Phosphate removal from aqueous solution by Fe-La binary (hydr)oxides: A study of batch experiments and mechanisms

In this study, Fe-La binary (hydr)oxides were prepared by a co-precipitation method for phosphate removal. Various techniques, including secondary electron microscopy with energy dispersive X-ray spectroscopy (SEM-EDX), powder X-ray diffraction (p-XRD) and Brunauer-Emmett-Teller (BET) surface area analysis, were employed to characterize the synthesized Fe-La binary (hydr)oxides. Batch experiments indicated that the performance of phosphate removal by Fe-La binary (hydr)oxides was excellent and increased with increasing La contents. The kinetics study showed that the adsorption was rapid and described better by the pseudo-second-order equation. The maximum adsorption capacities of Fe/La 3:1, Fe/La 1:1 and Fe/La 1:3 binary (hydr)oxides at pH 4.0 calculated by Langmuir model were 49.02, 69.44 and 136.99 mg/g, respectively. The uptake of phosphate was highly affected by solution pH and signicantly reduced with the increase of pH value. The analyses of p-XRD, Fourier transform infrared spectroscopy (FTIR) and X-ray photoelectron spectroscopy (XPS) suggested that the predominant mechanisms of phosphate removal involved surface hydroxyl exchange reactions and co-precipitation of released La 3+ and phosphate ions, which resulted into the formation of amorphous phase of rhabdophane (LaPO 4 ∙ 0.5H 2 O). The results show great potential for the application on the treatment of phosphate decontamination for their high eciency of phosphate removal.


Introduction
Phosphorus (P) is an essential nutritional element for the growth and development of algae and other living organisms, while excessive amounts of phosphate due to the inputs from urban and agricultural land runoff have a risk of causing eutrophication, harmful algal blooms and subsequent deterioration of aquatic environments and ecosystems (Conley et al. 2009). Consequently, it is indispensable to remove the excessive phosphate from water bodies. Various treatment technologies for phosphate removal including membrane ltration, adsorption, chemical precipitation, ion-exchange, and biological processes, have been developed over the past decades (Pan et al. 2020, Zhang et al. 2021. Compared with other removal processes, adsorption is attractive due to its high e ciency, cost-effectiveness and sample operation (Qu et al. 2020, Vu &Wu 2020. To date, various kinds of functionalized metal (hydr)oxidebased adsorbents, such as biochar-loaded Ce 3+ -enriched ultra-ne ceria nanoparticles, NaLa(CO 3 ) 2 decorated magnetic biochar, Fe-Al-Mn trimetal oxide adsorbent, graphene-lanthanum composite, magnetic porous biochar supported La(OH) 3 and lanthanum hydroxide nanorods, have been explored and exhibited enhanced phosphate removal (Chen et al. 2015, Lu et al. 2013, Qu et al. 2020, Zhang et al. 2021. In particular, different bimetallic hydroxide adsorbents have been employed due to their high e ciency for phosphate removal, separation convenience and highly e cient reusability. For example, magnetic La(OH) 3 /Fe 3 O 4 nanocomposites have exhibited a high sorption capacity of 83.5 mg P/g, and strong selectivity for phosphate removal . Vn et al. synthesized magnetic porous NiLa-layered double oxides with improved phosphate adsorption characteristics and antibacterial activities (Vu &Wu 2020). Zhang et al. reported an Fe-Mn binary oxide adsorbent with an Fe/Mn molar ratio of 6:1 prepared by a simultaneous oxidation and co-precipitation process (Zhang et al. 2009).
Recently, lanthanum (La), a rare earth element, has received enormous attention for its environmental friendliness, low cost and excellent phosphate removal performance. A number of La-based adsorbents have been prepared, including La-doped silica spheres, La(III)-bentonite/chitosan composites, lanthanum modi ed bentonite clay, a nanoporous material referred to as La 100 SBA-15 and La(OH) 3 loaded magnetic mesoporous nanospheres (Chen et al. 2019, Dithmer et al. 2015, Huang et al. 2014, Xu et al. 2020, Yang et al. 2011. La-based bimetallic hydroxide adsorbents have also been designed and developed (Qu et al. 2020, Vu &Wu 2020). Xu et al. obtained lanthanum/aluminum-hydroxide composites via co-precipitation and its mechanisms of phosphate removal were investigated by X-ray absorption ne structure (XAFS) spectroscopy (Xu et al. 2017). The results indicated that phosphate preferentially bonded with Al under weakly acidic conditions while tended to associate with La under alkaline conditions, leading to the formation of the inner sphere bidentate-binuclear complexes between La(III) and phosphate. However, less attention has been paid to the phosphate removal by La-Fe bimetallic hydroxide adsorbents. In one study, the activation of lattice oxygen in LaFe (oxy)hydroxides for phosphorus removal and their underlying removal mechanisms were determined by X-ray absorption near edge structure (XANES) and X-ray photoelectron spectroscopy (XPS) (Yu et al. 2019). In that study, Laincorporated La-Fe bimetallic hydroxides were employed. However, the Fe-incorporated La-Fe bimetallic hydroxides were ignored. Therefore, there are still some practical limitations that need to be further researched.
In this paper, a series of Fe-La binary (hydr)oxides with different Fe/La molar ratios were prepared by a facile co-precipitation method and characterized by scanning electron microscope with energy-dispersive X-ray spectrometer (SEM-EDX), Brunauer-Emmett-Teller (BET) surface area analysis, powder X-ray diffraction (p-XRD) and the pH of the point of zero charge (pH pzc ). Their performance of phosphate removal was studied by adsorption kinetics, adsorption isotherms, the effect of pH, interactive effects with co-existing ions and reusability, and the removal mechanisms were investigated by p-XRD, Fourier transform infrared spectroscopy (FTIR) and XPS analyses.

