Siderophore-inspired chelator hijacks uranium from aqueous medium

Over millennia, nature has evolved an ability to selectively recognize and sequester specific metal ions by employing a wide variety of supramolecular chelators. Iron-specific molecular carriers—siderophores—are noteworthy for their structural elegance, while exhibiting some of the strongest and most selective binding towards a specific metal ion. Development of simple uranyl (UO22+) recognition motifs possessing siderophore-like selectivity, however, presents a challenge. Herein we report a comprehensive theoretical, crystallographic and spectroscopic studies on the UO22+ binding with a non-toxic siderophore-inspired chelator, 2,6-bis[hydroxy(methyl)amino]-4-morpholino-1,3,5-triazine (H2BHT). The optimal pKa values and structural preorganization endow H2BHT with one of the highest uranyl binding affinity and selectivity among molecular chelators. The results of small-molecule standards are validated by a proof-of-principle development of the H2BHT-functionalized polymeric adsorbent material that affords high uranium uptake capacity even in the presence of competing vanadium (V) ions in aqueous medium.

A chieving a balance between responsible and conscientious environmental stewardship of available natural resources and the ever-increasing demand for raw materials required to sustain higher standards of living is the quintessential requirement of sustainable development. Unlike terrestrial sources of minerals (mines, salt lakes etc.), seawater is almost an inexhaustible resource of base elements such as Na and Cl. Unsurprisingly, less-abundant elements present in the seawater, such as Mg, K, Ca, continue to be mined from oceans on an industrial scale at a significantly lower cost to the environment than alternative land-mining operations 1 . Numerous strategies have been proposed to recover even less-abundant high value elements from seawater, such as copper (Cu), nickel (Ni), vanadium (V), and uranium (U), however, their implementation remains elusive 2 . Uranium extraction from seawater especially has attracted a significant amount of attention from researchers in recent years, because of its predominant use as a fuel to produce nuclear power 3 . Despite its low dissolved concentration of only~3 mg per ton of seawater, an estimated total content of uranium amounts to 4 billion tons, over 1000 times more than is available in all known terrestrial sources 4 . This almost inexhaustible supply of accessible uranium could fuel nuclear power generation for millennia even with the demand significantly increasing over time 5 .
The pace of the development and commercialization of new adsorbent materials capable of cost-efficient and selective recovery of uranium from seawater has been slow, despite numerous attempts based on layered inorganic materials 6,7 , engineered proteins 8,9 , chelating resins [10][11][12] , and polymeric adsorbent materials [13][14][15] that are currently in the development. The most successful adsorbents based on amidoxime and imide-dioxime (H 3 IDO) functional groups (Fig. 1a) have shown some of the highest uranium adsorption capacities in excess of 4 g per gram of adsorbent 13 . Functional scaffolds based on amidoximes and imide-dioxime have emerged as widely used molecular receptors for uranyl ion (UO 2 2+ ) binding [16][17][18][19] , as these are the functional groups predominantly formed in hydroxylamine-treated nitrile-based adsorbents, and particularly in polyacrylonitrile adsorbents. We and others have previously described the performance of adsorbents based on hydrophilic surface-grafted polyacrylonitrile 4,14,15,20,21 . However, the selectivity of these and other structurally-related materials still suffers from a much higher affinity towards vanadium (V) ions present in the seawater over uranium (VI), therefore requiring use of expensive reagents under harsh conditions for adsorbent regeneration and limiting their real-life implementation 22 . The development of more selective adsorbents containing superior functionalities is still required to overcome this bottleneck, especially materials that take advantage of environment friendly and non-toxic functionalities 23 .
