You are on page 1of 8

Separation and Purification Technology 95 (2012) 89–96

Contents lists available at SciVerse ScienceDirect

Separation and Purification Technology


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

Integration of nanofiltration, UV photolysis, and advanced oxidation processes


for the removal of hormones from surface water sources
Vanessa J. Pereira a,b, Joana Galinha a, Maria T. Barreto Crespo a,b, Cristina T. Matos c,⇑, João G. Crespo d
a
Instituto de Biologia Experimental e Tecnológica (IBET), Av. República, Qta. do Marquês (EAN), 2784-505 Oeiras, Portugal
b
Instituto de Tecnologia Química e Biológica (ITQB), Universidade Nova de Lisboa (UNL), Av. da República, Estação Agronómica Nacional, 2780-157 Oeiras, Portugal
c
Laboratório Nacional de Energia e Geologia, I.P., Unidade de Bioenergia, Estrada do Paço do Lumiar, 22, 1649-038 Lisboa, Portugal
d
REQUIMTE, Departamento de Química, Faculdade de Ciência e Tecnologia, Universidade Nova de Lisboa (UNL), 2829-516 Caparica, Portugal

a r t i c l e i n f o a b s t r a c t

Article history: This study evaluates the integration of nanofiltration with direct and indirect UV photolysis for drinking
Received 8 January 2012 water treatment in order to guarantee effective removal of different hormones with endocrine disruption
Received in revised form 16 March 2012 capabilities – mestranol, octylphenol, nonylphenol, progesterone, estrone, estriol, 17a-ethynylestradiol,
Accepted 16 April 2012
and b-estradiol – from real surface water matrices.
Available online 23 April 2012
The integration of nanofiltration previously to low pressure ultraviolet direct or indirect photolysis
reduces the level of turbidity as well as the micropollutant contamination levels in drinking water sup-
Keywords:
plies, due to rejection based on size exclusion and molecular interactions with the nanofiltration mem-
Hormones
Surface water
brane surface. The use of nanofiltration in the treatment of surface waters spiked with different
Nanofiltration hormones allowed their rejection at levels higher than 71% for all target hormones except estriol. Low
Low pressure direct photolysis pressure indirect photolysis with 100 mg/L of hydrogen peroxide was also efficient to degrade the
Advanced oxidation processes selected hormones with percent degradations higher than 74% achieved for all the hormones, except
nonylphenol (55%).
The integrated process (nanofiltration followed by direct photolysis or indirect photolysis) is extremely
efficient to remove all the target hormones from a real surface water matrix and guarantees the produc-
tion of water with extremely high chemical quality.
Ó 2012 Elsevier B.V. All rights reserved.

1. Introduction Ultraviolet (UV) radiation is gaining importance in drinking


water treatment facilities as a disinfection technique, although
Water sources are potential carriers of contaminants to human chlorine is often employed for final disinfection, so that a persistent
beings and may therefore constitute a direct threat to human residual is maintained in the water distribution system. The use of
health. Therefore, the production and distribution of drinking UV radiation alone (direct photolysis) or in advanced oxidation pro-
water with high chemical and microbiological quality is crucial cesses (AOP) – where UV may be used in combination with hydro-
and has been object of growing interest by water providers and gen peroxide (indirect photolysis) to produce highly reactive and
researchers. unselective hydroxyl radicals – has proved to be effective for the re-
Hormones were reported to occur at low levels in the aquatic moval of some organic pollutants (e.g. [12–14]), besides being ex-
environment (e.g. [1–6]). As an example, in surface water, hor- tremely effective against a wide range of waterborne pathogens,
mones such as estrone, b-estradiol, a-ethinylestradiol, progester- including Cryptosporidium and Giardia which are resistant to tradi-
one, octylphenol, nonylphenol, and estriol have been reported at tional disinfection techniques (e.g. [15]). The use of UV radiation
levels ranging from 0.2 to 110 ng/L [1,5,6]. Even though hormones can therefore also lead to a reduction of the chlorine dose often
are currently not regulated, they may be proposed for future drink- applied for final disinfection, decreasing the levels of disinfection
ing water regulations due to their potential to disrupt the normal by-products (DBPs) formed. The use of UV radiation to achieve
endocrine function and affect the physiological status of animals the degradation of hormones has been also reported in the litera-
[5]. Attention has been therefore, focused in recent years on the ture using different types of lamps (e.g. [8–11,16–19]). However,
effective removal of these compounds from drinking water sources UV radiation is not suitable for treatment of water with high levels
(e.g. [4,7–11]). of suspended solids, turbidity, color, or soluble organic matter.
Appropriate membrane filtration prior to the UV system can reduce
⇑ Corresponding author. Tel.: +351 210924600x4200; fax: +351 217163636. turbidity, which improves UV transmittance and reduces shielding
E-mail address: cristina.matos@lneg.pt (C.T. Matos). of microbial pathogens and chemical compounds by particulate

