Analytica Chimica Acta 559 (2006) 186–194

Behaviour of triazine herbicides and their hydroxylated and dealkylated

metabolites on a propazine-imprinted polymer
Comparative study in organic and aqueous media
R. Carabias-Martı́nez ∗ , E. Rodrı́guez-Gonzalo, E. Herrero-Hernández
Departamento de Quı́mica Analı́tica, Nutrición y Bromatologı́a, Facultad de Quı́mica, Universidad de Salamanca, 37008 Salamanca, Spain
Received 2 September 2005; received in revised form 25 November 2005; accepted 30 November 2005
Available online 17 January 2006

Abstract
A molecularly imprinted polymer (MIP) obtained by precipitation polymerization with methacrylic acid as functional monomer and propazine
as template was tested as a sorbent for the solid phase extraction of chloro- and methylthio-triazines and some of their dealkylated and hydroxylated
metabolites. In order to confirm the interaction between the functional monomer and the template, 1 H NMR (CD2 Cl2 ) analyses were made. It
was observed that the hydrogen and/or nitrogen of the amino group of the propazine were involved in the formation of hydrogen bonds with the
functional monomer.
The possibilities offered by the propazine-MIP for the extraction of triazine and derivatives were studied in organic and aqueous media and
different types of behaviour were observed. From organic medium, selective extractions of chloro- and methylthio-triazines were obtained after
optimization of the extraction process, with recoveries higher than 75% in both cases. However, the hydroxylated metabolites were not retained
on the propazine-MIP in organic medium.
In aqueous medium, the chloro-triazines and the dealkylated chloro-metabolites were retained on the propazine-MIP by specific interactions,
the recoveries obtained being a function of their polarity. Regarding the hydroxylated metabolites, these are retained on the propazine-MIP through
non-specific hydrogen-bridge interactions. The methylthio-triazines undergo poor molecular recognition in this medium.
© 2005 Elsevier B.V. All rights reserved.

Keywords: Molecularly imprinted polymers; Precipitation polymerization; Triazine herbicides; Dealkylated and hydroxylated metabolites

1. Introduction In water, parent triazine herbicides are subject to various

Triazine herbicides are some of the more commonly used
in Europe as selective pre- and post-emergence herbicides for
the control of broadleaf and grassy weeds in many agricultural
crops. Their prolonged use involves the risk of their retention
in crops and soils from which in turn, due to washing and
leaching processes, these substances pass to surface and ground
waters [1,2]. In previous work [3,4] evaluating these herbicides
in an agricultural zone close to the City of Salamanca (Spain),
we observed that surface and ground water samples contained
levels of atrazine, terbutryne and deethylatrazine (DEA) above
0.1 ␮g L−1 .
∗ Corresponding author. Fax: +923 294 574.
E-mail address: rcm@usal.es (R. Carabias-Martı́nez).
biotic and abiotic degradation processes such as photolysis,
oxidation, hydrolysis and biodegradation, leading to the dealky-
lation of amine groups, dechlorination, hydroxylation, deam-
ination, etc. [5,6]. For example, with respect to atrazine the
main degradation products in water are the dealkylated chloro-
metabolites: deethylatrazine (DEA) and deisopropylatrazine
(DIA) [7,8].
Analytical methods for the determination of pollutants in
environmental samples usually require a preconcentration step.
Solid-phase extraction (SPE) is a powerful tool for the pretreat-
ment of environmental waters and other kind of matrices, such
as biological fluids, plant tissue culture and forensic samples.
The main problem associated with SPE with ordinary stationary
phases is its low selectivity. More selective sorbents have been
developed, based on molecular recognition mechanisms. These
include immunosorbents that use selective antigen–antibody
interactions and molecularly imprinted polymers (MIPs).

0003-2670/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.aca.2005.11.081
 R. Carabias-Martı́nez et al. / Analytica Chimica Acta 559 (2006) 186–194 187

