No

Sa

b

c

a

ARRAA

KMDEUcsEPp

1

thesfb(bceucm

mbh

PT

0d

Analytica Chimica Acta 717 (2012) 77– 84

Contents lists available at SciVerse ScienceDirect

Analytica Chimica Acta

j ourna l ho me page: www.elsev ier .com/ locate /aca

atural dissolved organic matter affects electrospray ionization during analysisf emerging contaminants by mass spectrometry

amanthi Wickramasekaraa,c, Selene Hernández-Ruiza, Leif Abrell a,c, Robert Arnoldb, Jon Chorovera,c,∗

Department of Soil, Water and Environmental Science, University of Arizona, Tucson, AZ 85721, USADepartment of Chemical and Environmental Engineering, University of Arizona, Tucson, AZ 85721, USAArizona Laboratory for Emerging Contaminants, University of Arizona, Tucson, AZ 85721, USA

r t i c l e i n f o

rticle history:eceived 26 July 2011eceived in revised form 5 December 2011ccepted 11 December 2011vailable online 19 December 2011

eywords:atrix effectissolved organic matter

a b s t r a c t

Dissolved organic matter (DOM), present in many forms in water, can interfere with analysis of organiccontaminants by atmospheric pressure ionization–mass spectrometry. A quantitative analysis of thisinterference, or matrix effect, on organic contaminant target analyte measurements was carried out usingun-fractionated and fractionated dissolved natural organic matter from the Suwannee River, GA (SROM), astandard reference material, that was directly infused into the tandem mass spectrometer during multiplereaction monitoring (MRM) of a suite of endocrine disrupting compounds–pharmaceuticals and personalcare products (EDC/PPCPs). Most target analytes suffered signal suppression in the presence of bothfractionated and un-fractionated SROM, however greater interferences were measured with fractionated

lectrospray ionizationltra high pressure liquidhromatography-tandem masspectrometry (UHPLC–MSMS)ndocrine disrupting chemicals (EDCs)

relative to bulk SROM. This finding is consistent with the view of organic matter as a supramolecularassociation of low molecular mass components having separate charged and structural features revealedonly after dissociation.

© 2012 Elsevier B.V. All rights reserved.

harmaceuticals and personal careroducts (PPCPs)

. Introduction

Dissolved organic matter is a complex, heterogeneous mix-ure of organic molecules with variable molecular mass andydrophilicity. It can originate from the decay of terrestrial veg-tation and algal biomass, as well as from anthropogenic sourcesuch as treated wastewater and landfill leachate [1]. The traditionalractionation of humic substances is based on the operational solu-ility of humic acid (soluble under alkaline conditions), fulvic acidsoluble under alkaline and acidic conditions) and humin (insolu-le) [2]. Their further purification and isolation from biomolecularomponents has involved sorption onto hydrophobic resins [3]. Anmerging view of humic matter now appreciates the supramolec-

lar associations that occur in the naturally occurring in situomplex mixtures, where a diversity of relatively small molarass organic molecules (including biomolecular fragments derived

Abbreviations: SROM, Suwannee River organic matter; DOM, dissolved organicatter; DOC, dissolved organic carbon; HoA, hydrophobic acid; HoB, hydrophobic

ase; HoN, hydrophobic neutral; HiA, hydrophilic acid; HiB, hydrophilic base; HiN,ydrophilic neutral.∗ Corresponding author at: Department of Soil, Water and Environmental Science,

.O. Box No. 210038, University of Arizona, Tucson, AZ 85721-0038, USA.el.: +1 520 626 5635; fax: +1 520 626 9313.

E-mail address: chorover@cals.arizona.edu (J. Chorover).

003-2670/$ – see front matter © 2012 Elsevier B.V. All rights reserved.oi:10.1016/j.aca.2011.12.019

from proteins and nucleic acids) may be clustered via hydrogenbonding, cation bridging, and hydrophobic interaction [4]. Dis-solved organic matter (DOM) has been observed by transmissionelectron microscopy to self-organize into micelle-like microparti-cles (diameter 400–800 nm) [5]. High performance size exclusionchromatography [6] and diffusion-ordered 1H nuclear magneticresonance spectroscopy [7,8] have revealed the disruption of suchintermolecular associations via introduction of low molar massorganic acids with variable aliphatic chain lengths. We postulatethat such DOM associations may affect the extent to which thesenaturally occurring water matrix components interfere with themass spectrometry analysis and quantification of trace level organiccontaminants [9].