1383-5866/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.seppur.2012.04.013
90 V.J. Pereira et al. / Separation and Purification Technology 95 (2012) 89–96

matter before UV treatment. In addition, membrane materials with Table 1


defined surface chemical properties and highly controlled pore size Characterization of the surface water matrix used
(parameters are characterized in terms of average values
distribution, allows for the rejection of different micropollutants and the coefficient of variation is given in parenthesis).
and microorganisms. Nanofiltration is a process where the mem-
brane acts as a barrier depending on surface chemical interactions Temperature (°C) 17.4 (22.6%)
pH 7.83 (2.3%)
(hydrophobic and electrostatic) as well as on size exclusion mech- Total organic carbon (mg/L) 4 (15.7%)
anisms, due to a high control of membrane morphology. Different Turbidity (NTU) 12.4 (70%)
studies showed the possibility of using nanofiltration in the re- Color (mg/L Pt–Co) 10.5 (N/A)a
moval of hormones, and demonstrated that the rejections achieved Alkalinity (mg/L CaCO3) 69.2 (21.7%)
Total hardness (mg/L CaCO3) 131 (36.4%)
for these compounds are largely controlled by adsorption of the
compounds to the membrane [20–29]. a
Only one measurement made for the sample used
This work focused on the validation and comparison of inte- in the studies.
grated nanofiltration and low pressure direct and indirect UV pho-
tolysis for drinking water treatment in order to guarantee effective
The surface water samples were collected, filtered with a 5 lm
removal of hormones – mestranol, octylphenol, nonylphenol, pro-
filter cartridge, transported to the laboratory in a cooler at 4 °C, and
gesterone, estrone, estriol, 17a-ethynylestradiol, and b-estradiol –
kept refrigerated until used.
from real water sources. The rejection and adsorption obtained for
An Ultra-High-Speed Liquid Chromatography system (LaChrom-
each compound using nanofiltration are determined and discussed.
Ultra System, VWR-Hitachi) [32–34], was used for the detection of
In addition, the efficiency of the direct and indirect UV photolysis
hormones, as detailed in the Supplementary data.
processes using low pressure lamps – the most commonly used
lamps in drinking water treatment facilities – is evaluated in terms
of possible hormone competition effects, influence of the water 2.2. Nanofiltration
matrix composition on the degradation efficiency, and optimiza-
tion of the hydrogen peroxide concentrations used in the indirect The nanofiltration trials were conducted in a stirred dead end
photolysis experiments. Finally, the performance of the integrated cell (Metcell, manufactured by MET, UK) coupled to an HPLC pump
process – combination of nanofiltration with the direct and indi- that continuously feed the system (shown in Fig. 1) with surface
rect photolysis process – is evaluated. water spiked with 1 mg/L of the target hormones. The surface
water was pre-filtered with a 5 lm filter cartridge in order to re-
move particles. The membrane used in these trials was a Dow-
2. Experimental section Filmthec NF 270 with an effective area of 63.6 cm2 and a molecular
weight cut-off (MWCO) of proximally 400 Da. This membrane was
2.1. Chemical reagents and water characterization chosen since it allows the rejection of several surface water chem-
ical pollutants, TOC, color, and turbidity with a significantly high
The target hormones (mestranol, octylphenol, nonylphenol, water permeability (11 L/(m2 h bar)). The trials were performed
progesterone, estrone, estriol, 17a-ethynylestradiol, and b-estra- under batch conditions at a transmembranar pressure of 8 bar
diol) were selected due to their reported occurrence (e.g. [1–3,5]) and by permeating 500 mL of the initial volume 750 mL.
as well as their different chemical structure, different molecular The concentration of the selected hormones in the feed solu-
weight and, consequently, different chemical properties (shown tion, permeate, and retentate were analyzed using the liquid
in the Supplementary data). chromatography.
All hormones selected (mestranol, octylphenol, nonylphenol, The apparent rejection (Rapp) degree of the selected compounds
progesterone, estrone, estriol, 17a-ethynylestradiol, and b-estra- after nanofiltration (referenced in the text as rejection) was deter-
diol) were purchased as solids of the highest grade commercially mined using Eq. (1), with Cp and Cf referring to the concentrations
available (Sigma–Aldrich, St. Louis, MO). Stock solutions of the hor- of each target hormone in the permeate and feed, respectively:
 
