Thermodynamic Analysis and Experimental Investigation of Al and F Removal from Sulfuric Acid Leachate of Spent LiFePO 4 Battery Powder

: The co-precipitation thermodynamics of the Li + –Fe 2+ /Fe 3+ –Al 3+ –F − –SO 4 2 − –PO 4 3 − –H 2 O system at 298 K is studied, aiming to understand the precipitation characteristics. Based on the principle of simultaneous equilibrium and the mass action law, the missing Ksp values of AlF 3 and FeF 3 were estimated. The results of thermodynamic calculation demonstrate that Al 3+ and F − in the sulfuric acid leachate could be preferentially precipitated in the form of AlPO 4 and FeF 3 by the precise adjustment of the ﬁnal pH value. Only a small amount of P and Fe was lost by the precipitation of Fe 3 (PO 4 ) 2 · 8H 2 O, FePO 4 , and Fe(OH) 3 during the puriﬁcation process. Controlling the oxidation of ferrous ions effectively is of critical signiﬁcance for the loss reduction of P and Fe. Precipitation experiments at different pH value indicated that the concentration of Al 3+ and F − in the leachate decreased as the ﬁnal pH value rose from 3.05 to 3.90. When the ﬁnal pH value was around 3.75, aluminum and ﬂuoride ion impurities could be deeply puriﬁed, and the loss rate of phosphate ions and iron ions could be reduced as much as possible. Relevant research results can provide theoretical guidance for the puriﬁcation of leachate in the wet recycling process of lithium-ion batteries.


Introduction
Greenhouse gas emissions are severely aggravating global climate change, and thus all countries should formulate national policies to achieve zero carbon emissions [1,2]. China, for instance, has pledged to peak nationwide CO 2 emissions by 2030 and achieve carbon neutrality by 2060 [3,4]. Aimed at this goal, the zero-carbon industry represented by clean and renewable energy is facing a critical moment of accelerated development [5]. In the past decade, China has become the world's largest market of new energy vehicles (EV) as its EV stock reached 3,810,000 in 2019, nearly half of the global stock [6]. Given that the service life of power batteries is approximately 5-8 years, it is predicted that the cumulative decommissioning of powder batteries in China will reach 90.5 GWh from 2020 to 2022, of which spent LiFePO 4 batteries will account for 53.8% (48.7 GWh) [7].
Compared with traditional fuel vehicles, EV power battery systems use mineral resources such as lithium, cobalt, nickel, manganese, copper, aluminum, iron, phosphorus, and graphite [8]. Recycling valuable metal resources from discarded lithium batteries is the key to the sustainable development of the EV industry. On the other hand, if not professionally treated, potentially toxic components inside them would be directly disposed of sulfuric acid leachate within predetermined range in an inert atmosphere and monitor the changes of Al, F, P and Fe contents in the leachate.

Materials
Spent LiFePO 4 cathode powder was provided by a battery recycling company (Nantong, China), and it was separated from aluminum foils by thermal treatment and screening. The composition of collected spent LiFePO 4 cathode powder, determined by inductively coupled plasma optical emission spectroscopy (ICP-OES, SPECTROBLUE SOP) after leaching with aqua regia, is shown in Table 1. The sulfuric acid leaching solution is prepared by the leaching reaction of spent LiFePO 4 cathode powder with 2.5 M H 2 SO 4 under the conditions of S/L = 1/7, leaching temperature 75 • C, and leaching time of 2 h. Prior to the precipitation, a small amount of reductive iron powder was added to leachate to remove the copper ions by replacement. It should be noted that since reductive iron powder was used to remove the copper ion by replacement, the iron ions in the solution mainly existed in the form of Fe 2+ . The concentrations of metal ions (Fe, Li, Cu, Al, Na, K, Ca, Mg) and non-metal ions (P, S, Si) in the sulfuric acid leachate were examined by ICP-OES. The F content of sulfuric acid leachate was obtained through the fluoride ion selective electrode method. The composition of sulfuric acid leachate is shown in Table 2. Aside from main elements P, Fe, Li, and S, the composition of the solution also contains a small amount of impurity elements such as Al, F, Na, K, Ca, Si and Mg. Ascorbic acid and trisodium phosphate dodecahydrate used in the experiments were of analytical grade. The purity of high-purity argon gas was 99.99%.

