Advanced removal of phosphorus from urban sewage using chemical precipitation by Fe-Al composite coagulants

Phosphorus (P) removal is a significant issue in wastewater treatment. This study applies Fe-Al composite coagulant to the advanced treatment of different P forms in biological effluent. For 90% total P removal, the dosage of FeCl3-AlCl3 composite coagulant reduces by 27.19% and 43.28% than FeCl3 and AlCl3 only, respectively. Changes in effluent P forms could explain the phenomenon of composite coagulant dosage reduction. The suspended P in the effluent of composite coagulant is easier removed by precipitation than single coagulant. In this study, the hydrolysis speciations of Fe3+, Fe2+, and Al3+ at a pH range are calculated by Visual MINTEQ. Changes in the morphology of metal hydroxides correlate with P removal at pH 4–9. Besides, analyses of scanning electron microscope (SEM), Fourier transformed infrared (FTIR), and X-ray photoelectron spectroscopy (XPS) are performed on the coagulation precipitations. Fe2+ reacts directly with P to form flocs of Fe3(PO4)2, and Al2(SO4)3 assists in the sedimentation of the small-volume flocs. Al13 is a significant hydrolysis product of Al3+, and Fe and P would substitute for the peripheral AlVI of the Al13 structure to form stable Fe–O–Al covalent bonds.

In 2017, 95,400 tons of total phosphorus (TP) was released from domestic sources in China, which was the second source of TP after agricultural sources (released 212,000 tons of TP) 1 .Phosphorus (P) is one of the essential nutrients for plant growth.We have investigated the urban sewage plants around Zhengzhou city, and the average TP content of biological treatment effluent was 0.203 mg/l (Table SI1).However, slow-flowing water could be eutrophic when the P concentration exceeds 0.02 mg/l 2 .The main phenomena of eutrophication in water bodies include algae and plankton blooms and a large number of fish and shrimp deaths due to reduced dissolved oxygen 3,4 .Therefore, advanced removal of P from biotreated sewage effluent is an important method to prevent and control eutrophication in water bodies.
The main methods commonly used to remove P from wastewater are biological treatment, adsorption and chemical precipitation 5 .In the biological P removal treatment, microorganisms accumulate P beyond normal requirements for metabolic processes 6 .However, the TP removal efficiency of the biological treatment is frequently hindered by different operational and system constraints.In reports, the TP removal of biological treatment was susceptible to effects by the temperature, hydraulic retention time, and reaction volume [7][8][9] .In addition, TP in the effluent of biological treatment tends beyond the permissible limit.Falahati-Marvast and Karimi-Jashni reported that the optimal TP content of 0.7 mg/l in the effluent of a pilot-scale bioreactor exceeded the discharge limit of P for urban wastewater treatment plant (GB18919-2002) 10 .Therefore, physicochemical processes of P removal are usually combined with biological treatment in application practice.
Physicochemical methods such as adsorption and precipitation are common advanced P removal technology.However, the limitation of adsorption is the P sorption capacity of the absorbents 11 .Han et al. reported a decrease in the P removal efficiency of absorbent with a period of usage because of P saturation 12 .Chemical precipitation is one of the most common advanced P removal techniques.The metal salts, such as ferric chloride and alum, combine with P in the wastewater, and the coagulation flocs are removed by sedimentation or filtration.Chemical precipitation has been proven to be an effective process and is widely applied in urban wastewater treatment

Water sample
According to the Ammonium Molybdate Spectrophotometric Method (GB11893-89), the measured water sample should be acidified pretreatment to inhibit the effect of microbial metabolism on TP determination.However, acidification will change the P forms of the water sample 27 .In this study, to investigate the process of combining composite coagulants with different P forms in the actual wastewater, the water sample was obtained from the effluent of a pilot-scale self-cleaning activated bio-filter (Fig. SI1).Table 1 shows the water quality of the effluent when the self-cleaning activated bio-filter was stable.Comparing the P forms composition, effluent from the self-cleaning activated bio-filter could properly represent the urban sewage effluent (Table SI1).

Effect of coagulant dose on TP removal efficiency
A six-paddle stirrer (JJ-3A, Olabo) was used in the coagulation experiments.In each jar test, 300 ml of selfcleaning activated bio-filter effluent was added with 5, 10, 15, 20, 25, and 30 mg/l of composite coagulant solutions with Fe/Al mass ratios of 0.5, 1, and 2, respectively.The mixtures were rapidly stirred for 1 min at 200 rpm, followed by a slow stirring for 30 min at 30 rpm, then settling for 30 min at room temperature (25 ± 1 ℃).After settling, the supernatant samples were obtained to measure TP concentration and P forms.
We did not additionally add humic substances to the effluent of the self-cleaning activated bio-filter because humic substances had little effect on P removal by coagulants (Fig. SI2).

