Eom

NNa

b

S

a

ARRAA

KSSRMS

1

ivwaaldfe

pT

h0

Journal of Chromatography A, 1343 (2014) 55–62

Contents lists available at ScienceDirect

Journal of Chromatography A

j o ur na l ho me page: www.elsev ier .com/ locate /chroma

valuation of strong cation-exchange polymers for the determinationf drugs by solid-phase extraction–liquid chromatography–tandemass spectrometry�

úria Fontanalsa,∗, Núria Mirallesa, Norhayati Abdullahb, Arlene Daviesb,úria Gilarta, P.A.G. Cormackb

Departament de Química Analítica i Química Orgànica, Universitat Rovira i Virgili, Campus Sescelades Marcel lí Domingo, s/n, 43007 Tarragona, SpainWestCHEM, Department of Pure and Applied Chemistry, University of Strathclyde, Thomas Graham Building, 295 Cathedral Street, Glasgow G1 1XL,cotland, UK

r t i c l e i n f o

rticle history:eceived 27 November 2013eceived in revised form 17 March 2014ccepted 27 March 2014vailable online 4 April 2014

eywords:trong cation-exchangeolid-phase extraction

a b s t r a c t

This paper presents eight distinct strong cation-exchange resins, all of which were derived from precursorresins that had been synthesised using either precipitation polymerisation or non-aqueous dispersionpolymerisation. The precursor resins were transformed into the corresponding strong cation-exchangeresins by hypercrosslinking followed by polymer analogous reactions, to yield materials with high spe-cific surface areas and strong cation-exchange character. These novel resins were then evaluated asstrong cation-exchange (SCX) sorbents in the solid-phase extraction (SPE) of a group of drugs fromaqueous samples. Following preliminary experiments, the two best-performing resins were then eval-uated in solid-phase extraction–liquid chromatography–tandem mass spectrometry (SPE/LC–MS/MS)

ecoveryatrix effect

ewage

to determine a group of drugs from sewage samples. In general, use of these sorbents led to excellentrecovery values (75–100%) for most of the target drugs and negligible matrix effects (ME) (<20% ionsuppression/enhancement of the analyte signal), when 50 mL and 25 mL of effluent and influent sewagewater samples, respectively, were percolated through the resins. Finally, a validated method based onSPE/LC–MS/MS was used to quantify the target drugs present in different sewage samples.

© 2014 Elsevier B.V. All rights reserved.

. Introduction

In modern society, the numbers of drugs in widespread uses increasing. These drugs can be released into the environmentia sewage systems or incorrect disposal methods. Although theseaters pass through sewage treatment plants (STP), these drugs

re often not removed completely by the treatments processes. As result, they are often found in surface and wastewaters at ng/Levels [1,2]. In view of this, new analytical techniques should beeveloped. These analytical techniques include sample preparation

ollowed by liquid chromatography coupled with mass spectrom-try (LC–MS) or tandem mass spectrometry (LC–MS/MS), to enable

� Presented at the XIII Scientific Meeting of the Spanish Society of Chromatogra-hy and Related Techniques (SECyTA2013), 8–11 October 2013, Puerto de la Cruz,enerife, Canary Islands, Spain.∗ Corresponding author. Tel.: +34 977 55 86 29; fax: +34 977 55 84 46.

E-mail address: nuria.fontanals@urv.cat (N. Fontanals).

ttp://dx.doi.org/10.1016/j.chroma.2014.03.068021-9673/© 2014 Elsevier B.V. All rights reserved.

the determination of these drugs at the required low concentrationlevels [3–5].

Sample preparation must enrich the analytes, and should reducethe matrix effect (ME) on subsequent LC–MS analysis to obtainreliable and repeatable results. For liquid samples, some greensolvent-extraction techniques have been developed [6]; regardingto sorptive-extraction techniques, solid-phase extraction (SPE) isusually the technique of choice [7] because of its versatility aris-ing from the ready availability of different sorbent types. In recentyears, research into SPE sorbents has focused on improving capacity(enhancing the preconcentration factor) and selectivity (improv-ing the clean-up effectiveness) within a single material, leadingto the emergence of what are known as mixed-mode polymericsorbents. These sorbents combine a polymeric skeleton with ionicgroups, with two types of interactions available: reversed-phase(RP) (from the skeleton) and ion-exchange (from the ionic groups).

Mixed-mode sorbents are classified depending on whether theionic group attached to the resin is cationic or anionic, but alsowhether the ionic group is strong or weak. The most common ofthese ionic groups are sulfonic acids and carboxylic acids for strong

5 omato

aamsacioeett

ofpeTvmso

satOTmt8c

itmahfpatrtwm

SltrfntL

2

2

v(l(

6 N. Fontanals et al. / J. Chr

nd weak-cation exchange sorbents, respectively; and quaternarymines for strong anion-exchange, and tertiary, secondary and pri-ary amines for weak anion-exchange. A benefit of mixed-mode

orbents is that the ion-exchange interaction between the sorbentnd the analytes and/or interferences is turned on and off by theareful control of the pH of the washing and elution solvents, result-ng in the selective protonation or deprotonation of the analytesr interferences, and even the sorbent (in the case of weak ion-xchange sorbents). Thus, the interferences and analytes can beluted separately during the washing and elution steps, respec-ively, as a consequence of the careful, rational selection of pH andhe solvent in each SPE step [8].

