Ola

ND

a

ARRAA

KOWsIE

1

oloww

htedces

iiiidb[

0h

Journal of Chromatography A, 1286 (2013) 16– 21

Contents lists available at SciVerse ScienceDirect

Journal of Chromatography A

j our na l ho me p ag e: www.elsev ier .com/ locate /chroma

n-line weak cationic mixed-mode solid-phase extraction coupled toiquid chromatography–mass spectrometry to determine illicit drugst low concentration levels from environmental waters

úria Fontanals ∗, Francesc Borrull, Rosa M. Marcéepartament de Química Analítica i Química Orgànica, Universitat Rovira i Virgili, Campus Sescelades, Marcel·lí Domingo, s/n, 43007 Tarragona, Spain

r t i c l e i n f o

rticle history:eceived 27 November 2012eceived in revised form 21 February 2013ccepted 22 February 2013vailable online 27 February 2013

a b s t r a c t

This study presents a fully automated method based on on-line solid-phase extraction coupled to liquidchromatography with mass spectrometry detection (on-line-SPE-LC–MS) to determine illicit drugs inenvironmental water samples. The mixed-mode Oasis WCX sorbent used in an optimised protocol allowsthe addition of an effective washing step with 0.5 mL of methanol, which washed out all the interferencesretained by reversed-phase interactions and helped to reduce the matrix effect, while the cationic target

eywords:n-line solid-phase extractioneak cationic mixed-mode polymeric

orbent

analytes remained bound and could then be selectively eluted with recovery values near to 100%. Thismethod was successfully applied to the analysis of 10 mL of environmental water (river and sewage)spiked at low ng L−1 levels of the analytes, with recoveries from 50 to 100% for all studied analytes. Themethod was also validated with river water samples with figures of merit comparable to those of the

etho

llicit drugsnvironmental water samples

off-line SPE-LC–MS/MS m

. Introduction

In recent years, it has become a well-accepted fact that the usef a large number of drugs in various areas of our lives inevitablyeads to the release of these drugs into the environment and theirccurrence in the environment [1]. Recently, concern has grownith respect to the presence of illicit drugs and metabolites inastewater and surface water around the world [2–6].

In view of this, in recent years efforts in analytical chemistryave focused on the development of different analytical methodso determine these drugs at the low concentration levels found innvironmental samples [2]. Almost all of the published methods toetermine the illicit drugs in waters include a sample pretreatmentombined with liquid chromatography (LC) with mass spectrom-try (MS) or tandem MS using electrospray (ESI) as an ionisationource [2,5].

Sample treatment is necessary in order to enrich the analytes ofnterest and obtain good analyte recoveries, and to remove matrixnterferences that may compete with the target analytes in theonisation process in LC–MS. Off-line solid-phase extraction (SPE)s the most commonly used sample preparation technique when

ealing with environmental samples [2,5]. Different sorbents haveeen tested in illicit drug determination methods, such as Oasis HLB7] and the mixed-mode Oasis MCX [3,6,8,9], Strata-X-C [10] and

∗ Corresponding author. Tel.: +34 977 55 86 29; fax: +34 977 55 84 46.E-mail address: nuria.fontanals@urv.cat (N. Fontanals).

021-9673/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.chroma.2013.02.069

d.© 2013 Elsevier B.V. All rights reserved.

Evolute CX [11]. These mixed-mode sorbents combine ionic inter-actions with reversed-phase interactions, which under optimisedconditions permit a washing step involving organic solvent to beintroduced in the SPE process. This therefore improves the selec-tivity or/and removal of the interferences involved in the matrixeffect.

On-line SPE is another sample treatment which, compared to theoff-line mode, improves the method sensitivity and shortens thepretreatment and analysis time [12]. So far, the SPE sorbents testedin this way to determine illicit drugs are PLRP-S [13], an in-househypercrosslinked sorbent [14], and the mixed-mode Isolute HCX[15] and Strata-X-CW [16]. However, an appropriate washing stepwas not included in any of these applications and the enrichmentfactors and/or the ion suppression might be further improved if thisstep were incorporated.

