Application of titanium sulfate in a coagulation–ultrafiltration process: a comparison with aluminum sulfate and ferric sulfate

Abstract

This study focused on the floc characteristics and membrane fouling of Al₂(SO₄)₃ (AS), Fe₂(SO₄)₃ (FS) and Ti(SO₄)₂ (TS) in a coagulation–ultrafiltration (C–UF) process. Floc properties, including floc size, floc strength, recovery ability and structure, which would significantly affect membrane fouling, were analyzed. The effect of the solution pH on the floc properties and membrane fouling was also considered. The results showed that TS flocs with a compact structure, high strength and large size achieved the lowest extent of membrane fouling among these three coagulants. In addition, the floc properties and membrane fouling were also affected by the initial solution pH. TS formed the smallest and weakest flocs with an intermediate fractal dimension (D_f) value when the pH was 5, whereas TS formed medium-sized flocs with a compact structure at a pH of 7 and a pH of 9. Nevertheless, TS coagulation systems led to the least decrease in permeate flux under acidic, neutral and basic conditions in the ultrafiltration process, and the most obvious decline in permeate flux was caused by coagulation with FS. In addition, the results also indicated that the membrane fouling of the AS, FS and TS systems was the slightest at pH 7.

---

1. Introduction

Membrane filtration processes, especially ultrafiltration, have been widely used in water treatment owing to their good performance, small footprint and relatively low cost in the past few decades.¹ However, membrane fouling is one of the most serious problems that restrict the development of ultrafiltration.^1,2 Several pretreatment processes such as adsorption, coagulation, peroxidation, etc., have been used in combination with ultrafiltration. These technologies have been proved to be effective ways to reduce membrane fouling, especially coagulation pretreatment, which has been reported to be more effective than other pretreatment technologies in reducing membrane fouling in low-pressure membrane filtration.³ Key features of the filtration system such as the coagulant type may significantly affect the performance of a coagulation-membrane filtration process. Aluminum and iron salts are the most widely used metal coagulants in the treatment of water and wastewater owing to their high coagulation activity and low cost.⁴ However, there are still some drawbacks that limit the development of Al- and Fe-based coagulants. For example, it has been reported that residual turbidity due to aluminum in drinking water may lead to the possibility of some neurologic diseases such as Alzheimer's disease.⁵ The coagulation sludge of aluminum salts and iron salts is difficult to deal with. Recently, titanium salts have attracted attention owing to the reusability of their sludge and their coagulation performance as a new metal coagulant. Shon et al. found that the coagulation sludge of titanium tetrachloride could be reused to prepare TiO₂ that performed better than commercial P-25 TiO₂.⁶ Moreover, Zhao et al. and Wu et al. proved that titanium salts could achieve almost the same turbidity and removal efficiency of natural organic matter (NOM) compared with aluminum and iron salts. In addition, flocs formed by titanium salts were larger and denser than those formed by aluminum salts.^7,8 Floc properties such as their size, charge and fractal dimension significantly affected membrane permeability. Previous studies by Bagga et al. and Yu et al. suggested that resistance of the cake layer was one of the most significant mechanisms of the resisting force in flux decline.^9,10 Moreover, cake filtration was affected by physicochemical and structural properties of aggregation. In general, larger aggregates with a looser structure lead to less serious membrane fouling. Previous studies of C–UF processes also suggested that loose and porous flocs appeared to be favorable for membrane performance.^11,12 A previous study by Zhao et al. compared the different floc properties of aluminum, iron and titanium salts.⁷ They found that flocs in a coagulation system using titanium salts were found to be more compact with a larger value of fractal dimension.⁷ However, Zhao et al. only studied the coagulation performance and floc properties of titanium salts. There are still few studies that have considered the floc properties and membrane fouling of titanium coagulants. Moreover, the solution pH would also influence the floc properties and membrane fouling. However, there have been no reports on the effect of the solution pH on the floc properties and membrane fouling of titanium salts.

Therefore, the floc properties and membrane performance of Al₂(SO₄)₃, Fe₂(SO₄)₃ and Ti(SO₄)₂ in a coagulation–ultrafiltration process were compared in this study. Flocs formed by Al₂(SO₄)₃, Fe₂(SO₄)₃ and Ti(SO₄)₂ at different initial solution pH values were also investigated in terms of floc size, floc breakage/recovery factors and floc fractal dimension. In addition, membrane fouling at different solution pH values was also compared in this paper.

