Investigation on the adsorption of phosphorus in all fractions from sediment by modified maifanite

Sediment phosphorus (P) removal is crucial for the control of eutrophication, and the in-situ adsorption is an essential technique. In this study, modified maifanite (MMF) prepared by acidification, alkalization, salinization, calcination and combined modifications, respectively, were first applied to treat sediment P. The morphology and microstructure of MMF samples were characterized by X-ray fluorescence (XRF), Fourier transform infrared (FTIR), X-ray diffraction (XRD), scanning electron microscope (SEM) and Brunauer-Emmett-Teller (BET). Various adsorption parameters were tested, such as dosage of maifanite, time, operation pH and temperature. The adsorption mechanisms were also investigated and discussed. Results showed that CMMF-H2.5-400 (2.5 mol/L H2SO4 and calcined at 400 °C) exhibited the highest P adsorption capacity. Thus, it was selected as the in-situ adsorbent material to control the internal P loading. Under the optimal conditions of dynamic experiments, the adsorption rates of TP, IP, OP, Fe/Al-P and Ca-P by CMMF-H2.5-400 were 37.22%, 44.41%, 25.54%, 26.09% and 60.34%, respectively. The adsorption mechanisms analysis revealed that the adsorption of P onto CMMF-H2.5-400 mainly by ligand exchange. Results of this work indicated that the modification treatment could improve the adsorption capacity of maifanite, and CMMF-H2.5-400 could be further applied to eutrophication treatment.

Characterization of RMF and MMF. XRF. The XRF results (Table 1) showed that there were significant differences in the chemical compositions of RMF and various MMF samples, which were mainly composed of SiO 2 , Al 2 O 3 , Na 2 O, CaO and Fe 2 O 3 . The modification treatment resulted in an obvious increase in the content of SiO 2 and a significant decrease in MgO and P 2 O 5 of MMF samples. Typically, SiO 2 and Al 2 O 3 were the major compositions of these maifanite samples. Compared to the RMF, the content of Na 2 O in MMF-OH3.0 was increased. It indicated Na ion from 3.0 mol/L NaOH was intercalated in the layers of the RMF after the NaOH modification. This was also confirmed by XRD analyses.
FTIR. FTIR is a supporting tool to distinguish characteristic functional groups 38,39 . In the spectrum of RMF and MMF samples (Fig. 2), a broad intense peak near 3430 cm −1 was due to the stretching vibrations of O-H and N-H groups 39 . The broad peak at 3430 cm −1 in RMF was shifted to 3423 cm −1 , 3435 cm −1 , 3413 cm −1 , 3435 cm −1 and 3432 cm −1 in MMF-H2.5, MMF-OH3.0, MMF-La5.0, MMF-400, and CMMF-H2.5-400, respectively, suggesting that interfacial interaction between chemical reagent of the modification treatments and maifanite surface, and this was consistent with XRD results (Fig. 3). In the FTIR spectrum of the various maifanite samples, a prominent band at 1035 cm −1 could be assigned to a fundamental frequency of the PO 4 3− stretching 40 . The peak at 777 cm −1 was assigned to Mg-Fe-OH 41 , and it remained unchanged after the modification treatments. The peak at 589 cm −1 , probably assigned to the Fe-O bands 42 , was obviously unchanged, although with a low intensity, in the MMF samples.
XRD. XRD analyses of various maifanite samples were carried out to discuss the crystal structure and identity. XRD patterns (Fig. 3) revealed that the RMF and various MMF samples mainly contained Quartz (SiO 2 ) and Margarite (CaAl 2 (Si 2 Al 2 )O 10 (OH) 2 ). Additionally, it is obvious that the d 001 value increased after the modification treatment (Table 2), and this result was agreement with the previous results 43 . Comparing the intensity of obvious    diffraction peaks of RMF with MMF samples, there was no significant change. No obvious peaks of metal oxide were observed in the maifanite diffractograms, which indicated that the maifanite structures remained intact after the modification treatment and the metal oxide was well dispersed on the surface of maifanite 44 . XRD patterns ( Fig. 3(a,e,f)) of maifanite before and after calcining at 400 °C indicated that their main mineral diffraction peak position remained unchanged. It indicated that 400 °C was the appropriate temperature on calcination. The result was in accordance with that of Yang et al. (2011), who reported that the main mineral diffraction peak position of maifanite did not change after calcining at 500 °C 45 . On the other hand, some peaks were disappeared and weakened a little after the modification treatment, which might be caused by the decrease of impurity after the modification treatment. This was consistent with BET and FTIR results.