Matrix effects can occur in both electrospray ionization(ESI) [10] and atmospheric pressure chemical ionization (APCI)[11,12] sample introduction methods, but effects are generallygreater for ESI than for APCI [13,14]. The mechanism and ori-gins of the matrix effect are not fully understood but mayresult from a competition for charge between analyte and co-eluting matrix component(s) formed in the source region ofthe mass spectrometer [15,16]. Matrix effects may also arise

due to the presence of less volatile solutes that can decreasethe efficiency of droplet formation through co-precipitation ofanalytes, or by preventing droplets from reaching the criti-cal radius required for gas phase ion emission [13]. Matrix

7 ytica C

ecdtmcm

tlppteeuacTtan

tciswoa[tt

2

2

nXwAptaiwa4MtdRdAtauaCB

2

(

8 S. Wickramasekara et al. / Anal

ffects can also result from high concentrations of interferingompounds that increase the viscosity and surface tension ofroplets thereby reducing solvent evaporation and the ability ofhe analyte to reach the gas phase [17,18]. Molecules with higher

ass can mask the signal of smaller molecules by forming largeromplexes with different ionization potentials [19] and more polarolecules are more susceptible to ion suppression [20].Several prior studies have used a post-column infusion method

o evaluate matrix effects [11,20,21]. In this method, the ana-yte of interest is infused with a syringe pump into the mobilehase stream containing a chromatographically separated com-lex matrix eluted from an LC column [22]. It should be noted thathis post-column infusion method does not provide a quantitativevaluation of matrix effects observed on any specific analyte, andach analyte should be infused separately to appreciate its individ-al effect [11]. In this study we have used a post-column infusionpproach to quantitatively measure interferences on the identifi-ation and quantification individually of fourteen target analytes.hese analytes were selected because they represent a suite ofrace organic contaminants of emerging concern with widely vari-ble physico-chemical properties requiring both ESI positive andegative modes of analysis (Table 1).

Suwannee River natural organic matter (SROM) was chosen forhe current study because it is known to contain a diversity ofomplex organic molecules that is also relatively well character-zed [23]. In order to explore relationships between organic mattertructure and target analyte measurement interferences, SROMas separated via preparative sorption chromatography into six

perationally defined fractions – hydrophobic- and hydrophilic-cids, bases, and neutrals – using a multistep separation scheme24] that was slightly modified [25]. For SROM, four of these frac-ions were predominant and yielded sufficient material for use inhe mechanistic studies described below.

. Experimental

.1. Chemicals and materials

The SROM (cat #1R101) was purchased from the Inter-ational Humic Substances Society (IHSS, St. Paul, MN, USA).AD-8, Amberlist-15 and Amberlist-21 resins for fractionationere purchased from Sigma–Aldrich (St. Louis, MO). Bisphenol, salicylic acid and Tris(2-chloroethyl) phosphate (TCEP) wereurchased from TCI America (Portland, OR), N,N-diethyl-meta-oluamide (DEET), ibuprofen, triclosan and perfluorooctanesulfoniccid (PFOS) were purchased from Alfa Aesar (Lancshire, UK),opromide was purchased from USP (Rockville, MD), tonalide

as purchased from Toronto Research Chemicals (Toronto, CA)nd estrone, carbamazepine, ciprofloxacin, sulfamethoxazole and-nonylphenol were purchased from Sigma–Aldrich (St. Louis,O). Isotopically labeled internal standards were purchased from

wo sources. Ibuprofen-d3, sulfamethoxazole-d4, ciprofloxacin-8, triclosan-d3 and iopromide-d3 were purchased from Torontoesearch Chemicals (Toronto, CA), and 4-nonylphenol-d4, estrone-4, salicylic acid-d3, DEET-d7, carbamazepine-d10 and bisphenol-d16 were purchased from CDN Isotopes (Quebec, CA). Stock solu-

ions of target analytes and internal standards were prepared in 50% (v/v) aqueous MeOH. LCMS reagent water (J.T. Baker) wassed for diluting SROM and OM fractions. For the LC mobile phase,cetonitrile and formic acid (additive) were purchased from EMDhemicals (Gibbstown, NJ), and LCMS reagent water was from J.T.aker.

