Electrochemical Induced Calcium Phosphate Precipitation: Importance of Local pH.

Phosphorus (P) is an essential nutrient for living organisms and cannot be replaced or substituted. In this paper, we present a simple yet efficient membrane free electrochemical system for P removal and recovery as calcium phosphate (CaP). This method relies on in situ formation of hydroxide ions by electro mediated water reduction at a titanium cathode surface. The in situ raised pH at the cathode provides a local environment where CaP will become highly supersaturated. Therefore, homogeneous and heterogeneous nucleation of CaP occurs near and at the cathode surface. Because of the local high pH, the P removal behavior is not sensitive to bulk solution pH and therefore, efficient P removal was observed in three studied bulk solutions with pH of 4.0 (56.1%), 8.2 (57.4%), and 10.0 (48.4%) after 24 h of reaction time. While P removal efficiencies are not generally affected by bulk solution pH, the chemical-physical properties of CaP solids collected on the cathode are still related to bulk solution pH, as confirmed by structure characterizations. High initial solution pH promotes the formation of more crystalline products with relatively high Ca/P molar ratio. The Ca/P molar ratio increases from 1.30 (pH 4.0) to 1.38 (pH 8.2) and further increases to 1.55 (pH 10.0). The formation of CaP precipitates was a typical crystallization process, with an amorphous phase formed at the initial stage which then transforms to the most stable crystal phase, hydroxyapatite, which is inferred from the increased Ca/P molar ratio from 1.38 (day 1) to the theoretical 1.76 (day 11) and by the formation of needle-like crystals. Finally, we demonstrated the efficiency of this system for real wastewater. This, together with the fact that the electrochemical method can work at low bulk pH, without dosing chemicals and a need for a separation process, highlights the potential application of the electrochemical method for P removal and recovery.

Introduction

Phosphorus (P) is an irreplaceable nutrient, but it is also associated with eutrophication.[1−4] Indeed, on the one hand, a large amount of P is discharged to surface waters resulting in eutrophication due to limited recycling.[3] On the other hand, the quantity and the quality of P ore has declined in the past decades because of P rock mining for producing fertilizers.[1] Evelyn Desmid’s calculation which applied the data of U.S geological survey 2012, suggests that natural P reserves will be fully depleted in 372 years if current mining rates are maintained.[1] In addition, considering the uneven distribution of P rock reserves, there may arise a P shortage for countries that completely depend on importing P rock in the near future.[4,5] The potential P shortage along with P discharge associated eutrophication, has created increased awareness of the importance of P recycling.[1,4,6] For instance, the Swedish government has set a national goal to recover at least 40% of P in wastewater treatment plants.[7]

There are many P removal methods available,[1,8−10] but efficient and economically feasible P recovery methods are quite limited. Among the few methods, struvite (MgNH4PO4·6H2O) formation and precipitation is regarded as one of the most promising ways.[6,11−13] Struvite, which is a slow-release fertilizer, shows higher bioavailability than iron and aluminum phosphate.[10] However, it is necessary to supply a Mg source to assist struvite formation,[13−15] which makes the struvite process less economically attractive because of the low concentration of Mg2+ in wastewater.[15] Alternatively, calcium phosphate (CaP), which is the mined component in P rock, would be a better solution.[16,17] CaP solids can form without adding Ca2+ since there is often already sufficient Ca2+ in bodies of water.[18] Therefore, P recovery via CaP formation and precipitation is a preferred method, and has received a lot of attention.[17,19]

CaP precipitation is a very complex process. In general, the process is controlled by the chemical species in solution, including Ca and P concentrations and pH.[20−22] To induce CaP precipitation, the solution needs to be highly supersaturated. The typical way to create a supersaturated condition is by adding caustic soda to increase solution pH. However, because wastewater normally has a considerable buffering capacity because of the presence of organic acids and inorganic carbonates, significant base addition is needed in order to increase the bulk solution pH to a certain level that would induce CaP precipitation. For instance, as reported by Jaffer et al.,[23] the sodium hydroxide addition is accounted for up to 97% of the total chemical costs associated with P recovery by struvite formation method. Furthermore, the traditional chemical precipitation based methods produce a large quantity of sludge, which still needs to be treated before recycling.[24]

