Spacer-tuned diphosphine dioxides as preorganised co-ligands for synergistic lithium extraction with β-diketonates

Abstract

Synergistic liquid–liquid extraction (LLE) of Li⁺ commonly relies on β-diketonates in combination with neutral phosphine oxides such as tri-n-octylphosphine oxide (TOPO), yet the co-ligand is typically treated as an empirical additive rather than a design element. Here we show that commercially available diphosphine dioxides, featuring a preorganised PO⋯PO donor set and a spacer-defined bite, act as powerful co-ligands for Li⁺ extraction with 3-benzoyl-1,1,1-trifluoroacetone (HBTA) under mild pH conditions. Compared to TOPO, the best-performing PO⋯PO co-ligand enhances Li⁺ transfer while suppressing Na⁺/K⁺ co-extraction, consistent with altered solution speciation and stoichiometry. These results establish a chelating co-ligand (PO⋯PO) as a simple, modular strategy to control synergy in Li⁺ extraction systems based on classical CO⋯CO extractants such as HBTA.

---

Lithium has become a strategically vital element for the global economy because rechargeable lithium-ion batteries (LIBs) underpin consumer electronics, electric mobility, and grid-level storage.¹ The projected growth of lithium demand, driven largely by the expanding EV market, intensifies the need for scalable and selective lithium recovery from diverse resources.^2,3 Brine-based sources (salt-lake brines, geothermal brines, and oilfield brines) are particularly attractive due to their large reserves and comparable low operating costs,⁴ but lithium separation remains difficult because Li⁺ is present at low concentration and competes with the abundant alkali and alkaline-earth metal ions (Na⁺/K⁺/Mg²⁺/Ca²⁺).⁵

Among available separation technologies, liquid–liquid extraction (LLE) is appealing because it is operationally simple and compatible with continuous processing.⁶ Industrially relevant Li⁺ extraction typically relies on acidic extractants, including organophosphorus acids such as bis(2-ethylhexyl)phosphate (D2EHPA)^7,8 as well as β-diketones/acylpyrazolones and related O,O-chelators.^9–11 A key practical feature shared across many of these systems is that efficient Li⁺ transfer into the organic phase often requires neutral organophosphorus additives (e.g., phosphine oxides such as tri-n-octylphosphine oxide (TOPO)^12–15 or Cyanex 923),¹⁶ i.e. synergistic extraction. In D2EHPA-based systems, aggregation (including dimeric extractant motifs)¹⁷ and adduct formation with neutral donors are frequently invoked to rationalize performance enhancements, whereas for O,O-chelators the co-ligand can act as an additional donor that stabilizes charge-neutral Li-containing assemblies in the organic phase. Despite their widespread use, the molecular origin of synergism is still commonly treated empirically rather than as a tuneable structural parameter.

In this context, Scheme 1 summarises three conceptually distinct design strategies for synergistic Li⁺ extraction: (A) classical CO⋯CO extractants that rely on neutral P=O additives as empirical enhancers; (B) donor-set translation within the extractant itself (PO⋯CO), which embeds a phosphoryl donor as a built-in structural element; and (C) co-ligand bite engineering (PO⋯PO), where the extractant remains unchanged but synergism is tuned through a geometry-encoded co-ligand family.

Scheme 1 Three design strategies for synergistic Li⁺ extraction: (A) CO⋯CO extractants + neutral P=O additives, (B) PO⋯CO donor-set translation (our earlier work), and (C) this work: spacer-tuned PO⋯PO diphosphine dioxides for co-ligand bite engineering.