The present study explores a method to determine a groupof illicit drugs in complex environmental water samples that isbased on on-line SPE-LC–MS using a mixed-mode sorbent (OasisWCX) as the SPE sorbent which enables an effective washing stepto remove matrix interferences and overcome the ion suppres-sion/enhancement signal effect.

2. Materials and methods

2.1. Materials

All the illicit drugs and their metabolites studied: nicotine (NIC);cocaine (COC) and benzoylecgonine (BE); morphine (MOR) and

N. Fontanals et al. / J. Chromatogr. A 1286 (2013) 16– 21 17

Table 1Compound retention time, optimised fragmentor voltage and ions selected for quantification and confirmation in SIM mode.

Analytes pKa tR (min) Fragmentor voltage (V) Quantification ion (m/z) Confirmation ions (m/z)

Nicotine 9.0 2.4 100 163 132,106Morphine 8.3 4.5 175 286 165,153Dihydrocodeine 8.4 7.1 150 302 324,199Codeine 8.3 7.2 150 300 243,2156-Acetylmorphine 8.3 8.2 150 328 268,211BE 10.8 8.9 125 290 168,105

6(((1mAd

tAphaU

2

swtc(f

c(gplWf

2

wA3ttvtt

csp14wp

Cocaine 8.0 9.2 125

EDDP 7.7 9.9 100

Methadone 9.1 10.1 100

-acetylmorphine (AcMOR); codeine (Cod) and dihydrocodeineDHCod); and 2-ethylidene-1,5-dimethyl-3,3-diphenylpyrrolidoneEDDP) and methadone (MET), were obtained from CerilliantRound Rock, TX, USA) as solutions at a concentration of000 mg L−1 in methanol or acetonitrile. Working solutions of aixture of all compounds were prepared in 1:1 MeOH:H2O (v:v).ll the stock and working solutions were stored at −20 ◦C in theark.

Ultrapure reagent water purified by a water purification sys-em (Veolia, Sant Cugat del Vallés, Barcelona) was used throughout.cetonitrile (ACN) and methanol (MeOH) (both HPLC grade) wereurchased from SDS (Peypin, France). Analytical grade ammoniumydroxide (NH4OH) and formic acid (HCOOH), which were used todjust the pH, were purchased from Sigma–Aldrich (St. Louis, MO,SA).

.2. Instrumentation

The chromatographic system was an HP1100 series LC–MSelective detector (Agilent Technologies, Waldbronn, Germany)ith an ESI interface. It was equipped with a degasser, a qua-

ernary pump, a 20 �L loop injector and a column oven. Thehromatographic column was a Fused-CoreTM Ascentis Express C1850 mm × 4.6 mm) with a particle size of 2.7 �m (Supelco, Belle-onte, PA, USA).

The on-line solid-phase extraction system was connected to thehromatographic system by means of a six-port switching valveRheodyne, Cotati, CA, USA). An isocratic pump (Agilent Technolo-ies) was used to deliver the sample through a stainless steelrecolumn (dimensions 20 mm × 2 mm I.D.) fitted with 2 �m stain-

ess steel frits, all purchased from Upchurch Scientific (Oak Harbor,A, USA). This precolumn was packed manually using a packing

unnel with ∼30 mg of the sorbent.

.3. Chromatographic conditions

A binary mobile phase with gradient elution was used. Solvent Aas ultrapure water with 0.5% HCOOH (pH 2.5) and solvent B wasCN. The gradient profile was 10% solvent B increased to 15% in.5 min, then increased to 50% solvent B in 2.5 min, and increasedo 100% solvent B in 6 min, then held constant for 2 min, after whichhe mobile phase was returned to the initial conditions (10% sol-ent B) in 1 min (and held for 5 min to equilibrate the column forhe following analysis). The flow rate was 0.4 mL min−1 and theemperature of the column oven was set at 30 ◦C.

