4‐Phosphoryl Pyrazolones for Highly Selective Lithium Separation from Alkali Metal Ions

Abstract Effective receptors for the separation of Li+ from a mixture with other alkali metal ions under mild conditions remains an important challenge that could benefit from new approaches. In this study, it is demonstrated that the 4‐phosphoryl pyrazolones, HL 2‐HL 4, in the presence of the typical industrial organophosphorus co‐ligands tributylphosphine oxide (TBPO), tributylphosphate (TBP) and trioctylphosphine oxide (TOPO), are able to selectively recognise and extract lithium ions from aqueous solution. Structural investigations in solution as well as in the solid state reveal the existence of a series of multinuclear Li+ complexes that include dimers (TBPO, TBP) as well as rarely observed trimers (TOPO) and represent the first clear evidence for the synergistic role of the co‐ligands in the extraction process. Our findings are supported by detailed NMR, MS and extraction studies. Liquid‐liquid extraction in the presence of TOPO revealed an unprecedented high Li+ extraction efficiency (78 %) for HL 4 compared to the use of the industrially employed acylpyrazolone HL 1 (15 %) and benzoyl‐1,1,1‐trifluoroacetone (52 %) extractants. In addition, a high selectivity for Li+ over Na+, K+ and Cs+ under mild conditions (pH ∼8.2) confirms that HL 2‐HL 4 represent a new class of ligands that are very effective extractants for use in lithium separation.


Introduction
Lithium has attracted extensive commercial interest in recent decades since it is indispensable for the manufacture of lithiumion batteries (LIBs) and as a component of many other commercial goods that include glasses, ceramics, lubricants, and pharmaceuticals. [1] In particular, LIBs are being extensively used to achieve a reduced dependency of private and other transport systems on fossil fuels to curb global warming by reducing CO 2 emission, resulting in a rapid growth in the consumption of the earth's lithium resources. [2] Some predictions forecast that the global lithium demand will not be able to be fulfilled by 2023 without recycling [3] which is particularly problematic at present since the global recovery rate of lithium currently does not exceed 1 %. [4] In order to achieve carbonneutral and sustainable development goals, exploring efficient strategies for the recognition and recovery of lithium from diverse sources, such as from spent LIBs, brackish brines and seawater, will clearly help to increase the available lithium resources and contribute to alleviating environmental pressures.
Over recent years, various techniques have been employed for lithium separation, these include liquid-liquid extraction (LLE), [5] solid-liquid extraction (SLE), [6] as well as adsorption, [7] membrane [8] and electrochemical processes. [9] Due to its generally large processing capacity, high selectivity, and extraction efficiency, LLE has often been used and is considered to be the most promising process, especially for the separation of lithium from a mixture of other monovalent ions (such as sodium, potassium and cesium ions). [10] However, the selective binding of lithium by specific organic receptors remains challenging and so far only a small variety of such ligand systems have been developed and applied in LLE and SLE (Scheme 1). [5a-d,6,11] Organophosphorus compounds such as di-(2-ethylhexyl) phosphoric acid (D2EHPA) and β-diketones like benzoyl-1,1,1trifluoroacetone (HBTA), 4,4,4-trifluoro-1-(2-thienyl)-1,3-butanedione (HTTA) or α-acetyl-m-dodecylacetophenone (LIX54) have typically been employed for the extraction of Li + ; however these ligand systems require the presence of a neutral, industrial co-ligand(s) such as trioctylphosphine oxide (TOPO), mixtures of trialkylphosphine oxides (Cyanex923) or tributyl phosphate (TBP) and the synergistic mechanism of these coligands is still unclear at the molecular level. [5d,12] Due to its high hydration energy, Li + tends to be less readily extracted than other alkali metal ions. [6b,13] Furthermore, very basic conditions (pH > 11) are generally required for extraction from aqueous solution. [5b,12c] Macrocyclic receptors, particular crown-4 derivatives are also selective for lithium ions and have been widely investigated by SLE (Scheme 1). [5a,6,11b,d,14] For these, cavity size plays a vital role in lithium ion recognition. A range of ion-pair receptors have also been developed in recent decades for metal salt recognition. [15] While certain ditopic receptors have been demonstrated to be capable of extracting lithium salts selectively under SLE conditions, both the loading capacity together with the lithium selectivity tends to be less than ideal largely due to the previously-mentioned high Li + hydration energy; this hinders their use as successful LLE process reagents. [5a,6b,11b] Consequently, the development of more effective ligands that can be employed in SLE or LLE processes for Li + under mild conditions, while challenging, remains an important goal.
