Cfw

Pa

b

a

AA

KEACMM

1

tstbenfnwi[p

s

0d

Journal of Chromatography A, 1204 (2008) 197–203

Contents lists available at ScienceDirect

Journal of Chromatography A

journa l homepage: www.e lsev ier .com/ locate /chroma

omparison of electrospray ionization and atmospheric pressure photoionizationor coupling of micellar electrokinetic chromatographyith ion trap mass spectrometry

aul Hommersona,∗, Amjad M. Khanb, Gerhardus J. de Jonga, Govert W. Somsena

Department of Biomedical Analysis, Utrecht University, Utrecht, The NetherlandsAnalytical Chemistry, Process R&D, AstraZeneca, Macclesfield, UK

r t i c l e i n f o

rticle history:vailable online 15 April 2008

eywords:lectrospray ionizationtmospheric pressure photoionizationapillary electrophoresisass spectrometryicellar electrokinetic chromatography

a b s t r a c t

The performance of dopant-assisted atmospheric pressure photoionization (DA-APPI) and electrosprayionization (ESI) for the coupling of micellar electrokinetic chromatography (MEKC) with ion trap massspectrometry (ITMS) was compared using a set of test drugs comprising basic amines, steroids, esters,phenones and a quaternary ammonium compound. The influence of the surfactant sodium dodecyl sulfate(SDS) on analyte signals was studied by infusion of sample through the CE capillary into the respective ionsources. It was found that background electrolytes (BGEs) containing 20–50 mM SDS in 10 mM sodiumphosphate (pH 7.5) caused major ionization suppression for both polar and apolar compounds in ESI-MS, whereas APPI-MS signal intensities remained largely unaffected. ESI gave rise to the formation ofSDS clusters, which occasionally may cause space-charge effects in the ion trap. Furthermore, extensivesodium-adduct formation was observed for medium polar compounds with ESI-MS, whereas these com-

pounds were detected as their protonated molecules with APPI-MS. Using the BGE containing 20 mM SDS,MEKC-ESI-MS still provides slightly lower limits of detection (LODs) (2.6–3.1 �M) than MEKC-APPI-MS(4.3–6.4 �M) for basic amines. For less polar compounds, highest S/Ns were obtained with APPI-MS detec-tion (LODs, 4.5–71 �M). For BGEs containing 50 mM SDS, the limits of detection for MEKC-APPI-MS weremore favorable (factor 1.5–12) than MEKC-ESI-MS for nearly all tested drugs. Spray shield contamination

PPI-MMS, e

Hs(peep[tMo

by SDS was lower in DA-Acharacteristics for MEKC–

. Introduction

Micellar electrokinetic chromatography (MEKC) is an elec-rodriven separation technique characterized by the addition ofurfactants to the background electrolyte (BGE) [1]. The separa-ion of analytes in MEKC is based on a differential partitioningetween the micellar and aqueous phase, and on differences inlectrophoretic mobility, enabling simultaneous analysis of botheutral and charged solutes. Therefore, MEKC is particularly useful

or profiling studies in which the full nature of the sample compo-ents may be unknown. MEKC-UV is a widely accepted technique

hich found application into many areas of analytical science,

ncluding forensic, pharmaceutical and environmental analysis2,3]. For a further extension of the applicability of MEKC, its cou-ling with mass spectrometry (MS) would be highly desirable.

∗ Corresponding author at: Department of Biomedical Analysis, Utrecht Univer-ity, P.O. Box 80082, NL-3508 TB Utrecht, The Netherlands. Fax: +31 30 253 5180.

E-mail address: P.Hommerson@uu.nl (P. Hommerson).

[h[

srriM

021-9673/$ – see front matter © 2008 Elsevier B.V. All rights reserved.oi:10.1016/j.chroma.2008.04.017

S than in ESI-MS. It is concluded that DA-APPI shows the most favorablespecially when compounds of low polarity have to be analyzed.

© 2008 Elsevier B.V. All rights reserved.

owever, the direct coupling of MEKC with MS has always been con-idered problematic, as commonly used pseudo-stationary phasesPSPs), such as sodium dodecyl sulfate (SDS), cause ionization sup-ression, source contamination and/or high background signals inlectrospray ionization (ESI)-MS [4–6]. Over the past few years, sev-ral approaches to MEKC-ESI-MS have been developed includingartial-filling MEKC [7,8], and the use of reverse-migrating micelles9]. These modifications often require analyte-specific optimiza-ion and, therefore, are less suitable for the development of generic

EKC–MS methods. A promising approach to MEKC–MS is the usef special pseudo-stationary phases like, e.g., volatile surfactants10,11], or high molecular weight PSPs [12]. The latter type of PSPsave also been shown to be highly suitable for chiral MEKC–MS13–15].

