Table S3. The 1 H NMR of 3-Pd indicates that Pd is bonded directly to C (2) and that the C(2) proton is removed upon extraction 46 . The 1 H NMR spectra of 4 and 4-Pd are shown in Fig. S19. The NMR spectrum of 4-Pd shows all aromatic proton peaks corresponding to 4 and shifts in all proton peaks upon Pd(II) complexation. The 1 H NMR results of 4 and 4-Pd are collected in Table S4. The 1 H NMR of 4-Pd indicates that Pd is not bonded to C(2) upon complexation and does not form an SCS Pd-pincer complex 46 . In extractants 1 and 3 46 , the steric hindrance generated by the four methyl groups at the benzylic positions facilitates aromatic C(sp 2 )-H bond activation and promotes Pd-C(2) bond formation. Pd complexation at the C(2) carbon of the benzene ring and the sulphur moieties occurs via formation of stable five-membered chelate rings. Yamashina et al. 46 . reported the same SCS pincer complex upon reaction of 3 (C 18 H 26 O 4 S 2 ) with [PdCl 2 (MeCN) 2 ] in CHCl 3 at room temperature. They reported the crystal structure of the 3-Pd complex (C 18 H 25 ClO 4 PdS 2 ), which showed palladium complexation to the C(2) carbon and sulphur atoms to form five-membered chelate rings. Extractant 4 did not form an SCS pincer-type complex, but rather a conventional 4-Pd complex (C 14 H 18 Cl 2 O 4 PdS 2 ), upon reaction of 4 (C 14 H 18 O 4 S 2 ) with [PdCl 2 (MeCN) 2 ] in MeCN at room temperature. We have observed the same pattern of complexation in this work during liquid-liquid extraction of Pd(II) with 1-4. Based on the 1 H NMR studies and literature reports, it is concluded that extractants 1 and 3 remove Pd(II) at room temperature by forming a pincer complex via activation of an aromatic C(sp 2 )-H bond. Extractants 2 and 4 are not capable of activating an aromatic C(sp 2 )-H bond at room temperature and do not form a pincer complex. Schematic representations of the Pd(II) extraction mechanisms involving 1-4 are presented in Fig. 9.
Extraction percentages (E%) were calculated to ±5%. Solvent extraction studies were conducted using extractants diluted to the desired concentration in kerosene (10 mL) and the desired concentration of metal ions in acid solution (10 mL). The aqueous and organic phases were shaken thoroughly using a mechanical shaker at a speed of 300 rpm for 60 min except for studies of contact time. The aqueous and organic phases were separated with a separatory funnel. Metal ion concentrations in the aqueous phase were determined by inductively coupled plasmaatomic emission spectrometry (ICP-AES). Metal concentration and E% were calculated using Equations (1)

Preparation of acid leach liquors of automotive catalysts and their extraction studies.
Automotive catalysts procured from commercial resources in Japan served as the secondary source of PGMs. Automotive catalysts pellets were milled and pre-treated by hydrogen reduction. Samples were finely ground and sieved to less than 500 μm. A simple and effective PGMs leaching process developed by our group was adopted for the present study 43   where [Pd(II)] aq is the Pd(II) concentration in the aqueous solution after stripping, and [Pd(II)] org is the concentration in the organic phase before stripping. After stripping Pd(II), 10 mL of metal-free organic phases of 1 and 2 were washed with 20 mL of water and reused for five successive extraction/stripping cycles of automotive catalyst leach liquors under the same conditions.

Conclusions
Pd(II) extraction behaviour of SCS pincer ligands 1-4 in kerosene was studied in HCl and HNO 3  Extractants 2 and 4 do not form a pincer complex and extract Pd(II) solely via sulphur coordination. 1 is stable in HCl and HNO 3 , possesses good Pd(II) extraction capability from these media, and can be synthesized inexpensively in high yield at room temperature. 1 is suitable for recovering Pd(II) from acid leach liquors containing various oxidising agents and can be used successfully in place of DOS for selective separation of Pd(II) in PGM refineries. The present study reports our development of the novel inexpensive SCS pincer extractant 1 and an improved approach for the recovery of Pd(II) in current refining processes.