This study investigated the fate of selected pharmaceuticals and estrogens and the characteristics of bulk organic matter during pellet softening and proposed a possible hybridization with nanofiltration (NF) treatment. A groundwater softening system called pellet softening was used to remove calcium ions from groundwater by crystallizing calcium carbonate on the surface of sand grains that were used as seeding material. This crystallization was confirmed by X-ray powder diffraction, and X-ray fluorescence and scanning electron micrographs were used to characterize the surface of the sand grains during pellet softening. The fluorescence excitation–emission matrix showed that humic-like substances were slightly removed and that specific UV absorbance values decreased after pellet softening. The humic fraction determined by liquid chromatography-organic carbon detection was slightly more attenuated than the fractions of biopolymers, building blocks, low molecular weight acids, and low molecular weight neutrals. Therefore, the aromatic content per unit of dissolved organic carbon was preferentially attenuated during pellet softening. The average removal efficiencies of the three estrogens and 12 selected pharmaceuticals during the softening process were 59 and 5.7%, respectively. However, there was a greater reduction of pharmaceuticals during NF.

INTRODUCTION

The occurrence of organic micropollutants (OMPs) such as pharmaceuticals and estrogens, as well as their metabolites, and other personal care products has been detected in different water sources (Ternes et al. 1999, 2002; Heberer et al. 2002; Richardson 2003; Stackelberg et al. 2004; Kim et al. 2007; Benotti et al. 2009) with a wide range of concentrations from ng/L to μg/L (Heberer 2002; Mompelat et al. 2009). Many pharmaceuticals are not completely degraded in the human body (Dębska et al. 2004) and have broader effects when combined in an aquatic environment, even at low concentrations (Daughton & Ternes 1999). Jain et al. (2013) reported that some antiviral drugs are difficult to remove during conventional water treatment, and this can result in intermediates, which are difficult to degrade. Therefore, it is important to monitor OMPs in aquatic environments and water treatment processes. Many studies have been conducted to investigate novel water treatment systems or develop new materials that can remove OMPs more effectively in water treatment processes (Maeng et al. 2015). Furthermore, most studies performed on the fate or removal of OMPs were conducted with surface water, where OMPs are often detected, and limited studies were carried out for groundwater and bank filtrate during water softening (calcium and magnesium removal). It is important to monitor OMPs in bank filtrate or groundwater treatment processes since groundwater and bank filtrate can be contaminated by OMPs. Bank filtrate mostly flows from rivers to the aquifer, and the bank filtration systems are the hydraulic connections between the river and the aquifer.

Pellet softening by crystallizing calcium carbonate on the surface of sand grains in a fluidized bed reactor using lime, caustic soda, or soda ash has been used in the Netherlands as a key softening process to reduce the hardness of groundwater or bank filtrate, and its implementation as part of a centralized scheme was first verified at the Amsterdam Water Supply (Graveland et al. 1983). Pellet softening has received a great deal of attention worldwide (van Schagen et al. 2008b) and is now commonly used for groundwater softening in the Netherlands. In pellet softening, sand grains are used as the seeding material and commonly referred to as pellets, and calcium carbonate is crystallized on the surface of these sand grains through heterogeneous primary nucleation in a fluidized bed reactor (i.e., pellet reactor). Pellet softening is also considered in zero liquid discharge applications to recover more water and reduce concentrate in a reverse osmosis (RO) system (van Houwelingen et al. 2010).

In general, a pellet reactor comprises a cylindrical tank filled with the seeding material, usually sand grains (pellets) with a diameter of 0.15 to 0.4 mm, and the water is pumped upflow with constant water velocities of 60–100 m/h (Graveland et al. 1983; van Schagen et al. 2008a; Morgan et al. 2013). Withdrawing sand grains from the pellet reactor occurs after a period of softening or depends on groundwater characteristics. Surfaces of sand grains during pellet softening increase through the crystallization of calcium carbonate, and waste pellets are often reused for agricultural applications and steel industries (Van der Bruggen et al. 2009). Crystallization of calcium occurs once the pH is raised to approximately 9.0 by adding lime, caustic soda, or soda ash. Lime is used most often, since caustic soda and soda ash raise the sodium concentration in drinking water. The crystallization of calcium carbonate on the surface of sand grains leads to a gradual increase in their size, creating stratification with larger sand grains at lower levels of the reactor and smaller sand grains at upper levels. Larger sand grains are discharged through the bottom of the reactor, and fresh sand grains are introduced. Detailed information on the process and operation concepts are described in previous studies (van Dijk & Wilms 1991; Harms & Bruce Robinson 1992; van Schagen et al. 2008b).