In our earlier work, we demonstrated that 4-phosphoryl pyrazolones introduce a deliberate donor-set translation from classical CO⋯CO chelation to PO⋯CO binding, enabling Li⁺ recognition and extraction under mild conditions through well-defined di- and trinuclear Li motifs in the presence of neutral co-ligands (TBP/TBPO/TOPO).^18,19 Building on this concept, we now ask whether synergism can be made even more “designable” by shifting the neutral component itself from a generic additive to a geometry-encoded co-ligand family: PO⋯PO donors with controlled PO⋯PO separation (e.g., CH₂ vs. CH₂CH₂ spacers). In this way, the co-ligand becomes a structural handle that can bias aggregation, stabilize specific Li stoichiometries, and ultimately tune extraction efficiency and selectivity. Guided by this concept, we selected two commercially accessible, spacer-defined diphosphine dioxides as minimal PO⋯PO co-ligands (1 and 2, n = 1, 2; Scheme 1C) to test whether co-ligand geometry alone can modulate Li⁺ speciation and, in turn, extraction performance when the acidic extractant is kept constant. Diphosphine dioxides of this type have been reported previously and used in transition-metal coordination chemistry,^20,21 yet their potential as preorganised O-donor co-ligands for Li⁺ recognition and synergistic LLE has not been explored. We therefore combined 1 and 2 with the benchmark β-diketones 3-benzoyl-1,1,1-trifluoroacetone (HBTA) and compared their performance against the widely used monodentate additive TOPO under mild pH conditions.

Receptors 1 and 2 were obtained in high yield (>95%) by oxidizing the respective bisphosphines with H₂O₂ following adapted literature procedures (SI).^20,21 Their ability to transfer Li⁺ into an organic phase was first probed by contacting CH₂Cl₂ solutions of diposphine oxides (with and without the β-diketones) with solid LiClO₄, followed by multinuclear NMR analysis of the supernatant (Fig. 1a; full data in Fig. S17–S21, SI). For 1, the ³¹P single resonance at δ = 23.6 ppm shifted downfield to δ = 30.7 ppm upon contact with LiClO₄. The absence of resolved splitting is consistent with a symmetric interaction of the two PO groups of 1 with Li⁺ on the NMR timescale. The addition of HBTA to the organic phase did not induce an additional marked ³¹P shift (Fig. 1a), suggesting that the immediate PO environment in the dominant Li-containing species remains similar under these conditions. In the corresponding ⁷Li NMR spectra, no Li⁺ signal was detected in CH₂Cl₂ after contacting with LiClO₄ alone, whereas addition of 1 enabled detection of a broad resonance at δ = 0.85 ppm (ν_1/2 = 3.6 Hz, Fig. S18, SI), confirming Li⁺ transfer into the organic phase. Addition of HTBA did not produce a meaningful change in the Li⁺ chemical shift within experimental resolution (Fig. S18, SI). In line with our earlier observations for related Li⁺/O-donor assemblies,¹⁹ this does not exclude subtle changes in Li⁺ speciation and any such effects are better assessed by stoichiometry and structural data (vide infra).

Fig. 1 Cutout of ³¹P{¹H} spectra of 1 in CH₂Cl₂ before and after contact with LiClO₄, and for mixture of HBTA and 1 after contact with LiClO₄ (a), molecular structures of [Li(BTA)(1)·H₂O] linked via reciprocal hydrogen bonds of the coordinated water (b) and a repeating unit of {[Li(BTA)(2)]·(H₂O)_0.75·(DCM)_0.09}_n (c) (all carbon hydrogen atoms and solvate molecules are omitted for clarity, thermal ellipsoids are displayed at 50% probability, hydrogen bonds displayed as dashed lines).

Comparable experiments with 2 likewise showed a downfield shift of the ³¹P resonance from δ = 30.9 to 33.3 ppm when contacted with LiClO₄ (Fig. S20, SI). Upon addition of HBTA, a further small upfield shift to δ = 31.5 ppm was observed, consistent with a modest change in the PO environment in the Li-containing species. In ⁷Li NMR, Li⁺ was detected in the organic phase only in the presence of 2, giving a broad resonance at δ = 0.74 ppm (ν_1/2 = 3.4 Hz, Fig. S21, SI). The addition of HBTA resulted in a very broad resonance at δ = 0.94 ppm (ν_1/2 = 22 Hz) consistent with a noticeable change in the Li environment in the presence of 2 and HBTA.