Materials
All analytic grade reagents and chemicals including La(NO 3 ) 3 •6H 2 O, Fe(NO 3 ) 3 •9H 2 O, NaOH, KH 2 PO 4 , ascorbic acid and molybdenum acid ammonium were purchased from Sinopharm Chemical Reagent Co. (Shanghai, China) and used without further puri cation. All reaction glass volumetric asks and vessels used in the experiments were rst cleaned by soaking in a 5% HNO 3 solution for least 24 h and then rinsed several times with deionized (DI) water prior to testing. All working solutions were prepared using DI water. A phosphate stock solution with a concentration of 2.0 g/L was obtained by dissolving KH 2 PO 4 in DI water. Separate phosphate working solutions with various initial phosphate concentrations were freshly prepared by diluting the phosphate stock solution using DI water.

Preparation of adsorbents
Fe-La binary (hydr)oxides with various Fe/La molar ratios were synthesized via a facile co-precipitation method. In a typical synthesis, predetermined amounts of Fe(NO 3 ) 3 ·9H 2 O and La(NO 3 ) 3 ·6H 2 O were dissolved in 50 mL DI water under magnetic stirring to obtain a series of Fe-La nitrate solutions with different Fe/La molar ratios. Then, the pH of the solution was raised to ~ 10 within approximately 10 min and kept constant by the dropwise addition of a 1 mol/L NaOH solution. After adjusting the pH, the resulting suspensions were continuously stirred for 2 h. After that, the obtained precipitates were washed several times with DI water and freeze-dried for 24 h. The dried products were ground into a ne powder and stored in a desiccator for further use. Samples were labeled according to their initial Fe/La molar ratios, Fe/La 3:1, Fe/La 1:1 and Fe/La 1:3. Pure lanthanum hydroxide was also prepared, and details are provided in the Supplementary Material.