2. Materials and methods

2.1 Coagulants

In this research, HCl, NaOH, Al₂(SO₄)₃·18H₂O (AS), Fe₂(SO₄)₃ (FS) and Ti(SO₄)₂ (TS) were obtained from Sinopharm Chemical Reagent Co., Ltd., Shanghai, China. All reagents used were of analytical grade. Deionized water was used to prepare all solutions. The chemical structures of AS, FS and TS are shown in Table 1. In this study, an AS solution, an FS solution and a TS solution with an Al concentration, an Fe concentration and a Ti concentration of 2.0 g L⁻¹, 4.0 g L⁻¹ and 10.0 g L⁻¹, respectively, were prepared by dissolving the respective compounds in deionized water.

Table 1 Chemical structures of Al₂(SO₄)₃, Fe₂(SO₄)₃ and Ti(SO₄)₂

Name	Molar mass (g mol^−1)	2D structure
Al2(SO4)3	342.15
Fe2(SO4)3	399.88
Ti(SO4)2	239.99

---

2.2 Test water

Test water was prepared from humic acid (HA) and kaolin. A stock solution of HA (1 g L⁻¹) was prepared by dissolving 1.0 g HA with 0.4 g NaOH in 1 L deionized water under stirring for 1 h. Then, 5.0 g kaolin was dissolved in 1 L deionized water and then allowed to settle for 30 min in a 1 L measuring cylinder. Afterwards, the upper 500 mL of the solution was removed for use. The test water contained 10 mg L⁻¹ HA with an initial turbidity of 15 ± 0.5 NTU adjusted by kaolin. The properties of the test water were shown to be as follows: UV₂₅₄ absorbance = 0.310 ± 0.010, DOC = 3.35 ± 0.04 mg L⁻¹, pH = 8.15 ± 0.02. In addition, the initial pH of the test water was adjusted with 0.1 M HCl and 0.1 M NaOH solution.

2.3 Jar test

A jar test was conducted using a program-controlled jar test apparatus (ZR4-6, Zhongrun Water Industry Technology Development Co., Ltd., China) in six 1.5 L square beakers with 50 mm × 40 mm flat paddle impellers. Initially, 1 L test water was transferred to each beaker. Then, each beaker containing the test water was first stirred rapidly at 200 rpm for 1.5 min, followed by 15 min of slow stirring at 40 rpm, and finally allowed to settle for 20 min. A predetermined amount of coagulant was added at the beginning of rapid stirring. After settling, water samples of about 200 mL were collected from 2.0 cm below the solution surface for subsequent measurements. An unfiltered water sample was used for measurements of turbidity and zeta potential using a 2100P turbidimeter (Hach, USA) and a Zetasizer 3000HSa (Malvern Instruments, UK), respectively. The collected samples were filtered through 0.45 μm glass filter paper for measurements of UV₂₅₄ absorbance (Precision Scientific Instrument Co., Ltd., Shanghai, China) and DOC (TOC-VCPH, Shimadzu, Japan).

2.4 On-line monitoring of floc formation, breakage and regrowth

Experiments on floc formation, breakage and regrowth were programmed similarly to the jar test mentioned above. However, after slow stirring for 15 min, the suspension was exposed to a high shear rate (200 rpm) for 5 min, and then another 15 min of slow stirring at 40 rpm was reapplied for floc regrowth. The dynamic floc sizes during the whole process were measured by a Mastersizer 2000 (Malvern, UK). The suspension was monitored by drawing water through the test cell of the Mastersizer 2000 and back into the jar using a peristaltic pump (LEAD-1, Longer Precision Pump, China) on a return tube with an internal diameter of 5 mm at a flow rate of 2.0 L h⁻¹. The inflow and outflow tubes were placed opposite one another at a depth just above the impeller in the holding ports. Size distribution measurements were taken every 0.5 min and logged on a computer.

2.4.1 Floc size. In this study, the median volumetric diameter (d₅₀) was used to denote the floc size. In addition, the floc size distribution was also investigated using a Mastersizer 2000.

