Membrane electrolysis for separation of cobalt from terephthalic acid industrial wastewater

Hydrometallurgy 191 (2020) 105216

Contents lists available at ScienceDirect

Hydrometallurgy journal homepage: www.elsevier.com/locate/hydromet

Membrane electrolysis for separation of cobalt from terephthalic acid industrial wastewater

T

Rui Gaoa,e, Xochitl Dominguez Benettonb,e, Jeet Variaa, Bernd Meesc,e, Gijs Du Laingc,d, ⁎ Korneel Rabaeya,d, a

Center for Microbial Ecology and Technology (CMET), Faculty of Bioscience Engineering, Ghent University, Coupure Links 653, Ghent 9000, Belgium Separation and Conversion Technologies, Flemish Institute for Technological Research (VITO), Boeretang 200, Mol 2400, Belgium c Department of Green Chemistry and Technology, Faculty of Bioscience Engineering, Ghent University, Coupure Links 653, Ghent 9000, Belgium d CAPTURE, Coupure Links 653, Ghent 9000, Belgium e SIM vzw, Technologiepark 48, Zwijnaarde 9052, Belgium b

A R T I C LE I N FO

A B S T R A C T

Keywords: Metals organics separation Metal containing wastewater Wastewater treatment Metal recovery Organics recovery Metal catalyst

Recovery of valuable metals from wastewaters containing both metals and organics is challenging with current technologies, in part due to their interactions. Typical approaches are chemical intensive. Here, we developed a membrane electrolysis system coupled to an acidic and alkaline crystallizer to enable separate precipitation of the organics and metals without additional chemicals. The target industrial wastewater contained mainly purified terephthalic acid (PTA), benzoic acid (BA), p-Toluic acid (PA), cobalt (Co), and manganese (Mn). We examined the removal and recovery efficiency of PTA and cobalt from two types of synthetic stream and the real process stream using several configurations. The acidic crystallizer reached a removal efficiency of PTA of 98.7 ± 0.2% (Coulombic efficiency 99.71 ± 0.2%, pH 3.03 ± 0.18) in batch tests of the simple synthetic stream. The alkaline crystallizer achieved a cobalt recovery efficiency of 94.51 ± 0.21% (Coulombic efficiency 87.67 ± 0.31%, pH 11.37 ± 0.21) in batch tests of the simple synthetic stream (TPA and Co). Then, the system was operated continuously with complex synthetic stream (TPA, BA, PA, Co and Mn). The alkaline crystallizer achieved a cobalt recovery efficiency of 97.78 ± 0.02% (Coulombic efficiency 90.45 ± 0.17%)at pH 11.68 ± 0.02. The acidic crystallizer obtained a PTA removal efficiency of 61.2 ± 0.1% (Coulombic efficiency 62.3 ± 0.2%) over 144 h (pH 3.71 ± 0.03). A real stream was tested over 5 h runs in batch showing 31.1 ± 1.0% PTA (Coulombic efficiency 26.5 ± 0.2%) and 82.92 ± 0.22% cobalt removal (Coulombic efficiency 75.27 ± 0.31%) at pH 2.71 ± 0.12 and 8.07 ± 0.02, respectively. However, micron-scale precipitates were generated from real stream tests. To conclude, the membrane electrolysis cell coupled with acidic and alkaline crystallizers enabled simultaneous separation of PTA and cobalt as solid precipitates from a complex stream with no chemical addition. The efficiencies were lower with the real stream than the synthetic streams, showing the impact of matrix effects and the need to optimize the performance of the crystallizers.

1. Introduction Modern industrial chemical production leads to > 70,000 unique products (Isenberg, 2014; Adeola, 2011). In many cases metallic catalysts are used to enable organic conversions, leading to a co-occurrence of residual metals along with residual organics process streams and wastewaters. Typical metals applied in such chemical production processes are mainly transition metals, such as iron, silver, platinum, cobalt, zinc, manganese, titanium, ruthenium, and molybdenum (Srour et al., 2013; Kestenbaum et al., 2002; Browder et al., 1990). Relevant

organics include ethylene glycol, para-toluic acid, ethylene, hexane, toluene, xylenes, benzoic acid, trimethylbenzenes, vinyl acetate, amino acids, methanol and glucose (Mitchell and Waring, 2000; Chanda and Roy, 2006; Reviews, 2017; Svelle et al., 2006; Kleerebezem and Lettinga, 2000). The co-occurrence of metals and organics can lead to organo-metallic complexes with enhanced toxicity in the environment due to their complexity, high chemical stability, and low biodegradability (Kingsley et al., 1994; Lin et al., 2007). The removal efficiency of recalcitrant metals from industrial effluents is also reduced by the presence of organics due to formation of stable complexes (Abdul et al.,

⁎ Corresponding author at: Center for Microbial Ecology and Technology (CMET), Faculty of Bioscience Engineering, Ghent University, Coupure Links 653, Ghent 9000, Belgium. E-mail address: [email protected] (K. Rabaey).

https://doi.org/10.1016/j.hydromet.2019.105216 Received 13 June 2019; Received in revised form 11 October 2019; Accepted 27 November 2019 Available online 30 November 2019 0304-386X/ © 2019 Elsevier B.V. All rights reserved.

Hydrometallurgy 191 (2020) 105216

R. Gao, et al.

contains the hydroxide anions produced at the cathode to increase the pH in the alkaline crystallizer, is combined with part of the acidic crystallizer effluent and transported to the alkaline crystallizer to precipitate cobalt. The precipitation of cobalt from wastewater follows equations (Lu et al., 1997; Badawy et al., 2000; Benchettara and Benchettara, 2015; Haynes, 2012; Furlanetto and Formaro, 1995) (1–2). Presumable reactions for organics and other metals can be found in the Supporting Information (SI) Eqs. (S1–6).