Methods
On the basis of the simultaneous equilibrium principle and mass action law, the coprecipitation thermodynamics of the Li + -Fe 2+ /Fe 3+ -Al 3+ -F − -SO 4 2− -PO 4 3− -H 2 O system is simulated. The calculation process based on the Newton-Raphson iteration method was carried out by using Microsoft Excel. Owing to the lack of complete data on the activity component in the solution, the molar concentration of each species was used to replace the activity coefficient in the thermodynamic calculation.
The schematic plot of the precipitation experiment set-up is illustrated in Figure 1. In the bench-scale precipitation experiments, 250 mL of the sulfuric acid leachate was poured into a three-necked flask immersed in a water bath at 25 • C. Afterward, 0.5 g of ascorbic acid, a reducing agent, was added to the leachate to reduce ferric ions. Simultaneously, high-purity argon was introduced to prevent the oxidation of ferrous ions. Na 3 PO 4 ·12H 2 O was added to the sulfuric acid leachate in real time to adjust the pH value of the system and keep it constant at a predetermined value. After the predetermined time, the liquid-solid separation of the feed solution was carried out, and then the composition of purified solution was analyzed by ICP-OES and the fluoride ion selective electrode method. The precipitate was vacuum dried at 60 • C for 24 h and then weighed and characterized.
of the system and keep it constant at a predetermined value. After the predetermined time, the liquid-solid separation of the feed solution was carried out, and then the composition of purified solution was analyzed by ICP-OES and the fluoride ion selective electrode method. The precipitate was vacuum dried at 60 °C for 24 h and then weighed and characterized. The phase compositions of the precipitation were studied by X-ray diffraction (XRD, Rigaku D/max-2500, Rigaku Corporation, Tokyo, Japan) and Fourier transform infrared spectrometry (FT-IR, Nicolet 6700, Thermo Fisher Scientific, Waltham, USA). The elemental composition and chemical states on the surface of the electrode particles were detected by X-ray photoelectron spectroscopy (XPS, ESCALAB 250Xi, Thermo Fisher Scientific, Waltham, USA). The removal rate of Al 3+ and F − as well as the loss rate of Fe during the purification process were computed by Equation (1).
where (%) is the removal rate of Al 3+ and F − as well as the loss rate of Fe 2+ , (mL) is the volume of sulfuric acid leachate, (mL) is the volume of purified solution, (mg/L) is the concentration of Al 3+ , F − , and Fe 2+ in the sulfuric acid leachate, and (mg/L) is the concentration of Al 3+ , F − , and Fe 2+ in the purified solution.
The loss rate of PO4 3− in the purification process was calculated by Equation (2).
where (%) is the loss rate of PO4 3− , (mL) is the volume of sulfuric acid leachate, (mL) is the volume of purified solution, (mg/L) is the concentration of PO4 3− in the sulfuric acid leachate, (mg/L) is the concentration of PO4 3− in the purified solution, and (g) is the Na3PO4·12H2O dosage.
where R i (%) is the removal rate of Al 3+ and F − as well as the loss rate of Fe 2+ , V 0 (mL) is the volume of sulfuric acid leachate, V 1 (mL) is the volume of purified solution, C i 0 (mg/L) is the concentration of Al 3+ , F − , and Fe 2+ in the sulfuric acid leachate, and C i 1 (mg/L) is the concentration of Al 3+ , F − , and Fe 2+ in the purified solution.
