Journal of Water Process Engineering 25 (2018) 195–204

Contents lists available at ScienceDirect

Journal of Water Process Engineering

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

Assessing potential of nanofiltration and reverse osmosis for removal of T

toxic pharmaceuticals from water
⁎
K.P.M. Liconaa, , L.R. de O. Geaquintob, J.V. Nicolinic, N.G. Figueiredob, S.C. Chiapettab,
A.C. Habertc, L. Yokoyamaa
a
Universidade Federal do Rio de Janeiro, Programa de Pós-Graduação em Tecnologia de Processos Químicos e Bioquímicos – TPQB/EQ/UFRJ, Centro de Tecnologia - Av.
Horácio Macedo, 2030 - 101 - Cidade Universitária, Rio de Janeiro, RJ, 21941-450, Brazil
b
Instituto Nacional de Tecnologia, Av. Venezuela, 82, Saúde, Rio de Janeiro, RJ, 20081-312, Brazil
c
Universidade Federal do Rio de Janeiro, Programa de Pós-Graduação em Engenharia Química – PEQ/COPPE/UFRJ, Centro de Tecnologia - Av. Horácio Macedo, 2030 -
101 - Cidade Universitária, Rio de Janeiro, RJ, 21941-450, Brazil

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

Keywords: Concern about organic micropollutants, which are present in the environment at trace concentrations
Pharmaceutical compounds (ng L−1–μg L−1) is related to the adverse effects to organisms exposed to these substances. Pharmaceutically
Nanofiltration Active Compounds (PhCAs) may be present in natural waters and usually cannot be removed or degraded by
Reverse osmosis conventional water treatment processes. For this reason, treatment techniques, such as Nanofiltration (NF) and
Rejection efficiency
Reverse Osmosis (RO) are recommended to improve their removal. In this context, the present study aims to
Retention mechanisms
evaluate the removal of five Non-steroidal anti-inflammatory drugs (NSAIDs), analgesics and anti-pyretic:
acetaminophen, ibuprofen, dipyrone, diclofenac, and caffeine by NF and RO process. NF90 and BW30 mem-
branes were characterized by scanning electron microscopic (SEM), contact angle and zeta potential. Retention
of PhACs was evaluated considering pH feed solution and operating pressure. Results indicated that NF90
membrane was efficient to reach over 88% rejection for some selected PhACs. Best results were obtained at
20 bar and pH 5 with more than 90% of rejection. For nonionic compounds acetaminophen and caffeine, ex-
clusion by size is the main mechanism for rejection by NF90 membrane, whereas for anionic compounds ibu-
profen, dipyrone, and diclofenac, electrical exclusion predominated at pH 5 and 7. Rejection results with NF90
membrane show that hydrophobicity has an important role due to the adsorption on the membrane surface.
Conversely, lower rejections for hydrophilic compounds were observed due to the adsorption/diffusion me-
chanisms, both in NF90 and RO at pH 5.

1. Introduction organic compounds involved, there is a considerable group of emerging

Human health is increasingly dependent on pharmaceutical pro-
ducts. Also, water scarcity and water reuse have become essential issues
in water resource management worldwide, always considering the
conservation of natural water system as a final goal and primary driver
of the scientific community. As a result, they along with environmental
and public health authorities have been making a significant effort in
order to understand the fate of pharmaceutically active compounds
(PhACs) both in engineered urban wastewater treatment processes and
in the natural environment. This has been reflected in the high number
of scientific articles devoted to this subject in the last two decades. On
the other hand, there are no modification of legal regulations yet [1].
Although usually at very low concentrations compared to other the
organic micropollutants that have been detected in raw wastewaters of
various natures, in surface and groundwaters, in drinking water sources
and in some cases in treated waters as well. This group includes pes-
ticides, pharmaceutical products (PhACs), personal hygiene products
and several other industrial pollutants [2]. The highest possibility of
their occurrence is in industrial wastewater or sanitary sewage due to
inefficient treatment or even direct discharge without prior treatment,
which does not comply with legislation. Socioeconomic and other cul-
tural factors related to the indiscriminate marketing, consumption,
improper disposal and destination of medicines in the countries also
contribute negatively for the spread of these compounds in waters and
wastewaters [3].
Removal of some PhACs in conventional treatment stations is almost

⁎
Corresponding author.
E-mail address: karlaufs@hotmail.com (K.P.M. Licona).

