Effect of Ions on Removal of TiO 2 Nanoparticles by Coagulation and Microfiltration

Abstract

Removal of TiO₂ nanoparticles (NPs) from water by coagulation and microfiltration (MF) was investigated at various ionic species and strengths. Appropriate coagulant doses were determined based on the Derjaguin-Landau-Verwey-Overbeek (DLVO) theory, and coagulation with ferric chloride (FeCl₃) at 0.1–0.2 mM or poly-aluminum chloride (PACl; [Al₂(OH)_nCl_6-n]_m) at 0.2–0.4 mM improved removal of TiO₂ NPs to more than 90%, and, simultaneously, turbidity was significantly reduced. Using 0.2 mM FeCl₃ as a coagulant, the mean particle size of suspended TiO₂ NPs decreased from 145 to 43 nm after coagulation and sedimentation as a result of effective floc formation. Microfiltration (nominal pore size: 0.45 μm and 0.10 μm) of uncoagulated TiO₂ NP suspensions produced lower NP concentrations in the filtrate than that in the filtrate of pre-coagulated water. This is because pre-coagulation and settling removed large NPs in the feed water that reduce the number of smaller particles than the pore sizes by clogging or blocking membrane pores. Results suggest that removal of NPs by coagulation/sedimentation followed by membrane filtration may be less efficient than direct membrane filtration. Addition of salts (NaCl 0.3–15 mM or CaCl₂ 0.1–5 mM) to TiO₂ NP suspensions before coagulation led to significant reduction (0.28–0.47 mg/L) of TiO₂ concentrations in supernatant after sedimentation. Phosphate addition (0.01, 0.1, and 1.0 mM) increased TiO₂ NP concentration after FeCl₃ or PACl coagulation and settling, but TiO₂ NP removal rate by MF was improved by phosphate addition except for filtering TiO₂ NP suspension with 1.0 mM phosphate by a 0.45 μm membrane.

Introduction

Engineered nanoparticles (NPs) possess unique properties that have made them useful materials for an increasing number of applications. With over 1200 commercially available materials, these include applications such as cosmetics, paints, electronics, and cleaning supplies (Dunphy Guzmán et al., 2006; Reijnders, 2006; Wiesner et al., 2006). Among different NPs, TiO₂ is the most widely used, with production in the United States estimated at ∼40,000 metric tons/year (Savage et al., 2007).

There is concern regarding the potential for TiO₂ NP released into the environment to harm human health (Luo et al., 2014; Shakeel et al., 2016), and TiO₂ NP concentrations in surface waters in the United States have been estimated at 24.5 μg/L (Gottschalk et al., 2009). The proportion of TiO₂ NPs from total global production expected to enter waste water treatment plants is assumed to be large, with Keller et al. (2013) suggesting they account for 54% of total production. In addition, TiO₂ NPs have been detected in wastewater treatment plant effluents and in biosolids, demonstrating both the ability of NPs to be released into water sources in significant quantities and the inability of water treatment processes to remove them (Kiser et al., 2009; Westerhoff et al., 2013). However, most studies conducted to date have focused on the removal of NPs via conventional waste water treatment, which consists of a tandem process consisting of coagulation/flocculation/sedimentation (C/F/S). While some researchers have achieved high levels of NP removal (Kinsinger et al., 2015; Sousa et al., 2017), metal breakthrough in the effluent of these studies remains a concern, particularly since the level of contaminants that can cause harm to humans is not known.

Taking the precautionary principle in health effects, and considering that the amount of NPs and their associated ions released is only expected to increase over the coming years (Klaine et al., 2008), the development of practical methods for efficient NP removal remains an important consideration. In addition, the complex interactions of the factors that can influence NP removal in water treatment processes, such as water type, pH, the presence of natural organic matter (NOM), and coagulant type (Hwang et al., 2008; Hackenberg et al., 2010; Honda et al., 2014), are still not completely understood, and further research is required in this area. Moreover, the size, surface properties, and concentrations of NPs can also affect their removal during water treatment (Limbach et al., 2008; Brar et al., 2010; Yao et al., 2014).

Wang et al. (2013) used an initial TiO₂ NP concentration of 30 mg/L and found that polyferric sulfate as coagulant at 0.3 mM can achieve high removal rates (∼84%) with less pH reduction. Chalew et al. (2013) studied the removal efficiencies of various NPs at an initial concentration of 1 mg/L by the conventional treatment processes and found that the removal rates of TiO₂ NPs were between 92% and 97%. The removal efficiency of TiO₂ is affected by myriad factors such as the presence of NOM. For example, the removal efficiencies of TiO₂ NPs were found to be lower in tap water containing 3 mg/L of total organic carbon (TOC) than in buffered nanopure water for the same alum dosage (Zhang et al., 2008).