The direct coupling of SDS-MEKC and ESI-MS for the analy-

is of basic compounds has been reported [16–19]. These studiesevealed significant analyte ionization suppression by SDS but theesulting detection limits were still sufficient for relevant analyt-cal tasks like drug impurity profiling. An alternative strategy for

EKC–MS concerns the use of ionization techniques that show a

1 omato

basptosowa

nldlom[ibptcse

2

2

obsNf

BmTm(MDc

o(obw1iat

dmcaaclEv

2

F(

98 P. Hommerson et al. / J. Chr

etter compatibility with PSPs. Takada et al. introduced the use oftmospheric pressure chemical ionization (APCI) for MEKC–MS andhowed that buffer additives such as SDS did not significantly com-romise ionization efficiency of amine compounds [20]. However,he detection limits obtained were rather unfavorable. Recently,ur group described the hyphenation of MEKC and MS using atmo-pheric pressure photoionization (APPI) [21]. It appeared that APPIf amine model compounds was not significantly affected by SDS,hile detection limits down to 1 �g/ml (full scan mode) could be

chieved.Although MEKC–MS may be especially suitable for analysis of

eutral compounds, little quantitative data for this type of ana-ytes is currently available, neither with ESI-MS, nor with APPI-MSetection. Recent comparative studies of ESI and APPI for capil-

ary zone electrophoresis (CZE)-MS indicated that when volatiler low concentrations of nonvolatile BGEs are used, ESI providesore favorable detection limits than APPI for basic compounds

22,23]. APPI showed to be most promising when BGEs contain-ng strong ionization-suppressing agents, like surfactants, have toe used. Therefore, in the present study, a direct comparison of theerformance of ESI-MS and APPI-MS is made under MEKC condi-ions, i.e., with SDS added to the BGE, with a focus on neutral testompounds. The effect of the BGE on analyte ionization, analyteignal intensity, background signals and source contamination isvaluated and discussed.

. Experimental

.1. Chemicals and materials

Sodium dodecyl sulfate, sodium dihydrogenphosphate, dis-

dium hydrogenphosphate and sodium hydroxide were suppliedy Merck (Darmstadt, Germany). Formic acid, toluene and diphenylulfide were purchased from Sigma-Aldrich (Zwijndrecht, Theetherlands). Acetophenone and valerophenone were purchased

rom Fluka (Zwijndrecht, The Netherlands). Methanol was from

g51I

ig. 1. Molecular structures of (A) methyl atropine, (B) fluvoxamine, (C) mebeverine, (D) preH) propyl para-hydroxybenzoate (I) acetophenone and (J) valerophenone.

gr. A 1204 (2008) 197–203

iosolve (Valkenswaard, The Netherlands). Fluvoxamine andebeverine were obtained from Solvay Pharmaceuticals (Weesp,

he Netherlands). Methyl atropine, hydrocortisone, prednisone,ethyl 4-hydroxybenzoate (MOB) and propyl 4-hydroxybenzoate

POB) were from Fagron (Nieuwekerk a/d IJssel, The Netherlands).olecular structures of the test compounds are depicted in Fig. 1.eionized water was filtered and degassed before use. Fused-silicaapillaries were from BGB Analytik (Boekten, Switzerland).

For the APPI-MS and ESI-MS infusion experiments, solutionsf test compounds were prepared in 50 mM ammonium acetatepH 6.8) and 10 mM sodium phosphate (pH 7.5) containing 0, 20r 50 mM SDS. All BGEs were filtered through a 0.20 �m filterefore use. Analyte concentrations in ESI-MS infusion experimentsere 5 �M for methyl atropine, mebeverine and fluvoxamine, and

00 �M for the additional test compounds, whereas in APPI-MSnfusion experiments the concentrations were 100 �M for methyltropine, mebeverine and fluvoxamine and 500 �M for the addi-ional test compounds.

For MEKC–MS experiments, test compounds were individuallyissolved in water-BGE (1:1, v/v), at concentrations of 100 �M forethyl atropine, fluvoxamine and mebeverine, 200 �M for hydro-

ortisone and prednisone, 400 �M for MOB, POB, acetophenonend valerophenone and 800 �M for diphenyl sulfide. Additionally,test mixture of these compounds was prepared at the same con-

entrations in water–BGE (1:1, v/v). The composition of the sheathiquid was methanol–water–formic acid (75:25:0.1, v/v/v) forSI-MS and methanol–water–toluene–formic acid (75:25:5:0.05,/v/v/v) for APPI-MS.

