1 Introduction

The past two decades have witnessed a revival of interest in low-valent uranium chemistry, such as the unusual reactions mediated by uranium(III) [1,2,3,4,5,6,7,8]. As further development in this field relies on the availability of uranium(III) compounds, the establishment of their convenient syntheses has drawn much attention [9,10,11,12].

Uranium(III) species (e.g., UI3 and U(OTf)3) and their solvent adducts are typically prepared from uranium metal [9, 11, 13,14,15], while other routes such as the reduction of uranium(IV) by metals [16, 17] or other agents [18,19,20,21,22,23,24,25,26] are less common. These alternative routes are appropriate for the synthesis of compounds such as UL4(Ph4B)3 (L = amide derivative) [18], LiU2Cl7(THF)4.5 (THF = tetrahydrofuran) [21], and U(L′)3 (L′ = N,N-dimethylaminodiboranate) [25] but are not practical in other cases because of the difficulty of separating the spent reducing agent from uranium(III) products. The most convenient and common reductant for this purpose is zinc amalgam (Zn(Hg)) [26], which has been used for the preparation of uranium(III) species from uranium(IV) and uranyl(VI) in aqueous [27, 28] and organic solutions [29, 30] and for the study of their electrochemical and spectroscopic properties. Previously, we evaluated the stability of uranium(III) prepared by the reduction of U(OTf)4 with Zn(Hg) in THF and found that the uranium(III) oxidation state could be maintained for more than a week when the solution was kept in contact with Zn(Hg), whereas gradual oxidation was observed in the absence of Zn(Hg) [31]. Therefore, uranium(III) compounds with various counter ions can be conveniently prepared if zinc(II) eluted from Zn(Hg) can be rapidly removed from uranium(III) solutions in organic solvents.

Selective reduction relies on the differences in solute redox potentials and has been used in numerous bulk electrolysis-based electroseparations, as exemplified by the removal of > 96 wt% of aluminum impurities from LiCl–KCl melts [32] and that of ~ 95% of CaCl2 from CaCl2–LiCl–KCl melts [33] prior to lithium(I) reduction as well as by the separation of uranium from lanthanides in LiCl–KCl melts on a liquid gallium electrode [34]. The above technique was also used for the removal of non-metallic impurities such as oxygen, sulfur, and selenium from molten copper [35] and the removal of bromate using boron-doped diamond electrodes [36]. Although the bulk electrolysis of uranium solutions has been widely studied for various purposes, e.g., direct recovery for spent nuclear fuel reprocessing [34, 37,38,39,40,41,42], the solution-phase Electroseparation of zinc(II) from uranium(III) remains underexplored.

The electrochemistry of zinc and low-valent uranium has been studied using the mercury (Hg) electrode in aqueous and organic solutions. The behavior of zinc(II) on this electrode can be simply explained by amalgamation and dissolution irrespective of the solution composition [43,44,45], while the electrode reactions of low-valent uranium are strongly influenced by both the solvent and the electrolyte. In polarographic studies of UCl4, two cathodic waves assigned to the reversible reduction of uranium(IV) to uranium(III) and irreversible uranium amalgamation were observed at − 1.13 and − 1.93 V vs. the standard calomel electrode (− 1.52 and − 2.32 V vs. Ag/Ag+) in dimethylformamide (DMF), respectively; whereas both of these reactions were irreversible in dimethyl sulfoxide (DMSO) [46]. Two irreversible cathodic waves were also observed on a hanging mercury electrode (HME) when a solution of U(ClO4)4 in DMSO was probed by cyclic voltammetry (CV) [47]. In the bulk electrolysis of uranium(IV) in DMF, the electrode reactions are also strongly dependent on the electrode material and counter ion, e.g., in the case of the Hg pool electrode and perchlorate as the counter ion, the electrode potential stays at − 1.4 V vs. Fc/Fc+ (− 1.4 V vs. Ag/Ag+), whereas a plateau at − 2.2 V vs. Fc/Fc+ (− 2.2 V vs. Ag/Ag+) is observed for triflate as the counter ion. In both cases, the intensity of the uranium(IV) absorption band decreases with time at the plateau, whereas that of the uranium(III) band initially increases and then saturates or decreases at the same plateau [48].