Siderophores, a class of naturally occurring (Fig. 1b) nitrogenand oxygen-based donor group containing chelating compounds for iron sequestration present in bacteria and fungi that exhibit some of the highest binding affinities towards the metal, have long been recognized as potential ligands for f-block elements [24][25][26] . Artificial siderophores show record affinities towards iron, however, only a handful of f-element complexes have been characterized 27 . There have been speculations that iron in its +3 oxidation state (Fe 3+ ) and uranyl ion (UO 2 2+ ) interact very similarly with hydroxamate-based ligands, and therefore could be used as a proxy for one another 28 . Inspired by this hypothesis, as well as the desire to increase the selectivity of the adsorbents towards uranium, especially over vanadium, we have selected non-toxic and easy to prepare 2,6-bis[hydroxy(methyl)amino]-4morpholino-1,3,5-triazine (H 2 BHT) 29,30 -an artificial siderophore-as a ligand of choice (Fig. 1c).
Herein we report a comprehensive computational and experimental study on uranium binding by H 2 BHT in solution and in solid state, starting from small-molecule investigations and ending up with a developed polymeric adsorbent material. Quantum chemical calculations corroborated by potentiometric titrations establish exceptionally high uranyl binding affinity and selectivity of the H 2 BHT, making it one of the most compelling and unique functional units for uranium extraction and remediation technologies to date. We provide synthetic details and evidence that H 2 BHT-functionalized polymers have high uranium adsorption capacity even in the presence of competing vanadium ions, presenting a novel approach towards the rational design and synthesis of economically attractive tailor-made adsorbents for selective metal ion recognition and sequestration from aqueous media.

Results
Crystallographic and quantum chemical calculation studies. In order to find the representative binding mode between H 2 BHT chelator 31 and uranium, the 2,6-bis[hydroxy(methyl)amino]-4morpholino-1,3,5-triazine ligand was synthesized and tested for complexation with uranyl (Supplementary Methods). Single crystals of UO 2 (BHT) (Fig. 2a) were readily obtained from the stoichiometric aqueous solution of uranyl nitrate and H 2 BHT on standing over the period of three weeks, however, all attempts to isolate crystals of UO 2 (BHT) 2 2− suitable for X-ray diffraction studies were unsuccessful 32 . As seen from Fig. 2a, H 2 BHT coordinates to uranium in a tridentate fashion, similar to its coordination with other metals, however, the density functional theory (DFT) calculations at the M06/SSC/6-311 + + G** level of theory indicate that binding modes can vary significantly depending on protonation states of the ligand ( Supplementary Fig. 1). In agreement with the crystallographic data (Supplementary Tables 1, 2), the DFT optimized UO 2 (BHT) complex represents a neutral five-coordinate uranyl species (one H 2 BHT and two water molecules in the first coordination sphere) with relatively short Having established the most stable structures of the uranyl complexes in solution, we proceeded to calculate the key thermodynamic parameters, which then were used in our computational protocol 34,35 for predicting stability constants (log β theor ). This procedure enables us to carry out in silico prediction of log β for uranyl complexes with the H 2 BHT ligand and compare the obtained values with those of uranyl systems with the imide-dioxime (H 3 IDO) functionality, which is reputedly responsible for the extraction of uranium from seawater using the current generation of amidoxime-derived sorbents 4 . The computational results (log β theor ) in Table 1 indicate very similar binding strengths of H 2 BHT and H 3 IDO. On one hand, the electron-withdrawing effect of the aromatic triazine ring in H 2 BHT lowers the electron density and so the basicity of its hydroxylamine groups, which would result in a weaker complexation ability of H 2 BHT with UO 2 2+ than that of H 3 IDO. On the other hand, this electron-withdrawing effect can be offset by the presence of electron-donating methyl groups at the nitrogen atoms in H 2 BHT, explaining comparable uranyl binding affinities of the triazine hydroxylamine and imide-dioxime ligands. Natural bond orbital (NBO) analysis further confirms strong uranyl binding by both functionalities, revealing effective metal-ligand interactions via dative σ-bonds for the neutral 1:1 H 2 BHT and H 3 IDO uranyl complexes (Supplementary Table 3). The protonated UO 2 (HBHT) + , UO 2 (HBHT)(BHT) − , and UO 2 (HBHT) 2 complexes are generally 3-4 orders of magnitude weaker than the respective uranyl complexes formed with imide-dioxime. This trend can be rationalized by the structural differences between H 2 BHT and H 3 IDO (Fig. 1a, c). While in the imide-dioxime complexes the protons of both oxime groups can undergo transfer from    2 45. the oxygen to nitrogen atom to increase the binding strength with uranyl, a similar tautomeric rearrangement in H 2 BHT is prevented by methyl groups.