In addition to conventional water treatment utilizing C/F/S, advanced water treatment using low-pressure membrane (LPM) filtration has increased in popularity in recent years (Huang et al., 2009). In fact, to the best of our knowledge, there is only one report in the literature investigating the effects of membrane filtration on simulated NP removal filtration experiments (Chalew et al., 2013). In their study, waters taken from different sources were spiked with 1 mg/L TiO₂ NPs, achieving removal rates of 56–100% and 96–100% for microfiltration (MF) and ultrafiltration, respectively. However, in this study, filtration was conducted using both coagulated and uncoagulated, that is, control, TiO₂ NPs in feed waters, which was assumed to be simulating C/F/S combined with membrane filtration as a surrogate of sand filtration. This is important, because the efficiency of filtration depends on the NP size, which will vary depending on coagulation and flocculation processes. These processes, in turn, are greatly affected by the operating conditions, such as the coagulant and water matrix. Hence, filtration from a fresh batch of untreated NPs is unlikely to yield the same results as that achieved after water treatment by C/F/S. Therefore, this study was conducted to gain further insights into simulated advanced water treatments consisting of C/F/S in conjunction with membrane filtration, and to make comparisons with direct filtration of untreated TiO₂ NP suspensions. This was because it is important to report that, while flocculated NPs could be settled down and removed by the C/F/S process, a small fraction of the suspended NPs remaining in the supernatant might be more difficult to be removed by a subsequent MF process, compared to filtering uncoagulated water. We also investigated the change in particle size distribution in C/F/S experiments with ferric chloride (FeCl₃) as a coagulant. Finally, the effects of various salts (NaCl, CaCl₂, and K₂HPO₄) on TiO₂ NP removal by coagulation, sedimentation, and MF were studied.

Materials and Methods

Preparation of TiO₂ NP suspensions

TiO₂ NPs were purchased from Sigma Aldrich Co., LLC (Saint Louis, MO). The as-received TiO₂ NPs were in a powdered form. To prepare the TiO₂ suspensions, 50 mg of TiO₂ was added to 500 mL Milli-Q water, sonicated at 150 W for 30 min, and then diluted to different concentrations by adding Milli-Q water in different solutions. Solutions of NaHCO₃ (50 mM), FeCl₃ (50 mM), and poly-aluminum chloride (PACl) (99 mM PACl as Al) ([Al₂(OH)_nCl_6-n]_m) (1 ≤ n ≤ 5, m ≤ 10) were separately prepared.

Preparation of NaCl, CaCl₂, and phosphate solutions for TiO₂ NP removal by jar tests

Salt solutions were prepared as follows. An accurate amount of NaCl (5.844 g) or CaCl₂ (11.098 g) was dissolved in 100 mL of Milli-Q water, forming 1 M solutions. A stock solution of the phosphate solution was then prepared with anhydrous K₂HPO₄ (Kishida Chemical Co., Ltd.) and then diluted eight times to 125 mM. For the experiment with added phosphate, 100 mL of a 125 mg/L TiO₂ suspension was transferred to a flask and diluted to 500 mL with 1.25 mM phosphate, forming a suspension with final concentrations of 25 mg/L TiO₂ and 1 mM phosphate. The mixture was then vigorously stirred for 12 h to ensure the equilibrium adsorption of phosphate on TiO₂. The NaCl and CaCl₂ solutions used in the jar tests were prepared using the same method.

Jar tests for determining removal efficiencies

NaHCO₃ (0.3 mmol/L for 0.1 mM FeCl₃ or 0.6 mmol/L for 0.2 mM FeCl₃) was added to the TiO₂ NP solution (300 mL; 25 mg/L) before addition of coagulant to neutralize acid released and maintain the solution pH within optimum levels. The coagulant doses (FeCl₃ 0.1 or 0.2 mM and PACl 0.2 or 0.4 mM) were determined by preliminary experiments. For jar tests with added salts, 1 mL of 50 mM FeCl₃ was added into 500 mL of TiO₂ NP suspension to produce 0.1 mM Fe as coagulant. Rapid mixing at 200 rpm with a mechanical stirrer was conducted for 2 min, followed by slow mixing at 30 rpm for 30 min, after which the aggregated flocs were allowed to undergo sedimentation for 30 min. Triplicate samples were taken for water quality (residual turbidity) and TiO₂ NP measurements (TiO₂ NP concentration, mean particle size), as well as the zeta potential of the solution at each of the following stages: before coagulant addition, after rapid mixing, and after sedimentation.

Membrane filtration removal efficiency experiments

The stirred filtration unit with a capacity of 50 mL (Amicon 8050; Millipore Corp., Billerica, MA) was used in the membrane filtration experiments. Two hydrophilic polyvinylidene difluoride (PVDF) filters with the same diameter of 47 mm, but different pore sizes of 0.10 μm (VVLP04700) and 0.45 μm (HVLP04700), were purchased from Millipore Corp. Approximately 50 mL of feed water was pressured into a cylindrical container with the membrane on the bottom. The periphery of a membrane (47 mm in diameter) was cut to make it into 45 mm in diameter so as to be fitted in the filter housing. The first and last 10 mL of filtered water were discarded; thus, around 30 mL of filtered water was collected after filtration.

Analytical methods

Particle size distribution was evaluated by NP tracking analysis (LM10; NanoSight, United Kingdom) to measure NPs larger than 10 nm. The surface charge of the NPs was measured using an ELS SF-8000 particle size analyzer (OTSUKA Electronics, Japan).

Concentration of TiO₂ NPs in suspension (1–25 mg/L) was measured using a turbidity meter (2100Q; HACH). For concentrations lower than 1 mg/L, the data were analyzed with more sensitive instrumentation, namely, inductively coupled plasma-mass spectrometry (ICP-MS) (7500; Agilent). Accurate determination of NP concentrations in previous studies has often been hampered by a lack of such sensitive instruments (Zhang et al., 2008).

Liquid and solid samples were digested using the HNO₃/H₂SO₄ acid digestion method for Ti as described by Standard Method 3030 G for water and wastewater analysis (APHA, 2005), which was also used in other studies (Kiser et al., 2009). All solutions were heated to 160°C for 10 min, held for 5 min, and then slowly heated to 175°C over 7 min, where they were held for 10 min. Following digestion, all solutions were diluted to 50 mL with Milli-Q water. The performance of the microwave acid digestion was tested using triplicate 1000 μg/L Ti solutions diluted from Ti standard solution following the same steps as for the TiO₂ NP samples. The solution was digested in the microwave and then diluted to 50 mL with Milli-Q water. The results showed that the analyzed concentration was 997.97 ± 8.15 μg/L (n = 3), indicating that loss and interference during microwave digestion were negligible (recovered mass was 99.8%).

