Journal of Environmental Management 246 (2019) 920–928

Contents lists available at ScienceDirect

Journal of Environmental Management

journal homepage: www.elsevier.com/locate/jenvman

Research article

Evaluation of five filter media in column experiment on the removal of T

selected organic micropollutants and phosphorus from household
wastewater
Wen Zhanga,∗, Pablo Gago-Ferrerob, Qiuju Gaoc, Lutz Ahrensb, Kristin Blumc, Ande Rostvallb,
Berndt Björleniusd, Patrik L. Anderssonc, Karin Wibergb, Peter Haglundc, Gunno Renmana
a
Dept. of Sustainable Development, Environmental Science and Engineering, KTH Royal Institute of Technology, Teknikringen 10B, SE-10044, Stockholm, Sweden
b
Dept. of Aquatic Sciences and Assessment, Swedish University of Agricultural Sciences, Box 7050, SE-75007, Uppsala, Sweden
c
Dept. of Chemistry, Umeå University, Linnaeus väg 6, SE-90187, Umeå, Sweden
d
Dept. of Biotechnology, KTH Royal Institute of Technology, AlbaNova University Centre, SE-10691, Stockholm, Sweden

A R T I C LE I N FO A B S T R A C T

Keywords: A bench-scale column experiment was performed to study the removal of 31 selected organic micropollutants
Micropollutant (MPs) and phosphorus by lignite, xyloid lignite (Xylit), granular activated carbon (GAC), Polonite® and sand over
Phosphorus a period of 12 weeks. In total 29 out of the 31 MPs showed removal efficiency > 90% by GAC with an average
Sorbent removal of 97 ± 6%. Xylit and lignite were less efficient with an average removal of 80 ± 28% and
Surface functional group
68 ± 29%, respectively. The removal efficiency was found to be impacted by the characterization of the sor-
Physicochemical property
bents and physicochemical properties of the compounds, as well as the interaction between the sorbents and
Removal mechanism
compounds. For instance, Xylit and lignite performed well for relatively hydrophobic (log octanol/water par-
tition coefficient (Kow) ≥3) MPs, while the removal efficiency of moderately hydrophilic, highly hydrophilic and
negatively charged MPs were lower. The organic sorbents were found to have more functional groups at their
surfaces, which might explain the higher adsorption of MPs to these sorbents. The removal of several MPs
improved after four weeks in sand, Xylit, GAC and lignite which may be related to increased biological activity
and biofilm development. GAC and sand had limited ability to remove phosphorus (12 ± 27% and 14 ± 2%,
respectively), while the calcium-silicate material Polonite® precipitated phosphorus efficiently and increased the
total phosphorus removal from 12% to 96% after the GAC filter.

1. Introduction wastewater treatment plants (WWTPs) and OSSFs, which may threaten
ecosystem health as well as drinking water sources (Tröger et al., 2018).
Effluents from on-site sewage facilities (OSSFs) normally contain a Wastewater treatment techniques are traditionally optimized for
wide range of micropollutants (MPs), such as per- and polyfluoroalkyl removal of nutrients (e.g. phosphorus and nitrogen) and pathogens,
substances (PFASs), pesticides, pharmaceuticals, personal care pro- whereas potentially toxic MPs may be poorly or insufficiently treated
ducts, biocides, organophosphorus flame retardants, plasticisers, rubber (Blum et al., 2017; Gros et al., 2017). Advanced treatment methods are
additives and food additives (Blum et al., 2017; Gros et al., 2017). being developed for the removal of MPs in WWTPs (Kårelid et al.,
Several MPs have been found to be poorly removed by OSSFs, for in- 2017); however, less attention has been brought to corresponding
stance pesticides (e.g. dichlorobenzamide (BAM) and terbutryn) and techniques for OSSFs. Soil-based systems are one of the most widely
organophosphorus flame retardants (e.g. tributylphosphate and tris(2- applied treatment methods for OSSFs. Adsorption and biodegradation
chloro-ethyl)phosphate) (Blum et al., 2017; Gros et al., 2017). Varying are important processes for MP removal in soil-based systems as mi-
concentrations of MPs have been identified in recipient surface water croorganisms present in wastewater can biodegrade some MPs (Bester
(Fisher et al., 2016; Gago-Ferrero et al., 2017; Tröger et al., 2018) and et al., 2011; Dalahmeh et al., 2018). Moreover, biofilm may be formed
groundwater (Kreuzinger et al., 2004) originating from large-scale after a period of operation, which may stimulate degradation and

∗
Corresponding author.
E-mail addresses: zhangw@kth.se (W. Zhang), pablo.gago.ferrero@gmail.com (P. Gago-Ferrero), qiuju.gao@ri.se (Q. Gao), lutz.ahrens@slu.se (L. Ahrens),
kristin.maria.blum@gmail.com (K. Blum), ande.rostvall@outlook.com (A. Rostvall), berndtb@kth.se (B. Björlenius), patrik.andersson@umu.se (P.L. Andersson),
Karin.wiberg@slu.se (K. Wiberg), peter.haglund@umu.se (P. Haglund), gunno@kth.se (G. Renman).

https://doi.org/10.1016/j.jenvman.2019.05.137
Received 13 December 2018; Received in revised form 24 April 2019; Accepted 28 May 2019
Available online 04 July 2019
0301-4797/ © 2019 Elsevier Ltd. All rights reserved.
W. Zhang, et al. Journal of Environmental Management 246 (2019) 920–928

additionally change the texture of the sorbent and its surface charge

Mátrai Erömü, Bükkábrány, Hungary

and thus improve adsorption of certain cation contaminants (Méndez-

Ecofiltration Nordic AB, Sweden

Chemviron Carbon AB, Sweden
Díaz et al., 2012). The physicochemical properties of MPs have been
found to be correlated to their removal efficiency (RE) on different
sorbents (Rostvall et al., 2018), e.g. there is a positive relationship

Rådasand AB, Sweden

Eloy Water, Belgium
between the hydrophobicity of the MPs and their adsorption rate as
shown in both laboratory scale experiment and field studies (Stevens-
Garmon et al., 2011; Blum et al., 2017). However, interactions between

Supplier
the surface functional groups of the sorbents and sorbates strongly in-
fluence the binding mechanism, which in turn may be dependent on the
pH and ionic strength (Stevens-Garmon et al., 2011; Anumol et al.,
2015). The charge of the sorbent can affect the adsorption of ionizable
MPs, e.g. a positively charged surface will tend to attract negatively
charged compounds and vice versa (Martínez-Hernández et al., 2014).

Average pore size (nm)

The removal of nutrients by soil-based system may show high variation
(Kholoma et al., 2016; Kauppinen et al., 2014). Phosphorus, which is an
important parameter of eutrophication, cannot be well removed by the
traditional soil-based systems and therefore needs to be enhanced by
other technologies (Kholoma et al., 2016). Therefore, nutrients removal

15c
17c
17c
23c
2.2
by soil-based system should be taken in consideration when striving for
improved removal of MPs removal.
The aim of the present study was to compare the performance of
GAC, Xylit, lignite, Rådasand and Polonite® in their removal of 31 MPs

Pore volume (cm3 g−1)

(covering a wide range of physicochemical properties) and nutrients in
wastewater from OSSFs during a 12 week long-term column experi-
ment. In a previous study, we found that the coal-based organic sor-
bents achieved better performance in removal of a set of 19 MPs than
the natural wood fibre and the inorganic sorbents (Zhang et al., 2018).