https://doi.org/10.1016/j.jwpe.2018.08.002
Received 10 March 2018; Received in revised form 22 July 2018; Accepted 5 August 2018
Available online 11 August 2018
2214-7144/ © 2018 Elsevier Ltd. All rights reserved.
K.P.M. Licona et al.
impossible due to several factors, including low volatility, different
hydrophobicity, complex structures, recalcitrance, very low con-
centrations, influence of microorganisms and interaction with other
solutes and separation media. These compounds remain dissolved in
urban effluents or may connect to sludge which later is released into the
surrounding aquatic environment and soil; in other words, a lot of
water-soluble substances can be transported and distributed more easily
in the water cycle, causing impact on the environment and human
beings [4,5].
There is still little knowledge about the impact that can be created
by this exposure and what are the long-term effects on human health
and environment [6]. Nevertheless, some consequences are already
known: a) antibiotics have been blamed for the development of bac-
terial resistance in the environment; b) endocrine disruptors affect the
reproductive and endocrine system of aquatic organisms, inducing
genotoxocity, acute and chronic toxicity, for example causing the
feminization of male fish that live in contaminated rivers [7], and c)
anti-inflammatory drug diclofenac induces chronic toxic effects on the
fish kidneys after exposure at the dose of 0.001 mg L−1 [3].
Emerging micropollutants from anthropogenic origin are one of the
primary pollutant sources. Particularly speaking of pharmaceuticals,
they are persistent compounds that present high potential for bioac-
cumulation, low biodegradability and concentration in the range of
ng L−1 to μg L−1 in aquatic environments. These facts indicate that
conventional treatment technologies do not entirely remove them,
which leads to their subsequent release into the aquatic environment
through discharge and/or reuse applications. Therefore, there is a need
to investigate alternative technologies for their removal in municipal
wastewater treatment [2,8].
Due to this concern, different physicochemical processes have been
proposed as a tertiary treatment for effluents from water treatment
plants. Among them are membrane separation processes, which have
proved ability to remove these pollutants, achieving high removal rates
for pharmaceutical compounds. Membrane separation processes (MSP)
have been increasingly used in the treatment of reused water, drinking
water and water for industrial purpose and their use has increased
exceptionally due to technological, economic and environmental needs
[8].
Nanofiltration (NF) membranes have come a long way since their
beginning in the late 1980s. NF is a MSP that uses pressure gradient as
the driving force. With properties between those of ultrafiltration (UF)
and reverse osmosis (RO) membranes, NF membranes have attracted
interest due to their versatility as a separation process. Their pore size is
typically in the order of 1 nm, which corresponds to a molecular weight
cut-off (MWCO) in the range of 100–5000 Da. Therefore, this type of
process may seem interesting for organic micropollutants removal. NF
membranes also exhibit a moderate level of charge due to the dis-
sociation of surface functional groups or the adsorption of charged
solutes. For example, polymeric NF membranes contain ionizable
groups such as carboxylic and sulfonic acid groups which result in
charged surfaces in the presence of an aqueous feed solution. NF
membranes operate with no phase change and typically at low pres-
sures, reaching high rejections of multivalent inorganic salts and small
organic molecules. This makes the separation process highly competi-
tive regarding selectivity and cost-benefit when compared to traditional
separations. Thus, NF has found wide range of application across of
industrial sectors including water and wastewater treatment, pharma-
ceutical and biotechnological processes, and food engineering to name
a few. As the use of NF technology becomes increasingly widespread,
efforts will increase for more effective separation projects with the ul-
timate aim of reducing costs [6,9,10].
Although RO has a capacity of separation similar to that of NF, with
high solute rejection rates and, in some cases, being a concurrent pro-
cess, it has a higher operating cost than NF, usually, because of the need
of high operating pressure (to overcome lower permeability) and a
greater fixed investment. Membranes used in RO processes do not have
196
Journal of Water Process Engineering 25 (2018) 195–204
pores or convective flow, and the separation occurs by the sorption-
diffusion mechanism in which species permeate and are solubilized in
the material that composes the membrane and, then, diffuses through
their thickness moved by a gradient of chemical potential until the
desorption step on the permeate side [9].
Some studies in literature report the employment of NF and RO
membranes for the removal of different endocrine disruptors and PhACs
in real matrices [11–16]. According of these studies, high removals of
compounds can be obtained by the NF and RO membranes. Many stu-
dies investigate the removal of organic contaminants using NF and RO
in real matrices spiked with PhACs (treated municipal wastewaters,
treated water from drinking water plant, surface waters, among others),
and this is the main limitation to study the contribution of each com-
pound separately for the process performance, because of the influence
from the other compounds and from organic matter existing in these
matrices [10,12–14,17–21].
In the last decades, attention was given to the interactions between
contaminants, membranes and components in the water that impact the
rejection of solutes with different geometries and physicochemical
properties [9]. Membranes can remove PhACs either by size exclusion,
electrostatic repulsion, hydrophobic interaction or adsorption
[10,12,14,19,22–24]. It is noteworthy that the interaction of the solute
with water, as well as the nature of the polymer responsible for the
separation, are also fundamental factors for the efficiency of the pro-
cess, thus it is necessary to know the properties and characteristics of
the polymers that constitute the polymer matrix so that the membrane
separation process can be optimized [5]. In addition, membrane oper-
ating conditions, such as feed rate, flow, recovery, pressure and tem-
perature, are fundamental for the operation and rejection values [25].
Size exclusion is always considered the most important mechanism
of rejection. Often, the main rejections are expected from membranes
that have molecular weight cut off (MWCO), for organic contaminants
with a higher molecular mass than the MWCO. Molecular mass is not a
correct prediction when the molecules are charged or have a high hy-
drophobicity [26].