At present, mixed-mode sorbents are one of the main focusesf research for manufacturers and companies. One of the reasonsor this is, generally, the need for cleaner extracts from SPE and, inarticular, avoidance of ion-suppression/enhancement when thesextracts are injected into LC–MS or LC–MS/MS systems [4,9–11].herefore, despite being relatively new, they have been applied inarious fields to extract different types of analytes in a selectiveanner from the matrix interference usually present in complex

amples, such as those of biological, foodstuff and environmentalrigin [8,9]; thus, it is a constant evolving field.

In recent years, several SPE sorbent companies have launchedtrong or weak cation-exchange (SCX or WCX, respectively),nd strong or weak anion-exchange (SAX or WAX, respec-ively) variants of such well-known sorbent precursors, includingasis (Waters), Strata (Phenomenex), Bond Elut Plexa (Agilentechnologies), and Evolute (Biotage), among others. These com-ercially available mixed-mode sorbents are characterised by

heir macroreticular structures, specific surface areas from 600 to00 m2/g, mean particle diameters from 50 to 100 �m and, in someases, a degree of hydrophilicity [8].

To improve the features of the mixed-mode sorbents availablen the market, our research group has been working on improvinghe morphological properties (by exploiting hypercrosslinked poly-

er microspheres pioneered by Davankov in the 1970s [12,13])nd the introduction of ionic moieties, so that they can exhibitigher levels of RP and ionic interactions with the analytes. So

ar, we have prepared and evaluated, via SPE, hypercrosslinkedolymer microspheres modified with 1,2-diethylamine and piper-zine moieties, dimethylbutylamine and carboxylic acid moietieso impart WAX, SAX and WCX character onto the sorbents. Theesults obtained with these resins were promising and better thanhose from commercial available sorbents [8]. These suitable resultsere attributed to the enhanced structural properties of theseixed-mode sorbents.In view of this, we present eight different in-house prepared

CX materials. These materials differ in terms of their morpho-ogical properties and ion-exchange capacity, which are attributedo the different synthetic approaches used during their prepa-ation. We have evaluated these different SCX materials in theollowing terms: capacity to enrich target analytes and effective-ess in cleaning-up the interferences in the matrix samples; abilityo reduce the ME encountered when analytes are determined byC–MS/MS from complex environmental samples.

. Experimental part

.1. Materials

The reagents used for the polymer synthesis were para-

inylbenzyl chloride (VBC) (95% grade), divinylbenzene (DVB)80% grade), styrene (St) (99% grade), ethylene glycol dimethacry-ate (EGDMA) (98% grade), poly(N-vinylpyrrolidone) (PVP) 55MW ∼ 55,000) and Triton X-305, all supplied by Sigma-Aldrich

gr. A 1343 (2014) 55–62

(Steinheim, Germany). The monomers were purified by pass-ing them through a short column of neutral alumina. 2,2′-Azobis(isobutyronitrile) (AIBN), supplied by BDH (Poole, U.K.)was recrystallised from acetone at low temperature. Anhydrous1,2-dichloroethane (DCE), iron(III) chloride, chlorosulfonic acid,lauric acid and tetraethyl ammonium bromide (TEABr) (or sodiumchloride—NaCl) were supplied by Sigma-Aldrich; all were of highpurity as supplied and not purified further prior to use.

The additional reagents used in the preparation of the HXLNAD-SCX sorbents were 1,2-dichloroethane (DCE) (99.8% grade) andconcentrated sulfuric acid (95–97%), both supplied by Sigma-Aldrich, and distilled water. All reagents were used as received.

We selected some therapeutic drugs with basic and acidicgroups to evaluate the performance of the different sorbents. Theyincluded: trimethoprim, caffeine, antipyrine, atenolol, ranitidine,metoprolol, propranolol, carbamazepine, clofibric acid, salicylicacid, ibuprofen and diclofenac. They were obtained from Sigma-Aldrich. Standard stock solutions of each analyte were preparedat 1000 mg/L in methanol (MeOH). We also selected the followingillicit drugs: morphine, cocaine, methadone and codeine, and theirmetabolites, 6-acetylmorphine, benzoylecgonine, 2-ethylidene-1,5-dimethyl-3,3-diphenylpyrrolidine (EDDP) and dihydrocodeine,respectively. They were purchased from Cerilliant (Round Rock,TX, USA) as solutions in MeOH at a concentration of 1000 mg/L.A mixed solution of all analytes in MeOH at 50 mg/L was preparedweekly. All standard solutions were stored at −20 ◦C. Working solu-tions were prepared daily by an appropriate dilution of the mixedsolution with ultrapure water, which was obtained from a waterpurification system (Veolia, Sant Cugat del Vallès, Spain). Table 1presents the pKa values of these analytes.

Acetonitrile (ACN) and MeOH were of HPLC grade and fromProlabo (Llinars del Vallès, Spain). Nitrogen (N2) was suppliedby Carburos Metálicos (Tarragona, Spain). Hydrochloric acid (HCl)(37%), formic acid (HCOOH) (≥95%), ammonium hydroxide solu-tion (NH4OH) (28%) and sodium hydroxide (NaOH) (≥98%) werepurchased from Sigma-Aldrich.

2.2. Resin synthesis

Two type of polymerisations, namely non-aqueous dispersionpolymerisation (NAD) and precipitation polymerisation (PP) wereadopted to prepare the precursor particles [14].