0.020c
0.010c
0.002c
0.022c
c
However, GACs and xyloid lignite were not efficient in phosphorus 0.51
removal (Zhang et al., 2018. A dual-column system combining GAC and
Polonite® was therefore tested in the present study. The ultimate goal
was to identify possible removal mechanisms of MPs and phosphorus
Surface area (m2 g−1)

and find a suitable material or combination of materials with high ca-

pacity to remove a multitude of MPs in wastewater from OSSFs while
simultaneously also removing phosphorus.

2. Material and methods

a

5.3c
2.5c
0.6c
3.8c
950

2.1. Sorbents

Three coal-based organic sorbents (Envirocarb™ 207 EA, Xylit, lig-

Particle size (mm)

nite) and two inorganic sorbents (Rådasand, Polonite®) were tested in

this study (Table 1). Envirocarb™ 207 EA is a GAC manufactured from
bituminous coal. Xylit is xyloid lignite, which is a commercial product
0.7–1.0a

marketed as a sorbent for the removal of organic matters from waste-

2–4b
a

2–6a
3–4

20b

water. The lignite used in this experiment was obtained from a mining
and power plant. Rådasand is a sieved natural sand excavated from the
Råda esker in Sweden. Polonite® is calcium-rich material that is used in
OSSFs to remove and recycle phosphorus.
Natural wood fibres derived from lignite
Bituminous coal; Envirocarb™ 207 EA

2.2. Target compounds

Filter materials used in the column experiment.

Rådasand: Quartz and feldspar

The 31 MPs evaluated in this study comprised: an artificial sweet-

Calcium silicate bedrock

ener, biocides/pesticides (n = 2), fragrances (n = 4), organophosphates

Data obtained from Gamage (2017).
Measured in the laboratory at KTH.

(n = 5), perfluoroalkyl substances (PFASs) (n = 3), pharmaceuticals

(n = 9), a plasticiser, a polymer impurity, a preservative, a rubber ad-
ditive, a surfactant and UV stabilisers (n = 2). The selected compounds
Brown coal

cover a wide range of chemical classes and physicochemical properties

Material

Provided by supplier.

(Table 2).

2.3. Column experiment

The column experiment was carried out over a 12-week period in a

Polonite®
Sorbent

dark, temperature-controlled room at 15 ± 3 °C. Stainless steel col-

Lignite
Table 1

Sand
Xylit
GAC

umns (diameter 5 cm) were filled with a 50 cm layer of sorbent: GAC,

b
a

Xylit, lignite, sand or Polonite. The base of the columns was fitted with

921
W. Zhang, et al. Journal of Environmental Management 246 (2019) 920–928

Table 2
Chemical classes, names and abbreviations of the 31 MPs analysed in this study.
Class Analyte Abbreviation Formula log Kow pKa

Artificial sweetener Sucralose SL C12H19Cl3O8 −1.0ᵃ 11.8ᵃ

Biocides/pesticides Hexachlorobenzene HCB C6Cl6 5.9ᵇ N/A
N,N-diethyl-m-toluamide DEET C12H17NO 6.3ᵇ N/A
Fragrances Galaxolide HHCB C18H26O 4.5ᵇ N/A
Musk xylene MX C12H15N3O6 4.3ᵇ N/A
Musk ketone MK C14H18N2O5 6.3ᵇ N/A
Tonalide AHTN C18H26O 3.8ᵇ N/A
Organophosphates Tributylphosphate TBP C12H27O4P 3.6ᵇ N/A
Tris(1,3-dichloro-2-propyl)phosphate TDCPP C9H15Cl6O4P 4.7ᵇ N/A
Triphenylphosphate TPP C18H15O4P 3.0ᵇ N/A
Tris(2-butoxyethyl)phosphate TBEP C18H39O7P 1.6ᵇ N/A
Tris(2-chloro-ethyl)phosphate TCEP C9H18Cl3O4P 7.0ᵃ −3.27ᵃ
Perfluoroalkyl Perfluorooctanesulfonate PFOS C8HF17O3S 7.6ᵃ 6.52ᵃ
substances (PFAS) Perfluorobutanesulfonate PFBS C4HF9O3S 2.4ᵃ 0.14ᵃ
Perfluorooctanesulfonamide FOSA C8H2F17NO2S 2.3ᵃ 0.67ᵃ
Pharmaceuticals Acetaminophen APAP C8H9NO2 0.46ᵃ 9.40ᵃ
Caffeine CF C8H10N4O2 −0.07ᵃ 10.40ᵃ
Carbamazepine CBZ C15H12N2O 2.3ᵃ 4.17ᵃ
Diclofenac DF C14H11Cl2NO2 4.0ᵃ 4.20ᵃ
Ibuprofen IP C13H18O2 3.8ᵃ 4.40ᵃ
Lamotrigine LAT C9H7Cl2N5 1.0ᵃ 5.70ᵃ
Losartan LOT C22H23ClN6O 4.0ᵃ 4.10ᵃ
Metoprolol MEP C15H25NO3 1.7ᵃ 9.70ᵃ
Oxazepam OP C15H11ClN2O2 3.3ᵃ 1.70ᵃ
Plasticiser N-butylbenzenesulfonamide NBBS C10H15NO2S 2.3ᵇ N/A
Polymer impurity Bisphenol A BPA C15H16O2 3.6ᵇ 9.6ᶜ
Preservative Propylparaben PP C10H12O3 3.0ᵃ 8.4ᵃ
Rubber additive 2-(methylthio)-benzothiazole MTBT C8H7NS2 3.2ᵇ N/A
Surfactant 2,4,7,9-tetramethyl-5-decyn-4,7-diol TMDD C14H26O2 3.6ᵇ 13.5c
UV stabilisers Benzophenone BPH C13H10O 3.1ᵇ N/A
Octocrylene OC C24H27NO2 6.9ᵇ N/A

N/A: These compounds are not ionizable.

a
Obtained from Rostvall et al. (2018).
b
Obtained from the U.S. Environmental Protection Agency's EPISuite 4.1 (Blum et al., 2017)
c
Marvin was used for drawing, displaying and characterizing chemical structures, substructures and reactions, Marvin 17.21.0, ChemAxon (https://www.
chemaxon.com).