.2. SROM fractionation

SROM was fractionated into hydrophilic (Hi) and hydrophobicHo) acids (A), neutrals (N) and bases (B) to identify which fractions

himica Acta 717 (2012) 77– 84

interfere with analysis of different target analytes. All resins werecleaned and pre-conditioned according to the method describedby Allard et al. [26]. The fractionation procedure was performedaccording to methods developed by Leenheer [24] and modified byChefetz et al. [25]. Briefly, 120 mg of SROM dissolved in 200 mL ofMilliQ water (pH 6.5) was carried through the fractionation process.Dissolved organic carbon (DOC) of the solution was measured priorto the experiment. The SROM solution was mixed with an excessof hydrophobic XAD-8 resin (0.1 mL of resin per mg of carbon) ina glass bottle and shaken for 24 h to allow adsorption. Non-boundfractions were separated by vacuum filtration through a 0.45 �mhydrophilic polypropylene filter. Desorption of the HoB fractionfrom the XAD-8 resin was achieved by agitation with 0.14 L kg−1

(volume of solvent per unit mass of the resin) of 0.1 M HCl for 24 hfollowed by 0.43 L kg−1 of 0.01 M HCl for 2 h. Vacuum filtration wasused to separate the desorbed HoB fraction from the resin. Sorptionof HoA onto XAD-8 was accomplished by acidifying the remainingDOM solution (following HoB removal) to pH 2 and mixing withXAD-8 resin by overnight shaking. New hydrophilic non-boundfractions were eluted from the XAD-8 resin by vacuum filtration.The HoA fraction was desorbed from the resin with 0.14 L kg−1 of0.1 M NaOH and rinsed with 0.43 L kg−1 of H2O. To desorb the HoNfraction, 1 L kg−1 of MeOH was incubated with the resin overnightin an orbital shaker. The HoN fraction was collected by vacuumfiltration. The HoN fraction was reduced in a Turbo Evaporator(BiotageAB, Charlotte, NC) and reconstituted in water for furtheranalysis.

Remaining hydrophilic fractions in the DOM solution wereadsorbed onto an A-15 resin (amount calculated based on the elec-trical conductivity (EC) of the DOM solution) by shaking for 24 h.Solution containing fractions HiA and HiN was then eluted fromthe resin by vacuum filtration. The HiB fraction was desorbed fromA-15 resin by shaking with 0.3 L kg−1 of 0.1 M NH4OH for 24 h. Theremaining two non-bound fractions, HiA and HiN, were separatedby shaking the solution with A-21 resin (amount of the resin wascalculated based on the pH and EC of the solution) for 24 h. UnboundHiN fraction was then separated by vacuum filtration. The HiA frac-tion was desorbed from A-21 resin by shaking with 1 L kg−1 of 1 MNaOH for 24 h. Amount of DOC recovered in each fraction was mea-sured using a Shimadzu TOC-VCSH total organic carbon analyzer(Columbia, MD). The pH of fractionated and bulk SROM was mea-sured and adjusted to 6.5 (equal to LCMS reagent water, J.T. Baker)using HCl and CH3CO2NH4.

2.3. Full scan experiment with fractionated and un-fractionatedorganic matter

Full scan spectra were obtained by directly infusing the frac-tionated and un-fractionated organic matter into a Quattro PremierXE triple quadrupole mass spectrometer (Waters, Milford, MA)equipped with electrospray ionization source. Organic matter wasinfused by syringe pump with a 250 �L Hamilton syringe at a flowrate of 20 �L min−1 into a UHPLC (ultra high pressure liquid chro-matography) stream of 50% (v/v) aqueous ACN at 0.05 mL min−1

through a T-junction. Spectra were obtained by ESI negative ionmode with capillary and cone voltage of 2.4 kV and 25 V, respec-tively. The spectra were acquired with a mass range of 100–500 m/zand a total of 600–800 spectra were averaged for each mass spec-trum.

2.4. Matrix effect analyses by UHPLC–MSMS

A Waters Acquity UHPLCTM system (Waters, Milford, MA) con-sisting of Acquity UHPLCTM binary solvent manager and AcquityUHPLC BEH C18 column (1.7 �m; 2.1 mm × 50 mm) (Waters, Mil-ford, MA) was used for the separation analysis. A mixture of

S. Wickramasekara et al. / Analytica Chimica Acta 717 (2012) 77– 84 79

Table 1Properties and MRM conditions for target analytes.