Studies of advanced water treatment systems used to remove OMPs have found advanced oxidation processes (e.g., O3/H2O2 and UV/H2O2) and high pressure membranes (e.g., nanofiltration (NF) and RO membranes) to be effective for certain contaminants of emerging concern (CEC) (Huber et al. 2003; Verliefde et al. 2009; Yangali-Quintanilla et al. 2010; Ramasundaram et al. 2013; Wang et al. 2015). However, questions still remain regarding the fate of OMPs in conventional groundwater or bank filtrate treatment processes. Central softening has been utilized in the Netherlands for more than 30 years, yet no literature is currently available on the removal of pharmaceuticals and estrogens. Conventional chemical softening processes are not designed for the removal of OMPs. However, it is important to determine the level of post-treatment required to compensate for limitations of the softening process for the removal of CECs and to investigate bulk organic matter characteristics in order to understand the effects of natural organic matter (NOM) on post-treatment steps for the removal of OMPs, since NOM is known as one of the prominent interfering substances and can form disinfection byproducts (DBPs). To the best of our knowledge, no study has investigated the removal of pharmaceuticals, the removal of estrogens, or NOM characteristics during pellet softening.

A pellet softening system is designed to remove calcium through the crystallization of calcium carbonate on the surface of pellets, not specifically for removing pharmaceuticals. Therefore, adequate hybridization with other advanced processes should be considered for drinking water treatment processes. The development of hybrid systems has enhanced the efficiency of OMP removal during water treatment (Sudhakaran et al. 2013). An integrated membrane system preceded by one or more non-membrane-based treatment processes (Schippers et al. 2004) can act as a multi-barrier system for removing both total hardness and OMPs.

NF is an effective treatment process to remove OMPs whose molecular weights range between 150 and 500 Da (Snyder et al. 2003), but the removal is dependent on the NF membrane properties, such as molecular weight cut-off (MWCO), pore size, surface charge, hydrophobicity–hydrophilicity, and surface roughness (Xu et al. 2005). Several previous studies reported the performance of NF for the removal of pharmaceuticals (Radjenović et al. 2008; Omidvar et al. 2015) and investigated influential factors during operation, such as pH, ionic strength, transmembrane pressure, NOM (Zazouli et al. 2009), and fouling (Yangali-Quintanilla et al. 2009). Like other advanced water treatments, NF is not always effective for the removal of all OMPs. Therefore, it is important to investigate physicochemical properties of target contaminants when water utilities select proper water treatment for these compounds.

To overcome the limitations of pellet softening for the removal of OMPs, NF is a synergistic approach for hybridization, not only to attenuate OMPs but also total hardness and membrane foulants (e.g., calcium sulfate scaling) in the NF system. In this study, an NF system was considered as a potential hybridization of pellet softening to overcome the low removal of OMPs during pellet softening.

This study determined the behavior of 12 selected pharmaceuticals and three endocrine-disrupting compounds, including lipid regulators, stimulants, anticonvulsants, analgesics, non-steroidal anti-inflammatory drugs, and estrogens, during pellet softening. NF was also conducted, and its performance was compared to pellet softening. Further, this study investigated the NOM characteristics during pellet softening using advanced organic matter characterization tools, which to the authors' best knowledge, have not been previously reported.

MATERIALS AND METHODS

Experimental setup

A laboratory scale pellet reactor was used with a plexiglass column with an internal diameter of 15 mm, a height of 3,000 mm, and a nozzle at the bottom for the injection of sodium hydroxide (NaOH, 1.5 N). Sand grains were used as seeding material to fill the column to a height of 700 mm. A similar configuration was reported previously (Mahvi et al. 2005). Different upflow velocities of 50, 100, and 130 m/h were tested to determine the effect of water velocity on the removal of total hardness. Groundwater was used from a vertical well located at Sejong University (Seoul, Korea), and calcium chloride (CaCl2; Sigma-Aldrich, Kyunggido, Korea) was used to increase total hardness concentration in the groundwater (Table 1).

Table 1

Characteristics of influent used in this study

pH Turbidity (NTU) Alkalinity as CaCO3 (mg/L) Dissolved organic carbon (mg/L) Total hardness as CaCO3 (mg/L) 
7.2–7.5 1.6 ± 0.5  4.5 ± 0.5 176 ± 11 
pH Turbidity (NTU) Alkalinity as CaCO3 (mg/L) Dissolved organic carbon (mg/L) Total hardness as CaCO3 (mg/L) 
7.2–7.5 1.6 ± 0.5  4.5 ± 0.5 176 ± 11 

An NF membrane system was used to compare the removal performance of pharmaceuticals and estrogens. The NF system consisted of nine parallel membrane pressure vessels composing a spiral-wound polyamide membrane (NE4040-90; Woongjin Chemical Co. Ltd, Korea) with a total effective area of 71.1 m2, and showed 90% rejection of sodium chloride (NaCl; molecular weight cut-off at approximately 200 Da).

Total hardness, alkalinity, and pH measurements

Total hardness was measured according to standard methods: the complexometric titration method with ethylenediaminetetraacetic acid (APHA 2012). The pH was measured using a pH meter (HI 8424; Hanna Instruments, USA).