Flow injection analysis (FIA) was carried out to find the optimumonditions for each compound in the ESI. The average conditionselected for the optimum performance of the ESI interface in theositive mode were: nebuliser pressure 40 psi, drying gas flow rate

3 L min−1, drying gas temperature 350 ◦C, and capillary voltage000 V. Fragmentation voltages (optimised range was 50–250 V)ere defined individually and the specific values for each com-ound are listed in Table 1. The ions selected for quantifying the

304 182,105278 249,234265 310,245

samples are also listed in Table 1. In SIM mode, the most abundantion, which all analytes corresponds to [M + H]+ with the exceptionof methadone, was used for quantification and two other ions wereused for confirmation.

2.4. Solid-phase extraction

The commercially available Oasis WCX (particle size of30 �m), which is a mixed-mode polymeric sorbent based ona poly(vinylpyrrolidone-divinylbenzene) skeleton and modifiedwith carboxylic acid moieties (that can impart weak-cationic andreversed-phase interactions), was laboratory packed (∼30 mg) intoa 20 mm × 2 mm I.D. stainless-steel precolumn used for on-linetrace enrichment in the SPE process.

The final protocol was as follows: the SPE precolumn was con-ditioned with 5 mL of MeOH and 5 mL of ultrapure water adjustedat pH 7; 10 ml of water sample adjusted to pH 7 was loadedthrough the conditioned precolumn. The flow-rate was 3 mL min−1

throughout all of these steps. The sorbent was then washed bypassing 0.5 mL of MeOH through the precolumn at 1 mL min−1. Theretained analytes were desorbed using the mobile phase in the gra-dient profile of the chromatographic system, and in the back-flushmode to reduce band-broadening.

Environmental water samples (river water and effluent waterfrom sewage treatment plant) were filtered through 0.45 �m nylonmembranes (Osmonics Inc.) before the preconcentration step toeliminate the particulate matter, after which they were adjusted topH 7 with HCOOH.

3. Results and discussion

3.1. LC–MS conditions

Chromatographic separation was performed using a Fused-CoreTM Ascentis Express C18 column with a particle technology thatenables efficiency and speed of the separation pertinent to sub-2 �m particles to be achieved while maintaining the back pressures.Two different organic solvents were tested: ACN and 1% HCOOH inACN to establish how the ionisation is affected and hence the signalof all of the analytes. In the end, ACN on its own was selected as nosignificant improvement was seen with the addition of acidic solu-tion to the organic mobile phase. Table 1 lists the retention time ofthe studied analytes under the optimum separation conditions. Italso details both the fragmentor voltage and the ions selected forcompound quantification and confirmation.

Instrumental (LC–MS) linerarity was good for all the compoundswhen they were directly injected at low �g L−1 levels. The linearrange was 1–500 �g L−1 for all of the compounds with the exception

of morphine (5–1000 �g L−1) and acetylmorphine (1–1000 �g L−1).The limits of detection (LODs) calculated at a signal-to-noise ratio(S/N) of 3, were 0.2 �g L−1 for all of the compounds, except formorphine (1 �g L−1).

18 N. Fontanals et al. / J. Chromatogr. A 1286 (2013) 16– 21

Table 2% Recovery values of the studied analytes when the sample was loaded at 5 different pHs, either without a washing step or with a washing step involving 0.5 mL of MeOH.For experimental conditions, see text.

% RecoverypH 3 pH 4.5 pH 7 pH 10 pH 12

Sorbent state Neutral Neutral Negative Negative NegativeAnalyte state Positive Positive Positive Positive/neutral Neutral

Analytes No wash Wash No wash Wash No wash Wash No wash Wash No wash Wash

Nicotine 20 0 2 0 30 82 0 0 1 0Morphine 12 0 114 0 108 105 99 99 16 0Dihydrocodeine 54 30 108 4 110 105 39 88 82 0Codeine 42 4 97 3 88 108 35 91 64 06-Acetylmorphine 57 4 87 5 72 84 34 88 5 0BE 17 0 80 0 84 90 61 0 16 0Cocaine 87 3 63 39 93 97 63 103 17 0EDDP 66 3 33 67 99 106 77 80 2 2

%

3

twsWmmv

db

bm

3

twewt

fTaiie

T%Ls

Methadone 32 3 4 37

Relative standard deviations (RSD) (n = 3) were lower than 6%.