To identify plausible coordination motifs underlying the observed synergy, Li⁺ complexes were synthesised from HBTA, 1 or 2, and LiOH·H₂O in CH₂Cl₂ (SI). Single crystals suitable for X-ray analyses of [Li(BTA)(1)·H₂O] were obtained by slow diffusion of pentane into a saturated CH₂Cl₂, while crystals of {[Li(BTA)(2)]·(H₂O)_0.75·(DCM)_0.09}_n formed by slow evaporation of from toluene/CH₂Cl₂.

In both structures (Fig. 1b and c), Li⁺ is four-coordinate and bound by the O,O-chelate of deprotonated BTA⁻, forming the expected six-membered chelate ring. In [Li(BTA)(1)·H₂O], the coordination sphere is completed by one phosphoryl oxygen from 1 and one bound water molecule (Fig. 1b). The second PO group of 1 does not bind to Li⁺ directly, but engages in hydrogen bonding with the coordinated water; discrete mononuclear units are linked in the lattice via reciprocal hydrogen bonds. In contrast, 2 bridges between Li centres through both phosphoryl oxygen atoms, generating repeating [Li⋯OP–(CH₂)₂–PO⋯Li] connections and an extended 1D chain (Fig. 1c). The Li–O(P) bond lengths to the phosphine oxide donors span a comparatively wide range (1.874(3) and 1.937(3) Å), while the Li–O distances to the BTA-derived O atoms are 1.929(3) and 1.901(3) Å, consistent with an overall more distorted tetrahedral coordination environment in the polymeric motif. By comparison, in [Li(BTA)(1)·H₂O] the four Li–O contacts are much more uniform (Li–O1 1.914(3), Li–O2 1.915(3), Li–O3 1.919(2), Li–O5 1.910(2) Å), in line with a more regular local coordination sphere for the discrete complex.

The performance of 1 and 2 as PO⋯PO co-ligands was evaluated in LLE using HBTA as the acidic O,O-chelator, with TOPO included as benchmark (SI). Because practical brines and many real feed streams are acid to slightly basic, we selected pH 8.5 as a representative condition (Tris/HCl buffer). Under these conditions, control experiments confirmed that HBTA is required for Li⁺ transfer into the organic phase, consistent with its role as the charge-balancing anion in the extracted Li-containing species. With HBTA present, co-ligand identity has a decisive impact: 1 affords 58% Li⁺ extraction, substantially higher than 2 (26%) and TOPO (22%) under otherwise identical conditions (Fig. 2a). These results directly support the central premise of Scheme 1C: co-ligand geometry/bite is a strong handle to tune synergy, even when the acidic extractant (HBTA) is unchanged.

Fig. 2 (a) Percent extraction of Li(I) of the co-ligands 1, 2 and TOPO in the absence and presence of HBTA and (b) Li(I) extraction in dependence of the concentration of co ligand. Experimental conditions: [LiCl] = 0.01 M, [NH₄Cl] = 0.1 M, pH = 8.5 (Tris/HCl buffer), (a) [co-ligand] = 0.1 M, (b) [co-ligand] = 0.05–0.18 M, [HBTA] = 0.1 M in CHCl₃, 300 K, extraction time = 12 h.

To probe the composition of Li-containing species in the organic phase, slope analyses were performed (SI, eqn (S4)–(S6)), plotting log D–log[co-ligand]_(org) (Fig. 2b). For TOPO, a slope of ∼2 agrees with literature²² and is consistent with formation of [Li(BTA)(TOPO)₂] under the extraction conditions. In contrast, 1 gives a slope of ∼1, indicating a dominant 1 : 1 Li⁺ : co-ligand contribution, in line with the solid-state formulation [Li(BTA)(1)·H₂O] (noting that coordinated water may be replaced by solvent and/or additional donors in the organic phase). For 2, an intermediate slope of ∼1.5 suggests mixed stoichiometries (between predominant 1 : 1 and 1 : 2 Li⁺ : 2 contributions), consistent with the structural tendency of 2 to connect Li centres (chain propagation) while still allowing termination by additional donors in solution to form organo-soluble adducts.