The detection limit of this method is 0.36 μg/L (Ti) by measuring seven samples with a concentration of 1 μg/L diluted from a standard solution, which is similar to the detection limit of 0.35 μg/L obtained in previous research using HCl/HNO₃ digestion (Chalew et al., 2013).

Results and Discussion

TiO₂ NP removal efficiency by coagulation

Derjaguin-Landau-Verwey-Overbeek (DLVO) theory explains the tendency of colloids to agglomerate or remain discrete by forming a net interaction energy curve, and the height of the barrier indicates how stable the system is (Trefalt et al., 2014). The addition of 0.1–0.2 mM FeCl₃ or 0.2–0.4 mM Al (III) (PACl) as coagulants to a mixture of TiO₂ NPs led to disappearance of the energy barrier present in the NPs dissolved in Milli-Q water (Fig. 1; left and right side, respectively), suggesting that coagulant addition at these concentrations was effective for particle destabilization and agglomeration.

FIG. 1.

Derjaguin-Landau-Verwey-Overbeek (DLVO) interaction energy after dosing with FeCl₃ (left) and PACl (right) as coagulant. PACl, poly-aluminum chloride; FeCl₃, ferric chloride.

Having determined appropriate doses of coagulants based on the DLVO theory, we investigated the effects of these two coagulants (FeCl₃ and PACl) on solution turbidity, TiO₂ NP concentration, mean NP size, and zeta potential (Figs. 2a–d and 3a–d, respectively) at three different stages: before addition of coagulant, after rapid mixing (coagulation), and after settling (sedimentation). TiO₂ NPs can be removed through coagulation if they are enmeshed by flocs as they settle out of the water in a process called sweep flocculation (Crittenden et al., 2005). Alternatively, coagulants added to the water may affect the stability of TiO₂ NPs by producing positively charged hydrolytic species (Al³⁺, Fe³⁺, etc.) that neutralize negative surface charges on NPs (Westerhoff et al., 2011), resulting in greater NP aggregation because electrostatic repulsion between negatively charged particles is mitigated (Petosa et al., 2010).

FIG. 2.

Parameter changes during jar test (n = 3) (FeCl₃ as coagulant). Values with bars indicate mean and standard deviation. *Not quantified. (a) Turbidity, (b) TiO₂ concentration, (c) particle size, and (d) Zeta potential of TiO₂.

FIG. 3.

Parameter changes during jar test (n = 3) (PACl as coagulant). *Not quantified. (a) Turbidity, (b) TiO₂ concentration, (c) particle size, and (d) Zeta potential of TiO₂.

An approximate measure of the extent to which the TiO₂ NPs have been removed can be achieved by monitoring the turbidity level. Indeed, optimal coagulant doses are often determined based on the turbidity level, which is used as a surrogate for NOM reduction. The turbidity before the addition of the iron coagulant (Fig. 2a) was nearly the same in all the experimental conditions between 92 and 96 nephelometric turbidity unit (NTU), only depending on the concentration of NaHCO₃ added initially. The turbidity decreased significantly after sedimentation with FeCl₃ as coagulant, with a 98–99% reduction being achieved, while the turbidity reduction was much lower (26%) in the control experiments (Fig. 2a). The mean particle size varies as NPs aggregate into larger flocs and is an indicator of the level of aggregation. Even after rapid mixing, the particle size did not change much in the control experiment (Fig. 2c). These findings indicate that rapid mixing in the absence of coagulant does not lead to coagulation and, therefore, TiO₂ NPs are stable in Milli-Q water. However, when 0.2 mM FeCl₃ was used, the particle size increased significantly after rapid mixing in a process in which stabilized NPs aggregated together to form larger flocs that eventually settled and precipitated during the sedimentation process. Thus, the remaining amount of colloidal material left in the suspensions after the sedimentation was very small, with a turbidity level of only 1–2% of that before the addition of coagulant. Following sedimentation, only small negatively charged particles that were not neutralized by the positively charged coagulant [Fe(III)] and which did not aggregate into flocs remained, leading to a reduction in the mean particle size in this stage. In the jar test with 0.1 mM Fe, the zeta potential changed to slightly positive after rapid mixing and then to slightly negative after sedimentation (Fig. 2d). These results indicated that the surface charge of TiO₂ NPs became positive in response to addition of excess coagulant, but that those particles with an overall negative charge, which did not bind to the coagulant, remained suspended after sedimentation. Despite this, the energy barrier remained unaffected, so that the attractive forces between particles became dominant. However, when 0.2 mM Fe was used, the zeta potential remained positive after sedimentation. Considering the lower TiO₂ NP concentration achieved with 0.2 mM Fe, these results indicate that, to neutralize negatively charged TiO₂ NPs, an excess amount of coagulant must be added, making the average zeta potential positive after sedimentation.