To shed light on the electrochemical behavior of low-valent uranium triflate, we probed the electrode reactions of U(OTf)4 by voltammetry, using the standard HME and that where Hg was substituted by Zn(Hg). This study used DMF with its large negative potential window [49] as a solvent and tetrabutylammonium triflate (TBAOTf) as a supporting electrolyte. For comparison, we also probed the bulk electrolysis of U(OTf)4 in the presence of zinc(II) in DMF containing TBAOTf on both the pool Hg and the Zn(Hg) electrodes. The Electroseparation of zinc(II) from the uranium(III) solution prepared by the reduction of U(OTf)4 on Zn(Hg) in DMF was studied for the first time to evaluate the possibility of the convenient preparation of spent reductant-free uranium(III) starting materials.

2 Experimental

2.1 Materials

DMF was purchased from Fujifilm Wako Pure Chemical Corporation and distilled under reduced pressure before use. TBAOTf was purchased from Tokyo Kasei Kogyo Co., Ltd. (Tokyo, Japan) and used without further purification. Other chemicals, except for UO3, were used as received (Fujifilm Wako Pure Chemical Corporation). Deionized water (18 MΩ cm, Academic A10 model, Milli-Q, USA) was used in all experiments. Zn(Hg) was prepared as described elsewhere [28]. Briefly, zinc tips (3 g) were reacted with Hg (100 g) in 0.5 M aqueous H2SO4 for 1 h at 343 K, and the resulting product was washed with dry DMF (5 × 20 mL) to remove residual water.

U(OTf)4 was prepared as follows. UO3 (5.72 g) was dissolved in aqueous triflic acid (12.3 mL CF3SO3H + 10 mL H2O) at 353 K and the solution was electroreduced on a Hg cathode at a constant voltage to afford uranium(IV). The product obtained after the evaporation of the obtained solution was dried in vacuo for 12 h at 453 K and for a further 3 h in a flow of dry N2 gas. The resulting green dendritic crystals were examined by elemental analysis (Flash EA1112 CHN analyzer, Thermo Quest Italia) and inductively coupled plasma-atomic emission spectrometry (ICP-AES; ICP-7500, Shimadzu Corp., Japan), and their composition was determined as U(OTf)4 (calcd.: U, 28.53 wt%; C, 5.76 wt%; H, 0.00 wt%; found: U, 28.36 wt%; C, 6.14 wt%; H, 0.00 wt%).

The uranium(III) solution used for electrochemical zinc(II) removal was prepared by the reduction of U(OTf)4 over Zn(Hg) in DMF and the obtained solution was promptly used to determine the molar extinction coefficients of the uranium(III) absorption bands (Fig. S1).

2.2 Methods

All experiments were carried out inside an Ar-filled glove box (< 1 ppm O2 and H2O) at room temperature. CV and normal pulse voltammetry (NPV) measurements were performed using a three-electrode system (HZ-3000 and HZ-7000, Hokuto Denkou Corp., Japan). The working electrode (WE) was either the HME (Yanako Co. Ltd., Japan) or a Zn(Hg) electrode obtained by substituting Hg in the HME for Zn(Hg) (electrolyte contact area ≈ 1.4 mm2). The counter electrode (CE) was a 10 mm × 10 mm Pt plate. The Ag/Ag+ reference electrode (RE) comprised a silver wire and a solution of AgNO3 (0.01 M) and TBAOTf (0.1 M) in freshly prepared DMF. Bulk electrolysis and the Electroseparation of zinc(II) were carried out using the setup illustrated in Fig. 1. The pool WE (diameter = 43 mm), CE, and RE (described above) were used. In addition to the electrodes, the system comprised a beaker-type cell (30 mL) and an anion-exchange membrane (Selemion APS, Asahi Glass Engineering Co., Ltd., Japan). The anolyte was circulated in a flow-type optical cell (path length = 2 mm) installed outside the glove box for in situ visible–near-IR spectroscopic measurements (UV-3100PC, Shimadzu). Constant current electrolysis was performed at − 5 or − 10 mA, and WE potentials were recorded. Each solution (50 µL) was sampled at regular intervals and analyzed using ICP-AES to quantify total uranium and zinc(II) concentrations.