For NAD polymerisations, PVP-55 (6% w/w), Triton X-305 (2%w/w), AIBN (2–6% w/w), styrene (50–90% w/w) (all the percent-ages are relative to the total mass of monomer in the monomerfeed), half of the VBC and 47.5 mL of ethanol were added into a500 mL five-necked, round-bottomed flask fitted with an overheadstirrer, condenser and nitrogen inlet. Once a homogenous solutionhad formed at room temperature, the solution was bubbled withnitrogen gas at room temperature for 30 min. The flask was thenplaced into an oil bath set at 70 ◦C, and stirred mechanically usinga four-bladed PTFE stirrer at 160 rpm. EGDMA (1% w/w relativeto total mass of monomer in the feed) and the second half of theVBC was dissolved in a second portion of ethanol (47.5 mL) at 70 ◦Cunder nitrogen. One hour after the start of the polymerisation, thehot solution containing EGDMA and VBC was added dropwise intothe reaction vessel. The reaction was continued for a further 24 hunder constant agitation. The particles that were obtained werecentrifuged for 10 min at 3000 rpm and then washed 2 times inethanol and 2 times in methanol (the particles were suspended inthe appropriate wash solvent and centrifuged between each wash-ing step). The particles were filtered using vacuum filtration on a

0.2 �m nylon membrane filter and dried overnight in vacuo at 40 ◦C.

For the PP polymerisations, the comonomers (75% (w/w) VBCand 25% (w/w) DVB) (2% w/v total monomer in feed relativeto solvent) and AIBN (2 mol% relative to polymerisable double

N. Fontanals et al. / J. Chromatogr. A 1343 (2014) 55–62 57

Table 1Analyte information (pKa and retention times) and MS/MS parameters employed for the MRM acquisition for their determination.

Analyte pKaa tR(min) Ionis. mode Cone volt. (V) Parent ion (m/z) Prod. ion (m/z) Collision energy (eV)

Morphine 8.3 1.8 + 125 286 152 50165 50

Atenolol 9.4 2.4 125 267 145 25190 10

Ranitidine 8.4 2.7 + 100 315 176 5130 15

Dihydrocodeine 8.4 4.2 + 150 302 199 25128 50

Codeine 8.3 4.8 + 150 300 165 50153 50

6-Acetylmorphine 8.3 5.1 + 150 328 165 50211 25

Trimethoprim 7.0 6.2 + 125 291 230 15123 25

Caffeine 14.0 6.4 + 125 195 138 15110 25

BE 10.8 7.9 + 125 290 168 153.2 105 25

Metoprolol 9.4 8.1 + 125 268 116 15159 15

Cocaine 8.0 8.4 + 125 304 182 1582 25

Antipyrine 9.86 8.7 + 100 189 145 30115 30

Propranolol 9.5 10.5 + 125 260 116 15183 15

Salicylic acid 3.1 12.3 − 75 137 93 1565 30

EDDP 7.7 12.6 + 150 278 234 25249 15

Methadone 9.1 13.0 + 100 310 265 5105 25

Carbamazepine 13.94 13.6 + 150 237 179 35193 35

Clofibric acid 3.2 14.7 − 75 213 127 1085 5

b(0iUpfrtd

trsratwsewo

cwaltc

Diclofenac 4.2 15.4 − 75

Ibuprofen 4.4 15.6 − 75

onds) were added to ACN (200 mL) in a polypropylene bottle250 mL). The monomer solution was deoxygenated with N2 at◦C and then the bottle placed on a low-profile roller (Stovall)

n a temperature-controllable incubator (Stuart Scientific, Surrey,K). The temperature was ramped from ambient to 60 ◦C over aeriod of ∼2 h and the polymerisation allowed to proceed at 60 ◦Cor a further 46 h. The resulting particles were separated from theeaction medium by filtration on a 0.2 �m nylon membrane fil-er, washed successively with MeOH, toluene and acetone, beforerying in vacuo at 40 ◦C overnight.

For the hypercrosslinking reactions the NAD or PP precursor par-icles (∼1.2 g) were added to 1,2-dichloroethane (DCE) (40 mL) in aound-bottomed flask (100 mL) and were left to swell fully under atream of N2 at room temperature for 1 h. FeCl3 (1:1 mole ratio withespect to the CH2Cl content of particles) in DCE (40 mL) was addednd the mixture heated at 80 ◦C for 18 h. The hypercrosslinked par-icles (HXLPP and HXLNAD for PP and NAD precursors, respectively)ere recovered as described above and washed with MeOH and

everal times with aqueous HNO3 (pH 1). The particles were thenxtracted overnight with acetone in a Soxhlet extractor and wereashed again with MeOH and diethyl ether before drying in vacuo

vernight at 40 ◦C.For the sulfonation procedure, either HXLPP or HXLNAD parti-

les were charged to a three-necked, round-bottomed flask fittedith an overhead mechanical stirrer and a reflux condenser, under

n N2 atmosphere. Anhydrous DCE (30 mL) was added. This waseft for 1 h to wet the beads. For sulfonation of the HXLPP particles,he lauroyl sulfate solution (obtained by reaction of lauric acid andhlorosulfonic acid in cyclohexane (3 mL) under an N2 atmosphere

294 250 5214 15

205 161 5

and stirred at room temperature for 1 h [15]) was added to theHXLPP beads via syringe and the reaction was heated to 50 ◦C, withstirring, for 24 h. After 24 h, the product was recovered by filtrationon a 0.2 �m nylon filter membrane and washed with petroleumether (b.p. 60–80 ◦C), and further extracted overnight in a Soxhletapparatus with the same solvent. The product, in the form of abrown powder, was then washed with diethyl ether (b.p. 30–40 ◦C)and oven-dried in vacuo at 40 ◦C for 24 h. For sulfonation of theHXLNAD particles, sulfuric acid was added to the HXLNAD beadsvia syringe and the mixture was then heated rapidly to 60 ◦C undernitrogen, with continuous mechanical stirring during the reaction.The sulfonated hypercrosslinked particles were then allowed tocool on an ice–water bath to quench the reaction, washed with anexcess of distilled water until the pH of the filtrate was neutral, andthen filtered using vacuum filtration on a 0.2 �m nylon membranefilter. They were then dried overnight in vacuo at 40 ◦C.