2000). Available treatment technologies have been developed to deal with organo-metallic complexes, including adsorption with granular activated carbon, ion exchange resin or biochar (Park et al., 2007; Duwiejuah et al., 2017; Guarino et al., 1988; Qiu et al., 2010), electrocoagulation combined with electro Fenton (Yasri and Gunasekaran, 2017; Kabdaşli et al., 2010; Kabdaşlı et al., 2012), membrane filtration (Sharma and Sanghi, 2012; Landaburu-Aguirre et al., 2009; Fu and Wang, 2011), advanced oxidation with UV irradiation or TiO2 (Awaleh and Soubaneh, 2014), combined reverse osmosis and electro-winning (Zhang et al., 2008), biosorption coupled to biodegradation (Cobas et al., 2015), ozonation (Huang et al., 2016), wetland technology with bacterial pre-treatment (Chandra et al., 2008), and electrodialysis (Chen, 2004; Abou-Shady et al., 2012; Banasiak and Schäfer, 2009). Although many of these technologies achieve considerable removal efficiency of metals or organics, the technologies have several limitations. Besides having low removal efficiencies due to complexation of organics, acid and base are often added for metal precipitation or pH adjustment. However, high operational costs, high-energy consumption, handling costs and the long-term environmental impacts of sludge disposal due to the chemicals used (Kurniawan et al., 2006; Aziz et al., 2008). Membrane electrolysis (ME) is an electrochemical technology used, for example, by the chlor-alkali industry for caustic production (MeliánMartel et al., 2011; Savari et al., 2008). Briefly, a fluid waste stream can be brought through an anode and/or a cathode, which are separated by an ion selective membrane. The placement of electrodes in the stream to-be-treated enables modifying the pH of this stream without chemical addition (Xu et al., 2015; Liu et al., 2015; Bormann and Moebius, 1992). Also, ME typically has a low maintenance cost (Regel-Rosocka, 2010; Csicsovszki et al., 2005; Paidar et al., 2016). ME is different from electrodialysis in that in-situ pH control enables removal of the resources rather than concentrating them in a second stream. ME is known to work well with non-charged, higher molecular mass, and less mobile ionic species, and as such can be applied for selective removal of pollutants in wastewater, such as metals, suspended solids, and aromatic organic compounds (Paidar et al., 2016; Applications of Adsorption and Ion Exchange Chromatography in Waste Water Treatment, 2017). Therefore, with reduced chemical costs and a feasible treatment for selective recovery of metals and organics, we have expanded the ME system with acidic and alkaline crystallizers. We aim to separate charged low molecular mass toxic organics while simultaneously recovering the dissolved metals contained in a typical industrial wastewater, without introducing hazardous chemicals. To achieve this, we work for a model wastewater, which is discharged by an industry producing purified terephthalic acid (PTA) at high temperatures (80, 100 and 120 °C) as the raw material of polyethylene terephthalate (PET) for the polyester industry to produce PET-bottles and film, containing mainly purified terephthalic acid (PTA), benzoic acid, p-Toluic acid, cobalt and manganese at concentrations given in Table 1. The recovery and recycling of both PTA and the metal catalysts has positive implications on chemical and water savings, besides rendering effluents with higher environmental quality. The protons produced in the anode compartment of ME cell are transported to an acidic crystallizer receiving the system influent to decrease the pH to below the pKa (3.51, 4.82) (Harper and Janik, 1970) of PTA. This leads to PTA precipitation in the acidic crystallizer. Simultaneously, the cathode effluent, which

Concentration (mg L−1)

Cobalt Manganese Purified Terephthalic acid (PTA) Benzoic acid p-Toluic acid

10 7 3000 280 770

(1)

4Co (OH)2 (aq) + 2H2 O + O2 ⇌ 4Co (OH)3 (s)

(2)

Several configurations were tested in preliminary tests with the synthetic stream with and without multiple ions. The optimized system configuration was determined on simpler synthetic streams to establish key parameters such as current density, pH variations, and flow rate before using a real stream. 2. Materials and methods 2.1. Medium composition For the development of the process, a synthetic wastewater was prepared containing 1.42 g L−1 Na2SO4, 0.14 g L−1 CoSO4.7H2O, and 0.2 g L−1 PTA (Thermofisher, USA). This is here referred as the “single ion” synthetic stream, since only PTA and cobalt are targeted. A synthetic wastewater with multiple ions (“multiple ions” stream) was prepared with 0.2 g L-1 PTA, 0.35 g L−1 benzoic acid (BA), 0.35 g L−1 p-toluic acid (PA), 1.7 g L−1 Na2SO4, 0.14 g L−1 CoSO4.7H2O and 0.13 g L−1 MnSO4.5H2O. Both synthetic streams were adjusted to pH 5.5 ± 0.5 using 0.1 M NaOH and both had an initial conductivity of 2.51 ± 0.02 mS cm−1. A solution of 1.42 g L−1 Na2SO4 with pH 7.52 ± 0.05 and conductivity 2.81 ± 0.03 mS cm−1 was used at the onset of each test as a supporting electrolyte for anolyte and catholyte. Experiments were performed at 80 °C, as this is the temperature of the real waste stream. The real wastewater stream was obtained from BP Chembel, and had a pH of 3.08 and conductivity 1.85 mS cm−1. It was stored at 4 °C and reheated prior to use. The total chemical oxygen demand (COD) of the wastewater was 3800 mg O2 L−1 of which 1900 mg O2 L−1 was considered to be dissolved after filtering with a 0.45 μm filter. Table 1 shows the target components in the real stream: 2.2. Experimental approach Before evaluating the separation of PTA and cobalt from the target real stream, tests were performed on the single ion synthetic stream, in batch mode, to establish an optimal configuration with regards to the removal and recovery efficiencies of PTA and cobalt. In an initial set of experiments, we used system I (Supporting Information (SI) Fig. S1), which included two membranes separating anode and cathode (an anion exchange membrane between anode and middle compartment, and a cation exchange membrane between middle compartment and cathode) and directed the influent fluid through the middle compartment to simultaneously separate the cationic metals and the anionic organics from the synthetic stream. System II (Supporting Information (SI) Fig. S2) used only one membrane (Cation exchange membrane) separating anode and cathode chamber. Before running the experiment with system II, a pretreatment was firstly done using only supporting electrolyte containing Na2SO4 to decrease or increase the pH in anode and cathode, correspondingly. Then, the single ion stream was firstly fed into an acidic crystallizer to precipitate PTA and then circulated to the anode chamber. At the same time, cobalt ions were transferred to cathode chamber and recovered as precipitate. After preliminary tests with system I and system II, an optimized system III (Fig. 1) combined membrane electrolysis with both an acidic and alkaline crystallizers, wherein the both the synthetic streams and real stream were treated. In

Table 1 Main compositions in the target real stream. Target components

Co2 + (aq) + 2OH− ⇌ Co (OH)2 (s)

2

Hydrometallurgy 191 (2020) 105216

R. Gao, et al.

Fig. 1. Schematic of System III ( two compartment cell with an acidic and alkaline crystallizer),

with flow rate 0.6 L h

with flow rate 0.1 L h

with flow

rate 0.5 L h-1; 1: Influent, 2: Influent with outlet of anode chamber; 3: Outlet of acidic crystallizer; 4: Influent to the alkaline crystallizer; 5: Outlet of Cathode chamber; 6: Outlet of alkaline crystallizer; 7: Outlet of Anode chamber.

2.4. Reactor operation

system III, PTA and cobalt were removed from single and multi-ions synthetic streams and recovered separately as precipitates in the two different crystallizers. The pretreatment operated in system II was also performed by system III.