Fig. 1
figure 1

Schematic setup used for bulk electrolysis

Full size image

3 Results and discussion

3.1 Voltammetric investigation of U(OTf)4 reduction on Hg and Zn(Hg) electrodes

Figure 2 presents the cyclic voltammograms of 5 mM Zn(OTf)2 and 5 mM U(OTf)4 in DMF containing 0.1 M TBAOTf and that of DMF containing 0.1 M TBAOTf (for the potential window survey) recorded on the HME. The potential was scanned at 0.2 V s−1 in the cathodic direction starting from the open-circuit potential. The voltammogram of the supporting electrolyte showed that its potential window was sufficient to evaluate the electrode reactions of low-valent uranium triflate and zinc amalgamation. The voltammogram of Zn(OTf)2 showed a single redox wave (c1′/a1′; Zn2+ + 2e− \(\rightleftarrows\) Zn(Hg)) around − 1.4 V vs. Ag/Ag+ (hereinafter, all potentials are referenced to Ag/Ag+ if not stated otherwise; Fig. 2a). In contrast, the voltammogram of U(OTf)4 exhibited three cathodic waves (c1, c2, and c3 at − 1.48, − 1.81, and − 2.65 V, respectively) and one anodic wave (a1 at − 1.4 V) that disappeared when the potential range was extended beyond − 2.1 V in the negative direction (Fig. 2b). The a1 peak current recorded at the reversal potential of − 1.7 V (1.25 µA) increased when the potential was set to − 2.1 V (1.72 µA). These results suggest that waves c1 and c2 can be attributed to the reduction of two different uranium(IV) species. Moreover, the subsequent reaction of wave c2 was assigned to the degradation of the uranium(III) species generated in wave c2, which increased the content of uranium(III) species oxidized in wave a1. The reduction of U(OTf)4 in DMF probably involved the following reactions (coordinated DMF molecules are omitted for clarity).

Fig. 2
figure 2

Cyclic voltammograms of a 5 mM Zn(OTf)2 in DMF containing 0.1 M TBAOTf, b 5 mM U(OTf)4 in DMF containing 0.1 M TBAOTf (solid line) + the results of three consecutive scans (inset) and DMF containing 0.1 M TBAOTf (dotted line) recorded on the HME at 0.2 V s−1. The sweeps were started at the open-circuit potential and proceeded in the cathodic direction

Full size image
$${\text{U}}^{{{\text{4}} + }} + {\text{ OTf}} \rightleftarrows \left[ {{\text{U}}\left( {{\text{OTf}}} \right)} \right]^{{{\text{3}} + }} ,$$
(1)
$${\text{U}}^{{{\text{4}} + }} + e^{ - } \rightleftarrows {\text{U}}^{{{\text{3}} + }} \left( {{\text{forward}}:{\text{ a1}},{\text{ backward}}:{\text{ c1}}} \right),$$
(2)
$$\left[ {{\text{U}}\left( {{\text{OTf}}} \right)} \right]^{{{\text{3}} + }} + e^{ - } \left( {{\text{c2}}} \right){\text{ }} \to {\text{ }}\left[ {{\text{U}}\left( {{\text{OTf}}} \right)} \right]^{{{\text{2}} + }} \to {\text{ U}}^{{{\text{3}} + }} + {\text{ OTf}}^{ - } ,$$
(3)
$${\text{U}}^{{{\text{3}} + }} + {\text{ 3}}e^{ - } \left( {{\text{c3}}} \right){\text{ }} \to {\text{ U}}\left( {{\text{Hg}}} \right).$$
(4)