stainless steel grids (1 mm mesh). The feed water to the columns was columns (five sorbents plus control). The feed water was sampled twice
taken from the outlet of a septic tank sewage treatment system in a week in order to monitor potential changes, first directly after the
Brottby, located 35 km north of Stockholm, Sweden, to which five spiking process (day 0) and then from the remaining water in the
households were connected. A 22 L batch of wastewater was transferred container at the end of each week (day 7).
to a glass container, spiked with 1 mL of a standard mixture containing Feed water and effluent water samples from the 12 weeks of column
AHTN, BPA, BPH, HCB, HHCB, MK, MTBT, MX, NBBS, TBEP, TBP, operation were analysed every week for dissolved organic carbon
TCEP, TDCPP, TMDD, TPP and OC and 3.7 mL of a standard mixture (DOC), total phosphorus (Ptot), ammonium nitrogen (NH4-N), pH,
containing APAP, CBZ, CF, DEET, DF, FOSA, IP, LAT, LOT, MEP, OP, conductivity, turbidity and dissolved oxygen. In addition, feed water
PFBS, PFOS, PP and SL (Table 2). Details on the spiked chemicals and and effluent samples from week 1, week 2, week 4, week 8 and week 12
their concentrations are given in Table S1 in the Appendix: Supple- were analysed for MP concentrations. The mean concentrations of MPs,
mentary material (SM). The spiked water was kept in the refrigerator at DOC and Ptot at day 0 and day 7 (feedwater) were used for calculating
4 °C during the whole experimental period. the weekly RE.
Spiked wastewater was pumped (pump 1) to the GAC, Xylit, lignite
and sand columns (Fig. S1). Effluent from these four columns was 2.4. Analytical methodology
collected through funnels into glass bottles, which were covered with
aluminium foil to prevent photolytic degradation. Half of the treated Analyses of AHTN, BPA, BPH, HCB, HHCB, MK, MTBT, MX, NBBS,
effluent from the GAC column was pumped (pump 2) further to a Po- TBEP, TBP, TCEP, TDCPP, TMDD, TPP and OC (see Table 2) were
lonite® column (GAC + P) which had the same dimensions as the other conducted using solid phase extraction (SPE) and gas chromatography
columns. An empty reference column, receiving unspiked distilled (Agilent Technologies, Palo Alto, CA) coupled to high resolution mass
water, was used to control MP background levels. The reference column spectrometry (GC/HRMS) (Autospec Ultima MS, Waters Corporation,
was tilted to ensure water contact with the inner wall of the column. Milford, MA) as described in Blum et al. (2017, 2018). Prior to in-
The total pumping time per 24-h period was 10.5 h for pump 1 and strumental analysis, water sample were filtered using glass fibre filters
5.25 h for pump 2 (pumping schedule given in Table S2). The feed (GF/F with pore size 0.7 μm), and the filtrates were extracted by SPE.
water bottle was stirred slowly by an agitator during the pumping Before SPE extraction, the filtered samples were spiked with corre-
period. The intended surface load in the columns was around 0.39 m sponding isotopically labelled internal standards to compensate any
day−1. loss of target analytes during subsequent pre-treatment. SPE was con-
In order to avoid bias in sampling caused by occasional substance ducted using Oasis® HLB cartridges (200 mg, 6 mL, Waters, Milford,
concentration peaks, effluent samples were collected three times per MA, USA). Before water sample loading, the SPE cartridges were con-
week and stored at 4 °C. At the end of each week, the effluent samples ditioned with dichloromethane, followed by acetonitrile and Milli-Q
were pooled to a weekly composite sample for each of the six types of water. The water samples were loaded onto the cartridge at

922
W. Zhang, et al. Journal of Environmental Management 246 (2019) 920–928

10 mL min−1 using positive pressure. Elution of analytes was performed 2.7. Determination of surface functional groups on the sorbents
with dichloromethane/acetonitrile (80:20, v/v) followed by di-
chloromethane (Blum et al., 2017). The cartridges were washed with The surface functional groups on lignite, Xylit, sand and Polonite®
Milli-Q water and then dried under vacuum. For GC/HRMS analysis, were determined by Fourier transform infrared (FTIR) spectro-
separation of analytes was accomplished using a ZB-5MS plus fused photometry performed on a Spectrum 100 instrument (PerkinElmer,
silica capillary column (60 m, 0.25 mm ID, 0.25 μm film thickness, Shelton, USA), using the potassium bromide (KBr) pellet method for
Phenomenex, USA). Electron ionisation was performed at 70 eV in se- both unused and used sorbent samples. A 3 mg sample of sorbent was
lected ion recording mode, and the MS was tuned to a resolution mixed with 300 mg KBr and pressed under 8 ton pressure for 60 s to
of > 10,000. More details of the instrumental method and operational create a pellet. Samples of used sorbent were taken between 20 cm and
conditions are described in Blum et al. (2017). 30 cm from the top of the sorbents bed. The spectrometer was fitted
The analysis of APAP, CBZ, CF, DEET, DF, FOSA, IP, LAT, LOT, MEP, with a single reflection (attenuated total reflection, ATR) Golden gate
OP, PFBS, PFOS, PP and SL (Table 2), was conducted using SPE and accessory unit (Graseby Specac Ltd, Kent, England) and a triglycine
ultra performance liquid chromatography (UHPLC) coupled to a triple sulphate (TGS) detector. Spectra were obtained from 16 scans and
quadrupole mass spectrometer (MS/MS) (TSQ Quantiva, Thermo Sci- 4.0 cm−1 resolution. The spectra were within the range
entific, Waltham, MA, USA), as previously described by Rostvall et al. 400–4000 cm−1 and were analysed using software from PerkinElmer.
(2018). After filtration using GF/F filter (as described above), the fil- Studies on surface functional groups on GAC was performed using both
tered samples were spiked with isotopically labelled internal standards particle and KBr pellet method, but the spectra obtained were not valid
to compensate any loss of target analytes during subsequent sample pre- since most infrared light was adsorbed by GAC. Different types of GAC
treatment. Extraction by SPE was conducted using HLB cartridges (as have been well characterized before (Moreno-Castilla, 2004; Seredych
described above). The cartridge was conditioned with methanol fol- et al., 2008; Putra et al., 2009), therefore the GAC data from the pre-
lowed by Milli-Q water. After loading the sample extract, the cartridge vious studies were used for comparison with data obtained for the four
was rinsed with Milli-Q water and then eluted with methanol. For sorbents in the present study.
UHPLC-MS/MS analysis, chromatographic separation was performed
on an Acquity BEH C18 column (50 mm_2.1 mm, 1.7 mm, Waters Cor- 2.8. Prediction of phosphorus removal
poration, Manchester, UK) and was preceded by a guard column of the
same packaging material. Acquisition was performed in positive and The chemical equilibrium software Visual MINTEQ 3.1 (Gustafsson,
negative electrospray ionisation mode simultaneously. The aqueous 2013) was applied to determine possible precipitation of phosphorus by
phase consisted of 5 mM ammonium acetate buffer, and the organic GAC, lignite, sand and Polonite®. Xylit has similar chemical composition
phase was acetonitrile. The GC/HRMS and LC/MS/MS methods used to GAC, consisting mainly of carbon and oxygen, and was therefore not
have been validated in our previous work (Blum et al., 2017, 2018 and included in this analysis. The saturation index was calculated using the
Rostvall et., 2018). default equilibrium constants for complexes and precipitates in the
Analysis of DOC was carried out with a TOC-L analyzer (Shimadzu, Visual MINTEQ database. The predicted PO4-P removal was then
Kyoto, Japan). NH4-N and Ptot were analysed using Seal Analytical AA3 compared with the actual experimental results, in order to find a rea-
Autoanalyzer. sonable explanation for phosphorus removal.

2.5. Calculations and statistical analysis 2.9. Quality assurance and quality control

The RE was calculated for the DOC, Ptot and MPs according to Random grab samples (n = 15) were taken from the outflow of all
equation (1) (Eq. (01)): columns in week 5, week 7 and week 9, for analysis of basic wastewater
parameters (pH, turbidity and conductivity and levels of DOC, PO4-P
Ceff ⎞ and NH4-N). The results were compared against those obtained for the
RE = ⎛1 −
⎜ × 100%
⎟