mones were prepared in high pressure liquid chromatography Cp
(HPLC) grade methanol, stored at 20 °C and diluted in laboratory Rapp ð%Þ ¼ 100  1  ð1Þ
Cf
grade water (LGW) or surface water as required. The stock solu-
tions used were as concentrated as possible to minimize the vol- Several hormones analyzed were expected to adsorb at the
ume of methanol added to the samples. A maximum of 0.2% (v/v) membrane surface. The hormones’ percent of adsorption (A) to
of methanol was present in the samples. the membrane surface was therefore determined using Eq. (2),
The concentration of the hormones used in these experiments where Cr is the concentration of each target hormone in the reten-
were higher (1 mg/L) than what is typically found in environmen- tate, whereas Vp, Vr, and Vf stand for the volumes of permeate,
tal samples in order to follow the degradation of the compounds by retentate, and feed used in each nanofiltration experiment,
direct injection using high pressure liquid chromatography (HPLC). respectively.
Bovine liver catalase was obtained from Sigma–Aldrich and  
CpV p þ CrV r
used to quench the hydrogen peroxide (one unit decomposed Að%Þ ¼ 100  1  ð2Þ
Cf V f
1 lmol of H2O2). All the other reagents used were analytical grade.
The surface water used was supplied by the drinking water util-
ity Empresa Portuguesa das Águas Livres (located in the center of 2.3. UV photolysis
Portugal) that treats and supplies drinking water to approximately
2.5 million people. The surface water matrix used was character- 2.3.1. Direct photolysis
ized (in a previous study [30]), in terms of temperature, pH [31], Low pressure (LP) ultraviolet (UV) photolysis experiments were
total organic carbon (European Standard EN 1484:1997), turbidity conducted in a collimated beam laboratory-scale reactor (Trojan
(ISO 7027:1990), alkalinity [31], and total hardness [31]. The aver- Technologies Inc., Canada) using a low pressure Hg lamp that emits
age values and coefficient of variation (standard deviation divided mainly monochromatic light at 254 nm (Fig. 1).
by the mean expressed as a percentage) of the different parameters Hundred milliliter of laboratory grade water or a natural water
are described in Table 1. matrix – surface water – were spiked with the appropriate volume
V.J. Pereira et al. / Separation and Purification Technology 95 (2012) 89–96 91

Fig. 1. Nanofiltration and low pressure UV batch systems used; 1, nanofiltration module (metcell); 2, mixers; 3, nanofiltration permeate; 4, scale; 5, low pressure UV lamp
housing.

of the hormones stock solutions to achieve concentrations of 1 mg/ compounds were spiked in surface water with different hydrogen
L. This concentration was set in order to follow the degradation of peroxide concentrations (0, 20, 40, 60, 80, and 100 mg/L).
the compounds, spiked individually in different matrices (labora- At the determined exposure times, 200 lL of sample were taken
tory grade water and surface water) and as mixtures (in surface to vials containing catalase to quench the residual hydrogen perox-
water), by direct injection using the liquid chromatography system. ide. The residual H2O2 was determined at the beginning and end of
50 mL of sample were placed in a Petri dish and continuously stir- each experiment using the I 3 method described by Klassen et al.
red beneath the LP/UV source. The remaining 50 mL were used as [36].
control and kept in the dark, under identical experimental condi-
tions, in order to determine possible changes during the photolysis 2.4. Multi-barrier approach
reaction. All experiments were conducted at room temperature
(21 ± 2 °C). In order to study the efficiency of combining both nanofiltration
The lamp irradiance was measured using a calibrated radiome- and UV direct or UV indirect photolysis, as a multi-barrier system
ter (IL1700, International Light, Newburyport, MA) which was able to increase the removal of the target hormones, a nanofiltra-
placed at the same height of the water level in the Petri dish and tion permeate – obtained in the conditions described in Section
the solution transmittance was measured by a UV photometer 2.2 – was submitted to direct and indirect photolysis, using the
(P254C, Trojan Technologies Inc.). UV fluences of approximately experimental procedures described in Sections 2.3.1 and 2.3.2.
0, 40, 100, 500, 750, 1000, and 1500 mJ/cm2 were selected, taking
into account the radiometer reading as well as petri, reflection,
3. Results and discussion
water, and divergence factors as described by Bolton and Linden
[35]. These fluences were used to establish the corresponding
3.1. Nanofiltration
exposure times at which 200 lL of sample were taken to quantify
the concentration (C) of the target hormones.
In order to study the performance of a nanofiltration membrane
Fluence-based pseudo-first order rate constants (kf) were deter-
mined for the target pollutants using direct and indirect photolysis typically used in water treatment (NF270) for the rejection of the
target hormones, three experiments were conducted at 8 bar by
from the slope of a linear regression described by Eq. (3).
permeating 67% of the initial feed volume under batch conditions.
½hormone Table 2 depicts the rejections achieved for each hormone. Even
ln ¼ kf  UV fluence ð3Þ
½hormone0 
where [hormone0] is the initial concentration of the each hormone Table 2
spiked in solution and [hormone] is the concentration of each hor- Rejection degrees obtained for each hormone with nanofiltration.
mone at a given UV fluence. Hormones Rejection degree (%)
17a-Ethynylestradiol 92 ± 3
2.3.2. Indirect photolysis 17b-Estradiol 71 ± 8
The experiments conducted to address the efficiency of indirect Estriol 38 ± 1
low pressure photolysis to degrade hormones were conducted sim- Estrone 82 ± 3
ilarly to direct photolysis experiments, with the addition of 40 mg/ Progestrone >95 ± 0
Mestranol >95 ± 0
L hydrogen peroxide (H2O2) to the hormones spiked individually
4-Nonylphenol >91 ± 0
and as mixtures into laboratory grade water. In addition, to opti- 4-Octylphenol >89 ± 1
mize the removal of the selected hormones, mixtures of these
92 V.J. Pereira et al. / Separation and Purification Technology 95 (2012) 89–96