Before operating tests, the synthetic stream and real stream were heated (Stirrer MST Basic, lab World yellow line, Belgium) to 80 °C. The electrolyte in the alkaline crystallizer was heated to 45 ± 5 °C for balancing temperature of system. For the preliminary tests with the system I (three compartment electrochemical cell), the pH of the catholyte or anolyte (1 L) was firstly adjusted to 9 or 2, using 1 M NaOH or 0.5 M H2SO4, respectively. After that, the single ion synthetic stream (1 L) was sent to the middle compartment. All compartments were recirculated at 0.1 L min−1. Tests were performed at 50 A m−2 in batch mode and samples were taken every 30 min at the outlet of the middle, anode, and cathode chamber. For the preliminary tests with system II (Supporting Information (SI) Fig. S2), prior to the start the electrolytes were sent through the electrode compartments and the corresponding acidic crystallizer at room temperature (25 °C) for 3 h at 50 A m−2, with an internal recirculation rate of 0.1 L min−1, which enabled decreasing the pH in an acidic crystallizer and anode chamber, and increasing pH in cathode chamber and buffer tank. Then, the synthetic stream was introduced first to the bottom of a 4 L Perspex conical funnel acidic crystallizer (40 cm × 25 cm × 25 cm) at a flow rate of 0.02 L min−1. After precipitation in the acidic crystallizer, the single ion stream was transferred to the anode chamber at a rate of 0.02 L min−1 and internally recirculated through the anode chamber at a flow rate of 0.1 L min−1. The catholyte was recirculated through the cathode at 0.1 L min−1. Influent and effluent of both anode and cathode were sampled. System III (Fig. 1) was developed based on the performance of system I and system II. Pre-treatment with system III was performed both for the anode, acidic crystallizer, cathode and alkaline crystallizer at 50 A m−2 with a recirculation rate of 0.008 L min−1 over 7 h. The 4 L Perspex conical funnel was used as acidic crystallizer and a 5 L Duran Schott bottle (Sigma-Aldrich, Germany) was used as an alkaline crystallizer. The synthetic stream was sent to the acidic crystallizer at a flow rate of 0.01 L min−1, the effluent from the acidic crystallizer was sent to both anode chamber and alkaline crystallizer at a flow rate of 0.008 L min−1 and 0.002 L min−1, respectively. In batch mode, the stream from the anode was sent back to acidic crystallizer and mixed with the synthetic stream influent for precipitation. At the same time, the mixture of the effluent from cathode compartment and acidic

2.3. Electrochemical cell The electrochemical cell was comprised of plexiglass frames bolted together between two plexiglass side plates as described previously (Desloover et al., 2015). System I had three frames (Supporting Information (SI) Fig. S1) (inner dimensions 5 cm × 20 cm × 2 cm), the anode compartment was separated from the middle compartment by an anion exchange membrane (AEM, AMI-7001 T, Membranes International, USA), the middle from the cathode by a cation exchange membrane (CEM, CMI-7001 T, Membranes International, USA) (Lin et al., 2016; Vaiopoulou et al., 2016), and the frames had internal dimensions of 5 cm × 20 cm × 2 cm. A titanium (Ti) electrode coated with iridium mixed metal oxide (Ir MMO) mesh was used as anode (dimensions: 5 cm × 20 cm × 2 mm, Magneto Special Anodes BV, The Netherlands). The cathode was a stainless steel fiber felt (5 cm × 20 cm × 1 mm, 100 μm filtering rate, Lier Filter Ltd., China). These layers were sandwiched between rubber sheet seals and held together between the two bolted plexiglass plates. Both electrodes were positioned close to the AEM or CEM with a projected surface area of 5 × 20 cm2. For the electrochemical cell applied in system II and system III, a lower volume two-chamber electrochemical cell was used (internal dimensions 1.25 cm × 8 cm × 2 cm). Anode and cathode were each 3 mm diameter of A titanium (Ti) electrode coated with iridium mixed metal oxide (Ir MMO) electrode rod with 8.5 cm active length (Magneto Special Anodes BV, The Netherlands). A Ag/AgCl reference electrode (3 M KCl, +0.197 vs. SHE/V, BASi, France) was used to measure the applied potential at the respective electrodes. The cell voltage and applied potential at either electrode were measured using a digital multimeter (BST-9205 M, MacManiack, The Netherlands). The constant current was fixed by a power supply (LABPS3005D, Velleman, Belgium), the voltage of system was recorded by a data acquisition system (LABPS3005D PC software V2.5).

3

Hydrometallurgy 191 (2020) 105216

R. Gao, et al.

At the same time, the concentration of cobalt ions decreased in the middle compartment, but the cobalt was not found in the solution circulating through the cathode compartment (Supporting Information (SI) Table S1). It has been shown in previous studies (Simons, 1985; Oda and Yawataya, 1968) that multiple reactions can occur adjacent to the surface of a cation exchange membrane, where hydroxyl ions concentrate. From EDS analysis of the CEM, cobalt ions became trapped in the membrane rather than transferring to cathode chamber (Supporting Information (SI) Fig. S5). One of the major issues observed with system I was the increase in voltage over time, from 7.57 ± 0.03 to 27.8 ± 0.2 V, which relates to the depletion of ions and thus a decrease of conductivity in the middle compartment from 3.87 ± 0.02 to 2.64 ± 0.02 mS cm−1 (Support information (SI) Fig. S6). Besides, the voltage increased was probably attributed to a rising internal resistance of the cell due to the cobalt trapped in the CEM in short-term batch test. Therefore, the System I was improved with an extra acidic crystallizer to prevent PTA precipitate from anode chamber, and investigating the cobalt in middle and cathode chamber through long-term test. System II was operated at 50 A m−2 for 270 min per batch cycle, and resulted in a decrease in the pH of anode compartment from 4.88 ± 0.03 to 2.37 ± 0.01 (Support information (SI) Fig. S7). PTA precipitate was observed in both the acidic crystallizer and the anode compartment (pH 4.57 ± 0.01) as a white solid after 30 min operation. The cobalt ions still remained on the CEM instead of migrating to the cathode compartment over the longer experiment time, which was confirmed by SEM and EDS image of the CEM (Supporting Information (SI) Fig. S8a, S8b). According to the EDS spectrum analysis from duplicated experiments (Supporting Information (SI) Fig. S9a, S9b), the atomic concentration of the cobalt on the surface of CEM was 9.15 ± 1.45% respect to total atomic concentration of the membrane. Besides, the atomic concentration of oxygen and carbon were 65.8 ± 0.7 and 23.6 ± 0.5%, respectively. The conductivity (Supporting Information (SI) Fig. S10) in the cathode compartment rose from 5.03 ± 0.02 to 16.8 ± 0.1 mS cm−1 due to increased pH. At the same time, conductivity in the anode compartment maintained between 2.90 ± 0.06 and 3.65 ± 0.02 mS cm−1 instead of a significant increase caused by the precipitation of PTA scavenging protons. Therefore, the voltage observed in system II decreased over time from 7.20 ± 0.02 to 6.59 ± 0.03 V, due to the increase of the conductivity in both anode and cathode compartments. To prevent accumulation of cobalt on the CEM and PTA precipitation in the anode compartment, System III was developed for simultaneous separation of PTA and cobalt. In System III, an alkaline crystallizer was introduced to precipitate cobalt out of cathode chamber rather than trapping in CEM. Besides, the pre-treatment time of System III was increased from 3 h (System II) to 7 h to maintain the initial pH below the pKa of PTA.