Acylpyrazolones, derived from the β-diketone family, are widely used for the coordination of various metal ions due to their generally strong chelating ability. [16] Despite a considerable number of reports covering the coordination of such receptors Scheme 1. Previous work: a selection of ligands employed for the recognition and isolation of Li + ; this work: the 4-phosphoryl pyrazolone ligands and their observed binding motifs with Li + in the presence of solvent or organophosphorus co-ligands.
for d-and f-block elements, [17] investigations of the coordination of s-block elements (and especially alkali metals) are rare. [18] We found 4-phosphoryl pyrazolones can be promising alternatives to acylpyrazolones since they benefit from potentially higher functionality due to the presence of the conveniently adjustable (and versatile) phosphoryl moiety. Recently, we reported the successful employment of a 4-phosphoryl pyrazolone for coordination of a series of f-block elements. [19] Based on our findings, we anticipated that this ligand class would make very suitable and selective lithium-ion receptors.
Herein, we report the synthesis of a series of new 4phosphoryl pyrazolone ligands with an adaptable bite size between the chelating O-donor atoms, which is able to be controlled by variation of the steric demand of the substituents (Figure 1). We studied the coordination behavior of these new ligands towards Li + in the presence of different neutral organophosphorus co-ligands and also compared the behavior of 1-(5hydroxy-3-methyl-1-(4-nitrophenyl)-1H-pyrazol-4-yl)ethan-1-one (HL 1 ) towards this ion. X-ray single crystal structure analyses of the corresponding solid complexes were performed in order to provide a first insight into the coordination modes of the ligands and co-ligands towards the lithium cation. Detailed NMR studies were also employed to examine complex formation in solution. Further, a series of LLE and SLE experiments were also used to probe the extraction ability as well as the composition of the extracted species. Overall, our results show both a very high extraction efficiency towards Li + as well as high selectivity over Na + , K + , and Cs + making this ligand class very promising reagents for use in lithium separation processes.
Lithium complexes are known to adopt a wide range of structural motifs, both in solution and in the solid state, [27] with the structures often influenced by solvation effects. [27a,28] Single crystals suitable for X-ray investigation of the acetonitrilesolvated dimer complexes [Li 2 (L 2 ) 2 (CH 3 CN) 2 ] (4), [Li 2 (L 3 ) 2 (CH 3 CN) 2 ] (5), and [Li 2 (L 4 ) 2 (CH 3 CN) 2 ] (6) were obtained by slowly diffusing dry Et 2 O into a solution of the respective crude Li + complexes dissolved in CH 3 CN at À 30°C in the glove-box. [23] The structures obtained are depicted in Figure 2. In all three structures, each Li + is coordinated by three oxygen atoms from two ligands and a nitrogen donor from CH 3 CN to give each lithium center a distorted tetrahedral geometry. One oxygen donor from each pyrazolone unit bridges both Li + centers to yield a planar (LiO) 2 four-membered ring. Each six-membered (Li···OÀ PÀ CÀ CÀ O) chelate ring is slightly twisted and forms part of a step-ladder type arrangement (Figure 2e). The Li···O interactions involving the phosphoryl units are slightly shorter (1.870(4)-1.905(4) Å) than those to the pyrazolone oxygen atoms (1.935(2)-1.986(4) Å) ( Table 2). All Li···O contacts are in the typical range observed for other binuclear Li + complexes. [27b,29] In contrast, [Li(L 1 (7) is mononuclear when crystallized from EtOH, with two solvent molecules completing the tetrahedral coordination geometry of the lithium center ( Figure S62).