Variable amounts of the sulfonation reagent were added so thatwe obtained resins with different loadings of sulfonic acid groups.Table 2 details all the resin compositions.

The HXLPP-SCX and HXLNAD-SCX resins were characterised bymeasuring their specific surface areas using N2 sorption isothermdata generated on a Micromeritics ASAP 2000 porosimeter. Micro-sphere diameters and particle size distributions were calculatedusing ImageJ software from the image analysis of 100 individualparticles in scanning electron microscopy (SEM) images, which

were acquired using a Cambridge Instruments Stereoscan 90instrument. The ion-exchange capacity (IEC) was calculated bythe titration method using NaCl or TEABr. The carbon, hydrogen,sulfur, and nitrogen contents of the polymers were obtained by

58 N. Fontanals et al. / J. Chromatogr. A 1343 (2014) 55–62

Table 2Properties of the strong cation-exchange resins evaluated.

Polymer type Material sample code IECa (mmol/g) SSAb (m2/g) Mean particle diameter (�m)

HXLNAD HXLNAD-SCXa 1.3 900 3–4HXLNAD-SCXb 0.6 1020 2–3HXLNAD-SCXc 2.3 640 2–3HXLNAD-SCXd 1.9 770 2–3HXLNAD-SCXe 2.1 330 2–3

HXLPP HXLPP-SCXa 1.7 1070 4–6HXLPP-SCXb 2.0 1160 2–3HXLPP-SCXc 2.8 1370 4–6

eyi

2

sdw

Lliai

(A

2

wfl3tc

e

simwrttwc

2

ltbscSsd

a Ion-exchange capacity.b Specific surface area.

lemental microanalysis using a Perkin Elmer 2400 Series II anal-ser. The characterisation data obtained from the resin evaluations detailed in Table 2.

.3. Instrumentation

For the SPE evaluation, an Agilent Technologies 1100 series HPLCystem, equipped with a quaternary pump, UV detector, a solventegasser unit, a 20 �L loop manual injector and a column heateras used.

For the evaluation of MEs and analysis of sewage samples, anC–MS/MS system was used. The instrument consisted of an Agi-ent 1200 series LC, equipped with an automatic injector (volumenjected was 50 �L), a degasser, a binary pump and a column oven;nd a 6410 series triple quadrupole mass spectrometer using an ESInterface from Agilent Technologies (Waldbronn, Germany).

The analytical column was an Ascentis Express C18100 mm × 4.6 mm i.d.) with 2.7 �m particle size from Sigma-ldrich.

.4. Chromatographic conditions

A binary mobile phase was a mixture of two solvents: ultrapureater with 1% HCOOH (pH 3) (solvent A) and ACN (solvent B). Theow rate was 0.6 mL/min. The column temperature was kept at0 ◦C. The gradient was as follows: 7% B to 28% B in 9 min, increasedo 100% B in 5 min, constant for 1 min and then decreased to initialonditions in 1 min.

The wavelength used to detect all the compounds during thentire analysis was 210 nm.

The MS/MS parameters were optimised under flow injection oftandard solutions of each compound. Both positive and negativeonisation modes were applied to enable a simultaneous deter-

ination of the studied analytes. The optimum source conditionsere as follows: nebuliser pressure of 45 psi, drying gas (N2), flow

ate of 12 L/min, source temperature of 350 ◦C and a capillary poten-ial of 4000 V. To obtain two multiple reaction monitoring (MRM)ransitions for each compounds, cone voltage and collision energiesere optimised. Table 1 collects the optimum conditions for each

ompound.

.5. SPE protocol

All the in-house prepared hypercrosslinked SCX sorbents wereaboratory packed (60 mg) into 6 mL polyethylene cartridges withwo frits (a metal frit of 2 �m pore size at the bottom of the sor-ent bed and a polyethylene frit of 10 �m pore size at the top of theorbent bed). Each cartridge was used throughout all the study. The

artridges were placed in an SPE manifold (Teknokroma, Barcelona,pain) connected to a vacuum pump. The SPE procedure was theame for all the sorbents. First of all, the cartridges were precon-itioned with 5 mL of MeOH and 5 mL of ultrapure water adjusted

at pH 3 with HCl. Subsequently, the samples, also adjusted to pH 3(HCl), were passed through the cartridges at a flow rate of 5 mL/min.The cartridges were then washed with 5 mL of MeOH. Finally, theanalytes were eluted from the cartridges using 5 mL of 5% NH4OHin MeOH. Prior to the LC injection, the extracts were evaporatedto dryness under a gentle stream of N2 and redissolved in 1 mL ofultrapure water adjusted to pH 3 with HCOOH.

Sewage (influent and effluent) water samples from a treatmentplant in the Tarragona surrounding area were collected in pre-cleaned amber glass bottles from a wastewater treatment plant(WWTP). Subsequently, the samples were filtered using a 0.22 �mnylon membrane (Supelco, Bellefonte, PA, USA) to eliminate theparticulate matter prior to the preconcentration step, acidified topH 3 (HCl) and stored at 4 ◦C until analysis.

The quality assurance and quality control procedures includedanalysis of solvent blanks and duplicates of the samples. One proce-dural blank was run every five sewage samples to assess potentialsample contamination [16].