Organic matter characteristics

A fluorescence excitation–emission matrix (EEM) was used to characterize the bulk organic matter characteristics with a fluorescence spectrophotometer (LS50B; PerkinElmer, USA). EEM spectra were obtained at scanned excitation wavelengths (ex) between 200 and 400 nm at 10-nm intervals and emission wavelengths (em) between 280 and 600 nm at 0.5-nm intervals. The water Raman data at 348 nm were regularly checked to confirm the stability of the lamp in the fluorescence spectrophotometer, and EEM fluorescence spectra were corrected using blank subtraction. The EEM depicted bulk organic matter characteristics with respect to four peaks, which were determined by fluorescence intensity at distinct ex and em wavelengths: tryptophan protein-like peaks, T1 (ex/em = 220–240 nm/330–360 nm), tryptophan protein-like peaks, T2 (ex/em = 270–280 nm/330–360 nm); humic-like peaks, A (ex/em = 230–260 nm/400–450 nm); and humic acid-like peaks, C (ex/em = 300–340 nm/400–450 nm). Liquid chromatography and organic carbon detection (LC-OCD; DOC-LABOR, Germany), with a detection limit for LC-OCD at <1–50 ppb, was used. A Gräntzel thin-film reactor was used in the LC-OCD system to oxidize dissolved organic matter, resulting in CO2 production, which was detected by a gas analyzer (Ultramat 6; Siemens, Germany). LC-OCD was used to characterize bulk organic matter into five organic matter fractions (biopolymers, humic substances, building blocks, low molecular weight (MW) acids, and low MW neutrals), all of which were determined with respect to the organic carbon concentration (C mg/L). A detailed description of LC-OCD has been reported elsewhere (Huber et al. 2011). Specific UV absorbance (SUVA), the ratio between UV absorbance and dissolved organic carbon, was used to investigate changes in the aromaticity of organic matter during pellet softening.

Pharmaceuticals and estrogens

Twelve pharmaceuticals (gemfibrozil, bezafibrate, clofibric acid, caffeine, carbamazepine, ibuprofen, naproxen, phenacetine, acetaminophen, fenoprofen, ketoprofen, and pentoxifylline) and three estrogens (17α-ethinylestradiol or EE2, estrone or E1, and 17β-estradiol or E2) were spiked into groundwater at a concentration of 2 μg/L. The selected pharmaceuticals were pre-concentrated by performing online solid phase extraction with an EQUAN MAX™ system (Thermo Fisher Scientific, CA, USA), based on column switching techniques using a trap column and the analytical columns hypersil gold aQ (20 mm, 12 μm particle size) and hypersil gold (50 mm, 1.9 μm particle size; Thermo Fisher Scientific). The sample delivery system comprised a CTC PAL auto sampler manufactured by CTC Analytics (Zwingen, Switzerland) with six-port switching valves and a quaternary load pump. The measurement of mass spectrometry was performed on an Orbitrap Exactive model (Thermo Fisher Scientific), using ESI mode. The limit of quantification (LOQ) for the selected pharmaceuticals was <30 ng/L. EE2, E1, and E2 were determined by gas chromatography-mass spectrometry (GC-2010; Shimadzu, Japan) using an HP-5 type capillary column. C18 SPE cartridges (HLB C18 cartridge 6 cc, 1 g; Oasis, USA) were used to enrich the estrogens. The LOQ for E2, E1, and EE2 was 30 ng/L. The procedure used for selected estrogens is described elsewhere (Liu et al. 2004).

Composition and surface characterization of pellets during pellet softening

Sand grains before and after pellet softening were characterized by X-ray fluorescence (XRF) (RIX2100; Rigaku, Japan) with an Rh source of 40 kV and 75 mA, X-ray powder diffraction (XRD; X-ray source: Cu source, 40 kV, 200 mA with Cu Kα radiation, λ = 1.5406 Å, Dmax2500, Rigaku, Japan), and scanning electron micrographs (JSM-6390; JEOL).

RESULTS AND DISCUSSION

Performance of pellet softening at different total hardness loadings and upflow velocities

A total hardness removal efficiency from 47 to 80% was observed when the pellet reactor was fed total hardness concentrations from 176 to 673 mg/L of calcium carbonate (CaCO3) at a pH of 9; efficiency increased when the total hardness concentration increased (Table 2). The removal efficiency of total hardness can be enhanced by increasing the pH, but our aim was a total groundwater hardness between 90 and 110 mg/L of CaCO3 after softening, so as to retain mineral nutrition in water. Upflow velocities of 50, 100, and 130 m/h were tested to investigate the effect of velocity on the removal of total hardness. Chen et al. (2000) reported good formation of calcium at a pH of 9 and an upflow velocity of 100 m/h. However, our study found no significant difference in total hardness removal at upflow velocities of 50, 100, and 130 m/h, and an upflow velocity over 130 m/h was not considered due to the fine particles discharging from the pellet reactor.