using a membrane with a MWCO (approximately 400 Da) higher even though the equilibrium rejection after membrane saturation
than the molecular weight of the selected compounds the rejection of 17b-estradiol may decrease the overall value of rejection re-
degrees achieved were high (more than 71%) for all the selected mains high [28]. Additionally, Sanches et al. [29] reported high
hormones except for estriol (38%). For the hormones progesterone nanofiltration rejections of pesticides and hormones under ex-
and mestranol (both with detection limits of 50 ppb by direct tended adsorption conditions.
injection) as well as 4-nonyphenol and 4-octyphenol (both with
detection limits of approximately 100 ppb by direct injection) the 3.2. Direct and indirect UV photolysis
minimum rejection was calculated according with their detection
limits. Their true rejection is therefore expected to be higher than The two main parameters that influence the direct photolysis of
89%. a compound are the decadic molar absorption coefficient and the
The rejection degrees obtained (Table 2) were a result of molec- quantum yield, shown in Table 3.
ular exclusion and adsorption of the hormones to the membrane The decadic molar absorption coefficient (e(k)) measures the
material. probability that a compound will absorb light at a defined wave-
Estriol and estrone are less hydrophobic, present higher solubil- length (k) and was determined in this study by measuring the
ity and are therefore expected to adsorb to a lesser extent to the absorbance (a), at 254 nm, of solutions spiked with each of the tar-
membrane surface and natural organic matter [37] comparatively get compounds at concentrations ranging from 10 to 100 lM ([hor-
to the other hormones, which explains the lower rejections ob- mones]) using a 1 cm path length (z) according to Eq. (4):
served. The rejections achieved for hormones, such as 17b-estra-
a ¼ eðkÞ  ½hormones  z ð4Þ
diol and estrone, correlate well with values reported in the
literature with an NF200 [21] and an TFC-SR2 (from Koch Mem- For a compound to be photolabile it needs to have the capacity to
brane Systems) for estrone [37]. Comerton et al. [38] reported a absorb photons of the incident light. The decadic molar absorption
similar low value of rejection for estriol (48.1 ± 11.6) from filtered coefficient results presented in Table 3 therefore show that proges-
Lake Ontario water using an NF270 membrane, while lower rejec- terone will absorb light at the 254 nm wavelength emitted by LP
tion percentages compared to the ones obtained in this study were lamps, and was therefore expected to show the highest direct pho-
reported for 17b-estradiol, estrone, and 17a-ethynyl estradiol (that tolysis degradation (as confirmed below in Fig. 3).The quantum
ranged from 31.7% to 68.6%). Yoon et al. [39] reported high rejec- yield (U) for degradation of each target hormone represents the
tion percentages (44–93%) by using a ESNA membrane (from ratio between the total number of molecules of the compound
Hydranautics) for 26 endocrine disrupting compounds (EDCs), transformed to the total number of photons absorbed by the solu-
pharmaceutically active compounds and personal care products. tion due to the compound’s presence. This parameter is shown in
The rejection of the hormones was found to be largely depen- Table 3 and was experimentally determined as the ratio between
dent on their adsorption to the membrane material. As demon- the pseudo-first-order rate constant (k0 d) and the specific rate of
strated in Fig. 2 the overall removal of hormones from water is light absorbed by the compound at 254 nm (Ks), as detailed in
controlled by their adsorption. Fig. 2 shows the dependence of Eqs. (5) and (6). The degradation kinetics of the hormones by LP di-
the adsorption (calculated using Eq. (2)) and rejection degrees (cal- rect photolysis is a function of the parameters shown in Eqs. (5) and
culated using Eq. (1)) with the logarithmic value of the hormones (6) [12]:
partition coefficient between octanol and water, which is a mea- !
d½hormone 0
X
sure of the hydrophobicity of the compounds.  ¼ kd ½hormone ¼ K s ðkÞ /½hormone ð5Þ
High adsorption values (which varied between 40% and 80%) dt k
were obtained for all the selected hormones, except for 17b-estra-
with
diol. The observed dependence of adsorption with the octanol–
water partition coefficient of the selected hormones shown in Eop ðkÞeðkÞ½1  10aðkÞz 
Fig. 2 is in agreement with the literature [38–40]. K s ðkÞ ¼ ð6Þ
aðkÞz
McCallum et al. [24], reported that the adsorption on an NF-270
membrane follows a Freundlich adsorption isotherm, and that Ks(k) represents the specific rate of light absorption by the com-
pound, Eop ðkÞ the incident photon irradiance, e(k) the decadic molar
absorption coefficient (described above), a(k) the solution absor-
bance, z the solution depth in the Petri dish, and U the quantum
yield. Since LP lamps emit mainly monochromatic light, all the
parameters described above were experimentally obtained at
254 nm.
The highest quantum yield values were obtained for progester-
one and estrone (Table 3). Even though estrone exhibits a low dec-
adic molar absorption coefficient, its higher quantum yield