2.4.2 Floc breakage factor and recovery factor. The floc breakage and recovery factors were used to compare the relative breakage and regrowth of flocs in this study.¹³ The breakage factor indicates the ability of flocs to resist rupture in a certain velocity gradient. Larger values of the breakage factor indicate that flocs are stronger than those with smaller values. Likewise, the regrowth capacity of flocs was evaluated in terms of their recovery factors. Flocs with larger recovery factors have greater regrowth ability after breakage.

The factors were calculated as follows:¹³

where d₁, d₂ and d₃ are the sizes of flocs in the steady phase before breakage, after breakage and after the regrowth phase, respectively.

2.4.3 Floc dimension. The floc fractal dimension can be measured by the method of laser light scattering. During light scattering, a light beam was scattered by the particles in a water sample, and the scattering of light by the particles was correlated with their sizes and at a constant angle regardless of which part was hit. The smaller the particles were, the larger were the angles of the scattered light. The Malvern Mastersizer, which has an array of photosensitive detectors at different angles between 0.01° and 40.6°, could detect the light scattered by the samples.¹⁴ The light scattering technique involves the measurement of the light intensity I as a function of the scatter vector Q. Q is defined as the difference between the incident and scattered wave vectors of the radiation beam in the medium, which is given by eqn (3):where n is the refractive index of the suspending medium, θ is the scattering angle, and λ is the wavelength of the radiation in a vacuum (m).

Flocs are mass fractal objects, which can be described by the relationship between their mass M, a characteristic measure of their size L, and their mass fractal dimension D_f. For independently scattering aggregates, I is related to Q and the fractal dimension D_f is expressed by eqn (4):

I ∝ Q^−D_f (4)

where D_f is the mass fractal dimension, which can be determined from the slope of a plot of I as a function of Q on a log–log scale.

2.5 Coagulation–ultrafiltration procedure

Ultrafiltration experiments were conducted using a dead-end batch unit with a stirring capacity of 300 mL. There was an opening designed for the addition of pretreated samples at the top of the cell. After coagulation, the effluent without sedimentation was gently transferred from the coagulation tank to the dead-end filtration unit and filtered through an ultrafiltration membrane. Slow agitation was applied to create a uniform suspension. A constant pressure of 150 kPa was provided by nitrogen. The mass of the permeate was measured by an electronic balance (MSU5201S-000-D0; Sartorius AG, Germany), which was connected to a computer with a data acquisition system. In addition, the cumulative mass was recorded every 10 s. The UF membrane (Mosu, China) used in this study was a polyethersulfone (PES) flat sheet membrane with a molecular weight cutoff (MWCO) of 100 kDa. The effective area of the membrane was 50.24 cm². A fresh piece of membrane was used in each of these experiments. In addition, the membrane was soaked in deionized water for at least 24 h before use. A modified fouling index (MFI) was designed to measure the fouling potential of the microfiltration membrane and ultrafiltration membrane, and details can be found in previous studies:^15,16where V is the filtrate volume, t is the filtration time, ΔP is the pressure applied across the membrane, R_m is the resistance of the membrane, A is the available area of the membrane, η is the viscosity of water, α is the specific resistance of the cake deposited and C_b is the concentration of particles in the feed water. The MFI value is determined from the gradient of the linear portion of a plot of t/V versus V. A higher MFI value means more serious membrane fouling.¹⁷

3. Results and discussion

3.1 Pretreatment and zeta potential

Initially, a series of jar tests were conducted to determine the optimum dosage of coagulant by treating HA–kaolin test water without adjustment of the pH. The residual turbidity and UV₂₅₄ and DOC removal efficiencies were used to determine the proper dosages of different coagulants, as shown in Table 2. The zeta potentials at the optimal dosages were also measured and are shown in Table 2. The optimal amounts of AS, FS and TS were 10 mg L⁻¹, 26 mg L⁻¹ and 32 mg L⁻¹, respectively. In addition, the coagulation performance, including the residual turbidity and UV₂₅₄ and DOC removal efficiencies, under conditions of different solution pH (5–9) with the chosen optimal dosages of AS, FS and TS is shown in Fig. 1(a) and (b). The zeta potentials of coagulation systems using AS, FS and TS at different pH values were also measured, as shown in Fig. 1(c).