Table 2

Removal of total hardness by pellet softening at different total hardness and upflow velocities (n = 4)

Total hardness (mg/L as CaCO3)
 
Removal % Upflow velocity (m/h) Total hardness (mg/L as CaCO3)
 
Removal % 
Influent Effluent Influent Effluent 
176 ± 11 93 ± 31 47 50 510 ± 23 93 ± 20 82 
284 ± 9 100 ± 27 65 100 510 ± 23 93 ± 18 82 
522 ± 47 109 ± 53 79 130 510 ± 23 102 ± 34 80 
673 ± 15 134 ± 75 80     
Total hardness (mg/L as CaCO3)
 
Removal % Upflow velocity (m/h) Total hardness (mg/L as CaCO3)
 
Removal % 
Influent Effluent Influent Effluent 
176 ± 11 93 ± 31 47 50 510 ± 23 93 ± 20 82 
284 ± 9 100 ± 27 65 100 510 ± 23 93 ± 18 82 
522 ± 47 109 ± 53 79 130 510 ± 23 102 ± 34 80 
673 ± 15 134 ± 75 80     

The crystallization of calcium on sand grains was confirmed by XRF analysis (Table 3). The chemical composition of sand grains before pellet softening was dominated by silicon dioxide (SiO2) at 74.3%, followed by aluminum oxide (Al2O3) at 14.4%, and calcium oxide (CaO) at 1.27%, which is the typical composition of silica sand. The composition changed dramatically after softening to 19.1% SiO2, 3.5% Al2O3, and 73.3% CaO, which indicated that calcium was successfully crystallized and dominated the surface of sand grains. Figure 1 shows scanning electron microscope images of sand grains before and after softening at an upflow velocity of 100 m/h, and the results obtained from XRD analysis that confirmed the crystallization of calcium are shown in Figure 2. XRD patterns of sand grains showed that calcium peaks appeared only after softening. Results of XRD and XRF analyses confirmed the formation of CaCO3 on the surface of sand grains through crystallization.
Table 3

Chemical composition of sand (before and after softening) (wt.%) at total hardness 500 mg/L as CaCO3

Components Before pellet softening, % After pellet softening, % 
Na22.32 0.551 
MgO 0.827 0.596 
Al2O3 14.4 3.5 
SiO2 74.3 19.1 
P2O5 0.056 0.026 
SO3 0.015 0.075 
K25.03 1.27 
CaO 1.27 73.3 
TiO2 0.178 0.122 
Fe2O3 1.47 1.2 
NiO 0.005 0.05 
ZrO2 0.004 0.015 
SrO 0.017 0.068 
PbO 0.003 0.005 
ZnO 0.004 0.061 
Rb20.011 0.010 
MnO 0.023 0.043 
CuO 0.001 0.015 
Nb2O5 0.001 – 
BaO 0.079 – 
Components Before pellet softening, % After pellet softening, % 
Na22.32 0.551 
MgO 0.827 0.596 
Al2O3 14.4 3.5 
SiO2 74.3 19.1 
P2O5 0.056 0.026 
SO3 0.015 0.075 
K25.03 1.27 
CaO 1.27 73.3 
TiO2 0.178 0.122 
Fe2O3 1.47 1.2 
NiO 0.005 0.05 
ZrO2 0.004 0.015 
SrO 0.017 0.068 
PbO 0.003 0.005 
ZnO 0.004 0.061 
Rb20.011 0.010 
MnO 0.023 0.043 
CuO 0.001 0.015 
Nb2O5 0.001 – 
BaO 0.079 – 
Figure 1

Scanning electron microscope images showing pellet at different scale bars 500, 100, and 10 μm; before softening (a, c, and e) and after softening (b, d, and f) (total hardness 673 mg/L as CaCO3 at pH 9 and upflow velocity 100 m/h).

Figure 1

Scanning electron microscope images showing pellet at different scale bars 500, 100, and 10 μm; before softening (a, c, and e) and after softening (b, d, and f) (total hardness 673 mg/L as CaCO3 at pH 9 and upflow velocity 100 m/h).

Figure 2

X-ray diffraction patterns of pellet before (a) and after (b) pellet softening (total hardness 673 mg/L as CaCO3 at pH 9 and upflow velocity 100 m/h).

Figure 2

X-ray diffraction patterns of pellet before (a) and after (b) pellet softening (total hardness 673 mg/L as CaCO3 at pH 9 and upflow velocity 100 m/h).

