Increasing the circularity of the copper metallurgical industry: Recovery of Sb(III) and Bi(III) from hydrochloric solutions by integration of solvating organophosphorous extractants and selective precipitation

1 Increasing the circularity of the copper metallurgical industry: recovery of Sb(III) and Bi(III) from hydrochloric solutions by integration of solvating organophosphorous extractants and selective precipitation N. Benabdallah1, D. Luo2, M. Hadj Youcef1, J. Lopez2, M. Fernández de Labastida2, A.M. Sastre2, C.A. Valderrama2, J.L. Cortina2,3* 1 Laboratoire de Chimie et d’Electrochimie des Complexes Métalliques (LCECM), Département de Génie Chimique, Faculté de Chimie, Université des Sciences et de la Technologie d’OranMohamed Boudiaf (USTOMB), Oran, Algérie 2 Chemical Engineering Department and Barcelona Research Center for Multiscale Science and Engineering, UPC-BarcelonaTECH, C/ Eduard Maristany, 10-14 (Campus DiagonalBesòs), 08930 Barcelona, Spain 3Water Technology Center CETaqua, Carretera d’Esplugues 75, 08940 Cornellà de Llobregat, Spain *Corresponding author. e-mail: [email protected]du (J.L. Cortina) Revised Manuscript (clean for typesetting) Click here to view linked References 2 Abstract The lack of high-purity ores has made the copper industry exploit low-impurity ores containing As, Sb and Bi which can affect the final product quality due to the floating insoluble arsenates. This is usually solved by treating the electrorefining electrolyte in a polishing stage devoted to removing Sb and Bi using aminophosphonic resins. Once the resin is saturated, it is regenerated using 6 M HCl, and the eluate produced is treated with CaO(s) to precipitate both Sb and Bi. However, this stream is of interest due to the inclusion of both elements in the Critical Raw Material list from the European Union. Therefore, the linear management option must be substituted by a circular approach. This work aims to develop such approach by the selective separation and recovery of Sb and Bi from HCl stream polluted with As. The method presented was based on the use of using a solvating mixture of alkylphosphine oxides commercialized as Cyanex 923 dissolved in kerosene and 1-decanol (10%) as phase modifier. The separation factors of Sb(III) and Bi(III) from As(V) were evaluated and optimized as function of both extractant concentration and the aqueous to organic phase ratio. Results showed that at the lowest Cyanex 923 concentration (0.15 mol/L) and working at an A/O ratio of 1/3, it could be possible to extract 39% Bi and 78% Sb with As co-extraction below 2.5%. Both Sb and Bi were easily recovered (>90%) using 8M HNO3 as stripping agent. Following a treatment using NaOH/NaCl or NaOH, both elements could be recovered either as oxychlorides (SbOCl(s) and BiOCl(s)) or oxides (Sb2O3(s) and BisO3(s)). Keywords: antimony; bismuth; arsenic; HCl extraction; Cyanex 923; critical raw materials 3 1. Introduction The importance of copper has led to a continuous search for more efficient production methods capable of yielding high-purity copper by pyrometallurgical processing methods such as smelting [1]. Over 75% of the copper metal produced comes from the smelting of copper sulfide concentrates. After copper is isolated from sulfide concentrates, it is deposited into copper anodes (approx. 99% pure). These anodes contain a variety of metal impurities that are soluble in the molten copper, such as precious metals, Ni, Pb, Fe, Se, Te, As, Sb, Sn and Bi, and thus further refining is required. Moreover, aside from the overall purity of copper (99.99% for wire and electrical operations) [2], some impurities, in particular, must be kept to a minimum (e.g. bismuth present even at µg/kg levels can make copper too brittle to pull wire [3]). Therefore, as in other metal processing industries, smelting alone is insufficient to generate such high purity copper, and electrochemical processes (e.g. electrorefining) are required During the electrorefining process (80% of world production), anode impurities, most frequently As, Bi, Fe and Sb, are dissolved along with copper from the anode into the electrolyte [4,5]. These impurities can accumulate in the electrolyte or form different types of anode slimes which can adhere to the anode, deposit on the bottom of the electrolytic cell or float within the electrolyte [2,6–8]. Atlantic Copper [9] determined that the ratio of concentrations ([As]/([Sb]+[Bi])) should be kept within a 20-35 range as the optimal working conditions to prevent these undesirable phenomena from occurring in the copper tank electrolytes and hence avoiding problems related to copper cathode quality (e.g. to prevent the occurrence of floating slimes (lower value) and precipitation of insoluble arsenates (higher value)). This can be achieved by controlling these impurities in the anodes and their extraction in the liberator cells, in which Sb and Bi are removed together with As. The anode composition determines that it might be necessary to remove Sb and Bi, but not As from the electrolyte [4]. Copper plants 4 along the world are experiencing a significant increase in the impurities content (mainly Sb and Bi) in the ores and concentrates processed. Thus, it is essential to keep impurities below the levels accepted by national and international standards for the copper industry and to decrease the total emission of harmful components. In order to assure the cathode quality, the electrolyte solution must be bled and treated, and different removal of Sb and Bi options have been evaluated. The rationale of the processes developed was based on the fact that both elements are present in 150-200 g/l H2SO4 solutions as cationic forms (SbO+ and BiO+). The control of Sb and Bi in these electrolytes has been investigated using different polishing option as precipitation [10], activated carbon adsorption [11–15], chelating ion exchange (IX) resins [11,16–19], material recognition technology (MRT) adsorbents [20] and solvent