Table 2 Coagulation performance of AS, FS and TS

Index	AS	FS	TS
Residual turbidity (NTU)	1.25	1.42	1.03
UV254 removal efficiency (%)	87.5	88.5	89.2
DOC removal efficiency (%)	49.2	51.3	52.3
Zeta potential (mV)	1.3	−7.7	−13.3
Effluent pH	6.89	6.23	5.18

---

Fig. 1 Coagulation performance of AS, FS and TS: (a) effect of solution pH on effluent pH and residual turbidity; (b) effect of solution pH on UV₂₅₄ removal and DOC removal; (c) effect of solution pH on zeta potential.

The UV₂₅₄ and DOC removal efficiencies gradually increased as the dosage increased, and the minimum dosages of AS, FS and TS that achieved a HA removal efficiency of above 85% were 10 mg L⁻¹, 26 mg L⁻¹ and 32 mg L⁻¹, respectively. In addition, from a pH of 5 to a pH of 9, the HA removal efficiency increased first and then decreased. The maximum HA removal efficiencies of AS, FS and TS were achieved at pH values of 6, 7 and 7, respectively. In general, the results showed that the zeta potential increased as the coagulant dosage increased. For the AS coagulation system, HA–kaolin particles were neutralized by AS and became positively charged when the dosage was above 8 mg L⁻¹. However, for the FS and TS coagulation systems, the zeta potential gradually approached zero but still remained negative when a high HA removal efficiency was achieved. In addition, the zeta potential of FS–HA–kaolin flocs increased faster than that of TS–HA–kaolin flocs. The zeta potentials at the optimal dosages of AS, FS and TS were 1.3 mV, −7.7 mV and −13.3 mV, respectively. However, the UV₂₅₄ removal efficiencies achieved by AS, FS and TS were almost the same, which revealed that the coagulation mechanisms of these three coagulants were different. Charge neutralization and sweep flocculation may be dominant in the AS coagulation process, whereas precipitation charge neutralization and sweep flocculation may be dominant in the FS and TS coagulation systems. The zeta potentials of AS, FS and TS flocs were also significantly affected by the initial pH of raw water (Fig. 1(c)). The zeta potentials of AS, FS and TS flocs all decreased numerically as the pH increased. Under acidic conditions, NOM in water is mostly removed by the charge neutralization and complexation effects of positively charged metal species.¹⁸ The zeta potentials of AS, FS and TS flocs were all above zero in acidic conditions, which means that sweep flocculation gradually became dominant as the pH increased.

3.2 Floc properties in coagulation by AS, FS and TS

Floc growth, breakage and regrowth curves of AS, FS and TS at their respective optimal dosages in HA–kaolin water without pH buffering are shown in Fig. 2. In the cases of the AS and FS coagulation systems, the floc size increased rapidly in the first few minutes, followed by gentle growth, and finally reached a plateau. However, with the TS coagulation system, the floc size decreased slightly in the last several minutes of the growth phase. Similar results were found by other researchers who employed TiCl₄ in the treatment of HA–kaolin water.⁷ In addition, when a high shear rate (200 rpm) was applied after the floc growth stage, a sharp decrease in floc size was observed. Afterwards, when another slow stirring rate was reapplied, the floc size gradually increased again but could not reach the initial size observed before breakage.

Fig. 2 Floc growth, breakage and regrowth curves of AS, FS and TS flocs without pH buffering.

During the growth period, it was obvious that the floc growth rate was in the following order: TS > FS > AS. The maximum size of the TS floc was attained in the first 3 min; the FS floc reached its maximum size in 5 min; meanwhile, the AS floc grew slowly and reached a steady state after 7 min. It has been suggested that flocs formed by sweep flocculation grow faster than those formed by charge neutralization.¹⁹ This was consistent with the results for zeta potential, which indicated that sweep flocculation played a significant role in the TS and FS coagulation processes. The floc size distributions of AS, FS and TS flocs in a steady state are shown in Fig. 3. The floc sizes of FS (531 μm) and TS (504 μm) were similar, whereas the floc size of AS (235 μm) was obviously smaller than those of FS and TS. This was consistent with the results of a previous study.⁷ The floc size of TS was larger than that of FS at the initial stage, but then the TS floc size gradually decreased during the coagulation process at the formation stage. It has been suggested that flocs formed by charge neutralization are smaller than those formed by sweep flocculation. These results were consistent with previous results for zeta potential.²⁰

Fig. 3 Particle size distribution of AS, FS and TS flocs.