Table 3
Decadic molar absorption coefficient (e254 nm) and quantum yield (U) values of the
target hormones.

Compound e254 nm (M1 cm1) U (mol einstein1)


Estriol 1071 0.40
17b-Estradiol 2779 0.06
Estrone 2215 5.45
Octylphenol 3035 0.51
Nonylphenol 5900 0.27
17a-Ethynylestradiol 646 0.09
Progestrone 11265 2.82
Fig. 2. Dependence of hormones adsorption (to the nanofiltration membrane) and Mestranol 1255 1.45
their rejection degrees on the octanol–water partition coefficient (log Kow).
V.J. Pereira et al. / Separation and Purification Technology 95 (2012) 89–96 93

Fig. 3. Direct (LP/UV) and indirect photolysis (LP/UV + 40 mg/L H2O2) fluence based degradation rate constants (kf) obtained for hormones spiked individually and as
mixtures in laboratory grade water (LGW) and surface water (SW).

explains why it can be removed by direct photolysis from surface contribute to a decrease in the direct photolysis of the hormones
water to a greater extent than most of the other target hormones due to the interference of these parameters as light scavengers,
(shown below in Fig. 3). The value of quantum yield obtained in for all the target hormones except estriol. On the contrary, we
this study for 17b-estradiol (0.06 mol einstein1) agrees well with could observe a slight increase of the fluence based rate constants
the value reported by Rosenfeldt and Linden (0.04 mol einstein1) reported for mestranol, nonylphenol, progesterone, estrone, 17a-
[8]. Even though the quantum yield reported in this study for 17a- ethynyl-estradiol, and b-estradiol that can be explained due to
ethynylestradiol is higher than what was reported in the literature the photolysis of natural organic matter that can be leading to
(0.026 mol einstein1) [8,18], our study corroborates the findings the production of hydroxyl radicals. These findings concur with
that insignificant removal of both hormones (17b-estradiol and results obtained by Lin and Reinhard [11] when testing the photo-
17a-ethynylestradiol) is expected by low pressure direct degradation of 17b-estradiol, estriol, estrone, and 17a-ethinylest-
photolysis. radiol using a very different photosimulator (with emitted
The performance of direct and indirect UV photolysis for the wavelengths in the range of 290–700 nm). In addition, Zhang
degradation of the selected hormones was evaluated in terms of et al. [16] also reported an enhanced degradation of estrone and
competition effects, influence of water matrix composition, and 17b-estradiol in the presence of humic acids, Caupos et al. [10]
hydrogen peroxide concentrations. reported an photodegradation enhancement of estrone in the
Fig. 3 presents the direct and indirect photolysis fluence based presence of fulvic acids, and Leech et al. [9] reported a significant
rate constants, calculated for the selected hormones spiked indi- increase in the photodegradation of 17b-estradiol as a function of
vidually and as mixtures in laboratory grade water. In the indirect the dissolved organic carbon derived from humic acids of the
photolysis experiments, 40 mg/L of hydrogen peroxide was added Suwannee River with a threshold attained at approximately
to the laboratory grade water spiked with the hormones. An exam- 5 mg/L.
ple of one of the fluence related degradation profiles obtained and Since the results obtained in laboratory grade water (Fig. 3)
used to determine the degradation rate constant is presented in the showed that an increase in the overall photolysis rate constants
Supplementary data. could be expected using indirect photolysis, additional experi-
The results obtained show that higher degradation rate con- ments were conducted with the objective of understanding the ef-
stants of the selected hormones can be achieved by indirect pho- fect of using different hydrogen peroxide concentrations: 0, 20, 40,
tolysis, when 40 mg/L of hydrogen peroxide is used. Similar 60, 80, and 100 mg/L (Fig. 4), on the degradation of the hormones
degradation rate constants were obtained for the selected hor- spiked to the surface water.
mones spiked individually and as mixtures in laboratory grade For all the selected hormones, except estrone and progesterone,
water. Therefore, the competition for the UV light did not induce there is a clear advantage of adding hydrogen peroxide to increase
a significant effect on photolysis, for the studied hormones concen- the degradation rate constants (Fig. 4). Indirect photolysis using
trations and mixture combinations tested. 20 mg/L enhanced the degradation of nonylphenol and mestranol,
To understand the influence of the matrix compositions of real use of 80 mg/L enhanced the degradation of octylphenol, whereas
waters on the UV degradation rate constants, direct photolysis 100 mg/L of hydrogen peroxide enhanced the degradation of es-
experiments were also conducted using surface water. Fig. 3 shows triol, 17b-estradiol, 17a-ethynylestradiol, progesterone, and mes-
a comparison between the direct photolysis degradation rate con- tranol. The addition of 100 mg/L hydrogen peroxide seems to
stants obtained in laboratory grade water and surface water. optimize the overall removal of the group of hormones selected.
Very similar degradation rate constants were obtained for all Previous studies also reported an increase in the degradation rates
the selected hormones in laboratory grade water and surface obtained with increasing concentrations of hydrogen peroxide for
water, showing that the higher levels of total organic matter as 17a-ethynylestradiol and 17b-estradiol [8], as well as estrone
well as suspended and dissolved solids in surface water did not and 17b-estradiol [16]. These results are explained by the high
94 V.J. Pereira et al. / Separation and Purification Technology 95 (2012) 89–96