3. Results and discussion

Eight distinct SCX materials were prepared with variations inthe ion-exchange capacity and specific surface area. Table 2 detailsthe characterisation data for all the materials studied. In-exchangecapacities are in the range 0.6–2.8 mmol/g. Specific surface areasfrom ∼300–1400 m2/g, and the mean particle diameter were in thelow micro range (2–6 �m is typical).

3.1. Evaluation of the SPE performance

The SPE protocol used was adapted from the method recom-mended by different manufactures [8], and also based on ourprevious experiences with the handling and use of ion-exchangeresins. The sample was adjusted to pH 3, as all of the basic analytesexist in their protonated form at this pH. During the loading step,the basic analytes are retained on the sorbent by both reversed-phase (RP) and ionic interactions. Any acidic compounds in thesample will also be retained by RP mechanisms; however, there willbe no additional ionic interactions with the acidic compounds. Inthe washing step, MeOH breaks the RP interactions which bind theanalytes to the SCX materials. It was not necessary to use any morethan a 5 mL volume of MeOH since, even with 5 mL of methanol,trimethoprim (a basic compound) started to elute in the washingstep already. With specific regards to the elution step, this was per-formed using a solution of NH4OH in MeOH. The NH4OH serves tobreak any ionic interactions between the analytes and the sorbent,while the MeOH prevents any new RP interactions from develop-ing. We evaluated two different proportions of basic additive: 5%

and 10% NH4OH in MeOH. However, an increase in the proportionof NH4OH from 5% to 10% had no significant effect on the recoveryof the basic compounds, leading to the conclusion that 5% NH4OHin MeOH was a suitable composition for the elution solvent.

omatogr. A 1343 (2014) 55–62 59

eae

eiiertrr

HtiactrfbftdtbaNfrsbotttyc

pttgiast

s

tttib(cn

3

as

Table 3Matrix effect and recovery values for target compounds in effluent and influentwastewaters samples using both HXLNAD-SCXb and HXLPP-SCXc as sorbents inSPE/LC–MS/MS.

HXLNAD-SCXb HXLPP-SCXc

Effluent Influent Effluent Influent

%R %ME %R %ME %R %ME %R %ME

Morphine 52 33 49 30 48 16 43 12Atenolol 117 12 73 16 94 17 74 21Ranitidine 92 13 98 17 117 8 95 −4Dihydrocodeine 115 3 99 −6 97 5 85 −7Codeine 86 6 80 8 74 9 70 76-Acetylmorphine 93 2 109 −5 83 8 95 7Trimethoprim 41 10 32 30 22 17 31 10Caffeine 73 24 107 39 64 18 70 14Metoprolol 120 6 104 6 101 5 86 10BE 99 5 101 −3 98 4 93 3Cocaine 80 11 75 6 69 5 62 5Antipyrine 92 3 69 6 95 4 67 3Propanolol 97 10 82 7 96 7 95 9EDDP 75 −15 62 −10 85 −8 63 −8

N. Fontanals et al. / J. Chr

We then evaluated, as part of a preliminary study, each of theight materials by loading with 50 mL sample of the drug mixture at

level of 0.5 ppm, washing with 5 mL volumes of MeOH and finallyluting with 3 × 5 mL of a 5% NH4OH in MeOH.

The compounds eluted via the washing step were all acidic, asxpected, with the exception of carbamazepine. Carbamazepines basic and should therefore, in principle, be retained byon-exchange interactions; however, its aromatic ring structurenhances its retention with the resins through RP interactionsather than SCX interactions. The remainder of the basic analytesested eluted as expected during the elution step. Fig. 1 shows theecovery values obtained with the eight different SCX resins for aepresentative group of basic compounds.

Among the five HXLNAD-SCX resins tested, HXLNAD-SCXa andXLNAD-SCXb gave the highest recoveries for all the analytes

ested; only trimethoprim and morphine experienced a decreasen its recovery (for instance, trimethoprim recoveries were 65%nd 55% for HXLNAD-SCXa and HXLNAD-SCXb, respectively). In anyase, trimethoprim recoveries were even lower (values not higherhan 15%) for the rest of HXLNAD-SCX resins tested. These lowecoveries for trimethoprim were, as mentioned already, due to itsractionation between the washing and elution step, which mighte attributed to the weak ion-exchange interactions with the sul-onic groups in the resins. The poorer recoveries of the analytes inhe HXLNAD-SCXc, HXLNAD-SCXd and HXLNAD-SCXe resins wereue to either their lack of retention during the loading step orhe fact that they were washed out during the washing step. Thisehaviour is in agreement with their specific surface areas whichre lower than the other two resins (i.e., HXLNAD-SCXa and HXL-ADb). Therefore, these three resins were ruled out as candidates

or further evaluation. In addition, although the HXLNAD-SCXaesin provided successful results, it was discarded as a candidateince it was necessary to elute the analytes retained on the sor-ent with up to 15 mL of elution solution. However, in the restf HXLNAD-SCX sorbents tested, the elution was complete withhe first 5 mL of elution solution. The higher the elution volumehe more diluted the analytes and the lower the sensitivity or, inhe case of evaporating the eluate to dryness, the longer the anal-sis time. Thus, the HXLNAD-SCXa resin was also ruled out as aandidate.

With regards to the three HXLPP-SCX resins, they, in general,rovided recoveries ranging from 70% to 100% for all the analytesested, with the exception of caffeine where the recoveries droppedo 60% for HXLPP-SCXb and to 40% for HXLPP-SCXa. HXLPP-SCXcave rise to the best recoveries for all the analytes tested (as shownn Fig. 1). These results are in accordance with the morphologicalnd chemical properties of the resin since it combines the largestpecific surface area (1370 m2/g) and the highest ion-exchange con-ent (2.8 mmol/g).