Different authors have employed MIPs to extract triazines
from different matrices such as water [9–11], vegetable extracts
[12], beef liver [13], urine, apple extract [14] and soil [15]
and also an atrazine sensor based on MIP modified electrodes
[16] has been developed. Most of these works attempt to study
the retention of parent compounds (chloro- and methylthio-
triazine herbicides) using MIPs obtained by bulk polymeriza-
tion. Although in some cases dealkylated chloro-metabolites
have also been included [11,17], few works have addressed
the behaviour of hydroxylated metabolites. Only the works of
Chapuis et al. [15,18] included hydroxyterbutylazine among
the target analytes for retention studies on a terbutylazine-MIP
obtained by bulk polymerization. As far as we are aware, there
are no reports that have addressed the retention behaviour of
N-dealkylated hydroxy-metabolites.
The aim of this work was to check the possibilities of
MIPs prepared by precipitation polymerization for the molec-
ular recognition of different triazinic derivatives (chloro- and
methylthio-triazines and some of their hydroxylated, dealky-
lated and dealkylated hydroxy-metabolites) when the sample
passing through the sorbent is organic or aqueous. The mecha-
nism of the retention in both media was studied and the results
were compared with those reported by other authors with MIPs
obtained by bulk polymerization.
The selectivity of the MIP for the specific retention of these
metabolites in organic and aqueous solutions was assessed
against non-triazine herbicides in frequent use: lenacil, chlor-
sulfuron, isoproturon and linuron.
2. Experimental
2.1. Chemicals
The herbicides and the metabolites were obtained from Dr.
Ehrenstorfer (Augsburg, Germany) and were used without fur-
ther purification. The triazine compounds are listed in Table 1.
Other herbicides were: chlorsulfuron (Cls) 1-(2-chlorophenyl-
sulfonyl)-3-(4-methoxy-6-methyl-1,3,5-triazin-2-yl)urea CAS
RN [64902-72-3]; isoproturon (Ipn) 3-(4-isopropylphenyl)-
1,1-dimethylurea CAS RN [34123-59-6]; linuron (Lin) 3-
(3,4-dichlorophenyl)-1-methoxy-1-methylurea CAS RN [330-
55-2] and lenacil (Len) 3-cyclohexyl-1,5,6,7-tetrahydrocyclo-
pentapyrimidine-2,4(3H)-dione, CAS RN [2164-08-1].
Stock solutions of each herbicide and metabolite were
prepared in acetonitrile at 500 ␮g mL−1 except in the case
of the hydroxylated metabolites which were prepared in
acetonitrile–0.1 M hydrochloric acid (80:20, v/v) at a concen-
tration of 200 ␮g mL−1 .
Methacrylic acid (MAA), ethylene glycol dimethacrylate
(EGDMA) and 2,2 -Azobis(2-methyl-propionitrile) (AIBN)
were obtained from Acros Organics (Geel, Belgium). The
organic solvents, toluene, dichloromethane, acetonitrile and
methanol were of HPLC grade (Merck) and were used as
received. Ultra-high quality UHQ water (purity 18.2 M
cm at
25 ◦ C) was obtained with an Elgastat UHQ water purification
system (Elga Ltd., Bucks, UK). All other chemicals were of
analytical reagent grade.
2.2. Synthesis and characterization of the imprinted
polymer
The procedure followed to obtain the MIP by precipitation
was as follows: the template, propazine (1 mmol) and the func-
tional monomer, MAA (4 mmol), were added to a glass tube
and after 5 min EGDMA (20 mmol), toluene (30 mL) and AIBN
(0.2 mmol) were added. The solution was degassed with nitro-
gen for 5 min and the tubes were closed and sealed under this
atmosphere. Microspheres were obtained by precipitation poly-
merization in a water bath at 60 ◦ C for 10 h with stirring at
350 rpm. The resulting microspheres were washed twice with
acetone to remove the fine particles. Template was removed by
Soxhlet extraction with a methanol:acetic acid (9:1, v/v) mixture
until no template was found in the washing solution (analyzed
by HPLC-DAD UV).
Blank polymers or non-imprinted polymers (NIP) were pre-
pared in the same way but without the addition of the template
to the polymerization mixture.
Characterization of the imprinted polymer was made by
nitrogen sorption and scanning electron microscopy measure-
ments. Nitrogen sorption measurements were performed on a
Gemini 2375 (Micromeritics Instrument Corporation, Norcross,
GA) at the Department of Inorganic Chemistry (University

Table 1
General structure and some properties of triazine herbicides and metabolites considered in this work
Compounds R1 R2 R3 log Kow a pKa