.2. SPE optimisation

Although the most commonly used mixed-mode sorbents forhe off-line solid-phase extraction of illicit drugs have been thoseith strong cation-exchange character [3,6,8–11], in this case we

elected a weak cationic exchange (WCX) sorbent, such as OasisCX, which is also feasible with this type of basic compounds;oreover, the elution solvent used in the WCX protocols [17]atches the LC mobile phase, and, therefore, it is much more con-

enient in the on-line SPE mode.Once the sorbent was selected, we had to optimise the SPE con-

itions so that the ionic as well as the reversed-phase interactionsetween the sorbent and the analytes were promoted.

The initial experiments for the SPE optimisation were performedy percolating 10 mL of ultrapure water spiked with the analyteixture at 200 ng L−1 through the precolumn.

.2.1. Loading sample conditionsWe tested five different sample pHs (3, 4.5, 7, 10 and 12) for

he sample, and we performed experiments with and without aashing step (with 0.5 mL of MeOH) after the loading step. The

xperiments with the washing step were carried out to confirm athich pHs the ionic interactions between the carboxylic groups in

he sorbent and the amino groups of the analytes occur.Table 2 shows the recovery values for all the analytes at all dif-

erent pHs tested with and without washing step. As we can see inable 2, on one hand, when the washing step is not included, all the

nalytes present some retention (with differences in the % recover-es) in all different pHs tested, which means that the reversed-phasenteractions between the sorbent and the analytes occurred to somextent.

able 3 Recovery values of the analytes studied when a 10 mL sample: ultrapure water, Ebre riC–MS analysed with a washing step involving 0.5 mL of MeOH; and validation parameteee text.

Analyte %Rultrapurea %Rriver

a %REWWTPa

Nicotine 100 90 74

Morphine 92 105 115

Dihydrocodeine 107 95 102

Codeine 101 88 114

6-Acetylmorphine 93 74 66

BE 82 63 51

Cocaine 92 67 63

EDDP 97 71 104

Methadone 100 80 100

a % RSD (n = 3) were lower than 7% for ultrapure water, 16% for river water and 17% forb Values obtained when the sample was spiked at 10 ng L−1 (n = 3).

98 99 74 71 1 1

On the other hand, when the washing step was included, onlyanalytes in samples adjusted at pH 7 and 10 (that are the pHs atwhich the ionic interactions are expected to occur) presented neartotal recovery values. At pH 10, however, the recoveries of mostanalytes (particularly nicotine, BE, EDDP and methadone) werelower than at pH 7, since these ionic interactions might start todiminish as the pKa of the analytes (pKa from 8 to 10) and theworking pH (10) are too close.

The retention of the analytes in the sorbent when the sample isadjusted at pHs 3, 4.5 and 12 (and including the washing step) islower with recovery values of about 0% in most instances, exceptin the sample adjusted to pH 4.5 for cocaine, EDDP and methadone,and also except dihydrocodeine at pH 3. These near 0% recoveriesfurther confirm that at these pHs, only reversed-phase interactionsoccur, that are not strong enough to retain such polar compounds.

In the end, the recoveries obtained at pH 7 are the highest. Theseresults are as expected since at pH 7, the sorbent is in its anionicform and the analytes are in their cationic form, suitable for thepromotion of ionic interactions. In view of this, the sample wasfixed at pH 7.