After this preliminary study, two highly promising resins wereelected for further experiments: HXLNAD-SCXb and HXLPP-SCXc.

For experiments involving ultrapure water samples, the break-hrough volume was studied for the analytes percolated throughhe HXLNAD-SCXb and HXLPP-SCXc resins. As a compromise for allhe basic analytes tested, a sample volume of up to 250 mL (whichs consistent with the quantity of SPE resins packed—60 mg) cane percolated through both resins without losses of the analytesrecovery values ranging from 60% to 100% in all cases). So far, in theomparison of the performance of HXLPP-SCX with HXLNAD-SCX,o differentiation in terms of %recovery was found.

.2. Recoveries and matrix effect in sewage waters

When dealing with complex samples, such as sewage waters,part from the recovery results per se, the ME issue should be con-idered, and several experiments were conducted in this sense to

Methadone 75 7 42 −4 67 7 45 6

cover both issues. First, a blank sample was analysed in order tosubtract the possible signal of existing analytes that appeared inall instances at low concentration levels. Then, the ME was eval-uated, which was calculated by comparing the signal obtained forthe analytes (S2) when spiked at 50 �g/L to the blank extract ofthe SPE of either 50 mL of effluent or 25 mL of influent sewagesamples to the signal obtained for these analytes at the same con-centration in the injection solution (S1), as described elsewhere [9](%ME = (100 − (S2/S1)) × 100).

The ME values for both type of samples analysed (i.e., effluentand influent sewage) are detailed in Table 3, where it can be seenthat most of the analytes showed ion-suppression (positive val-ues of %ME). In general, the ME is not higher than 10% for most ofthe analytes in both type of samples, and for both resins tested.Just as was the case for morphine, with atenolol, trimethoprimand caffeine the ME rises to 20% at most when using HXLPP-SCXc.When using HXLNAD-SCXb, the ME values of trimethoprim, caf-feine and morphine are about 30% (ion-suppression) in influentwastewater samples (and morphine for both type of sample). Thesevalues are satisfactory and lower than the usual values encoun-tered when these types of environmental samples are analysed bySPE/LC–MS/MS [4,17–19]. For instance, Caban et al. [4] reported val-ues of ion-suppression from 18 to 41% when a group of �-blockers(including atenolol, metoprolol and propranolol) were extractedfrom composite (effluent and influent) sewage samples. The lowME values encountered in the present study are due to the clean-upstep performed (with 5 mL of MeOH), which simplifies the com-plexity of the matrix. In fact, in order to check how the washingstep helps to diminish the ME, a set of experiments was con-ducted in which the washing step was not included. In general,the MEs values increased, especially for the first eluting analytes.For instance, morphine presented values of ion-suppression from52% to 57% in each of the four cases, and for dihydrocodeine onHXLNAD-SCXb the ion-suppression effect increased dramaticallyto 65% for effluent wastewater and to 60% for influent water.Similar results were reported [10,20,21] when different mixed-mode resins were used as SPE sorbents to extract a similar groupof drugs and the SPE protocol included a washing step involv-ing an organic solvent. As an example, morphine presented anear 50% level of ion-enhancement when the clean-up was not

included using Oasis WCX as sorbent, whereas it presented a ion-suppression of 10% when a clean-up that included MeOH wasincluded [20].

60 N. Fontanals et al. / J. Chromatogr. A 1343 (2014) 55–62

ased o

vwwttwtabsoItn

frO1HIhcs5thetpsat

%

Fig. 1. Preliminary evaluation of the eight SCX resins b

Once it was confirmed that the ME was negligible, the recoveryalues (Table 3) for each resin and type of sample were calculatedhen 50 mL of effluent and 25 mL of influent spiked at 2 and 4 �g/L,ith the analyte mixture, respectively, were loaded onto the car-

ridges. All the values were, in general, from 75% to 100% for bothypes of resins and, as expected, slightly lower for influent waste-ater samples (due to the matrix complexity). We should just stress

he drop in the recoveries for morphine (near 50%), which comesbout because this compound is one of the most-affected by the MEut also because its polarity might hinder its proper retention in theorbent, but also for methadone in the influent samples with valuesf 24% and 45% for HXLNAD-SCXb and HXLPP-SCXc, respectively.n addition, and as was reported for the ultrapure water samples,rimethoprim was just partially retained and displayed recoveriesot higher than 40%.

In spite of these examples of low recoveries, the extraction per-ormance of both resins were good, and comparable to the resultseported already where similar illicit drugs were extracted usingasis MCX [21,22] (recovery values on average range from 60 to30%), or better when therapeutic drugs were extracted using OasisLB [18,23] (recovery values on average range from 20 to 85%).

t should be mentioned that in the reported studies cited hereigher sample volumes (up to 250 mL) of wastewater were per-olated through cartridges packed with 150 mg of sorbent; thishould be equivalent in terms of capacity in the present case when0 mL of sample were loaded through 60 mg of packed resin. Thus,he morphology of the tested resins (i.e., low particle size andypercrosslinked structure) also helps in the performance of thextraction. This fact is also supported in a previous study [24] wherewo non-hypercrosslinked SCX resins prepared by an alternativeolymerisation protocol which gave irregular particles presentedimilar results in %ME values and %R under the same SPE conditions

s used in the present study, but with 200 mg of polymer packed inhe cartridge.