Potentiometric titrations and NMR spectroscopic studies. The H 2 BHT ligand protonation constants (pK a ) were determined in a 0.5 M NaCl ionic medium (Supplementary Table 4), and are in excellent agreement with previously reported values in other media 33 . It is worth noting that H 2 BHT is more acidic than imide-dioxime, with pK a1 = 8.0 ± 0.3, indicating that in oceanic environment with pH 8.0-8.3 approximately half the ligand is already in a mono-deprotonated state that can facilitate metal ion complexation 23,36,37 . Potentiometric titrations were performed to verify the presence of theoretically predicted complexes in the uranyl-H 2 BHT aqueous systems and the values of their respective equilibrium constants log β expt . The modeled speciation diagram at a 1:2 ratio as a function of pH is shown in Fig. 2b and a representative titration curve is provided in Fig. 2c, while the stability constants are summarized in Table 1. Our results in Fig. 2b Table 5). The simulated speciation diagrams in Supplementary Fig. 2 indicate that there is no V(BHT) 2 + complex formation over the entire pH range even in the presence of a large excess of the ligand. It is worth noting that in addition to vanadium, iron ions (Fe 3+ ) form strong complexes with imide-dioxime, exhibiting high stability constants, e.g., log β (Fe(HIDO) 2 − ) = 36.0 39 , which is 8.5 orders of magnitude higher than log β for the corresponding UO 2 (HIDO) 2  To shed an additional light on the solution-state uranyl-ligand complexation behavior, NMR experiments were performed to corroborate the results of potentiometric titrations and computations pertaining to solution speciation in the H 2 BHT-uranyl system ( Supplementary Fig. 3). In aqueous solution, the UO 2 (BHT) complex is the predominant species formed under acidic to neutral conditions (Fig. 2b). However, once crystallized from solution as UO 2 (BHT) (crystal structure), the complex becomes insoluble in neutral or acidic water, indicative of a neutral, low-polarity species. Surprisingly, upon addition of base (3 equivalents, final pH = 12) to the UO 2 (BHT) complex in water, it re-dissolves. We propose that the complex is deprotonated in solution forming the anionic complex UO 2 (BHT)(OH) − (Table 1). This monoanionic species is not present in speciation models from the potentiometric titrations. However, the NMR experiments were only performed with a 1:2 or larger uranyl to ligand ratio. When a formation constant of log β expt = 7.0 for UO 2 (BHT)(OH) − is added to the model, the complex is not observed at 1:2 or higher uranyl to ligand ratios, however, at a 1:1 ratio UO 2 (BHT)(OH) − appears at pH = 9 and is the only species present in the solution at pH = 12. Further addition of the ligand results in the solution becoming paler in color, and the NMR spectra confirm the formation of new species, that we assign as UO 2 (BHT) 2 2− , with excess amount of free ligand and very small amounts of the UO 2 (BHT) complex still present. The 13 C spectra is consistent with the formation of UO 2 (BHT) 2 2− species in solution ( Supplementary Fig. 4), with 2D 1 H-13 C HSQC spectrum further corroborating this assignment ( Supplementary Fig. 5). The results of potentiometric titrations and NMR spectroscopy were further supplemented by UV-Visible absorption studies (Supplementary Fig. 6 and Supplementary Note 2), providing additional confirmation of the observed speciation in solution over a wide pH range.