Characterization of adsorbents
The morphologies and surface element distributions of the samples were characterized by a eld emission scanning electron microscope with energy-dispersive X-ray spectrometer (SEM-EDX) (Phenom LE, Netherlands). The speci c surface areas and pore structures of the materials were measured by N 2 adsorption/desorption using a Micromeritics ASAP 2020 surface area analyzer (Micromeritics ASAP 2020, USA) at 77 K and were calculated by the Brunauer-Emmett-Teller (BET) method with the adsorption branches of the isotherms and Barrett-Joyner-Halenda (BJH) model with the desorption curves, respectively. FTIR spectra were recorded on a Nicolet Nexus 870 FT-IR spectrometer using a transmission model (Nicolet, USA) with the KBr pellet technique. In order to investigate the surface chemistry compositions of the materials before and after phosphate adsorption, X-ray photoelectron spectra (XPS) were collected on an X-ray photoelectron spectra spectrometer (ESCALAB 250Xi, Thermo Fisher, America) equipped with a monochromatized Al Kα X-ray source (hν = 1486.6 eV) and a hemispherical electron analyzer. The C 1s photoelectron peak (284.6 eV) was used for calibrating the binding energy. A nonlinear least-squares curve tting program was employed in the data analysis using XPSPEAK41 Software. The spectral bands were deconvoluted with the subtraction of a nonlinear Shirleytype background and a Gaussian (20%)-Lorentzian (80%) mixed function. Powder X-ray diffraction (p-XRD) patterns of the samples were carried out on a D-8 Advance X-ray diffractometer (Bruker-AXS, Germany) using Cu K radiation (λ = 1.5406 Å) with a step size of 0.1° over a 2θ range from 5° to 80°w ith the electric current operated at 40 mA and the voltage being xed at 40 kV. The pH of the point of zero charge (pH pzc ) was identi ed by measuring the ζ-potential with a zeta potential analyzer (Nano Zetasizer 2000, Malvern Co., UK).

Batch adsorption experiments
Phosphate removal experiments by the materials were conducted at room temperature (25 ± 1°C) with the target phosphate concentration and an adsorbent dose of 0.5 g/L unless otherwise noted. Before the adsorbent was added, the pH of the solution was adjusted to the required values (± 0.2) and maintained at target pH values throughout the experiments by addition the NaOH or HCl solution (1 mol/L). After the addition of the adsorbents, the mixture suspensions were shaken in a thermostatic shaker at 180 rpm. Following the reaction period, samples were collected through a 0.22 µm membrane syringe lter for phosphate concentration analysis. The residual phosphate concentrations of the supernatant solutions were determined by a molybdenum-blue ascorbic acid method with a UV-Vis spectrophotometer at 880 nm (UV-2550, Shimadzu, Japan) (Chen et al. 2015). All experiments were repeated in triplicate, and the average values were used to analyze the results. The removal e ciency (RE) was determined as follows: where the initial concentration of the phosphate solution is C 0 (mg/L) and the equilibrium concentration of the phosphate solution is C e (mg/L), the volume of the phosphate solution is V (L) and the weight of the adsorbent used in the experiments is m (g).
The adsorption capacity of phosphate removed by the adsorbent (q e , mg•g −1 ) was calculated as the difference between the initial and residual phosphate concentrations in solution divided by the weight of the adsorbent as followed: where C 0 and C e are same as mentioned above, V is the volume of the phosphate solution (L) and m is the weight of the adsorbent (g).
Adsorption isotherms, adsorption kinetics and the effect of pH and coexisting anions were performed in order to evaluate the performance of phosphate adsorption by the adsorbents. The phosphate adsorption kinetics were carried out by adding 0.125 g of Fe-La binary (hydr)oxides to a 250 mL phosphate solution with an initial concentration of 50 mg/L at pH 4.0. Approximately 2 mL samples were taken out at predetermined times and ltered immediately. The adsorption isotherm experiments were investigated by adding 0.05 g of adsorbent into 100 mL of phosphate solutions with different initial concentrations (30.0-80.0 mg•L −1 ) at pH 4.0 for 24 h. The in uence of solution pH on phosphate adsorption was measured by adding 0.05 g of Fe-La binary (hydr)oxides to phosphate solutions of 50 mg/L under the different pH values ranging from 3 to 11 for 24 h. To explore the effect of common coexisting anions, 0.05 g of Fe-La binary (hydr)oxides were added to 100 mL phosphate solutions (50 mg/L) in the presence and absence of NaCl, Na 2 SO 4 , NaHCO 3 , and NaCH 3 COO, respectively, and the initial concentrations of the coexisting anions were set in the range of 10-100 mmol•L −1 . Specially, the CH 3 COO − anion is a representative of natural organic acids (NOAs) in the tests. For the reusability study, adsorption and desorption experiments were carried out for four cycles at pH 7.0, and the desorption experiments were conducted over 4 h using 0.2 mol/L NaOH solution. The initial phosphate concentration was 50.0 mg/L, and the adsorbent dose was 0.5 g/L