When comparing HXLPP-SCXb and HXLNAD-SCXc in terms ofrecovery and %ME, we cannot draw any significant conclusion

n the %recovery values for four selected basic analytes.

since the results are quite similar, although the strong cation-exchange capacity for HXLNAD-SCXb is lower than HXLPP-SCXc.In any case the ion-exchange capacity must be enough for theaccessibility of the analytes and their proper retention. ResinHXLNAD-SCXb may potentially provide slightly better results, thuswe selected it for further experiments.

3.3. Method performance and analysis of samples

The overall analytical method was evaluated for effluent sewage(Table 3) and considered linearity, limits of quantification (LOQs),limits of detection (LODs), repeatability and reproducibility.

Six-point calibration curves with matrix were constructed witha linearity ranging from 25 to 5000 ng/L for all compounds stud-ied with the exception of morphine, ranitidine, dihydrocodeineand EDDP, with calibration curve range from 50 to 5000 ng/L, andantipyrine (100 to 5000 ng/L). The calibration curves are linear withdetermination coefficients (r2) exceeding 0.999 on average.

The LOQs for each compound were taken as the lowest con-centration level of the calibration curve. The LODs, calculated asthe signal-to-noise ratio (S/N) of 3, or when the compounds werepresent in real samples the LODs were estimated from instrumen-tal parameters and by taking into account the recovery and MEfor each compound. The LODs were 5 ng/L, for most of the com-pounds studied, but 2 ng/L for BE, cocaine and propranolol; 10 ng/Lfor morphine, ranitidine and EDDP; and, 100 ng/L for trimethoprim(which might be attributed to the low recoveries achieved duringthe extraction). These LOQs are similar to those reported in the liter-ature [21,25,26] where higher volumes of sample were percolatedbut the signals were more affected by ME which, in turn, negatesthe gain in sensitivity.

The repeatability and reproducibility, expressed as a % of rel-

ative standard deviation (%RSD), were determined by spiking fivereplicates at two different levels (LOQs and 100 ng/L) for each typeof sample; the results obtained were less than 12% and 19%, respec-tively.

N. Fontanals et al. / J. Chromatogr. A 1343 (2014) 55–62 61

1x10

1

+ MRM (267.0 -> 145.0)

2x10

5

+ MRM (315.0-> 176.0)

2x10

0

2

+ MRM (300.0-> 165.0)

1x10

0

2

+ MRM (328.0-> 165.0)

4x10

1

+ MRM (195.0 -> 138.0)

3x10

0

1

+ MRM (290.0 -> 168.0)

3x10

0

2.5

+ MRM (304.0-> 182.0)

2x10

0

1

+ MRM (189.0-> 145.0)

3x10

0

2

4+ MRM (260.0-> 116.0)

vs. Acquisition Time (min)1 2 3 4 5 6 7 8 9 10 11

AAtteennoollooll

RRaanniittiiddiinnee

CCooddeeiinnee

66--aacceettyyllmmoorrpphhiinnee

CCaaff ffeeiinnee

BBEE

CCooccaaiinnee

AAnntt iippyyrriinnee

PPrroopprraannoollooll

ge sample. For experimental conditions, see the text.

taoDiLlFswtatdittjp

Table 4Concentrations of analytes found in influent and effluent wastewater samples(n = 5) when the samples were analysed by SPE/LC-MS/MS using the HXLNAD-SCXbsorbent.

Concentration (ng/L)

Effluent Influent

Morphine <LOQ–157 <LOQAtenolol 869–1323 217–563Ranitidine <LOQ <LOQ–136Dihydrocodeine <LOQ <LOQCodeine <LOQ–85 <LOQ–2926-Acetylmorphine <LOQ–87 <LOQ–124Trimethoprim <LOQ <LOQCaffeine 124–437 982–4577Metoprolol <LOQ–215 <LOQ–62BE 331–3805 248–1311Cocaine 108–344 70–602

Counts

Fig. 2. MRM chromatograms of an influent sewa

Finally, the developed SPE/LC–MS/MS method was applied forhe determination of drugs in influent and effluent samples fromn urban STP with samples taken on different days. The presencef the analytes found was confirmed according to the Commissionecision 2002/657/EC [27]. All the analytes studied were found

n the analysed samples, but some of them at levels below theOQs. Table 4 presents the range of concentration for each ana-yte for both effluent and influent sewage samples. As an example,ig. 2 shows representative MRM chromatograms from the analy-is of one of the influent sewage samples. These values are in lineith those reported at similar STPs [28–31]. We should note that

he low concentration of trimethoprim found in all the samplesnalysed might be attributed to the low recoveries achieved withhe present method. In addition, we should mention that somerugs are present at a higher concentration in effluent than in

nfluent water. This might be due to the fact that the sampling of

he influent and effluent sewage was not performed in the sameime period, but also due to a possible conversion of their con-ugated metabolite to the original substance after the treatmentrocesses.

Antipyrine 35–370 <LOQ–73Propanolol 75–273 <LOQ–336EDDP <LOQ–84 <LOQ–213Methadone <LOQ–180 <LOQ–74

6 omato

4

acaeph

tSp

Ltres

ts

A

(tNtfaRwi

A

fj

[[

[[

[

[[[[

[

[[

[

[

[

[

[

[[[

2 N. Fontanals et al. / J. Chr

. Conclusions

Eight different resins were prepared using different syntheticpproaches; these resins boasted differences in their ion-exchangeapacities and morphological properties. The benefits arising from

combination of properties became apparent when the resins werexploited as SCX sorbents. The resins that presented the best SCXerformance were those that had high ion-exchange capacity andigh specific surface area.

The SCX resin properties were attractive and enabled the reten-ion of a group of therapeutic and illicit drugs from sewage samples.imultaneously, they prevented the ME encountered when com-lex samples are analysed by LC–MS/MS.