extraction (SX) [21–24]. However, only chelating ionexchange resins and MRT adsorbents were initially installed in several Japanese copper electrorefineries (Saganoseki, Hitachi, Tamano), allowing the removal of both Sb and Bi, keeping its concentrations below 0.5 g/L in the electrolyte solution [20,25,26]. Since then, this process was extended abroad Japan [27–29]. When IX resins are used as polishing systems, electrolyte impurities are selectively removed and, in a subsequent step, are regenerated with a concentrated HCl solution (e.g. up to 6 M) [17,30,31], promoting the formation of anionic chloride complexes (e.g. MCl4-). The most common option is a linear management treatment where the HCl eluate is typically limed and the sludge generated is disposed as toxic waste [25]. As an alternative, the HCl eluate is neutralized with Ca(OH)2 to collect Sb and Bi as their respective metal hydroxides forms sold as a raw material for producing electrical-quality bismuth [32]. Traditionally, Sb is mainly obtained from stibnite (Sb2S3(s)) either using a pyrometallurgical or hydrometallurgical route. The first one is the most applied in the applied in industry, allowing 5 to obtain pure commercial Sb (>99.5%). Nevertheless, the process needs a huge energy consumption and generates SO2(g) and an arsenic-alkali residue [35]. In the hydrometallurgical route, different strategies can be used to recover Sb, such as electrolysis or electrowinning [36], hydrolysis [37], solvent extraction [38], ion-exchange [25] and/or precipitation/crystallization [39]. Similarly to Sb, Bi can be obtained by a pyrometallurgical or hydrometallurgical route from bismuthinite (Bi2S3). Due to the volatility of Bi and its corresponding oxides, the hydrometallurgical route is preferred. In this case, Bi can be recovered using electrowinning or hydrolysis [40–42]. Nevertheless, only a few studies are focused on the recovery of Sb and Bi from these wastes. The absence for such lack of knowledge is based on the speciation changes suffering Sb. For instance, Sb is mainly present as Sb(III) [33] and forms anionic complexes in 3-6M HCl solutions such as SbCl4-. However, scarce data is found for the speciation of Sb(V), although some studies postulated the formation of SbCl6- [34]. Most of the adsorbents, IX resins and SX reagents developed for the polishing stage are not applicable as the chemistry was based on the extraction of the corresponding cationic forms (SbO+, BiO+). It should be mentioned the use of a MRT sorbent (Superlig 240) for anions removal, which have been applied to the selective recovery of Bi(III) from the valorization of anode slimes after a leaching stage with HCl solutions at the Kucc Magna electrorefinery (Salt Lake City, USA) [28]. SuperLig 240 is a polyacrylate resin containing a proprietary selective chelating functional group for anions as the expected BiCl6-3 in the concentrated HCl solutions. BiCl6-3 is then eluted by 9M H2SO4 solutions at room temperature to be recovered as Bi(HSO4)3(s). Ruiz et al. [13] evaluated at pilot plant scale the integration of ion-exchange resins to purify electrorefining solutions. Due to the speciation of both Sb and Bi in the electrorefining electrolyte (SbO+ and BiO+), the solution passed first through an amino-phosphonic ion exchange (AMPIX) (TP260, cation exchange resin). Once the resin was saturated, it was regenerated using 20-30% HCl. Due to 6 changes in speciation of both elements, present as SbCl4and BiCl4in HCl media, the eluate was then treated with a tertiary amine anion exchange resin (MP62). However, the process was not finally implemented at full scale as the regeneration process of the MP62 resin with 50% H2SO4 solutions was not effective. Accordingly, the use of SX could be the potential alternative using extractants developed for the removal of anions, as could be tertiary (e.g. Alamine 336), quaternary (Aliquat 336) amines, or solvating reagents (e.g TBP, TOPO, Cyanex 921, 923 and 925), although most of the applications described are for H2SO4 solutions [22,24]. The present study is centred on the development of a process for the recovery of the addedvalue Sb(III) and Bi(III) and removal of As(V) by valorizing solutions produced during copper electro-refinery processes, specifically from the effluents generated on the regeneration of the AMPIX polishing resins. While Sb and Bi are targeted for their recovery and separation because of their intrinsic economic value due to their scarcity (both of them are listed as Critical Raw Materials), As is considered a hazardous and toxic impurity affecting the quality of potential by-products and its safe separation from Sb and Bi is desired, ideally as an environmentally stable species. Therefore, a recovery route for separating Sb(III) and Bi(III) from As(V) in HCl aqueous media was developed based on the use of an organic phase containing a mixture of alkyl-phosphine oxides (Cyanex 923), as solvating extraction agent. Initially, tests were carried out with synthetic solutions containing Sb/Bi/As and later with real samples from the regeneration of the AMPIX resins at Atlantic Copper Plant (Huelva, Spain). In a second stage, the re-extraction of the loaded organic phase, wherein most of the As remained in the acidic aqueous medium, was evaluated with strong acids (e.g.H2SO4/HNO3) to evaluate its selective separation in the stripping stage. As solvating reagents are characterized for their ability on the extraction of strong acids and taking into account that the ion-exchange eluates are highly HCl concentrated streams, the influence of the HCl extraction on the global Bi(III)/Sb(III) extraction processes was evaluated. 