The floc breakage and recovery factors of AS, FS and TS are shown in Table 3. The floc breakage factors of AS and TS were significantly higher than that of FS. It can be inferred that the inter-particle strengths of AS and TS flocs were greater than that of FS flocs. Therefore, flocs formed by AS and TS were more resistant to a high shear rate than FS flocs in the subsequent treatment process. The floc recovery factors of AS, FS and TS varied in the following order: AS > FS > TS, which was in agreement with a previous study that found that flocs formed by titanium salts were difficult to recover after exposure to a high shear rate.⁷ Previous studies also suggested that flocs formed by a charge neutralization effect were easier to recover after breakage by a high shear rate.^21,22 The high recovery ability of AS flocs also proved that a charge neutralization effect played an important role in the AS humic acid–kaolin coagulation process.

Table 3 Floc breakage factors, recovery factors and fractal dimensions of AS, FS and TS flocs

Coagulant	AS	FS	TS
Bf (%)	41.35	29.65	46.21
Rf (%)	40.08	19.53	13.95
Df	2.01	2.09	2.29

---

The floc fractal dimension (D_f) is also a crucial indicator of floc properties that significantly affect the cake layer in membrane filtration processes. The floc fractal dimensions of AS, FS and TS before breakage are shown in Table 3. It was obvious that AS flocs with a low D_f value had a loose and open structure. Flocs formed by FS and TS were more compact with higher D_f values. Wang et al. suggested that flocs formed by charge neutralization coagulation were much looser than those formed by sweep coagulation.²³ Thus, the results for zeta potential and floc properties indicated that charge neutralization was not the only mechanism in coagulation by FS and TS, and that sweep coagulation might also have a significant effect.

3.3 Effect of solution pH on floc properties of AS, FS and TS flocs

The floc properties of AS, FS and TS flocs at various initial solution pH values were also analyzed using a Mastersizer 2000. The growth, breakage and regrowth curves of AS, FS and TS flocs at pH values of 5, 7 and 9 are shown in Fig. 4, which shows that the floc properties were dramatically affected by the solution pH.

Fig. 4 Floc growth, breakage and regrowth curves of AS, FS and TS flocs under various pH conditions: (a) pH 5; (b) pH 7; and (c) pH 9.

When the solution pH was 5, FS flocs attained the fastest growth rate, whereas flocs formed by AS and TS attained a similar growth rate. However, AS flocs were the largest in size among those of the three coagulants at a pH of 5 (Fig. 5). FS flocs grew fastest, but attained an intermediate floc size at a pH of 5. The TS floc size was the smallest at a pH of 5. Nevertheless, in neutral and alkaline conditions, the floc growth rates for these three coagulants were in the following order: TS > FS > AS. The TS floc size reached a maximum and then gradually decreased during slow stirring at pH values of 7 and 9. Therefore, the floc size of TS was slightly smaller than that of FS in the last few minutes of the slow stirring phase at pH values of 7 and 9. However, the floc size of AS was obviously smaller than those of FS and TS, which can also be observed in Fig. 5. The floc size distribution (PSD) curve for AS showed an apparent movement to the left compared with that for FS. As for the PSD curve for the TS floc, only a slight shift in the major peak was observed compared with the distance between the PSD curves for the AS floc and the FS floc.

Fig. 5 Floc size distribution of AS, FS and TS flocs under various pH conditions: (a) pH 5; (b) pH 7; and (c) pH 9.

After the floc growth stage, a high shear rate was applied, which resulted in a sudden decrease in floc size. After that, another slow stirring (40 rpm for 5 min) phase was applied to allow the broken floc to reaggregate. The floc breakage factors and recovery factors at various solution pH values were calculated using eqn (1) and (2), respectively. According to Fig. 6(a), the breakage factors at a pH of 5 were in the following order: AS > FS > TS. However, at pH values of 7 and 9, the breakage factor of the FS floc was smaller than those of the AS and TS flocs. As for the floc recovery factors, at the studied pH values, the floc recovery factors for these three coagulants were in the following order: AS > FS > TS. The recovery ability of TS flocs was the lowest, which agreed with a previous study that found that flocs formed by titanium salts were difficult to recover after exposure to a high shear rate.²⁴ In addition, it could be noted that the recovery factors of the TS floc were both below 10% at pH values of 7 and 9.