crystallizer was sent to alkaline crystallizer and internally recirculated at a flow rate of 0.008 L min−1. For the continuous tests (Supporting Information (SI) Fig. S3), both anode and cathode compartment were fed continuously to achieve a hydraulic retention time (HRT) of 4.2 h. The effluent was collected from the acidic and alkaline crystallizers. All experiments were performed in duplicate. The removal efficiency of PTA and cobalt and the coulombic efficiencies (CEPTA and CECobalt) were calculated based on the Supporting Information (SI) Eqs. S1 and S2, respectively. 2.5. Chemical and instrumental analysis The conductivity and pH were measured using a Consort EC and pH electrode, respectively (Consort, Turnhout, Belgium). The PTA concentration was determined by HPLC (P680, DIONEX) with UV-detector (UVD170U, DIONEX) at the wavelength of 240 nm (Murai et al., 1988). First, all samples were diluted to a concentration lower than 20 mg L−1 with methanol (20/80 v/v). The concentration of dissolved PTA was subsequently analyzed onto a C18 column (3 μm, 150 × 2 mm, Phenomenex B.V., Netherlands) kept at 40 °C. The analytical conditions used were as follows: flow 0.4 mL min− 1, eluent 98% methanol with 0.1% (v/v) Trifluoroacetic acid. The injection volume for each sample was 20 μL. Inductively Coupled Plasma Mass Spectrometry (ICP-MS) was used for analyzing the concentration of metals in samples (Vanhoe et al., 1989). Samples for ICP-MS (Perkin Elmer Nexion 350S, USA) were diluted 100 times in 1% HNO3 before analysis. Scanning electron microscopy (SEM) (Phenom ProX, Thermo fisher scientific, USA) images were recorded at an acceleration voltage of 15 keV with a JSM-7600F field emission SEM instrument. Energy-dispersive X-ray spectroscopy (EDS) was used to study the elemental composition of the cobalt and manganese precipitates obtained from experiments. Samples were mounted on aluminum stubs by means of a double-sided adhesive carbon tape and sputtered with Au to reduce charging effects. X-ray diffraction (XRD) was carried out for metals speciation analysis on a Bruker D8 Discover XRD system equipped with a Cu X-ray source (λ = 1.5406 Å) and a linear X-ray detector. During XRD analysis, the precipitate powder was spread on a silicon wafer and placed on the sample heating stage. The XRD sample was heated from room temperature to 450 K (176.85 °C) and cooled back to room temperature (RT) at a heating or cooling rate of 5 K/min. 3. Results and discussion 3.1. Initial system operation In system I, the synthetic stream was fed to the middle compartment to remove dissociated terephthalate (C8H4O42−)towards the anode, and cobalt ions (Co2+) towards the cathode. At 50 A m−2, the pH in the middle compartment initially decreased to 2.95 ± 0.03 in the first 60 min operation followed by an increase to 4.37 ± 0.04 after 90 min operation (Support information (SI) Fig. S4). Theoretically, the pH in the middle compartment should remain stable. However, the PTA precipitate in both anode and middle compartment was consuming protons, so there were less protons remaining in the middle compartment causing back diffusion to the middle compartment from the anode while much hydroxyl anions remained in the middle compartment. The pH in the middle compartment decreased in the first 60 min indicating that proton cross-over from the anode was higher than hydroxide backdiffusion from cathode (Lin et al., 2016). Terephthalate (PTA−) precipitated as a white solid (PTA, C8H4O4) and was first observed in both middle and anode compartments after 30 min, when the pH in the middle compartment reached 4.73 ± 0.02. Also, the pH in the anode (2.00 ± 0.12) was well below the pKa of PTA (4.82) (Thamer and Voigt, 1952) which exceeded its solubility resulting in its precipitation.

3.2. Simultaneous PTA and cobalt separation from synthetic streams The pH of the single ion synthetic stream, which was analyzed after heating at 80 °C, was 5.88 ± 0.01. After 7 h of pre-treatment with the background electrolyte (prepared at pH 7.52 ± 0.05), the initial pH of the sampling points at the outlet of the acidic crystallizer and the anode compartment were 3.58 ± 0.04 and 3.52 ± 0.02, respectively. The pH at the outlet of the acid crystallizer remained stable until the end of the experiment. The pH of the anode compartment outlet decreased to 3.03 ± 0.18 by the end of experiment. The pH at the alkaline crystallizer outlet was 11.0 ± 0.1 after the pre-treatment and this value was maintained until the end of the experiment, despite the influx of acidic crystallizer effluent. The initial pH at the cathode compartment outlet was 11.4 ± 0.1 after pre-treatment then increased to 11.7 ± 0.2 after 30 min operation and remained constant until the end of the experiment (Fig. 2A). The conductivity of all sampling points remained stable between 2.39 ± 0.05 to 3.13 ± 0.13 mS cm−1 (Fig. 2B). The concentration of terephthalate in the single ion synthetic stream 4

Hydrometallurgy 191 (2020) 105216

R. Gao, et al.

Fig. 2. Determination of performance of System III with the synthetic stream without multiple ions, pH (A), EC (B), concentration of terephthalate (C) and concentration of cobalt (D), 1: Influent, 2: Influent with outlet of anode chamber; 3: Outlet of acidic crystallizer; 4: Influent to the alkaline crystallizer; 5: Outlet of cathode chamber; 6: Outlet of alkaline crystallizer; 7: Outlet of Anode chamber.

influent was 199 ± 1 mg L−1 (Fig. 2C), while the concentration of terephthalate at the outlet of the acidic crystallizer was only 6.06 ± 0.27 mg L−1 after 7 h operation, which shows that the low pH of the acidic crystallizer enables a PTA removal efficiency of 97.1 ± 0.1%. At the same time, the concentration of dissolved terephthalate at the outlet of the anode compartment was 2.06 ± 0.88 mg L−1, which indicated that part of the terephthalate precipitated in the anode compartment (Fig. 2C). The overall removal efficiency of PTA after the anode compartment was thus 98.7 ± 0.2% (Coulombic efficiency 99.71 ± 0.2%). Compared to existing studies on PTA removal with chemical or biological methods, our system has a higher removal efficiency than could be achieved, for example, with adsorption materials, chemicals, and a mixed microbial community (Sun et al., 2018; Verma et al., 2017; 2013; Li et al., 2014), with the added benefit of the possibility to recycle the PTA precipitate into the industrial process. The initial concentration of cobalt in the single ion synthetic stream was 29.5 ± 2.50 mg L−1. After the acidic crystallizer, the concentration of cobalt in the stream sent to the alkaline crystallizer decreased to 24.5 ± 1.5 mg L−1 (Fig. 2D) showing complexation with PTA. Although most of the residual terephthalate in the stream out of acidic crystallizer was protonated at pH 3.41 ± 0.10 (pKa of PTA 3.51, 4.82) (Verma et al., 2013) and the PTA concentration was already down to 2.06 ± 0.88 mg L−1, in our modeling about Co speciation in stream (Supporting Information (SI), Fig. S11), at least 0.06 mg L−1 Co(terephthalate) and CoH(terephthalate)+ can be formed under the pH from 3 to 7. Thus, the loss of cobalt can be attributed to the complexation

with deprotonated PTA but we also consider that our crystallization is at this moment not optimal (quite disperse crystals) in which water can be trapped containing cobalt as well, or the system requires subsequent unit operations like gravimetric separations with a hydrocyclone or a centrifuge. The concentration of cobalt in the outlet of cathode compartment and alkaline crystallizer dropped below 5.00 mg L−1 (1.70 ± 0.10 mg L−1) and remained at this value until the end of the experiment (Fig. 2D). Cobalt precipitated as a light brown solid in the alkaline crystallizer in the first 3 h, after which a dark brown floccular precipitate accumulated in the bottom of the alkaline crystallizer. This lead to a recovery efficiency of cobalt of 94.5 ± 0.2% (Coulombic efficiency 87.67 ± 0.31%) from the alkaline crystallizer. Theoretically, cobalt ions first precipitate as Co(OH)2 (blue) at pH 8.6 at a temperature above 40 °C. However, Co3O4 or Co(OH)3 can be formed through oxidation as brownish precipitate at high potentials or in the presence of sufficient oxygen (Chivot et al., 2008; Pataki et al., 2013). The speciation of cobalt was detected by XRD analysis (Supporting Information (SI) Fig. S12) indicating an amorphous cobalt precipitate (an oxide or a mixture of elements). EDS spectrum analyses (Supporting Information (SI) Fig. S13) also indicated that the main composition of the precipitate from the alkaline crystallizer was cobalt. Unfortunately, the specific speciation of cobalt in the precipitate was not feasible due to the uncertainty in the measurements. Cobalt did not accumulate on the AEM as observed in the earlier configurations, this was confirmed by EDS. This indicates that an alkaline crystallizer can prevent cobalt trapped in the CEM and accumulation in the EC (Supporting Information (SI) Fig. S14).