Deisopropylhydroxyatrazine (DIHA) –OH –NH2 –NH–CH2 –CH3 −0.1 4.65

Deethylhydroxyatrazine (DEHA) –OH –NH–CH(CH3 )2 –NH2 0.2 4.57–4.75
Deisopropylatrazine (DIA) –Cl –NH2 –NH–CH2 –CH3 1.15 1.30–1.58
Deethylatrazine (DEA) –Cl –NH–CH(CH3 )2 –NH2 1.52 1.30–1.65
Hydroxyterbutylazine (HT) –OH –NH–C(CH3 )3 –NH–CH2 –CH3 – 5.2
Simazine (Smz) –Cl –NH–CH2 –CH3 –NH–CH2 –CH3 2.18 1.65
Atrazine (Atz) –Cl –NH–CH(CH3 )2 –NH–CH2 –CH3 2.5 1.68
Propazine (Ppz) –Cl –NH–CH(CH3 )2 –NH–CH(CH3 )2 2.93 1.85
Terbutylazine (Tbz) –Cl –NH–C(CH3 )3 –NH–CH2 –CH3 3.21 2.00
Prometryn (Pmn) –SCH3 –NH–CH(CH3 )2 –NH–CH(CH3 )2 3.51 4.05
Terbutryn (Tbn) –SCH3 –NH–C(CH3 )3 –NH–CH2 –CH3 3.65 4.38
a log Kow : n-octanol–water partition coefficients, defined as the ratio of the equilibrium concentrations of a dissolved substance in two immiscible solvents.
188 R. Carabias-Martı́nez et al. / Analytica Chimica Acta 559 (2006) 186–194
of Salamanca). The sample (ca. 80–100 mg) was previously
degassed in flowing nitrogen at 100 ◦ C for 12 h in a Flow-
Prep 060 (Micromeritics) in order to remove physisorbed water.
The specific surface area was evaluated using the BET method,
and specific pore volume and the average pore diameter were
assessed according to the BJH theory. The physical character-
istics of the propazine-MIP material are as follows: specific
surface area, 57.73 m2 /g; specific pore volume, 0.063 cm3 /g;
average pore diameter, 3.78 nm. The polymer particles had a
regular spherical morphology, with an average particle diameter
of ∼1 ␮m as determined by scanning electron microscopy using
a Zeiss DSM 949 (Zeiss, Oberkochen, Germany) device at the
Servicio de Microscopı́a Electrónica (University of Salamanca).
2.3. Procedure for preconcentrating triazines in organic
medium
For the experiments on extraction, 200 mg of polymer were
dry-packed in empty solid-phase extraction cartridges of 3 mL
and conditioned with 5.0 mL of a mixture acetonitrile:acetic acid
(9:1, v/v) and 5.0 mL of toluene. The MIP columns were then
loaded with 2.0 mL of a mixture of triazines and metabolites at
a concentration of 200 ␮g L−1 in toluene. After an exhaustive
drying step, 10.0 mL of dichloromethane was percolated through
the sorbent in order to eliminate non-specific interactions. The
target analytes were eluted from the cartridge with 10 mL of an
acetonitrile–acetic acid (9:1, v/v) mixture and analysed using
an HPLC-DAD UV system. Adjustable-volume pipettors with
disposable tips were used to load the sample, washing solvent
and eluent into the cartridge; percolation was accomplished by
applying a vacuum of −15 mmHg.
2.4. Procedure for preconcentrating triazines in aqueous
medium
A procedure similar to that describe in the previous section
was used for the extraction of triazines from aqueous sam-
ples, with the difference that in the final conditioning step the
solvent employed was water instead of toluene. A volume of
2.0 mL of aqueous sample was passed through the MIP, after
which an exhaustive drying step was implemented (under a vac-
uum of −15 mmHg for 30 min). For the washing step, 5 mL
of dichloromethane were employed. Analyte elution was per-
formed under the same conditions as those described in the
previous section.
2.5. Chromatographic conditions
HPLC-DAD UV was performed on an HP 1100 Series
chromatograph from Hewlett Packard (Waldbronn, Germany)
equipped with a quaternary pump, a membrane degasser, an
autosampler and a diode-array UV detector. The system was
controlled by a HP ChemStation, which also performed data
collection from the diode array UV detector (DAD UV) and
the quantitative measurements. The analytical column was a
250 mm × 4.0 mm i.d. Spherisorb S5 ODS2 packed with 5 ␮m
particles (Waters, Milford, MA, USA). The diode-array UV
detector was set at 210 and 245 nm. UV spectra were recorded
in the 190–400 nm range.
A gradient acetonitrile (solvent A)–5 mM phosphate buffer
pH 7.2 (solvent B) was used: from 5 to 45% of solvent A in
15 min, from 45 to 50% of solvent A in 15 min and returned to
initial conditions in 2 min, with 3 min for column equilibration.
The flow rate was 1 mL min−1 and the volume injected was
100 ␮L. The analytical column was thermostated at 25 ◦ C.
The elution order (retention time in brackets) was as follows:
DIHA (7.32 min), DEHA (8.46 min), DIA (11.20 min), DEA
(13.38 min), HT (14.91 min), simazine (17.15 min), atrazine
(19.73 min), propazine (23.64 min), terbutylazine (25.41 min),
prometryn (29.43 min) and terbutryn (30.87 min).
2.6. 1 H NMR study
Sample was prepared with a fixed concentration of propazine
(58.0 mmol L−1 ) in CD2 Cl2 and varying concentrations of MAA
(from 0 to 246 mmol L−1 ). 1 H NMR spectra were recorded at
room temperature on a Bruker Avance DRX 400 MHz NMR
spectrometer (Bruker, Karlsruhe, Germany).
3. Results and discussion
A MIP with propazine as template has been developed
and its use as sorbent for the solid-phase extraction of tri-
azine herbicides and some of their major metabolites has been
studied. The analytes included in this study belong to the
triazine family and can be classified in the following sub-
groups: chloro-triazines (simazine, atrazine, propazine and ter-
butylazine), methylthio-triazines (prometryn and terbutryn),
hydroxy-metabolites (hydroxyterbutylazine, HT) dealkylated
chloro-metabolites (deisopropylatrazine, DIA and deethyla-
trazine, DEA) and dealkylated hydroxy-metabolites (deiso-
propylhydroxyatrazine, DIHA and deethylhydroxyatrazine,