Verliefde et al. [10] evaluated the rejection efficiency of twenty
pharmaceutical compounds with different electrostatic charges of NF
membranes. Removal of the pharmaceutical compounds was partially
determined by exclusion by molecular weight, but the charge of the
compounds also appeared to play an essential role in the rejection of
neutral solutes. Ozaki and Li [24] studied the influence of molecular
mass on the retention of organic and inorganic compounds on reverse
osmosis membrane. The study concluded that polar compounds, either
positively or negatively charged, are more effectively removed com-
pared to less polar or neutral charged compounds because they interact
with the membrane surface.
The understanding of these interactions is essential for the im-
provement and selection of suitable membranes, as well as for the de-
velopment of tools that will allow the prediction of process efficiency
for a wide spectrum of micropollutants [9]. However, these effects have
not been properly discussed in the literature, mainly regarding the
evaluation of membrane performance at different pH and zeta potential
analyzes throughout the pH range.
To date, the intrinsic relationship between the physiochemical
properties of PhCAs at trace levels, the solution pH and the membrane
retention behavior remains poorly understood. Thus, this study aimed
to evaluate the rejection performance by NF and RO membranes for the
following PhCAs – dipyrone, paracetamol, diclofenac, ibuprofen and
caffeine – and to elucidate their removal mechanisms in different pH
conditions. These compounds represent different pharmaceutical
groups with quite distinct physicochemical properties and are com-
monly detected in aquatic samples. Paracetamol (N-4-hydro-
xyphenylacetamide) is one of the most commonly used non-narcotic
analgesic and antipyretic in the world. Diclofenac (2,4,6-di-
chlorophenyl-aminobenzeneacetic acid) has antirheumatic and anti-
inflammatory use. Ibuprofen is a derivative of 3-phenoxybenzeneacetic
K.P.M. Licona et al.
acid which is also one of the most widely used drugs worldwide.
Caffeine is a stimulant of the central nervous system that contributes to
the relief of some types of headaches, is a respiratory stimulant and an
analgesic potentiator. It is found in most medications [12,13,17]. So-
dium dipyrone, also known as metamizole, is a pyrazolonic derivative
that acts as analgesic and antipyretic agent, and is mainly marketed in
sodium form in different pharmaceutical formulations, such as: oral
solutions, tablets, injectables and suppositories.
The effect of surface properties, such as scanning electron micro-
scopy, zeta potential, hydrophilicity, on membrane performance was
deeply investigated using solutions of pure PhCAs, that is, one solution
for each PhAc, for each membrane process (NF and RO). Also as no-
velty, this study introduces the evaluation of removal of dypirone by
membranes, since there is no study so far in literature reporting the
treatment of water containing traces of this component by NF and RO
membranes.
2. Materials and methods
2.1. Chemical reagents
For instrumental analysis, the five pharmaceuticals selected as
model micropollutants for this study (acetaminophen, ibuprofen, di-
clofenac, dipyrone, caffeine) were all high purity grade (> 99%), pur-
chased from Sigma–Aldrich (see Tables 1 and 2). Solvents used for li-
quid chromatography analysis – acetonitrile (ACN), methanol (MeOH)
and Ammonium Acetate (CH3COONH4)) – were HPLC-grade. ACN and
MeOH were purchased from Baker and Ammonium Acetate was pur-
chased from Fluka. MILIQ water was used for the experiments, and
acetic acid (analytical grade) was used to adjust the pH of the ammo-
nium acetate to 4.
2.2. Scanning electron microscopy (SEM)
Morphology of NF90 and BW30 membranes was analyzed by
scanning electron microscopy (FEI Company Quanta 200).
Photomicrograph of the membrane surface was obtained after
membrane skin was carefully removed from the support and fractured
with liquid nitrogen, and then glued on stubs by carbon conductive tape
and covered with a thin layer of gold (conductor).
Table 1
Pharmaceutically active compounds (PhACs) investigated in this study.
Therapeutic Class Pharmaceutical active compound CASa number
Analgesics/anti-inflammatories Acetaminophen 103-90-2
Ibuprofen 15687-27-1
Diclofenac 15307-86-5
Dipyrone 68-89-3
Central Nervous System (CNS) Stimulant Caffeine 58-08-2
a
CAS: Chemical Abstracts Service.
197
Journal of Water Process Engineering 25 (2018) 195–204
2.3. Contact angle measurements
Static contact angle measurements were performed by the sessile-
drop method using Data Phisics OCA 15, Germany at room tempera-
ture. A droplet of ultrapure water was deposited on the membrane
surface. After reaching the equilibrium configuration, a static image of
the droplet was taken and the equipment software calculated the con-
tact angle. The measurements were performed at three different points
of the membrane surface to obtain the average contact angle value.
2.4. Zeta Potential of the membrane surface
Zeta potential of NF and RO membranes were determined by
streaming potential measurements with an electrokinetic analyzer,
SurPASS, AntonPaar. Two membrane samples of 55 mm × 25 mm were
inserted opposite from each other and separated by a spacer, forming a
thin slit channel. The streaming potential was resulted from the pres-
sure-driven flow of an electrolyte solution that passes through this
channel. Then the zeta potential was calculated using the Fairbrother-
Mastim equation, Eq. (1) [28].
Δφ ε ε λ R
= ζ 0 r ⎛ h h⎞
ΔP ηλ 0 ⎝ R ⎠ (1)
where ζ is the zeta potential (mV), Δφ is the measured streaming po-
tential in the flow cell (mV), ΔP is the applied pressure (mbar), ε0 is the
vacuum permittivity (F m−1), εr is the relative dielectric constant of the
electrolyte solution, λ 0 is the bulk conductivity of the circulating
electrolyte (mS m−1), and Rh and R are the measured electrical re-
sistances (mV A−1) across the flow channel filled with the saline re-
ference solution and filled with the electrolyte solution, respectively.
Prior to the analysis, the membranes were soaked in KCl 0.01 mol L−1
electrolyte solution for 24 h and the zeta potential measurements were
performed using the same electrolyte solution at pH ranging from 3 to
10 [29].
2.5. Bench-scale experiments
2.5.1. Sample preparation
A synthetic solution (1 L) with 10 mg L−1 was prepared for each
selected PhAC in ultrapure water. Ibuprofen and diclofenac were in-
itially diluted in 10 mL of methanol, because of water solubility lim-
itations. pH of the prepared solutions is around 5, for all the drugs
considered. For assays at pH 4, HCl (1 M) was used for adjustment, and
Formula Molecular Weight (Da) Molecular Structure
C8H9NO2 151.17
C13H18O2 206.29
C14H11Cl2NO2 296.15
C13H16N3NaO4S 333.34
C8H10N4O2 194.19
K.P.M. Licona et al. Journal of Water Process Engineering 25 (2018) 195–204