Chemistry-A European Journal
Since the bite-size of ligand [L 4 ] À in [TBA]L 4 is the smallest, we investigated the speciation chemistry of this salt following its reaction with LiClO 4 and co-ligand TOPO through extensive NMR ( 31 P, 1 H and 7 Li) studies. The latter included several concentration-dependent procedures such as the continuous variation, [31] mole ratio [32] and slope ratio [33] methods in order to shed light on the lithium complex composition and its dynamic behavior in solution. The continuous variation method turned out to be not suitable for this system due to the required large excess of TOPO and the resulting generation of a number of species ( Figure S64). [23] The mole ratio method proved much more suitable and thus was used instead. In a typical experiment, we gradually increased the concentration of TOPO in steps of 0.1 equiv. in a 7.2 mM solution of LiClO 4 /[TBA]L 4 in CD 2 Cl 2 at 300 K (Table S2) and recorded the corresponding 31 P, 1 H and 7 Li NMR spectra (Figure 5a-d; the 31 P and 7 Li chemical shifts are summarized in Table S3). [23] For reference, Figure 5a (bottom) shows the respective singlet resonances of the mixture of LiClO 4 /[TBA]L 4 (δ = 20.2 ppm) and co-ligand TOPO (δ = 46.7 ppm) dissolved in CD 2 Cl 2 . We assume that the 1 : 1 LiClO 4 /[TBA]L 4 mixture consists of the dimeric Li-complex [Li 2 (L 4 ) 2 (solv) 2 ], similar to the investigated complexes 4-6. Upon addition of TOPO, the resonances attributed to coordinated TOPO, that are expected to be observed between δ = 50-60 ppm, are extremely broad until the proportion of the ligand reaches 0.4 equiv., indicating a dynamic exchange process. Interestingly, the resonance that is attributed to the coordi-nated ligand [L 4 ] À (δ = 20.2 ppm) upon addition of 0.1 equiv. TOPO is slightly displaced by a high-field shifted shoulder indicating the formation of a new complex species. The latter becomes the dominant species when the proportion of the coligand reaches~0.3 equiv. (δ = 19.3 ppm; ν 1/2 = 112 Hz). With increasing concentration of TOPO, this resonance shifts back slightly to lower-field with a much sharper resonance at δ = 19.5 ppm (ν 1/2 = 51 Hz). The same sharpening is observed for the resonance attributed to the co-ligand (δ = 52.1 ppm; ν 1/2 = 154 Hz) until 1.0 equiv. is reached. The change in resonance in the 31 P NMR spectra for [L 4 ] À upon addition of TOPO is consistent with the occurrence of different species in solution. A deconvolution analysis of these phosphorus signals was performed using the deconvolution function of the commercially available topspin software designed primarily to analyze high-resolution small molecule NMR data (Figure 5c). [23,34] The observed changes of these overlapping resonances were analyzed for the 0.1 equiv. and 0.2 equiv. TOPO cases and showed that for 0.1 equiv. TOPO an integration ratio of 7 : 3 (Table S4) was present. This indicates that the new resonance involves 0.3 equiv. ligand as well as 0.1 equiv. TOPO. Upon the addition of 0.2 equiv. TOPO, this ratio increased to 4 : 6 (Table S5), consistent with the involvement of 0.6 equiv. of ligand and 0.2 equiv. of co-ligand TOPO. As already discussed, for a proportion of the co-ligand of~0.3 equiv. the corresponding singlet resonance occurs between δ = 19.1-19.3 ppm which is consistent with the Li + : [L 4 ] À : [TOPO] = 3 : 3 : 1 complex, [Li 3 (L 4 ) 3 (TOPO)] (13; δ = 19.2 ppm), occurring to represent the first rearrangement step from [Li 2 (L 4 ) 2 (solv) 2 ] as illustrated in Figure 5c. For comparison, a cut-out of the 31 P NMR spectra corresponding to 0.1 to 0.4 equiv. TOPO as well as for the trinuclear complex 13 in purple and, for reference, [Li 2 (L 4 ) 2 (TBPO) 2 ] (10) in orange are also shown (Figure 5b). The latter complex differs only in the alkyl chain-length of the TBPO (butyl) versus TOPO (octyl) and thus, the