5

Hydrometallurgy 191 (2020) 105216

R. Gao, et al.

crystallizer. It is possible to increase the removal efficiency by decreasing the temperature of the stream due to lower solubility of PTA at lower temperatures (Supporting Information (SI), Fig. S17) (Li et al., 2014; Takebayashi et al., 2012), in practice this would be an option due to anticipated crystallization within heat exchangers. Otherwise, an optimization of the acid crystallizer is necessary to minimize side effects in crystallizer or replacement by e.g., an in-line centrifuge. The pH at the outlet samples from both of the alkaline crystallizer and cathode compartment was 11.7 ± 0.1 after the pre-treatment and was stable until the end of the experiment (Fig. 3A). Cobalt precipitated as dark brown solids and accumulated at the bottom of alkaline crystallizer after 2 h. The concentration of cobalt in the outlet of cathode compartment and alkaline crystallizer decreased from 29.7 ± 0.6 mg L−1 to 1.10 ± 0.10 mg L−1 and remained at this value until the end of the experiment (Fig. 3D). This leads to a recovery efficiency of cobalt of 97.8 ± 0.1% (Coulombic efficiency 62.3 ± 0.2%) from the alkaline crystallizer. However, the precipitate that accumulated in the alkaline settler was not only cobalt but also manganese, since manganese can be precipitated as Mn(OH)2, Mn (OH)3, Mn(OH)4, Mn3O4 or Mn2O3 depending on different potentials when the pH is above 9 (Huang, 2016). This was confirmed by the XRD and EDS analysis (Fig. 3E, F). In EDS analysis, 19.1% of manganese and 18.5% of cobalt were identified from the precipitate sample from the alkaline crystallizer. Again, according to the XRD analysis, it is hard to identify the specific speciation of Mn and Co in the precipitate due to their amorphous nature. Furthermore, we also observed a black metal precipitate from the acidic settler and anode compartment, likely due to manganese oxidation to MnO2, which can occur at low pH and high potential (above 1.4 V) (Kao and Weibel, 1992; Tripathy et al., 2001). As shown in Fig. S15a-b, the main composition of manganese precipitate from acidic settler was Mn or mixture of Mn and Co in the anode compartment. Therefore, the concentrations of cobalt in the outlet of the acidic crystallizer and anode compartment decreased from 29.7 ± 0.6 mg L−1 to 11.2 ± 0.4 mg L−1 and 10.4 ± 0.4 mg L−1, respectively (Fig. 3D). The manganese precipitate in the acidic crystallizer and anode compartment presents a challenge to identify and separate the PTA precipitate. Also, the complexation of manganese and PTA might have an influence on removal of both metals and organics. Besides this, the low concentrations of divalent manganese (at 0.1 to 1 mg L−1) can dissolve in the aqueous solution instead of precipitate (HEM, 1963). Therefore, minimizing the concentration of the divalent manganese in the stream might reduce the MnO2 precipitation in the system. Electrodeposition of Mn under an acidic condition can be an effective option to lower MnO2 precipitate. Through Pourbaix diagram, oxidation potential of Mn2+ is 0.8 V vs. SHE which is lower than oxidation potential of water (1.23 V vs. SHE) (Klahr et al., 2012; Yi and Majid, 2018). Therefore, Mn2+ can be oxidized and precipitated as MnO2 before water is oxidized to produce proton. Then, pH of stream will be stable, and terephthalate will not be precipitated as pH is not at an optimal value. Hence, an extra electrolysis cell can be introduced into System III to separate manganese precipitate from PTA precipitate.

Table 2 Concentrations of compositions in the synthetic stream with multiple ions. Target components

Concentration (mg L−1)

Cobalt Manganese Terephthalic acid (TPA) Benzoic acid p-Toluic acid

30 10 200 350 350

The first single ion synthetic stream was limited in terms of metal and organic ions present and operating time; this is far from a realistic situation. To investigate the effects of multiple ions on PTA removal and cobalt recovery, a multiple ions synthetic stream containing more target ions was applied over a longer operating time with our system (Table 2). As shown in Fig. 3A, the initial pH of the synthetic stream was 4.94 ± 0.04 at 80 °C and remained stable until the end of the experiment. After 7 h of pre-treatment with background electrolyte, the initial pH at the outlet samples of the anode compartment and acidic crystallizer were 3.57 ± 0.02 and 3.34 ± 0.02, respectively (Fig. 3A). These gradually increased to a stable value of 4.09 ± 0.03 and 3.71 ± 0.03, respectively until the end of the experiment. However, the pH at the outlet samples of both acidic crystallizer and anode compartment increased more steadily comparing with the pH in the experiment with the synthetic stream only containing terephthalate and cobalt, as consumption of protons occurred by the buffering capacity of competitive organic anions in the stream. The conductivity at the outlets of the cathode and alkaline settler increased from 3.61 ± 0.04 to 4.83 ± 0.06 mS cm−1, (Fig. 3B) which had a similar trend to the previous experiments. However, the conductivity of the sampling points out of the anode compartment and acidic crystallizer decreased sharply from 4.44 ± 0.03 to 1.33 ± 0.05 mS cm−1 over the first 27 h. Afterward, the conductivity was stable at 1.22 ± 0.01 mS cm−1 until the end of experiment. These results can be attributed to multiple acids precipitation with protons produced by the electrochemical cell. The concentration of terephthalate in the influent of the multiple ions synthetic stream was 198 ± 1 mg L−1 (Fig. 3C); the initial concentration of terephthalate at the outlet of the acidic crystallizer was 46.6 ± 0.1 mg L−1, revealing that the acidic crystallizer can remove the majority of terephthalate, at 76.6 ± 0.1%. At the same time, the concentration of terephthalate at the anode outlet compartment decreased to 25.0 ± 0.1 mg L−1, which indicated that more precipitation occurred in the anode compartment (bringing it to a total of 87.4 ± 0.1% removal and coulombic efficiency achieved 81.2 ± 0.3%). However, at the end of the experiment, the concentration of terephthalate at the outlet of the acidic crystallizer increased to 132 ± 1 mg L−1 (33.5 ± 0.1%). Also, the concentration of terephthalate at the anode compartment outlet was 77.2 ± 0.1 mg L−1 (removal efficiency of 61.2 ± 0.1%, coulombic efficiency 62.3 ± 0.2%). The concentration of benzoate and p-Toluate in the outlet of acidic crystallizer and anode compartment was also analyzed and can be found in Supporting Information (SI) Fig. S15a-b. Besides, we observed precipitate accumulating around the anode electrode as a white foam. According to EDS analysis, the main composition of the white foam was organic precipitate (Supporting Information (SI), Fig. S16). The removal efficiency of PTA in this experiment is lower than the experiment without multiple ions. The removal efficiency of PTA decreased with complex organics after a long-term continuous experiment has been discussed by some researches (Kleerebezem et al., 1999; Ghimire and Wang, 2018; Jafarzadeh et al., 2006; Sandhwar and Prasad, 2017), however, none of them gave a brief explanation on it. The reasons might be attributed to a weak interaction between PTA and protons, consumption of considerable amount of charges during electrolysis through some side reactions (other organics precipitate and Mn precipitate) and the matrix effects of accumulated organics in the acidic