.2. CE system

CE was performed using a PrinCE CE system (Prince Technolo-ies, Emmen, The Netherlands) with a fused silica capillary of0 �m i.d. and a length of 90 cm. The capillaries were flushed withM sodium hydroxide (10 min) and water (10 min) prior to use.

njection of sample was carried out at a pressure of 35 mbar for

dnisone, (E) hydrocortisone, (F) diphenyl sulfide, (G) methyl para-hydroxybenzoate,

P. Hommerson et al. / J. Chromatogr. A 1204 (2008) 197–203 199

Table 1Main parameter settings for ESI and APPI

Conditions Electrospray ionization (ESI) Atmospheric pressure ionization (APPI)

Spray voltage 5000 V 1300 VCapillary exit voltage 114 V 110 VSkimmer voltage 40 V 35 VSheath liquid Methanol–water–formic acid (75:25:0.1, v/v/v) Methanol–water–toluene–formic acid (75:25:5:0.05, v/v/v/v)Sheath liquid flow rate 5 �l/min 15 �l/minDDNV

6wManwatkttH

P

wm

fsfevsae

2

Smh1etnwtirtwtamTrtcas

aswc

3

3

lEtTpaianarfs

3

a(edepaa[libAnaSattem

ry gas flow rate 5 l/minry gas temperature 200 ◦Cebulizing gas 15 psiaporizer temperature –

s. During injection, the nebulizing gas flow of the CE–MS sprayeras switched off. A separation voltage of 30 kV was applied for allEKC analyses. To minimize negative effects (e.g., peak broadening

nd loss of resolution) due to the hydrodynamic flow caused by theebulizing gas, a pressure of 40 or 70 mbar below ambient pressureas applied at the inlet vial during CE–ESI-MS and CE–APPI-MS

nalysis, respectively. These values were determined from the timebetween injection and detection of a test compound applying anown infusion pressure (Pinf) of 0–100 mbar. The pressure reduc-ion (Pred) required to cancel the hydrodynamic flow induced byhe nebulizer was then calculated by substitution into a rearrangedagen–Poiseuille equation:

red = Pinf − 32 × 107 �L2

d2t

ith pressure P in mbar, dynamic visocisity � in mPa s, length L in, diameter d in �m and time t in s.Prior to each analysis, the capillary was flushed with fresh buffer

or 2 min at 1500 mbar. During infusion experiments, the analyteolution under study was continuously introduced into the inter-ace via the CE capillary. For the ESI-MS and APPI-MS infusionxperiments a pressure of 100 and 70 mbar was applied to the inletial, respectively. Under these conditions, the resulting flows wereimilar. No electric field was applied during infusion in order tovoid differences in flow rate and analyte flux due to a generatedlectro-osmotic flow (EOF) and electrophoretic mobility.

.3. MS system

The CE system was coupled to an Agilent Technologies 1100eries LC/MSD SL ion-trap mass spectrometer (Waldbronn, Ger-any) equipped with an Agilent ESI or APPI source. The APPI source

oused a krypton discharge lamp emitting photons of 10.0 and0.6 eV perpendicularly to the nebulized and vaporized capillaryffluent. The coupling of ESI-MS and APPI-MS with CE was achievedhrough a coaxial sheath-flow CE-MS sprayer from Agilent Tech-ologies. For APPI-MS, the sprayer was mounted on a plastic spacerith a length of 36 mm, which was subsequently positioned on

he APPI source. Since the spacer is made of electrically insulat-ng material, an electric wire connected to the sprayer was used toestore the ground potential of the sprayer so as to ensure a func-ional electrical circuit for CE. Using either ion source, the CE systemas positioned such that the capillary inlet was at equal height with

he tip of the sprayer needle. Consequently, siphoning effects werevoided during injection when the nebulizer is switched off. Theajor operating parameters for each interface are listed in Table 1.

he instrument was operated in positive ion mode and the scan

ange was 100–440 m/z, unless otherwise indicated. MS parame-er settings, such as lenses and trap drive, were optimized for eachompound individually by infusion of analyte solutions in 50 mMmmonium acetate. The value for the ion-charge control (ICC) wasubsequently set at the highest value that allowed proper mass

pbtia

3 l/min200 ◦C25 psi300 ◦C

nalysis even in the presence of 50 mM SDS. In order to ensure con-tant interface performances, the ion sources were cleaned daily byiping the ionization chamber surface, spray shield and capillary

ap using a mixture of water–2-propanol (50:50, v/v).