7 2. Selective recovery of Sb and Bi from the HCl containing solutions generated in the regeneration of AMPIX polishing unit: speciation analysis and fundamentals The operation of the copper electrorefining stage is based on the control of the speciation of the main electrolyte impurities. In particular, the redox states of As and Sb have a great influence on the formation of the insoluble precipitates, which could affect copper cathode quality. As(V) and Sb(V) can form a series of arsenato-antimonic acids which can further react with As(III), Sb(III), and Bi(III) to form arsenato-antimonates [35]. As(V), Sb(III) Bi(III) can form arsenates [36] and Sb(V) plays a substantial role in the formation of floating slimes, which are amorphous and chemically undefined compounds that may contain Sb(III), Sb(V), Bi(III), As(V) and As(III) known as “floating slimes” [37–39]. Floating slimes are commonly avoided by controlling the total Sb concentration in the electrolyte below 0.5 g/L [29] and by maintaining the concentration of As in the electrolyte above 6–7 g/L and a [As]/([Sb]+[Bi]) molar ratio above 1.5–2 in the anodes [40]. There are several studies on the removal of Sb from copper electrolyte [41,42], but limited studies about the Sb speciation of the redox states. González de las Torres et al. [33] presented a speciation analysis of a pilot plant removing Sb and Bi from the electrolyte at the Atlantic Copper Refinery in Huelva, Spain (Figure 1). 8 Figure 1. Bi/Sb polishing stage of the copper tank electrorefining electrolyte of Atlantic Copper (Huelva, Spain) including a two stages treatment process: i) pre-treatment with Cu(s) shavings, and ii) ion-exchange column containing a aminophosphonic (AMPIX) resin The polishing treatment includes a two-step process in which the electrolyte is pumped through two columns in series. The first column contains copper shavings (Cu(s)) to promote the reduction of Fe(III) to Fe(II), and thus avoid the possible poisoning of the AMPIX resin with Fe(III). The second column contains the AMPIX resin (Lewatit MonoPlus TP 260), that removes Sb and Bi from the electrolyte. The operating configuration involves the conditioning of the resin with H2SO4, loading of the electrolyte, electrolyte displacement and backwash with H2SO4, regeneration of the resin with HCl and thiourea, displacement of HCl with H2SO4 at different pH (2, 4 and 6) and conditioning of the resin with recirculated H2SO4 from a previous stage. The IX stage allowed: i) quantitative removal of Sb(III) (>95%) and partial removal of Sb(V) (e.g. <60%); and, ii) minimum changes on the As(III) and As(V) concentrations. The evolution of Sb, Fe and As species in the electro refining electrolyte was monitored prior and after the installation of the Sb/Bi removal plant [33]. The results showed a ca. 45% decrease in 9 total Sb content, where Sb(III) concentration decreased from 0.18 to 0.09 g/ L and Sb(V) concentration diminished from 0.11 to 0.07 g/ L. The AMPIX resin also retained ca. 75% of the Bi(III) content. The total As increased during the period (from 7.7 to 9.0 g/L) due to changes in plant ore content being predominantly As(V) (ca. 93–95%). The total Fe concentration experienced little variation (0.9–1.1 g/ L) with Fe(II) being the main species (ca. 94–96%). However, the speciation of the stream generated in the regeneration of the AMPIX resin was not evaluated, but taking into account: i) the mass balance of the processing scheme (Figure 1) and; ii) the difficult elution of Sb(V) from AMPIX resins using concentrated HCl (4-6 M) in comparison with Sb(III) (ca 100% eluted) [17], it could be expected that As is mainly present as As(V) (ca. 94%) and Sb is mainly present as Sb(III) (>90%). Accordingly, the review of the geochemical database was centred as preliminary hypothesis in a chemical system containing HCl-Sb(V)-As(V)-Bi(III)-H2SO4 as major components (>10-2M). Other elements present in concentrations below 10-2 M were not considered in the modelling efforts, and their speciation in HCl solutions was not considered at this stage. Additionally, although Sb(III) and Bi(III) are complexed with HSO4-/SO4-2 anions, these species were not taken into account as the chloride concentration could be up to 400 times higher than the one of sulfate. The review of the geochemical data for Sb and Bi for natural water bodies has been thoroughly covered in the last decade by Filella et al. [43,44]. However, scarce efforts could be found for metallurgical and hydrometallurgical processes streams in both H2SO4 (the most common case) and HCl solutions (e.g. in the polishing stages of Sb and Bi with IX resins). The review of the published equilibrium data for Sb(III), As(V) and Bi(III) is summarized in Table 1. Solutions generated in the regeneration of the AMPIX resins are characterized by high concentrations of HCl (e.g. 3 to 6 M), however limited equilibrium data at this high ionic strength were found (i.e. Pitzer parameters). Therefore, modelling efforts were carried out using Hydra/Medusa and PhreeqC codes. Additionally, as most of the processes to be developed moved in the acidity 16 4. Results and discussion 4.1. Characterization of the IX eluate generated from the Atlantic Copper Refinery (Huelva, Spain) The chemical composition of the AMPIX resin eluate generated from the Atlantic Copper is summarized in Table 2. Values have a large variation as it depends on the operation of the polishing stage, and on the operation of the regeneration stage that is depending on the resin life in terms of sorption and desorption cycles, Table 2. Major and minor components concentrations in the HCl stream generated in the regeneration of the AMPIX resin along the polishing stage of the refining processes at Atlantic Copper (Huelva, Spain) Composition, mg/L pH SO4 Cl Ca Fe Cu As Sb Bi n.d. n.d. 63000 n.d. 6 348 1386 637 2429 n.d. n.d. 190500 n.d. 17 200 2254 6848 5143 n.d. n.d. 107120 n.d. 3 70 2254 7944 8388 -0.43 8240.00 115510 n.d. n.d. n.d. 3940 2790 3260 -0.58 7304.25 94385 256 2 272 1030 5455 8257 -0.47 16699.29 64359 n.d. n.d. n.d. 3617 1501 2273 -0.53 7757.60 n.d. 248 n.d. 255 1008 5278 8461 n.d. not determined 17 Values are the average of seven different samples collected along one year. The concentration of Sb(IIII) ranged from 9 