Results And Discussion
Characterization of materials Scanning electron microscopy The surface morphologies and element distributions of the Fe-La binary (hydr)oxides before and after phosphate adsorption were characterized by SEM-EDX. As illustrated in Fig. 1a, SEM images show that the Fe/La 3:1 adsorbent has rough surfaces and consists of irregular and aggregated small particles with different sizes. Groups of aggregated blocks covered with granules were formed on the surface of the Fe/La 1:1 adsorbent (Fig. 1b). For the Fe/La 1:3 adsorbent, the grains were consisted of nano ake aggregates and smaller particles of various sizes (Fig. 1c). With the decrease of the Fe/La ratios, an increasing number of morphologies and structures similar to those of pure La (hydr)oxide were observed on the surface of the adsorbents . After phosphate adsorption, the surface of the adsorbents became rougher and more irregular, and needle-like nanocrystals were formed, as presented in Fig. 1d-1f. The needle-like nanocrystals corresponding to CePO 4 were also observed after phosphate adsorption by biochar-loaded Ce 3+ -enriched ultra-ne ceria nanoparticles ). There were no obvious discrepancies among the phosphate-loaded adsorbents with different Fe/La molar ratios. Furthermore, the EDX surface element analysis of the adsorbents after phosphate removal ( Fig. 2) revealed that phosphate was successfully removed by the adsorbents. Furthermore, the contents of phosphate element on the surface were approximately 4.42 wt% for the Fe/La 3:1 adsorbent, 10.49 wt% for the Fe/La 1:1 adsorbent and 14.64 wt% for the Fe/La 1:3 adsorbent, indicating that the performance of phosphate adsorption was enhanced with the decrease of the Fe/La ratios. This result was further con rmed by the results of the batch adsorption experiments described below.
Speci c surface area The N 2 adsorption-desorption isotherms and the pore size distributions estimated by the BJH analysis using desorption data of the Fe-La binary (hydr)oxides are depicted in Fig. 3a. As shown in Fig. 3a, the adsorption-desorption isotherms of the Fe/La binary (hydr)oxides were attributed to the characteristics of type IV isotherms with H3-type hysteresis loops according to the IUPAC classi cation, suggesting that the multilayered physical adsorption of N 2 occurred between the aggregates of the adsorbents. These results indicate that the materials were possibly mesoporous, as presented in Fig. 3a.
The speci c surface areas, t-plot micropore areas, average pore diameters, and average pore volumes of Fe-La binary (hydr)oxides obtained from the N 2 adsorption-desorption isotherms are summarized in Table 1. It can be seen that the Fe/La 1:3 (hydr)oxides showed the highest speci c surface area of 137.1 m 2 /g, while the Fe/La 1:1 (hydr)oxides exhibited the lowest speci c surface area of 16.3 m 2 /g. For the tplot micropore area and average pore volume, the Fe-La binary (hydr)oxides showed similar trends with changes in the Fe/La ratios. These observations indicated that the incorporation of Fe and La metals on the Fe/La 1:1 (hydr)oxides was different from that of the other two type Fe-La (hydr)oxides (Yu et al. 2019). However, the average pore diameters were 35.2, 103.3 and 192.8 Å for the Fe/La 3:1, Fe/La 1:1 and Fe/La 1:3 (hydr)oxides, respectively, and increased with the increase in La contents, which may be attributed to the change of the particle shape from irregular and aggregated particles to nano akes, as observed by SEM in Fig. 1. Previous researches demonstrated that the pH pzc of ferrihydrite was pH 7-9 (Masue et al. 2007), and that the pH pzc of lanthanum hydroxide was pH 9-10.5 (Xu et al. 2017. The surface charge of the adsorbent is highly affected by solution pH. When the pH value of the solution is below their pH pzc , positive charges would be formed on the surface of the adsorbent, and the stronger electrostatic attraction between active sites of the adsorbent and phosphate is favorable for the e cient phosphate removal. On the other hand, the surface of the adsorbent is negatively charged when pH > pH pzc , which is not conducive to the uptake of phosphate due to the enhancement of electrostatic repulsion.