DEHA). Their chemical structures and some of their properties
are shown in Table 1.
The propazine-MIP was obtained by precipitation polymer-
ization; this way of obtaining the desired product is relatively
simple and has the advantage of not requiring a polymer grinding
step, which could alter some of the polymer’s cavities, prevent-
ing such sites from becoming available for later binding to the
analytes.
3.1. 1 H NMR study
Study of the intermolecular interaction between the template
and functional monomer prior to polymerization is interesting to
determine the recognition mechanism of the imprinted polymers
because the recognition of imprinted polymers is based on these
interactions.
Here the interactions between MAA and the template were
assessed experimentally using 1 H NMR analysis. The effect of
the MAA on the propazine spectrum was evaluated by recording
the spectra in the presence of different ratios of MAA/propazine.
Upon plotting the chemical shifts (δ) that occur upon varying
the concentration of MAA in CD2 Cl2 (Fig. 1(a)), it may be seen
 R. Carabias-Martı́nez et al. / Analytica Chimica Acta 559 (2006) 186–194
Fig. 1. 1 H NMR study in CD2 Cl2 . (a) Chemical shift, δ, of propazine hydrogens against the concentration of the functional monomer; (b) chemical shift, δ, of
hydrogen H3 of propazine against the functional monomer/template ratio.
that the only hydrogen that undergoes a shift corresponds to the
amine hydrogen. This suggests that amine hydrogen would be
involved in the formation of hydrogen bridges with the func-
tional monomer. When the shift is plotted against the functional
monomer/template ratio, it is seen that the stoichiometric ratio
between them is equal to unity (Fig. 1(b)). Bearing in mind
methacrylic acid is also a hydrogen donor, it is likely that the
nitrogen atom of the triazine ring will be involved in hydrogen
bonding as a hydrogen acceptor. The fact that the stoichiometry
observed was 1:1 suggests that one molecule of MAA would
bind to one molecule of propazine through the formation of two
hydrogen bonds.
3.2. Study of the retention process in organic medium
Initially, we studied the ability of the MIPs obtained by
precipitation polymerization to specifically recognise triazines
and metabolites from different organic media (toluene and
dichloromethane). This was evaluated working in parallel
with cartridges filled with imprinted polymer (propazine-MIP)
and blank or non-imprinted polymer (NIP), both obtained as
described in the experimental section. The loading step was
accomplished using a standard mixture of herbicides in toluene
or dichloromethane spiked with an approximate concentration
of a 200 ␮g L−1 of each of the triazines and metabolites studied.
In order to prevent non-specific interactions between the ana-
lytes and the MIP, a cartridge-washing step was implemented
using, initially, 10 mL of toluene because this was the solvent
employed to obtain the polymer. A mixture of acetonitrile and
acetic acid (9:1, v/v) was used to elute specifically retained ana-
lytes.
The recoveries obtained when the sample was passed through
the MIP dissolved in toluene were higher than those obtained
for samples in dichloromethane. These results are in good agree-
ment with those reported in literature [11,18] in which the best
medium for molecular recognition is that used for the synthesis
of the MIP.
189
For samples in toluene, different types of behaviour were
observed: (a) the chloro- and methylthio-triazines were specifi-
cally retained on the MIP, (b) the dealkylated chloro-metabolites
– DEA and DIA – were retained in both the imprinted and the
non-imprinted polymer and (c) the hydroxylated metabolites –
HT, DIHA and DEHA – were not retained neither on the MIP
nor on the NIP.
Accordingly, the results obtained indicate that toluene is a
suitable organic medium for the specific retention of triazines.
They also suggest that dichloromethane can be used as a good
washing solvent for eliminating non-specific interactions.
3.2.1. Influence of the amount of sorbent used
To check the influence of this in the recovery values, a series
of empty SPE cartridges were filled with different amounts of
polymer. After the cartridges had been conditioned, 2 mL of
toluene with 200 ␮g L−1 of each analyte were passed through
them. For most of the triazines, the recoveries obtained were not
significantly better when amounts of sorbent above 200 mg were
used, whereas for the dealkylated chloro-metabolites DIA and
DEA recovery continued to increase with the increase in polymer
mass. In the case of the hydroxylated metabolites DIHA, DEHA
and HT, the recoveries obtained were lower than 10% in all cases.
Taking into account the physical characteristics observed for
the imprinted polymer (see Section 2.2), the difficulty in passing
the sample through the system increases with the increase in
polymer mass. Accordingly, it is recommended that 200 mg of
polymer be used (Fig. 2).
3.2.2. Influence of the washing solvent
The washing solvents assayed here to remove non-specific
interactions were selected as a function of their dielectric con-
stant (␧) and the hydrogen bond parameter (HBP). These sol-
vents were toluene, dichloromethane and acetonitrile whose ␧
values are 2.5, 9.1 and 36.2, respectively, while their HBP values
are, respectively, 3.6, 2.2 and 7.0. Taking into account these lat-
ter values, acetonitrile would be expected to be the solvent that
190 R. Carabias-Martı́nez et al. / Analytica Chimica Acta 559 (2006) 186–194