to 11 g/L and Bi(III) from 2.3 to 8.4g/L. Values of other metals coextracted in the resin as Fe(II), Zn(II), Cu(II), Ni(II), Co(II), Ca(II) and Mg(II) are in the order of mg/L. The AMPIX resin is able to selectively extract elements in the (+III) oxidation valence (mainly Bi(III) and Sb(III), although Fe(III) and Al(III) are also co-extracted) from elements in the (+II) oxidation state in solutions of 200-250 g/L H2SO4. It is worth mentioning that the measured values of As(V) present as H3AsO4 (1.1 – 3.4 g/L) and S(VI) present as HSO4- (7-16 g/L), which are not expected to be adsorbed by the AMPIX resin. Its presence in the eluate could be associated with three potential phenomena: i) the partial or un-complete displacement of the electrolyte solution along the regeneration stages, as both components are present at high concentrations in the copper electrolyte, especially for the case of H2SO4 with 200 to 250 g/l; ii) adsorption on the polymer matrix through the “acid retardation phenomena” [48], and; iii) formation of As(V) mineral phases as BiAsO4(s) and SbAsO4(s) along the sorption stages due to the saturation of both minerals in the resin phase as it occurs in the copper tank electrolytes (e.g. formation of anode slimes) that latter are re-dissolved along the regeneration stage with HCl. Analysis with FSEM-EDAX of the internal section of AMPIX resins (Figure S2, Supplementary Information) identified mineral phases containing mainly: As, O, Bi and As, O, Sb, which support the last hypothesis. As it could be seen in the analytical data from EDAX analysis, other metals present in the electrolyte solutions such as Cu and Zn were also detected. 18 4.2. Extraction of Bi(III) and Sb(III) from HCl solutions generated in the regeneration of the AMPIX polishing unit by Cyanex 923 solutions in kerosene 4.2.1. Extraction of HCl Extraction of HCl acid using Cyanex 923 (S) as solvating extractant in kerosene was studied as a function of both HCl and extractant concentration. The experiments were performed by varying the concentration of HCl from 1 to 6 M and the one of Cyanex 923 between 0.05 to 1 mol/L at a 1:1 (O/A) phase ratio. The effect of Cyanex 923 concentration ([S]o) on hydrochloric acid extraction (%) at different concentrations is shown in Figure 3. In the range evaluated the extraction efficiency increased linearly with the concentration of Cyanex 923, reaching a maximum of 12.5±1.2% with a concentration of 1 mol/L of extractant and 4 mol/L of HCl. 19 Figure 3. Effect of Cyanex 923 concentration on the extraction of HCl at different concentrations. Room temperature (22°±1ºC), Phase ratio (O/A) = 1/1. Good agreement of the HCl extraction (%E) was found when compared with those previously published by Sarangi et al. [48] (Figure 4). They evaluated the extraction of HCl from 5 mol/L solutions with Cyanex 923 in kerosene for a 1:1 phase ratio (O/A), and the extraction of HCl increased from 9 to 27 % with the increase of Cyanex 923 concentration from 10% to 100%vol. The reported value at 10%(0.22) M is in agreement with the results of this study, as shown in Figure 3, with extraction values close to 10%. Figure 4. Comparison on acid extraction as a function of the concentration of Cyanex 923 (%vol). Data was taken from Sarangi et al. [48] The extraction of HCl by a solvating extractant (S) can be described by equation 6, where m is the number of solvating (S) molecules: 20 H+ + Cl– + m S(org) ⇔ H+ Cl– Sm(org) (6) The extraction equilibrium constant (KHCl) described in eq 6, could be expressed by eq. 7: KHCl = [H+ Cl– Sm(org) ] / ([H+] [Cl–] [S(org)]m) (7) While the HCl distribution coefficient (DHCl) could be calculated by eq. 8, DHCl = [H+ Cl - Sm(org)] / [ Cl–] (8) After integrating equations 7-8, the dependence of log DHCl as a function of the logarithm of the free concentration of S is described eq. 9. log DHCl – log [H+] = log Kex + m log [S(org)] (9) Therefore, the plot of (log D - log [H+]) versus log [S(org)], will lead to a slope m, which indicates the Cyanex 923 solvating number. HCl extraction data, shown in Figure S3, as logDHCl, were plotted as a function of log[S]org for the four different Cyanex 923 concentrations. A linear dependence was observed with slopes close to 1 (ca. 0.95 to 099), indicating that the extraction of HCl occurs through one molecule of Cyanex 923 as described in eq. 10: H+ + Cl– + S(org) ⇔ HCl·S(org) (10) The log KHCl values, also calculated from this plot, were found to decrease with increasing acid concentration (Table 3). Table 3. Liquid-liquid extraction constants of HCl by Cyanex 923 (KHCl) in kerosene calculated by linear regression analysis of log DHCl /log[S]org values from figure S3. CHCl(M) 1 2 4 6 log KHCl -1.00±0.05 -1.08±0.17 -1.14±0.04 -1.17±0.01 21 The extraction of acids by Cyanex 923 [49,50], tri-n-octylphosphine oxide and tris 2-(ethyl hexyl) amine [27] were traditionally centred on the recovery of H2SO4. However, extraction performance of HCl with different extractants as Alamine 336, Cyanex 923, Aliquat 336, tributyl phosphate (TBP), octylphosphine oxide and tris 2-(ethyl hexyl) amine have been also reported [27,48,51]. Results of this study are in agreement with the data reported by Sarangi et al. [48] for 5 M HCl in the range of 0.22 to 2.2 M Cyanex 923 with a logKHCl -1.1 (for m=1). Alguacil and López [49] when evaluating the extraction of 1 M HCl by using 0.025 to 0.5 M Cyanex 923 diluted in toluene and decane reported values of -0.95±0.09 and -1.00±0.11 (for m=1), respectively. Both values, even using a different solvent are in agreement with the determined value in kerosene for 1 M HCl solutions (-1.00±0.05). Regarding reneretarion of the solvating extractant, Sarangi et al. [48]. Although Alamine 336 was shown to be the better extractant for HCl, the acid could not be stripped from the loaded organic with water, while HCl was easily stripped from the loaded organics of Aliquat 336, TBP and Cyanex 923. 