Changes in bulk organic matter characteristics

Advanced organic matter characterization tools such as EEM, LC-OCD, and SUVA were utilized to gain a better understanding of changes in bulk organic matter characteristics during groundwater softening via pellet softening. The characteristics of NOM are important to investigate in water treatment systems where the disinfection is carried out via chlorine. NOM with more aromatic organic matter tends to form higher level form DBPs than NOM with a low content of aromatic organic matter (US EPA 2006). Russell et al. (2009) reported that the removal of NOM was observed during water softening through adsorption onto or co-precipitation with calcium and magnesium. However, no previous study has reported the changes in bulk organic matter characteristics during water softening using EEM and LC-OCD. Based on the EEM results, there was a reduction of fluorescence intensity observed in the humic-like peaks (C and A peaks), and the tryptophan-like peaks (T) were unchanged after pellet softening (Figure 3). The groundwater was contaminated by the pond located near the groundwater well; high concentrations of dissolved organic carbon and a tryptophan-like peak in the groundwater were detected. The dissolved organic carbon concentration commonly detected in the groundwater was between 0.5 and 1.0 mg/L; however, the groundwater was contaminated because the pond was contaminated by food waste during the time of this study. Baker et al. (2004) reported that wastewater or treated wastewater showed the highest tryptophan-like fluorescence intensity compared to waters that were less impacted by wastewater, and tryptophan-like peaks can be used as an indicator of organic pollution, along with other traditional water quality parameters, such as dissolved oxygen, pH, and turbidity (Khamis & Stevens 2013). Previous studies reported the preferential removal of aromatic organic matter during chemical softening (Liao & Randtke 1986; Thompson et al. 1997; Roalson et al. 2003). We also confirmed by EEM that humic-like substances (C and A), which represent aromatic organic matter characteristics, were preferentially attenuated during pellet softening. At a higher pH, humic acid in humic-like substances becomes more soluble, and more negative charges are generated in the acidic functional groups of humic substances (Uyguner-Demirel & Bekbolet 2011). Negative charges of humic substances could adsorb onto a positively charged calcium carbonate that crystallized on the surface of sand grains during pellet softening. However, tryptophan-like substances were not effectively attenuated because of the high pKa value of α-amino group (9.4), indicating less negative charge compared to that of humic-like substances.
Figure 3

Fluorescence EEM spectra of dissolved organic matter in groundwater before (a) and after (b) pellet softening.

Figure 3

Fluorescence EEM spectra of dissolved organic matter in groundwater before (a) and after (b) pellet softening.

LC-OCD analysis grouped bulk organic matter into five different organic matter fractions (biopolymers, humics, building blocks, low MW acids, and low MW neutrals) before and after pellet softening (Figure 4). The humic fraction determined by LC-OCD (humics 800–1,000 Da) was reduced slightly greater than the other four fractions, and low MW acids in very low concentrations were not attenuated. As expected, biopolymers, which were also detected as tryptophan-like substances in EEM, were not changed. SUVA values decreased from 5.7 to 3.4, 4.1, and 3.9 when initial total hardness concentrations were 673, 284, and 522 mg/L, respectively. As such, more positively charged metal ions (e.g., calcium ions) are available for binding negatively charged aromatic organic matter. Based on advanced organic matter characterization tools, a slightly higher removal was observed for humic-like substances during pellet softening.
Figure 4

Bulk organic matter fractions (biopolymers, humics, building blocks, low MW acids, and low MW neutrals) determined by LC-OCD at different total hardness concentrations during pellet softening (284, 522, and 673 mg/L as CaCO3).

Figure 4

Bulk organic matter fractions (biopolymers, humics, building blocks, low MW acids, and low MW neutrals) determined by LC-OCD at different total hardness concentrations during pellet softening (284, 522, and 673 mg/L as CaCO3).

Fate of pharmaceuticals and endocrine disrupting compounds

The selected pharmaceuticals were grouped into four categories according to usage in order to determine similarities in removal during pellet softening (lipid regulator, stimulants, analgesic, and nonsteroidal anti-inflammatory drugs). The 12 selected pharmaceuticals exhibited very low removal patterns and showed no similarities by group. Furthermore, there was no significant change in the removal of pharmaceuticals when the total hardness increased from 284 to 673 mg/L. Neither the degree of total hardness removal nor the initial total hardness concentration during pellet softening had any effect on the removal of the selected pharmaceuticals via crystallization or co-precipitation. The concentration of the selected pharmaceuticals was lower than the concentration of NOM in the groundwater, and most of the pharmaceuticals were of the hydrophilic and ionic species during pellet softening (pH of 9). The average removal efficiencies of the four different groups of pharmaceuticals during pellet softening were 4.4, 5.8, and 6.8% for total hardness concentrations of 284, 522, and 673 mg/L, respectively, indicating that pellet softening is ineffective for removing the selected pharmaceuticals. Previous studies reported similar results, in which chemical lime softening was ineffective (<25%) for the removal of pharmaceutical and personal care compounds (Westerhoff et al. 2005), and coagulation and lime softening processes were ineffective for the removal of antibiotics (Adams et al. 2002). This study found that pellet softening showed a similar performance to a conventional chemical softening process, and the calcium that crystallized on the surface of the sand grains had no influence on the removal of the selected pharmaceuticals. Many of the selected pharmaceuticals were not effectively removed (<7%) during pellet softening, but the removal efficiency of pharmaceuticals by NF (NaCl 90% and a MWCO of ∼200 Da) was significantly higher (Figure 5). For example, bezafibrate (MW: 361) was significantly rejected (>95%), which is at least five times higher than the rate of removal by pellet softening; also, the NF removal efficiencies of gemfibrozil (MW: 250), fenoprofen (MW: 242), ibuprofen (MW: 206), and carbamazepine (MW: 236) were 82, 80, 77, and 67%, respectively. NF rejected compounds with MWs greater than 300, primarily by steric hindrance, and did not completely reject compounds with a MW between 200 Da and 250 Da. Rejection of ionic compounds such as gemfibrozil (MW: 250), fenoprofen (MW: 242), and ibuprofen (MW: 206) was slightly higher than rejection of a neutral compound (carbamazepine, MW: 236), which may be attributed to electrostatic interactions (repulsion) between the charge of anionic pharmaceuticals and the negative charge of the NF membrane surface. The hybridization of pellet softening with NF may be a good multi-barrier approach for improving the performance of pellet softening and NF in drinking water treatment systems. In this study, the hardness removal by pellet softening was only 80% at the optimum conditions; therefore, about 20% of the calcium after softening can complex with anionic pharmaceuticals. Thus, the removal of anionic compounds that are not effectively rejected by NF can be improved by complexing with the remaining calcium, since NF is effective in the removal of divalent ions such as calcium and magnesium. However, it is important to consider the pretreatment process for organic matter when pellet softening is hybridized with NF. Pellet softening is not effective for the removal of organic matter; therefore, the dissolved organic carbon concentration like that used in this study would lead to serious organic fouling on the surface of the NF membrane.
Figure 5