⎝ Cin ⎠ (01) weekly samples, to check the representativeness of the weekly samples.
No significant differences (p > 0.05) were found between grab and
where Cin is the influent concentration and Ceff is the effluent con- weekly samples.
centration. To evaluate the potential impact from the experimental set-up, in-
Grab samples from week 5, week 7 and week 9 were statistically fluent and effluent samples from the reference column (operated with
evaluated against weekly samples, using SPSS 24 (IBM Corp.). Mann- distilled water) were analysed. A few MPs (viz. BPH, BPA and NBBS)
Whitney U tests were performed to compare the variation in the con- were detected in both the influent and effluent distilled water samples,
centrations of DOC, PO4-P, Ptot, NH4-N, and dissolved oxygen and in the but at insignificant concentrations compared with the spiked con-
values of pH, turbidity, conductivity and redox potential. One-way centrations of target chemicals (< 3%).
ANOVA and least significant difference (LSD) post-hoc tests were per-
formed to test whether the RE differed significantly between sorbents. 3. Results and discussion
The correlation between octanol/water partition coefficient (log Kow),
acid dissociation constant (pKa) and the RE of the MPs were analysed 3.1. Sorbent characterization
using Pearson correlation statistics.
3.1.1. Surface charge on sorbents and chemicals
2.6. Determination of point of zero charge (pHpzc) of the sorbents The presence of surface functional groups and the surface charge are
important factors affecting the adsorption capacity and the removal
The point of zero charge (pHpzc) of the sorbents was determined mechanism of the MPs (Vargas et al., 2011). During the 12 weeks of the
using a pH titration procedure (Putra et al., 2009). This involved pla- column experiment, the average pH value (with standard deviation
cing 50 cm3 of 0.01 M sodium chloride (NaCl) solution in Erlenmeyer (SD)) of the spiked feed water was 7.5 ± 0.2. The pHpzc on GAC can
flasks and adding 0.1 M hydrochloric acid (HCl) or 0.1 M sodium hy- range from acidic to basic depending on its physicochemical properties
droxide (NaOH) until the pH of solution within each flask was adjusted and the treatment method used during modification, and higher pHpzc
to a value between 2 and 12 (pHinitial). A 0.15 g sample of sorbent was (> 8) have been shown to exhibit higher adsorption capacity (Órfão
then added to each flask and the pH was measured after 48 h (pHfinal). et al., 2006; Mahmudov and Huang, 2010; Di Natale et al., 2013). The
The pH for which the pHfinal equals pHinitial is defined as the pHpzc. pHpzc of GAC and Polonite® was 10 and > 12, respectively, which is

923
W. Zhang, et al. Journal of Environmental Management 246 (2019) 920–928

Fig. 1. Fourier transform infrared (FTIR) spectra of virgin and used sorbents (depicted –ww (wastewater)): A) lignite, B) Xylit, C) sand and D) Polonite®.

higher than the pH of the feed water (Fig. S2) indicating an overall and Polonite® were different compared to lignite and Xylit. Many of the
positive surface charge on these sorbents. In contrast, the pHpzc for peaks were attributed to silicon containing groups, e.g. the peaks near
Xylit, lignite and sand (5.9, 2.6 and 5.9, respectively) were all lower 1100, 650 and 470 cm−1 are indicative of O-Si-O vibration, Si-H
than the feed water, indicating an overall negatively charged surface. rocking and Si-O vibration, respectively (Atalay et al., 2001). Several
Some of the target compounds will appear as positively or negatively functional groups (e.g. C]O and C-O in carboxylic acid; C-O in alcohol
charged ions depending on the pH in the solution and the pKa of the and ester) were lacking, suggesting that inorganic sorbents are less ef-
compound (Table 2). For instance, PFOS and diclofenac are negatively ficient in removal of organic MPs. Similar observations have been made
charged at pH 7.5. for other mineral-based filter materials (Gustafsson et al., 2009; Putra
et al., 2009).
The used lignite, Xylit, sand and Polonite® resembled the unused
3.1.2. Characterization of functional groups on unused and used sorbents
samples, but the intensity changed markedly and position changed
The peaks found in the FTIR spectra of unused lignite and Xylit were
slightly (Fig. 1). Peaks close to 1100 cm−1 and 602 and 663 cm−1 for
similar and indicate the presence of several functional groups (Fig. 1A
lignite were much weaker or disappeared. The reason for the changes
and B, Table S4). For instance, the C]O stretching peaks (close to
could be quite complex, since real wastewater spiked with multiple MPs
1710 cm−1) indicate the presence of carboxylic acid (Seredych et al.,
was used in the experiment and reactions such as adsorption, de-
2008), and the intense peaks near 1508-1609 cm−1 for Xylit and at
gradation, oxidation and precipitation may have affected the results.
1624 cm−1 for lignite were attributed to C]C aromatic stretching
(Putra et al., 2009). Furthermore, the peaks detected close to
1375 cm−1 were probably from C-H rocking in the aldehydes. Overall, 3.2. Removal of DOC and total phosphorus
the results for lignite and Xylit indicate the presence of hydroxyl and
carbonyl groups which can potentially bind MPs in the wastewater The mean concentration of DOC in the unspiked wastewater was
solution to the surface of the organic sorbents. Similar surface func- 68 ± 16 mg L−1, which increased up to 145 ± 26 mg L−1, after
tional groups have been identified in different types of GAC in previous spiking the solution with MPs. Therefore, 53% of the DOC came from
studies (Seredych et al., 2008; Putra et al., 2009; Gong et al., 2016). It the solvent of the spiking solution of MPs. The organic sorbents released
should, however, be noted that Xylit displayed several peaks between some carbon into the wastewater, so the initial concentration in the
1000 cm−1 and 1600 cm−1 that were not present in the spectra for spiked solution may not have represented the true DOC value that came
lignite (Fig. 1, Table S4). from the wastewater organic matters. For this reason, the DOC con-
In contrast, the surface functional groups observed for unused sand centration from week 1 was not considered in the analysis (Fig. 2A).

Fig. 2. Removal efficiency (%) of A) dissolved organic carbon (DOC) and B) total phosphorus (P) during the 12-week column experiment.