The loss rate of PO 4 3− in the purification process was calculated by Equation (2).
where L P (%) is the loss rate of PO 4 3− , V 0 (mL) is the volume of sulfuric acid leachate, V 1 (mL) is the volume of purified solution, C p 0 (mg/L) is the concentration of PO 4 3− in the sulfuric acid leachate, C p 1 (mg/L) is the concentration of PO 4 3− in the purified solution, and m p (g) is the Na 3 PO 4 ·12H 2 O dosage.

Ion Species and Their Thermodynamic Data
In the Li + -Fe 2+ /Fe 3+ -Al 3+ -PO 4 3− -SO 4 2− -F − -H 2 O system, the main species in the solution were assumed to be H + , OH − , F − , PO 4 Table 3.  3 precipitations formed as the pH value and ion concentration in the solution changed. The solubility product constants K sp of AlF 3 and FeF 3 , as well as their thermodynamic data ∆ f G i θ and ∆ f S i θ , have not been reported before in the literature. In this paper, the K sp value of AlF 3 and FeF 3 was estimated according to their solubility and the equilibrium model of gradual dissociation.

Calculation of Ksp Value of AlF 3
Assuming that a certain amount of AlF 3 was added to deionized water, when AlF 3 reached dissolution equilibrium, the main species in the solution and their equilibrium equations can be shown in Equations (38)-(43) in Table 2.
According to the simultaneous equilibrium principle and mass action law, the mass balance equations of F and Al are given as follows:  6 3− ] are defined as the equilibrium concentration for each species in equilibrium conditions.
According to the literature, the solubility of AlF 3 is 0.67 g/100 mL of water at 20 • C [34]. Below this temperature, when AlF 3 reached dissolution equilibrium, the The K sp definition of AlF 3 (s) is shown in Equation (48). The K sp value of AlF 3 (s) was calculated to be 4.53 × 10 −17 .

Calculation of Ksp Value of FeF 3
Given that a certain amount of FeF 3 was added to deionized water, when FeF 3 reached dissolution equilibrium, the main species in the solution and their equilibrium equations can be shown in Equations (26)- (28) in Table 2.
On the basis of the simultaneous equilibrium principle and mass action law, the mass balance equations of F and Fe are given as follows: The chemical equilibrium constant relationship of Equations (26)- (28) in Table 2 is substituted into Equations (49) and (50) to obtain Equations (51) and (52).
According to the literature, the solubility of FeF 3 is 0.091 g at 20 • C/100 mL of water [34]. In this case, when The K sp definition of FeF 3 (s) is presented in Equation (53). The K sp value of FeF 3 (s) was calculated to be 3.623 × 10 −25 .
[P] T = [PO 4 3− ] + [HPO 4 When a solid phase is formed in the system, each component in the solid phase stable region must satisfy not only Equations (3)-(16), (29)-(43), and (66)-(72) but also their corresponding constraint equations. The constraint equations for different solid phase stable regions are provided in Table 5.
[P] T = [PO 4 3− ] + [HPO 4 When a solid phase is formed in the system, each component in the solid phase stable region must meet Equations (3) Table 7.    (Figure 2b). AlPO4 was the main precipitate at pH = 1.52.6. At pH ≥2.6, the total concentration of iron ions decreased and that of phosphate ions decreased more sharply because the reaction of ferric ions and phosphate ions formed Fe3(PO4)2·8H2O precipitate. Accordingly, the precipitates generated in the solution included AlPO4 and Fe3(PO4)2·8H2O when the pH value was above 2.5, and the molar proportion of Fe3(PO4)2·8H2O increased as the final pH value increased. At pH ≥3.0, the molar proportion of Fe3(PO4)2·8H2O and AlPO4 remained stable, and Fe3(PO4)2·8H2O was the main precipitate, AlPO4 was the minor precipitate.   (Figure 3b). FeF3 was the only precipitate at pH ≤ 0.8. At pH = 01.4, the total concentration of fluoride ion continuously increased, suggesting that the increase in pH value within the range was not conducive to the formation of FeF3. At pH ≥0.8, the total concentration of phosphate ions and iron ions substantially decreased because of the formation of FePO4. At pH ≥1.4, the total concentration of aluminum ions began to decrease owing to the formation of AlPO4, whereas the concentration of fluoride ions also began to decrease because of the formation of FeF3. At pH = 1.03.9, FePO4 was the main precipitate, FeF3 and AlPO4 were the minor precipitate. At pH ≥3.9, the total concentration of phosphate ions started to increase, because of the disappearance of FePO4 and the mass formation of Fe(OH)3 with the increase in concentration of hydroxide ions. At pH ≥4.0, the molar proportion of FeF3, AlPO4, and Fe(OH)3 precipitate remained stable, and Fe(OH)3 was the main precipitate, FeF3 and AlPO4 was the minor precipitate.