3.2.2. Washing and elution conditionsAlthough during the sample pH optimisation we checked the

possibility of including a washing step that consisted of 0.5 mL ofMeOH, we further optimised this parameter by testing differentvolumes of MeOH that ranged from 0.25 to 2 mL. In general, noneof the anaytes were washed out after a washing step with any ofthe MeOH volumes tested.

Thus, the next step involved experiments with environmen-tal water samples instead of ultrapure water in order to checkthe influence of the matrix components in the washing step.The first experiments were conducted by percolating 10 mL of

ver water and effluent wastewater (EWWTP) spiked at 10 ng L−1 was on-line SPE-rs of the method with Ebre river water samples. For the experimental conditions,

Linear range (ng L−1) Repeat. %RSDb Reprod. %RSDb

2–1000 4.2 6.35–1000 3.1 6.82–500 3.1 1.82–500 5.5 2.82–1000 5.3 3.65–1000 5.4 8.25–500 3.1 15.82–500 4.7 8.52–500 4.1 7.0

effluent water.

N. Fontanals et al. / J. Chromatogr. A 1286 (2013) 16– 21 19

F t wasto

ewocim

vtbsstocs

LiAtitwpswi

ig. 1. % Matrix effects (%ME) when 10 mL of sample of river water (a) and effluenf MeOH, were on-line SPE-LC–MS analysed.

nvironmental water (river and wastewater) adjusted to pH 7 andashed with volumes of MeOH ranging from 0.25 to 2 ml. A volume

f 0.5 ml of MeOH was the maximum that avoided losses of theompounds after the washing step. Therefore, the volume of MeOHn the washing step should be fixed at 0.5 mL since the presence of

atrix components weakens the retention of the analytes.Next, we also tried other washing solvents including: ACN in

olumes ranging from 0.25 to 2 mL; 5 mL of pure water adjustedo pH 7; 5 mL of pure water adjusted to pH 10 (with NH4OH) anduffered at pH 10 (NH4OH/NH4Cl); and also these three aqueousolvents separately followed by 0.5 mL of pure MeOH. All of theseolvents provided similar or even slightly lower washing capacity ofhe studied analytes compared with the washing step with 0.5 mLf MeOH. Moreover, washing with an organic solvent would alwayslean the matrix from interferences more effectively. Therefore, weelected 0.5 mL of MeOH as the washing solvent.

Another issue to be considered is the suitability of the unsplitedC mobile phase as elution solvent for the SPE. The LC mobile phases a combination of ultrapure water with 0.5% HCOOH (as solvent) and ACN (as solvent B) in a gradient profile, which is expected

o be suitable as it combines the acidic additive (to perform ionicnteractions) and the strength of an organic solvent. We also testedhe combination of the same aqueous solvent A (ultrapure waterith 0.5% HCOOH) and ACN with 1% HCOOH (solvent B) as a mobile

hase, and we obtained similar results both in the chromatographiceparation and in the SPE elution. In view of this, we continuedorking with the initial mobile phase and gradient (as described

n Section 2).

ewater (b), either with a washing step or without a washing step involving 0.5 mL

3.2.3. Recovery studiesOnce the parameters that affect the SPE had been optimised,

the next step was to check how the recoveries are when 10 ml ofultrapure water was spiked with the analyte mixture at lower con-centration level such as 10 ng L−1. The recoveries are calculated asratio between obtained concentration after SPE and expected the-oretical concentration. It should be mentioned that the recoveryvalues (Table 3) in all the cases were near to 100%.

In general, these recovery values are similarly high to thosereported in a previous study where an on-line SPE column packedwith hypercrosslinked sorbent (high retention) was connectedon-line to a hydrophilic interaction LC (HILIC) column with MSdetection. In that case, however, the washing was not as effectiveas it consisted of 5 mL of water at pH 7 [14].