924
W. Zhang, et al. Journal of Environmental Management 246 (2019) 920–928

GAC achieved the highest removal of DOC (96 ± 1%), followed by the
dual filter system GAC + P (92 ± 2%) and Xylit (89 ± 3%) (Fig. 2A).
Activated carbon adsorption is one of the recommended technologies
for removal of dissolved organic matter from water (Gur-Reznik et al.,
2008) and Polonite has also shown to be able to remove organic matter
from wastewater (Nilsson et al., 2013). Lignite showed good removal
until week 5 (79 ± 3%), but the RE then declined to around 31% be-
tween week 8 and week 12. The lowest removal of DOC by sand was at
week 4, and then the RE increased from week 5 (Fig. 2A). That increase
was probably due to the growth of biofilm, which helped with biode-
gradation and/or adsorption of organic matters. However, increased
production of biomass in sand filters has shown little further advantage Fig. 3. Average removal efficiency (%) of the 31 MPs during the 12-week
to water purification (Campos et al., 2002). column experiment for five sorbents tested.
The concentration of Ptot at the influent was 7.4 ± 1.2 mg L−1. The
dual filter (GAC + Polonite® achieved the best removal of Ptot with an had adsorbed enough water and the HRT increased from 5-20 min to
average RE of 96 ± 2% (Fig. 2B). The average removal of Ptot by GAC 15–20 min (Table S3), which probably contributed to higher overall RE
was 12 ± 27%. Polonite® improved the phosphorus removal greatly. (Ejhed et al., 2018; Kim et al., 2016).
According to the saturation index calculated by Visual MINTEQ (Table The average removal of MPs by GAC, GAC + P and Xylit remained
S6), phosphorus was unable to precipitate in the GAC columns. All steady after week 4, while removal by lignite started to decrease. The
phosphate-related minerals are undersaturated and the major species main drivers for this downward trend for lignite were reduced removals
that may be formed in GAC columns are HPO42− and H2PO4− (Table of OC, APAP, DEET, LOT and PFOS (a reduction of 15%, 6%, 7%, 19%
S7). Thus, the removal of phosphorus was most likely by adsorption. and 66%, respectively) and desorption of DF, IP and PFBS after week 4.
Polonite® contains a large amount of calcium, which means that 82% of The concentrations of DF, IP and PFBS found in the effluent water
phosphorus in wastewater may form CaPO4− in the column (Table S7). samples were eventually higher than the inlet concentration, indicating
Phosphorus in Polonite® columns is generally bound to calcium and that lignite probably exceeded its sorption capacity for these chemicals
may precipitate as Ca3(PO4)2, Ca4H(PO4)3·3H2O and hydroxyapatite and released those previously adsorbed (McCleaf et al., 2017). A similar
(Ca5(PO4)3OH) (Gustafsson et al., 2009). reduction in removal of DOC and Ptot by lignite was observed at week 5
Similarly to the DOC removal, the highest RE of Ptot by lignite was (Fig. 2A and B). It should be mentioned that the observed decreasing
observed in the first five weeks (85 ± 3%) and then removal decreased removal efficiency from lignite could be associated with other factors,
sharply (Fig. 2B). Negative removal was observed from week 9, which and analytical uncertainty cannot be ruled out; however, the analytical
likely was caused by desorption. The assumption can be made that error induced from sample pre-treatment was minimized by the use of
lignite was exhausted after 5 weeks. Lignite contains a certain amount isotopically labelled internal standards before SPE (as described above).
of Al3+ and Ca2+, resulting in formation of AlHPO4+, CaHPO4 (aq) and
CaH2PO4+, but 96% of phosphorus was in H2PO4̄ form (Table S7).
Phosphate was not able to bind on calcium in the lignite column, but it 3.3.2. Target MPs removal efficiencies
can bind to aluminium and iron and thus precipitate to AlPO4·1.5H2O, The GAC column achieved the highest RE among all five sorbents
strengite ((FePO4·2H2O) and variscite (AlPO4·2H2O) (Eveborn et al., tested, with 29 of the 31 target MPs showing over 90% removal and
2009; Grzmil and Wronkowski, 2006). with an average removal efficiency of 97 ± 6% (Table 3, Fig. 4). In
The removal of Ptot by sand stabilised at 14 ± 2% after 6 weeks, i.e. agreement with other studies, GAC achieved high removal of a wide
after bacterial formed organic matter was observed in the effluent range of MPs (Clara et al., 2005; Derylo-Marczewska et al., 2017; Zhang
water. The removal mechanism probably shifted at that point from et al., 2018). The addition of a Polonite® filter after the GAC did not
adsorption to biological processes (cf. Campos et al., 2002). Sand further improve the removal of other MPs with an average RE of
contains a small amount of calcium, iron and aluminium that may be 94 ± 12%. However, OC was improved by 8–24%, which may be due
released into the water, leading to the formation of Ca3(PO4)2, FePO4 to competitive adsorption between MPs and organic matters in the GAC
and AlPO4. Xylit obtained a similar removal of Ptot as sand with average column (Jung et al., 2013; Hu et al., 2014; Mailler et al., 2015). The pH
removal of 14 ± 15%. in the Polonite® filter was about 11, which means that OC was nega-
tively charged that possibly increases the degree of adsorption of OC
3.3. Removal of the target MPs onto the positively charged Polonite® surface.
There was a greater variation in the RE of the 31 MPs for Xylit and
3.3.1. General time trends for MPs removal lignite, and a similar phenomenon was observed in our previous study
The average RE of the 31 MPs during the 12-week column experi- (Zhang et al., 2018). The pore size of Xylit and lignite was similar (14.7
ment remained comparatively steady for the GAC (96–98%) and and 16.7 nm, respectively) (Table 1), however lignite has a larger pore
GAC + P (91–97%) columns (Fig. 3). The fluctuation in RE was greater volume (0.02 cm3 g−1) and bigger specific surface area (5.3 m2 g−1)
over time for Xylit (73–85%) and lignite (60–76%). Sand had the most
fluctuating removal curve, with the RE varying between the sampling Table 3
Number of MPs removed by the five different sorbents tested and their removal
weeks (Fig. 3).
efficiency (n = 31).
Slightly increased RE of the MPs was observed for all sorbents in the
first two to four weeks of operation (Fig. 3), although DOC removal Removal Lignite Xylit GAC GAC + Polonite® Sand
decreased slightly in the GAC, lignite, Xylit and sand filters (Fig. 2A). efficiency

This may have been due to competition between bulk parts of DOC and > 90% 9 17 29 27 4
MPs (Mailler et al., 2015), since the wastewater, which was taken from 80% - 90% 7 5 1 1 1
effluent of a septic tank, had a high DOC concentration (on average, 50%–80% 8 6 1 2 5
68 mg L−1). Another explanation for the increased removal of MPs at < 50% 7 3 0 1 21
Average removal 68 ± 29 80 ± 28 97 ± 6 94 ± 12 43 ± 30
the first week may be the short hydraulic retention time (HRT) in the
efficiency
columns, since unsaturated flow was applied in the column experiment (%)
to mimic the flow of OSSFs. After one week of operation, the sorbents

925
W. Zhang, et al. Journal of Environmental Management 246 (2019) 920–928

Fig. 4. Twelve-week mean removal efficiency (n = 5) of 31 micropollutants in the five different sorbents tested. For compound abbreviations, see Table 2.

when compared with Xylit (0.01 cm3 g−1 and 2.5 m2 g−1), which (1.6 ≤ log Kow ≤ 2.6) than that of the chemicals mentioned in the
should favour the MP adsorption capacity. However, the mean removal previous paragraph, except for LOT and IP (log Kow = 4.0 and 3.8, re-
rate of MPs by Xylit was 80 ± 28% and 17 MPs showed more than 90% spectively). The removal of surfactant (TMDD) and artificial sweetener
removal. A few chemicals were poorly removed, for instance the (SL) was poor for lignite (11–35%) and Xylit (9–55%), which may due
average removal of PFOS was 25% and a negative removal rate of PFBS to their high water solubility (1.7–2.0 g L−1 and 23 g L−1, respectively)
and SL was observed (−8% and −9%, respectively, Fig. 4). Only 16 of (Guedez and Püttmann, 2011; Rostvall et al., 2018). Several other
the 31 MPs were more than 80% removed by lignite, while seven of the studies also confirmed that compounds with high hydrophobicity and
remaining MPs achieved less than 50% removal (Fig. 4, Table 3). The low solubility show a stronger tendency to be adsorbed and retained on
average RE of lignite was 68 ± 29%. A possible explanation for the less sorbent surfaces (Derylo-Marczewska et al., 2017; Kumar et al., 2007;
efficient removal for lignite may be its low pHpzc value (2.55), while the Moreno-Castilla, 2004).
pHpzc for Xylit is 5.93, which inhibit the adsorption of some MPs. An- A limited number of MPs were well removed by sand, including
other possible explanation is that Xylit has more surface function APAP, PP, MX, CF and HCB with a RE ranging from 88% to 100%.
groups than lignite, which provided more adsorption sites (Fig. 1, Table Among these five MPs, PP, MX and HCB are quite hydrophobic com-
S4). pounds with log Kow over 3 and thus their removal was most likely
The removal of MPs by sand was significantly different from that of attributable to adsorption instead of degradation (Blum et al., 2017).
other sorbents (p < 0.05, one-way ANOVA; Table S5). Most MPs were APAP was positively charged in the feed water, and thus ionic attrac-
poorly removed in the sand filter, with 21 out of the 31 chemicals tion may have helped in adsorption onto the sand, which has negatively
showing less than 50% removal, resulting in average RE of 43 ± 30% charged surfaces (Martínez-Hernández et al., 2014).
(Fig. 4, Table 3). Sand obtained very limited pore volume The exact mechanism involved within liquid phase adsorption is
(0.002 cm3 g−1) and surface area (0.6 m2 g−1), besides the surface complex. Hydrophobic compounds adsorb stronger to the filter material
functional groups were not favourable to adsorb MPs. The removal of compared with hydrophilic compounds due to the hydrophobic effect
MPs for sand was most likely due to biological activity than adsorption. (Blum, 2018). Pearson correlation coefficient between the RE and oc-
As showed in other studies, sand showed limited ability on adsorption, tanol-water partition coefficient (log Kow) of the MPs for GAC, Xylit,
while a few pharmaceuticals could be biodegraded in biofilm active lignite, sand and GAC + Polonite® was calculated. Lignite showed a
filters (Dalahmeh et al., 2018; D'Alessio et al., 2015). significant correlation (r = 0.41, p < 0.05) between log Kow and the
removal of MPs, whereas no significant correlation was found for GAC,
Xylit, sand and GAC + P.
3.4. Hydrophobic effect