Recently, (bio)electrochemical processes were suggested as next generation technologies for treating (in)organic polluted water[25] and recognized as an efficient strategy for nutrient removal and recovery from nutrient rich wastewater.[26] Though (bio)electrochemical reactions are quite complicated processes, they can be simply divided as anode oxidation and cathode reduction. Most environment related electrochemical applications depend on the processes at the anode. The well-established electro-Fenton method for degrading organic pollutants is a good example.[27] By contrast, the role of cathode mediated reduction, has just begun to be explored for remediation and recovery by environmental scientists.[28] The (bio)electrochemical induced P removal and recovery as struvite has been well-documented.[29−31] However, the electrochemical assisted struvite formation, like chemical precipitation still relies on dosing of costly Mg2+. Moreover, the importance of local pH at the cathode with respect to electrochemical P recovery has not been recognized yet. Most studies mention that the increased pH is responsible for the precipitation of phosphate salts but none, to the best of our knowledge, has investigated the role of local pH in detail. This is because it is difficult to measure the local pH directly, as there still are no reliable pH sensors for detecting the electrochemically induced local pH at the electrode surface, though there are some special designed lab tools.[32,33] Moreover, the importance of local pH was seemingly ignored. Some researchers equate bulk solution pH to local pH and therefore just record bulk solution pH and use it as the pH for phosphate salts precipitation.[29,34] Consequently, the reported results with respect to local pH varied from experiment to experiment. As an example, Wang et al.[29] reported the slight increase of pH near the cathode from 7.0 to 7.5 as the cause for pure struvite formation in their electrochemical system. However, the local pH can be much higher than can be measured.[33]

To the best of our knowledge the electrochemical induced CaP precipitation on the cathode has not been reported, in terms of P removal and recovery and at various bulk pH. Although CaP coverage of the cathode might seem unwanted, we see this as an opportunity to separate P from waste streams with low P concentrations. Therefore, the purpose of this study, was to evaluate the efficiency of a single electrochemical cell without membrane for P removal and recovery by forming CaP precipitates. The importance of a local high pH in the electrochemical cell was demonstrated by evaluating the performance of this system at low, higher and high bulk solution pH combined with theoretical calculations. Finally, the efficiency and the cost for treatment of real wastewater were addressed to evaluate the potential for this new P recovery system.

Materials and Methods

Materials

All chemicals used here were at least reagent-grade. Disodium monohydrogen phosphate (Na2HPO4) and sodium sulfate anhydrous (Na2SO4) were purchased from VWR (Leuven, Belgium). Calcium nitrate tetrahydrate (Ca(NO3)2·4H2O) was received from Merck (Germany). Electrodes were provided by MAGNETO Special Anodes BV (Schiedam, The Netherlands).

Electrolysis Setup

The electrochemical cell consisted of two compartments, one working cell (500 mL) for CaP precipitation and one tank cell (500 mL) for mixing and sampling. The total solution in the two compartments (1000 mL) is circulated with a pump at a flow rate of 100 mL/min. The anode material is platinum coated (50 g/m2) titanium mesh with a round shape (Ø 10 cm, thickness 0.1 cm) and it is perpendicularly welded to a 10 cm long Ti rod (Ø 0.3 cm). The cathode is a pure titanium plate similarly welded (grade A, Ø 8.2 cm, thickness 0.1 cm). A pH sensor was placed in the sampling tank to record bulk solution pH change. In some cases, the pH electrode (Ø 1.2 cm, Endress Hauser, Germany) was also placed near the cathode (∼1.0 mm), in order to record local pH. The pH sensors were calibrated weekly. The diagram of the set sup is shown in Supporting Information (SI) Figure S1.