Fig. 2. Influence of polymer mass on the recoveries. Sample: 2 mL of toluene spiked with 200 ␮g L−1 of each herbicide (n = 2). The MIP was washed with 10 mL
of toluene.

removes the greatest number of hydrogen bridges and hence
would suppress the greatest number of specific interactions in
the washing step; dichloromethane lies at the opposite end of this
range. The results obtained (Fig. 3) on using these washing sol-
vents on the MIP and in the blank (NIP) are consistent with the
HBP values of the solvents. When the sample was percolated into
the MIP and washing step was carried out with dichloromethane
or toluene, higher recovery values were obtained than when it
was carried out with acetonitrile (Fig. 3(a)). In contrast, when
the same experiment was performed using a NIP as sorbent and
dichloromethane as washing solvent, the recoveries obtained
were always lower than those achieved when washing was car-
ried out with toluene (Fig. 3(b)) or acetonitrile (Fig. 3(a)). Thus,
dichloromethane is the solvent of choice for the most efficient
removal of non-specific interactions (Fig. 3(c)).
To determine the selectivity of the imprinted polymer
against other analytes in organic medium and under the opti-
mum conditions found up to this point, four widely used
herbicides were employed: chlorsulfuron (log Kow = −0.99),
lenacil (log Kow = 2.31), isoproturon (log Kow = 2.50) and lin-
uron (log Kow = 3.0). All have chemical characteristics different
from those of the template and they cover a broad range of
polarities. Table 2 shows the recoveries obtained for chloro-
and methylthio-triazines and the dealkylated chloro-metabolites
DIA and DEA. As may be seen, none of the other herbicides
included to study selectivity was retained in any case. From
the results obtained it may be inferred that, when the sam-
ple is passed through the cartridge in toluene and is washed
with dichloromethane, the propazine-MIP selectively retains the
chloro- and methylthio-triazines, with similar recovery values
for both, ranging between 89 and 77%.
The behaviour of the methylthio-triazines on the propazine-
MIP obtained by precipitation observed here differs from that

Table 2
Recoveries of some triazines and non-related herbicides
Recoveries (%) ± S.D.a