Table 2
Physicochemical properties of the selected (PhACs) [27].
Compound log Kowa pKa Dipole moment (D) Molar volume Molecular Diffusion
(cm3/mol) radius (nm) coefficient
(cm2/s)

Acetaminophen 0.49 (Hydrophilic) 9.5 3.85 121.0 0.32 7.63·10−6

Ibuprofen 3.68 4.4 1.22 200.5 0.37 6.61·10−6
(Hydrophobic)
Diclofenac 4.64 (Hydrophobic) 4.1 2.51 207.0 0.46 5.41·10−6
Dipyrone −0.82 3.8 – – 0.45 –
(Hydrophilic)
Caffeine −0.04 (Hydrophilic) 10.4 3.40 133.9 0.36 6.80·10−6

a
Hydrophobic when log Kow > 2 [27].

for assays at pH 7, NaOH (1 M) was used for adjustment.

2.5.2. Nanofiltration and reverse osmosis system

Two commercial NF (NF90) and RO (BW30) membranes, purchased
from Dow Filmtec, were used in our study. The nominal molecular
weight cut-off (MWCO) of NF90 is in the range of 200–400 Da and of
BW30 ∼100. More information about the properties of the NF90 and
BW30 membranes is detailed in Table 3.
A laboratory-scale membrane filtration unit was used for all the
experiments as shown in Fig. 1. Effective membrane area of the flat
membrane cell is 28.26 cm2. Temperature of feed solution was kept
constant at 23 ± 0.5 °C for all the experiments. Before each experi-
ment, the NF and RO plants were washed with ultra-pure water for at
least 30 min. The experiments were performed in batch mode with re-
circulation. Membranes were soaked in pure water for 24 h. Soaked
Fig. 1. Experimental system of flow through the Membrane.
membranes were then compacted in NF system using ultrapure water at
maximum operation pressure (20 bar for NF and 30 bar for RO) until
constant permeate flow was reached. Membranes permeability was Where Cp concentration of solute in permeate flow and Cf concentra-
determined by Eq. (2) with ultra-pure water, at a flow rate of 40 L h−1, tion of solute in the feed flow. The concentrations of pharmaceuticals in
varying transmembrane pressure. both permeate and feed flow were measured using a High-Performance
Liquid Chromatography (HPLC).
Jw = Aw.(ΔP − Δπ) (2)
−2 −1
Where Jw is water permeate flux (L m h ), Aw is water perme- 2.5.3. Analytical determination
ability (L m−2 h−1 bar−1), and (ΔP − Δπ) are osmotic and hydraulic Concentration profiles of each compound were obtained by High-
pressure difference through the membrane, respectively. Feed tank was Performance Liquid Chromatography (HPLC) with a Diode-Array
filled with 1 L of synthetic solution obtained as detailed in item 2.4.1. Detector (DAD) and a reversed phase analytical column (Zorbax Eclipse
Finally, the experiment was initiated by discharging permeate and re- XDB-C18 2.1 × 100 mm, 3,5 μm).
circulating the concentrate flow into the feed tank. Samples of feed, The method consisted of a binary gradient of CH3COONH4 (10 mM)
permeate and concentrate flows were taken when permeate flow was (mobile phase A) and ACN (mobile phase B) at a flow rate of
constant. 1 mL min−1. Proportion of mobile phase B is maintained at 10% for
Observed rejection (R0) of pharmaceuticals and caffeine by the 5 min and then varied to 90% in 20 min, holding the value of 90% for
membrane was determined. The amount of solute rejected by the 2 min. Then, proportion of mobile phase B decreases to 10% in 3 min
membrane was calculated based on the following equation (Eq. (3)): and remains so for a further 2 min, totaling approximately 30 min of
analysis. Injection volume was 5 μL, and the column temperature was
Cp
R 0 (%) = ⎜⎛1− ⎟⎞. 100 maintained at 30 °C throughout the analysis. Acetic acid was used to
⎝ Cf ⎠ (3) adjust mobile phase A to pH 4. Detection wavelengths were 243 nm for
acetaminophen, 225 nm for ibuprofen, 280 nm for diclofenac and
Table 3 273 nm for dipyrone and caffeine. These five compounds selected as
Characteristics of NF and RO membranes. model pharmaceuticals have the advantage of solubility, stability, po-
larity and easy separation in the HPLC system, as well as appropriate
Supplier NF90 BW30
Dow Filmtec Dow Filmtec
light absorption for detection in a diode array detector (DAD). Stock
solutions containing concentration of 1000 μg mL−1 of the considered
MWCO (Da)a 200–400 ∼100 pharmaceuticals were prepared in methanol-water (1:1, v/v).
Max. temp. °Ca 45 45 Analytical curves were prepared from the stock solution with con-
pH rangesa 3–10 2–11
centrations ranging from 1 to 10 μg mL−1.
Average pore radius (nm)b 0.34–0.68 –
Saline Rejectiona 85–95% NaCl 99,5%NaCl
> 97% Na2SO4 3. Results and discussion
Material of Fullaromatic Fullaromatic
skin layera polyamide Polyamide
3.1. Membrane morphology
a
According to the membranes supplier.
b
[30,31]. Fig. 2(a) and (b) illustrate the SEM cross-sections of NF90 and BW30