Electrolysis Experiments

The electrochemical precipitation process was conducted with 0.6 mM P and 1.0 mM Ca under constant current (20 mA) conditions and at constant ionic strength mediated by 50 mM Na2SO4. The choice for a sulfate salt was made because it does not interfere with the precipitation of CaP and does not produce harmful chlorine gas as well. While the initial Ca concentration is close to its natural concentration, the initial P concentration was higher compared to real wastewater in order to collect sufficient solid samples for further characterization. Where appropriate, the bulk solution pH was adjusted by concentrated NaOH or HNO3. Unless specified, the electrolysis process was open to air and lasted for 24 h at room temperature. The bulk solution pH was monitored during the whole process and logged by a computer program (Liquisys M CPM 253, Endress + Hauser, Naarden, The Netherlands).

Calcium Phosphate Collection

After the reaction was stopped at a predetermined time, the solutions in the electro cell were carefully removed with a syringe as to not disturb CaP precipitates at the cathode surface, for the sake of solid characterization. Then the electrode with precipitates on its surface was air-dried in room temperature. After drying, CaP solids were harvested by light scraping. After sampling, the cathode was immersed in a 1.0 M HNO3 solution to remove any CaP remaining and then thoroughly rinsed with Milli-Q water and dried again for use.

Analytical Methods

The concentrations of P and Ca ions were analyzed by ICP-AES. X-ray diffraction (XRD) was used to identify the crystal structure (or absence thereof if amorphous) and collected on a Bruker D8 advanced diffractometer equipped with a Vantec position sensitive detector and with a Co Kα radiation (λ = 0.179 nm) over a range of 10–70° in 0.02 step sizes with an integration time of 0.5 s. Raman spectra were obtained using a LabRAM HR Raman spectrometer from Horiba Jobin Yvon to obtain bonding information on collected solids. This system is equipped with a mpc3000 laser emitting at 532.2 nm and an 800 mm focal length achromatic flat field monochromator (grating of 600 grooves·mm–1). The laser beam was focused on the sample with an Olympus Bx41 microscope equipped with a 50× objective lens, which gives a spot size ca.1–2 μm and resolution of 6 cm–1. The detector is a Synapse multichannel air cooled (−70 °C) CCD. The applied laser power was between 5 and 50 mW (using density filters). The measurement time varied 5–30 s. Finally, the data were processed with LabSpec software. The morphology of collected products and their elemental compositions were examined by a scanning electron microscope (SEM) coupled with energy dispersive X-ray spectroscopy (EDS) (JEOL-6480LV, JEOL Ltd., Japan). Samples were coated with gold using a JEOL JFC-1200 Fine coater at 10 Pa for 30 s.

Calculations

The degree of saturation (Ω) and saturation index (SI) of a solution regarding a mineral phase, are defined as follows:[35]Where IAP refers to the ion activity of the associated lattice ions and Ksp is the thermodynamic solubility product. The computer program visual MINTEQ[36] was applied to calculate SI, as an indication for the potential saturation of possible products. Ca and P fractions were acquired by using Hydra–Medusa database.[37]

Based on Faraday’s law of electrolysis assuming that the electricity consumed was 100% used for water reduction and meanwhile supposing the produced OH was not consumed by other occurring reactions and was homogeneously mixed in the local layer, the theoretical maximum local pH, with respect to the thickness of local layer (δ, m) and electrolysis time (t, s) can be calculated by eqs , 4, and 5:I electricity current (A); z number of electrons transferred in the reaction, z = 1; F Faraday constant 96,485 (C/mol); d diameter of cathode (d = 0.082 m). It should be noted here that the real local pH will be below the theoretical calculated value because the current efficiency is unlikely to reach 100% and the electrochemically produced H+ at anode will react with OH to a certain extent.