SEM.
To compare the morphology of RMF with MMF samples, the SEM analysis were performed. It was seen that the morphology of MMF samples changed obviously due to the modification treatment (Fig. 4). The surface of the RMF was smooth, and few pores were observed. Comparing with the RMF, the surface of the MMF samples was rougher and more pores were obtained, which indicated that a porous structure with irregularly defined channel was formed. Additionally, there were more flakes appeared in MMF samples. The SEM micrographs of MMF-H2.5 (Fig. 4b), MMF-OH3.0 (Fig. 4c), MMF-La5.0 (Fig. 4d) and CMMF-H2.5-400 ( Fig. 4f) showed that the activation of MMF samples resulted in smaller grain sizes caused by the dispersive effect of H + , Na + , La 3+ , respectively, on the MMF-H2.5, MMF-OH3.0, MMF-La5.0, and CMMF-H2.5-400 structures [46][47][48] . Calcination led to microporous, followed by the removal of surface water, bound water and water of hydration of CMMF-400 ( Fig. 4e) and CMMF-H2.5-400 ( Fig. 4f) 49,50 . These changes could improve the reactivity of maifanite and impart a higher capacity of adsorption.
Surface analysis. Generally, the BET equation has been applied to measure and compare the specific surface areas of a variety of porous materials 51 . The BET surface area was regarded as an important factor in determining the pore properties of the adsorbent materials 42 . The pH PZC , CEC, the maifanite basal plane diffractions d 001 , the specific surface (S BET ), the total pore volume (V t ), the volume of micropores (V mikro ), the external surface (S external ) and the average pore size (D p ) before and after the modification were given in Table 2. The results suggested that the modification caused the disintegration of maifanite structural. Based on these, it led to a significant increase in the S BET and an obvious decrease in the D p . It could provide more active sites for the adsorption reaction, and made the surface more available for the sediment P. These findings supported the SEM results. The pH PZC of    52 revealed that pH PZC had a significant impact in As(V) adsorption by natural zeolites. That's because the adsorption of multivalent cation occurred effectively at a pH below pH PZC . Compared to the RMF, the pH PZC values of MMF-H2.5, MMF-OH3.0, MMF-La5.0, CMMF-400 and CMMF-H2.5-400 were increased in different degrees, respectively. On the other hand, The CEC of MMF samples was found to be higher than RMF, and the CEC of CMMF-H2.5-400 was the maximum. The above information revealed that modification treatments could improve the sediment P adsorption capacity of maifanite.

P adsorption on maifanite by dynamic experiments.
Effect of maifanite dosage. The amount of the sorbent dosage plays a vital role in sediment P adsorption. Experiments were carried out with various dosages of RMF and CMMF-H2.5-400, respectively at 20 ± 2 °C, pH 7.0 ± 0.2 and shaken at 200 rpm for 12 h to investigate the effects of maifanite dosage. Figure 5a,b indicated that the adsorption effects on sediment P by CMMF-H2.5-400 were better than RMF, with the tendency of increased at first then decreased and stabilized with an increase of dosage (Fig. 5b). Compared to the RMF, CMMF-H2.5-400 showed more considerable micropore adsorption and higher ion exchange capacity due to the more microporous microstructure and superior charge number 31 . Figure 5a indicated that the adsorption quantity of P from the sediment increased with an increase of RMF dosages. The quantity of adsorption sites becomes more availability with an increase of sorbent dosage 53,54 . From Fig. 5b, the adsorption quantity of P first increased and then decreased with the increasing of CMMF-H2.5-400 dosages. Typically, the adsorption of P didn't have an immense change when CMMF-H2.5-400 The corresponding adsorption rates were 21.80%, 28.35%, 11.18%, 37.48% and 20.12%, respectively. Based on the above results, optimal dosages were fixed as 2 g and 12 g, respectively for CMMF-H2.5-400 and RMF and pursued in further investigations.