Adsorbent material synthesis and uranium sorption studies. As with other hydroxylamine-derived ligands, the direct incorporation of such species via free radical or cationic polymerization is not feasible due to radical quenching and strong nucleophilicity of the hydroxylamine group, the presence of relatively acidic protons precludes the use of anionic polymerization. Therefore, an alternative approach based on a stepwise construction of H 2 BHT core via post-synthetic functionalization has been developed instead 41 . The synthesis of H 2 BHT-decorated polymeric adsorbent was commenced with the polyethylene-graftpolyacrylic acid hollow gear fibers prepared employing our previously developed radiation induced graft polymerization (RIGP) technique ( Supplementary Fig. 7) 42 . The activation of carboxylic group with thionyl chloride under mild conditions provided acyl chloride-functionalized polymeric material. Subsequent coupling reaction with mono N-tert-butyl carbonyl (BOC) protected piperazine derivative resulted in the formation of the monoamide-monocarbamate intermediate. Next, the deprotection of the N-BOC substituted amine by treating with excess trifluoroacetic acid resulted in the formation of the corresponding ammonium trifluoroacetate, which upon addition of excess amounts of N,N-diisopropylethylamine (DIEA) and 1,3,5-trichlorotriazine afforded the aminodichlorotriazine advanced intermediate 43 . The final reaction of the polymer-bound aminodichlorotriazine with an excess amount of N-methyl hydroxylamine free base in tetrahydrofuran (prepared in situ by treating corresponding hydrochloride salt with DIEA), provided the H 2 BHT-functionalized polymeric adsorbent in five linear steps, constituting the first proof-of-principle synthesis of an artificial siderophore-embedded uranium adsorbent. The scanning electron microscopy (SEM) images reveal similar polymer morphologies for the pristine polyethylene-graft-polyacrylic acid (Fig. 3a) and the H 2 BHT-modified polymer (Fig. 3b), indicating that the initial structure of the fiber was not damaged during the synthesis process. In addition, the structure of the H 2 BHTpolymer was confirmed by Fourier-transform infrared spectroscopy (FT-IR) ( Supplementary Fig. 8), the incorporation of H 2 BHT ligand into the polymer structure was further corroborated by 13 C cross polarization/magic-angle spinning (CP/MAS) NMR spectroscopy (Supplementary Fig. 9). The overlay of spectra corresponding to the H 2 BHT molecule and the polymer, depicted in Fig. 3c, clearly shows the presence of the three triazine carbon atoms (A, B, and C) in the polymer. The signals from piperazine carbons F and F′, as well as the signals from N-methyl carbons D overlap with signals from the carbons present in the polyethylene and polyacrylamide backbone, falling in the same 10-60 ppm region, consistent with solution-state NMR of the ligand 29 .
With the H 2 BHT polymer adsorbent in hand, we investigated the uranium (VI) binding by this material using a variety of spectroscopic techniques, including X-ray photoelectron spectroscopy (XPS) and elemental distribution mapping via energydispersive X-ray (EDX) spectroscopy analysis. Uranium inclusion within the adsorbent is evident from the characteristic U 4 f signals in the XPS spectra of the uranyl reacted polymer sample ( Supplementary Fig. 10). The EDX elemental mapping results ( Supplementary Fig. 11) confirm the existence of uranium species on the surface of the H 2 BHT functionalized polymer with a homogeneous distribution of the extracted uranium throughout the material. Furthermore, FT-IR spectrum of the uranyl contacted sample (Supplementary Fig. 12) exhibits a significant red-shifted peak that corresponds to the antisymmetric stretch vibrational mode of [O = U = O] 2+ , indicating strong interactions between uranyl and the H 2 BHT functional groups in the developed polymeric adsorbent.