A few chemicals were well removed by all organic sorbents, in- 3.5. Other adsorption mechanisms
cluding fragrances (HHCB, MX, MK, AHTN), biocide (HCB), organo-
phosphates (TBP, TBEP, TDCPP, TPP), pharmaceutical (LAT, OP), The negatively charged PFASs, DF and IP showed a high RE using
rubber additive (MTBT), PFASs (FOSA), UV stabiliser (BPH) and pre- GAC (> 99%), which can be most likely explained by the positive
servative (PP) (Fig. 4). The average RE for these compounds by GAC, charge of GAC (Anumol et al., 2015; Zeng, 2015; Ávila et al., 2017). On
lignite and Xylit was 98 ± 2%, 90 ± 5% and 95 ± 4%, respectively. the other hand, the removal of these MPs was less efficient by the ne-
All of these chemicals except LAT are relatively hydrophobic (log Kow gatively charged Xylit and lignite (23%–80%). A charged MP molecule
≥3), which increases the possibility of adsorption (Luo et al., 2014; can electrostatically interact with an opposite charged surface thus
Derylo-Marczewska et al., 2017; Kumar et al., 2007). LAT is quite increase the RE. The removal of the negatively charged DF and IP by
persistent and poorly biodegradable (Bollmann et al., 2016) and, lignite was between 60% and 100% in the first four weeks, but negative
though the hydrophobicity was low, the removal was still most likely removal was observed for week 8, indicating desorption of DF and IP
due to adsorption. probably due to the negative charge of lignite (Fig. 4). Low adsorption
The removal of plasticiser (NBBS), pharmaceuticals (APAP, CF, CBZ, of DF due to its negative charge has been found in other studies (Ejhed
LOT, MEP, IP), pesticide (DEET), UV stabiliser (OC), organophosphate et al., 2018; Malmborg and Magnér, 2015). Most of the target MPs
(TCEP) and polymer impurity (BPA) was less efficient with lignite and appeared to have neutral charge in wastewater, although both nega-
Xylit (86 ± 9% and 60 ± 13%, respectively) than with GAC tively and positively charged chemicals exist. A correlation between pKa
(95 ± 9%). The log Kow value of these chemicals is in general lower of the MPs and their RE (n = 31) by different sorbents were not

926
W. Zhang, et al. Journal of Environmental Management 246 (2019) 920–928

significant (Pearson correlation, p > 0.05). Thus, the ionic attraction achieved average RE of 98 ± 2%, 90 ± 5% and 95 ± 4% for GAC,
between the MPs and the sorbents surface was not obvious. lignite and Xylit, respectively. MPs which are less hydrophobic, more
According to FTIR analyses, the functional groups of Xylit and lig- water soluble or negatively charged were less efficiently removed by
nite include carbonyl, carboxylic and hydroxyl groups. The hydrogen Xylit and lignite (67 ± 35%, 49 ± 26%), as compared to GAC. Sand
atoms of the functional groups can interact with nitrogen and oxygen showed low removal efficiencies for MPs with removal efficiencies
atoms of the target MPs in wastewater (Tong et al., 2016). Furthermore, of < 50% for 21 out of the 31 targeted MPs. Surface functional groups
the oxygen atoms of these groups can interact with hydrogen atoms of differ greatly between organic sorbents (GAC, Xylit, lignite) and in-
the –COOH, –OH and –NH2 groups in the MPs. In addition, a lot MPs organic sorbents (sand, Polonite®). Organic sorbents contain more sur-
have aromatic rings in their structures, the π-π dispersion interaction face functional groups at the surfaces, which helps in adsorption of
between the sorbent surface and the MPs should be considered in the MPs. Moreover, the negatively charged MPs (i.e. PFOS, DF and IP) were
adsorption processes as well (Vargas et al., 2011). more easily desorbed from the surface of lignite over time.
As mentioned in section 3.1.2, functional groups identified in GACs Polonite® removes phosphorus efficiently, as the calcium in
by several studies were also found in Xylit and lignite. The better re- Polonite® can react with phosphorus and precipitate. Lignite has po-
moval of MPs by GAC may also due to the large specific surface area tential for phosphorus removal by aluminium and iron binding, but its
(PriyaSwapna and Radha, 2017). GAC has a surface area around life span is too short to make this sorbent economically efficient. GAC
950 m2 g−1, whereas Xylit and lignite has 2.5 and 5.3 m2 g−1, respec- and sand have limited ability to remove phosphorus. The combination
tively (Table 1). A larger surface area provides more surface functional of a GAC and a Polonite® filter has merits for removal of both MP and
groups involved in the interaction with MPs thus increase the sorption phosphorus.
capacity (Aivalioti et al., 2012). The average pore size of GAC is 2.2 nm, The removal of Ptot and most MPs by lignite was high in the first
which is in a beneficial range for the removal of MPs compare with four weeks. After week 8, desorption of some MPs occurred (e.g.
Xylit and lignite (16.7 and 14.7 nm), since most of the sorption takes PFASs), resulting in negatively removal efficiencies. Sand alone cannot
place in micropores (PriyaSwapna and Radha, 2017; Li et al., 2002). remove MPs and nutrients efficiently, thus the soil-based system in
Besides, GAC obtained a much larger pore volume (0.51 cm3 g−1) in OSSFs needs to be enhanced. To achieve the dual removal of MPs and
contrast to the other sorbents, which is of importance for sorption as phosphorus, organic sorbents, such as GAC, should be complemented
well. by an extra phosphorus filter (e.g. Polonite®).

3.6. Biofilm formation Declaration of interest

Biofilm formation is another major issue affecting MP removal, as it None.