Batch adsorption experiments
Adsorption kinetics In order to estimate the rate of phosphate removal, the amounts of phosphate adsorbed onto Fe-La binary (hydr)oxides as a function of contact time were conducted at pH 4.0 with an initial phosphate concentration of 50.0 mg/L and the results are shown in Fig. 4a. It is clearly that the adsorption rates were rapid within the rst 6 h, and thereafter followed by a relatively slow adsorption step. Approximately 67.6%, 87.9% and 100% of the equilibrium adsorption capacities of Fe/La 3:1, Fe/La 1:1 and Fe/La 1:3 binary (hydr)oxides were achieved within the rst 6 h, respectively. It should be noted that nearly 99.3% of the total phosphate contents had been removed by the Fe/La 1:3 (hydr)oxides after 1.5 h of reaction. The initial rapid adsorption behavior might be attributed to the large amounts of unoccupied and freely available adsorption sites, which was favorable for the transportation of phosphate onto the surface of the adsorbents. With the decrease in the number of vacant sites, surface precipitation and intraparticle diffusion were the dominant mechanisms of phosphate adsorption, and the adsorption rate was slow (Lu et al. 2013).
The kinetic data of phosphate adsorption were tted using pseudo-rst-order model and pseudo-secondorder models in order to further understand the characteristics of the adsorption process. These two models can be expressed as: where q e (mg/g) and q t (mg/g) represent the amounts of phosphate adsorbed at time t and at equilibrium, respectively, and k 1 (h − 1 ) and k 2 (g/(mg•h)) are the rate constants of the pseudo-rst-order and pseudo-second-order models, respectively.
The tting parameters estimated from the kinetic models are represented in Table 2. It is obvious that the experimental data were described better by the pseudo-second-order equation model with higher values of correlation coe cient (R 2 > 0.99) being observed. This indicates that the dominant adsorption process could be attributed to chemisorption or chemical bonding between the active sites of the adsorbents and phosphate. In addition, the calculated k 2 values of phosphate removal derived from the pseudo-secondorder equation model for Fe/La 3:1, Fe/La 1:1 and Fe/La 1:3 binary (hydr)oxides were 0.0084, 0.0162 and 0.0333 g/(mg•h), respectively, indicating that the increase of the La contents in Fe-La binary (hydr)oxides not only increased the adsorption capacity at equilibrium, but also enhanced the adsorption rate.  Fig. 4b. It is seen that the amount of phosphate uptake increased as the initial concentration increased, and that the adsorption capacity increased signi cantly with the decrease of the Fe/La molar ratios. Both Langmuir and Freundlich adsorption isotherm models were employed to quantitatively describe the experimental data. The Langmuir and Freundlich isotherm models can be represented as follows (Wang et al. 2016, Zhang et al. 2021): where C e (mg/L) is the equilibrium concentration of phosphate, q e (mg/g) and q m (mg/g) are the equilibrium and maximum adsorption capacities, respectively, K L (L/mg) and K F (mg/g) are the adsorption constants for the Langmuir and Freundlich models, respectively, which could be used to roughly represent the a nity of phosphate adsorption. 1/n is a heterogeneity factor related to the adsorption intensity.
The adsorption isotherm tting curves are shown in Fig. 4b and the parameters acquired from the Langmuir and Freundlich adsorption isotherm models are listed in Table 3. As illustrated in Moreover, Table 4 summaries compared the maximum adsorption capacities of the Fe-La binary (hydr)oxides from this study with those of some other La-based adsorbents previously reported. As illustrated in Table 4, the Fe-La binary (hydr)oxides shows much better performance than most of the previously reported adsorbents, indicating their speci c potential for phosphate removal.