4.2.2. Extraction of As(V), Bi(III) and Sb(III) from single component solutions in 6 M HCl: influence of the extractant concentration The effect of Cyanex 923 concentration (from 0.05 to 0.5 mol/L) on the extraction of As(V), Bi(III) and Sb(III) from single component aqueous solutions in 6M HCl has been studied in the range 0.05 – 0.4 mol/L Cyanex 923 at an A/O ratio of 1:1 (Figure 5). 22 a b 23 c Figure 5. Effect of Cyanex 923 concentration on the extraction efficiency of a) As(V), b) Bi(III), and c) Sb(III) Room temperature (22°±1ºC), Phase ratio (O/A) = 1/1. As shown in Figures 6, the extraction efficiency of As(V), Bi(III) and Sb(III) increases with both Cyanex 923 and non-metal concentrations. Maximum extraction efficiencies of 26.6±1.6, 47.4±0.8 and 49.3±0.9 % were obtained for As(V) (0.05 mol/L), Bi(III) (0.02 mol/L) and Sb(III) (0.10 mol/L), at a concentration of 0.5 mol/L of Cyanex 923. Extraction efficiency was evaluated in the worst scenario of an excess molar ratio of 15 times of HCl to As(V), Bi(III) and Sb(III), and then the limiting factor of co-extraction of HCl (5.23±0.33% at 6 mol/L) could be reduced by optimization of the regeneration stage of the AMPIX resin. Dziwinski and Szymanowski [21] evaluated the use of a similar solvating extractant, Cyanex 925 in xylene, for extraction and separation of Sb(III) and Bi(III) from strong acids (HCl, HNO3 and H2SO4). Cyanex 925 contains five isomers of tri-octylphosphine oxide with branched and normal alkyl chains. Quantitative extraction of Sb(III) was achieved in the range 0.1-0.8 mol/L HCl and 0.011.0 mol/L H2SO4, while Bi(III) was extracted in the range 0.01-0.4 mol/L HCl and 0.1-1.0 24 mol/L HNO3. From the organic phase, Sb(III) was stripped back with 8.0 mol/L H2SO4 and Bi(III) with 2.0-3.0 mol/L HNO3. The extraction behavior of both elements was studied separately as a function of acid concentration, temperature, equilibrium time and stripping ability using various acids and bases. Based on these results, sequential methods for their separation from multicomponent mixtures were developed. According to speciation diagrams (see Figure 2), the main species of each element in HCl solutions (1 to 5 mol/L) are: a non-charged species for As(V) (H3AsO4), and two anionic species, BiCl4for Bi(III) and for SbCl4for Sb(III). The extraction of these three species from HCl solutions could be described by the two general extractions reactions described by eq. 11 and 12: As(V) H3AsO4 + mS(org) ⇔ H3AsO4·Sm(org) (11) Bi(III) and Sb(III) MCl4– + H+ + m Sorg ⇔ MCl4 H Sm(org) (12) The equilibrium constant (KM) are defined by eq. 13-14 respectively: KH3AsO4 = [HAsO4Sm(org)] / [H3AsO4][Sorg]m (13) KMCl4– = [MCl4HSm(org)] / [MCl4–][H+] [Sorg]m (14) Taking into account the distribution values DH3AO4 and DMCl4– and the extraction constant could be described by equations 15-16: log DH3AsO4 = log KH3AsO4 + m log [S]org (15) log D MCl4– + pH = log K MCl4– m log [S]org (16) According to equation 15-16, the plot of the logarithm of the distribution coefficient (D) of each non-metal as a function of the logarithm of the extractant concentration ([S]org) were built to 25 determine the number of solvating molecules (m) for each element (slope of the logD/log[S]org). The plots obtained are shown in Figure S4 (Supplementary information), and from the data, solvating numbers of the extraction reactions were calculated by linear regression analysis. Results indicated that As(V) and Bb(III) are extracted by H3AsO4´Sorg and HSbCl4.Srog as the slope values were close to one. Conversely, the slope values for Bi(III) were found to be around 1.4, implying the co-existence of two complexes with 1 and 2 molecules HBiCl4.Sorg and HBiCl4.S2org. Few data are available with extractants of the Cyanex family for the elements evaluated in the present study. Iyer and Dhadke [52] studied the extraction of Bi(III) and Sb(III) from 0.1-0.8 M mol/L HCl, 0.01-1.0 mol/L H2SO4 and 0.1-1.0 mol/L HNO3 and they proposed reactions involving two solvating molecules (m=2). Similarly, Ahmed et al. [53], when evaluating the extraction of Ga(III) by Cyanex 923 and Cyanex 925 in kerosene from 1-7 mol/L HCl also postulated two solvating molecules for both reagents (GaCl3S2, org). 4.2.3. Extraction of As(V), Bi(III) and Sb(III) from multicomponent synthetic aqueous solution in hydrochloric acid medium: influence of extractant concentration Experiments were carried out with multi-component synthetic solutions mimicking the effluent from the copper metallurgical plant. Selective separation of Bi(III) and Sb(III) from As(V) was achieved using different concentrations of Cyanex 923 in kerosene (0.05, 0.10 and 0.15 mol/L), from an aqueous phase containing 3.2 g/L As(V), 2.6 g/L Bi(III), 1.9 g/L Sb(III), 3 g/L Ca(II), 1 g/L Cu(II), 10 g/L H2SO4 and 3 mol/L HCl. The effect of the presence of Cu(II) and Ca(II) on the extraction performance was evaluated. The extraction experiments were carried out at different aqueous/organic phases ratios (A/O) of 1/1; 1/2 and 1/3. The extraction efficiency in absence of Cu(II) and Ca(II) for the different A/O ratios is shown in Figure 6. 