. Results and discussion

.1. Infusion experiments

Infusion experiments were performed to study the effect of ana-yte nature and BGE composition on the MS signal intensity usingSI-MS and APPI-MS detection. The analyte under study was con-inuously introduced through the CE capillary into the interface.he test compounds were selected to cover a wide range of com-ounds potentially separable by MEKC and included a quaternarymmonium compound (methyl atropine), basic amines (fluvoxam-ne and mebeverine), steroids (prednisone and hydrocortisone),lkyl esters of p-hydroxybenzoic acid (MOB and POB), alkylphe-ones (acetophenone and valerophenone) and diphenyl sulfide. Asstarting point, interface conditions, such as composition and flow

ate of the sheath liquid, and nebulizing gas pressure, that wereound suitable for MEKC-ESI-MS and MEKC-APPI-MS in previoustudies, were used [17,21].

.1.1. ESI-MSWith ESI-MS using a volatile BGE of 50 mM ammonium

cetate, the test compounds were detected as protonated molecule[M + H]+), except for methyl atropine which was detected asven-electron ion M+, and diphenyl sulfide which could not beetected by ESI-MS. When a BGE of sodium phosphate (pH 7.5) wasmployed, the basic compounds and quaternary ammonium com-ound were detected as [M + H]+ and M+, respectively, whereas thenalytes that do not form ions in aqueous solution, i.e. the steroids,lkylketones, MOB and POB, were detected as sodium adductsM + Na]+ upon ESI. This adduct formation can be attributed to theack of a basic group and the presence of a carbonyl or ester groupn these compounds. The ESI background spectrum was dominatedy several sodium phosphate clusters at m/z 143, 165, 263, and 383.ddition of SDS to the sodium phosphate BGE did not change theature of the formed ions; the steroids, the alkylphenones, MOBnd POB were still detected as sodium adduct. However, employingDS-containing BGEs, several analytes were occasionally detectedt increased m/z values (+0.2 to 1.0). These mass shifts are indica-ive for space-charge effects, which arise when the trap is filled withoo many ions [24]. As a result of this over-filling, the quadrupolelectrical field becomes distorted which, in addition to mass shifts,ay lead to significantly reduced signal intensities or even com-

lete failure to detect analyte ions. The space-charge effects maye related to the formation of SDS clusters which are observed inhe background spectrum when SDS-containing BGEs are infusednto the electrospray ion source (Fig. 2). Space-charge effects maylso lead to reduced signal intensities or even failure to detect

200 P. Hommerson et al. / J. Chromatogr. A 1204 (2008) 197–203

F1

atit

t(ccawtvinalrc

3

aDwfmcw

Fi

et[wpd(

aaptuwdepeote

Dbpbackground spectrum obtained during infusion of 50 mM of SDS in

Fa(

ig. 2. ESI mass spectrum obtained during continuous infusion of 50 mM SDS in0 mM sodium phosphate (pH 7.5).

nalyte ions. Therefore, in order to allow proper comparisons, ICCarget values were optimized for each compound individually dur-ng infusion experiments (see Section 2) to avoid over-filling of therap.

Fig. 3 shows the effect of the different BGEs on the rela-ive signal intensity obtained with ESI-MS for methyl atropinequaternary amine), mebeverine (basic compound) and hydro-ortisone and valerophenone (neutral compounds). For all testompounds, the BGE of 10 mM sodium phosphate (pH 7.5) causedsignal reduction of 50–70% with respect to the signal obtainedith the ammonium acetate BGE. When 20 mM SDS was added to

he BGE, the signals were 5–10% of the signal obtained with theolatile BGE, which is in agreement with previous studies [16]. Theonization suppression by SDS was slightly more pronounced foreutral compounds than for the quaternary ammonium compoundnd the basic amines. Increasing the SDS-concentration to 50 mMed to a further decrease in signal intensity, although the relativeeduction was modest compared to that observed when the SDSoncentration was increased from 0 to 20 mM.