Fig. 4. Hydrogen peroxide concentration effect in the overall fluence based degradation rate constant (kf) obtained in surface water (SW).

second-order rate constant for 17a-ethynylestradiol and OH radi- while percent degradations lower that 37% were obtained for all
cals reported by Huber et al. [19] and by Rosenfeldt and Linden the other selected hormones. Indirect photolysis with addition of
[8] (9.8 ± 1.2  109 M1 s1 and 1.08 ± 0.23  1010 M1 s1, respec- 100 mg/L of hydrogen peroxide did not impact extensively the deg-
tively) and the high second-order rate constant reported for 17b- radation of progesterone and estrone but significantly increased
estradiol and OH radicals (1.41 ± 0.33  1010 M1 s1) [8]. the percent degradation of all the other hormones. The advanced
The percent degradation of the selected hormones using a UV oxidation process therefore proved to be more suitable to achieve
fluence of 1500 mJ/cm2 was therefore compared using direct and degradation of a broad range of hormones.
indirect photolysis. Indirect photolysis was achieved by addition
of 100 mg/L of hydrogen peroxide to the surface water spiked with 3.3. Multi-barrier approach
the mixture of the selected hormones (Fig. 5).
The results obtained (Fig. 5), show that direct photolysis will A multi-barrier treatment approach was tested by combining
only be extremely efficient to degrade progesterone and estrone, nanofiltration with direct and indirect photolysis in order to

Fig. 5. Removal efficiency of the individual processes and the multi-barrier approach.
V.J. Pereira et al. / Separation and Purification Technology 95 (2012) 89–96 95