Pharmaceutical removal by pellet softening or NF.

Figure 5

Pharmaceutical removal by pellet softening or NF.

For estrogens, the removal efficiencies of E1, E2, and EE2 at a total hardness of 522 mg/L of CaCO3 and an upflow velocity of 100 m/h were 59 ± 12, 60 ± 21, and 57 ± 11%, respectively (Figure 6). The higher attenuation of estrogens may be due to hydrophobic characteristics (log Kow E1: 3.43, E2: 3.94, and EE2: 4.15) via adsorption onto sand grains. Estrogens with a log Kow between 3.43 and 4.15 are preferred for adsorption onto solids, and previous studies reported that adsorption was the dominant mechanism of estrogen removal (17β-estradiol, estriol, and testosterone) during soil passage (Mansell et al. 2004; Mansell & Drewes 2004; Maeng et al. 2013). Westerhoff et al. (2005) reported that neutral hydrophobic compounds, based on log Kow in 22 different endocrine-disrupting compounds and pharmaceutical personal care products, showed a relatively higher removal efficiency compared to compounds with low log Kow values during lime softening, resulting from their sorption onto small particles and precipitated solids. However, bezafibrate, with a log Kow of 4.25 and hydrophobic characteristics similar to estrogens, showed a low reduction during pellet softening. Bezafibrate becomes an ionic compound (pKa: 3.6) during pellet softening at a pH of 9, and the low reduction may be due to the electrostatic interaction between the negatively charged surface of sand grains and the anion in bezafibrate. Moreover, the occurrence of interactions between NOM in groundwater and the spiked estrogens should not be neglected since dissolved organic matter molecular composition and concentration affect the interactions of pharmaceuticals and personal care products in water (Hernandez-Ruiz et al. 2012).
Figure 6

Removal of selected estrogens (17α-ethinylestradiol (E1), estrone (E2), and 17β-estradiol (EE2)) by pellet softening (total hardness of 522 mg/L as CaCO3, flow velocity: 100 m/h).

Figure 6

Removal of selected estrogens (17α-ethinylestradiol (E1), estrone (E2), and 17β-estradiol (EE2)) by pellet softening (total hardness of 522 mg/L as CaCO3, flow velocity: 100 m/h).

Some reduction was observed in the selected estrogens during pellet softening. However, the activity of estrogens needs to be investigated to confirm complete removal because transformation products can retain estrogenic activity. Hammes et al. (2011) detected the bacterial colonization of pellets in a full-scale plant (Leiduin, The Netherlands) for the first time, and characterized the biomass on calcite pellets using adenosine triphosphate and denaturing gradient gel electrophoresis analysis. Further study is necessary to investigate any possible degradation of adsorbed estrogens onto sand grains and the desorption capacity of estrogens.

CONCLUSION

The removal of total hardness via crystallization with selected pharmaceuticals and estrogens during pellet softening was investigated, and the main conclusions are as follows:

  • The removal efficiencies of total hardness were 47, 65, 79, and 80% for groundwater with an initial total hardness of 176, 284, 522, and 673 mg/L of CaCO3 at a pH of 9, respectively. The chemical compositions determined by XRD and XRF confirmed that the crystallization of calcium successfully occurred on the surface of sand grains.

  • EEM showed that humic-like substances were preferentially removed during pellet softening, and tryptophan protein-like peaks were unchanged. The humic fraction determined by LC-OCD was slightly more attenuated than the fractions of biopolymers, building blocks, low MW acids, and low MW neutrals. SUVA values decreased, indicating that a lower concentration of aromatic organic matter remained after pellet softening. The negatively charged aromatic organic matter could adsorb onto the positively charged calcium carbonate that crystallized on the surface of sand grains during pellet softening.

