Theoretical and experimental investigation of phosphate removal from seawater by multi-stage coagulation

Phosphate removal from seawater is important for biofouling control on RO membrane because phosphorous is one of nutrients for microbial growth. This paper is based on the hypothesis that multi-stage coagulation results in better phosphate removal. Therefore, comparison of phosphate removal of one-step and three-step dose coagulation from the aspect of both the theoretical calculation and experimental results is investigated in this paper. The result of theoretical calculation based on Freundlich equation shows that ﬁ nal phosphate concentration of three-step dose i.e. 0.43 μ gP/L, is 10 times lower than that with one-step dose i.e. 4.47 μ gP/L. The experimental result shows that for the three-step dose, ﬁ nal phosphate concentration is 1.0 μ gP/L, which is lower than one-step dose (i.e 4.0 μ gP/L), but not as low as the theoretical calculated value (0.43 μ gP/L). This discrepancy between theoretical calculation and experimental result may be the impact of equilibrium phosphate concentration, different initial Fe:P molar and NOM competition between one-step dose and three-step dose coagulation. Although this discrepancy exists, experimental results still showed that multi-stage coagulation presented better phosphate removal in seawater to concentration levels that are lower than conventional coagulation. In other words, the problem of the high coagulant dosage in the pretreatment process while removing phosphate from seawater may be solved by application of multi-stage coagulation instead of conventional coagulation.


INTRODUCTION
Biofouling is the formation of biofilms on membrane surface by deposition and growth of micro-organisms from raw water. Microbial growth needs nutrients such as (carbon(C), nitrogen(N), and phosphorus(P)). Therefore, limiting these nutritional compositions in the feed solution has been introduced in attempts to prevent or restrict microbial growth, instead of killing microorganisms using chemical biocide (Kooij et  Jacobson (Jacobson et al. ) suggested that phosphate limitation in RO systems may be an option in controlling biofouling which was only linked to decreased cleaning frequency (Table 1). It shows that chemical cleaning of the RO units was much less frequent in the nonsummer months due to the lower phosphate concentration in these months than in summer both in Plant A and Plant B. And Vrouwenvelder (Vrouwenvelder et al. ) experimentally showed low phosphate concentration was able to control biofouling by measuring biomass concentration, which is summarized in Table 1 that when 'limiting P'(no P dose) in RO feed, biomass concentration of 2.5 × 10 2 pgATP/cm 2 is much lower than that of 6 × 10 4 pgATP/cm 2 with 20 μgP/L dose in RO feed water. Later, Chang-Min Kim investigated effects of phosphate limitation on performance of FO. By measuring biomass parameter of total cell number decrease, it was discovered that the phosphate limitation in FO feed water greatly reduced microbial growth (Table 1). Jacobson and Vrouwenvelder proved that phosphate limitation can control biofouling. However, the problem is that high chemical dosage in pretreatment process for the removal of phosphate causes high operational costs. Moreover, in the existing researches, only surface water and synthetic water as raw water are studied for phosphate limitation and no relevant research is currently available concerning the seawater.
Phosphorus removal started in the 1950s in response to the growing problem of eutrophication and the level of phosphorus needed to be reduced before discharging to surface water. Chemical precipitation was the initial method applied to remove phosphorus (Jiang & Graham ). Furthermore, biological phosphorus removal (Oehmen et al.  Biomass parameter: total cell number 3.0 × 10 6 (±5.5 × 10 5 ) cells/cm 2 been established gradually. Chemical precipitation and biological removal are generally suitable for removing phosphate with higher concentration and usually the con- Although three-parameter equations (Redlich-Peterson and Langmuie-Freundlich) often provide a better fit of the isotherm data, two-parameter equations (Freundlich, Langmuir, Temkin) are more widely used in the practice due to the convenience of evaluating two parameters than three parameters (Zeng et al. ).
In the case of phosphate removal by coagulation, two coagulants including ferric chloride (FeCl 3 .6H 2 O) and aluminum sulfate (Al 2 (SO 4 ) 3 .14H 2 O) are commonly used.

Raw water
Raw water was collected from the intake of a seawater-  Figure 1 shows the main steps of the whole experimental process.
One-step dosing and three-step dosing coagulation • The filter holder and syringe were soaked in (1 þ 4) HCl or 10% H 2 SO 4 for at least one hour and then rinsed with ultra pure water at least three times to take out any traces of phosphate.
• The 0.45 um filter was set into the filter holder.
• A certain amount of flocculated solution was drawn into the syringe, screw the filter set into the syringe then push it to filterate.

RESULTS AND DISCUSSION
Phosphate removal capacity based on adsorption isotherm The experimental results of phosphate removal by coagulation are presented in Table 3. It can be observed that the Where q is the equilibrium adsorption capacity (μgP/g); Ce is the equilibrium concentration of phosphate in aqueous phase (μgP/L); and the rest of the parameters are different isotherm constants which are determined by regression of experimental data. The parameters and correlation coefficients for each model at initial Fe;P molar ration of 5:1to 16:1 is shown in Table 4. It can be seen that correlation coefficient of Freundich and Langmuir isotherm is higher than Termkin isotherms which indicates data fitting was better with Freundich and Langmuir equation. Therefore, these two models are equally valid to fit experimental data, and Freundich equation is used for theoretical calculation in this paper.
Therefore, Freundlich equation is q ¼ 28,425Ce 0.963 for phosphate removal capacity for initial Fe:P molar ratio of 5:1. For initial Fe:P molar ratio of 0.5:1 to 5:1, the Freundlich equation is q ¼ 74,023 × 0.2794 (see Figure 2) which obtained by experimental data from Table 3.  Calculation of final phosphate concentration for coagulation with one-step dose Coagulation was performed at 1mgFe 3þ /L, with a volume of 1 L and m ¼ 1,000 μgFe 3þ /L. Assuming initial phosphate concentration C o ¼ 117 μgP/L, results in initial Fe:P molar ratio: i.e. 0.43 μgP/L, is 10 times lower than that with one-step dose i.e. 4.47 μgP/L. Therefore, theoretically phosphate removal of three-step dose is much better than that with one-step dose, at same total iron concentration.
Experimental results of final phosphate concentration of one-step and three-step dose coagulation The experimental results of final phosphate concentration for one-step and three-step dose are shown in Figure 3. It can be seen that experimental results of final phosphate concentration of one-step dose and three-step dose was different from calculation.
(1) For one-step dose, it can be observed that final phosphate concentration is 4.0 μgP/L which was almost the same as the results from calculation of 4.47 μgP/L.   Although the experimental value of final phosphate concentration by three-step coagulation is higher than the theoretical value, compared with the one-step coagulation under the same dosage of 1 mg Fe3 þ /L, phosphate removal by three-step coagulation is still better. In other words, if the same phosphate removal efficiency is expected to be achieved, the dosage of coagulant by three-step dose is lower than that by one-step dose, which will play an important role in saving the actual operating cost of pretreatment of seawater RO system.

ACKNOWLEDGEMENT
This research was supported by UNESCO-IHE and seawater-UF/RO pilot desalination plant in Zeeland Province, the Netherlands. The authors wish to acknowledge the assistance of colleagues in the laboratory of UNESCO-IHE and seawater-UF/RO pilot desalination plant in Zeeland Province, the Netherlands.

DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.