198
K.P.M. Licona et al.
Fig. 2. SEM images of the (a) NF90 and (b) BW30 membranes cross sections.
Fig. 3. Images of the contact angles formed by the deposition of the ultrapure water drop on the (a) NF90and (b) BW30 membranes at room temperature.
Fig. 4. Zeta potential of NF90 and BW30 membrane as a function of the pH in
0.01 mol L−1 KCl solution.
membranes, respectively. SEM clearly shows the morphological mi-
crostructure of the surface and cross sections of the membrane material,
which can be used to predict its performance. All membranes show a
characteristic morphology of asymmetric membrane, with an aromatic
polyamide top layer and a polysulfone porous sub-layer. These com-
mercial membranes are prepared as thin film composites with a thin
crosslinked polyamide selective barrier layer synthesized from m-phe-
nylenediamine and trimesoyl chloride. Because of this characteristic,
they acquire unique separation properties. [32].
3.2. Membrane surface properties
Contact angle is commonly employed to measure the hydrophilicity
199
Journal of Water Process Engineering 25 (2018) 195–204
Fig. 5. Chemical structure of polyamide membranes and ionization of func-
tional groups as a function of the pH.
of membranes, which is controlled by electrostatic and/or adsorption
interactions between the surface functional groups and water mole-
cules. Contact angle can be used to relate permeability flux and hy-
drophobic interactions of the PhACs investigated in this study. Fig. 3
shows the contact angle of NF90 (a) and BW30 (b) membranes, re-
spectively. It is possible to observe that membranes have hydrophilic
surface when they have contact angle smaller than 90°. Polyamide
membranes have surface functional groups, including carboxyl and
amino groups, which are attributed to the membrane fabrication
technique. The amino groups are due to interfacial polymerization of
polyamide with monomers as reactants. The unreacted monomers in
the polyamide matrix hydrolyze into carboxyl groups after the mem-
brane is immersed in water [33]. In the contact angle experiment, a
drop of ultrapure water (pH around 6) is deposited on the surface of the
membrane. In this pH, deprotonated carboxylic groups (pKa 1.8–2.4)
interact with water by hydrogen-bond [34], which gives a hydrophilic
surface. NF90 membrane has pores that confer greater hydrophilicity
K.P.M. Licona et al.
Fig. 6. Speciation diagrams of (a) Acetaminophen; (b) Ibuprofen; (c) Diclofenac; (d) Dipyrone and (e) Caffeine in water.
than that of dense BW30 membrane. Carboxyl and amino groups are
also located at the pore inner surfaces.
Electrostatic phenomena strongly influences the separation perfor-
mance of membranes due to surface charge, and zeta potential gives
some information about the net charge on the surface. Fig. 4 shows the
zeta potential of NF90 and BW30 membranes as a function of the pH.
The isoelectric point (IEP) of NF90 was close to pH 4 and for BW30 was
close to pH 3. For pH > i.e.p., zeta potential of both membranes is
negative due to deprotonation of the carboxyl groups (eCOOH) (pKa
1.8–2.4) on the skin layer [28]. Dissociation of carboxylic groups tends
to be complete with the increase of pH. The positive surface charge is a
200
Journal of Water Process Engineering 25 (2018) 195–204
result of the protonation of amino groups protonation. These groups
exhibit pKa at a range of 8.8–10.9 [35]. Fig. 5 illustrates the protona-
tion and deprotonation of functional groups according to the pH of the
solution.
A difference in the isoelectric point of the analyzed polyamide
membranes is observed in Fig. 4. Tang et al. suggest that the BW30
membrane is coated with a layer rich in alcohol functions such as
polyvinyl alcohol resulting from a dense membrane post-treatment
step. These groups (CeH) with very weak acid properties result in the
same behavior found in this experiment [36].
K.P.M. Licona et al.
Fig. 7. Rejection of the studied PhAcs as a function of pH during NF experiment (NF90 membrane): operation pressure of 5, 10, 15 and 20 bar.
3.3. Physicochemical properties of pharmaceuticals compounds (PhACs)
Functional groups strongly influence the biological and chemicals
activities of PhACs. Compounds selected for this study have distin-
guishing properties that makes them interesting to be compared, in-
cluding molecular weight, hydrophobicity, dissociation constants, di-
pole moment, molar volume, molecular radius and diffusion coefficient
(see Tables 1 and 2). To analyse hydrophilic or hydrophobic character
of the PhACs, octanol-water partition coefficient (KOW) can be used as
an indicator of the sorption properties of a micropollutant. This para-
meter does not consider pH and only reflects hydrophobic interactions.
However, hydrophobicity is a property related to electrostatic interac-
tions, surface complexation or hydrogen bonding, which depend on the
pH, which in turn depend on dissociation constant (pKa) of two func-
tional groups [37]. Compound dissociation constant and the pH influ-
ence chemical speciation of the PhCAs. Therefore, PhCAs separation
with membranes may differ significantly due to differences in charge
and physicochemical properties. Fig. 6 shows the microspecies dis-
tributions of Acetaminophen, Ibuprofen, Diclofenac, Dipyrone and
Caffeine molecules in ultrapure water.
Acetaminophen and Caffeine have similar pKa values. When the
pH < pKa, these species are in their protonated form (neutral form).
Acetaminophen contains a phenolic functional group as well as an
amide group, and above pKa point, this molecule shows a structure of a
free electron pair on the nitrogen atom, in the anionic form, as shown in
Fig. 6(a) [38]. The caffeine molecule (Fig. 6(e)) has a nitrogen atom
with a single pair that can be protonated in water at pH > pKa, re-
sulting in a positively charged molecule. At pH < pKa the molecule
will deprotonate said nitrogen, giving a neutral molecule, making it less
201
Journal of Water Process Engineering 25 (2018) 195–204
soluble in water [27].
Ibuprofen, Diclofenac, and Dipyrone have similar pKa values
(Figure (b), (c) e (d), respectively). At pH below its pKa value, these
compounds are neutral species. Above these pKa values, they acquire
negative charge [31]. Ibuprofen has a carboxyl group as a functional
group that acquires deprotonated form (eCOOe) at pH > pKa [39].
Diclofenac has weak acid properties with the NH and the carboxylic
group presence, which may act as either proton donors or proton ac-
ceptors in the water [40]. Dipyrone molecule exhibits anionic form
above pKa and is a hydrophilic molecule.
The pH of a solution can affect not only charge but also hydro-
phobicity and solubility of these substances [30]. The solubility and
hydrophobicity of these compounds reflect their speciation as a func-
tion of the pH. Acetaminophen and Caffeine are hydrophilic (log
Kow < 2) while Ibuprofen, Diclofenac, and Dipyrone are hydrophobic
(log Kow > 2) [27]. However, at higher pH values these compounds
become less hydrophobic, being more soluble in their anionic form,
which has been confirmed by other researchers [31]. For these reasons,
variations in charge and other physicochemical properties as a function
of pH may provide crucial information for remove these contaminants
by membrane technology.
3.4. Removal of the selected Pharmaceutical Substances by nanofiltration
and reverse osmosis
The performance of NF membranes was evaluated by measurement
of the PhACs’ rejection as a function of feed pH (4, 5 and 7) and op-
erating pressure (5, 10, 15 and 20 bar). Fig. 7 shows the results. The
absence of data that is observed for diclofenac at pH 4 happens because
K.P.M. Licona et al.
Fig. 8. Rejection of the studied PhAcs as a function of pH during RO experiment (BW30 membrane): operation pressure of 10, 15, 20 and 30 bar.
when HCl was added to adjust the pH, it degraded diclofenac, making
its analysis impossible.
Rejections of the five PhACs were above 88%. Higher rejections
were obtained at 20 bar and pH 5 (> 90%) for Ibuprofen and
Diclofenac. These results indicated that the NF90 membrane was effi-
cient when used to remove selected PhACs. Separation in a nanofil-
tration process may be favored by steric effects, size exclusion and/or
electrostatic interactions. Moreover, the selected PhACs exhibit the
additional separation mechanism of adsorption whenorganic matter is
mainly hydrophobic or has strong hydrogen-bonding characteristics,
which makes it readily adsorbed on the membranes surface [32,34].
Acetaminophen and caffeine show similar characteristics regarding
rejection as function of pH. In the pH range studied in this work (4–7),
these compounds are in their non-ionic form. Thereby, there is no
electrostatic interaction with the membrane, which is negatively
charged. For the range of operating pressures studied, the rejection did
not vary with the pH of the feed solution. This indicates that size effects
of molecules and membrane are predominant. The differences observed
between acetaminophen and caffeine are due to the different sizes of
their molecules, which causes caffeine to be more rejected by NF90
membranes. Acetaminophen and caffeine are hydrophilic compounds.
In water solutions these compounds are solvated and, consequently,
their effective diameter becomes larger. Therefore, they can be rejected
more effectively by steric effects [41]. For BW30 membrane, it was
observed low rejection of acetaminophen and caffeine, in comparison to
other PhACs, mainly at pH 5 in all range of operation pressure. Acet-
aminophen and caffeine are hydrophilic compounds, whose preferential
orientation favors its absorption/diffusion on the membrane surface
and inside the pores. This was previously demonstrated for other
202
Journal of Water Process Engineering 25 (2018) 195–204
hydrophilic compounds such as endocrine disrupting compounds
(EDCs) [30,42]. Therefore, acetaminophen and caffeine accumulated
on the BW30 membrane surface due to size exclusion and could even-
tually diffuse through the membrane polymer matrix towards the
permeate side. In the pH 5, these compounds are highly hydrophilic
[27], pointing to the absorption/diffusion phenomenon, and explaining
the lower rejections in RO membrane (R ∼ 92% at pH 5) compared to
rejections of other pharmaceuticals, as shown in Fig. 8. For the other
PhACs compounds, such as ibuprofen, diclofenac, and dipyrone, the
rejection through the membrane BW30 was higher than 98%.
For diclofenac, pH of feed has not had a significant influence on the
NF90 membrane rejection at all pressures evaluated. Retention of di-
clofenac by the NF90 membrane seems to be largely influenced by
steric exclusion due the smaller pore size (0.34 to 0.68 nm) [30,31] of
this membrane compared to the diclofenac molecule. However, it is
possible that there is contribution from electrostatic interactions to
diclofenac rejection, attributed to an anionic form of diclofenac mole-
cule at pH 5 and 7. The slight decrease in retention efficiency around
pH 5 can be attributed to this effect.
Ibuprofen and dipyrone exist primarily as a neutral species at
pH < pKa value (pH 4.4). Thus, exclusion by size was the only se-
paration mechanism at pH 4. As shown in Table 2, ibuprofen is a hy-
drophobic compound and dipyrone is a hydrophilic compound. It is
noteworthy that in the absence of electrostatic repulsion, hydrophobic
interaction between ibuprofen and the membrane surface becomes
more prominent, improving rejection by adsorption. At pH 5, Ibuprofen
is in its neutral (50%) and anionic (50%) form. When pH is increased to
7, there is a contribution of electrostatic repulsion, which increases the
retention of the ibuprofen molecule. Besides, the hydrophobicity of
K.P.M. Licona et al. Journal of Water Process Engineering 25 (2018) 195–204