3.3. Performance of PTA and cobalt separation with real wastewater Finally, the real stream (Table 1) was tested over 5 h periods due to a limited volume of real stream available; thus, this should be considered as a proof of concept only. As shown in Fig. 4A, the pH of the real stream was 3.14 ± 0.11, which was lower than the prepared synthetic streams. Prior to the experiment, considerable organics precipitation was observed in the real stream due to the low-temperature storage and low pH. Reheating to 80 °C for 3 h before experiment was sufficient to re-dissolved these organics completely (Harper and Janik, 1970). After the pre-treatment, the pH of the sampling points out of the anode compartment and acidic settler was only 2.71 ± 0.12. The initial conductivity of all sampling points was 2.21 ± 0.41 mS cm−1 but 6

Hydrometallurgy 191 (2020) 105216

R. Gao, et al.

Fig. 3. Determination of performance of System III with the synthetic stream with multiple ions, pH (A), EC (B), concentration of terephthalate (C) and concentration of cobalt (D), 1: Influent, 2: Influent with outlet of anode chamber; 3: Outlet of acidic crystallizer; 4: Influent to the alkaline crystallizer; 5: Outlet of cathode chamber; 6: Outlet of alkaline crystallizer; 7: Outlet of Anode chamber; 8: Effluent out of the acidic crystallizer; 9: Effluent out of the alkaline crystallizer. The Scanning electron microscopy (SEM) analysis of cobalt and manganese precipitate from alkaline crystallizer (E), the X-ray spectroscopy (EDS) analysis of cobalt and manganese precipitate from alkaline crystallizer.

gradually decreased to 1.75 ± 0.08 mS cm−1 over 5 h (Fig. 4B). The concentration of dissolved PTA decreased to 180 ± 1 mg L−1 from 248 ± 1 mg L−1, after acidic crystallizer (Fig. 4C), which indicated that the acidic settler was still capable of PTA removal even with the more complex real stream. The removal efficiency of PTA from acidic settler was only 25.1 ± 0.1%, which is similar to the synthetic stream with multiple ions. There is a clear need for optimization. The concentration of PTA from the sampling point out of the anode compartment was 170 ± 1 mg L−1, which revealed the PTA can be further

precipitated in the anode compartment (bringing total removal to 31.1 ± 1.0% and coulombic efficiency reached 26.5 ± 0.2%). A limited amount of solid precipitate of PTA was observed in the bottom of the acidic crystallizer. The removal efficiency of PTA obtained from the real stream experiment was thus overall lower compared with synthetic stream experiments applied in this study, we attribute this to the presence of more types of organic acids in the real stream (> 10). Future experiments should focus on longer term operation, high current densities, and improving the crystallizer design. 7

Hydrometallurgy 191 (2020) 105216

R. Gao, et al.

Fig. 4. Determination of performance of System III with the real stream, pH (A), EC (B), concentration of terephthalate (C) and concentration of cobalt (D), 1: Influent, 2: Influent with outlet of anode chamber; 3: Outlet of acidic crystallizer; 4: Influent to the alkaline crystallizer; 5: Outlet of cathode chamber; 6: Outlet of alkaline crystallizer; 7: Outlet of Anode chamber; 8: Effluent out of the acidic crystallizer; 9: Effluent out of the alkaline crystallizer.

Information (SI), Fig. S14), most of cobalt can be precipitated as Co (OH)2 at low temperature (30 °C) over a pH 10.8. Therefore, cooling temperature down can be an option in our next research to increase recovery efficiency of cobalt before sending stream to cathode chamber.

The pH of the outlet sample of alkaline crystallizer was 7.85 ± 0.05 initially but steadily decreased to 7.31 ± 0.02 by the end of the experiment. The pH from the cathode compartment was stable at 8.07 ± 0.02 until the end of experiment (Fig. 4A). There was no solid precipitate observed in the alkaline crystallizer at the pH below 9 (Chivot et al., 2008). However, we observed that the stream in the alkaline crystallizer turned yellow after 27 h. The initial concentration of cobalt in the process stream was 10 mg L−1. After the cathode compartment, the concentration of cobalt decreased to 2.41 ± 0.05 mg L−1. The concentration of cobalt out of the alkaline crystallizer was 2.22 ± 0.02 mg L−1(brings removal efficiency of cobalt to 82.92 ± 0.22% and coulombic efficiency reached 75.27 ± 0.31%). Although the concentration of cobalt in the stream decreased over time, the solid precipitate was only visually observed on the surface of a filter used on the samples from the alkaline settler, highlighting that only very small particles formed the size of which was not determined here. Manganese precipitates were not observed in both acidic and alkaline crystallizer. A key bottleneck of the real stream experiment is the low concentration of cobalt in the influent here, which was lower than in earlier measurement points. The precipitation efficiency of cobalt decreased with decreasing the initial concentration of cobalt in the stream from 0.001 to 0.002 mg L−1 (Aktas et al., 2013). Besides, the precipitation we obtained in the alkaline crystallizer such as Co(OH)2 and Co3O4 are amorphous precipitation. Part of amorphous precipitation might be sent to cathode chamber then precipitated further. Therefore, we would like to consider crystallization mechanism of Co(OH)2 and Co3O4 for compositional analysis and crystals in alkaline crystallizer. Through the model of the dissolved cobalt in target stream sample at different temperatures and over a pH range (Supplementary

4. Conclusion A membrane electrolysis system combined with acidic and alkaline crystallizers was developed and tested. The novel system design resulted in the simultaneous separation of PTA and cobalt without additional chemicals or adsorption materials. The system achieved an excellent removal efficiency of PTA 98.7 ± 0.2% (Coulombic efficiency 99.71 ± 0.2%) in a simple stream a lower efficiency of 61.2 ± 0.1% (Coulombic efficiency 62.3 ± 0.2%) with multiple ions. Also, the recovery efficiency of cobalt from the stream with or without multiple ions is 94.5 ± 0.2% (Coulombic efficiency 87.67 ± 0.31%) and 97.8 ± 0.1% (Coulombic efficiency 62.3 ± 0.2%), respectively. The system performed better relative to reported systems applied in the treatment of the synthetic stream containing single ions (PTA and cobalt), due to the absence of additional acids or base and extra electrochemical or biological reactors. However, the system with the real process stream achieved a low removal efficiency of PTA at 31.1 ± 1.0% (bring coulombic efficiency to 26.5 ± 0.2%). The cobalt in the process stream is hard to be precipitate under the conditions applied, into valuable forms. Therefore, the system still needs further study to improve the stability of the system performance with real process streams. Also, further studies should assess prevention of the manganese precipitate in the acidic crystallizer (or co-precipitation with cobalt into valuable mixed metal oxides or hydroxides) and the 8

Hydrometallurgy 191 (2020) 105216

R. Gao, et al.

influence of manganese on PTA removal. The speciation of cobalt in the cobalt precipitate could be further studied to investigate the impact of membrane electrolysis on the metal speciation.