To assess selectivity under competitive conditions relevant to brines, LLE experiments were conducted with equimolar Li⁺/Na⁺/K⁺ in the aqueous phase (Fig. 3a). The Li⁺ extraction efficiencies remain essentially unchanged (58% for 1, 26% for 2), while co-extraction of Na⁺ and K⁺ is very low: for 1 only trace transfer (<1%) is detected, whereas 2 shows slightly higher, but still minor, Na⁺/K⁺ extraction (ca. 3–4%). These data highlight that 1 combines high Li⁺ transfer with strong suppression of Na⁺/K⁺ co-extraction under the chosen conditions.

Fig. 3 (a) Percent extraction of Li(I), Na(I), and K(I) competitively extracted and (b) Reusability test for liquid–liquid extraction of Li(I) by 1 and 2 in presence of HBTA. Conditions for (a): [LiCl] = 0.01 M, [NaCl] = 0.01 M, [KCl] = 0.01 M, [NH₄Cl] = 0.1 M, pH = 8.5 (Tris/HCl buffer, 12 h and for b): [LiCl] = 0.01 M, [NH₄Cl] = 0.1 M, pH = 8.5 (Tris/HCl buffer), [1] = 0.1 M, [2] = 0.1 M [HBTA] = 0.1 M in CHCl₃, 300 K, overnight; for stripping: 0.5 M HCl, 300 K, 1 h.

Recyclability was evaluated over four extraction/stripping cycles. After Li⁺ loading at pH 8.5, stripping with 0.5 M HCl (V(aq) : V(org) = 1 : 1) releases Li⁺ quantitatively and regenerates the organic phase. Across four cycles, extraction efficiencies remain stable (ca. 55–56% for 1 and 24–27% for 2; Fig. 3b), and stripping is essentially complete in each cycle (Table S1, SI), demonstrating operational robustness.

In addition, selected follow-up experiments were conducted with 1 in the presence of HBTA. In experiments examining the effect of pH, a clear dependence was observed, as expected. While extraction is negligible at a pH of 6 and below, a steady increase to 69% extraction of Li⁺ at a pH of 9 is observed (Fig. S23, SI). Initial studies on the influence of Mg²⁺ and Ca²⁺ show that these ions are preferentially extracted (Table S2, SI).

In summary, we demonstrate that simple diphosphine dioxides 1 and 2 act as effective PO⋯PO co-ligands for synergistic Li⁺ extraction with the benchmark β-diketone HBTA under mild pH conditions. Solid-state structures reveal distinct coordination outcomes: 1 supports discrete Li-containing units, whereas 2 readily generates chain-like Li⋯O(P) connectivity. In LLE, these differences translate into markedly different performance: 1 provides substantially enhanced Li⁺ transfer relative to 2 and TOPO and strongly suppresses Na⁺/K⁺ co-extraction. Presumably, the shorter CH₂ spacer in 1 allows the formation of discrete Li⁺ complexes in solution, which leads to increased extraction. Slope analyses support altered stoichiometry/speciation relative to TOPO. Collectively, these results establish chelating co-ligand with spacer-defined PO⋯PO donors as a straightforward and modular strategy to tune synergism in Li⁺ extraction systems built on classical CO⋯CO-type acidic extractants. In further experiments the separation of Li⁺ from earth alkali metals such as Mg²⁺ and Ca²⁺ should be in focus to determine whether separation via pH regulation is possible, as reported in the literature.¹⁹

Conflicts of interest

There are no conflicts to declare.

Data availability

Supplementary information (SI): methods and procedures, detailed synthetic procedure, copy of NMR and additional characterisation data, detailed procedure and used equation for the LLE experiments; crystallographic details. See DOI: https://doi.org/10.1039/d6dt00738d.

CCDC 2538696 ([Li(BTA)(1)·H₂O]) and 2538695 ({[Li(BTA)(2)]·(H₂O)_0.75·(DCM)_0.09}_n) contain the supplementary crystallographic data for this paper.^23a,b

Acknowledgements

This work was financially supported by the China Scholarship Council (CSC No. 202204910352) and the German Federal Ministry for Economic Affairs and Climate Action (Swell 03ETE042C). The authors thank Prof. Michael Ruck (from TU Dresden) for access to the Rigaku Oxford Diffraction XtaLAB Synergy diffractometer.