Next we tested uranyl ion extraction ability from aqueous solutions by varying metal concentrations, to shed light on the affinity of this material towards uranium (VI) species. The adsorption isotherm (pH of the solution was optimized at 6, Supplementary Fig. 13) was obtained to assess the overall capacity of the adsorbent by allowing it to equilibrate with solutions containing varying UO 2 2+ concentrations ranging from 1 to 250 ppm and a phase ratio of 0.25 mg mL −1 . The resulting uranium adsorption isotherm in Fig. 3d is in excellent agreement with the Langmuir model, exhibiting a correlation coefficient R 2 = 0.988. As was envisioned from small-molecule studies, the H 2 BHT adsorbent material has high affinity towards uranyl ion in aqueous solution with the maximum uptake capacity of 105 mg g −1 with an equilibrium concentration of 205 ppm (experimental data used to plot the uranium sorption isotherm are given in Supplementary Table 6 adsorption test was performed to probe the selectivity profile of the developed material towards uranium (VI) versus vanadium (V) species in solution. The adsorbent sample (6.5 mg) was stirred overnight in 200 mL solution (pH 6) with equal concentrations (10 ppm) of uranium (UO 2 (NO 3 ) 2 .6H 2 O) and vanadium (Na 3 VO 4 ). Impressively, the uranium concentration dropped significantly from 10 ppm to~0.15 ppm despite the high concentration of competing vanadium (V) species in solution (Supplementary Tables 7-9 and Supplementary Note 3). From these results, we conclude that the H 2 BHT polymeric chelator has superior binding characteristics for uranium (VI) over vanadium (V), affording selective recognition of uranyl in the mixture of metal ions. In contrast to the currently used state-of-the-art polyamidoxime adsorbents 44 , which possess very low recyclability due to highly acidic conditions required during the elution process 40 33,38,45 , while exhibiting almost identical affinity towards uranyl ions. As a result, the adsorbed uranium could be easily recovered from the polymer using simple aqueous carbonate treatment, while the adsorbent material can be readily regenerated by aqueous base treatment and further reused. In addition to the successful synthesis and testing of the adsorbent, complete agreement between computational and experimental results from small-molecule studies have been validated by a proof-of-principle synthesis of the adsorbent material that exhibits the same features as small-molecule ligand. This study ushers in a practical approach towards the polymer design and synthesis of adsorbent materials decorated with the tailor-made ligands. Current efforts on extending this strategy to applications in water remediation and sequestration of other valuable metals are ongoing and will undoubtedly result in even deeper understanding of the underlying factors affecting the selectivity and efficiency of metal sequestration from complex aqueous media.

Methods
Materials and measurements. Commercially available reagents were purchased and were used without additional purification. 1 H NMR spectra were recorded on a Bruker Avance-400 (400 MHz) spectrometer. Chemical shifts are expressed in ppm downfield from TMS at δ = 0 ppm, and J values are given in Hz. 13 C (100.5 MHz) cross-polarization magic-angle spinning (CP-MAS) NMR experiments were recorded on a Bruker Avance-400 (400 MHz) spectrometer equipped with a magicangle spin probe in a 4-mm ZrO 2 rotor. Scanning electron microscopy (SEM) was performed on JEOL SM-6060 Scanning Electron Microscope. IR spectra were recorded on a Perkin-Elmer Frontier FT-IR spectrometer. ICP-OES was performed on a Perkin-Elmer Elan DRC II Quadrupole instrument. EDX mapping were performed on a Hitachi SU 8000. XPS spectra were obtained on a Thermo ESCALAB 250 with Al Kα irradiation at θ = 90°for X-ray sources, and the binding energies were calibrated using the C1s peak at 284.9 eV. Details on the synthesis, potentiometric titrations, UV-Vis, NMR, X-ray crystallography, and computational studies are given in the Supplementary Methods.

Data availability
The X-ray crystallographic coordinates for UO 2 (BHT) structure reported in this study have been deposited at the Cambridge Crystallographic Data Centre (CCDC), under deposition number 1854619. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif. All data supporting the findings discussed here are available within the paper and its Supplementary Information files, or from the corresponding authors upon request. Additional data can be provided by the authors upon reasonable request.