Results and Discussion

Effects of Initial Bulk pH (pH0)

As a proof of principle, recovery of P in the electrochemical system was evaluated at three pH values including background solution pH (pH0 ∼ 8.2) after mixing of all chemicals, weak acidic (pH0 4.0) and alkaline (pH0 10.0) conditions. As can be seen from Figure A, under open circuit conditions, only 20% of P was removed in the case of pH0 10.0 and there was no obvious P removal at pH0 4.0 and 8.2. For pH0 4.0, the solution was undersaturated with respect to hydroxyapatite (HAP) (SIHAP = –15.5) and with respect to any calcium solid species like gypsum (Figure B). In addition, the calculation of the species distribution showed that nearly 87% of Ca existed as dissolved CaSO4 and P was almost 100% present as H2PO4 (SI Figure S2). Therefore, it is not surprising that no CaP precipitated from solution at pH0 4.0. In terms of pH0 8.2, while the thermodynamic calculation indicates the solution is supersaturated with respect to HAP (SIHAP = 8.6) and the fraction calculation also suggests the formation of HAP (SI Figure S2), no visible precipitates were found in reactors. Actually, many lakes are also supersaturated with respect to HAP without HAP being found in the lake sediments.[35] Indeed, thermodynamic predictions for precipitation of certain solids do not imply that they are kinetically favorable. The precipitation rate may be too slow to be observed and precipitation may progress via the Ostwald Step Rule. Interestingly, it was found that the application of a low current (20 mA, current density corresponds to 3.79 A/m2) makes a big difference for removal of P. The P removal efficiencies jumped to over 48% in all cases; 56.1%, 57.4%, and 48.4% of P was removed at pH0 4.0, pH0 8.2 and pH0 10.0 respectively (Figure A) within 24 h. We found that approximately 50% of Ca was removed as well. The simultaneous removal behavior of Ca and P indicates the removal as CaP precipitates. The precipitated solids were characterized with XRD and Raman spectroscopy. The Raman spectrum (Figure C) of the three samples almost all show internal bands of CaP, including a main ν1 PO43– peak around 955 cm–1 and well isolated ν2 PO43– (∼425 cm–1) and ν4 PO43– (∼590 cm–1) peaks, which clearly demonstrates the formation of CaP.[38,39] Interestingly, the XRD patterns (Figure D) suggest amorphous products are produced in acid and neutral solution as confirmed by the lack of sharp peaks and the presence of a broad peak around 38° though at pH0 10.0, a relatively more crystalline product is formed. The sharp peak around 30° indicates the presence of more crystalline CaP phases. However, the product is still dominantly amorphous. Most of the sharp peaks of pH0 10.0, unfortunately, is attributed to Na2SO4 because the electrode was air-dried without rinsing.

Figure 1

(A) Effects of initial pH on P and Ca removal efficiency. (B) Supersaturation index calculated from Visual MINTEQ. (C) Raman and (D) XRD patterns of recovered solid products. (E) Change of solution pH in open and closed circuit. Conditions: [Ca(NO3)2·4H2O] = 1.0 mM; [Na2SO4] = 50 mM;[Na2HPO4] = 0.6 mM; Current = 20 mA, Time = 24 h.

While it is not surprising that P was removed in an alkaline solution, the high removal efficiency of P at pH0 4.0 was not expected. As seen from Figure B, the solution at pH0 4.0 is undersaturated for all possible CaP products. The only factor that can contribute to the increase of SI here could be the increase of pH. By contrast, Figure E shows the solution pH decreases largely for pH0 8.2 and pH0 10.0, in which the solution pH drops to 4.6 and 4.0 respectively. Regarding pH0 4.0, it also declines to 3.4 after 24 h’ reaction. It should be noted here that under open circuit the solution pHs also drop to some extent due to equilibration with atmospheric CO2 in an open system (Figure E). In conclusion, it may be reasonable to infer that bulk solution pH is not that important, in terms of P removal efficiency.