A novel SPE method using HXLNAD-SCXb followed byC–MS/MS was validated with excellent linearity, limits of quan-ification (LOQs), limits of detection (LODs), repeatability andeproducibility being observed. Thereafter, the SPE method wasxtended successfully to the analysis of different types of sewageamples with the target drugs present.

These novel SCX sorbents presented are an alternative to fur-her exploit in the determination of contaminants from complexamples.

cknowledgements

The authors thank the Ministry of Science and InnovationCTQ2011-24179) and the Department of Innovation, Universi-ies and Enterprises (Project 2009 SGR 223) for financial support.. Gilart would also like to thank the Department of Innova-

ion, Universities and Enterprises and the European Social Fundor a predoctoral grant (FI-DGR 2011). N. Abdullah would like tocknowledge funding from the Department of Chemical and Natureesources Engineering, University of Malaysia, Pahang. A. Daviesould like to acknowledge funding from the Engineering and Phys-

cal Sciences Research Council (UK).

ppendix A. Supplementary data

Supplementary data associated with this article can beound, in the online version, at http://dx.doi.org/10.1016/.chroma.2014.03.068.

[

[

gr. A 1343 (2014) 55–62

References

[1] D. Fatta-Kassinos, S. Meric, A. Nikolaou, Anal. Bioanal. Chem. 399 (2011) 251.[2] M. Petrovic, M. Farré, M. López de Alda, S. Pérez, C. Postigo, M. Köck, J. Radjen-

ovic, M. Gros, D. Barceló, J. Chromatogr. A 1217 (2010) 4004.[3] K. Wille, J.L. Kiebooms, M. Claessens, K. Rappé, J. Bussche, H. Noppe, N. Praet,

E. Wulf, P. Caeter, C. Janssen, H. Brabander, L. Vanhaecke, Anal. Bioanal. Chem.400 (2011) 1459.

[4] M. Caban, N. Migowska, P. Stepnowski, M. Kwiatkowski, J. Kumirska, J. Chro-matogr. A 1258 (2012) 117.

[5] S.D. Richardson, Anal. Chem. 84 (2011) 747.[6] M. Tobiszewski, A. Mechlinska, B. Zygmunt, J. Namiesnik, TrAC, Trends Anal.

Chem. 28 (2009) 943.[7] L. Ramos, J. Chromatogr. A 1221 (2012) 84.[8] N. Fontanals, P.A.G. Cormack, R.M. Marcé, F. Borrull, Trends Anal. Chem. 29

(2010) 765.[9] I. Marchi, S. Rudaz, J.-L. Veuthey, J. Pharm. Biomed. Anal. 49 (2009) 459.10] P. Vazquez-Roig, C. Blasco, Y. Picó, Trends Anal. Chem. 50 (2013) 65.11] R. Pal, M. Megharaj, K.P. Kirkbride, R. Naidu, Sci. Total Environ. 463–464 (2013)

1079.12] V.A. Davankov, S.V. Rogozhin, M.P. Tsyurupa, US Patent 3729457, 1971.13] M.P. Tsyurupa, Z.K. Blinnikova, Y.A. Davidovich, S.E. Lyubimov, A.V. Naumkin,

V.A. Davankov, React. Funct. Polym. 72 (2012) 973.14] N. Fontanals, P. Manesiotis, D.C. Sherrington, P.A.G. Cormack, Adv. Mater. 20

(2008) 1298.15] P.A.G. Cormack, A. Davies, N. Fontanals, React. Funct. Polym. 72 (2012) 939.16] P. Konieczka, J. Namiesnik, J. Chromatogr. A 1217 (2010) 882.17] N. Gilart, R.M. Marcé, F. Borrull, N. Fontanals, J. Sep. Sci. 35 (2012) 875.18] B. Kasprzyk-Hordern, R.M. Dinsdale, A.J. Guwy, J. Chromatogr. A 1161 (2007)

132.19] E.Y. Ordónez, J.B. Quintana, R. Rodil, R. Cela, J. Chromatogr. A 1256 (2012)

197.20] N. Fontanals, F. Borrull, R.M. Marcé, J. Chromatogr. A 1286 (2013) 16.21] I. González-Marino, J.B. Quintana, I. Rodríguez, M. González-Díez, R. Cela, Anal.

Chem. 84 (2011) 1708.22] L. Bijlsma, J.V. Sancho, E. Pitarch, M. Ibánez, F. Hernández, J. Chromatogr. A 1216

(2009) 3078.23] B. Kasprzyk-Hordern, R. Dinsdale, A. Guwy, Anal. Bioanal. Chem. 391 (2008)

1293.24] N. Gilart, P.A.G. Cormack, R.M. Marcé, N. Fontanals, F. Borrull, J. Chromatogr. A

1325 (2014) 137.25] R. López-Serna, B. Kasprzyk-Hordern, M. Petrovic, D. Barceló, Anal. Bioanal.

Chem. 405 (2013) 5859.26] S. Caldas, C. Bolzan, J. Guilherme, M. Silveira, A. Escarrone, E. Primel, Environ.

Sci. Pollut. Res. 20 (2013) 5855.27] T.E. Commission, Off. J. Eur. Commun. (2002) 8.28] C. Postigo, M.J. López de Alda, D. Barceló, Trends Anal. Chem. 27 (2008) 1053.29] B. Petrie, E.J. McAdam, M.D. Scrimshaw, J.N. Lester, E. Cartmell, Trends Anal.

Chem. 49 (2013) 145.30] C.I. Kosma, D.A. Lambropoulou, T.A. Albanis, Sci. Total Environ. 466–467 (2014)

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

410.