At this point, we also compared the Oasis WCX performance toOasis HLB sorbent (same polymeric skeleton, but without pendingcarboxylic groups). 10 mL of sample adjusted at pH 7 and spikedat 10 ng L−1 of the analyte mixture were percolated through theOasis HLB cartridge with or without being followed by a wash-ing step involving 0.5 mL of MeOH. The results of Oasis HLB, evenwithout applying a washing step, were much lower (recovery val-ues ranging from 20% to 60% in most of the instances) than thoseobtained with Oasis WCX, which also confirms that the retentionsof a sorbent with merely reversed-phase interactions are not strong

enough to interact properly with analytes that are so polar. Theseresults are in agreement with a study [13] in which 5 mL of ultra-pure water spiked at 1000 ng L−1 of a similar group of illicit drugswere loaded through Oasis HLB and followed by on-line LC elution,

2 omatogr. A 1286 (2013) 16– 21

wc

3

3

smitSgrvsaSbffpttsescc(los1ai%ssTcapetTciaieas5waoslouospMs

2 4 6 8 10 120

codeine, 74.4 ng L-1

cocaine, 34.8 ng L-1

EDDP, 35.9 ng L-1

methadone, 18.7 ng L-1

0 %

100 %0 %

100 %0 %

100 %0 %

100 %

0 N. Fontanals et al. / J. Chr

hich provided recoveries from 30% to 50% for all the commonompounds.

.3. Application to environmental samples

.3.1. Matrix effect and recovery studiesIn a first step, we evaluated the matrix effect (ME) since the ESI

ource is highly susceptible to other components present in theatrix, and may result in signal suppression/enhancement lead-

ng to incorrect quantification results. When off-line SPE is used,he procedure to evaluate the ME is well-established [18] (i.e.PE environmental water extracts are usually spiked with the tar-et analytes and MEs are evaluated for each analyte calculatingesponses in comparison to those of reference standards in sol-ent). However, in on-line SPE, this procedure cannot be appliedince the extract is directly transferred to the analytical columnnd it is not possible for it to be spiked with the targets after thePE. To solve this issue, some authors [13,16,19] evaluate the MEy comparing the response of the analytes of spiked real samplesrom the SPE-LC determination with the response of the analytesrom spiked ultrapure grade water after the SPE-LC. However, in ouroint of view, this model evaluates both the losses resulting fromhe ME as well as those due to the SPE itself. Therefore, we evaluatehe ME by percolating 10 mL of non-spiked environmental waterample (river and wastewater) through the SPE cartridge, and thenluting the sorbent with the mobile phase (as usual) and, at theame time, direct injecting the pure standard into the LC (R2) via aommon injection. In this way, the analytes entered to the columnompletely, together with the matrix components. The R2 obtainedafter subtracting the response corresponding to the native ana-ytes present in real samples) were compared to the response (R1)btained in a separated run by direct injection of the analytes inolvent. Thus, the matrix effect percentage (%ME) is calculated as00 × R2/R1. Fig. 1 summarises the %ME evaluated in the river (A)nd effluent waste water (B) including or not the washing step dur-ng SPE in both instances. In the case of river water (Fig. 1A), theME is similar in both protocols (with and without the washingtep) and near to 100%, except for morphine and cocaine, whichhowed signal enhancement in the SPE without the washing step.he differences in %ME wastewater samples (Fig. 1B) arose whenomparing protocols with and without the washing step, with EDDPnd methadone being the analytes most affected with signal sup-ression of 92 and 95%, respectively; and BE and cocaine with signalnhancement of 256 and 160%, respectively. Nevertheless, none ofhe analytes showed any of these effects after the washing step.hus, the inclusion of the washing step in the SPE procedure isompletely beneficial for minimising MEs. Therefore, when includ-ng the washing step, the ME can be considered negligible and wessumed so in further analysis. These values of ME were foundn other studies to determine similar drugs. For instance, Postigot al. [13] included internal standards for quantification as well asutomatic correction of extraction recoveries to address a signaluppression from 94% for BE to 70% for 6-acetylmorphine when

mL of wastewater spiked with a similar group of illicit drugsere analysed with on-line SPE-LC–MS/MS. Similar strategies were