observations are related by Taheran et al., and Agenson et al. [5,43]. In

contrast, hydrophilic dipyrone molecule can absorb/diffuse through the
polyamide matrix, reducing the observed rejection. At pH 7, the elec-
trostatic effects of dipyrone and surface membrane are more expressive,
because of an increase of the negative zeta potential of the membrane
with pH. Fig. 9 illustrates the main phenomena involved in the reten-
tion of the compounds acetaminophen, ibuprofen, diclofenac, dipyrone
and caffeine at the range of pH studied in this work by NF90 and BW30
membranes. The rejection was very similar whitin the pressure range
for both membranes. This indicates that the PhACs flux also increases
with solvent flux.

4. Conclusion

The performance of NF/RO process for PhACs removal from water

at different pH conditions was elucidated. It is summarized as follows:

1 NF produced a practically clean permeate containing less than 12%

of pharmaceuticals originally present in water.
2 PhAC rejection by membrane is influenced by a number of para-
meters including membrane MWCO, porosity, morphology, charge
and hydrophobicity. Molecular Weight, molecular size, charge, and
hydrophobicity of the PhACs as well as that of feed water must also
be taken into account.
3 Membranes showed high efficiency in the removal of the selected
pharmaceutical compounds. Permeation of the components is de-
fined by the competition between size exclusion, electric repulsion,
hydrophobic interactions and sorption/diffusion mechanisms of the
hydrophilic compounds, being related mainly to the pH of the
medium.
4 For NF90, rejection of the acetaminophen and caffeine is in-
dependent of pH for the range of evaluated (4–7), because these
components are in neutral form. The main mechanism of rejection is
exclusion by size. For diclofenac, the rejection at pH 4 and 5 occurs
due to the competition between the steric and the electrostatic
mechanisms. Rejection of ibuprofen and dipyrone is highly depen-
dent on the pH of the medium. Hydrophobic mechanisms are
markedly dominant for rejection of ibuprofen, unlike dipyrone
which is a hydrophilic molecule and therefore mechanisms of ex-
clusion by electrostatic repulsion dominate.
5 RO membranes can remove efficiently almost all PhACs studied,
with more than 98% rejection, except for the acetaminophen and
caffeine. These are smaller molecules with high hydrophilicity and
reduction of rejection occurs by sorption/diffusion mechanisms,
depending on pH. Rejection by BW30 membrane is less influenced
by electrostatic and hydrophobic effects in comparison to NF90
membrane.