MIP NIP

DIA 96 ± 11 66 ± 5
DEA 93 ± 7 43 ± 6
Simazine 87 ± 10 5 ± 0.7
Atrazine 79 ± 8 –
Propazine 83 ± 9 –
Terbutylazine 77 ± 10 –
Prometryn 89 ± 7 –
Terbutryn 84 ± 9 –
Chlorsulfuron 2± 0.3 –
Lenacil – –
Isoproturon – –
Linuron – –
Fig. 3. Influence of washing solvent in the recoveries of the analytes from
organic medium. Washing solvent: (a) acetonitrile, (b) toluene and (c) Sample: 2 mL of toluene spiked with 200 ␮g L−1 . Washing solvent: 10 mL of
dichloromethane. Sample: 2 mL of toluene spiked with 200 ␮g L−1 of each her- dichloromethane. (–) Not detected.
a S.D., standard deviations for n = 3.
bicide (n = 3).
 R. Carabias-Martı́nez et al. / Analytica Chimica Acta 559 (2006) 186–194
reported by other authors [17] for imprinted polymers obtained
by bulk polymerization, in which the –SCH3 group of the
methylthio-triazines seems to exert a steric effect that hin-
ders their specific retention on the polymer. This difference in
behaviour seems to be related to the way in which the polymer
was obtained.
Regarding the dealkylated chloro-metabolites DIA and DEA,
recoveries higher than 93% were obtained but it was observed
that these compounds were retained by specific and non-specific
interactions. In contrast, the hydroxylated metabolites DIHA,
DEHA and HT were not retained on the MIP. Similar results
have been reported by other authors [18] using an imprinted
polymer obtained by bulk polymerization.
3.3. Study of the process of retention in aqueous medium
With a view to being able to use the MIP for the pretreat-
ment of environmental waters contaminated by herbicides, the
behaviour of this imprinted polymer in the preconcentration of
triazines from aqueous samples was studied.
The behaviour of different chloro- and methylthio-triazines
in aqueous medium using MIPs obtained by bulk polymerization
has been described [10,11], but few authors have addressed the
joint behaviour of the hydroxylated metabolites of these com-
pounds [18].
The propazine-MIP was exposed to a standard aqueous solu-
tion containing the eleven herbicides. After an exhaustive drying
step, the cartridge was washed with toluene and eluted with a
mixture of acetonitrile–acetic acid (9:1, v/v) (Fig. 4(a)). When
the same experiment was carried out using a non-imprinted poly-
mer (NIP) as sorbent, the recoveries were very similar to those
obtained when extraction was performed with the imprinted
Fig. 4. Influence of washing solvent in the recoveries of the analytes from aque-
ous medium. Washing solvent: 10 mL of (a) toluene and (b) dichloromethane.
Sample: 2 mL of water spiked with 200 ␮g L−1 of each herbicide (n = 3).
191
polymer (Fig. 4(a)), suggesting that the interactions occurring
on the MIP were not specific.
Bearing in mind the results obtained previously, the ability
of dichloromethane as a washing solvent to disrupt non-specific
interactions was studied when triazines were passed through
the cartridges in aqueous medium. The recoveries obtained
(Fig. 4(b)) showed that by using dichloromethane as the wash-
ing solvent all the triazines were removed from the NIP with the
exception of the hydroxylated metabolites, regardless of whether
they were dealkylated or not. Thus, under the experimental
conditions employed three different types of behaviour were
observed: (a) the chloro-triazines and the dealkylated chloro-
metabolites – DIA and DEA – were retained by specific inter-
actions; (b) the hydroxylated metabolites – DIHA, DEHA and
HT – were retained by non-specific interactions on the poly-
meric matrix; (c) the methylthio-triazines were not extracted,
instead they were washed out of the cartridge by the action of
dichloromethane. The retention of the hydroxylated metabolites
by the NIP seems to be related to the negative charge generated
by the hydroxyl groups, which may act as donors and acceptors
of hydrogen bridges with the molecules of the NIP [15].
3.3.1. Effect of the drying step
The retention of the analytes on a sorbent from an aque-
ous medium may be strongly affected by the presence of water
embedded in the sorbent after passage of the sample. Accord-
ingly, we next studied how the drying time affected the retention
of the analytes in both the MIP and NIP. The sorbent was dried by
applying a vacuum of −15 mmHg. Fig. 5 shows the recoveries
Fig. 5. Effect of MIP and NIP drying time. Sample: 2 mL of water spiked
with 200 ␮g L−1 of each herbicide (n = 3). Washing solvent: 10 mL of
dichloromethane. Elution with 10 mL of a mixture acetonitrile:acetic acid (9:1,
v/v).
192 R. Carabias-Martı́nez et al. / Analytica Chimica Acta 559 (2006) 186–194

Table 3
Influence of the volume of dichlorometane used in the washing step
Percentage of analyte removed in each fractiona

First fraction, Second Third fraction, Fourth Fifth fraction, Total

2 mL fraction, 3 mL 5 mL fraction 5 mL 5 mL washed

Triazines and metabolites

DIHA – – – – – –
DEHA – – – – – –
DIA – – – – – –
DEA – – – – – –
HT – – – – – –
Simazine – – – 23 14 37
Atrazine – – 10 24 16 50
Propazine 18 22 13 16 11 80
Terbutylazine 25 21 11 15 8 80
Prometryn 39 27 15 7 5 93
Terbutryn 54 26 10 3 – 93
Other herbicides
Chlorsulfuron 63 25 – – – 88
Lenacil 62 24 – – – 86
Isoproturon 88 – – – – 88
Linuron 96 – – – – 96

Sample: 2 mL of water spiked with 200 ␮g L−1 of each herbicide.

a Mean value of two replicates.