32 of the organophosphorus extractants dibutyl butyl phosphonate (DBBP) and bis(2-ethylhexyl) phosphoric acid (D2EHPA) for As removal from synthetic Cu electrorefining solutions. The aqueous phase was characterized by 200 g/L H2SO4, 30 g/L Cu and 3 g/L As, among others. Authors studied the effect of extractant concentration, sulphuric acid concentration, arsenic concentration and the O/A ratio. They were able to remove 90% of the As in the solution working with 0.2 M D2EHPA and 2 M DBBP, while 76% was stripped using 0.4 M Na2SO4. However, authors observed losses of extractant in the aqueous phase with mixtures with low D2EHPA contents. Fan et al. [57] evaluated the recovery of Bi from a pickling solution (HNO3based) using 2.94 mol/L TBP diluted in kerosene. Crude bismuth contained 0.66 g/g Bi, 0.10 g/g Sb, 0.09 g/g Pb and 0.001 g/g As. Neither leachate composition nor the acid concentrations used for leaching were reported. During extractions, using one stage at O/A ratio 2, the authors were able to extract 98.4% Bi, 14.4% Sb, 4.6% Pb and 24.8% As. After scrubbing with saturated NaNO3 solution, 100% Pb and 81% As were removed, with low losses of Bi (0.6%) and Sb (2.2%). Then, an stripping was performed using 0.5 mol/L NaOH, allowing to recover 96.2% of the Bi in the organic phase. Artzer et al. [58] studied the performance of two phosphonic acid ester extractants (REX-1 and REX-2 from Basf) from copper electrorefining cells (150160 g/L H2SO4, 40-50 g/L Cu, 5-8 g/L As, 0.1-0.2 g/L Sb, 0.08-0.16 g/L Bi). Authors evaluated different compositions of the electrolyte and the effect of the mixtures REX-1 and REX-2 on extraction performance. They were able to achieve 85%-95% of Sb and 60-70% of Bi in the “live” electrolyte samples, with low Cu and As co-extraction (<10%). However, when REX-1 was present, a better stripping performance was achieved. Authors concluded that mixtures 25/75 REX-1/REX-2 could be preferred for high Sb and low Bi content, while 75/25 REX1/REX-2 could be preferred for the opposite case. 33 Table 4. Comparison of the experimental data with the one previously published in literature concerning As, Bi and Sb recovery using organic extractants Solution composition (g/L) Extraction Scrubbing Stripping Ref. Organic extractant Experimental conditions % Experimental conditions % Experimental conditions % HCl: 94 As: 1.0 Bi: 8.1 Sb: 5.4 0.15 mol/L Cyanex 923 diluted in kerosene containing 10% 1-decanol O/A: 3/1 As: 2.7±0.1 Bi: 38.5±1.5 Sb: 77.6±4.5 - - 8 M HNO3 O/A: 1/1 As: 86.7±6.6 Bi: 87.2±1.7 Sb: 95.6±5.4 This work H2SO4: 1022 As: 24 97 vol% TBP 3-stage Countercurrent O/A: 0.74 H2SO4: 31.4 As: 83.7 2-stage Countercurrent O/A: 4.0 Pure water H2SO4: 83.6 As: 24.9 4-stage Countercurrent O/A: 2.0 Pure water H2SO4: 100 As: 89.7 [54] H2SO4: 198 HCl: 18.5 Bi: 0.09 Sb: 0.25 Supported liquid membrane (PVDF) containing Cyanex 921 - Bi: 97.2 Sb: 95.4 - - 0.5% w/v tartaric acid Bi: 100 Sb: 100 [55] H2SO4: 200 Cu: 30 As: 3 Sb: 0.1 0.2 M D2EHPA and 2 M DBBP Batch O/A: 3 As: 90 Sb: 70 - - Water As: 67 [56] 0.4 M Na2SO4 As: 76 * 2.94 mol/L TBP in kerosene One-stage O/A: 2 Bi: 98.4 Sb: 14.4 Pb: 4.6 As: 24.8 Saturated NaNO3 solution Bi: 0.55 Sb: 2.22 Pb: 100 As: 81.0 0.5 mol/L NaOH Bi: 96.2 Sb: 30.4 Pb: - As: 0 [57] H2SO4: 158 Cu: 40.8 As: 7.9 Sb: 0.19 Bi: 0.08 25 wt.% REX-1, 75 wt.% REX-2 in Orfom SX-12 diluent O/A: 1 Bi: 62 Sb: 95 As: <5 - - 400 g/L H2SO4 O/A: 1 Bi: 62 Sb: 15 As: <5 [58] * not reported 34 4.5. Sb (III) and Bi(III) recovery process from ion-exchange eluate generated in the polishing stage of the copper electro-refining stage The separation results obtained along the study could be used to define the chemical basis for the selective recovery of Sb(III) and Bi(III) by the integration of three steps: i) liquid-liquid extraction using Cyanex 923 in kerosene to extract selectively Sb(III) and Bi(II) from As(V) where a fraction of HCl is co-extracted, ii) the stripping stage of the loaded organic phase containing Sb(III) and Bi(III) using HNO3, and, iii) the precipitation of Sb(III) and Bi(III) from HNO3 to obtain Bi2O3(s)/Sb2O3(s) in free-chloride solution or the corresponding BiOCl(s)/SbOCl(s) if solution is containing chloride ions (e.g. NaCl) for commercial uses [59,60]. Figure 10 describes the integration of the three stages previously mentioned. Figure 10. Proposed treatment of the AMPIX eluate including i) liquid-liquid extraction using Cyanex 923; ii) reextraction of Sb(III) and Bi(III) by using HNO3, iii) selective precipitation of SbOCl(s) and BiOCl(s) with NaOH, and iv) stabilization of As(v) as scodorite (FeAsO4 ⋅ 2H2O(s)). 35 The process described in Figure 10 provides a stabilization route for the As(V) separated stream based on its precipitation as scorodite (FeAsO4⋅2H2O(s)) using a mixture of FeSO4(s) and CaO(s) as has been described elsewhere [61]. As it is considered a state-of-the-art management option for streams containing As, it was not evaluated in this study. In comparison to published studies on the literature regarding the recovery of Sb(III) and Bi(III) at an industrial scale, the most preferred option is the use of selective precipitation, MRT or sorbents. For example, Sole et al [32] reported the effort Hitachi and Tamano refineries (Japan) using an AMPIX resin to selectively adsorb Sb and Bi from the electrolyte [62]. Following elution with HCl, the eluate is neutralized with Ca(OH)2(s) to collect Sb(III) and Bi(III) as the respective metal hydroxides. The neutralized sludge containing these mixtures of Bi(III)/Sb(III) is sold as a raw material for the production of electrical-quality bismuth. It has been also reported the use of MRT sorbents for Bi(III) removal from the IX eluate [63]. The Port Kembla copper refinery (Australia) uses SuperLig 83 to selectively remove Bi(III) from the electrolyte, which is later recovered from the resin with H2SO4, and pure bismuth sulfate is produced as a saleable product. In any of the cases, the issue of the presence of As and its further management route were not discussed. Recently, Kim et al. [28] described the recovery of Bi(III) and Sb(III) from the valorization of anode slimes, where after some initial stages of Cu recovery and Se removal, the solid containing Sb(III) and Bi(III) was treated with two different MRT resins and concluded that it could be an alternative to produce both metals from wastes. From the separation and pre-concentration options evaluated, the liquid-liquid extraction options are promising from an economic, environmental and sustainability point of view. The use of combination of liquid-liquid extraction systems and selective precipitation as main processes for Bi(III)/Sb(III) recovery from IX eluates is the most relevant options. Even limited research projects, as it is the case of this work, proposes the selective precipitation as a technological option for the Sb(III)/Bi(III) removal from IX eluates, but its high removal ratios 36 justify further research. Pilot-scale experiments for Bi(III)/Sb(III) recovery from IX eluates had already been conducted and this emerging option has shown economic liability. Further research is needed to demonstrate the performance of evaluated schemes at full scale. 