Fig. 6 Floc breakage (a) and recovery (b) factors of AS, FS and TS flocs under various pH conditions.

The D_f values of AS, FS and TS flocs under different solution pH conditions before breakage were also calculated by eqn (3) and (4). According to Table 4, it can be noted that the D_f values of flocs were also significantly affected by the solution pH. At a pH of below 5, the D_f value of the FS floc (2.11) was slightly higher than those of the AS floc (2.03) and TS floc (2.04). However, at pH values of 7 and 9, the D_f values of the TS floc (2.32 and 2.39, respectively) were significantly higher than those of the AS floc (2.03 and 2.04, respectively) and FS floc (2.07 and 2.09, respectively). Therefore, flocs formed by FS were more compact than those formed by AS and TS under acidic conditions, whereas TS flocs were much more compact than those formed by AS and FS in neutral and basic conditions. In addition, it can also be found that only small changes occurred in the structure of AS and FS flocs when the initial solution pH was varied. However, the variation in the structure of the TS floc was much more obvious when the pH value changed from 5 to 9.

Table 4 Floc fractal dimension values of AS, FS and TS flocs under various solution pH conditions

pH	AS	FS	TS
5	2.03	2.11	2.04
7	2.03	2.07	2.32
9	2.05	2.09	2.39

---

When the pH was slightly lower than that of the minimum solubility of the metal ion, the hydrolysates of the metal coagulants were mainly medium polymers and monomers with a positive charge. When the pH was higher than that of the minimum solubility, the main hydrolysates of the coagulants were polymers with a high molar mass or colloidal/precipitated species.^25,26 In addition, when the solution pH was low, it had been suggested that complexation and adsorption charge neutralization effects of the medium polymers and monomers played primary roles. However, when the solution pH was higher, the main coagulation mechanism was adsorption or sweep flocculation by polymers with a high molar mass or colloidal/precipitated species.^25,26 A previous study also indicated that flocs formed by sweep flocculation had a higher growth rate and larger size than those formed by charge neutralization. Flocs formed by sweep flocculation were also more compact than those formed by charge neutralization.²⁷ Li et al. suggested that floc strength was in the following order: bridging > charge neutralization > sweep flocculation.²⁸ In addition, a previous study also proved that flocs formed by charge neutralization had higher recovery ability than those formed by other mechanisms. The structure of inter-particle bonding in a floc can also be revealed by floc breakage and recovery ability. Flocs that aggregated together by physical bonding were easier to recover after breakage. Hence, it can also be inferred that coagulation by AS was more dominated by charge neutralization than coagulation by FS and TS, in which sweep coagulation played a more significant role.

3.4 Membrane fouling of AS, FS and TS in C–UF treatment process

The membrane fouling of different coagulants was determined by a series of UF experiments. The normalized permeate flux (J/J₀) was used to represent the performance of the UF membrane and the results are shown in Fig. 7. Fig. 7 shows declining trends in the permeate flux for HA–kaolin water (without pH buffering) pretreated by AS, FS and TS at their respective optimal dosages (10 mg L⁻¹, 26 mg L⁻¹ and 32 mg L⁻¹, respectively). The permeate fluxes for these three coagulants all decreased dramatically initially, and then decreased slowly to reach a plateau state.

Fig. 7 Change in flux with time for AS, FS and TS under pre-coagulation conditions.

It is obvious that the decline in flux in coagulation by TS was slower than those in coagulation by AS and FS. The final normalized permeate flux for TS was about 0.416, which was slightly higher than those for AS (about 0.377) and FS (about 0.357). In addition, the MFI of each coagulation–UF system was calculated by eqn (5) and the results are also shown in Fig. 7. The MFI values for the various coagulants were in the following order: FS > AS > TS, which was consistent with the results mentioned above. That may be induced by the strength, size and structure of the flocs as demonstrated by the above experiments. Although the flocs formed by TS were denser and more compact, the least serious membrane fouling was observed in the TS C–UF system. This may be ascribed to its greater floc strength and larger floc size. However, the most serious membrane fouling, namely, that in the FS C–UF system may be induced by the compact floc structure and weak floc.^29,30 AS formed the floc that was smallest but had the loosest structure and achieved an intermediate level of membrane fouling.