References

[1] J. Mamo, M.J. García-Galán, M. Stefania, S. Rodríguez-Mozaz, D. Barceló,

Fig. 9. Illustration of mechanisms involved in the rejection of studied PhAcs by
NF90 membrane at (a) pH 4, (b) pH 5 and (c) pH 7. (d) Major separation
mechanisms involved in the rejection of studied PhAcs by BW30 membrane.
ibuprofen decreases significantly as the pH of solution increases [31].
This suggests that hydrophobic adsorption of this compound to the
NF90 membrane can be weakened, reducing rejection values. Similar
H. Monclús, I. Rodriguez-Roda, J. Comas, Fate of pharmaceuticals and their
transformation products in integrated membrane systems for wastewater reclama-
tion, Chem. Eng. J. 331 (2018) 450–461.
[2] J.L. Acero, F.J. Benitez, F. Teva, A.I. Leal, Retention of emerging micropollutants
from UP water and a municipal secondary effluent by ultrafiltration and nanofil-
tration, Chem. Eng. J. 163 (2010) 264–272.
[3] J.L. Tambosi, L.Y. Yamanaka, H.J. José, R. de Fátima Peralta Muniz, H.F. Schôder,
Recent research data on the removal of pharmaceuticals from sewage treatment
plants (STP), Quím. Nova 33 (2010) 411–420.
[4] I. Oller, S. Malato, J.A. Sánchez-Pérez, Combination of advanced oxidation pro-
cesses and biological treatments for wastewater decontamination—a review, Sci.
Total Environ. 409 (2011) 4141–4166.
[5] M. Taheran, S.K. Brar, M. Verma, R.Y. Surampalli, T.C. Zhang, J.R. Valero, Review:
membrane processes for removal of pharmaceutically active compounds (PhACs)
from water and wastewaters, Sci. Total Environ. 547 (2016) 60–77.
[6] G.B. Crestana, J.H. Silva, Fármacos residuais: panorama de um cenário negli-
genciado, Rev. Int. de Dir. e Cid. 9 (2011) 55–65.
[7] C.G.A. Silva, C.H. Collins, Aplicações de cromatografia líquida de alta eficiência
para o estudo de poluentes orgânicos emergentes, Quím. Nova 4 (2011) 665–676.