chemical shifts for both dimers are comparable. Therefore, we assume that with  Figure 5d, with the full spectra presented in Figure S65 as well as by the 7 Li NMR spectra (Figure 6a). The respective resonances are significantly broadened preventing a similar deconvolution analysis to that employed for the 31 P NMR spectra; however, the same shift trends are observed and are comparable to those of the trinuclear complex 13 (referenced to [Li 2   ( Figure 6b). In addition, the interconversion between these complexes is reversible as shown by the respective 31 P NMR spectra following the alternating stoichiometric additions of LiClO 4 /[TBA]L 4 and TOPO and, in particular, the observed resonance shifts that are attributed to the presence of the binuclear (δ = 19.5 ppm; 2 : 2 : 2) and trinuclear (δ = 19.3 ppm; 3 : 3 : 1) species in solution (Figure 6c).
The speciation behavior of the Li + complexes in solution was further investigated by obtaining the electrospray ionization mass spectra (ESI-MS) [35] of 12 and 13 dissolved in both neat methanol and in LiCl/methanol solution (Figures 7 and  S66). [23] The attributed peaks were assigned according to their simulated isotope patterns, as well as from recorded daughter experiments (Table 3, Figures S67-S70). For both complexes comparable peaks were observed, which can be attributed to the presence of similar complex species (including trinuclear species).
In subsequent experiments, the ESI-MS spectra of 13 in the presence of a varying excess of TOPO were recorded (Figure 7). [23] With increasing TOPO concentration (5-20 equiv.), a significant decrease in the relative intensity of the m/z = 1346.4 peak, attributed to the species [Li 2 · L 4 3 ] À , most likely originated form the trinuclear species [Li 3 · L 4 3 ], was observed, whereas the relative intensity of the peak at m/z = 895.2, attributed to [Li · L 4 2 ] À , remains the dominant peak. This observation indicates that an excess of TOPO facilitates the conversion of the trinuclear species into other species, which is in line with the proposed behavior observed in the NMR experiments.
To investigate the ability of HL 2-4 to extract Li + , LLE experiments were performed in the presence and absence of TOPO (Figures 8 and S71) at pH 8.2. For comparison, the acylpyrazolone ligand HL 1 as well as the widely employed HBTA reagent were also included in these studies. The Li + extraction for all ligands in the absence of TOPO was essentially negligible (< 5%, Figure S71). In contrast, in the presence of TOPO, Li + extraction of 78 % was observed for HL 4 followed by 70 % for HL 3 and 61 % for HL 2 . These three 4-phosphoryl pyrazolone ligands clearly outperform the acylpyrazolone ligand HL 1 (15 % extraction) and HBTA (52 % extraction) under similar conditions. As mentioned above, the slope ratio method is also an important strategy for use in solution studies. Herein, slope analyses of the respective extraction data using logD-logc L(org) diagrams were performed ( Figure S72) in order to investigate the composition of the extracted species corresponding to HL 3 , HL 4 and HBTA in the organic phases. [33] For this, extraction experiments that involved variation of the ligand and co-ligand in excess TOPO were performed using aqueous solutions of 0.01 M LiCl (see Eqs. (S3)-(S5) in Section S9). [23] For HBTA a  [a] n = 3 in 12 and n = 4 in 13. In order to probe if a single complex is formed in the organic phase under 'saturation' conditions (corresponding to a constant co-ligand concentration) in the presence of excess metal, a second set of LLE experiments was performed. [36] These studies can reveal the composition of the extracted species when the co-ligands' presence is in shortfall under LLE conditions. The experimental data are shown in Figures S73 and  S74 in Section S10 and the results are summarized in Table 4.