5. Conclusions It has been demonstrated that Cyanex 923 solutions in kerosene have shown the selective extraction of Sb(III) and Bi(III) although has been initially developed and applied for H2SO4 solutions. Results had shown higher selectivity of Cyanex 923 towards Sb(III) and Bi(III) over As(V) and specially over HCl. The quasi-selective extraction of Sb(III) and Bi(III), minimizing the As co-extraction, could be achieved by reducing the concentration of Cyanex 923 up to values of 0.10 M. Although three extraction stages could be necessary, the increasing of extraction stages is having as benefit the reduction of the residual contents of As(V) in the extraction stages. Regeneration of Cyanex 923 should be performed preferentially with 2M HNO3, as the use of H2SO4 is facing limitations on the solubility of Bi(III) in concentrated H2SO4 solutions. From the results of the study, it has been demonstrated the possibility to valorize a waste that is generated in the pyrometallurgical copper industry. The developed process allows to move from a linear management model based on CaO(s) addition to a circular model based on the use of solvent extraction to selectively recover Sb and Bi. Both elements are listed as Critical Raw Materials by the EU, but also included in the Strategic Element List of USA. By applying the proposed treatment to the Atlantic Copper metallurgical complex, which produces 284000 tn pure copper per year, it could be possible to recover up to 130 tn/year of Bi and 100 tn/year of Sb. 37 Finally, it should be mentioned that potential recovery schemes of the excess of HCl from the eluate or from the stripping solutions should be analyzed in subsequent stages because of its economic concerns in the consumption of neutralization chemicals as NaOH and to promote circularity schemes to recover HCl on-site (e.g. recovery in the AMPIX resin regeneration stages). The use of cheapest alkali reagents as CaO(s) or CaCO3(s) is limited by the coextraction of H2SO4 by Cyanex 923. Acknowledgments This research was supported by the RECOPP Project (PI-19119, H2020-EIT Raw Materials), W4V project (PID2020-114401RB-C21) financed by the Spanish Ministry of Science and Innovation, by the R2MIT project (CTM2017-85346-R) financed by the Spanish Ministry of Economy and Competitiveness, and by the Catalan Government (ref. 2017-SGR-312), Spain. Nabil Benabdallah gratefully acknowledges to the Ministry of Higher Education and Scientific Research, Algeria its PhD fellowship (PNE), the University of Science and Technology of Oran (USTOMB), Algeria and the EEBE, Universitat Politècnica de Catalunya. The authors are also thankful to the General Directorate of Scientific Research and Technological Development of Algeria for sponsoring this work. D. Luo acknowledges China Scholarship Council for her PhD grant (CSC 202008420263). Additionally, the authors acknowledge the Open Innovation - Research Translation and Applied Knowledge Exchange in Practice through UniversityIndustry Cooperation (OpenInnoTrain), Grant agreement number (GAN): 823971, H2020MSCA-RISE-2018-823971. Finally, authors thanks to A. Mejías, I. Ruiz and G. Rios from Atlantic Copper, samples supply of the ion-exchange eluate as well and the information provided on the analytical characterization techniques as well as the inputs provided along the development of the experimental work. 38 References [1] M. Schlesinger, M. King, K.C. Sole, W.G. Davenport, Extractive metallurgy of copper, 5th ed., Elsevier Science, 2011. https://doi.org/https://doi.org/10.1016/C2010-0-648413. [2] International Copper Study Group, The World Copper Factbook 2021, (2021). https://icsg.org/wp-content/uploads/2021/11/ICSG-Factbook-2021.pdf (accessed January 12, 2022). [3] B. Wesstrom, The effects of high antimony in electrolyte, in: Bill Davenport Honorary Symposium, Conference of Metallurgists, Canadiuan Istitute of Mining, Metallurgy and Petroleum, Montreal, 2014: p. 8527. [4] X. Wang, Q. Chen, Z. Yin, M. Wang, B. Xiao, F. Zhang, Homogeneous precipitation of As, Sb and Bi impurities in copper electrolyte during electrorefining, Hydrometallurgy. 105 (2011) 355–358. https://doi.org/10.1016/j.hydromet.2010.10.004. [5] X. Wang, X. Wang, B. Liu, M. Wang, H. Wang, X. Liu, S. Zhou, Promotion of copper electrolyte self-purification with antimonic oxides, Hydrometallurgy. 175 (2018) 28–34. https://doi.org/10.1016/j.hydromet.2017.10.028. [6] J.B. Hiskey, Mechanism and thermodynamics of floating slimes formation, in: T.T. Chen Honorary Symposium on Hydrometalturgy, Electrometallurgy and Materials Characterizatio, TMS (The Minerals, Metals & Materials Society), 2012: pp. 101–112. https://doi.org/10.1002/9781118364833.ch8. [7] S. Jafari, M. Kiviluoma, T. Kalliomäki, E. Klindtworth, A.T. Aji, J. Aromaa, B.P. Wilson, M. Lundström, Effect of typical impurities for the formation of floating slimes in copper electrorefining, Int J Miner Process. 168 (2017) 109–115. https://doi.org/10.1016/j.minpro.2017.09.016. [8] R.R. Moskalyk, A.M. Alfantazi, Review of copper pyrometallurgical practice: Today and tomorrow, Miner Eng. 16 (2003) 893–919. https://doi.org/10.1016/j.mineng.2003.08.002. [9] G. Ríos, R. Ramírez, C. Arbizu, Management of Electrolyte Impurities at the Atlantic Copper Refinery, in: GDMB (Ed.), Copper 2010, 2010: pp. 1599–1616. 39 [10] O.V.J. Hyvarinen, Process for Selective Removal of Bismuth and Antimony from an Electrolyte, Especially in Electrolytic Refining of Copper, US4157946A, 1977. [11] K. Ando, N. Tsuchida, Recovering Bi and Sb from electrolyte in copper electrorefining, The Journal of The Minerals, Metals & Materials Society (TMS). 