Effect of stirring time.
To investigate the effects of stirring time on P adsorption by maifanite, the experiments were carried out with 2 g CMMF-H2.5-400 or 12 g RMF at 20 ± 2 °C, pH 7.0 ± 0.2 and shaken at 200 rpm. Figure 5c,d presented the results. After the initial rapid reaction, the adsorption quantities of sediment P by CMMF-H2.5-400 or RMF decreased gradually. The P adsorption amounts by CMMF-H2.5-400 and RMF respectively increased from 2 h to 4 h and then decreased after 4 h. The adsorption amount of P from sediment reached the maximum at 4 h by CMMF-H2.5-400 and RMF, respectively. Thus, 4 h was chosen as the optimal stirring time to investigate the effects of pH and operation temperature on the P adsorption by CMMF-H2.5-400 and RMF, respectively. Especially, the adsorption quantities of IP and Fe/Al-P by CMMF-H2.5-400 and RMF increased markedly (P < 0.05), respectively.
Influence of pH. The operation pH is a key factor that controls the P adsorption from sediments 56 . To investigate the influence of extremely high or low pH value on adsorption of sediment P, the experiments were carried out at different pH conditions ranged from 2.0 to 12.0. Figure 5e,f confirmed that the operation pH influenced the adsorption capacity: the adsorption of TP by CMMF-H2.5-400 reached the maximum at the optimal pH (pH opt ) 2.0 and decreased with the increasing pH value. The adsorption of TP by RMF increased with the increasing pH (2.0-6.0), then decreased when pH exceeded 6.0 (P < 0.05). The adsorption amounts of TP, IP, OP, Fe/Al-P and Ca-P by CMMF-H2. 5  These results were agreement with the previous reports 31, 57 . The effects of P adsorption in various pH were due to a series of mechanisms, including chemical interaction, ligand exchange, electrostatic attraction/repulsion and coagulation/precipitation 58 .
Effect of temperature. Operation temperature is a significant factor influencing the P adsorption in sediment, which could remarkably improve p release 59,60 . Shallow lakes are usually isothermal, and the sediment is susceptible to temperature variations 61 . Thus, the temperature may have a larger influence on adsorption P in sediment from shallow lakes. The results were depicted in Fig. 5g,h. As temperatures increased from 5 °C to 20 °C, the P adsorption efficiencies of CMMF-H2.5-400 and RMF, respectively, increased remarkably (P < 0.05). High temperatures facilitated the P adsorption confirming that the sediment P adsorption on the maifanite samples was an endothermic reaction. This result was agreement with the previous reports 61, 62

Static adsorption experiments.
To simulate the P adsorption under in-situ treatment, static experiments were carried out with time from 0 d to 30 d. Figure 6 depicted the adsorption of sediment P by RMF and CMMF-H2.5-400, respectively, under static conditions. The P adsorption on CMMF-H2.5-400 included quick, slow and dynamic balance adsorption steps (P < 0.05). Firstly, the quick adsorption step mainly occurred from 0 d to 10 d, and then followed by a slower second step (10 d-18 d). Furthermore, there was no obvious difference in adsorption quantities of P in all fractions from the sediment after 18 d (Fig. 6b). The adsorption quantity of sediment P by CMMF-H2.5-400 reached the peak value at 18 d. On the other side, the amounts of P adsorption on RMF increased rapidly within 20 d and then decreased slowly. Finally, the adsorption amounts of P barely achieved to a real equilibrium within the selected time (P < 0.05). The adsorption quantities of sediment P by RMF reached the maximum at 20 d (Fig. 6a) Figure 6 also showed that the P adsorption capacity of CMMF-H2.5-400 was higher than RMF. Furthermore, it indicated that the modification treatment could improve the P adsorption capacity of maifanite, and this result was in agreement with our previous study, which reported that the P adsorption capacity of modified bentonite granules (MBGs) was higher than raw bentonite granules (RBGs) 31 .