may help in the biodegradation (Bester et al., 2011) and sorption onto
the biofilm (Dalahmeh et al., 2018). Bacterially formed organic matter Acknowledgements
was found in the effluent water from GAC, lignite and sand columns
after 30, 36 and 18 days operation, and a biofilm was possibly formed This study was supported by the Swedish Research Council FORMAS
in these filters. The formation of a biofilm in Polonite® filter was through the project RedMic (Reduction of diffuse emissions of micro-
probably inhibited due to the low organic matter content (on average, pollutants from on-site sewage facilities) (216-2012-2101). The authors
5.5 ± 2.1 mg L−1) and high pH (on average, 11.4 ± 0.8). This might gratefully acknowledge the facilities and technical assistance of the
explain that the RE did not improve over time using GAC + Polonite® Umeå Core Facility for Electron Microscopy (UCEM) at the Chemical
filter. Biological Centre (KBC), Umeå University.
The removal of several MPs (≥5%) in GAC (n = 2), Xylit (n = 5),
lignite (n = 11) and sand (n = 13) columns increased when comparing Appendix A. Supplementary data
their removal efficiencies between week 4 and week 8, especially in
sand column (Fig. 4). The downward trend of the average RE for sand at Supplementary data to this article can be found online at https://
week 12 was due to the negatively removal of BPH, MTBT, NBBS and doi.org/10.1016/j.jenvman.2019.05.137.
TCEP, the average RE of these MPs by sand was 66% at week 8, how-
ever, decreased to −30% at week 12 (Fig. 4). When comparing the References
removal of MPs by sand over time, BPA, MK, TBP, TBEP TDCPP, TPP
and TMDD were found to be better removed at both week 8 and week Aivalioti, M., Pothoulaki, D., Papoulias, P., Gidarakos, E., 2012. Removal of BTEX, MTBE
12 compared to week 4, with an average increase of 38%. The best and TAME from aqueous solutions by adsorption onto raw and thermally treated
lignite. J. Hazard Mater. 207–208, 136–146.
improvement of the removal was observed for TBEP, TBP and TMDD Anumol, T., Sgroi, M., Park, M., Roccaro, P., Snyder, S.A., 2015. Predicting trace organic
(73%, 59% and 43%, respectively), whereas it was smaller for BPA, MK, compound breakthrough in granular activated carbon using fluorescence and UV
TDCPP and TPP (15–28%). absorbance as surrogates. Water Res. 76, 76–87.
Atalay, S., Adiguzel, H.I., Atalay, F., 2001. Infrared absorption study of Fe2O3–CaO–SiO2
The increase of the RE achieved in the Xylit, GAC and lignite col- glass ceramics. Mater. Sci. Eng. A 304–306, 796–799.
umns was not as much as in the sand column, except for TBP, TCEP, Ávila, C., Pelissari, C., Sezerino, P.H., Sgroi, M., Roccaro, P., García, J., 2017.
TDCPP, CBZ and OP in lignite columns, where removal increased by Enhancement of total nitrogen removal through effluent recirculation and fate of
PPCPs in a hybrid constructed wetland system treating urban wastewater. Sci. Total
12%, 23%, 8%, 12% and 13%, respectively, in week 8 and week 12
Environ. 584–585, 414–425.
compared to week 4. In addition, the RE of TBP, BPH, HCB and OP Bester, K., Banzhaf, S., Burkhardt, M., Janzen, N., Niederstrasser, B., Scheytt, T., 2011.
slightly increased in Xylit and GAC columns (2–4%) after week 4. The Activated soil filters for removal of biocides from contaminated run-off and waste-
waters. Chemosphere 85, 1233–1240.
improved removal of TMDD, TBP and BPH may due to adsorption onto
Blum, K.M., 2018. Targeted and Untargeted Analysis of Organic Contaminants from On-
the developing biofilm, since these chemicals are quite hydrophobic Site Sewage Treatment Facilities: Removal, Fate and Environmental Impact. National
(log Kow > 3.1), and thus may be removed more easily by adsorption. Library of Sweden), SwePub9789176018361.
Blum, K.M., Andersson, P.L., Renman, G., Ahrens, L., Gros, M., Wiberg, K., Haglund, P.,
2017. Non-target screening and prioritization of potentially persistent, bioaccumu-
4. Conclusions lating and toxic domestic wastewater contaminants and their removal in on-site and
large-scale sewage treatment plants. Sci. Total Environ. 575, 265–275.
GAC achieved the best removal of the target MPs (on average Bollmann, F.A., Seitz, W., Prasse, C., Lucke, T., Schulz, W., Ternes, T., 2016. Occurrence
and fate of amisulpride, sulpiride, and lamotrigine in municipal wastewater treat-
98 ± 2%), followed by Xylit (80 ± 28%) and lignite (68 ± 29%). In ment plants with biological treatment and ozonation. J. Hazard Mater. 320, 204–215.
general, more hydrophobic MPs (log Kow ≥ 3) were better removed,