Effect of pH on phosphate removal
Solution pH is one of the most important factors in phosphate removal because it can affect both the surface charge of adsorbents (Fig. 3b) and phosphate species (Fig. S1). Herein, the effect of solution pH on phosphate removal by Fe-La binary (hydr)oxides was investigated in the pH range of 3-11 with an initial phosphate concentration of 50 mg/L and the results are presented in Fig. 4c. Apparently, the uptake of phosphate by the Fe-La binary (hydr)oxides was remarkably dependent on solution pH. The greatest degree of adsorption was observed under acidic conditions, and the removal e ciency reduced dramatically with the increase of solution pH. The decline of phosphate adsorption capacity with the increase of pH has also been reported in previous literatures (Lu et al. 2013, Zhang et al. 2021). However, this trend is signi cantly distinct with that by La(OH) 3 /Fe 3 O 4 nanocomposite, by which the highest level of phosphate removal was observed between pH 4.3 and 6 ). This discrepancy might be due to the formation of surface co-precipitation of lanthanum phosphate under acidic conditions , Yu et al. 2019. Furthermore, the phosphate capacity of Fe-La binary (hydr)oxides increased with the decrease of the Fe/La molar ratios with respect to the same pH value. This observation was similar to those of adsorption kinetics (Fig. 4a) and adsorption isotherms (Fig. 4b), and was further con rmed by EDX surface elemental analysis (Fig. 2).
As given in Fig. S1, monovalent H 2 PO 4 − was the main species of phosphate in the pH range of 3-6, and the surfaces of the adsorbents were fully protonated. The decrease of pH was favourable for protonation. Consequently, the positively charged adsorbents could effectively interact with the negatively charged H 2 PO 4 − through electrostatic attractions, thus leading to the enhanced phosphate sorption capacity under acidic conditions. With the increase in solution pH, divalent HPO 4 2− became the dominant species in the pH range of 7-11, and the surfaces of the adsorbents were deprotonated. The surface charge of the adsorbents could become negative when the pH value of the solution was above the pH pzc of the adsorbent. The negatively charged adsorption sites and the negatively charged HPO 4 2− gradually promoted the electrostatic repulsion, which would be unfavourable for phosphate removal. What's more, the lower a nity of the surface of La(OH) 3 for HPO 4 2− compared to that for H 2 PO 4 − due to the hydroxylation of lanthanum ions also resulted into the sharply decrease of the phosphate removal under alkaline conditions , Zhang et al. 2021).
In addition, the release of La and Fe in solution after phosphate adsorption at different pH values was evaluated as shown in Fig . Therefore, a rather low solubility product constant (K sp = 3.7 × 10 − 23 ) suggests that the insoluble lanthanum phosphate (LaPO 4 ) precipitation was formed under acidic conditions, which was further discussed as below.