Compounda CAS # Internal standardb Log Kowc pKa MRM transition CVd/CEe

Tonalide 21145-77-7 NA 6.4 NA 259.33 > 175.12 26/204-Nonylphenol 104-40-5 4-Nonylphenol d4 5.1 7.06 219.44 > 106.26 45/20Triclosan 3380-34-5 Triclosan d3 4.5 8.14 289.09 > 35.24 21/6Bisphenol A 80-05-7 Bisphenol A d16 3.3 10 227.19 > 212.47 35/19Estrone 53-16-7 Estrone d4 3.3 10.4 269.22 > 145.54 49/48Carbamazepine 298-46-4 Carbamazepine d10 2.6 13.9 237.25 > 194.45 35/21DEET 134-62-3 DEET d7 2 0.67 192.17 > 119.00 30/18Ibuprofen 15687-27-1 Ibuprofen d3 1.9 4.91 205.32 > 161.35 13/6Sulfamethoxazole 723-46-6 Sulfamethoxazole d4 0.9 1.83/5.57 254.26 > 155.93 25/16TCEP 115-96-8 NA 0.9 NA 286.93 > 98.89 30/21Iopromide 73334-07-3 Iopromide d3 0 10.2 789.22 > 127.20 25/38PFOS 2795-39-3 NA −1.1 −3.3 498.87 > 99.31 68/50Ciprofloxacin 85721-33-1 Ciprofloxacin d8 −1.2 6.09/8.74 332.34 > 288.16 41/18Salicylic acid 69-72-7 Salicylic acid d6 −1.7 2.97 137.09 > 93.27 29/16

a Compounds are arranged according to the decreasing order of log Kow value.b Internal standard used for the quantification.c Octanol water partition coefficient.d Cone voltage (CV).e Collision energy (CE).

nt des

fwbwtDosfvt1sa

(supirdauwrtMs

M

Fig. 1. Infusion experime

ourteen target analytes (ca. 1 mg L−1 each; Table 1) was spikedith eleven deuterated internal standards (ca. 500 �g L−1 each)

efore mass spectrometric analysis. Matrix effects experimentsere carried out by T-mixing a 50 �L min−1 infused flow of frac-

ionated or un-fractionated SROM (at 1, 5, 10, 15, 25, or 50 mg L−1 ofOC) with the eluent stream (300 �L min−1) from a 5 �L injectionf the target analyte mixture (Fig. 1). A binary solvent gradient con-isting of water (A) and acetonitrile (B) (containing 0.1% formic acidor the positive ion mode) was used for UHPLC separation. The sol-ent gradient was initiated with 10% B that was linearly increasedo 40% over 2 min, to 60% over 2 min, to 95% over 1 min, and then to00% over 1 min, where it was held for 0.5 min. The solvent compo-ition was then returned to the starting level (10% B) over 0.5 minnd held at 10% B for 1 min, giving a total run time of 10 min.

A Quattro Premier XE triple quadrupole mass spectrometerWaters, Milford, MA) equipped with an electrospray ionizationource was used for the analyses. A 10 mL Hamilton syringe wassed to infuse the fractionated or un-fractionated SROM by syringeump. Analyses were performed in both positive and negative

onization mode using a capillary voltage of 3.2 kV and 2.9 kV,espectively. The ion source temperature was kept at 110 ◦C andesolvation gas temperature was set to 350 ◦C. Nitrogen was useds both cone gas and desolvation gas, and high purity argon wassed as the collision gas. MassLynx 4.1 (Waters, Milford, MA) soft-are was used to control instruments and record data. Multiple

eaction monitoring (MRM) was used to selectively identify thearget analytes with a dwell time of 100 ms per ion pair (Table 1).

atrix effects are expressed as a percentage of target analyte signal

uppression (or enhancement) by the following equation (Eq. (1))

atrix effect (ME)% = PA(OM) − PA(W)PA(W)

× 100 (1)

ign with UHPLC–MSMS.

where PA(OM) is the peak area of the target analyte in the presenceof dissolved organic matter and PA(W) is the peak area of the targetanalyte in the presence of water alone (blank) [27].

3. Results and discussion

3.1. SROM fractionation

3.1.1. YieldSROM contains a heterogeneous and complex mixture of organic

molecules. This is confirmed by previously reported data [28]. BulkSROM was fractionated, using a resin-based method, into six differ-ent fractions. Approximately 60% by mass of the bulk SROM carbonwas recovered as fraction HoA (Fig. 2), in agreement with liter-ature reports [29,30]. A significantly greater proportion of SROMwas recovered as hydrophobic relative to hydrophilic fractions,also consistent with prior literature reports for DOM of terres-trial vegetation origin [31]. Only 3% of the SROM was recoveredas basic fractions, which is characteristic of the low recovery ofbasic fraction of the organic matter, originating from natural waters[29,32,33].

3.1.2. Full scan mass spectra comparison for different DOMfractions

Full scan mass spectra obtained for bulk and fractionated (HoA,HiA and HoN – low percentage recovery fractions were not ana-lyzed) SROM are distinct (Fig. 3). SROM masses were distributedaround 400 m/z, in agreement with literature reported full scan

mass spectra for SROM [34]. This mass distribution can be a resultof multiply charged ions and adducts formed in the ESI sourceregion [3]. The largest fraction by percent recovery, HoA, showedthree mass distributions centered around 163, 265 and 369 m/z.