References

1. L. Peiseler, V. Schenker, K. Schatzmann, S. Pfister, V. Wood and T. Schmidt, Nat. Commun., 2024, 15, 10301.
2. F. Maisel, C. Neef, F. Marscheider-Weidemann and N. F. Nissen, Resour., Conserv. Recycl., 2023, 192, 106920.
3. C. Xu, Q. Dai, L. Gaines, M. Hu, A. Tukker and B. Steubing, Commun. Mater., 2020, 1, 99.
4. H. Nikkhah, D. Ipekçi, W. Xiang, Z. Stoll, P. Xu, B. Li, J. R. McCutcheon and B. Beykal, Chem. Eng. J., 2024, 498, 155349.
5. M. L. Vera, W. R. Torres, C. I. Galli, A. Chagnes and V. Flexer, Nat. Rev. Earth Environ., 2023, 4, 149–165.
6. F. J. Alguacil and J. I. Robla, Desalin. Water Treat., 2022, 265, 71–93.
7. L.-Y. Yu, K.-J. Wu and C.-H. He, Hydrometallurgy, 2023, 220, 106102.
8. S. Sun, Y. Deng, J. Chen, R. Qin, Z. Du, Q. Wen and Y. Meng, Chem. Eng. Sci., 2026, 323, 123212.
9. Z. Li and K. Binnemans, AIChE J., 2020, 66, e16246.
10. T. Kanagasundaram, O. Murphy, M. N. Haji and J. J. Wilson, Coord. Chem. Rev., 2024, 509, 215727.
11. J. Zhang, M. Wenzel, L. Yang, R. C. Luckay and J. J. Weigand, Front. Chem. Eng., 2023, 5, 1271045.
12. H. M. Akbar, S. Habib, R. Akoumeh, E. Mahdi, M. Al-Ejji, A. Altaee and A. H. Hawari, Water, Air, Soil Pollut., 2025, 236, 723.
13. J. Jirattisak, A. Imyim and N. Bhawawet, Chemosphere, 2025, 385, 144556.
14. M. Xing, F. Zhang, P. Zheng, Y. Wang, L. Zhao and C. Xu, J. Cleaner Prod., 2024, 434, 139998.
15. J. Cao, Z. Meng, J. Zhang, Q. Meng, M. Wei, L. Wang, H. Wu, M. Wenzel, J. J. Weigand, H. Liu and L. Yang, ACS Sustainable Chem. Eng., 2025, 13, 17902–17912.
16. K. Ishimori, H. Imura and K. Ohashi, Anal. Chim. Acta, 2002, 454, 241–247.
17. Y. Song, Z. Zhao and L. He, Sep. Purif. Technol., 2020, 249, 117161.
18. J. Zhang, M. Wenzel, J. Steup, G. Schaper, F. Hennersdorf, H. Du, S. Zheng, L. F. Lindoy and J. J. Weigand, Chem. – Eur. J., 2022, 28, e202103640.
19. J. Zhang, N. Tanjedrew, M. Wenzel, P. Royla, H. Du, S. Kiatisevi, L. F. Lindoy and J. J. Weigand, Angew. Chem., Int. Ed., 2023, 62, e202216011.
20. F. Mani and M. Bacci, Inorg. Chim. Acta, 1972, 6, 487–490.
21. B. J. Brisdon, J. Chem. Soc., Dalton Trans., 1972, 2247–2250.
22. L. Zhang, L. Li, D. Shi, J. Li, X. Peng and F. Nie, Sep. Purif. Technol., 2017, 188, 167–173.
23. (a) CCDC 2538695: Experimental Crystal Structure Determination, 2026,  DOI:10.5517/ccdc.csd.cc2r6qd3; (b) CCDC 2538696: Experimental Crystal Structure Determination, 2026,  DOI:10.5517/ccdc.csd.cc2r6qf4.

---

Footnote

† Both authors contributed equally and share first authorship.

---