The trend in turbidity with the addition of PACl (Fig. 3a) was the same as that of FeCl_3, indicating that the addition of either FeCl₃ or PACl effectively removed TiO₂ NPs from aqueous solutions. However, there are some differences between these two coagulants. The particle size was larger with PACl at 0.2 mM than with FeCl₃ at the same concentration (Fig. 3c). Moreover, a negative zeta potential was observed after sedimentation with both 0.1 mM FeCl₃ and 0.2 mM PACl, in contrast to 0.2 mM FeCl₃, where a positive zeta potential was observed (Fig. 3d). However, for 0.4 mM PACl, the charge was positive. These results indicated that, at the same dose, FeCl₃ leads to greater neutralization of the surface charge than PACl. In addition, the slightly lower pH observed for both concentrations of FeCl₃ than PACl might have resulted in a slightly higher surface charge on the TiO₂ NPs (Figs. 2d and 3d). This means that there is no significant difference in coagulation capacity between FeCl₃ and PACl at 0.2 mM. The difference in particle sizes after rapid mixing and settling is affected by the rates of aggregation and settling. Since we were not able to measure these two factors separately, we cannot show each of these two rates. However, it is postulated that, in addition to the interaction energy as estimated by DLVO in Fig. 1, aggregate/floc size and density influence the rate of aggregation. As iron is heavier than aluminum, it is postulated that iron aggregate could settle more rapidly than aluminum aggregate.

Figure 4 shows the changes in the particle size distribution during the course of the coagulation experiment with 0.2 mM FeCl₃. Before the coagulant was added, the TiO₂ NPs were monodispersed, showing a unimodal distribution with a single peak at 145 nm (Fig. 4a). After the rapid mixing stage during which coagulation occurred, the peak at 145 nm disappeared, while a multimodal distribution appeared with three other peaks, at 338, 424, and 699 nm. These findings are in agreement with the results of Domingos et al. (2010), who also reported multimodality during the aggregation process. This increase in particle size is indicative of TiO₂ NPs gradually coagulating to form larger aggregates. Following sedimentation, larger aggregates precipitated out of the solution, leaving behind smaller unaggregated clusters dissolved in the mixture, and the mean particle size decreased to 43 nm, which was lower than that before coagulant addition (145 nm). After coagulation, a total of 4 × 10⁶ particles/mL were detected in solution, which was about 6.6% of the total particle number before the addition of 0.2 mM FeCl₃.

FIG. 4.

Particle size distribution change during the jar test. (a) Particle concentration 0.61 × 10⁸ particles/mL, (b) Particle concentration 5.32 × 10⁸particles/mL, (c) Particle concentration 0.04 × 10⁸ particles/mL. Particle size distribution was measured by the nanoparticle tracking analysis (NTA) method (Nanosight) and expressed as the number-based distribution. Coagulant concentration—Fe 0.2 mM. (a) Before coagulant dose, (b) after rapid mixing, and (c) after settling.

As flocs settled in solution and the aggregate size increased, the density decreased. During this process, larger aggregates may be less densely packed than smaller ones, and there may be void volumes between the smaller aggregates.

TiO₂ NP removal by coagulation and membrane filtration

In past studies, TiO₂ NPs produced a higher removal rate than other species such as Zn and Ag NPs. One study demonstrated that TiO₂ NPs could be removed via C/F/S processes with a 3–8% breakthrough (92–97% removal rate) (Chalew et al., 2013), while another study showed the NPs could be eliminated with an 84% removal rate (Wang et al., 2013). In our study, we obtained the removal rates of 80–90%, which further confirms the ability of C/F/S processes to remove a large proportion of TiO₂ NPs.

To the best of our knowledge, the only study conducted to date that has investigated the effects of membrane filtration on NP removal has been carried out by Chalew et al. (2013); however, their study did not consider the effects of membrane filtration when carried out as part of a tandem process following C/F/S. Even though conventional treatment has been determined to be effective for the elimination of NPs with large removal rates, there is a concern that the resulting concentrations might still be high enough to harm aquatic life (Chalew et al., 2013); in addition, such concentrations could still potentially be harmful to humans. Therefore, considering that NP use and release into aquatic systems are expected to increase over the coming years, there is a need to improve conventional water treatment processes (C/F/S) to maximize the removal of potentially hazardous NPs.

To further understand the removal of NPs through advanced water treatment by LPM filtration (Huang et al., 2009), we passed a series of (untreated) TiO₂ solutions ranging in concentration from 0.2 to 25 mg/L through a 0.45 μm PVDF MF membrane. Following filtration, the TiO₂ NP concentrations decreased to 17–121 μg/L, and the removal efficiencies in all cases were over 90% (Fig. 5), although a slight increase in the removal rate was observed when the feed water concentration increased.

FIG. 5.

Removal efficiency of different concentrations of TiO₂ filtered by 0.45 μm polyvinylidene difluoride membrane (n = 3).

To gain further insight into the viability of LPM filtration for practical applications in advanced water treatment, solutions treated by C/F/S with 0.1 mM FeCl₃ were passed through 0.45 μm PVDF and 0.1 μm PVDF filters, and the concentrations of the resulting filtrate were determined by ICP-MS. These results were compared to those of analogous experiments in which 25 mg/L of untreated TiO₂ particles was used (Table 1). Two factors contributed to the overall concentration found in the filtrate, the size of the NPs that come in contact with the membrane filter and the total concentration of the initial solution. As shown in Fig. 4, there is a considerable reduction in NP size after C/F/S (from 145 nm before treatment with 0.2 mM FeCl₃ to 43 nm after sedimentation). Hence, beyond a certain pore size, particles may become too small to be filtered by the membrane; therefore, membranes with appropriate pore sizes need to be chosen.

Table 1.

Concentrations (μg/L) of TiO₂ After Filtration with Various Membranes and TiO₂ Supernatant After Slow Mixing (n = 3) (Ferric Chloride 0.1 mM)

Feed water before filtration	0.45 μm (PVDF)	0.1 μm (PVDF)	0.05 μm (poly-carbonate)
25 mg/L TiO₂	120.81 ± 3.54	62.78 ± 3.34	31.26 ± 2.11
Sedimentation supernatant^a	160.37 ± 19.81	72.26 ± 2.25	54.75 ± 1.98

After treating 25 mg/L TiO₂ with 0.1 mM FeCl₃.