.1.2. DA-APPI-MSIonization efficiencies in CE-APPI-MS can be enhanced by the

ddition of a so-called dopant to the sheath liquid [21,23,25].opants are compounds with relatively low ionization energieshich are readily ionized by the vacuum UV (VUV) light (123.9 nm)

rom the krypton lamp. Upon ionization, a dopant radical cationay bring about analyte ion formation by proton-transfer or

harge-exchange, either directly, or through intermediate reactionsith solvent molecules [26]. Acetone was found to be primarily

1sBr

ig. 3. Relative ESI-MS responses of methyl atropine, mebeverine, hydrocortisone andmmonium acetate (pH 6.8), (B) 10 mM sodium phosphate (pH 7.5), (C) 20 mM SDS in 10pH 7.5). Analyte concentrations: methyl atropine and mebeverine, 5 �M; hydrocortisone

ig. 4. DA-APPI mass spectrum obtained during continuous infusion of 50 mM SDSn 10 mM sodium phosphate (pH 7.5). For experimental conditions, see Table 1.

ffective for the ionization of relatively polar compounds, whereasoluene was found suitable for both polar and nonpolar compounds25]. For a broad applicability of MEKC–MS, a versatile dopantould be most useful and, therefore, toluene was selected in theresent study for dopant-assisted (DA) APPI. Furthermore, the con-uctivity of the sheath liquid was enhanced by the addition of 0.05%v/v) of formic acid.

When the test compounds were infused in the ammoniumcetate BGE, the apolar diphenyl sulfide was detected by DA-APPIs radical cation (M+•), whereas the basic and neutral com-ounds were all detected as protonated molecules ([M + H]+) andhe quaternary ammonium compound methyl atropine remainedndetected. The latter observation is in line with previous studieshich indicated that quaternary ammonium compounds can beetected by APPI-MS only when specific interface conditions aremployed [23,27]. The nature of the ions formed for the test com-ounds with DA-APPI did not change when nonvolatile BGEs weremployed, even when they contained SDS. Mass shifts were notbserved during the DA-APPI-MS infusion experiments, indicatinghat the ion trap performance is not compromised by space-chargeffects.

With all tested BGEs, the background spectra obtained duringA-APPI-MS revealed one major signal at m/z 92 which is causedy the molecular ion of toluene. In DA-APPI-MS no sodium phos-hate or SDS clusters were observed. Fig. 4 shows the DA-APPI-MS

0 mM sodium phosphate (pH 7.5). The spectrum shows a minorignal at m/z 449 which was only observed with SDS-containingGEs, and was not detected when toluene was either left out, oreplaced by acetone. The SDS clusters formed in ESI (m/z 311, 559

valerophenone as studied by continuous infusion using various BGEs. (A) 50 mMmM sodium phosphate (pH 7.5) and (D) 50 mM SDS in 10 mM sodium phosphateand valerophenone, 100 �M. For other experimental conditions, see Section 2.

matogr. A 1204 (2008) 197–203 201

atgltaitrspsaw

DTsoiopaSfodtlbt(

3

dtwlMi

fiwa

Table 2Concentration-normalized S/Ns obtained during MEKC–MSa

S/N (×10−2 �M−1)

ESI DA-APPI

20 mM SDS 50 mM SDS 20 mM SDS 50 mM SDS

Mebeverine 113 34 47 22Fluvoxamine 98 30 70 45Methyl atropine 67 23 <3.0 <3.0Hydrocortisone 14b 4.2b 67 36Prednisone 6.8b 3.1b 54 37MOB 12b 2.7b 4.2 2.7POB 42b 8.5b 30 18Acetophenone <0.8 <0.8 4.2 2.3Valerophenone 4.4b <0.8 64 43Diphenyl sulfide <0.4 <0.4 23 12

t

adtoS6bboS5

aot4omd(iGt

Fa7

P. Hommerson et al. / J. Chro

nd 887) were not observed in DA-APPI. This fits the conceptionhat SDS-cluster formation occurs in the liquid rather than in theas-phase [28]. Recent APPI studies show that photon-independentiquid-phase ionization processes may occur during vaporization ofhe sample [27,29]. With our CE-APPI-MS set-up, we could observenalyte ion formation by this mechanism only when the MS cap-llary voltage was set in the range 500–800 V, whereas analyteshat are ionized through VUV excitation, typically showed optimumesponses at voltages above 1000 V [27]. After switching the lightource off and lowering the MS capillary voltage to 600 V in ourresent study, the spectra of SDS-containing BGEs indeed showedeveral SDS clusters. This indicates that in DA-APPI-MS, SDS clustersre actually formed, but not transmitted into the mass spectrometerhen using the optimum MS settings.