evaluate the efficiency of the combined process for the removal of impact the degradation of most of the target hormones. A slight in-
the target hormones from surface water. In order to assess the crease of the direct photolysis rate constants were obtained in sur-
multi-barrier approach performance, the nanofiltration permeate face water compared to laboratory grade water probably due to
from the experiments described in Section 2.2 was subject to direct reaction of the LP UV light with natural organic matter present in
UV photolysis and indirect photolysis. The results obtained are the water and the consequent production of hydroxyl radicals. Ex-
summarized in Fig. 5. tremely similar direct and indirect photolysis rate constants were
As shown in Fig. 5 and described in Section 3.1, nanofiltration also obtained in surface water for individual compounds when com-
showed lower rejection degrees for estriol (38%), 17b-estradiol pared to mixtures, showing that competition for the ultraviolet light
(71%) and estrone (82%) when compared with the other hormones does not seem to impact the photolysis of hormones significantly.
studied. Further treatment using direct and indirect photolysis in- Both treatment processes tested – nanofiltration or low pres-
creased the removal of these hormones from surface water. sure advanced oxidation processes – could be used individually
The comparison of the results achieved with the combined pro- and are expected to provide a high removal degree of the target
cesses with the ones achieved with the individual processes dem- hormones. However, as demonstrated, the multi-barrier approach
onstrates that the multi-barrier approach may increase the (integration of nanofiltration with direct or indirect UV photolysis)
efficiency of the water treatment for the removal of certain com- can increase the overall hormone removal efficiency. The integra-
pounds such as 17b-estradiol and estrone as well as other organic tion of nanofiltration with direct UV photolysis is considered the
contaminants with similar properties. most efficient solution since it prevents the addition of hydrogen
The use of nanofiltration prior to the UV photolysis potentiated peroxide. The use of these treatment processes prior to the com-
the increase of the UV degradation efficiency, which was more pro- mon final disinfection has an additional advantage of lowering
nounced for the case of estriol. After nanofiltration, UV photolysis the chlorine dose needed to achieve final disinfection due to the
allowed a higher degradation of estriol when compared to the proven disinfection efficiency of photolysis needed and thus the le-
experiments where only UV photolysis was used. This can be ex- vel of disinfection by-products in the water that form by reaction
plained due to the removal of color and turbidity by nanofiltration, of chlorine with the natural organic matter and have been linked
which improves UV transmittance and reduces shielding of chem- to adverse health effects. In addition, nanofiltration highly de-
ical compounds by particulate matter. The nanofiltration perme- creases the natural organic matter present in the water and other
ates presented levels of color and turbidity below the detection particulate matter that may interfere with photolysis.
limits (<1 mg/L Pt–Co and <0.2NTU, respectively) and 90% UV This multi-barrier approach, the combination of nanofiltration
transmittance. with LP/UV photolysis, guarantees the production of water with
Even though the higher removal of the different hormones was extremely high chemical quality able to cope with drinking water
obtained by the combined nanofiltration and indirect photolysis microbial outbreaks (e.g. Cryptosporidium outbreaks) and future
using hydrogen peroxide, the combination of nanofiltration and di- more stringent regulations in terms of both chemical pollutants
rect photolysis proved to be sufficient for achieving high hormone and microbiological agents.
removals (close to the ones obtained for the combined process
with the H2O2 addition). Therefore, the combination of nanofiltra- Acknowledgments
tion and low pressure direct photolysis will prevent the addition of
H2O2, contributing to the overall economic efficiency and sustain- The authors thank Filipa Santos for technical assistance. We also
ability of the treatment process and preventing secondary water thank Trojan Technologies Inc. through Linha d’Água for providing
chemical contamination by possible by-product formation due to the collimated beam bench-scale reactor and VWR for the
reactions of the highly reactive OH radicals produced during indi- LaChromUltra system. Vanessa J. Pereira thanks Fundação para a
rect photolysis. Ciência e a Tecnologia for the grant BPD/26990/2006. Financial
support from the EEA Financial Mechanism, Empresa Portuguesa
das Águas Livres (EPAL), Município de Almada, and IBET (through
4. Conclusions project PT0012), is gratefully acknowledged. This work was sup-
ported by Fundação para a Ciência e a Tecnologia through the grant
Nanofiltration allowed for the rejection of the selected hor- PEst-OE/EQB/LA0004/2011.
mones with efficiencies higher than 71% (except estriol 38%) from
surface water, even using a membrane (NF270) with a molecular
Appendix A. Supplementary data
weight cut-off higher than the molecular weight of the target com-
pounds (mestranol, octylphenol, nonylphenol, progesterone, es-
Supplementary data associated with this article can be found,
trone, estriol, 17a-ethynylestradiol, and b-estradiol), which
in the online version, at http://dx.doi.org/10.1016/j.seppur.
allows for a significant high water treatment rate (approximately
2012.04.013.
90 l/(m2 h) for a transmembrane pressure of 8 bar), when com-
pared with nanofiltration membranes with lower molecular
References
weight cut-off. Hydrophobic interactions were found to be impor-
tant mechanisms governing the rejection of these hormones. High- [1] D.W. Kolpin, E.T. Furlong, M.T. Meyer, E.M. Thurman, et al., Environmental
er rejection efficiencies of estriol could probably be obtained using Science & Technology 36 (2002) 1202–1211.
a different membrane with a lower molecular weight cut-off. How- [2] B.D. Stanford, H.S. Weinberg, Journal of Chromatography A 1176 (2007) 26–36.
[3] T.A. Ternes, M. Stumpf, J. Mueller, K. Haberer, et al., Science of the Total
ever, the use of such membrane will result in lower water treat- Environment 225 (1999) 81–90.
ment rates. [4] B.D. Stanford, H.S. Weinberg, Environmental Science & Technology 44 (2010)
Low pressure direct photolysis with an ultraviolet fluence of 2994–3001.
[5] S.A. Snyder, T.L. Keith, D.A. Verbrugge, E.M. Snyder, et al., Environmental
1500 mJ/cm2 proved to be extremely effective to degrade only
Science & Technology 33 (1999) 2814–2820.
two of the selected hormones – progesterone and estrone. To [6] W.F. Jardim, C.C. Montagner, I.C. Pescara, G.A. Umbuzeiro, et al., Separation and
achieve higher removal efficiencies of all the other selected hor- Purification Technology 84 (2012) 3–8.
mones, indirect photolysis should be used with the addition of [7] I. Gültekin, N.H. Ince, Journal of Environmental Management 85 (2007) 816–
832.
100 mg/L of hydrogen peroxide, to produce the highly reactive [8] E.J. Rosenfeldt, K.G. Linden, Environmental Science & Technology 38 (2004)
and unselective hydroxyl radicals that were found to profoundly 5476–5483.
96 V.J. Pereira et al. / Separation and Purification Technology 95 (2012) 89–96