Regeneration and reusability
To estimate the reusability of the Fe-La binary (hydr)oxides, their regeneration using 0.2 mol/L NaOH solution and re-adsorption were performed. Four reusability cycles were carried out, and the results were depicted in Fig. 4d. The adsorption capacities of cycle 0 correspond to those of the origin adsorbents.
Apparently, the adsorption capacities of the adsorbents signi cantly decreased after the rst cycle and exhibited the lower and more stable adsorption capacities after two cycles, which could be attributed to the formation of the insoluble lanthanum phosphate (LaPO 4 ) co-precipitation.

Effect of co-existing anions
Environmental phosphate always coexists with various anions in natural water environments or industrial e uents. The presence of competitive anions may interfere with phosphate adsorption. In this study, the effects of Cl − , SO 4 2− , HCO 3 − , and acetate (Ac − ), as a representative of NOAs, on phosphate removal by the Fe-La binary (hydr)oxides were investigated at pH 7.0 with an initial phosphate concentration of 50.0 mg/L. As presented in Fig. 5, it is obvious that the uptake of phosphate was negligibly affected by the presence of Cl − , and SO 4 2− , even at concentrations of up to 100 mmol/L. however, there was slight interference from HCO 3 − , and acetate. Similar effects with respect to these competing ions on phosphate removal have been reported in previous studies ).

Analysis of Fe-La binary (hydr)oxides before and after phosphate adsorption
In view of the above experimental results, a series of analytical techniques, including p-XRD, FTIR and XPS, were employed to characterize the Fe-La binary (hydr)oxides before and after phosphate adsorption in order to further investigate the mechanisms of phosphate removal.
XRD analysis p-XRD spectra were used to elucidate the phase compositions of the Fe-La binary (hydr)oxides before and after phosphate adsorption. Figure 6a presents the p-XRD patterns of these materials, which are in were destroyed due to the interaction between phosphate and Fe-La binary (hydr)oxides. These results reveal that the formation of rhabdophane was the main mechanism of phosphate removal.