Fig. 5 shows the effect of the BGE composition on the relativeA-APPI-MS signal intensities for a number of test compounds.he DA-APPI-MS responses were not affected by the presence ofodium phosphate or SDS. This observation holds for both typesf gas-phase ions that can be formed by DA-APPI-MS, i.e., rad-cal cations and protonated molecules. Apparently, the presencef sodium phosphate or SDS does not significantly affect the gas-hase reactions involved in DA-APPI. The deleterious effect of SDSs observed in ESI-MS can be attributed to the adverse influence ofDS on spray droplet formation and analyte ion evaporation as alsoollows from fundamental ESI studies [30]. With regard to the effectf SDS on the ionization process, it can be concluded that DA-APPIetection is more advantageous than ESI-MS. The detection limitshat may be achieved in MEKC–MS, however, depend on both abso-ute signal intensities and noise levels, and are also affected by bandroadening processes. Therefore, in order to make a useful quanti-ative evaluation of ESI and DA-APPI, analyte signal-to-noise ratiosS/Ns) were compared under MEKC–MS conditions (Section 3.2).

.2. MEKC-ESI-MS vs. MEKC-APPI-MS

For a quantitative comparison of ESI-MS and DA-APPI-MS asetection methods in MEKC, optimized MS parameters (e.g. lenses,rap drive) and ICC target values were used for each test compound,hile the interface parameters were kept at predefined values as

isted in Table 1. Each compound was analyzed individually byEKC-ESI-MS and MEKC-APPI-MS using BGEs of 20 and 50 mM SDS

n 10 mM sodium phosphate (pH 7.5).With ESI-MS detection, no signal was observed for diphenyl sul-

de and acetophenone, whereas the less polar test compoundsere invariably detected as their sodium adducts. DA-APPI-MS

llowed the detection of the apolar diphenyl sulfide but methyl

tylsw

ig. 5. Relative DA-APPI-MS responses of mebeverine, hydrocortisone, valerophenone andmmonium acetate (pH 6.8), (B) 10 mM sodium phosphate (pH 7.5), (C) 20 mM SDS in 10 m.5). Analyte concentrations: mebeverine, 100 �M; hydrocortisone and valerophenone, 50

a Concentration-normalized S/Ns are calculated as S/N divided by analyte concen-ration in �M. Results are expressed as mean value of three replicate measurements.

b Detected as sodium adduct.

tropine could not be detected. The plate numbers were analyte-ependent, but ESI-MS generally yielded slightly higher efficiencieshan APPI-MS. Highest plate numbers (up to 140,000) werebtained with MEKC-ESI-MS using the BGE containing 20 mMDS, whereas MEKC-DA-APPI-MS yielded typical plate numbers of0,000–90,000. This slight loss in separation efficiency has alsoeen observed in comparative CZE-MS studies and can probablye attributed to the higher nebulizing gas pressure required forptimal DA-APPI-MS performance [23]. Using the BGE with 20 mMDS, the test compounds migrated within 6–14 min, whereas with0 mM SDS the analyte migration times were 8–20 min.

Table 2 lists the concentration-normalized signal-to-noise ratioss determined from the extracted ion chromatograms (XICs)btained with the SDS-containing BGEs for ESI and DA-APPI. Rela-ive standard deviations (RSDs) of the S/Ns (n = 3) were between.2 and 29%. In MEKC-ESI-MS, best signal-to-noise-ratios werebserved for the amine compounds mebeverine, fluvoxamine andethyl atropine. When the BGE with 20 mM SDS was used, the

etection limits for mebeverine and fluvoxamine, were 2.6 �M1.1 �g/ml) and 3.1 �M (1.0 �g/ml), respectively. These values aren line with those found in an earlier MEKC-ESI-MS study [16].enerally, the S/Ns obtained with ESI-MS detection for the neu-

ral compounds were considerably lower than those obtained for

he amine compounds. Compared to other neutral compounds, POBielded a high S/N, and this appeared to be caused by a relativelyow noise level in the m/z 203 trace, and not by a higher signal inten-ity. When the SDS concentration was raised to 50 mM, S/N levelsere decreased by a factor 3–5 with ESI-MS detection. This loss

diphenyl sulfide as studied by continuous infusion using various BGEs. (A) 50 mMM sodium phosphate (pH 7.5) and (D) 50 mM SDS in 10 mM sodium phosphate (pH0 �M; diphenyl sulfide, 800 �M. For other experimental conditions, see Section 2.