Analysis of phosphorus forms
The analysis of P forms is divided into two steps: converting the P forms of interest to dissolved orthophosphate and determining the concentration of dissolved orthophosphate 28 .The collected water sample of the coagulation experiment was digested at 120 ℃ for 30 min, and then the TP was measured by colorimetry with a UV-vis spectrophotometer (UV-6300, Mapada).Total dissolved P was determined by colorimetry through a 0.45 μm filter membrane before digestion.Total suspended P was calculated as the difference between the TP and total dissolved P. The total orthophosphate was measured by direct colorimetry, and the total dissolved orthophosphate was measured by colorimetry through a 0.45 μm filter membrane without digestion.The total organic P was calculated as the difference between the TP and orthophosphate 29 .The analysis process of the P forms is shown in Fig. 1.

Analysis of hydrolysis speciation
Visual MINTEQ 3.1 was used to calculate the distribution coefficients (δ) of Me 3+ , Me(OH) 2+ , Me(OH) 2 + , Me(OH) 3 , and the polynuclear hydroxyl complexes of Al 3+ , Fe 3+ and Fe 2+ at a pH range.In this study, the setup conditions of the software were pH range from 1 to 13, temperature 25 ℃, ionic strength 0.001, and initial metal ions concentration 1 mol 30 .

Analysis of coagulation precipitate
After settling, the coagulation precipitates were carefully taken from the beaker and dried for several hours.The morphology and structure of the precipitates were observed with a scanning electron microscope (Apreo 2, Thermo Fisher).The characteristics of the precipitates were analyzed by an FTIR spectrometer (Nicolet iS50, www.nature.com/scientificreports/Thermo Fisher).The chemical compositions of the precipitates were analyzed by an X-ray photoelectron spectrometer (K-Alpha, Thermo Fisher).

Effect of coagulant dosage
In advanced P removal, coagulant dosage is essential in affecting TP removal efficiency, application cost, and toxicity in the effluent.Figure 2a illustrates that the TP removal rate of the FeCl 3 -AlCl 3 composite coagulant is higher than FeCl 3 and AlCl 3 only at 10-30 mg/l.Compared with single coagulants of FeCl 3 and AlCl 3 , the FeCl 3 -AlCl 3 composite coagulant can significantly reduce the dosage and cost of P removal by chemical precipitation.For the 90% TP removal, the optimal dosages of single FeCl 3 , AlCl 3 , and FeCl 3 -AlCl 3 are 30.01,38.52, and 21.85 mg/l, respectively.The TP removal rate of AlCl 3 is lower than FeCl 3 .As a result, Fe 3+ have a higher affinity for P and hydrolyses more rapidly than Al 3+31 .At the same time, the TP removal rates of the composite coagulant at the dosage of 5 mg/l and 30 mg/l (32.08%, 93.89%) are approaching the FeCl 3 only (28.64%, 89.99%), respectively.Yang et al. reported that the excess hybrid coagulant had a less beneficial effect on turbidity removal 32 .In reports, coexisting anions had a limited influence on P removal by physicochemical methods 33 .However, the TP removal rates of the FeCl 3 -Al 2 (SO 4 ) 3 and the FeCl 3 -PAC are less than the FeCl 3 -AlCl 3 (Fig. SI3).This phenomenon may be attributed to the reduced solubility of the precipitate due to the common ion effect.Figure 2b shows that the TP removal rate of the FeSO 4 -Al 2 (SO 4 ) 3 composite coagulant at the FeSO 4 /Al 2 (SO 4 ) 3 mass ratio of 2 is significantly higher than single FeSO 4 and Al 2 (SO 4 ) 3 .The TP removal rate of FeSO 4 -Al 2 (SO 4 ) 3 composite coagulant is 83.71% higher than FeSO 4 (72.2%) and Al 2 (SO 4 ) 3 (48.48%)at the dosage of 15 mg/l.With reducing the FeSO 4 /Al 2 (SO 4 ) 3 mass ratio, the TP removal rate of the FeSO 4 -Al 2 (SO 4 ) 3 composite coagulant shows a decreasing tendency.The Fe/Al mass ratio significantly impacts the P removal efficiency of Fe 2+ -based composite coagulants.For the 90% TP removal, the optimal dosages of single FeSO 4 , Al 2 (SO 4 ) 3 , and FeSO 4 -Al 2 (SO 4 ) 3 are 25.43, 41.05, and 18.25 mg/l, respectively.Guan et al. reported that the application of Fe 2+ with metal ions could increase the surface charge and produce more precipitated ferrous hydroxide or ferric hydroxide 34 .Compared with FeSO 4 -Al 2 (SO 4 ) 3 , the TP removal rates of the FeSO 4 -AlCl 3 and FeSO 4 -PAC are less improved than the single FeSO 4 (Fig. SI4).
We have investigated the relationship between injection orders of Fe-Al composite coagulants and P removal (Fig. SI5).There is little influence of injection order on P removal efficiency, so we will not discuss the injection order in the following.We have also tested the P removal efficiencies of PFS-AlCl 3 and PFS-Al 2 (SO 4 ) 3 , and the test results show that FeCl 3 -AlCl 3 and FeSO 4 -Al 2 (SO 4 ) 3 have the most obvious promotion effect on P removal (Fig. SI6).