49 (1997) 49–51. https://doi.org/10.1007/s11837-997-0033-4. [12] P. Navarro, F.J. Alguacil, Adsorption of antimony and arsenic from a copper electrorefining solution onto activated carbon, Hydrometallurgy. 66 (2002) 101–105. https://doi.org/10.1016/S0304-386X(02)00108-1. [13] I. Ruiz, G. Rios, C. Arbizu, I. Burke, U. Hanschke, Pilot tests on Bismuth and Antimony removal from electrolyte at Atlantic Copper refinery. Atlantic Copper SL (Subsidiary of Freeport McMoran Copper and Gold), in: Proceedings of European Metallurgical Conference 2013, 2013: pp. 1–15. [14] K. Toyabe, C. Segawa, H. Sato, Impurity control of electrolyte at Sumitomo Niihama Copper Refinery, in: Electrorefining and Winning of Copper, 1987: pp. 117–128. [15] X. Wang, Q. Chen, Z. Yin, P. Zhang, Z. Long, Z. Su, Removal of impurities from copper electrolyte with adsorbent containing antimony, Hydrometallurgy. 69 (2003) 39–44. https://doi.org/10.1016/S0304-386X(03)00026-4. [16] P. Navarro, J. Simpson, F.J. Alguacil, Removal of antimony (III) from copper in sulphuric acid solutions by solvent extraction with LIX 1104SM, Hydrometallurgy. 53 (1999) 121–131. https://doi.org/10.1016/S0304-386X(99)00033-X. [17] P.A. Riveros, J.E. Dutrizac, R. Lastra, A study of the ion exchange removal of antimony(III) and antimony(V) from copper electrolytes, Canadian Metallurgical Quarterly. 47 (2008) 307–316. https://doi.org/10.1179/cmq.2008.47.3.307. [18] H. Sato, Method of removing antimony from an antimony-containing copper electrolyte, US-4444666-A, 1984. [19] R.M. Cunningham, J.V. Calara, M.G. King, Removal of antimony and bismuth from copper electrolyte development of a commercial plant at Amarillo copper refinery, in: EPD Congress 1997, Proceedings of Sessions and Symposia Held at the TMS Annual Meeting, Minerals, Metals & Materials Society: Warrendale, Orlando, 1997: pp. 453– 460. 40 [20] R.L. Bruening, J.B. Dale, N.E. Izatt, W.; Young, The application of molecular recognition technology (MRT) for the removal of impurities and the recovery of metals in copper electrorefining and electrowinning, in: Symposium Series - South African Institute of Mining and Metallurgy, 1996: pp. 45–54. [21] E. Dziwinski, J. Szymanowski, Composition of Cyanex 925, Solvent Extraction and Ion Exchange. 17 (1999) 1515–1522. [22] J. Szymanowski, Removal of Toxic Elements from Copper Electrolyte by Solvent Extraction, Mineral Processing and Extractive Metallurgy Review. 18 (1998) 389–418. https://doi.org/10.1080/08827509808914162. [23] C.M. Sevrain, M. Berchel, H. Couthon, P.A. Jaffrès, Phosphonic acid: Preparation and applications, Beilstein Journal of Organic Chemistry. 13 (2017) 2186–2213. https://doi.org/10.3762/bjoc.13.219. [24] S. Wang, Impurity control and removal in copper tankhouse operations, The Journal of The Minerals, Metals & Materials Society (TMS). 56 (2004) 34–37. https://doi.org/10.1007/s11837-004-0089-3. [25] F. Arroyo-Torralvo, A. Rodríguez-Almansa, I. Ruiz, I. González, G. Ríos, C. FernándezPereira, L.F. Vilches-Arenas, Optimizing operating conditions in an ion-exchange column treatment applied to the removal of Sb and Bi impurities from an electrolyte of a copper electro-refining plant, Hydrometallurgy. 171 (2017) 285–297. https://doi.org/10.1016/j.hydromet.2017.06.009. [26] A.B. Botelho Junior, A.D.A. Vicente, D.C.R. Espinosa, J.A.S. Tenório, Effect of iron oxidation state for copper recovery from nickel laterite leach solution using chelating resin, Sep Sci Technol. 55 (2020) 788–798. https://doi.org/10.1080/01496395.2019.1574828. [27] M. Cox, D. Flett, T. Velea, C. Vasiliu, Impurity removal from copper tankhouse liquors by solvent extraction, in: International Solvent Extraction Conference, Cape Town, South Africa, 2002: pp. 17–21. [28] D. Kim, S. Wang, Bismuth recovery from hydrochloric acid solution, Canadian Metallurgical Quarterly. 47 (2008) 317–326. https://doi.org/10.1179/cmq.2008.47.3.317. 41 [29] R.L. Bruening, J.B. Dale, N.E. Izatt, S.R. Izatt, Commercial separations in the copper industry using molecular recognition technology, in: Proceedings of the International Symposium Honoring Professor Ian M. Ritchie, Minerals, Metals & Materials Society, 2003: pp. 729–739. [30] D.B. Dreisinger, B.J.Y. Leong, B.R. Saito, E.-Sells. P.G., The solvent extraction and ion exchange removal of As, Sb and Bi from copper sulphate-sulphuric acid solutions, in: SME, 1993: pp. 801–815. [31] D.B. Dreisinger, B.J.Y. Scholey, Ion exchange removal of antimony and bismuth from copper refinery electrolytes, in: Proceedings of the Copper 95 International Conference, Santiago (Chile), 1995: pp. 26–29. [32] K.C. Sole, M.B. Mooiman, E. Hardwick, Ion Exchange in Hydrometallurgical Processing: An Overview and Selected Applications, Separation and Purification Reviews. 47 (2018) 159–178. https://doi.org/10.1080/15422119.2017.1354304. [33] A.I. González de las Torres, M.S. Moats, G. Ríos, A. Rodríguez Almansa, D. SánchezRodas, Removal of Sb impurities in copper electrolyte and evaluation of As and Fe species in an electrorefining plant, Metals (Basel). 11 (2021). https://doi.org/10.3390/met11060902. [34] L.Y. Kovalenko, F.A. Yaroshenko, V.A. Burmistrov, T.N. Isaeva, D.M. Galimov, Thermolysis of Hydrated Antimony Pentoxide, Inorganic Materials. 55 (2019) 586–592. https://doi.org/10.1134/S0020168519060086. [35] F.X. Xiao, D. Cao, J.W. Mao, X.N. Shen, F.Z. Ren, Role of Sb(V) in removal of As, Sb and Bi impurities from copper electrolyte, Transactions of Nonferrous Metals Society of China (English Edition). 24 (2013) 271–278. https://doi.org/10.1016/S10036326(14)63057-0. [36] F.X. Xiao, J.W. Mao, D. Cao, X.N. Shen, F.Z. Ren, Formation of antimonate in coprecipitation reaction of As, Sb and Bi in copper electrolytes, Miner Eng. 35 (2012) 9– 15. https://doi.org/10.1016/j.mineng.2012.05.001. [37] S. Beauchemin, T.T. Chen, J.E. Dutrizac, Behaviour of antimony and bismuth in copper electrorefining circuits, Canadian Metallurgical Quarterly. 47 (2008) 9–25. https://doi.org/10.1179/cmq.2008.47.1.9.