Precipitation Experiments
Based on the results of above thermodynamic modeling calculation, the sulfuric acid leachate of spent LiFePO4 battery powder was employed to conduct bench-scale experiments to validate the feasibility of the separation and purification technology under the conditions of the reaction temperature of 25 °C, reaction time of 5 min, and agitation rate of 400 rpm. Figure 4 shows the effects of Na3PO4·12H2O dosage on the final pH value and precipitation behavior of each element in the solution. Figure 4a depicts the relationship between Na3PO4·12H2O dosage and the final pH value of the 250 mL solution. As indicated in Figure 4b,c, the final pH value of the solution had a considerable effect on the removal of aluminum and fluorine impurities by chemical precipitation. When the amount of Na3PO4·12H2O added was increased from 21.4 g to 25.3 g, the final pH value increased from 3.05 to 3.77. Correspondingly, the concentration of residual aluminum ions in the solution decreased from 890.5 mg/L to 27.55 mg/L. Moreover, the removal rate of aluminum ions increased from 55.48% to 98.62%, the concentration of residual fluoride ions decreased from 490.5 mg/L to 46.0 mg/L, and the removal rate of aluminum ions increased from 51.1% to 95.41%. By comparison, when the final pH value increased to 3.90, the concentration of residual aluminum and fluoride ions only slightly decreased from 27.55 mg/L to 15.95 mg/L and from 46.0 mg/L to 39.78 mg/L, respectively. The effects of the final pH value on PO4 3− , Fe 2+ , and Li + in the solution are illustrated in Figure  4d-f, respectively. As presented in Figure 4d, e, the loss rate of PO4 3− and Fe 2+ increased, when the final pH value increased from 3.05 to 3.77. Combined with the results of thermodynamic calculation, the increase in the loss rate of PO4 3− and Fe 2+ should be resulted by the formation of FeF3, AlPO4, Fe3(PO4)2·8H2O, FePO4, and Fe(OH)3. When the final pH value increased from 3.05 to 3.90, the loss rate of Li + slightly increased from 0.21% to 2.28% (Figure 4f). The high final pH value probably resulted in an increase in the amount of precipitation that contained lithium ions. In conclusion, the final pH value at about 3.77 could deeply purify the aluminum ions and fluoride ions, as well as reduce the loss rate of phosphate ions and iron ions as much as possible.