obtained on using 3 or 30 min for sorbent drying. As the drying
time increased so did the recoveries of all the analytes, this effect
being more pronounced in the case of the chloro-triazines. When
the experiments were carried out with a NIP, the effect was less
marked and a slight decrease in the recoveries of the hydroxy-
lated metabolites was observed upon increasing the drying time
and the low retention of the chloro-triazines was eliminated.
3.3.2. Effect of volume of washing solvent on MIP
selectivity
The selectivity of the imprinted polymer for specifically
extracting triazines from aqueous solutions was assessed against
the above non-triazine herbicides—lenacil, chlorsulfuron, iso-
proturon and linuron. To accomplish this, the MIP was washed
successively with increasing volumes of dichloromethane and
the fractions thus obtained were analysed chromatographically
(Table 3). It was observed that the non-triazine herbicides
were eluted efficiently (>80%) in the first two fractions of
dichloromethane. In contrast, in the case of the triazine ana-
lytes the order of elution in the washing step was inversely
related to their hydrophobicity (Table 1). Thus, the methylthio-
triazines prometryn and terbutryn were removed at 39% and
54%, respectively, in the first dichloromethane fraction and
were almost completely washed from the MIP with 10 mL of
dichloromethane.
Chlorsulfuron, with a log Kow value of −0.99, lower than
that of DIA (log Kow = 1.15), was eluted from the sorbent when a
washing volume of 2 mL of dichloromethane was used, whereas
DIA was not washed out even with 20 mL.
These results show that washing with dichloromethane has
two effects: on one hand, the solvent washes analytes that do
not belong to the triazine family because they are retained on
the polymer through hydrophobic interactions and, on the other,
it generates an organic medium in which the specific inter-
actions are developed that facilitate the retention of triazines
on the MIP. The competition between these two effects has
unfavourable results for the more apolar triazines and the larger
ones, the methylthio-triazines, which elute in the organic wash-
ing medium before developing specific interactions with the
MIP.
These results led us to employ a compromise solution in
which, using a washing step with 5 mL of dichloromethane,
the chloro-triazines and the dealkylated chloro-metabolites can
interact in a specific way with the MIP, recoveries ranging
between 38% for terbutylazine and 96% for DIA being obtained.
In the case of the methylthio-triazines, the recovery values were
lower than 12% since these were the most apolar compounds.
For the hydroxylated metabolites, DIHA, DEHA and HT, recov-
eries higher than 92% were obtained, although for these analytes
no specific recognition by the MIP was observed and their reten-
tion must be due non-specific hydrophilic interactions (hydrogen
bridges) with the polymeric matrix, which are not eliminated by
dichloromethane.
The results indicate that when the sample is passed through
the cartridge in aqueous medium the analytes are retained
by non-selective interactions whereas later washing with
dichloromethane generates an organic medium in which non-
selective interactions are converted into that are specific for the
chloro-triazines and dealkylated chloro-metabolites. The recov-
ery values obtained for the methylthio-triazines should not be
related to a greater retention of these analytes but, instead, to
the fact that they are the most apolar of the triazines studied
and are therefore removed more easily in the washing step with
dichloromethane.
 R. Carabias-Martı́nez et al. / Analytica Chimica Acta 559 (2006) 186–194 193

Table 4 such that its log Kow value would be expected to lie between
Behaviour of triazine herbicides and related metabolites on the propazine-MIP these values. However, the behaviour of HT in both the MIP and
Analytes Organic Aqueous NIP was similar to that of compounds of greater polarity, such
medium medium as DIHA (log Kow = −0.1) and DEHA (log Kow = 0.2).
MIP NIP MIP NIP The propazine-MIP exhibited selective binding for chloro-
Chloro-triazines: triazines and dealkylated chloro-metabolites. It can also be used
Simazine +++ − ++ − in the extraction of the hydroxy- and dealkylated hydroxy-
Atrazazine derivatives of the triazines since these compounds were retained
Propazine in a non-specific way by hydrogen bridge interactions that were
Terbutylazine
not eliminated in the step involving washing with dichloro-
Methylthio-triazines: methane.
Prometryn +++ − − −
Terbutryn
4. Conclusions
Hydroxy-metabolites:
Hydroxyterbutylazine (HT) − − +++ +++
The propazine-MIP prepared by precipitation polymeriza-
Dealkylated chloro-metabolites: tion showed some advantages over MIPs obtained by bulk
Deisopropylatrazine (DIA) +++ ++ +++ −
polymerization. It exhibited highly selective binding for tri-
Deethylatrazine (DEA)
azines, dealkylated chloro-metabolites, dealkylated hydroxy-
Dealkylated hydroxy-metabolites: metabolites and hydroxy-metabolites and thus, it can be used
Deisopropylhydroxyatrazine (DIHA) − − +++ +++
Deethylhydroxyatrazine (DEHA)
to preconcentrate these analytes. However, different types of
behaviour were observed as a function of the sample medium
Recoveries: +++, >80%; 80% > ++ > 50%; − <20%. and of the type of analyte being studied. Accordingly, depend-
ing on the type of triazine to be preconcentrated, the appropriate
3.4. Behaviour in organic medium versus aqueous medium medium should organic or aqueous. Additionally, it may be con-
cluded that the behaviour of the hydroxy-metabolites (HT) and
The possibilities of using the propazine-MIP as a selective dealkylated hydroxy-metabolites (DEHA and DIHA) is mainly
sorbent for triazines and metabolites depend on the extraction governed by the presence of the hydroxyl group and hence the
medium and on the type of analyte being studied. A summary three metabolites show similar behaviour.
of the behaviour of the MIP under the conditions employed in Other non-related herbicides (lenacil, chlorsulfuron, isopro-
the present work can be found in Table 4. turon, linuron) were not specifically retained.