80 S. Wickramasekara et al. / Analytica C

Ft

TmttcTcow

were observed from post column infusion of different SROM frac-

ig. 2. Mass percent recovery (based on DOC measurement) of the six organic frac-ions obtained from bulk SROM.

hese m/z values are not due to multiple charging of the sameolecule. HiA and HoN fractions showed simpler spectra than did

he HoA fraction. The hydrophilic fraction is known to contain pro-eins, amino acids, amino sugars, and also hydrophilic acids thatan be formed by the oxidative degradation of hydrophobic acids.hese components evidently produce a simpler mass spectrum

ompared to that of the HoA fraction [35,36]. The full scan spectrumf the HiA fraction contained abundant odd-numbered m/z peakshich indicate molecular formulae with zero or an even number of

Fig. 3. Full scan mass spectra for (A) SROM, (B) HiA fraction, (C) HoA fract

himica Acta 717 (2012) 77– 84

nitrogen atoms [3]. The HoN fraction, which was difficult to ionizeby negative mode ESI, probably consists mainly of hydrocarbonsand tannins [35]. Nonpolar, low molecular weight compoundspresent in water may be responsible for the abundant peaksobserved in the full scan spectrum of the HoN fraction, but fur-ther selected fragmentation experiments are needed to positivelyidentify such compounds.

3.2. Matrix effect analysis by post column infusion method

Post column infusion of different SROM concentrations duringMRM analysis of chromatographed target analytes allowed us toquantify matrix effects caused by the presence of DOM. Matrixeffects were primarily manifest as ion suppression for most targetanalytes (Fig. 4). Panels in Fig. 4 are arranged in order of decreasinglog Kow value (i.e., decreasing hydrophobicity) of target analytes.The pH values of fractions HoB and HoA were adjusted to 6.5 withHCl and CH3CO2NH4 prior to analysis in order to maintain a con-sistent pH for all fractions tested.

Although ionization suppression is known to occur during theelectrospray process, effects of the mixing time between SROM(fractions) and target analytes were examined. Target analytes hadless than 20 ms to interact with the infused SROM solution duringthe post column infusion experiment. Nonetheless, significant sig-nal suppression, and some signal enhancement, of target analytes

tions. We examined the effect of residence time on the amount ofsignal suppression/enhancement by changing the tube length afterthe “T” junction and confirmed that increasing the mixing time does

ion and (D) HoN fraction obtained in ESI negative ionization mode.

S. Wickramasekara et al. / Analytica Chimica Acta 717 (2012) 77– 84 81

F on of dl .

n(

iiScaom

ig. 4. Calculated matrix effects for target analytes arising from post column infusiog Kow values (left column top to bottom, followed by right column top to bottom)

ot significantly change the magnitude or sign of matrix effect (Eq.1)) caused by the organic matter.

For most of the target analytes, matrix effects arising fromndividual SROM fractions were more pronounced than effects aris-ng from the un-fractionated SROM (Fig. 4). To the extent thatROM behaves as an aggregated mixture of low molecular mass

omponents [4], the disruption of this assembly during the fraction-tion process is expected to expose physiochemical characteristicsf the individual fractions that are otherwise masked by inter-olecular association. Through fractionation, interior chemical

ifferent fractions and bulk SROM. Compounds are arranged in order of decreasing

functionalities may be revealed resulting in a chemical behaviorthat is distinct from the bulk. Newly revealed features of fraction-ated, and dissociated DOM, may contribute to signal suppressionin the electrospray process by changing the viscosity and surfacetension of the ESI droplets [17,37], resulting in less efficient tar-get analyte ionization. Electrospray ionization reactions depend

on ionization energy and proton affinity of all molecules presentat the ESI interface. Ability of the target analyte to become, andremain, ionized may also depend on competition for charge withinterfering DOM species. Signal enhancement that was observed

82 S. Wickramasekara et al. / Analytica Chimica Acta 717 (2012) 77– 84

F asis) fm

ftd

auaNacwnrashptpwtit

mgfpHe

ig. 5. Ion suppression of target analytes arising from (A) 15 mg L−1 HiA (carbon batrix effect).

or carbamazepine and nonylphenol with lower DOM concentra-ions may be due to charge stabilization by DOM species presenturing the electrospray process [27,38].