Characterization of maifanite after adsorption.
Ion-exchange occurred and altered the elements content of CMMF-H2.5-400 and RMF, respectively, during the adsorption process (  Figure 7(A) depicted the FTIR spectra of CMMF-H2.5-400 and RMF after adsorption. Comparing the spectrum of CMMF-H2.5-400 after adsorption with that of CMMF-H2.5-400 before adsorption (Fig. 2b), a new adsorption peak emerged at 1393 cm −1 , which referred to the O-H bending vibration with Fe (III), Al (III) species present on the surface of CMMF-H2.5-400 after adsorption 51 . It indicated that Fe (III), Al (III) species were intercalated into interlayers of CMMF-H2.5-400 by adsorption process. Additionally, the peaks of CMMF-H2.5-400 at 3432 cm −1 , 1634 cm −1 and 1093 cm −1 were shifted to 3413 cm −1 , 1627 cm −1 and 1085 cm −1 , respectively. For the RMF, a new adsorption peak emerged at 3568 cm −1 , which were attributed to  Table 3. The main chemical compositions of (a) RMF and (b) CMMF-H2.5-400 after adsorption (wt.%). broad OH-stretching in RMF after adsorption. On the other hand, the peaks of RMF at 3430 cm −1 and 776 cm −1 were shifted to 3420 cm −1 and 778 cm −1 , respectively. Comparing the Fig. 7(A) with Fig. 2, it could be seen that the peaks observed on the maifanite samples were obviously unchanged after adsorption. XRD patterns of RMF and CMMF-H2.5-400 after adsorption were shown in Fig. 7(B). Comparing the intensity of obvious diffraction peaks of maifanite before and after adsorption, a reasonable shifting of peaks was observed from 20 to 70° 2 theta indicating that the adsorption of sediment P on maifanite granules changed the peaks of chemical composition intensities on the maifanite granules. According to the XRD spectra, it was thought that there was a ligand exchange between sediment P in all fractions and maifanite granules. The results of XRD in Figs 3 and 7(B) clearly reveal the presence of phosphate salts on the structure of adsorbent material after the adsorption process 63 . The d 001 values of CMMF-H2.5-400 and RMF were 3.73 nm and 3.07 nm, respectively, before adsorption (Table 2), and then increased to 3.91 nm and 3.13 nm after adsorption, respectively (Table 4). SEM images were used to examine the surface morphology of RMF and CMMF-H2.5-400 before and after sediment P adsorption, respectively. Figure 7(C) exhibited the SEM images of RMF and CMMF-H2.5-400 after adsorption. Comparing the SEM images of maifanite samples before and after adsorption, some rough exterior and fresh cavities emerged after adsorption. The micrograph obtained after adsorption indicated that the flakes of the phosphate were observed on the adsorbent surface ( Fig. 7(C)) 64,65 . Furthermore, the pores of the particles of the adsorbent have been covered with adsorbate 63 . After the adsorption process, the SEM micrographs of CMMF-H2.5-400 and RMF revealed the formation of metal-hydroxyl-phosphate ligand (Yang et al., 2009). Table 4 confirmed that adsorption caused the disintegration of maifanite structural and led to a significant increase in the S BET . These findings supported the SEM results.

Adsorption mechanisms. Phosphate is adsorbed onto clay minerals via electrostatic, ligand exchange, and
Lewis acid-base interaction [66][67][68] . Surface hydroxyl groups are protonated in the ligand exchange process at low pH. That's because, compared to the hydroxyl groups, -OH 2 + is easier to displace from the metal binding sites 43 . Therefore, it was likely that the adsorption of phosphate onto CMMF-H2.5-400 mainly using ligand exchange. This result was in agreement with our previous study, which reported that MBG adsorbed phosphorus mainly by anionic coordination exchange adsorption 31 . On the CMMF-H2.5-400 samples, phosphate replaced the hydroxyl groups, which were then released into the solution. The adsorption of phosphate could be speculated to take place as follows: the phosphate in the sediment was first transferred to the sites on the adsorbent; then, chemical complexation/ion exchange occurred at the active sites 69,70 .
Comparing the SEM images of the CMMF-H2.5-400 before and after adsorption process, we observed an aggregated morphology and some large flakes in CMMF-H2.5-400 samples after adsorption process. Additionally, the size of the intra-particle voids was decreased due to a stacking structure formed by some thin lamellas (Table 4). These results revealed that phosphate did adsorb onto the CMMF-H2.5-400 surface and it could be combined in the form of oxygen bridge 43,63 . Meanwhile, the hydroxyl and hydration base could be swop out 31 .