927
W. Zhang, et al. Journal of Environmental Management 246 (2019) 920–928

Campos, L.C., Su, M.F.J., Graham, N.J.D., Smith, S.R., 2002. Biomass development in Kreuzinger, N., Clara, M., Strenn, B., Vogel, B., 2004. Investigation on the behaviour of
slow sand filters. Water Res. 36 (18), 4543–4551. selected pharmaceuticals in the groundwater after infiltration of treated wastewater.
Clara, M., Strenn, B., Gans, O., Martinez, E., Kreuzinger, N., Kroiss, H., 2005. Removal of Water Sci. Technol. 50 (2), 221–228.
selected pharmaceuticals, fragrances and endocrine disrupting compounds in a Kumar, A., Kumar, S., Kumar, S., Gupta, D.V., 2007. Adsorption of phenol and 4-ni-
membrane bioreactor and conventional wastewater treatment plants. Water Res. 39 trophenol on granular activated carbon in basal salt medium: equilibrium and ki-
(19), 4797–4807. netics. J. Hazard Mater. 147 (1), 155–166.
Dalahmeh, S., Ahrens, L., Gros, M., Wiberg, K., Pell, M., 2018. Potential of biochar filters Li, L., Quinlivan, P.A., Knappe Detlef, R.U., 2002. Effects of activated carbon surface
for onsite sewage treatment: adsorption and biological degradation of pharmaceu- chemistry and pore structure on the adsorption of organic contaminants from aqu-
ticals in laboratory filters with active, inactive and no biofilm. Sci. ATotal Enviro. eous solution. Carbon 40, 2085–2100.
612, 192–201. Luo, Y., Guo, W., Ngo, H.H., Nghiem, L.D., Hai, F.I., Zhang, J., Liang, S., 2014. A review
Derylo-Marczewska, A., Blachnio, M., Marczewski, A.W., Swiatkowski, A. Buczek B., on the occurrence of micropollutants in the aquatic environment and their fate and
2017. Adsorption of chlorophenoxy pesticides on activated carbon with gradually removal during wastewater treatment. Sci. Total Environ. 473–474 619-614.
removed external particle layers. Chem. Eng. J. 308, 408–418. Mahmudov, R., Huang, C.P., 2010. Perchlorate removal by activated carbon adsorption.
Di Natale, F., Erto, A., Lancia, A., 2013. Desorption of arsenic from exhaust activated Separ. Purif. Technol. 70 (3), 329–337.
carbons used for water purification. J. Hazard Mater. 260, 451–458. Mailler, R., Gasperi, J., Coquet, Y., Deshayes, S., Zedek, S., Cren-Olivé, C., Cartiser, N.,
D'Alessio, M., Yoneyama, B., Kirs, M., Kisand, V., Ray, C., 2015. Pharmaceutically active Eudes, V., Bressy, A., Caupos, E., Moilleron, R., Chebbo, G., Rocher, V., 2015. Study
compounds: their removal during slow sand filtration and their impact on slow sand of a large scale powdered activated carbon pilot: removals of a wide range of
filtration bacterial removal. Sci. Total Environ. 524–525, 124–135. emerging and priority micropollutants from wastewater treatment plant effluents.
Ejhed, H., Fång, J., Hansen, K., Graae, L., Rahmberg, M., Magnér, J., Dorgeloh, E., Plaza, Water Res. 72, 315–330.
G., 2018. The effect of hydraulic retention time in onsite wastewater treatment and Malmborg, J., Magnér, J., 2015. Pharmaceutical residues in sewage sludge: effect of sa-
removal of pharmaceuticals, hormones and phenolic utility substances. Sci. Total nitization and anaerobic digestion. J. Environ. Manag. 153, 1–10.
Environ. 618, 250–261. Martínez-Hernández, V., Meffe, R., Herrera, S., Arranz, E., de Bustamante, I., 2014.
Eveborn, D., Gustafsson, J.P., Hesterberg, D., Hillier, S., 2009. XANES speciation of P in Sorption/desorption of non-hydrophobic and ionisable pharmaceutical and personal
environmental samples: an assessment of filter media for on-site wastewater treat- care products from reclaimed water onto/from a natural sediment. Sci. Total Environ.
ment. Environ. Sci. Technol. 43 (17), 6515–6521. 472, 273–281.
Fisher, I.J., Phillips, P.J., Colella, K.M., Fisher, S.C., Tagliaferri, T., Foreman, W.T., McCleaf, P., Englund, S., Östlund, A., Lindegren, K., Wiberg, K., Ahrens, L., 2017.
Furlong, E.T., 2016. The impact of onsite wastewater disposal systems on ground- Removal efficiency of multiple perfluoroalkyl substances (PFASs) in drinking water
water in areas inundated by Hurricane Sandy in New York and New Jersey. Mar. using granular activated carbon (GAC) and anion exchange (AE) column tests. Water
Pollut. Bull. 107 (2), 509–517. Res. 120, 77–87.
Gago-Ferrero, P., Gros, M., Athrens, L., Wiberg, K., 2017. Impact of on-site, small and Méndez-Díaz, J.D., Mahmoud, M.A.D., Rivera-Utrilla, J., Sánchez-Polo, M., Bautista-
large scale wastewater treatment facilities on levels and fate of pharmaceuticals, Toledo, I., 2012. Adsorption/bioadsorption of phthalic acid, an organic micro-
personal care products, artificial sweeteners, pesticides, and perfluoroalkyl sub- pollutant present in landfill leachates, on activated carbons. J. Colloid Interface Sci.
stances in recipient waters. Sci. Total Environ. 601–602, 1289–1297. 369, 358–365.
Gamage, M.G.D.S., 2017. Reduction of organic micro-pollutants in sewage water – a Moreno-Castilla, C., 2004. Adsorption of organic molecules from aqueous solutions on
structure-adsorption relationship study and detailed characterization of natural ad- carbon materials. Carbon 42 (1), 83–94.
sorbent. DiVA.org: umu-130409. Nilsson, C., Renman, G., Westholm, L.J., Renman, A., Drizo, A., 2013. Effect of organic
Gong, Z., Li, S., Ma, J., Zhang, X., 2016. Synthesis of recyclable powdered activated load on phosphorus and bacteria removal from wastewater using alkaline filter ma-
carbon with temperature responsive polymer for bisphenol A removal. Separ. Purif. terials. Water Res. 47 (16), 6289–6297.
Technol. 157, 131–140. Órfão, J.J.M., Silva, A.I.M., Pereira, J.C.V., Barata, S.A., Fonseca, I.M., Faria, P.C.C.,
Gros, M., Blum, K.M., Jernstedt, H., Renman, G., Rodríguez-Mozaz, S., Haglund, P., Pereira, M.F.R., 2006. Adsorption of a reactive dye on chemically modified activated
Andersson, P.L., Wiberg, K., Ahrens, L., 2017. Screening and prioritization of mi- carbons—influence of pH. J. Colloid Interface Sci. 296 (2), 480–489.
cropollutants in wastewaters from on-site sewage treatment facilities. J. Hazard Priya, S., Swapna and Radha, K.V., 2017. A review on the adsorption studies of tetra-
Mater. 328, 37–45. cycline onto various types of adsorbents. Chem. Eng. Commun. 204 (8), 821–839.
Grzmil, B., Wronkowski, J., 2006. Removal of phosphates and fluorides from industrial Putra, E.K., Pranowo, R., Sunarso, J., Indraswati, N., Ismadji, S., 2009. Performance of
wastewater. Desalination 189 (1–3), 261–268. activated carbon and bentonite for adsorption of amoxicillin from wastewater: me-
Guedez, A.A., Püttmann, W., 2011. Occurrence and fate of TMDD in wastewater treat- chanisms, isotherms and kinetics. Water Res. 43 (9), 2419–2430.
ment plants in Germany. Water Res. 45 (16), 5313–5322. Rostvall, A., Zhang, W., Dürig, W., Renman, G., Wiberg, K., Ahrens, L., Gago-Ferrero, P.,
Gur-Reznik, S., Katz, I., Dosoretz, C.G., 2008. Removal of dissolved organic matter by 2018. Removal of pharmaceuticals, perfluoroalkyl substances and other micro-
granular-activated carbon adsorption as a pretreatment to reverse osmosis of mem- pollutants from wastewater using lignite, Xylit, sand, granular activated carbon
brane bioreactor effluents. Water Res. 42 (6–7), 1595–1605. (GAC) and GAC+Polonite® in column tests - role of physicochemical properties.
Gustafsson, J.P., 2013. Visual MINTEQ version 3.1. https://vminteq.lwr.kth.se/. Water Res. 137, 97–106.
Gustafsson, J.P., Renman, A., Renman, G., Poll, K., 2009. Phosphate removal by mineral- Seredych, M., Hulicova-Jurcakova, D., Lu, G.Q., Bandosz, T.J., 2008. Surface functional
based sorbents used in filters for small-scale wastewater treatment. Water Res. 42 (1), groups of carbons and the effects of their chemical character, density and accessibility
189–197. to ions on electrochemical performance. Carbon 46 (11), 1475–1488.
Hu, J., Martin, A., Shang, R., Siegers, W., Cornelissen, E., Heijman, B., Rietveld, L., 2014. Stevens-Garmon, J., Drewes, J.E., Khan, S.J., McDonald, J.A., Dickenson, E.R.V., 2011.
Anionic exchange for NOM removal and the effects on micropollutant adsorption Sorption of emerging trace organic compounds onto wastewater sludge solids. Water
competition on activated carbon. Separ. Purif. Technol. 129, 25–31. Res. 45, 3417–3426.
Jung, C., Park, J., Lim, K.H., Park, S., Heo, J., Her, N., Oh, J., Yun, S., Yoon, Y., 2013. Tong, K., Lin, A., Ji, G., Wang, D., Wang, X., 2016. The effects of adsorbing organic
Adsorption of selected endocrine disrupting compounds and pharmaceuticals on pollutants from super heavy oil wastewater by lignite activated coke. J. Hazard
activated biochars. J. Hazard Mater. 263, 702–710. Mater. 308, 113–119.
Kårelid, V., Larsson, G., Björlenius, B., 2017. Pilot-scale removal of pharmaceuticals in Tröger, R., Klöckner, P., Ahrens, L., Wiberg, K., 2018. Micropollutants in drinking water
municipal wastewater: comparison of granular and powdered activated carbon from source to tap - method development and application of a multiresidue screening
treatment at three wastewater treatment plants. J. Environ. Manag. 193, 491–502. method. Sci. Total Environ. 627, 1404–1432.
Kauppinen, A., Martikainen, K., Matikka, V., Veijalainen, A.M., Pitkänen, T., Heinonen- Vargas, A., Cazetta, A., Kunita, M.H., Silva, T.L., Almeida, V.C., 2011. Adsorption of
Tanski, H., Miettinen, I.T., 2014. Sand filters for removal of microbes and nutrients methylene blue on activated carbon produced from flamboyant pods (Delonix regia):
from wastewater during a one-year pilot study in a cold temperate climate. J. study of adsorption isotherms and kinetic models. Chem. Eng. J. 168 (2), 722–730.
Environ. Manag. 133, 206–213. Zeng, E.Y., 2015. Persistent organic pollutants (POPs): analytical techniques. 67.
Kholoma, E., Renman, G., Renman, A., 2016. Phosphorus removal from wastewater by Environmental Fate and Biological Effects, pp. 1–660.
field-scale fortified filter beds during a one-year study. Environ. Technol. 37 (23), Zhang, W., Renman, G., Blum, K.M., Gros, M., Ahrens, L., Jernstedt, H., Wiberg, K.,
2953–2963. Andersson, P.L., Björlenius, B., 2018. Removal of micropollutants and nutrients in
Kim, E., Jung, C., Han, J., Her, N., Park, C.M., Jang, M., Son, A., Yoon, Y., 2016. Sorptive household wastewater using organic and inorganic sorbents. Desali. Water Treat.
removal of selected emerging contaminants using biochar in aqueous solution. J. Ind. 120, 88–108.
Eng. Chem. 36, 364–371.

928