The greatest signal suppression measured for most of the targetnalytes occurred in the presence of the HiA fraction. Intermolec-lar interactions (e.g., bridging) between hydrophilic moleculesre notably greater than between hydrophobic molecules [39–41].egative mode ion suppression of the most hydrophilic targetnalytes in the presence of HiA fraction may be explained byharge-masking from supramolecular organic matter complexesith net negative charge [4]. Another possible mechanistic expla-ation for negative ion mode signal suppression is charge–chargeepulsion during the electrospray process, effectively reducing themount of ions entering the mass analyzers [42]. Target analytestudied containing a phenolic group in their structure tend to be lessydrophilic and hence were subject to less ion suppression in theresence of 15 mg L−1 HiA (carbon basis) fraction [Fig. 5(A)]. Posi-ive mode ion suppression may result by proton abstraction fromositively charged target analytes to organic matter complexesith a net negative charge. Overall, target compounds analyzed in

he positive ionization mode suffered from greater ion suppressionn the presence of the HiA fraction compared to those analyzed inhe negative ion mode (except for ibuprofen and iopromide).

An opposite trend of ion suppression is shown in Fig. 5(B) withatrix effects from the 15 mg L−1 HoB (carbon basis) fraction. Tar-

et compounds analyzed in the negative ionization mode suffered

rom greater ion suppression in the presence of HoB fraction com-ared to target compounds analyzed in the positive ion mode. TheoB fraction predominantly contains weakly basic aromatic moi-ties such as aromatic amines [35,43]. Protonated amine groups

raction, (B) 15 mg L−1 HoB (carbon basis) fraction (arranged in order of increasing

present in the HoB fraction under normal LC conditions (at pH 6.0),can donate a proton to negatively charged target analyte ions (innegative mode ESI) effectively neutralizing the target analyte andsuppressing the negative ionization mode signal.

Matrix effect data were obtained experimentally by normalizingto the pH of the 50 mg L−1 SROM solution. However, bulk SROM andfractionated DOM samples have different ionic strengths. In orderto evaluate the potential effect of DOM background electrolyte,control experiments were carried out by infusing NaCl solutioninstead of DOM. Ionic strengths of tested NaCl solutions were var-ied between 0 and 250 �S cm−1 to encompass an even greaterrange of conductivity values than those measured from infused,fractionated DOM samples. Infusion of NaCl solutions also showedinterferences with both positive and negative mode electrosprayionization. In the positive ionization mode, sodiated ions producedwill have different parent and daughter ion masses, thereby reduc-ing the signal intensity of the analyte, and competing chloride ionscan acquire protons from the charged analyte molecules, therebyneutralizing the analyte. In the negative ionization mode, neutral-ization of the target molecules can result in signal suppression.

Charged DOM and inorganic ions both contribute to the totalionic strength of the solution. Constantopoulos et al. [44] studiedthe effect of salt concentration on analyte response using electro-spray ionization mass spectrometry and compared that with theEnke’s equilibrium partitioning model. They observed an increasinganalyte response corresponding to an increase in NaCl concen-

tration from 10−6 to 10−4 M, but at 10−3 M NaCl the analytesignal response decreased. Our experimental NaCl concentrations(data not shown) were similar, within the range of 10−4–10−3 Mand our experiments demonstrated that DOM, as well as the

tica C

emnNcctbitMfitps

imaEtawoie

4

sh[Sc[iLumawbatarscatdo(dmpsaueDom

[[

[

[[

[

[

[

[[

[

[[[

[[[

[

[

[

[

[

[[

[

[[[[

S. Wickramasekara et al. / Analy

lectrolytes present in solution, contributed to the measuredatrix effect on analyte response. Constantopoulos et al. offer alter-

ative explanations for the changing analyte response at differentaCl concentrations. Reduced analyte response at the highest NaCloncentration may result from higher droplet charge densities thatause repulsive forces which spread out the spray diameter, effec-ively sending fewer ions into the mass spectrometer. At lower,ut increasing, NaCl concentrations the analyte response may also

ncrease as a result of better analyte ion desolvation determined byhe analyte ion location in the droplet electrical double layer [44].

ore hydrophobic analytes with higher desolvation energies willractionate to the compact (outer droplet) layer, especially withncreasing droplet solution ionic strength, where they will be par-ially desolvated on the droplet surface before entering the gashase. Also, greater solution conductivity produces smaller dropletizes which will desolvate more efficiently [45].

Inclusion of isotopically labeled internal standards (isotopologs)n the analysis method is an accepted approach to help overcome

atrix effects [15,46]. In our experiments eleven different deuter-ted internal standards were used for the target analyte mixtureSI MSMS analysis. Just as individual environmental samples con-aining matrix components might be expected to produce differentmounts of matrix effects, this was observed in our experimentsith different DOM fractions. These differences were also borne

ut as decreased (or increased) recoveries of isotopically labelednternal standards. No particular trend in ionization suppression ornhancement of the internal standards as a group was observed.