3.4.1. Organic medium Acknowledgements

In organic medium the propazine-MIP prepared by precipi-
tation polymerization exhibited highly selective binding for the
chloro- and methylthio-triazines, with recoveries higher than
75% in both cases.
The dealkylated chloro-metabolites – DIA and DEA – are
retained by a combination of specific and non-specific interac-
tions, although these latter can be taken advantage of to carry
out extraction since they were not eliminated in the washing
step. However, in this medium it was not possible to extract the
hydroxy-derivatives (DIHA, DEHA and HT) since these do not
interact with the MIP in a specific way and their non-specific
interactions with the polymeric matrix are weak.
3.4.2. Aqueous medium
In aqueous medium, choice of the volume of the washing sol-
vent is more critical than in organic medium. This is because, on
one hand, the medium eliminates (washes out) the non-specific
hydrophobic interactions, and, on the other, it provides a suitable
medium for molecular recognition of the target analytes.
In this medium, the behaviour of the hydroxy-metabolites
(HT) and dealkylated hydroxy-metabolites (DEHA and DIHA)
is mainly governed by the presence of the hydroxyl group and
these three compounds show similar behaviour despite their
differences in polarity. Thus, chromatographically HT elutes
between DEA (log Kow = 1.52) and simazine (log Kow = 2.18),
This work was supported by the MCYT-DGI (Project BQU-
2002-02314) and by the Consejerı́a de Cultura y Turismo, Junta
de Castilla y León, and the European Union (European Social
Fund, Project SA044/02).
References
[1] L.W. Canter, R.C. Knox, D.M. Fairchild, Ground Water Quality Protec-
tion, Lewis, Michigan, 1987.
[2] J.E. Barbash, E.A. Resek, Pesticides in Ground Water. Distribution,
Trends and Governing Factors, Ann Arbor, Michigan, 1996.
[3] R. Carabias-Martı́nez, E. Rodrı́guez-Gonzalo, M.E. Fernández-Laespada,
L. Calvo Seronero, Water Res. 37 (2003) 928.
[4] R. Carabias-Martı́nez, E. Rodrı́guez-Gonzalo, E. Herrero-Hernández, F.J.
Sánchez-San Román, M.G. Prado-Flores, J. Chromatogr. A 950 (2002)
157.
[5] D.W. Kolpin, E.M. Thurman, D.A. Goolsby, Environ. Sci. Technol. 30
(1996) 335.
[6] P. Schmitt, A.W. Garrison, D. Freitag, A. Kettrup, J. Chromatogr. A 723
(1996) 169.
[7] E.M. Thurman, M.T. Meyer, M.S. Mills, L.R. Zimmerman, C.A. Perry,
D.A. Goolsby, Environ. Sci. Technol. 28 (1994) 2267.
[8] M.S. Mills, E.M. Thurman, Environ. Sci. Technol. 28 (1994) 600.
[9] R. Koeber, C. Fleisher, F. Lanza, K.S. Boos, B. Sellergren, D. Barceló,
Anal. Chem. 73 (2001) 2437.
[10] T. Pap, V. Horvath, A. Tolokan, G. Horvai, B. Sellergren, J. Chromatogr.
A 973 (2002) 1.
194 R. Carabias-Martı́nez et al. / Analytica Chimica Acta 559 (2006) 186–194

[11] I. Ferrer, F. Lanza, A. Tolokan, V. Horvath, B. Sellergren, G. Horvai,
D. Barcelo, Anal. Chem. 72 (2000) 3934.
[12] C. Cacho, E. Turiel, A. Martin-Esteban, C. Perez-Conde, C. Camara,
Anal. Bioanal. Chem. 376 (2003) 491.
[13] M.T. Muldoon, L.H. Stanker, Anal. Chem. 69 (1997) 803.
[14] B. Bjarnason, L. Chimuka, O. Ramstrom, Anal. Chem. 71 (1999)
2152.
[15] F. Chapuis, V. Pichon, F. Lanza, B. Sellergren, M.C. Hennion, J. Chro-
matogr. B 804 (2004) 93.
[16] R. Shoji, T. Takeuchi, I. Kubo, Anal. Chem. 75 (2003) 4882.
[17] E. Turiel, A. Martı́n-Esteban, P. Fernández, C. Pérez-Conde, C. Cámara,
Anal. Chem. 73 (2001) 5133.
[18] F. Chapuis, V. Pichon, F. Lanza, S. Sellergren, M.C. Hennion, J. Chro-
matogr. A 999 (2003) 23.