Variations of phosphorus forms
Investigating the change of P forms under coagulation treatment contributes to exploring the combining process of coagulant with P. Figure 3a and b illustrate that the dissolved orthophosphate and suspended organic P are the primary P forms in the self-cleaning activated bio-filter effluent.Metal salts as coagulants are added to the wastewater to form crystalline precipitates, which adsorb the dissolved P on the surface of the precipitates and transform to suspended P 35 .The suspended P is subsequently separated by gravity.The dissolved orthophosphate is mostly removed after dosing 15 mg/l coagulant, and the FeSO 4 -Al 2 (SO 4 ) 3 composite coagulant achieved the highest dissolved orthophosphate removal rate of 96.76%.An increase of suspended orthophosphate in the coagulation effluent indicates that the coagulant combines with dissolved orthophosphate to form suspended orthophosphate flocs 36 .There are 76.79% and 57.14% removal of suspended organic P under AlCl 3 and Al 2 (SO 4 ) 3 treatment, respectively.The removal of suspended organic P is mainly by adsorption-bridging and sweep coagulation 37 .Compared with FeSO 4 of 25% removal for suspended organic P, it indicates the removal of suspended P by FeSO 4 through ionic layer compression and electrical neutralization.The suspended orthophosphate content of the FeSO 4 coagulation effluent is 0.185 mg/l, and the FeSO 4 -Al 2 (SO 4 ) 3 composite coagulant In this study, we observed the increased dissolved organic P concentration of the coagulation effluent, with a 6.6 times growth rate of dissolved organic P by FeCl 3 .This phenomenon is due to the limitation of the P forms analytical method, which ignores that a proportion of inorganic P, such as polyphosphates, cannot be measured by direct colorimetry 28 .

Effect of pH
pH depends on the degree of reaction between hydroxyl and metal ions, thus affecting the bridging flocculation 38 .Figure 4a and b show that the TP removal rates of FeCl 3 -AlCl 3 and FeSO 4 -Al 2 (SO 4 ) 3 composite coagulants are higher than single coagulants (FeCl 3 , FeSO 4 , AlCl 3 , and Al 2 (SO 4 ) 3 ) when 6 < pH < 9. Li et al. reported that composite coagulants would hydrolyze to produce long, complex, and stable reaction bonds, which are difficult to destroy by the change in pH 39 .The optimal TP removal of FeCl 3 -AlCl 3 and FeSO 4 -Al 2 (SO 4 ) 3 are 91.31% and 86.82% at pH 5 and 7, respectively.In a weakly acidic or neutral solution, the adsorption sites on the surface of the hydroxide produced by the hydrolysis of metal ions could adsorb a large amount of P.However, the production of metal hydroxides from coagulants would be inhibited in acidic or alkaline solutions.The composite coagulants show a wider pH range with better P removal efficiency.The Fe/Al mass ratio is negatively correlated with the TP removal rate at pH 4-9.The TP removal rates of FeSO 4 -Al 2 (SO 4 ) 3 are 86.55% and 21.5%, corresponding to the FeSO 4 /Al 2 (SO 4 ) 3 mass ratio of 2 and 0.5, respectively.FeSO 4 is prone to change in pH because Fe 2+ reacts with OH -in the solution to form soluble Fe(OH) 2 , which is easily oxidized to Fe(OH) 3 by dissolved oxygen in wastewater.The oxidation reaction is inhibited at low pH due to insufficient OH -in the wastewater, resulting in the decreased removal of TP 40 .pH affects the hydrolysis product of metal ions and P form in solutions 41 .
Figure 5a shows that the hydrolysis products of Fe 3+ are mainly mononuclear hydroxides such as Fe(OH) 2 + , Fe(OH) 4 -or polynuclear hydroxyl complexes such as Fe 3 (OH) 4 5+ .Figure 5b shows that the hydrolysis products  www.nature.com/scientificreports/ of Fe 2+ are mainly Fe 2+ and mononuclear hydroxides such as FeOH + and Fe(OH) 3 -.Fe 2+ has lower P removal efficiency than Fe 3+ in practical applications because Fe 2+ does not hydrolyze to produce polynuclear hydroxyl complexes 43 .Figure 5c shows that the main hydrolysis products of Al 3+ are Al(OH) 4 -and Al 3 (OH) 4 5+ , which are similar to the hydrolysis products of Fe 3+ .Compared to Fig. 4a and b, the optimal TP removal rates of Al salts are achieved at the pH range of 5-6, corresponding to the main hydrolysis product of Al 3+ is Al 3 (OH) 4