HEM, J.D., 1963. Chemical equilibria affecting the behavior of manganese in natural water. Hydrol. Sci. J. 8 (3), 30–37. Huang, H.-H., 2016. The Eh-pH diagram and its advances. Metals 6 (1), 23. Huang, X., et al., 2016. Coupled Cu(II)-EDTA degradation and Cu(II) removal from acidic wastewater by ozonation: performance, products and pathways. Chem. Eng. J. 299, 23–29. Isenberg, A.C., 2014. The Oxford Handbook of Environmental History. Oxford University Press. Abdul, J.C., et al., 2000. Simultaneous recovery of heavy metals and degradation of organic species – copper and ethylenediaminetetra-acetic acid (EDTA). J. Chem. Technol. Biotechnol. 75 (5), 353–358. Jafarzadeh, M., et al., 2006. Petrochemical wastewater treatment using an anaerobic hybrid reactor. Pak. J. Biol. Sci. 9 (6), 1037–1042. Kabdaşli, I., et al., 2010. Organic matter and heavy metal removals from complexed metal plating effluent by the combined electrocoagulation/Fenton process. Water Sci. Technol. 61 (10), 2617–2624. Kabdaşlı, I., et al., 2012. Electrocoagulation applications for industrial wastewaters: a critical review. Environ. Technol. Rev. 1 (1), 2–45. Kao, W.H., Weibel, V.J., 1992. Electrochemical oxidation of manganese(II) at a platinum electrode. J. Appl. Electrochem. 22 (1), 21–27. Kestenbaum, H., et al., 2002. Silver-catalyzed oxidation of ethylene to ethylene oxide in a microreaction system. Ind. Eng. Chem. Res. 41 (4), 710–719. Kingsley, J.P., A.K. Roby, and L.M. Litz, Terephthalic acid production. 1994, Google Patents. Klahr, B., et al., 2012. Electrochemical and photoelectrochemical investigation of water oxidation with hematite electrodes. Energy Environ. Sci. 5 (6), 7626–7636. Kleerebezem, R., Lettinga, G., 2000. High-rate anaerobic treatment of purified terephthalic acid wastewater. Water Sci. Technol. 42, 259–268. Kleerebezem, R., Hulshoff Pol, L.W., Lettinga, G., 1999. The role of benzoate in anaerobic degradation of terephthalate. Appl. Environ. Microbiol. 65 (3), 1161–1167. Kurniawan, T.A., et al., 2006. Physico–chemical treatment techniques for wastewater laden with heavy metals. Chem. Eng. J. 118 (1–2), 83–98. Landaburu-Aguirre, J., et al., 2009. The removal of zinc from synthetic wastewaters by micellar-enhanced ultrafiltration: statistical design of experiments. Desalination 240 (1–3), 262–269. Li, X.K., et al., 2014. Performance and microbial community profiles in an anaerobic reactor treating with simulated PTA wastewater: from mesophilic to thermophilic temperature. Water Res. 61, 57–66. Lin, R., R.S. O'meadhra, and R.B. Sheppard, Process for the production of purified terephthalic acid. 2007, Google Patents. Lin, H.-W., et al., 2016. Direct anodic hydrochloric acid and cathodic caustic production during water electrolysis. Sci. Rep. 6, 20494. Liu, X., et al., 2015. Study on extraction of lithium from salt lake brine by membrane electrolysis. Desalination 376, 35–40. Lu, J., Dreisinger, D., Cooper, W., 1997. Cobalt precipitation by reduction with sodium borohydride. Hydrometallurgy 45 (3), 305–322. Melián-Martel, N., Sadhwani, J.J., Ovidio Pérez, S., 2011. Báez, Saline waste disposal reuse for desalination plants for the chlor-alkali industry. The particular case of pozo izquierdo SWRO desalination plant. Desalination 281 (1), 35–41. Mitchell, S., Waring, R., 2000. Ullmanns Encyclopedia of Industrial Chemistry. WileyVCH Verlag, Weinheim, Germany. Murai, S., et al., 1988. Rapid determination of norepinephrine, dopamine, serotonin, their precursor amino acids, and related metabolites in discrete brain areas of mice within ten minutes by HPLC with electrochemical detection. J. Neurochem. 50 (2), 473–479. Oda, Y., Yawataya, T., 1968. Neutrality-disturbance phenomenon of membrane-solution systems. Desalination 5 (2), 129–138. Paidar, M., Fateev, V., Bouzek, K., 2016. Membrane electrolysis—history, current status and perspective. Electrochim. Acta 209, 737–756. Park, H.G., et al., 2007. Activated carbon-containing alginate adsorbent for the simultaneous removal of heavy metals and toxic organics. Process Biochem. 42 (10), 1371–1377. Pataki, L., et al., 2013. Basic Analytical Chemistry. Elsevier Science. Qiu, T., Han, S., Wu, Y., 2010. Recovery of Co (II) and Mn (II) from pure terephthalic acid wastewater. J. Chem. Eng. Data 55 (7), 2399–2404. Regel-Rosocka, M., 2010. A review on methods of regeneration of spent pickling solutions from steel processing. J. Hazard. Mater. 177 (1), 57–69. Reviews, C., 2017. Environmental Sociology. (Cram101). Sandhwar, V.K., Prasad, B., 2017. A comparative study of electrochemical treatment of Para-Toluic acid (p-TA) from aqueous solution: process optimization and sludge analysis. Water Conserv. Sci. Eng. 1 (4), 257–270. Savari, S., Sachdeva, S., Kumar, A., 2008. Electrolysis of sodium chloride using composite poly(styrene-co-divinylbenzene) cation exchange membranes. J. Membr. Sci. 310 (1–2), 246–261. Sharma, S.K., Sanghi, R., 2012. Wastewater Reuse and Management. Springer, Netherlands. Simons, R., 1985. Water splitting in ion exchange membranes. Electrochim. Acta 30 (3), 275–282. Srour, H., et al., 2013. Metal-catalyzed asymmetric sulfoxidation, epoxidation and hydroxylation by hydrogen peroxide. Coord. Chem. Rev. 257 (21−22), 3030–3050. Sun, F., et al., 2018. High efficient alternating anaerobic/aerobic process for polyester resin wastewater treatment: performance and microbial community structure. Biochem. Eng. J. 138, 121–130. Svelle, S., et al., 2006. Conversion of methanol into hydrocarbons over zeolite H-ZSM-5: Ethene formation is mechanistically separated from the formation of higher alkenes. J. Am. Chem. Soc. 128 (46), 14770–14771. Takebayashi, Y., et al., 2012. Solubility of terephthalic acid in subcritical water. J. Chem.