[9] D.M. Leech, M.T. Snyder, R.G. Wetzel, Science of the Total Environment 407 [26] A.J.C. Semião, A.I. Schäfer, Journal of Membrane Science 381 (2011) 132–141.
(2009) 2087–2092. [27] B. Van der Bruggen, M. Mänttäri, M. Nyström, Separation and Purification
[10] E. Caupos, P. Mazellier, J.-P. Croue, Water Research 45 (2011) 3341–3350. Technology 63 (2008) 251–263.
[11] A.Y.-C. Lin, M. Reinhard, Environmental Toxicology and Chemistry 24 (2005) [28] A.I. Schäfer, I. Akanyeti, A.J.C. Semião, Advances in Colloid and Interface
1303–1309. Science 164 (2011) 100–117.
[12] C. Sharpless, K. Linden, Environmental Science & Technology 37 (2003) 1933– [29] S. Sanches, A. Penetra, A. Rodrigues, E. Ferreira, et al., Separation and
1940. Purification Technology 94 (2012) 44–53.
[13] V.J. Pereira, K.G. Linden, H.S. Weinberg, Water Research 41 (2007) 4413–4423. [30] V.J. Pereira, M.C. Basílio, D. Fernandes, M. Domingues, et al., Water Research 43
[14] S. Sanches, M.T. Barreto Crespo, V.J. Pereira, Water Research 44 (2010) 1809– (2009) 3813–3819.
1818. [31] AWWA, Standard Methods for the Examination of Water and Wastewater,
[15] K. Linden, G. Shin, G. Faubert, W. Cairns, M. Sobsey, Environmental Science & Washington, DC, 1995.
Technology 36 (2002) 2519–2522. [32] M. Mezcua, A. Aguera, J. Lliberia, M. Cortes, et al., Journal of Chromatography A
[16] Y. Zhang, J.L. Zhou, B. Ning, Water Research 41 (2007) 19–26. 1109 (2006) 222–227.
[17] F.J. Benitez, F.J. Real, J.L. Acero, C. Garcia, Journal of Hazardous Materials 138 [33] R. Plumb, J. Castro-Perez, J. Granger, I. Beattie, et al., Rapid Communications in
(2006) 278–287. Mass Spectrometry 18 (2004) 2331–2337.
[18] S. Canonica, L. Meunier, U. von Gunten, Water Research 42 (2008) 121–128. [34] R. Plumb, TRAC-Trends in Analytical Chemistry, 2004, 23, V-V.
[19] M.M. Huber, S. Canonica, G.Y. Park, U. Von Gunten, Environmental Science & [35] J.R. Bolton, K.G. Linden, Journal of Environmental Engineering – ASCE 129
Technology 37 (2003) 1016–1024. (2003) 209–215.
[20] A.M. Comerton, R.C. Andrews, D.M. Bagley, P. Yang, Journal of Membrane [36] N.V. Klassen, D. Marchington, H.C.E. McGowan, Analytical Chemistry 66 (1994)
Science 303 (2007) 267–277. 2921–2925.
[21] I. Koyuncu, O.A. Arikan, M.R. Wiesner, C. Rice, Journal of Membrane Science [37] A.I. Schäfer, L.D. Nghiem, A. Meier, P.A. Neale, Separation and Purification
309 (2008) 94–101. Technology 73 (2010) 179–187.
[22] V. Yangali-Quintanilla, A. Sadmani, M. McConville, M. Kennedy, G. Amy, Water [38] A.M. Comerton, R.C. Andrews, D.M. Bagley, C. Hao, Journal of Membrane
Research 43 (2009) 2349–2362. Science 313 (2008) 323–335.
[23] B. Van der Bruggen, L. Braeken, C. Vandecasteele, Desalination 147 (2002) [39] Y. Yoon, P. Westerhoff, S.A. Snyder, E.C. Wert, Journal of Membrane Science
281–288. 270 (2006) 88–100.
[24] E.A. McCallum, H. Hyung, T.A. Do, C.-H. Huang, J.-H. Kim, Journal of Membrane [40] L.D. Nghiem, A.I. Schäfer, M. Elimelech, Environmental Science & Technology
Science 319 (2008) 38–43. 38 (2004) 1888–1896.
[25] A. Schäfer, L. Nghiem, T. Waite, Environmental Science & Technology 37 (2003)
182–188.

You might also like