202 P. Hommerson et al. / J. Chromatogr. A 1204 (2008) 197–203

Fig. 6. Extracted-ion traces obtained by (A) MEKC-ESI-MS and (B) MEKC-APPI-MS of a test mixture of methyl atropine (100 �M; m/z 304), acetophenone (400 �M; m/z 121),M prednm GE, 50s

isuahBi

fpDl5fi

AEtcwMeai

ef5

cadsvoTNapFwiep

OB (400 �M; m/z 153), POB (160 �M; m/z 181), valerophenone (400 �M; m/z 163),/z 430), fluvoxamine (100 �M; m/z 319), and diphenyl sulfide (800 �M; m/z 186). B

ee Section 2.

n sensitivity could mainly be attributed to a decrease in absoluteignal (i.e. peak height), as the XIC noise levels virtually remainednchanged. Infusion experiments predicted a loss of signal of aboutfactor of 2 (see Section 3.1.1). The additional decrease in peak

eight is caused by the longer migration times obtained with theGE containing 50 mM of SDS. In CE-MS peak heights decrease with

ncreasing migration time [23].With DA-APPI-MS detection, most favorable S/Ns were obtained

or the basic amines fluvoxamine and mebeverine, the steroids,rednisone and hydrocortisone, and POB and valerophenone. TheA-APPI-MS signals for MOB and acetophenone were relatively low,

eading to poor S/Ns. Increasing the SDS concentration from 20 to0 mM overall led to a decrease of S/N of a factor 2, which can beully attributed to the longer migration times and not to enhancedonization suppression.

A comparison of S/Ns obtained by MEKC-ESI-MS and MEKC-DA-PPI-MS (Table 2) indicates that despite ionization suppression,SI-MS is still slightly more favorable than DA-APPI-MS for the polarest compounds when 20 mM of SDS is used. However, with an SDSoncentration of 50 mM, the S/Ns for the polar analytes obtained

ith MEKC-DA-APPI-MS become similar to those obtained withEKC-ESI-MS. For the neutral test compounds, DA-APPI-MS gen-

rally yields better S/Ns than ESI-MS in either BGE. MOB and POBre an exception to this trend as their S/Ns were somewhat highern ESI-MS than in DA-APPI-MS when the BGE with 20 mM SDS was

otvte

isone (200 �M; m/z 359) hydrocortisone (200 �M; m/z 363), mebeverine (100 �M;mM SDS in 10 mM sodium phosphate (pH 7.5). For other experimental conditions,

mployed. However, DA-APPI-MS showed better detection limitsor MOB and POB when the SDS concentration was increased to0 mM.

To study the feasibility of analyzing compounds of differentharacter in a single run, a mixture of the test compounds wasnalyzed by MEKC-ESI-MS and MEKC-APPI-MS. The interface con-itions as listed in Table 1 were used, whereas MS parameterettings were a compromise of the optimized values for the indi-idual test compounds. Extracted-ion electropherograms (XIEs)btained with the BGE containing 50 mM of SDS are shown in Fig. 6.he test compounds migrated within a time range of 8–20 min.ot all compounds could be baseline separated, but no furtherttempts to optimize the separation were made. Although the MSarameter settings may be non-optimal for individual compounds,ig. 6 clearly demonstrates that MEKC-ESI-MS is a feasible approachhen polar compounds are analyzed, but shows a poor applicabil-

ty for less polar analytes. MEKC-APPI-MS appears more versatilenabling simultaneous analysis of a range of polar and apolar com-ounds.

Another factor that has to be considered for the evaluation

f ESI-MS and DA-APPI-MS as detection methods for MEKC ishe contamination of the ion source and MS ion-optics by non-olatile BGE constituents. Despite the gradual contamination ofhe ion sources and spray shield observed with MEKC-APPI-MS andspecially MEKC-ESI-MS, it was found that once-a-day cleaning of

mato

taapflosaigdaaotbaitaSttp

4

AoprtwatboeaA4etisS

camu2n

A

Pal

R

[[[

[

[[[[

[

[[[[

[

[

[[[27] P. Hommerson, A.M. Khan, T. Bristow, W. Niessen, G.J. de Jong, G.W. Somsen,