  • In general, most of the 12 selected pharmaceuticals were not effectively attenuated during pellet softening, but a relatively higher amount of estrogens were removed compared to pharmaceuticals. It is expected that the hydrophobic characteristics of neutral compounds such as E1, E2, and EE2 led to higher removal during pellet softening and were considered to be important factors.

  • The removal efficiencies of bezafibrate, gemfibrozil, fenoprofen, ibuprofen, and carbamazepine by NF were 99, 82, 80, 77, and 67%, respectively, and were significantly higher than removal efficiencies by pellet softening. The rejection of ionic compounds such as gemfibrozil, fenoprofen, and ibuprofen was slightly higher than the rejection of a neutral compound (carbamazepine), which may be attributed to the electrostatic interactions between the negative charge of anionic pharmaceuticals and negative charge of the NF membrane surface.

ACKNOWLEDGEMENTS

This study was supported by the Korea Ministry of Environment as a part of ‘The Eco-Innovation project’ (Global Top Project) under Grant GT-SWS-11-01-006-00 and the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (2014R1A1A1037670).

REFERENCES

REFERENCES
APHA
2012
Standard Methods for the Examination of Water and Wastewater
,
22nd edn
.
American Public Health Association
,
Washington, DC
,
USA
.
Baker
A.
Ward
D.
Lieten
S. H.
Periera
R.
Simpson
E. C.
Slater
M.
2004
Measurement of protein-like fluorescence in river and waste water using a handheld spectrophotometer
.
Water Res.
38
,
2934
2938
.
Benotti
M. J.
Trenholm
R. A.
Vanderford
B. J.
Holady
J. C.
Stanford
B. D.
Snyder
S. A.
2009
Pharmaceuticals and endocrine disrupting compounds in U.S. drinking water
.
Environ. Sci. Technol.
43
(
3
),
597
603
.
Chen
Y.-H.
Yeh
H.-H.
Tsai
M.-C.
Lai
W. L.
2000
The application of fluidized bed crystallization in drinking water softening
.
J. Chin. Inst. Environ. Eng.
10
(
3
),
177
184
.
Dębska
J.
Kot-Wasik
A.
Namieśnik
J.
2004
Fate and analysis of pharmaceutical residues in the aquatic environment
.
Cri. Rev. Anal. Chem.
34
(
1
),
51
67
.
Graveland
A.
Van Dijk
J. C.
De Moel
P. J.
Oomen
J. H. C. M.
1983
Developments in water softening by means of pellet reactors
.
J. Am. Water Works Assoc.
75
(
12
),
619
625
.
Hammes
F.
Boon
N.
Vital
M.
Ross
P.
Magic-Knezev
A.
Dignum
M.
2011
Bacterial colonization of pellet softening reactors used during drinking water treatment
.
Appl. Environ. Microb.
77
(
3
),
1041
1048
.
Harms
W.
Bruce Robinson
R.
1992
Softening by fluidized bed crystallizers
.
J. Environ. Eng.
118
(
4
),
513
529
.
Heberer
T.
Reddersen
K.
Mechlinski
A.
2002
From municipal sewage to drinking water: Fate and removal of pharmaceutical residues in the aquatic environment in urban areas
.
Water Sci. Technol.
46
(
3
),
81
88
.
Huber
M. M.
Canonica
S.
Park
G.-Y.
Gunten
U. V.
2003
Oxidation of pharmaceuticals during ozonation and advanced oxidation processes
.
Environ. Sci. Technol.
37
(
5
),
1016
1024
.
Jain
S.
Kumar
P.
Vyas
R. K.
Pandit
P.
Dalai
A. K.
2013
Occurrence and removal of antiviral drugs in environment: a review
.
Water Air Soil Pollut.
224
,
1410
.
Khamis
K.
Stevens
R.
2013
The use of tryptophan-like fluorescence as an indicator of organic pollution
.
Available from: http://www.envirotech-online.com/. November/December, 27–28
.
Maeng
S. K.
Shrama
S. K.
Lee
J. W.
Amy
G.
2013
Fate of 17β-estradiol and 17α-ethinylestradiol in batch and column studies simulating managed aquifer recharge
.
J. Water Supply Res. Technol.-AQUA
62
(
7
),
409
416
.
Mahvi
A. H.
Shafiee
F.
Naddafi
K.
2005
Feasibility study of crystallization process for water softening in a pellet reactor
.
Int. J. Environ. Sci. Technol.
1
(
4
),
301
304
.
Mansell
J.
Drewes
J. E.
2004
Fate of steroidal hormones during soil-aquifer treatment
.
Ground Water Monit. R.
24
(
2
),
94
101
.
Mansell
J.
Drewes
J. E.
Rauch
T.
2004