Declaration of Competing Interest The author declares no conflict of interest. Acknowledgements All authors thank the support from the Flemish SIM MaRes programme, under grant agreement no. 150626 (Get-A-Met project). The authors thank Xu Zhang, Linsey Lapeire and Davy Deduytsche for the electrochemical, SEM-EDX and XRD analyses. The authors thank Amanda K. Luther and Xiaoyan Sun for critically reading the manuscript. The authors have no competing interests to declare. Appendix A. Supplementary data Supporting Information to this article can be found online at https://doi.org/10.1016/j.hydromet.2019.105216. References Abou-Shady, A., et al., 2012. Effect of pH on separation of Pb (II) and NO3− from aqueous solutions using electrodialysis. Desalination 285, 46–53. Adeola, F.O., 2011. Hazardous Wastes, Industrial Disasters, and Environmental Health Risks: Local and Global Environmental Struggles. Palgrave Macmillan US. Aktas, S., et al., 2013. Investigation of cobalt sulfate precipitation by alcohol and influencing factors. Miner. Metall. Process. 30 (3), 174–179. Applications of Adsorption and Ion Exchange Chromatography in Waste Water Treatment. Materials Research Forum LLC. Awaleh, M.O., Soubaneh, Y.D., 2014. Waste water treatment in chemical industries: the concept and current technologies. Hydrol. Curr. Res. 5 (1), 1. Aziz, H.A., Adlan, M.N., Ariffin, K.S., 2008. Heavy metals (Cd, Pb, Zn, Ni, Cu and Cr (III)) removal from water in Malaysia: post treatment by high quality limestone. Bioresour. Technol. 99 (6), 1578–1583. Badawy, W., Al-Kharafi, F., Al-Ajmi, J., 2000. Electrochemical behaviour of cobalt in aqueous solutions of different pH. J. Appl. Electrochem. 30 (6), 693–704. Banasiak, L.J., Schäfer, A.I., 2009. Removal of boron, fluoride and nitrate by electrodialysis in the presence of organic matter. J. Membr. Sci. 334 (1), 101–109. Benchettara, A., Benchettara, A., 2015. Improved D-glucose electro-oxidation at a binary catalyst of nickel and cobalt oxide modified glassy carbon electrode. Russ. J. Electrochem. 51 (9), 881–885. Bormann, G., Moebius, A., 1992. Membrane electrolysis for maintenence of hard chromium plating baths. A contribution to drastic reduction of effluent and sludge volumes. Galvanotechnik 83 (1), 93–96. Browder, L.W., J.C. Medlin, and A.T. Spaugh Jr, Extraction process for removal of impurities from terephthalic acid filtrate. 1990, Google Patents. Chanda, M., Roy, S.K., 2006. Plastics Technology Handbook. CRC press. Chandra, R., et al., 2008. Bacterial pretreatment enhances removal of heavy metals during treatment of post-methanated distillery effluent by Typha angustata L. J. Environ. Manag. 88 (4), 1016–1024. Chen, G., 2004. Electrochemical technologies in wastewater treatment. Sep. Purif. Technol. 38 (1), 11–41. Chivot, J., et al., 2008. New insight in the behaviour of Co–H2O system at 25–150°C, based on revised Pourbaix diagrams. Corros. Sci. 50 (1), 62–69. Cobas, M., et al., 2015. Removal of metal and organic pollutants from wastewater by a sequential selective technique. Bioresour. Technol. 213, 2–10. Csicsovszki, G., Kékesi, T., Török, T.I., 2005. Selective recovery of Zn and Fe from spent pickling solutions by the combination of anion exchange and membrane electrowinning techniques. Hydrometallurgy 77 (1–2), 19–28. Desloover, J., et al., 2015. Electrochemical nutrient recovery enables ammonia toxicity control and biogas desulfurization in anaerobic digestion. Environ. Sci. Technol. 49 (2), 948–955. Duwiejuah, A., Cobbina, S., Bakobie, N., 2017. Review of eco-friendly biochar used in the removal of trace metals on aqueous phases. Int. J. 5 (2), 27–40. Fu, F., Wang, Q., 2011. Removal of heavy metal ions from wastewaters: a review. J. Environ. Manag. 92 (3), 407–418. Furlanetto, G., Formaro, L., 1995. Precipitation of spherical Co3O4 particles. J. Colloid Interface Sci. 170 (1), 169–175. Ghimire, N., Wang, S., 2018. Biological treatment of petrochemical wastewater. In: Petroleum Chemicals-Recent Insight. IntechOpen. Guarino, C.F., et al., 1988. Activated carbon as an advanced treatment for petrochemical wastewaters. Water Sci. Technol. 20 (10), 115–130. Harper, J.J., Janik, P., 1970. Terephthalic acid solubility. J. Chem. Eng. Data 15 (3), 439–440. Haynes, W.M., 2012. CRC Handbook of Chemistry and Physics, 93rd ed. Taylor & Francis.

9

Hydrometallurgy 191 (2020) 105216

R. Gao, et al.

Verma, S., Prasad, B., Mishra, I.M., 2017. Treatment of purified terephthalic acid wastewater using a bio-waste-adsorbent bagasse fly ash (BFA). Environ. Sci. Pollut. Res. 24 (2), 1953–1966. Xu, J., et al., 2015. In-line and selective phase separation of medium-chain carboxylic acids using membrane electrolysis. Chem. Commun. 51 (31), 6847–6850. Yasri, N.G., Gunasekaran, S., 2017. Electrochemical Technologies for Environmental Remediation. In: Anjum, N.A., Gill, S.S., Tuteja, N. (Eds.), Enhancing Cleanup of Environmental Pollutants: Volume 2: Non-Biological Approaches. Springer International Publishing, Cham, pp. 5–73. Yi, C., Majid, S., 2018. The Electrochemical Performance of Deposited Manganese OxideBased Film as Electrode Material for Electrochemical Capacitor Application. Zhang, L., et al., 2008. Electro-winning and membrane combination technical for treatment and reuse PCB manufactory sewage. In: Proceedings of the 2nd International Conference on Asian-European Environmental Technology and Knowledge Transfer.

Eng. Data 57 (6), 1810–1816. Thamer, B.J., Voigt, A.F., 1952. The spectrophotometric determination of overlapping dissociation constants of dibasic acids. J. Phys. Chem. 56 (2), 225–232. Tripathy, B.C., Singh, P., Muir, D.M., 2001. Effect of manganese(II) and boric acid on the electrowinning of cobalt from acidic sulfate solutions. Metall. Mater. Trans. B Process Metall. Mater. Process. Sci. 32 (3), 395–399. Vaiopoulou, E., et al., 2016. Electrochemical sulfide removal and caustic recovery from spent caustic streams. Water Res. 92, 38–43. Vanhoe, H., et al., 1989. Determination of iron, cobalt, copper, zinc, rubidium, molybdenum, and cesium in human serum by inductively coupled plasma mass spectrometry. Anal. Chem. 61 (17), 1851–1857. Verma, S., Prasad, B., Mishra, I.M., 2013. Treatment of petrochemical wastewater by acid precipitation and carbon adsorption. J. Hazardous Toxic Radioact. Waste 18 (3) (04014013).

10