5+
. It indicates that the hydroxides produced by Fe 3+ and Al 3+ play a significant role in P adsorption, and Fe 2+ combines directly with P to form the precipitate of Fe 3 (PO 4 ) 2 .

SEM analysis
Figure 6a shows that the structure of the FeCl 3 coagulation precipitate is porous and loose with an irregular and partly smooth surface.Moreover, the FeSO 4 coagulation precipitate consists of Fe 3 (PO 4 ) 2 particles with a diameter of around 0.1 μm (Fig. 6b). Figure 6c shows that the morphology of the AlCl 3 is petal-like with an average length of 0.25 μm, similar to threadlike NaCl crystal 44 .The distinctions in the structure of the AlCl 3 and Al 2 (SO 4 ) 3 (Fig. 6d) precipitates are due to the different anions, such as Cl -and SO 4 2-, involved in combining the coagulant with the contaminant 45 .Figure 6e shows that the structure of the FeCl 3 -AlCl 3 coagulation precipitates is significantly different from the FeCl 3 and AlCl 3 .The change in structure indicates that the composite coagulant modifies the binding mode of P. Figure 6f shows that the structure of the FeSO 4 precipitate is similar to the precipitate of FeSO 4 -Al 2 (SO 4 ) 3 , which consists of several 0.1 μm diameter Fe 3 (PO 4 ) 2 particles.According to Sect."Variations of phosphorus forms", it is reasonable to speculate that the P removal process of FeSO 4 -Al 2 (SO 4 ) 3 composite coagulant is mainly the reaction of Fe 2+ with P to form Fe 3 (PO 4 ) 2 flocs, and the hydrolysis product of the Al 2 (SO 4 ) 3 performs the adsorption-bridging and sweep coagulation to promote the Fe 3 (PO 4 ) 2 flocs settling.

FTIR analysis
As shown in Fig. 7, the stretching vibration at 3300 cm -1 is assigned to the O-H, which is due to the absorbed water and hydroxyl group on the surface of the metal ions hydrolysis product or the adsorbed substance.The strong bending vibration at 1600 cm -1 is assigned to physically adsorbed H 2 O 14 .The strong bending vibrations at 1510 to 1210 cm -1 are assigned to the NO 3 , which is due to the absorption of NO 3 -from the nitrogen source of the simulated wastewater and biological metabolism.The stretching vibration at 1000 cm -1 is assigned to the P-OH due to the phosphorus absorption by the polynuclear hydroxyl complexes of the metal ions hydrolysis produced.For the Fe 3+ /Fe 2+ salts and the Fe-Al composite coagulants coagulation precipitates, the stretching vibration at 830 cm -1 is assigned to the P-O.It indicates that Fe-O-P is formed due to the direct reaction of the Fe 3+ /Fe 2+ with PO 4 3-.And for the Al 3+ salt coagulation precipitate, the absorption band at 535 cm -1 is assigned to the Al-O vibrations of aluminum in the octahedral coordination 46 .