Chemistry-A European Journal
When TOPO was employed as co-ligand, from the extraction data an approximate [TOPO] : [metal + ligand] ratio of 1 : 3 was determined, indicating that both ligands can extract Li + by the formation of a complex with 0.33 equiv. TOPO. This result is consistent with the stoichiometry present in the corresponding X-ray structure (3 : 3 : 1 (Li + : [L n ] À : TOPO)) and that obtained by the NMR mole ratio method in the presence of a shortfall of TOPO. In contrast, the studies employing TBPO as co-ligand reveal inflection points at approximately 1 (Table 4), in keeping with a stoichiometry of the extracted species of 1 : 1 : 1 (Li + : [L n ] À : TBPO), most likely of type [Li 2 (L n ) 2 (TBPO) 2 ], which is consistent with the ratio observed in the corresponding X-ray structures. For HBTA, an inflection at 1.88 was obtained ( Figure S73c, Table 4), indicating a stoichiometry for the extracted species of 1 : 1 : 2 (Li + : [BTA] À : TOPO), which parallels the result of the extraction studies as well as for the reported behavior of HTTA in Li + extraction. [12b] UV-vis and 31 P NMR studies were performed in the presence of [TBA]L 4 in order to probe the potential selectivity of [L 4 ] À for Li + over Na + , K + and Cs + .
[TBA]L 4 was used to avoid the presence (and detection) of simple deprotonation processes during complex formation. For the UV-vis studies, 5 mM [TBA]L 4 was mixed with equimolar quantities of LiCl, NaCl, KCl, and CsCl in MeCN for 4 h. Images of the resulting solutions and the corresponding UV-vis spectra are depicted in Figure 9. Notably, only the LiCl-containing system exhibited a color change (from red to pale yellow) while for the NaCl, KCl and CsCl systems no color change was evident. These naked eye observations correlate well with the respective spectra, with the spectra both in the absence and presence of NaCl, KCl or CsCl being very similar while that of [TBA]L 4 plus LiCl differs. In this latter case the spectrum shows the absence of the prominent shoulder at 413 nm observed in the spectrum of [TBA]L 4 . There is also a bathochromic shift of 27 nm from the absorption maximum for [TBA]L 4 at 347 nm to 374 nm in the LiCl case. In contrast, only minor spectral changes are evident for the NaCl, KCl and CsCl  systems, in accordance with the preferred recognition of Li + by the [L 4 ] À anion. The above selective Li + binding was further investigated by NMR studies using an equimolar mixture of [TBA]L 4 and TOPO (3 mM each) in CD 3 CN. This solution was subsequently treated with an equimolar amount of CsCl, with the procedure repeated with KCl, NaCl and LiCl substituted for CsCl. After each addition, the reaction mixture was stirred for 18 h in an inert atmosphere. The 31 P NMR spectra in CD 3 CN of the respective mixtures are shown in Figure 10. Upon the addition of CsCl and KCl a slight downfield shift for the [L 4 ] À resonance was observed (Figure 10), whereas practically no shifts of the 31 P resonance of TOPO were evident. After the addition of NaCl, the 31 P resonance of [L 4 ] À shifts by approximately 0.4 ppm. However, a significantly increased downfield shift of the 31 P resonance of [L 4 ] À (1.3 ppm) was observed when LiCl was added, which points to a clear selectivity for Li + over the other alkali metal ions. Greater differences occur for the 31 P resonance of TOPO. On addition of LiCl to the solution, a downfield shift of 3.4 ppm was observed, and the peak broadens. This observation is in accord with the phosphine oxide moiety of TOPO also participating in the coordination of the Li + . The NMR selectivity studies under LLE conditions also confirm a clear selectivity for Li + ( Figure S75). In an extension of the above studies, competitive SLE experiments in the presence of 50 times excess of the alkali chloride salts were carried out employing [TBA]L 4 and a mixture of [TBA]L 4 and TOPO in CHCl 3 . [23] The loadings for all four metal ions in both experiments are depicted in Figure 11a. A loading of 79 % for Li + was obtained while Na + , K + , and Cs + are effectively not loaded (all < 0.1 %). When TOPO is present, the  