dopted when the off-line SPE was performed with protocols with-ut [7,20] or with less effective washing step, involving an aqueousolution [6,8,21]. On the other hand, Baker et al. [3] reported muchess contribution of the ME for the determination of a similar groupf illicit drugs in wastewater when the off-line SPE was performedsing Oasis MCX and included a washing step composed of 2 mLf acidic aqueous solution followed by 2 mL of acidic methanolic

olution. Even, improved results for Oasis MCX in the off-line SPErotocol were reported when the washing step involves 2 ml ofeOH and were compared to similar protocols without washing

tep using Oasis MCX and Oasis HLB [9].

Fig. 2. SIM chromatograms obtained when 10 mL of an effluent wastewater wason-line SPE-LC–MS analysed after a washing step that involves 0.5 mL of MeOH.

Once it has been checked that the ME using this protocol isnegligible, the next step was to quantify the recoveries obtainedwith complex matrices. Table 3 details the recovery values obtainedwhen 10 mL of river or effluent water from sewage treatment plant(STP) spiked at 10 ng L−1 were percolated through the SPE columnand following the optimised protocol. As we can see, the recoveriesfor environmental water are good as most of them are close to 100%.Only in the cases of 6-acetylmorphine (74% for river and 66% foreffluent), BE (63% for river and 51% for effluent) and cocaine (67% forriver and 63% for effluent), the recoveries slightly decreased, whichmight be due to competition between the interefences present incomplex samples and target analytes for the sorbent.

3.3.2. Method validationThe on-line SPE-LC–MS method was then validated using 10 mL

of water samples taken from the Ebre River and following the wholeprocedure developed in the SIM mode. Table 3 also reports theperformance of the method. The calibration curve was linear inthe concentration range with determination coefficients (r2) higherthan 0.996. The limits of quantification (LOQs) for each compoundwere taken as the lowest concentration level of the calibrationcurve, which ranged from 2 to 5 ng L−1. The LODs (S/N > 3) were0.5 ng L−1 for all the compounds except for BE and cocaine (1 ng L−1)and morphine (2 ng L−1). The sensitivity of the method is enhancedin an on-line system because the whole sample is directly trans-ferred to the chromatographic system, rather than one aliquot ofthe final extract, as occurs in an off-line system. Thus, the LOQs andLODs reported with the present methodology are comparable tothose found with off-line SPE-LC–MS/MS systems [8,21].

The repeatability and reproducibility between days were deter-mined by spiking three replicates of river water sample at 10 ng L−1,and the results obtained, expressed as a % of relative standard devi-ation (%RSD), were less than 6 and 16%, respectively.

omato

3

rurfrwitaMc(wseao

4

Ssmwm

evtitn

[

[[

[[[

[

[

[

[19] R. López-Serna, S. Pérez, A. Ginebreda, M. Petrovic, D. Barceló, Talanta 83 (2010)

N. Fontanals et al. / J. Chr

.3.3. Analyses of environmental samplesWe then used this validated method to analyse different envi-

onmental samples. To quantify the effluent water samples, wesed the same calibration curve as for river water because theecoveries obtained with both sample matrices were similar. Dif-erent wastewater samples from the tertiary treatment (n = 3) andiver water samples (n = 3) were analysed, and different analytesere found, the presence of which was confirmed by a difference

n ratio of ion abundances lower than ±20% as well as retentionime of the analytes that should match ±5%. Nicotine and EDDPppeared in river water samples at concentrations below the LOQs.orphine (5–5.4 ng L−1, n = 2), dihydrocodeine (8.9 ng L−1, n = 1),

odeine (74.4 ng L−1, n = 1), cocaine (24.4–43.6 ng L−1, n = 3), EDDP35.9–85.4 ng L−1, n = 3) and methadone (15.9–18.7 ng L−1, n = 3)ere found in the STP. These results are in agreement with other

tudies that also found these analytes in similar STPs [4,7,8]. As anxample, Fig. 2 shows representative SIM chromatograms from thenalysis, obtained under optimum methodology conditions, of onef these samples from STP.