3.5 Effect of solution pH on membrane fouling of AS, FS and TS

UF experiments were conducted by filtering HA–kaolin water pretreated by AS, FS and TS at various solution pH values. The declining trends in permeate flux caused by coagulation of water by these three coagulants at pH values of 5, 7 and 9 are shown in Fig. 8. It can be noted that the permeate fluxes decreased quickly first and then the rate of reduction became slower. It was obvious that the degree of membrane fouling was affected by the solution pH. At a pH of 5, the permeate fluxes for AS and FS decreased faster than that for TS, and the final permeate fluxes followed the order of TS > AS > FS. At a pH of 7, the permeate flux for TS first decreased faster than that for AS, but slower than that for FS. However, the final permeate flux for TS was the largest among these three coagulants. At a pH of 9, the results were similar to those at a pH of 7, while the gap between different coagulants was narrower. In addition, the MFI values from UF experiments on AS, FS and TS in different pH conditions were calculated and are shown in Fig. 8. It can be noted that regardless of the solution pH, the MFI for AS, FS and TS increased in the following order: TS < AS < FS. In conclusion, all the coagulants exhibited their respective minimum extents of membrane fouling at a pH of 7.

Fig. 8 Change in flux with time for AS, FS and TS under various pH conditions.

The membrane fouling was markedly influenced by the floc properties, in particular the cake layer, which mainly caused the reversible fouling.^4,30,31 At a pH of 5, the floc size of AS was slightly larger than that of TS, but the gap was very small. Coagulation by AS only led to an intermediate extent of membrane fouling. A previous study proved that floc size was not sufficient to explain membrane performance.³² Membrane performance could also be affected by floc strength and floc structure. The minimum extent of membrane fouling for coagulation by TS may arise from the loose floc structure. Lee et al. found that specific membrane resistance could decrease with an increase in floc size and reduction in floc fractal dimension.¹¹ At pH values of 7 and 9, the floc formed by FS attained the lowest floc strength according to Fig. 6(a). Moreover, the intermediate D_f of the floc and its largest floc size did not alleviate membrane fouling. All these three coagulation systems achieved their lowest extent of membrane fouling at a pH of 7. This may be due to the charge neutralization effect of metal coagulants under neutral conditions. A previous study suggested that, as a result of a charge neutralization effect, coagulated water usually exhibited a slower decline in membrane permeate flux compared with that due to sweep coagulation.⁴ In conclusion, TS flocs had the largest size, the highest D_f value and intermediate floc strength, which led to a minimum extent of membrane fouling.

4. Conclusion

The floc properties and membrane fouling of TS in a coagulation–ultrafiltration process were comparatively investigated in this work. From this research, it could be found that in neutral and basic conditions, TS formed larger and stronger flocs with the lowest recovery ability. A small and strong floc with a high recovery ability was formed by AS. The FS floc attained an intermediate size and was weaker than the AS and TS flocs. In acidic conditions, the floc formed by FS was the largest. The breakage factors were in the following order: AS > FS > TS. In addition, the membrane fouling of TS was the slightest among these three coagulants in the investigated range of pH values. The most serious membrane fouling was caused by the FS coagulation system.

Acknowledgements

This work was supported by a grant from the Chinese National Natural Science Foundation (No. 51278283) and grants from the Tai Shan Scholar Foundation (No. ts201511003).