Importance of Local pH

A phenomenon that we observed during our experiments is that precipitates just formed at/near the surface of the cathode. This points to different conditions at the cathode surface than in the bulk solution. The possible differences could be pH, Ca, and P concentration, which determines the saturation of CaP species in our system. Indeed, electro migration could transfer negative ions to the anode and positive ions to cathode. However, because the relative low concentration of Ca2+ compared to electrolyte (50 mM Na2SO4), it is unlikely that Ca2+ can be enriched to such extent that it can increase the saturation state of CaP. Also, if electro migration plays an important role here, the P concentration in the vicinity of cathode surface should decline correspondingly. Therefore, it is concluded that electro migration of ions does not play a crucial role in this system. The only possible reason for precipitation should then be attributed to the production of OH by electrochemical mediated water reduction at the cathode:Though the produced OH will diffuse to the bulk solution and the diffusion rate will increase with mixing rate, the relatively high pH in the very vicinity of cathode will not disappear.[40] While we did not have special pH sensors to record local pH, an attempt was made to measure the local pH by a general pH sensor. Indeed, a big difference was found between bulk solution pH and the so measured local pH, as shown in Figure A. For example, in 1 h, while the solution pH dropped to 7.4 from 8.2, the local pH went up to 9.9. However, as the measurement of local pH by this method is sensitive to the distance between the sensor and cathode, it is difficult to record a consistent pH. Consequently, the trend of local pH changes a lot. Indeed, though we did not measure the exact thickness of the precipitation layer, it is supposed that the local crystallization zone ranges to less than 1 mm away from the cathode surface, which was proven by a simple test. When we put a glass plate (26 × 26 × 1 mm) on the cathode surface, covering 12.8% of the cathode, there was no precipitates initiated from the glass surface. This showed that CaP precipitation just take places in the local region of the Ti cathode where the surface pH is much higher than the bulk solution pH because of the electrochemical production of hydroxide ions. Considering the size of a regular pH sensor as used in our experiments and the thickness of the reaction zone where a high local pH is created, it is evident that the local pH cannot be recorded consistently with a common pH electrode. Nevertheless, there is no doubt that local pH is much higher than bulk solution pH. In addition to measuring the local pH directly, an attempt was made to calculate the local pH theoretically. The production of OH corresponds to the electricity consumed with time elapse and can be calculated by Faraday’s law. The calculation results (Figure B) suggest that the local pH decreases with the thickness of local diffusion layer and it can reach pH values as high as 13.2 and 14.5 theoretically for an assumed maximum thickness of the local diffusion layer of 1 mm and after, respectively, 1 and 24 h electrolysis. The local pH can be even higher if we assume a smaller local diffusion layer. The theoretical calculation along with the fact that CaP only forms in the vicinity of and on the cathode surface indicated that the electrochemically induced high local pH is indeed responsible for the phosphate precipitation.

Figure 2

(A) The measured and (B) theoretically calculated local pH. Conditions: [Ca(NO3)2·4H2O] = 1.0 mM; [Na2SO4] = 50 mM; [Na2HPO4] = 0.6 mM; Current = 20 mA, Time = 24 h.

Crystallization Mechanism

As discussed above, the bulk solution pH values in the electrochemical system are not as important as in traditional chemical precipitation processes. This is attributed to the electrochemically created difference between bulk solution pH and the local pH at the vicinity of cathode. A possible CaP formation and precipitation mechanism based on the increase of local pH is suggested here. For the first step, the consumption of electrons by cathode mediated water reduction, created the high local pH (see eq ). Meanwhile, dihydrogen phosphate (H2PO4) reacts to monohydrogen phosphate (HPO42) and phosphate (PO43) via acid–base reactions in the local area.In the second step, homogeneous nucleation of CaP occurs because of the increased thermodynamic driving force and the declined solubility of CaP salts, both resulting from the high local pH. It should be noted that the Ti cathode might also provide a favorable surface for CaP nucleation in this system. Even so, it takes more than 4 h to see macroscopic precipitates. These then promote the growth and precipitation of precursor phases of HAP. The formed precipitates were weakly attached to the cathode surface via electrostatic interactions and continued to growing.[41] Gradually, the precipitates covered the cathode surface. One may worry that covering the cathode surface with CaP precipitates will increase the resistance and will corrupt the local pH and thus under constant current conditions, the cell potential would increase a lot. However, this phenomenon was not observed in our system, probably because the surface is not completely blocked as a result of the formation of hydrogen bubbles that resulted in small channels through the CaP layer. In addition, because of the design of our electrodes, the bottom side (or even the rod) of the cathode can work equally well when the top of the cathode is covered.