Conclusions
In this study, MMF samples were prepared by various modification methods and applied to adsorb sediment P in all fractions for the first time. The results revealed that the modification treatment could improve the adsorption capacity of maifanite and CMMF-H2.5-400 was selected as the optimal in-suit adsorption material. The results of adsorption experiments showed that the dosage of maifanite, adsorption time, operation pH and operation temperature were the main factors influencing the adsorption performance on sediment P of CMMF-H2.5-400. Under the optimal conditions of dynamic experiments, the adsorption rates of TP, IP, OP, Fe/Al-P and Ca-P by CMMF-H2.5-400 were 37.22%, 44.41%, 25.54%, 26.09% and 60.34%, respectively. The adsorption mechanisms analyses revealed that the adsorption of phosphate onto CMMF-H2.5-400 mainly by ligand exchange. The above information indicated that CMMF-H2.5-400 exhibited a promising adsorption capacity on sediment P and could be further applied to reduce internal P loading in the eutrophic lakes. Furthermore, a combined technology of MMF and other ecological methods would be a significative orientation to treat sediment P.

Methods
Study site and sampling. West Lake (30°14′45′′N, 120°08′30′′E) is located on the western side of Hangzhou City, China. West Lake, which has been listed in the World Heritage Site in 2011, is a typical eutrophic lake with an area of 6.5 km 2 and a mean depth of 2.27 m 30 . The sediment in West Lake is unstable, at the risk of releasing P to West lake 71 . Currently, developing and applying an effective in-situ technology for sediment P control was urgent.
The sampling site is located in a severe eutrophic region (30°23′16′′N, 120°13′18′′E) in Xiaonan Lake, one sub lake of West Lake. The surface lake sediments, at a depth of 0-10 cm, were collected by a Peterson grab sampler (model HNM1-2) on May 14th, 2017. The sediment samples were then stored in plastic bags. After transportation to the laboratory, the sediment samples were air-dried. The content of TP, TN, NH 4 + -N and NO 2 -N of overlying water above sediments in the lake was 0.06 ± 0.01 mg/L, 3.57 ± 0.03 mg/L, 0.15 ± 0.02 mg/L and 0.12 ± 0.01 mg/L, respectively. The pH values of the interstitial water and overlying water were 8.07 and 7.9, respectively.   Analytic methods. The Standards and Measurements and Testing (SMT) protocol 72 were used for determining P fractions. The P fractions can be characterized as TP (total P), IP (inorganic P), OP (organic P), Fe/ Al-P (P bound to Al, Fe, and manganese (Mn) oxides extracted by NaOH) and Ca-P (P bound to calcium (Ca) extracted by HCl). The presence of metals in sediment can mediate the transport of P. Each P fraction concentration was measured directly using the ammonium molybdate spectrophotometric method with an UV-visible spectrophotometer at 700 nm (DR4000/U, HACH company, USA). The cation exchange capacity (CEC) of the sediment samples was analyzed using NH 4

Combined modifications
The optimal acidification modification, alkalization modification, salinization modification, and calcination modification methods were calcined at the optimal temperature, respectively. After that, the combined MMF samples were cooled to room temperature for further studies. of maifanite granules to adsorb the cations, was analyzed using the ammonium acetate method. The specific surface areas were calculated by nitrogen adsorption using the Brunauer-Emmett-Teller (BET) equation on an analyzer (ASAP 2020 M, America). The t-plot method was applied to gain the volume of micropores and the surface of mesopores together with the external surface. The total pore volume was derived from the nitrogen volume adsorbed at the relative pressure p/p 0 → 1 73 . The batch equilibrium techniques were applied to estimate the points of zero charge (pH pzc ) of maifanite samples 52 . All the chemicals and reagents used were analytical grade. All glassware and sample bottles were presoaked before use in diluted HCl solution for at least 12 h followed by washing with deionized water and drying in oven. Deionized water was used for preparing solutions. Data analysis. The amount of P adsorbed on the maifanite granules and the adsorbed efficiency (A) were calculated using the following equations: where q is the adsorption quantity of P per unit weight of maifanite samples (mg·kg −1 ), and q 0 and q e (mg·kg −1 ) are the initial and final P quantity, respectively. All treatments were conducted in triplicate. OriginPro 8.0 (OriginLab Corporation, Northampton, MA, USA) was used to plot various figures. All statistical analyses were estimated using SPSS 18.0 (SPSS software, IBM, USA). Analyses of the variance (ANOVA, one factor) were applied to test the significant differences between the dependent variables (the adsorption quantity of P) and independent variables (the corresponding adsorption parameter). The difference was considered statistically significant when the significance level was smaller than 0.05.