Precipitation Experiments
Based on the results of above thermodynamic modeling calculation, the sulfuric acid leachate of spent LiFePO 4 battery powder was employed to conduct bench-scale experiments to validate the feasibility of the separation and purification technology under the conditions of the reaction temperature of 25 • C, reaction time of 5 min, and agitation rate of 400 rpm. Figure 4 shows the effects of Na 3 PO 4 ·12H 2 O dosage on the final pH value and precipitation behavior of each element in the solution. Figure 4a depicts the relationship between Na 3 PO 4 ·12H 2 O dosage and the final pH value of the 250 mL solution. As indicated in Figure 4b,c, the final pH value of the solution had a considerable effect on the removal of aluminum and fluorine impurities by chemical precipitation. When the amount of Na 3 PO 4 ·12H 2 O added was increased from 21.4 g to 25.3 g, the final pH value increased from 3.05 to 3.77. Correspondingly, the concentration of residual aluminum ions in the solution decreased from 890.5 mg/L to 27.55 mg/L. Moreover, the removal rate of aluminum ions increased from 55.48% to 98.62%, the concentration of residual fluoride ions decreased from 490.5 mg/L to 46.0 mg/L, and the removal rate of aluminum ions increased from 51.1% to 95.41%. By comparison, when the final pH value increased to 3.90, the concentration of residual aluminum and fluoride ions only slightly decreased from 27.55 mg/L to 15.95 mg/L and from 46.0 mg/L to 39.78 mg/L, respectively. The effects of the final pH value on PO 4 3− , Fe 2+ , and Li + in the solution are illustrated in Figure 4d-f, respectively. As presented in Figure 4d, e, the loss rate of PO 4 3− and Fe 2+ increased, when the final pH value increased from 3.05 to 3.77. Combined with the results of thermodynamic calculation, the increase in the loss rate of PO 4 3− and Fe 2+ should be resulted by the formation of FeF 3 , AlPO 4 , Fe 3 (PO 4 ) 2 ·8H 2 O, FePO 4 , and Fe(OH) 3 . When the final pH value increased from 3.05 to 3.90, the loss rate of Li + slightly increased from 0.21% to 2.28% (Figure 4f). The high final pH value probably resulted in an increase in the amount of precipitation that contained lithium ions. In conclusion, the final pH value at about 3.77 could deeply purify the aluminum ions and fluoride ions, as well as reduce the loss rate of phosphate ions and iron ions as much as possible.
The XRD patterns and FT-IR spectra of precipitation obtained at different final pH values are plotted in Figure 5. As shown in Figure 5a, no obvious characteristic diffraction peak appeared on the XRD pattern, demonstrating that the precipitation had an amorphous structure. Figure 5b shows the FTIR spectra of precipitation. The wide absorption peak at 3424 cm −1 and the small absorption peak at 1630 cm −1 corresponded to the O-H bond stretching and bending vibrations in the precipitation, respectively [12]. The strong absorption peak at 1090 cm −1 was attributed to the P-O stretching vibrations in PO 4 3− [37]. The small absorption peak at 617 cm −1 represented the Al-O stretching vibrations in tetrahedral AlO 4 5− [38], and the absorption peak at 530 cm −1 corresponded to the Fe-O stretching vibrations [38]. According to the FTIR spectra of precipitation and the results of thermodynamic simulation calculation, the main compositions of precipitation were AlPO 4 , Fe 3 (PO 4 ) 2 ·8H 2 O, FePO 4 , and Fe(OH) 3 . The XRD patterns and FT-IR spectra of precipitation obtained at different final pH values are plotted in Figure 5. As shown in Figure 5a, no obvious characteristic diffraction peak appeared on the XRD pattern, demonstrating that the precipitation had an amorphous structure. Figure 5b shows the FTIR spectra of precipitation. The wide absorption peak at 3424 cm −1 and the small absorption peak at 1630 cm −1 corresponded to the O-H bond stretching and bending vibrations in the precipitation, respectively [12]. The strong absorption peak at 1090 cm −1 was attributed to the P-O stretching vibrations in PO4 3− [37]. The small absorption peak at 617 cm −1 represented the Al-O stretching vibrations in tetrahedral AlO4 5− [38], and the absorption peak at 530 cm −1 corresponded to the Fe-O stretching vibrations [38]. According to the FTIR spectra of precipitation and the results of thermodynamic simulation calculation, the main compositions of precipitation were AlPO4, Fe3(PO4)2·8H2O, FePO4, and Fe(OH)3. The SEM-EDS spectra of precipitation under the optimal conditions are presented in Figure 6. Figure 6a shows that the precipitation was inside small irregular particles. Figure 6b is a partially enlarged map of precipitation, and its EDS mapping indicated that the precipitation contained various elements, such as P, Fe, O, Al, and F.    The XRD patterns and FT-IR spectra of precipitation obtained at different final pH values are plotted in Figure 5. As shown in Figure 5a, no obvious characteristic diffraction peak appeared on the XRD pattern, demonstrating that the precipitation had an amorphous structure. Figure 5b shows the FTIR spectra of precipitation. The wide absorption peak at 3424 cm −1 and the small absorption peak at 1630 cm −1 corresponded to the O-H bond stretching and bending vibrations in the precipitation, respectively [12]. The strong absorption peak at 1090 cm −1 was attributed to the P-O stretching vibrations in PO4 3− [37]. The small absorption peak at 617 cm −1 represented the Al-O stretching vibrations in tetrahedral AlO4 5− [38], and the absorption peak at 530 cm −1 corresponded to the Fe-O stretching vibrations [38]. According to the FTIR spectra of precipitation and the results of thermodynamic simulation calculation, the main compositions of precipitation were AlPO4, Fe3(PO4)2·8H2O, FePO4, and Fe(OH)3. The SEM-EDS spectra of precipitation under the optimal conditions are presented in Figure 6. Figure 6a shows that the precipitation was inside small irregular particles. Figure 6b is a partially enlarged map of precipitation, and its EDS mapping indicated that the precipitation contained various elements, such as P, Fe, O, Al, and F.   The SEM-EDS spectra of precipitation under the optimal conditions are presented in Figure 6. Figure 6a shows that the precipitation was inside small irregular particles. Figure 6b is a partially enlarged map of precipitation, and its EDS mapping indicated that the precipitation contained various elements, such as P, Fe, O, Al, and F.
The phase composition and surface chemical states of precipitation was further characterized via XPS analysis (Figure 7). The P 2p spectrum in Figure 7a shows that the peaks corresponding to the binding energies of P 2p 3/2 and P 2p 1/2 were located at 133.6 and 134.4 eV, respectively, which were caused by spin-orbit coupling. This result was consistent with the reported binding energies of P-O in PO 4 3− [39]. The Fe 2P spectrum in Figure 7b shows that the spin-orbit doublets of Fe 2p 3 [40]. The results indicated that the iron element in the precipitation existed in the form of Fe 2+ and Fe 3+ , and the Fe 3+ /Fe 2+ molar ratio was 1.35. Owing to the overlapping photoelectron spectra of Li 1s and Fe 3p, Gaussian peak fitting was performed for Li 1s (Figure 7c). The characteristic peak of Li 1s located at 56.6 eV was attributed to Li + [40]. The O 1s spectrum was fitted with two peaks (Figure 7d). The strong peak at 531.3 eV was related to the P-O of phosphate groups, whereas the weak peak at 532.8 eV was associated with O-H. The characteristic peaks of Al 2p that appeared at 74.7 and 75.1 eV were related to Al-O from AlPO 4 (Figure 7e), indicating that the aluminum element in the precipitation existed in the form of AlPO 4 . Figure 7f shows the spectrum of F 1s. The characteristic peak of F 1s centered at 68.5 eV as ascribed to Fe-F. Therefore, the fluorine element in the precipitation