All these reactions except for reaction (4) were sufficiently fast to maintain reproducibility between voltammograms across several scans (Fig. 2, inset). Moreover, the electrode reaction of the uranium(IV) solvate complex (reaction (2)) was quasi-reversible (Figs. S2 and S3, Table S1), and this complex accounted for only 24% of all uranium(IV) species in solution (Fig. S4). However, this reaction may have substantially overlapped with zinc amalgamation/dissolution.

To compare the electrode reactions of pure and zinc-amalgamated Hg, CV measurements were also conducted on Zn(Hg) for 5 mM U(OTf)4 in DMF containing 0.1 M TBAOTf and for DMF containing 0.1 M TBAOTf (Fig. 3). The potential scans were carried out as in the case of the HME, namely in the cathodic direction from the open-circuit potential at a sweep rate of 0.2 V s−1, and showed that the positive side of the potential window was restricted by the onset of zinc(II) dissolution from Zn(Hg) at − 1.55 V. The voltammogram of the U(OTf)4 solution exhibited three cathodic waves (c1′, c2, and c3 at − 1.52, − 1.80, and − 2.60 V, respectively) with peak potentials consistent with those observed for the HME (Fig. 2). However, the c1′ peak current of the second and third cycles exceeded that of the first cycle (Fig. 3, inset), in contrast to wave c1 on HME (Fig. 2 inset). Moreover, a similar peak was observed during the potential window measurement. Based on these results, wave c1′ was assigned to zinc amalgamation. It is important to note that the competing reduction of the uranium(IV) solvate complex (wave c1) was expected to occur at the same potential.

Fig. 3
figure 3

Cyclic voltammograms of 5 mM U(OTf)4 in DMF containing 0.1 M TBAOTf + the results of three consecutive scans (inset) and of DMF containing 0.1 M TBAOTf (dotted line) recorded at 0.2 V s−1 on Zn(Hg). The sweeps were started at the open-circuit potential and proceeded in the cathodic direction

Full size image

3.2 Bulk electrolysis of U(OTf)4 in DMF on Hg and Zn(Hg) electrodes

To investigate the electrode reduction of low-valent uranium triflate and zinc(II) on the large-area WE, we performed the bulk electrolysis of U(OTf)4 on Hg and Zn(Hg) in DMF and compared the results with those of Voltammetric experiments.

Figure 4 shows the solution-phase behavior of uranium(III) and zinc(II) (i.e., the evolution of [U3+]/[U]total and [Zn2+]) and that of WE potential during the bulk electrolysis of 6.4 mM U(OTf)4 (hereinafter abbreviated as “without Zn(OTf)2”) and a mixture of 6.7 mM U(OTf)4 + 4.6 mM Zn(OTf)2 (“with Zn(OTf)2”) in DMF containing 0.1 M TBAOTf at − 10 mA. [U3+] was calculated from the intensity of uranium(III) absorption at 11,062 cm− 1 measured during electrolysis (Fig. S5) and the corresponding molar extinction coefficient (estimated as 436 M− 1 cm− 1; Fig. S1).

Fig. 4
figure 4

Changes in (a) [Zn2+] and [U3+]/[U]total and (b) the electrode potential of Hg during constant current (− 10 mA) electrolytic reduction of 6.4 mM U(OTf)4 (red) or a mixture of 6.7 mM U(OTf)4 + 4.6 mM Zn(OTf)2 (black) in DMF containing 0.1 M TBAOTf. (Color figure online)