References

1. W. Yu and N. J. D. Graham, J. Membr. Sci., 2015, 473, 283–291.
2. W. Yuan and A. L. Zydney, Environ. Sci. Technol., 2000, 34, 5043–5050.
3. J. Liu, B. Liu, T. Liu, Y. Bai and S. Yu, Desalination, 2014, 333, 126–133.
4. W.-z. Yu, N. Graham, H.-j. Liu and J.-h. Qu, Chem. Eng. J., 2013, 234, 158–165.
5. Z. Yang, B. Gao, W. Xu, B. Cao and Q. Yue, J. Hazard. Mater., 2011, 189, 203–210.
6. H. K. Shon, S. Vigneswaran, I. S. Kim, J. Cho, G. J. Kim, J. B. Kim and J. H. Kim, Environ. Sci. Technol., 2007, 41, 1372–1377.
7. Y. X. Zhao, B. Y. Gao, H. K. Shon, B. C. Cao and J. H. Kim, J. Hazard. Mater., 2011, 185, 1536–1542.
8. Y.-F. Wu, W. Liu, N.-Y. Gao and T. Tao, Water Res., 2011, 45, 3704–3711.
9. A. Bagga, S. Chellam and D. A. Clifford, J. Membr. Sci., 2008, 309, 82–93.
10. W. Yu, T. Liu, J. Gregory, L. Campos, G. Li and J. Qu, J. Membr. Sci., 2011, 385–386, 194–199.
11. S. A. Lee, A. G. Fane and T. D. Waite, Environ. Sci. Technol., 2005, 39, 6477–6486.
12. P.-K. Park, C.-H. Lee and S. Lee, Environ. Sci. Technol., 2006, 40, 2699–2705.
13. P. Jarvis, B. Jefferson, J. Gregory and S. A. Parsons, Water Res., 2005, 39, 3121–3137.
14. P. Jarvis, B. Jefferson and S. A. Parsons, Environ. Sci. Technol., 2005, 39, 2307–2314.
15. S. F. E. Boerlage, M. D. Kennedy, M. P. Aniye, E. Abogrean, Z. S. Tarawneh and J. C. Schippers, J. Membr. Sci., 2003, 211, 271–289.
16. S. F. E. Boerlage, M. D. Kennedy, M. R. Dickson, D. E. Y. El-Hodali and J. C. Schippers, J. Membr. Sci., 2002, 197, 1–21.
17. R. Mao, Y. Wang, B. Zhang, W. Xu, M. Dong and B. Gao, Desalination, 2013, 314, 161–168.
18. J. Duan and J. Gregory, Adv. Colloid Interface Sci., 2003, 100–102, 475–502.
19. F. Xiao, P. Yi, X.-R. Pan, B.-J. Zhang and C. Lee, Desalination, 2010, 250, 902–907.
20. Y. X. Zhao, B. Y. Gao, Q. B. Qi, Y. Wang, S. Phuntsho, J. H. Kim, Q. Y. Yue, Q. Li and H. K. Shon, J. Hazard. Mater., 2013, 258–259, 84–92.
21. M. I. Aguilar, J. Sáez, M. Lloréns, A. Soler and J. F. Ortuño, Water Res., 2003, 37, 2233–2241.
22. V. Chaignon, B. S. Lartiges, A. El Samrani and C. Mustin, Water Res., 2002, 36, 676–684.
23. J. Wang, J. Guan, S. R. Santiwong and T. D. Waite, Desalination, 2010, 258, 19–27.
24. Y. X. Zhao, B. Y. Gao, H. K. Shon, B. C. Cao and J. H. Kim, J. Hazard. Mater., 2011, 185, 1536–1542.
25. M. Yan, D. Wang, J. Qu, J. Ni and C. W. K. Chow, Water Res., 2008, 42, 2278–2286.
26. A. Matilainen, M. Vepsäläinen and M. Sillanpää, Adv. Colloid Interface Sci., 2010, 159, 189–197.
27. S.-H. Kim, B.-H. Moon and H.-I. Lee, Microchem. J., 2001, 68, 197–203.
28. T. Li, Z. Zhu, D. Wang, C. Yao and H. Tang, Powder Technol., 2006, 168, 104–110.
29. A. Touffet, J. Baron, B. Welte, M. Joyeux, B. Teychene and H. Gallard, J. Membr. Sci., 2015, 489, 284–291.
30. T. D. Waite, A. I. Schäfer, A. G. Fane and A. Heuer, J. Colloid Interface Sci., 1999, 212, 264–274.
31. J. Nan, M. Yao, Q. Li, D. Zhan, T. Chen, Z. Wang and H. Li, RSC Adv., 2016, 6, 163–173.
32. M. Kimura, Y. Matsui, S. Saito, T. Takahashi, M. Nakagawa, N. Shirasaki and T. Matsushita, J. Membr. Sci., 2015, 477, 115–122.

---