existed as FeF 3 . The phase composition and surface chemical states of precipitation was further characterized via XPS analysis (Figure 7). The P 2p spectrum in Figure 7a shows that the peaks corresponding to the binding energies of P 2p3/2 and P 2p1/2 were located at 133.6 and 134.4 eV, respectively, which were caused by spin-orbit coupling. This result was consistent with the reported binding energies of P-O in PO4 3− [39]. The Fe 2P spectrum in Figure 7b shows that the spin-orbit doublets of Fe 2p3/2 and Fe 2p1/2 were situated at approximately 712.0 and 725.0 eV, respectively. Each of them consisted of two main peaks (709.4 and 711.3 eV for Fe 2p3/2, 722.5 and 724.4 eV for Fe 2p1/2) and two satellites (713.9 and 717.3 eV for Fe 2p3/2, 727.0 and 730.4 eV for Fe 2p1/2) [40]. The results indicated that the iron element in the precipitation existed in the form of Fe 2+ and Fe 3+ , and the Fe 3+ /Fe 2+ molar ratio was 1.35. Owing to the overlapping photoelectron spectra of Li 1s and Fe 3p, Gaussian peak fitting was performed for Li 1s (Figure 7c). The characteristic peak of Li 1s located at 56.6 eV was attributed to Li + [40]. The O 1s spectrum was fitted with two peaks (Figure 7d). The strong peak at 531.3 eV was related to the P-O of phosphate groups, whereas the weak peak at 532.8 eV was associated with O-H. The characteristic peaks of Al 2p that appeared at 74.7 and 75.1 eV were related to Al-O from AlPO4 (Figure 7e), indicating that the aluminum element in the precipitation existed in the form of AlPO4. Figure 7f shows the spectrum of F 1s. The characteristic peak of F 1s centered at 68.5 eV as ascribed to Fe-F. Therefore, the fluorine element in the precipitation existed as FeF3.

Conclusions
The evolution law of the total equilibrium concentration of Al, Fe, P, and F, and the phase composition of precipitates at different final pH values are explained using the thermodynamic mathematical model. By adjusting the final pH value of the Li + -Fe 2+ /Fe 3+ -PO4 3− -SO4 2− -H2O system, Al and F elements were precipitated in solid phases of AlPO4, and FeF3, whereas P and Fe elements were co-precipitated as Fe3(PO4)2·8H2O, FePO4, and Fe(OH)3, respectively. Among them, Fe3(PO4)2·8H2O is the main precipitate, and the precipitate also contains a small amount of FePO4, Fe(OH)3, AlPO4, and FeF3. Therefore, selective precipitation of Al 3+ and Fcan be precisely controlled by pH of the leaching solution, at the same time, the loss rate of P and Fe can be reduced. The removal mechanism of Al and F was further revealed by characterizing the precipitation via XRD, FTIR, SEM-EDS, and XPS. Results indicated that the precipitation mainly existed in the form of AlPO4, Fe3(PO4)2·8H2O, FePO4, Fe(OH)3, and FeF3, consistent with the results of thermodynamic analysis.

Conclusions
The evolution law of the total equilibrium concentration of Al, Fe, P, and F, and the phase composition of precipitates at different final pH values are explained using the thermodynamic mathematical model. By adjusting the final pH value of the Li + -Fe 2+ /Fe 3+ -PO 4 3− -SO 4 2− -H 2 O system, Al and F elements were precipitated in solid phases of AlPO 4 , and FeF 3 , whereas P and Fe elements were co-precipitated as Fe 3 (PO 4 ) 2 ·8H 2 O, FePO 4 , and Fe(OH) 3 , respectively. Among them, Fe 3 (PO 4 ) 2 ·8H 2 O is the main precipitate, and the precipitate also contains a small amount of FePO 4 , Fe(OH) 3 , AlPO 4 , and FeF 3 . Therefore, selective precipitation of Al 3+ and Fcan be precisely controlled by pH of the leaching solution, at the same time, the loss rate of P and Fe can be reduced. The removal mechanism of Al and F was further revealed by characterizing the precipitation via XRD, FTIR, SEM-EDS, and XPS. Results indicated that the precipitation mainly existed in the form of AlPO 4 , Fe 3 (PO 4 ) 2 ·8H 2 O, FePO 4 , Fe(OH) 3 , and FeF 3 , consistent with the results of thermodynamic analysis.