FeCl₃, ferric chloride; PVDF, polyvinylidene difluoride.

The concentration of prefiltrated solution can have an unexpected effect in the filtrate concentration at sufficiently high NP concentrations. Hence, after filtering the solutions at the same initial concentration of 25 mg/L through a 0.45 μm MF membrane, the TiO₂ NP concentrations in filtrate were 120.8 ± 3.5 μg/L for the uncoagulated solution and 160.3 ± 19.8 μg/L for the C/F/S-treated solution. This result could be ascribed to two factors; namely, differences in sizes of suspended particles and concentrations of TiO₂ NP in coagulated and uncoagulated solutions that were fed to the membrane. As mentioned above, suspended particles in the feed water after coagulation/sedimentation were small enough to pass through the MF membrane, resulting in a filtrate with a greater concentration of smaller sized TiO₂ NP particles. Moreover, particles in the uncoagulated feed water had sizes greater than the MF membrane pores and thus could block the membrane pores, making particles smaller than the membrane pores unable to pass through the pores. In addition, the uncoagulated feed water had higher concentrations of NPs than coagulated ones and thus prone to cause clog. This mechanism is estimated to lead to a reduction in the number of particles in the filtrate. On the contrary, the coagulated solutions had smaller NP sizes with much lower initial NP concentrations, making it difficult to block the membrane pores; accordingly, the smaller particles than the membrane pores passed through the membrane.

These results suggest that at high TiO₂ NP concentrations (25 mg/L), TiO₂ removal by C/F/S might be less efficient than direct filtration through a membrane because of the smaller particle sizes and much lower initial concentrations of solutions subjected to treatment. However, such high concentrations of NPs would not be encountered in practice in large-scale water treatment plants; therefore, it is reasonable to assume that under actual concentrations found in water, removal of NPs by conventional C/F/S in tandem with membrane filtration may be less efficient than using membrane filtration alone (Table 1).

Effects of NaCl and CaCl₂ on removal efficiency by coagulation

The existence of NaCl and CaCl₂ in solution has been shown to enhance the aggregation of TiO₂ NPs. Figure 6 shows the effects of ions (i.e., NaCl and CaCl₂) on the removal efficiency of TiO₂ NPs by 0.1 mM FeCl₃.

FIG. 6.

Comparison of TiO₂ concentration in supernatant and zeta potential change before dosing coagulant and 30 min after gravity settling (FeCl₃ 0.1 mM).

With the addition of NaCl or CaCl₂ into a solution containing an initial TiO₂ NP concentration of 25 mg/L, the final TiO₂ NP concentrations after sedimentation were all lower than under control conditions and <1 mg/L (Fig. 6; Table 2). These results mirror those obtained by Chalew et al. (2013), who achieved turbidity values of 2–4 after coagulation. Although these values were slightly higher than those observed in the present study, they used water from natural sources that contained more types of ions than were evaluated in the present study.

Table 2.

Concentration of TiO₂ Nanoparticles After Sedimentation with Addition of Different Kinds of Ions

Coagulation conditions	Supernatant concentration after sedimentation (μg/L) (n = 3)	Standard deviation (μg/L)
Fe 0.1 mM	534	67
Fe 0.2 mM	232	19
Al 0.3 mM	1013	5
Al 0.4 mM	145	7
NaCl 0.3 mM	247	104
NaCl 15 mM	108	12
CaCl₂ 0.1 mM	301	9
CaCl₂ 5 mM	109	5
P 0.01 mM Fe 0.1 mM	597	20
P 0.1 mM Fe 0.1 mM	946	3
P 1 mM Fe 0.1 mM	5352	29
P 0.01 mM Al 0.4 mM	241	6
P 0.1 mM Al 0.4 mM	335	5
P 1 mM Al 0.4 mM	387	11

Overall, it can be concluded that the addition of NaCl and CaCl₂ before coagulation led to greater destabilization of TiO₂ NPs, resulting in greater particle aggregation, floc formation and sedimentation, and therefore reduced final NP concentrations. These findings are in accordance with those of a study by Wang et al. (2013), who found that increased ionic strengths led to enhancement of the coagulation effect. In the present study, addition of NaCl or CaCl₂ resulted in modest reductions in NP concentrations after sedimentation, suggesting that the effects of increased ionic strength are likely complex, probably depending on a variety of other factors such as water type, pH, and TOC.

Effects of phosphate on removal efficiency by coagulation and membrane filtration

There is evidence that phosphate anions can be attached onto the surface of TiO₂ NPs by multisite adsorption, affecting particle stability (Kang et al., 2011). Based on these findings, it is likely that phosphate addition would affect particle aggregation and therefore the removal rate of TiO₂ NPs.

To investigate the effects of phosphate addition on the TiO₂ NP removal rate, a series of experiments were conducted with 0.1 mM FeCl₃ as the coagulant and phosphate in concentrations of 0 (control experiment), 0.01, 0.1, and 1 mM (Table 2). Figure 7 shows the corresponding concentrations after sedimentation in the presence and absence of membrane filtration with 0.45 and 0.10 μM PVDF membrane filters. Phosphate was found to have a clear inhibitory effect on NP removal, with greater concentrations leading to smaller NP removal levels after sedimentation. Use of 0.45 and 0.10 μM PVDF membrane filters led to significant reductions in NP concentrations compared to unfiltered supernatant solutions. With the 0.10 μM filter, the amount removed increased gradually with increasing phosphate concentrations, from 74.3% in the control experiment up 96.7% in response to a phosphate concentration of 1 mM (Table 3). Passing the supernatant through a 0.45 μM filter led to significant reductions at phosphate concentrations of 0.01 and 0.1 mM (82.1% and 88.9%, respectively), although the reductions at 0 and 1 mM were lower (44.7% and 41.1%, respectively). These results suggest that the factors influencing NP removal by PO₄³⁻ addition are multifold and complex. Specifically:

FIG. 7.