203
K.P.M. Licona et al.
[8] D.L. Oatley-Radcliffe, M. Walters, T.J. Ainscough, P.M. Williams, A. Wahab
Mohammad, N. Hilal, Nanofiltration membranes and processes: a review of re-
search trends over the past decade, J. Water Process Eng. 19 (2017) 164–171.
[9] S. Sanches, C.F. Galinha, M.T.B. Crespo, V.J. Pereira, J.G. Crespo, Assessment of
phenomena underlying the removal of micropollutants during water treatment by
nanofiltration using multivariate statistical analysis, Sep. Purif. Technol. 118
(2013) 377–386.
[10] A. Verliefde, E. Cornelissen, G. Amy, B. Van Der Bruggen, H. Van Dijk, Priority
organic micropollutants in water sources in Flanders and the Netherlands and as-
sessment of removal possibilities with nanofiltration, Environ. Pollut. 146 (2007)
281–289.
[11] Q. Sui, J. Huang, S. Deng, G. Yu, Q. Fan, Occurrence and removal of pharmaceu-
ticals, caffeine and DEET in wastewater treatment plants of Beijing, China, Water
Res. 44 (2010) 417–426.
[12] X. Yang, R.C. Flowers, H.S. Weinberg, P.C. Singer, Occurrence and removal of
pharmaceuticals and personal care products (PPCPs) in an advanced wastewater
reclamation plant, Water Res. 45 (2011) 5218–5228.
[13] N.K. Stamatis, I.K. Konstantinou, Occurrence and removal of emerging pharma-
ceutical, personal care compounds and caffeine tracer in municipal sewage treat-
ment plant in Western Greece, J. Environ. Sci. Health B 48 (2013) 800–813.
[14] A. Azaïs, J. Mendret, E. Petit, S. Brosillon, Evidence of solute-solute interactions and
cake enhanced concentration polarization during removal of pharmaceuticals from
urban wastewater by nanofiltration, Water Res. 104 (2016) 156–167.
[15] J. Radjenovic, M. Petrovic, F. Ventura, D. Barceló, Rejection of pharmaceuticals in
nanofiltation and reverse osmosis membrane in drinking water treatment, Water
Res. 42 (2008) 3601–3610.
[16] C. Bellona, J. Drewes, G. Oekler, J. Luna, G. Filteau, G. Amy, Comparing nanofil-
tration and reverse osmosis for drinking water augmentation, J. Am. Water Works
Assoc. 100 (2008) 102–116.
[17] J.H. Al-Rifai, C.L. Gabelish, A.I. Schafer, Occurrence of pharmaceutically active and
non-steroidal estrogenic compounds in three different wastewater recycling
schemes in Australia, Chemosphere 69 (2007) 803–815.
[18] J. Radjenovic, M. Petrovic, D. Barceló, Analysis of pharmaceuticals in wastewater
and removal using a membrane bioreactor, Anal. Bioanal. Chem. 387 (2007)
1365–1377.
[19] A.M. Comerton, R.C. Andrews, D.M. Bagley, C. Haoc, The rejection of endocrine
disrupting and pharmaceutically active compounds by NF and RO membranes as a
function of compound and water matrix properties, J. Membr. Sci. 313 (2008)
323–335.
[20] Q. Sui, J. Huang, S. Deng, G. Yu, Q. Fan, Occurrence and removal of pharmaceu-
ticals, caffeine and DEET in wastewater treatment plants of Beijing, China, Water
Res. 44 (2010) 417–426.
[21] S. Hajibabania, A. Verliefde, J.A. Mc Donald, S.J. Khand, P. Le-Clecha, Fate of trace
organic compounds during treatment by nanofiltration, J. Membr. Sci. 373 (2011)
130–139.
[22] S.O. Ganiyu, E.D.V. Hullebusch, M. Cretin, G. Esposito, M.A. Oturan, Coupling of
membrane filtration and advanced oxidation processes for removal of pharmaceu-
tical residues: a critical review, Sep. Purif. Technol. 156 (Part 3) (2015) 891–914.
[23] A.R.D. Verliefde, E.R. Cornelissen, S.G.J. Heijmana, I. Petrinic, T. Luxbacherd,
G.L. Amye, B. Van der Bruggen, J.C. van Dijka, Influence of membrane fouling by
(pretreated) surface water on rejection of pharmaceutically active compounds
(PhACs) by nanofiltration membranes, J. Membr. Sci. 330 (2009) 90–103.
[24] H. Ozaki, H. Li, Rejection of organic compounds by ultra-low pressure reverse os-
mosis membrane, Water Res. 36 (2002) 123–130.
[25] I. Miralles-Cuevas, A. Oller, J.A.R. Aguirre, S.S. Pérez, M. Rodríguez, Removal of
pharmaceuticals at microg L_1 by combined nanofiltration and mild solar photo-
Fenton S, Chem. Eng. J. 239 (2014) 68–74.
204
Journal of Water Process Engineering 25 (2018) 195–204
[26] Y. Kiso, Y. Sugiura, T. Kitao, K. Nishimura, Effects of hydrophobicity and molecular
size on rejection of aromatic pesticides with nanofiltration membranes, J. Membr.
Sci. 192 (2001) 1–10.
[27] J. Garcia-Ivars, L. Martella, M. Massella, C. Carbonell-Alcaina, M.-I. Alcaina-
Miranda, M.-I. Iborra-Clar, Nanofiltration as tertiary treatment method for re-
moving trace pharmaceutically active compounds in wastewater from wastewater
treatment plants, Water Res. 125 (2017) 360–373.
[28] J.V. Nicolini, C.P. Borges, H.C. Ferraz, Selective rejection of ions and correlation
with surface properties of nanofiltrationmembranes, Sep. Purif. Technol. 171
(2016) 238–247.
[29] C. Bellona, J. Drewes, The role of membrane surface charge and solute physico-
chemical properties in the rejection of organic acids by NF membranes, J. Membr.
Sci. 249 (2005) 227–234.
[30] L.D. Nghiem, A.I. Scha¨fer, M. Elimelech, Removal of natural hormones by nano-
filtration membranes: measurement, modeling and mechanisms, Environ. Sci.
Technol. 38 (2004) 1888–1896.
[31] L.D. Nghiem, S. Hawkes, Effects of membrane fouling on the nanofil-
trationofpharmaceutically active compounds (PhACs): mechanisms and role of
membrane pore size, Sep. Purif. Technol. 57 (2007) 176–184.
[32] W. Xie, G.M. Geise, B.D. Freeman, H. Lee, G. Byun, J.E. McGrath, Polyamide in-
terfacial composite membranes prepared from m-phenylenediamine, trimesoyl
chloride and a new disulfonateddiamine, J. Membr. Sci. 403–404 (2012) 152–161.
[33] Y. Mo, A. Tiraferri, N.Y. Yip, A. Adout, X. Huang, M. Elimelech, Improved anti-
fouling properties of plyamidenanaofiltration membranes by reducing the density
of surface carboxyl groups, Environ. Sci. Technol. 46 (2012) 13253–13261.
[34] Q. Li, X. Pan, Z. Qu, X. Zhao, Y. Jin, H. Dai, B. Yang, X. Wang, Understanding the
dependence of contact angles of commercially RO membranes on external condi-
tions and surface features, Desalination 309 (2013) 38–45.
[35] A.E. Childress, M. Elimelech, Effect of solution chemistry on the surface charge of
polymeric reverse osmosis and nanofiltration membranes, J. Membr. Sci. 119
(1996) 253–268.
[36] C.Y. Tang, Y.-N. Kwon, J.O. Leckie, Probing the nano-and micro-scales of reverse
osmosis membranes – a comprehensive characterization of physiochemical prop-
erties of uncoated and coated membranes by XPS, TEM, ATR-FTIR, and streaming
potential measurements, J. Membr. Sci. 287 (2007) 146–156.
[37] S.R. Wegst-Uhrich, D.A.G. Navarro, L. Zimmermann, D.S. Aga, Assessing antibiotic
sorption in soil: a literature review and new case studies on sulfonamides and
macrolides, Chem. Cent. J. 8 (2014) 1–12.
[38] V. Bernal, A. Erto, L. Giraldo, C.M. Pirajan, Effect of solution on the adsorption of
paracetamol on chemically modified activated carbons, Molecules 22 (2017) 1032.
[39] S. Evoli, D.L. Mobley, R. Guzzi, B. Rizzati, Mutiple binding modes of ibuprofen in
human serum albumin identified by absolute binding free energy calculations, Phys.
Chem. Phys. 18 (2016) 32358–32368.
[40] K.A.K. Musa, L.A. Eriksson, Photodegradation mechanism of the common non-
steroid anti-inflammatory drug diclofenac and its carbozole photoproduct, Phys.
Chem. Phys. 11 (2009) 4601–4610.
[41] L. Braeken, R. Ramaekers, Y. Zhang, G. Maes, B.V.D. Bruggen, C. Vandecasteele,
Influence of hydrophobicity on retention in nanofiltration of aqueous solutions
containing organic compounds, J. Membr. Sci. 252 (2005) 195–203.
[42] K. Kimura, G. Amy, J.E. Drewes, T. Heberer, T.-U. Kim, Y. Watanabe, Rejection of
organic micropollutants (disinfection by-products, endocrine disrupting com-
pounds, and pharmaceutically active compounds) by NF/RO membranes, J. Membr.
Sci. 227 (2004) 113–121.
[43] K.O. Agenson, J.I. Oh, T. Urase, Retention of a wide variety of organic pollutants by
different nanofiltration/reverse osmosis membranes: controlling parameters of
process, J. Membr. Sci. 225 (2003) 91–103.