The possible intermediate phases, including amorphous calcium phosphate (ACP), brushite (CaHPO4·2H2O, DCPD), and octacalcium phosphate (Ca8(HPO4)2(PO4)4·5H2O, OCP) can be involved in the crystallization process (see eqs , 8, and 9). However, we were not able to characterize all possible species mentioned. The associated initial phase in our system was demonstrated as ACP by the absence of peaks in the corresponding XRD patterns. The typical broad peak at 2θ = 38° confirms the formation of ACP as a precursor (Figure A). Regarding ACP, there is no defined chemical formula yet but normally the formula of Ca9(PO4)6·nH2O is used since Posner and Betts proposed that structure.[42] However, the Ca/P molar ratio (1.38, SI Figure S3) in our system is lower than the proposed value and therefore, the formula of CaH(PO4)·nH2O is suggested. The formation of ACP in our system can be expressed as given in eq . In addition, carbonate, which could originate from atmospheric CO2 under alkaline conditions, might also be incorporated or precipitate as calcium carbonate. However, both XRD and Raman data cannot confirm the presence of CaCO3. The formation of ACP in our system agrees with Ostwald rule,[43] which foresees that the crystallization process is initiated by the formation of least thermodynamically stable phase. Indeed, though thermodynamics predict HAP formation, the direct formation of HAP (eq ) was not observed. This is because the formation of HAP is much slower than that of either ACP or OCP, and during simultaneous phase formation, a larger percentage of the kinetically favored species may be observed, even though it has a much smaller thermodynamic driving force.[44] At constant temperature, the transformation kinetics is a function of only pH because pH regulates both the dissolution of precursor phases and the formation of the early HAP nuclei.[44] In our system, the cathode mediated water reduction regulates the production of OH. Therefore, we thought that when the electrolysis time is increased, the initially formed ACP and other intermediate CaP phases may transform to HAP via eq .

Figure 3

(A) XRD patterns, (B) SEM images and (C) Raman spectrum of samples collected under different reaction days. (D) Ca and P concentration change with time elapse. Conditions: [Ca(NO3)2·4H2O] = 1.0 mM; [Na2SO4] = 50 mM;[Na2HPO4] = 0.6 mM; Current = 20 mA, pH0 8.2; Time = 1 day to 11days.

To check if HAP can form eventually, we increased the electrolysis time up to 11 days. The results, however, illustrate that even after a period of 4 days, the products were still dominantly amorphous (Figure A). This indicates that the precursor phase can be maintained for a long period. However, we found that though the phase does not change, the solids particle size increased, as can be seen from SEM images shown in Figure B. Note that these SEM images have the same magnification factor (×10 000). In addition to the growth of particles, the corresponding Ca/P molar ratio also increases to 1.44 (see SI Figure S3). However, on day 7, both the morphology and phase changed. The XRD data (Figure A) along with the typical needle-like shape[45] (Figure B) demonstrates the formation of HAP on the seventh day. The good match with peaks around 13°, 30° and the four peaks in the range of 2θ 38° to 42° for commercial HAP confirms the transformation to HAP. The Ca/P molar ratio (1.66) on day 7 also agrees well with theoretical Ca/P ratio (1.67). On day 11, the particle size increased again and the Ca/P ratio reached 1.76, but the morphology remained need-like. The phase transformation to HAP can be further supported by Raman data (Figure C), where the ν1 PO43– band shifted from 955 cm–1 typical for ACP (day 1 and 4) to 963 cm–1 that is for HAP (day 7 and 11).[46] In addition to solid characterization and analysis, the changes of Ca and P concentrations in the bulk solution also support the phase transformation. Figure C shows the removal trend of Ca and P from solution. Both P and Ca concentrations decreased with electrolysis time. After 7 days, more than 90% P and Ca precipitated from solution. Specifically, at the end of all reaction periods, the removal efficiency of P is higher than of Ca, but the difference for 7 days and 11 days (3.1% ± 0.3) is much lower than for day 1–7 (9.5% ± 1.0). This result suggests that low Ca/P molar ratio products (ACP) are formed initially on day 1 (Ca/P = 1.38) and day 4 (Ca/P = 1.44) and later transformed into high ratio (1.66 and 1.76 for day 7 and 11 respectively) product (HAP), thanks to the continuous production of OH at the cathode surface. Because the initial molar ratio of Ca (1.0 mM) to P (0.6 mM) is 1.67 and therefore the formation of low ratio Ca/P products will result in the relatively lower use of Ca. To conclude, the formation of HAP in the electrochemical system is identified as a typical crystallization process, starting with an amorphous phase followed by the precursors and finally transformed to the thermodynamically most stable phase (HAP).