XPS analysis
C, O, Fe, Al and P are the primary constituent elements of coagulation precipitate (Fig. SI8).As shown in Fig. 8a and b, the O1s spectra of the precipitate are deconvoluted into three peaks by the XPSPEAK41.The red peak at the bonding energy of 530 eV is assigned to O in Fe-O-Fe and Al-O-Al, which is due to Fe 3+ , Fe 2+ , and Al 3+ directly binding with O.The blue peak at the bonding energy of 532 eV is assigned to O in Fe-O-H and Al-O-H, which is due to Fe and Al binding with O in the hydroxyl group 47 .The yellow peak at the bonding energy of 533.3 eV is assigned to O in H 2 O due to the adsorption of the hydroxyl complexes.The relative area of the peak represents the content of the elemental form in the precipitate, demonstrating that the primary components of the coagulation precipitate are Fe(OH) 3 , Al(OH) 3 , and polynuclear hydroxides 48 .The peak of P2p spectra of the coagulation precipitate is located at 133.4 eV, which is located between the peaks of the Fe salts precipitate (at 133.14 eV) and the Al salts precipitate (at 133.75 eV) (Fig. SI9). Figure 8c and d show the Fe2p spectra of the coagulation precipitate.It indicates that the peaks at the bonding energies of ~ 711 eV and ~ 724.5 eV are assigned to the Fe2p 3/2 and Fe2p 1/2 , respectively.The positions of the Fe2p 3/2 and Fe2p 1/2 peaks are determined by the element valence state of the Fe 49,50 .The energy separation between Fe2p3/2 and Fe2p 1/2 is 13.5 eV, which is in agreement with FePO 4 in the report 51 .After the deconvolution, the peaks at the bonding energy of 711.05 eV and 724.84 eV are assigned to FeOOH, and the peak at the bonding energy of 718.55 eV is assigned to Fe(OH) 3 52 .Based on Fig. 8e and f, the Al2p spectra of the coagulation precipitate are deconvoluted into two peaks at the bonding energies of 73.89 eV and 74.64 eV, which are assigned to tetrahedrally coordinated Al (Al IV ) and octahedrally coordinated Al (Al VI ), respectively.The Al IV has a lower bonding energy than the Al VI53 .During the hydrolysis process of Al 3+ , the metastable [AlO 4 Al 12 (OH) 24 (H 2 O) 12 ] 7+ (Al 13 ) is a significant intermediate, with the structure of a central Al IV is surrounded by 12 peripheral Al VI54 .Therefore, the Al VI /Al IV ratio of the Al(OH) 3 is 12 theoretically.However, in this study, the Al VI /Al IV ratio of the coagulation precipitate formed by AlCl 3 and Al 2 (SO 4 ) 3 are 7.52 and 11.22, respectively.It indicates less content of voluminous Al 13 , and Al(OH) 3 is the major Al species in the precipitate.In addition, the presence of P may impede the formation of Al VI55 .Compared to the coagulation precipitate formed by the single AlCl 3 and Al 2 (SO 4 ) 3 , the Al VI content decreased by 25.44% and 26.02% in the precipitate of FeCl 3 -AlCl 3 and FeSO 4 -Al 2 (SO 4 ) 3 composite coagulants, respectively.However, the content of the Al IV remains constant.It indicates that the Fe isomorphous substitutes for the peripheral Al VI and is involved in the coordination process of the Al 13 with the P. The substitution process is illustrated in Fig. 9.

Conclusion
In this study, we surveyed the effect of Fe-Al composite coagulants on the removal of different P forms and discussed the mechanism of Fe-Al composite coagulants to enhance the removal rate of TP.The main conclusions are as follows: (1) Compared with single coagulants, the TP removal rate of Fe-Al composite coagulants significantly improved.The coagulant combines with dissolved orthophosphate to form suspended orthophosphate and sedimentation.(2) Fe-Al composite coagulants have a higher optimal TP removal rate than single coagulants when 6 < pH < 9.
Polynuclear hydroxyl complexes are the primary hydrolysis product of Fe and Al salts coagulants at pH 6.
The adsorption-bridging effect of the metal hydroxides hydrolyzed by Fe 3+ and Al 3+ plays a significant role in P removal.(3) FeSO 4 reacts readily with P to form non-settling Fe 3 (PO 4 ) 2 flocs, and Al 2 (SO 4 ) 3 can promote the sedimentation of the small-volume flocs in FeSO 4 -Al 2 (SO 4 ) 3 composite coagulant.Fe isomorphous substitutes for the peripheral Al VI and is involved in the coordination process of the Al 13 with the P.
In conclusion, Fe-Al composite coagulants are efficient and feasible processes to remove low P concentrations in urban sewage.

Figure 1 .
Figure 1.Steps of analysis of phosphate forms.

Figure 3 .
Figure 3.Effect of coagulants on the phosphorus forms in biological effluent.

Table 1 .
Water quality of self-cleaning activated bio-filter.