TiO₂ concentrations after gravity settling and filtration (FeCl₃ 0.1 mM) (n = 3).

Table 3.

Removal Rates of TiO₂ Nanoparticles in Supernatants by Membrane Filtration (%)

	Phosphate concentration (mM)
Filter pore size (μm)	0	0.01	0.1	1.0
0.45	44.7	82.1	88.9	41.1
0.10	74.3	87.6	94.0	96.7

Data are taken from Fig. 7. FeCl₃ 0.1 mM.

(1) PO₄³⁻ can absorb onto TiO₂ NPs, contributing further to the negative charge of the particles and facilitating electrostatic repulsion between the NPs.

(2) PO₄³⁻ can bind to and neutralize positively charged Fe-hydroxide species, effectively eliminating the capacity of such iron species to act as coagulants.

(3) As the TiO₂-NP concentration increases in the supernatant after sedimentation with increasing phosphate concentrations, removal of the NPs during membrane filtration could increase at sufficiently high concentrations as a result of pore clogging.

Based on these mechanisms, the results from Fig. 7 and Table 3 could be interpreted as follows:

(1) The lower NP removal rates exhibited in the control experiment with both membrane filters suggest that coagulation in the absence of phosphate leads to smaller particle sizes.

(2) The effect of PO₄ addition at low dosages, that is, 0.01 and 0.1 mM, is not obvious in the TiO₂ NP concentrations after membrane filtration because PO₄ prevents aggregation of TiO₂ NPs at low concentration by rendering TiO₂ NP surface negatively charged. At a high concentration of PO₄ (1 mM), TiO₂ NP concentrations increased significantly both after coagulation/sedimentation and after membrane filtration. This may be because an increased ionic strength by high concentration of PO₄ overwhelmed the dispersing effects of PO₄, which promoted aggregation of TiO₂ NP. As a result, there remained very few large TiO₂ NPs that can clog/block the membrane pores to prevent passage of small TiO₂ NPs through the membrane pores. The reaction between Fe (III) and PO₄³⁻ is very rapid (completed in <1 s), and may be considered instantaneous (Parsons and Berry, 2004). The increase in supernatant NP concentration with increasing phosphate concentration suggests that phosphate can act as an inhibitor of the coagulant, with a possible mechanism occurring via the fast adsorption of the anion to Fe-hydroxide species. The sharp increase in NP concentration after sedimentation between phosphate concentrations of 0.1 and 1 mM (Fig. 7) indicates that the inhibitory process may be particularly potent after the 1 mM level.

Experiments were conducted to determine the effects of PO₄³⁻ addition on NP removal via PACl as the coagulating agent (Fig. 8; Table 2). The results showed that, comparatively, there was a much weaker effect on the level of NPs remaining after sedimentation relative to the experiments with FeCl₃ (Fig. 7). When compared to the control experiment, the concentration of NPs after sedimentation decreased at 0.01 mM PO₄³⁻, while it increased after filtration using 0.45 μm filters in the 0.01 mM and 1.0 mM PO₄³⁻ experiments. However, the differences between the runs (both filtered and unfiltered) with added phosphate and the control experiments were not statistically significant; therefore, clear trends could not be established.

FIG. 8.

TiO₂ concentrations after gravity settling and filtration (PACl 0.4 mM as Al) (n = 3).

When FeCl₃ was used as coagulant, addition of phosphate at 0–0.1 mM to the mixture led to a steady reduction in the removal rate, although the decrease was much more pronounced at the highest concentration of 1 mM. In a previous experiment investigating the stability of TiO₂ NPs, phosphate at 1 mM was determined to be capable of stabilizing the NPs (Kang et al., 2011), indicating that similar concentrations of phosphate can hinder the coagulation process. Satisfactory removal of NPs was achieved with the 0.45 μm PVDF filter, except in the control experiment and at the highest phosphate concentration. It should be noted that a concentration of 1 mM using the 0.1 μm filter still yielded 174 μg/L TiO₂ NPs in the filtrate, which was the highest content of any experiment with this filter. Based on these findings, it is reasonable to assume that, apart from any possible iron hydroxide/phosphate interactions, higher phosphate concentrations could have a stabilizing effect on the NPs and lead to larger amounts of smaller sized particles being formed.

Ability of phosphate to interact with iron oxides has been widely documented, but reports of the performance of aluminum for phosphate removal are much rarer (Valsami-Jones, 2004). This may suggest that iron can form more stable complexes with ions such as phosphate in water. Even though iron remains an efficient precipitant for phosphorus, complexation with other species, such as NOM present in water (Weng et al., 2012), could still affect the overall sedimentation process.

Conclusions

More than 90% of TiO₂ NPs was removed from a solution with an initial concentration of 25 mg/L by coagulation and sedimentation when FeCl₃ and PACl were used as coagulants. The coagulant doses were sufficient to eliminate the energy barrier required for particle aggregation, as determined by DLVO models. However, it was determined that the minimum dose of coagulant required to induce the desired change in particle size distribution after the coagulation process was 0.1 mM FeCl₃ or 0.2 mM PACl. At the same molar concentration (0.2 mM), FeCl₃ showed better removal efficiency of TiO₂ NPs than PACl. The changes in zeta potential observed with FeCl₃ and PACl as coagulants appear to be in conformity with this result. When FeCl₃ was used as a coagulant, the zeta potential changed from negative before the addition of coagulant to slightly positive after sedimentation; however, when PACl was used the values remained negative during these processes.