Electrochemical Recovery of Phosphorus in Real Wastewater

Besides studying the efficiency and the precipitation mechanism using simulated solutions with various bulk pH, the efficiency of electrochemical P precipitation for real wastewater was investigated and compared with conventional chemical precipitation, in terms of efficiency and cost. Detailed information about the wastewater compositions, experimental methods and cost calculation are provided in the Supporting Information (SI) (See the texts and Table S1).

Figure gives a summary of the results of electrochemical and chemical precipitation. In electrochemical precipitation system, after a period of 24, 48, and 72 h, the P concentration decreased from 8.0 to 4.3, 3.1, and 2.3 mg/L respectively. This corresponds to a removal efficiency of 42.8% in 24 h, 62.1% in 48 h and 71.5% in 72 h. Though the wastewater has a complicated matrix (see SI Table S1) and a much lower P concentration, the removal efficiency is comparable to the simulated solutions. This is probably due to the role of Mg and Ca. In the wastewater, the removal of P may result from both calcium phosphate and magnesium phosphate precipitation. This was concluded from the simultaneous removal of P, Ca and Mg (Figure A). At the same time, we found the concentration of inorganic carbon also decreased from 166 to 115 mg/L (SI Figure S4). This points to formation and precipitation of CaCO3 and MgCO3 or a mixed phase. The contribution of CaCO3 was also reported on P removal from wastewater by CaP precipitation.[16,34] In addition to the coprecipitation of carbonate salts, the heavy metal ions in the wastewater, which we did not address in this paper, might be removed via adsorption or incorporation, as reported in a previous study on struvite formation from urine.[47] Hence, for the purpose of P recycling for use in fertilizer, the behavior of toxic ions in the phosphate recovery process should be investigated in detail. Ideally, heavy metal ions (i.e., Zn, Cu) could be incorporated and work as micro nutrients, but their contents should be kept below the standard for P fertilizers. A more in depth study on the fate and behavior of coexisting components and the corresponding effects on the possible application of products is ongoing.

Figure 4

(A) Concentration change and removal efficiency of P, Ca and Mg in real wastewater by electrochemical precipitation. (B) Removal efficiency of P by conventional chemical precipitation under different solution pH adjusted by sodium hydroxide.

Electrochemical P precipitation was also compared with conventional chemical precipitation, in terms of efficiency and cost. Clearly, as shown in Figure , as expected, chemical precipitation is more efficient than electrochemical precipitation regarding removal efficiency. After adjusting the solution pH ≥ 10, over 78.8% P (Figure B) was removed from the solution. It should be noted that the P removal refers to the P removal after filtration through 0.45 μm membrane and therefore this value is higher than the precipitation efficiency (see Figure B), as the formed products do not have a good settling rate. For example, the removal efficiency of P is 93.9% at pH 11 but the corresponding precipitation efficiency is only 67.8%. Hence, in chemical precipitation process, a follow up separation process is needed. However, in the electrochemical system, because the precipitation only happens near and on the cathode surface, removal and separation are simultaneous. The extra separation process is therefore avoided.

For cost comparison, we only considered the electricity cost in the electrochemical system and the caustic soda cost for the chemical precipitation system. After normalizing the cost as €/kg P, the cost of electrochemical precipitation is 41 €/kg P, which is comparable to chemical precipitation. The cost of chemical precipitation depends on the solution pH and varies from 18.9 to 61.1 €/kg P. The lowest cost is achieved at pH 10. As the cost of the two methods are of the same magnitude, we believe optimization of the electrochemical process can make the process economically viable.