Removal mechanisms of endocrine disrupting compounds (steroids) during soil aquifer treatment
.
Water Sci. Technol.
50
(
2
),
229
237
.
Morgan
C.
Bevington
C.
Levin
D.
Robinson
P.
Davis
P.
Abbott
J.
Simkins
P.
2013
Water Sensitive Urban Design in the UK
.
Ideas for built environment practitioners
.
CIRIA report C723
.
Omidvar
M.
Soltanieh
M.
Mousavi
S. M.
Saljoughi
E.
Moarefian
A.
Saffaran
H.
2015
Preparation of hydrophilic nanofiltration membranes for removal of pharmaceuticals from water
.
J. Environ. Health Sci. Eng.
13
,
13
42
.
Ramasundaram
S.
Yoo
H. N.
Song
K. G.
Lee
J.
Choi
K. J.
Hong
S. W.
2013
Titanium dioxide nanofibers integrated stainless steel filter for photocatalytic degradation of pharmaceutical compounds
.
J. Hazard. Mater.
258–259
(
15
),
124
132
.
Richardson
S. D.
2003
Disinfection by-products and other emerging contaminants in drinking water
.
TrAC Trends Anal. Chem.
22
(
10
),
666
684
.
Roalson
S. R.
Kweon
J.
Lawler
D. F.
Speitel
G. E.
2003
Enhanced softening: effect of lime dose and chemical addition
.
J. Am. Water Works Assoc.
95
(
11
),
97
109
.
Russell
C. G.
Lawler
D. F.
Speitel
G. E.
Katz
L. E.
2009
Effect of softening precipitate composition and surface characteristics on natural organic matter adsorption
.
Environ. Sci. Technol.
43
(
20
),
7837
7842
.
Schippers
J. C.
Kruithof
J. C.
Nederlof
M. M.
Hofman
J. A. M. H.
2004
Integrated membrane systems, AWWA Research Foundation
.
Report 990899
,
AWWA
,
Denver, CO
,
USA
.
Stackelberg
P. E.
Furlong
E. T.
Meyer
M. T.
Zaugg
S. D.
Henderson
A. K.
Reissman
D. B.
2004
Persistence of pharmaceutical compounds and other organic wastewater contaminants in a conventional drinking-water-treatment plant
.
Sci. Total Environ.
329
(
1–3
),
99
113
.
Ternes
T. A.
Stumpf
M.
Mueller
J.
Haberer
K.
Wilken
R. D.
Servos
M.
1999
Behavior and occurrence of estrogens in municipal sewage treatment plants-I. Investigations in Germany, Canada and Brazil
.
Sci. Total Environ.
225
(
1–2
),
81
90
.
Ternes
T. A.
Meisenheimer
M.
Mcdowell
D.
Sacher
F.
Brauch
H.-J.
Haist-Gulde
B.
Preuss
G.
Wilme
U.
Zulei-Seibert
N.
2002
Removal of pharmaceuticals during drinking water treatment
.
Environ. Sci. Technol.
36
(
17
),
3855
3863
.
Thompson
J. D.
White
M. C.
Harrington
G. W.
Singer
P. C.
1997
Enhanced softening: factors influencing DBP precursor removal
.
J. Am. Water Works Assoc.
89
(
6
),
94
105
.
US EPA
2006
For the final stage 2 disinfectants and disinfection byproducts rule. Initial distribution system evaluation guidance manual, EPA 815-B-06-002
.
Van der Bruggen
B.
Goossens
H.
Everard
P. A.
Stemgée
K.
Rogge
W.
2009
Cost-benefit analysis of central softening for production of drinking water
.
J. Environ. Manage.
91
(
2
),
541
549
.
van Dijk
J. C.
Wilms
D.
1991
Water treatment without waste material – fundamentals and state of the art of pellet softening
.
J. Water Supply Res. Technol. AQUA
40
(
5
),
263
280
.
van Houwelingen
G.
Bond
R.
Seacord
T.
Fessler
E.
2010
Experiences with pellet reactor softening as pretreatment for inland desalination in the USA
.
Desalin. Water Treat.
13
(
1–3
),
259
266
.
van Schagen
K. M.
Rietveld
L. C.
Babuška
R.
Baars
E.
2008a
Control of the fluidised bed in the pellet softening process
.
Chem. Eng. Sci.
63
(
5
),
1390
1400
.
van Schagen
K. M.
Rietveld
L. C.
Babuška
R.
Kramer
O. J. I.
2008b
Model-based operational constraints for fluidised bed crystallisation
.
Water Res.
42
(
1–2
),
327
337
.
Verliefde
A. R. D.
Cornelissen
E. R.
Heijman
S. G. J.
Petrinic
I.
Luxbacher
T.
Amy
G. L.
Van Der Bruggen
B.
van Dijk
J. C.
2009
Influence of membrane fouling by (pretreated) surface water on rejection of pharmaceutically active compounds (PhACs) by nanofiltration membranes
.
J. Membr. Sci.
330
(
1–2
),
90
103
.
Xu
P.
Drewes
J. E.
Bellona
C.
Amy
G.
Kim
T.-U.
Adam
M.
Heberer
T.
2005
Rejection of emerging organic micropollutants in nanofiltration-reverse osmosis membrane applications
.
Water Environ. Res.
77
(
1
),
40
48
.