Full size image

In the case of “without Zn(OTf)2,” the [U3+]/[U]total ratio saturated at < 0.2 about 20 min after the onset of electrolysis. Similar results were obtained using other uranium(III) absorption bands (at 8130, 9242, 9469, 10,121, 13,440, 16,025, and 18,587 cm− 1), whereas the uranium(IV) bands (at 7000, 9500, and 12,500 cm− 1) monotonically lost intensity (Fig. S5). The WE potential curve showed one precipitous shoulder (− 1.8 V) and one plateau (− 2.3 V). There was no signal around − 1.4 V related to uranium(IV) solvate reduction (reaction (2)). The shoulder, assigned to the reduction of [U(OTf)]3+ (reaction (3)), was too narrow for this reduction to be complete, as assumed from previous Voltammetric investigations. Moreover, the plateau potential of − 2.3 V, corresponding to the continuous reduction of uranium(IV) and the preservation of uranium(III) species on the plateau following their first generation, has never been observed in CV and NPV experiments. These results indicate that the reduction behavior of U(OTf)4 on the large-area WE (diameter = 43 mm) used in bulk electrolysis is significantly different from that on the small-area WE (area ≈ 1.4 mm2) used in Voltammetric experiments.

In the case of “with Zn(OTf)2,” [Zn2+] rapidly decreased over the first 50 min, subsequently more slowly decreasing to 0.1 mM. The [U3+]/[U]total ratio started to increase after 40 min and saturated at 0.25 after 70 min. On the other hand, the WE potential curve showed one shoulder (− 1.8 V) and two plateaus (− 1.5 and − 2.3 V), respectively. The first plateau potential well matched that of zinc amalgamation observed in CV experiments and the period agreed with the abovementioned decrease of [Zn2+]. Moreover, the reactions of the other plateau and shoulder were essentially identical to those observed “without Zn(OTf)2,” whereas the time of [U(OTf)]3+ reduction was obviously longer in the latter case.

Figure 5 presents the changes in [U3+]/[U]total, [Zn2+], and WE potential during the bulk electrolysis of 5.2 mM U(OTf)4 in DMF containing 0.1 M TBAOTf with Zn(Hg) as the WE at − 5 mA. The behavior of [Zn2+] indicated that the slight occurrence of zinc amalgamation was followed by zinc dissolution from Zn(Hg) in the first 20–25 min. Simultaneously, uranium(III) was generated up to [U3+]/[U]total = 0.44 by the reduction of uranium(IV) (Fig. S6) and the [U3+]/[U]total ratio saturated at 0.66 after 45 min. The WE potential curve exhibited a gradual slope from − 1.5 to − 1.8 V and one plateau at − 2.3 V. These results indicated that the potential slope observed for the first 25 min corresponded to the competing reactions of zinc amalgamation and the electrolytic reduction of uranium(IV) to uranium(III). However, when zinc amalgamation was almost complete, the electrolytic generation of uranium(III) became difficult and finally stopped, whereas the reduction of uranium(IV) monotonically continued (Fig. S6) in the plateau region at − 2.3 V.

Fig. 5
figure 5

Changes in (a) [Zn2+] and [U3+]/[U]total and (b) the electrode potential of Zn(Hg) during constant current (− 5 mA) electrolytic reduction of 5.2 mM U(OTf)4 in DMF containing 0.1 M TBAOTf

Full size image

The results of bulk electrolysis experiments demonstrated that the WE potential of zinc amalgamation was consistent with the results of Voltammetric experiments, in contrast to that of the electrochemical reactions of low-valent uranium triflate. Notably, the former potential was more positive than the latter, especially for the plateau potential at − 2.3 V, which was related to uranium(III) degradation.

3.3 Electroseparation of zinc(II) from uranium(III) solution on Zn(Hg)

After the reduction of 4.5 mM U(OTf)4 in DMF by Zn(Hg), we attempted the Electroseparation of zinc(II) from the uranium(III) solution by bulk electrolysis on Zn(Hg).