loading percentage for Li + was observed to marginally increase to 83 % (but still lies within experimental error of 79 %) while the loading of Na + , K + and Cs + remains negligible (< 0.1 %). These results again demonstrate the high selectivity of [TBA]L 4 for Li + over the other alkali metal ions under the SLE conditions employed. Moreover, at best, only a marginal synergistic effect is observed in the presence of the co-ligand TOPO. To investigate whether the 4-phosphoryl pyrazolone ligands are also capable of extracting Li + under competitive conditions, LLE experiments in the presence of 10-fold excess of Na + , K + and Cs + were carried out using HL 3 and HL 4 (Figure 11b). The results show close to undifferentiated (high) Li + extraction for both ligands (72 % for HL 3 and 77 % for HL 4 ), whereas any extraction of the competing metal ions is very limited. For Cs + an extraction of 5 % by HL 3 and 6 % by HL 4 was observed, while Na + and K + were extracted in trace amounts (< 1%). These results demonstrate the impressive high affinity and selectivity of these ligand systems towards Li + , which are clearly illustrated from the extraction process under LLE conditions given in Figure 11c. The bite size between two chelating O-donor of these ligands may provide a suitable fit for the coordination with Li + which make them remarkably selective for the capture of this ion. The formation of complexes involving the co-ligand TOPO results in the capability to extract Li + into the organic phase. It should be noted that there are only a few synthetic receptors known that can extract Li + from water -a situation arising from the previously mentioned high enthalpy of hydration of this ion, [5a,6b,11b] and the often need for very basic conditions (pH > 11). [5b,12c]

Conclusion
Despite a wealth of previous reports on reagents for lithium ion coordination and extraction in recent decades, the development of more effective reagents for use under mild conditions has remained challenging. This study demonstrates that the present 4-phosphoryl pyrazolones represent a new class of ligands that are excellent reagents for the recognition and extraction of lithium ions in the presence of excess sodium, potassium and cesium ions. Single crystal X-ray analyses confirm the formation of binuclear Li + complexes in the solid state in which co-coordination of MeCN, TBPO, or TBP occurs to yield a 2 : 2 : 2 (Li + : [L n ] À : solvent/co-ligand) stoichiometry in each case. In contrast, in the presence of TOPO, unusual trinuclear complexes are formed in which the three metal centers are bridged by the O-donor atom of TOPO to yield a tetrahedral [Li 3 O]-core. Detailed NMR and MS studies reveal the presence of these polynuclear species in solution. Their dynamic aggregation and reversible conversion in two steps were clearly monitored by multinuclear NMR spectroscopy. Detailed slope analysis and loading experiments under LLE conditions further confirmed the corresponding active species dependence of coligands. It is worth noting that the overall explorations provide a good example of the use of solid and solution studies for understanding the coordination mechanism in mixed-ligand systems. We propose that the present study provides a power-ful basis, not only for the development of further simple lithium-selective receptor systems that display high extraction efficiency at moderately basic pH values, but also for related dand f-block metal complexation/extraction studies -including for achieving rare earth separation. The ready functionalization of the present (new) class of N,O,P-donor phosphoryl pyrazolone ligands will clearly also facilitate the successful exploitation of these opportunities. Studies of the above type are planned for the near future.

Experimental Section
General experimental procedures for the synthesis of all compounds, characterization, liquid-liquid extraction, solid-liquid extraction and X-ray crystallography are described in the Supporting Information.
Deposition Number 2101594 (for HL 2 ), 2101595 (for HL 3  contain the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service.