. Conclusions

This work presents a fully automated method based on on-linePE-LC–MS to determine illicit drugs from environmental wateramples. The main benefit of this method is the inclusion of aixed-mode Oasis WCX as SPE sorbent, which allows an effectiveashing step with 0.5 mL of MeOH while maintaining the enrich-ent factors.After the optimisation, the method was sensitive and selective

nough to analyse different environmental water samples, pro-iding high SPE recoveries and quantification at low ng L−1 for all

he studied compounds. Moreover, the effectiveness of the wash-ng step allows the removal of matrix interferences to the degreehat the matrix effect was considered negligible, and it was notecessary to balance the signal.

[[

gr. A 1286 (2013) 16– 21 21

Acknowledgements

The authors thank the Ministry of Science and Innovation(Projects CTQ 2011-24179 and CTM 2011-28765-CO2-01) and theDepartment of Innovation, Universities and Enterprise (Project2009 SGR 223) for the financial support.

References

[1] W.W. Buchberger, J. Chromatogr. A 1218 (2011) 603.[2] C. Postigo, M.J. López de Alda, D. Barceló, TrAC Trends Anal. Chem. 27 (2008)

1053.[3] D.R. Baker, B. Kasprzyk-Hordern, J. Chromatogr. A 1218 (2011) 1620.[4] C. Postigo, M. López de Alda, D. Barceló, Environ. Int. 37 (2011) 49.[5] A.L.N. van Nuijs, S. Castiglioni, I. Tarcomnicu, C. Postigo, M.L. de Alda, H. Neels,

E. Zuccato, D. Barcelo, A. Covaci, Sci. Total Environ. 409 (2011) 3564.[6] L. Bijlsma, J.V. Sancho, E. Pitarch, M. Ibánez, F. Hernández, J. Chromatogr. A 1216

(2009) 3078.[7] M.R. Boleda, M.T. Galceran, F. Ventura, J. Chromatogr. A 1175 (2007) 38.[8] M. Pedrouzo, F. Borrull, E. Pocurull, R.M. Marcé, J. Sep. Sci. 34 (2011) 1091.[9] I. González-Marino, J.B. Quintana, I. Rodríguez, M. González-Díez, R. Cela, Anal.

Chem. 84 (2011) 1708.10] K. Bisceglia, A. Roberts, M. Schantz, K. Lippa, Anal. Bioanal. Chem. 398 (2010)

2701.11] G. Dowling, L. Regan, J. Pharm. Biomed. Anal. 54 (2011) 1136.12] S. Rodríguez-Mozaz, M.J. López de Alda, D. Barceló, J. Chromatogr. A 1152 (2007)

97.13] C. Postigo, M.J. López de Alda, D. Barceló, Anal. Chem. 80 (2008) 3123.14] N. Fontanals, R.M. Marcé, F. Borrull, J. Chromatogr. A 1218 (2011) 5975.15] M. Bjork, M. Nielsen, L. Markussen, H. Klinke, K. Linnet, Anal. Bioanal. Chem.

396 (2010) 2393.16] U. Chiuminatto, F. Gosetti, P. Dossetto, E. Mazzucco, D. Zampieri, E. Robotti, M.C.

Gennaro, E. Marengo, Anal. Chem. 82 (2010) 5636.17] N. Fontanals, R.M. Marce, F. Borrull, TrAC Trends Anal. Chem. 29 (2010)

765.18] M. Pedrouzo, F. Borrull, R.M. Marcé, E. Pocurull, J. Chromatogr. A 1216 (2009)

6994.

410.20] J.-D. Berset, R. Brenneisen, C. Mathieu, Chemosphere 81 (2010) 859.21] A.L.N. van Nuijs, I. Tarcomnicu, L. Bervoets, R. Blust, P. Jorens, H. Neels, A. Covaci,

Anal. Bioanal. Chem. 395 (2009) 819.