Figure 6 shows the concentrations of uranium(III) and zinc(II) ([U3+]/[U]total and [Zn2+]) and the WE potential as functions of time, revealing that [Zn2+] monotonically decreased and reached 0.13 mM at 50 min. In contrast, the [U3+]/[U]total ratio was stable at 0.99 for 35 min and was comparable to that obtained just after reduction by Zn(Hg) ([U3+]/[U]total = 1.00 at 40 min), subsequently experiencing a slight decrease. Similar results were obtained from other uranium(III) absorption bands (Fig. S7). The WE potential curve showed that zinc amalgamation occurring at the first plateau of − 1.52 V, as assigned by CV experiments, exclusively proceeded up to 35 min. At times of 35–40 min, the potential shifted in the negative direction from the first plateau to the second plateau located at approximately − 2.3 V. In this range, spectroscopic observations indicated the occurrence of competitive zinc amalgamation and uranium(IV) (presumably U(OTf)3+) reduction. The second plateau (t > 40 min) reflected the degradation of uranium(III) without uranium amalgamation corresponding to the potential of − 2.6 V (reaction (4)).

Fig. 6
figure 6

Changes in (a) [Zn2+] and [U3+]/[U]total and (b) the electrode potential of Zn(Hg) during the Electroseparation of zinc(II) from DMF containing 4.5 mM uranium triflate after reduction by Zn(Hg)

Full size image

The coulombic efficiency (ηC) of zinc(II) Electroseparation (i.e., zinc amalgamation) is estimated as:

$$\eta _{{\text{C}}} ~ = ~\frac{{\left( {C_{{i - 1}} - C_{i} } \right)V}}{{(q_{i} - q_{{i - 1}} )/nF}},$$
(5)

where Ci and Ci−1 are the concentrations of [Zn2+] in samples i and i–1 (i = 1, 2, …), V is the solution volume (L), qi and qi−1 are the integral quantities of electricity consumed during electrolysis (C), F is the Faraday constant, and n is the stoichiometric number of electrons involved in the electrode reaction (n = 2 for the reduction of zinc(II)). Simultaneously, the zinc(II) separation ratio is estimated as:

$${\text{Separation ratio }} = {\text{ }}\left( {{\text{1 }} - C_{i} /C_{0} } \right){\text{ }} \times {\text{ 1}}00\% ,$$
(6)

where C0 is [Zn2+] at the beginning of electrolysis. Table 1 lists the thus obtained values, revealing that the highest ηC equaled 0.93, while the final zinc(II) separation ratio equaled 93.4%. These results demonstrate that almost quantitative zinc(II) removal from the uranium(III) solution occurred under the chosen conditions.

Table 1 Coulombic efficiency (ηC) and zinc(II) separation ratios achieved at different times
Full size table

4 Conclusion

The reduction of U(OTf)4 in DMF containing TBAOTf was studied on the pristine HME and on the HME where Hg was replaced with Zn(Hg). The three cathodic waves observed in the former case were assigned to the reduction of a uranium(IV) solvate complex and [U(OTf)]3+ (− 1.48 and − 1.81 V, respectively) and uranium amalgamation (− 2.65 V). The redox wave of the uranium(IV) solvate complex overlapped with those of zinc(II) dissolution from Zn(Hg) and zinc amalgamation. The electrode reactions of low-valent uranium triflate on the large-area WE (used for bulk electrolysis) were critically different from those observed on the small-area WE (used for Voltammetric experiments), in contrast to the electrode reaction of zinc(II). Notably, a new plateau potential related to uranium(III) degradation was obtained around − 2.3 V instead of a lack of uranium(IV) solvate complex reduction under all conditions. These results demonstrated that zinc amalgamation proceeded prior to uranium(III) degradation. The Electroseparation of zinc(II) from the uranium(III) solution prepared by the reduction of U(OTf)4 using Zn(Hg) in DMF was conducted in a quantitative manner, with the final zinc(II) separation ratio equaling 93.4%. The results show that Electroseparation has sufficient potential for the preparation of spent reductant-free uranium(III) starting materials. We are currently attempting to demonstrate this method using THF, which is typical synthetic solvent, and subsequent synthesis of uranium(III) complexes.