The addition of salts (NaCl and CaCl₂) at concentrations ranging from 0.1 to 15 mM led to modest reductions in NP concentrations after sedimentation, which is in accordance with previous reports showing that increased ionic strengths lead to a reduction in particle concentrations. When FeCl₃ was used as a coagulant, higher phosphate concentrations led to increasingly higher supernatant concentrations after sedimentation, although use of 0.45 or 0.10 μm filters still led to lower concentrations than under control conditions. Use of the 0.10 μm filter with the same coagulant led to greater NP removal rates as phosphate concentration increased, while no clear pattern was established with the 0.45 μm filter. When PACl was used as the coagulant in the absence of a filtration device, there was no significant difference between treatment with phosphate and the control. After passing treated (C/F/S) or untreated TiO₂ NP suspensions through 0.45, 0.1, or 0.05 μm membrane filters, the untreated filtrates were found to have a lower NP concentration compared to the filtrated supernatant after sedimentation. These findings suggest that floc formation after C/F/S leads to a concomitant increase in the concentration of small particles in the supernatant, yielding higher filtrate concentrations.

Overall, the results indicate that, in addition to the concentration and type of coagulant used for C/F/S processes, other factors that should be considered when predicting TiO₂ NP removal levels are the presence and concentration of ions in source water such as Na⁺, Ca²⁺, and PO₄³⁻. Therefore, there could also potentially be a seasonal effect on removal rates of NPs because of variations in ion concentrations, which may depend on seasonal precipitation.

Footnotes

Acknowledgments

The authors thank Mr. Kazuyoshi Fujimura for his technical assistance with the measurements of physical properties in the TiO₂ NPs. This study was supported by a Grant-in-Aid for Scientific Research (#26303013) by the Japan Society for the Promotion of Science (JSPS). The author, C.Z., is thankful for the joint program provided by the China Scholarship Council (CSC, 2009617060) and financial support from the Japanese government (Monbukagakusho: MEXT) scholarship.

Author Disclosure Statement

No competing financial interests exist.

References

APHA, AWWA, WEF (2005) Standard Methods for the Examination of Water and Wastewater (21st ed.). Washington, DC: American Public Health Association.

Brar

S.K.

, Verma

, Tyagi

R.D.

, and Surampalli

R.Y.

(2010). Engineered nanoparticles in wastewater and wastewater sludge—evidence and impacts. Waste Manag. 30, 504.

Chalew

T.E.A.

, Ajmani

G.S.

, Huang

, and Schwab

K.J.

(2013). Evaluating nanoparticle breakthrough during drinking water treatment. Environ. Health Perspect., 121, 1161.

Crittenden

J.C.

, Trussell

R.R.

, Hand

D.W.

, Howe

K.J.

, and Tchobanoglous

(2005). Water Treatment: Principles and Design. Second edition. Hoboken, NJ: John Wiley and Sons.

Domingos

R.F.

, Peyrot

, and Wilkinson

K.J.

(2010). Aggregation of titanium dioxide nanoparticles: Role of calcium and phosphate. Environ. Chem., 7, 61.

Dunphy Guzmán

K.A.

, Taylor

M.R.

, and Banfield

J.F.

(2006). Environmental risks of nanotechnology: National nanotechnology initiative funding, 2000–2004. Environ. Sci. Technol., 40, 1401.

Gottschalk

, Sonderer

, Scholz

R.W.

, and Nowack

(2009). Modeled environmental concentrations of engineered nanomaterials (TiO₂, ZnO, Ag, CNT, fullerenes) for different regions. Environ. Sci. Technol., 43, 9216.

Hackenberg

, Scherzed

, Kessler

, Hummel

, Technau

, Froelich

, and Kleinsasser

(2011). Silver nanoparticles: evaluation of DNA damage, toxicity and functional impairment in human mesenchymal stem cells. Toxicol Lett, 201, 27.

Honda

R.J.

, Keene

, Daniels

, and Walker

S.L.

(2014). Removal of TiO₂ nanoparticles during primary water treatment: Role of coagulant type, dose, and nanoparticle concentration. Environ. Eng. Sci., 31, 127.

10.

Huang

, Schwab

, and Jacangelo

J.G.

(2009). Pretreatment for low pressure membranes in water treatment: A review. Environ. Sci. Technol., 43, 3011.

11.

Hwang

E.T.

, Lee

J.H.

, Chae

Y.J.

, Kim

Y.S.

, Kim

B.C.

, Sang

B.I.

, et al. (2008). Analysis of the toxic mode of action of silver nanoparticles using stress‐specific bioluminescent bacteria. Small, 4, 746.

12.

Kang

S.A.

, Li

, Lee

H.E.

, and Lee

J.Y.

(2011). Phosphate uptake by TiO₂: Batch studies and NMR spectroscopic evidence for multisite adsorption. J. Colloid Interface Sci., 364, 455.

13.

Keller

, McFerran

, Lazareva

, and Suh

(2013). Global life cycle releases of engineered nanomaterials. J. Nanopart. Res., 15, 1692.

14.

Kinsinger

, Honda

, Keene

, and Walker

S.L.

(2015). Titanium dioxide nanoparticle removal in primary prefiltration stages of water treatment: Role of coating, natural organic matter, source water, and solution chemistry. Environ. Eng. Sci., 32, 1.