FTIR analysis
FTIR spectra of the Fe-La binary (hydr)oxides before and after phosphate adsorption are depicted in Fig.   S3. As illustrated in Fig. S3, it is common sense for us that broad vibration bands located at 3433 and 1634 cm − 1 are observed for all materials, which can be ascribed to the stretching vibration of adsorbed water molecules and the bending vibration of the -OH group, respectively , Yu et al. 2019). As shown in Fig. 6b, peaks of pristine (oxy)hydroxides located at 462 cm − 1 correspond to the stretching of Fe − O bonds, and weak band appearing at 1050 cm − 1 can be attributed to the bending vibrations of La − OH bonds (Li et al. 2012, Zhang et al. 2012. Peaks centered at 1388 cm − 1 are assigned to the vibration mode of NO 3 − anions due to the residues from the raw synthetic materials of La(NO 3 ) 3 •6H 2 O according to the previous literatures (Li et al. 2012), and its intensity signi cantly decreased after phosphate removal. However, Zhang et al. demonstrated that the vibration bands at 1490 cm − 1 and 1394 cm − 1 were the characteristics of the lanthanum (hydr)oxide ).
Furthermore, these peaks have also been reported by other binary (hydr)oxides, such as Mn-La composites (Yu &Chen 2014). Therefore, the attribution of the peaks at 1480 cm − 1 and 1388 cm − 1 needs further research. After phosphate adsorption, new strong vibration bands at 1055 cm − 1 and new shoulder vibration bands at 1018 cm − 1 are observed and can be respectively assigned to the asymmetric and the symmetric stretching vibrations of P − O bonds, respectively. Two other new peaks appeare at 617 cm − 1 and 538 cm − 1 , which belonged to the bending vibrations of O − P bonds . The FTIR results suggest that chemical interactions between phosphate and the Fe-La binary (hydr)oxides occurred, resulting in the formation of inner-sphere complexes (Chen et al. 2019).
XPS analysis X-ray Photoelectron Spectroscopy is considered as a powerful technique for the characterization of the surface electronic structure and bonding con guration of materials. As illustrated in Fig. 7a, the wide scan XPS spectra of the pristine Fe-La binary (hydr)oxides indicate that Fe, La and O are present in the composites. After phosphate adsorption, the appearance of the P 2p characteristic peak at ~ 133.8 eV demonstrates the successful loading of phosphate on La-Fe binary hydroxides. Figure 7b and 7c present the XPS spectra of La 3d and Fe 2p before and after phosphate adsorption, respectively. The XPS spectra of La 3d are mainly consist of La 3d 5/2 and La 3d 3/2 due to the spin-orbit splitting . The representative peaks of La 3d located at ~ 835.1 and ~ 839.0 eV and those centered at ~ 852.2 and ~ 855.9 eV can be assigned to La 3d 5/2 and La 3d 3/2 , respectively (Yu et al. 2019, Yu &Chen 2014 with the increase of Fe contents, and to lower values with the decrease of Fe contents. These discrepancies in the change of the binding energies as the variety of the metal composition indicate that the electronic migration between La and P due to the adsorption could be different from that between Fe and P. These divergences further resulted in irregular energy changes in the binding energies corresponding to the peaks of P 2p, which were located at 133.4 eV for the Fe/La 3:1 (hydr)oxides, 134.4 eV for the Fe/La 1:1 (hydr)oxides and 133.8 eV for the Fe/La 1:3 (hydr)oxides (Fig. 7d).
The high resolution scans of O 1s spectra of the Fe-La binary (hydr)oxides before and after phosphate adsorption were employed to further investigate the mechanisms of phosphate adsorption. As depicted in Fig. 8, the binding energies corresponding to the peaks of O 1s after phosphate removal shifted from 530.8 eV to 531.4 eV for the Fe/La 3:1 (hydr)oxides, from 532.0 eV to 531.7 eV for the Fe/La 1:1 (hydr)oxides and from 531.7 eV to 531.0 eV for the Fe/La 1:3 (hydr)oxides.  (Lu et al. 2013).

Removal mechanisms
Based on the above analysis, the main mechanisms of phosphate removal by Fe-La binary (hydr)oxides are proposed to be based on the formation of rhabdophane through surface hydroxyl exchange reactions and co-precipitation of La 3+ and phosphate ions. Surface hydroxyl exchange reactions were the predominate adsorption mechanisms and can be described as following (where ≡ S represents the surface): Under acidic and neutral conditions, H 2 PO 4 − is the predominant species in solution, and La 3+ ions were released into solution. In addition to surface hydroxyl exchange reactions, co-precipitation of released La 3+ and phosphate ions led to the formation of insoluble rhabdophane as following: Under alkaline conditions, the release of La decreased dramatically, and surface hydroxyl exchange reactions became the main removal mechanisms, resulting into a sharp reduction in removal e ciency with the increase of solution pH.

Conclusions
In summary, Fe-La binary (hydr)oxides were prepared by a co-precipitation method and their performances of phosphate removal were studied. The binary (hydr)oxides exhibited excellent performances in phosphate removal. The adsorption capacity of phosphate increased with the increase of La contents, and the calculated Langmuir capacities of Fe/La 3:1, Fe/La 1:1 and Fe/La 1:3 binary (hydr)oxides at pH 4.0 were 49.02, 69.44 and 136.99 mg/g, respectively. The removal e ciency was remarkably dependent on solution pH and signi cantly reduced with the increase of pH. Based on the analyses of p-XRD, FTIR and XPS, the formation of amorphous rhabdophane phase through surface hydroxyl exchange reactions and co-precipitation of released La 3+ and phosphate ions were the dominant removal mechanisms. The high removal e ciencies of Fe-La binary (hydr)oxides indicates their speci c potential on the treatment of P-contaminated water.