. Conclusions

Suwannee River organic matter (SROM) was fractionated intoix operationally defined pools and, in agreement with past studies,ydrophobic acids (HoA) comprised the highest yielding fraction29]. Full scan mass spectra obtained for both un-fractionatedROM and the HoA fraction showed similar m/z peak distributions,onsistent with the prevalence of hydrophobic acids in the SROM47]. HoA, HiA and HiN fractions were used for infusion exper-ments. When target analyte samples were co-injected into theCMSMS mobile phase stream in the presence of fractionated andn-fractionated SROM, target analytes were subjected to variableatrix effects. Differences in ion suppression (or enhancement)

rose due to physiochemical properties of the target analytes, asell as because of reactive functional groups potentially revealed

y DOM fractionation. Almost all of the tested DOM fractionsffected target analyte electrospray ionization to a greater degreehan did bulk DOM, possibly because of disruption of molecularggregates that predominate in the bulk form but that are dis-upted upon fractionation. The discovery that greater ionizationuppression arises from DOM fractions than from bulk DOM isonsistent with the conceptualization of natural organic matter as

supramolecular association of low molecular mass componentshat will reveal their separate charged and structural features uponissociation [4]. The greatest signal suppression observed for mostf the target analytes arose from co-infusion of the hydrophilic acidHiA) fraction. Ionization suppression by the HiA fraction may beue to masking of the target analyte charge by complexed organicatter [35,41]. Most of the target compounds analyzed in the ESI

ositive mode in the presence of un-fractionated SROM showedignal suppression possibly arising from proton transfer to neg-tively charged SROM rendering the target analyte neutral andn-detectable. While elucidation of matrix effect mechanisms is

nabled by examining different fractions of DOM, in natural waters,OM exists in bulk, un-fractionated form. Improved understandingf DOM matrix effects on emerging contaminant quantification byass spectrometry will, therefore, require a parallel set of advances

[[[

himica Acta 717 (2012) 77– 84 83

in our understanding of intermolecular associations of DOM com-ponents themselves.

Acknowledgements

Research support was provided by the Water Research Foun-dation (Award #4269), the Binational Agricultural Research andDevelopment (BARD) fund (Grant # IS-3822-06), and Universityof Arizona Water Sustainability Program. The comments and viewsdetailed herein may not necessarily reflect the views of the WaterResearch Foundation, its officers, directors, affiliates, or agents.Analyses in the Arizona Laboratory for Emerging Contaminantswere supported by NSF CBET 0722579.

References

[1] R. Sutton, G. Sposito, M.S. Diallo, H.R. Schulten, Environ. Toxicol. Chem. 24(2005) 1902.

[2] F. Amery, C. Vanmoorleghem, E. Smolders, Eur. J. Soil Sci. 60 (2009) 956.[3] J.A. Leenheer, T.I. Noyes, C.E. Rostad, M.L. Davisson, Biogeochemistry 69 (2004)

125.[4] R. Sutton, G. Sposito, Environ. Sci. Technol. 39 (2005) 9009.[5] M. Kerner, H. Hohenberg, S. Ertl, M. Reckermann, A. Spitzy, Nature 422 (2003)

150.[6] A. Piccolo, Soil Sci. 166 (2001) 810.[7] A.J. Simpson, Magn. Reson. Chem. 40 (2002) S72.[8] D. Smejkalova, A. Piccolo, Environ. Sci. Technol. 42 (2008) 699.[9] Z.H. Rivera, E. Oosterink, L. Rietveld, F. Schoutsen, L. Stolker, Anal. Chim. Acta

700 (2011) 114.10] F. Worrall, T. Besien, D.W. Kolpin, Sci. Total Environ. 299 (2002) 131.11] H. Trufelli, P. Palma, G. Famiglini, A. Cappiello, Mass Spectrom. Rev. 30 (2011)

491.12] H.R. Liang, R.L. Foltz, M. Meng, P. Bennett, Rapid Commun. Mass Spectrom. 17

(2003) 2815.13] T.M. Annesley, Clin. Chem. 49 (2003) 1041.14] A. Apffel, S. Fischer, G. Goldberg, P.C. Goodley, F.E. Kuhlmann, J. Chromatogr. A

712 (1995) 177.15] B.K. Matuszewski, M.L. Constanzer, C.M. Chavez-Eng, Anal. Chem. 75 (2003)

3019.16] R. King, R. Bonfiglio, C. Fernandez-Metzler, C. Miller-Stein, T. Olah, J. Am. Soc.