15.

Kiser

M.A.

, Westerhoff

, Benn

, Wang

, Perez-Rivera

, and Hristovski

(2009). Titanium nanomaterial removal and release from wastewater treatment plants. Environ. Sci. Technol., 43, 6757.

16.

Klaine

S.J.

, Alvarez

P.J.

, Batley

G.E.

, Fernandes

T.F.

, Handy

R.D.

, Lyon

D.Y.

, Mahendra

, McLaughlin

M.J.

, and Lead

J.R.

(2008). Nanomaterials in the environment: behavior, fate, bioavailability, and effects. Environ. Toxicol. Chem., 27, 1825.

17.

Limbach

L.K.

, Bereiter

, Müller

, Krebs

, Gälli

, and Stark

W.J.

(2008). Removal of oxide nanoparticles in a model wastewater treatment plant: Influence of agglomeration and surfactants on clearing efficiency. Environ. Sci. Technol., 42, 5828.

18.

Luo

, Guo

, Ngo

H.H.

, et al. (2014). A review on the occurrence of micropollutants in the aquatic environment and their fate and removal during wastewater treatment. Sci Total Environ. 473, 619.

19.

Parsons

S.A.

, and Berry

T.-A.

(2004). Chemical phosphorus removal. In Valsami-Jones

, Ed., Phosphorus in Environmental Technologies: Principles and Applications. London, United Kingdom: International Water Association (IWA Publishing), p. 261.

20.

Petosa

A.R.

, Jaisi

D.P.

, Quevedo

I.R.

, Elimelech

, and Tufenkji

(2010). Aggregation and deposition of engineered nanomaterials in aquatic environments: Role of physicochemical interactions. Environ. Sci. Technol., 44, 6532.

21.

Reijnders

(2006). Cleaner nanotechnology and hazard reduction of manufactured nanoparticles. J. Clean Prod., 14, 124.

22.

Savage

, Thomas

T.A.

, and Duncan

J.S.

(2007). Nanotechnology applications and implications research supported by the US Environmental Protection Agency STAR grants program. J. Environ. Monit., 9, 1046.

23.

Shakeel

, Shabbir

F.J.S.

, Asghar

M.S.

, Khan

M.S.

, and Chaudhry

A.S.

(2016). Toxicity of nano-titanium dioxide (TiO₂-NP) through various routes of exposure: A review. Biol. Trace Elem. Res., 172, 1.

24.

Sousa

V.S.

, Corniciuc

, and Teixeira

M.R.

(2017). The effect of TiO₂ nanoparticles removal on drinking water quality produced by conventional treatment C/F/S. Water Res. 109, 1.

25.

Trefalt

, Ruiz-Cabello

F.J.M.

, and Borkovec

(2014). Interaction forces, heteroaggregation, and deposition involving charged colloidal particles. J. Phys. Chem. B, 118, 6346.

26.

Valsami-Jones

(Ed.) (2004). Phosphorus in Environmental Technologies: Principles and Applications. London, United Kingdom: International Water Association, IWA Publishing.

27.

Wang

H.T.

, Ye

Y.Y.

, Qi

, Li

F.T.

, and Tang

Y.L.

(2013). Removal of titanium dioxide nanoparticles by coagulation: Effects of coagulants, typical ions, alkalinity and natural organic matters. Water Sci. Technol., 68, 1137.

28.

Weng

, Van Riemsdijk

W.H.

, and Hiemstra

(2012). Factors controlling phosphate interaction with iron oxides. J. Environ. Qual., 41, 628.

29.

Westerhoff

, Song

, Hristovski

, and Kiser

M.A.

(2011). Occurrence and removal of titanium at full scale wastewater treatment plants: Implications for TiO₂ nanomaterials. J. Environ. Monit., 13, 1195.

30.

Westerhoff

P.K.

, Kiser

M.A.

, and Hristovski

(2013). Nanomaterial removal and transformation during biological wastewater treatment. Environ. Eng. Sci., 30, 109.

31.

Wiesner

M.R.

, Lowry

G.V.

, Alvarez

, Dionysiou

, and Biswas

(2006). Assessing the risks of manufactured nanomaterials. Environ. Sci. Technol., 40, 4336.

32.

Yao

, Nan

, and Chen

(2014). Effect of particle size distribution on turbidity under various water quality levels during flocculation processes. Desalination, 354, 116.

33.

Zhang

, Chen

, Westerhoff

, Hristovski

, and Crittenden

J.C.

(2008). Stability of commercial metal oxide nanoparticles in water. Water Res. 42, 2204.

Effect of Ions on Removal of TiO 2 Nanoparticles by Coagulation and Microfiltration

Abstract

Abstract

Introduction

Materials and Methods

Preparation of TiO2 NP suspensions

Preparation of NaCl, CaCl2, and phosphate solutions for TiO2 NP removal by jar tests

Jar tests for determining removal efficiencies

Membrane filtration removal efficiency experiments

Analytical methods

Results and Discussion

TiO2 NP removal efficiency by coagulation

TiO2 NP removal by coagulation and membrane filtration

Effects of NaCl and CaCl2 on removal efficiency by coagulation

Effects of phosphate on removal efficiency by coagulation and membrane filtration

Conclusions

Footnotes

Acknowledgments

Author Disclosure Statement

References

Preparation of TiO₂ NP suspensions

Preparation of NaCl, CaCl₂, and phosphate solutions for TiO₂ NP removal by jar tests

TiO₂ NP removal efficiency by coagulation

TiO₂ NP removal by coagulation and membrane filtration

Effects of NaCl and CaCl₂ on removal efficiency by coagulation