Mass Spectrom. 11 (2000) 942.17] J.P. Antignac, K. de Wasch, F. Monteau, H. De Brabander, F. Andre, B. Le Bizec,

Anal. Chim. Acta 529 (2005) 129.18] C.R. Mallet, Z.L. Lu, J.R. Mazzeo, Rapid Commun. Mass Spectrom. 18 (2004) 49.19] J.L. Sterner, M.V. Johnston, G.R. Nicol, D.P. Ridge, J. Mass Spectrom. 35 (2000)

385.20] R. Bonfiglio, R.C. King, T.V. Olah, K. Merkle, Rapid Commun. Mass Spectrom. 13

(1999) 1175.21] Y. Gru, R. Colin, P. Le Cloirec, J. AOAC Int. 93 (2010) 1020.22] H. Stahnke, T. Reemtsma, L. Alder, Anal. Chem. 81 (2009) 2185.23] G.V.K. Jean-Phillipe Croue, M.M. Benjamin, Characterization of Natural Organic

Matter in Drinking Water, AWWA Research Foundation, 2000.24] J.A. Leenheer, Environ. Sci. Technol. 15 (1981) 578.25] B. Chefetz, Y. Hadar, Y. Chen, Acta Hydrochim. Hydrobiol. 26 (1998) 172.26] B. Allard, H. Borén, A. Grimvall, Humic Substances in the Aquatic and Terrestrial

Environment, Springer-Verlag, Berlin, New York, 1991.27] F. Gosetti, E. Mazzucco, D. Zampieri, M.C. Gennaro, J. Chromatogr. A 1217 (2010)

3929.28] S.B. Schwede-Thomas, Y.P. Chin, K.J. Dria, P. Hatcher, E. Kaiser, B. Sulzberger,

Aquat. Sci. 67 (2005) 61.29] T. Ratpukdi, J.A. Rice, G. Chilom, A. Bezbaruah, E. Khan, Water Environ. Res. 81

(2009) 2299.30] P.F. Greenwood, J.A. Leenheer, C. McIntyre, L. Berwick, P.D. Franzmann, Org.

Geochem. 37 (2006) 597.31] M. Schumacher, I. Christl, R.D. Vogt, K. Barmettler, C. Jacobsen, R. Kretzschmar,

Biogeochemistry 80 (2006) 263.32] G.V. Korshin, M.M. Benjamin, R.S. Sletten, Water Res. 31 (1997) 1643.33] J. Swietlik, A. Dabrowska, U. Raczyk-Stanislawiak, J. Nawrocki, Water Res. 38

(2004) 547.34] A. Gaspar, E.V. Kunenkov, R. Lock, M. Desor, I. Perminova, P. Schmitt-Kopplin,

Rapid Commun. Mass Spectrom. 23 (2009) 683.35] J.A. Leenheer, J.P. Croue, Environ. Sci. Technol. 37 (2003) 18a.36] G. Guggenberger, W. Zech, Sci. Total Environ. 152 (1994) 37.37] R.C. Deng, P. Williams, Anal. Chem. 66 (1994) 1890.38] D. Remane, D.K. Wissenbach, M.R. Meyer, H.H. Maurer, Rapid Commun. Mass

Spectrom. 24 (2010) 859.39] A.A. MacKay, B. Canterbury, J. Environ. Qual. 34 (2005) 1964.40] L. Aristilde, G. Sposito, Environ. Toxicol. Chem. 29 (2010) 90.41] M.A. Zazouli, M. Ulbricht, S. Nasseri, H. Susanto, J. Iran, Environ. Health 7 (2010)

15.

8 ytica C

[

[[

4 S. Wickramasekara et al. / Anal

42] Z.R. Wu, W.Q. Gao, M.A. Phelps, D. Wu, D.D. Miller, J.T. Dalton, Anal. Chem. 76(2004) 839.

43] J. Dilling, K. Kaiser, Water Res. 36 (2002) 5037.44] T.L. Constantopoulos, G.S. Jackson, C.G. Enke, J. Am. Soc. Mass Spectrom. 10

(1999) 625.

[[

[

himica Acta 717 (2012) 77– 84

45] J.F. Delamora, I.G. Loscertales, J. Fluid Mech. 260 (1994) 155.46] B.J. Vanderford, S.A. Snyder, Environ. Sci. Technol. 40 (2006)

7312.47] E. Perdue, J. Ritchie, in: H.D. Holland, K.K. Turekian (Eds.), Dissolved Organic

Matter in Freshwaters, Elsevier, Amsterdam, 2004.