P. Hommerson et al. / J. Chro

he ion source and spray shield appeared sufficient to preservenalyte signal intensities. Although the use of nonvolatile saltsnd surfactants should preferably be avoided using atmosphericressure ionization sources, MEKC–MS benefits from very low CEow rates (∼100 nl/min). Additionally, in the present set-up, therthogonal position of the CE sprayer with respect to the masspectrometer inlet, sampling of neutral BGE constituents is largelyvoided thereby reducing contamination of the spray shield andon optics. Nevertheless, during MEKC-ESI-MS, the spray shieldradually became contaminated by sodium phosphate and SDSepositions. This was also observed in MEKC-DA-APPI-MS, albeit tomuch lower extent. From the degree of spray shield contaminationnd the number of SDS-related background signals (Figs. 2 and 4)ne may conclude that ion-optics contamination is more substan-ial in MEKC-ESI-MS than in MEKC-APPI-MS. This conclusion coulde rationalized by taking the specific interface characteristics intoccount. In ESI, the sprayer tip is positioned relatively close to thenlet of the mass spectrometer and a relatively substantial part ofhe spray is sampled. In DA-APPI, the CE sprayer is located furtherway from the inlet and the sample is first vaporized. AlthoughDS clusters are in fact formed in DA-APPI during vaporization ofhe sample, these clusters are not transmitted towards the inlet athe conditions used in this study. Furthermore, the photoionizationrocess itself hardly produces SDS-related ions.

. Conclusions

This study has compared the performance of ESI-MS and DA-PPI-MS as detection techniques for MEKC. Whereas the presencef 20–50 mM of SDS in the BGE results in major ionization sup-ression of polar and medium polar compounds in ESI, DA-APPIesponses are not affected by BGE constituents. In DA-APPI-MS, alsohe type of ions formed was not influenced by the BGE composition,hereas in ESI-MS, neutral compounds were detected as sodium

dducts in the presence of SDS or sodium phosphate. This implieshat detection of neutral compounds by ESI-MS is strongly affectedy the specific composition of the BGE. As a consequence, a changef BGE may yield analyte ions at different m/z values and with differ-nt signal intensities. With ESI-MS detection, several SDS clustersre formed and detected in the background spectrum, whereas DA-PPI-MS spectra shows only one SDS-related background ion at m/z49. The formation of SDS clusters may also result in space-charge

ffects when ion-trap MS is used. In this case, a reduction of the ICCarget value is indicated, although this results in a loss of sensitiv-ty. In this respect, a (triple) quadrupole instrument may be moreuited for the coupling of MEKC with ESI-MS. When using 20 mM ofDS in the BGE, MEKC-ESI-MS yields slightly higher S/Ns for polar

[[[

gr. A 1204 (2008) 197–203 203

ompounds than MEKC-DA-APPI-MS. With 50 mM of SDS, or whennalysing less polar compounds, S/Ns obtained with DA-APPI areore favourable. Overall, it can be concluded that APPI is partic-

larly useful for MEKC–MS when SDS concentrations higher than0 mM are used and a wide range of compounds (i.e. polar/apolar,eutral/charged) has to be analyzed.

cknowledgements

This study was financially supported by AstraZeneca, GlobalR&D, External Science Group (Macclesfield, UK). We thank Dr. Ger-rd P. Rozing of Agilent Technologies (Waldbronn, Germany) for theoan of the APPI source.

eferences

[1] S. Terabe, Anal. Chem. 76 (2004) 240A.[2] I.S. Lurie, J. Chromatogr. A 780 (1997) 265.[3] T.J. Pappas, M. Gayton-Ely, L.A. Holland, Electrophoresis 26 (2005) 719.[4] K.L. Rundlett, D.W. Armstrong, Anal. Chem. 68 (1996) 3493.[5] W. Lu, G.K. Poon, P.L. Carmichael, R.B. Cole, Anal. Chem. 68 (1996) 668.[6] S.E.G. Dekkers, U.R. Tjaden, J. van der Greef, J. Chromatogr. A 712 (1995) 201.[7] W.M. Nelson, Q. Tang, A. Kamel Harrata, C.S. Lee, J. Chromatogr. A 749 (1996)

219.[8] P.G. Muijselaar, K. Otsuka, S. Terabe, J. Chromatogr. A 802 (1998) 3.[9] L. Yang, A. Kamel Harrata, C.S. Lee, Anal. Chem. 69 (1997) 1820.10] Y. Ishihama, H. Katayama, N. Asakawa, Anal. Biochem. 287 (2000) 45.11] P. Petersson, M. Jornten-Karlsson, M. Stalebro, Electrophoresis 24 (2003) 999.12] H. Ozaki, N. Itou, S. Terabe, Y. Takada, M. Sakairi, H. Koizumi, J. Chromatogr. A

716 (1995) 69.13] S.A.A. Rizvi, J. Zheng, R.P. Apkarian, S.N. Dublin, S.A. Shamsi, Anal. Chem. 79

(2007) 879.14] J. Hou, S.A.A. Rizvi, J. Zheng, S.A. Shamsi, Electrophoresis 27 (2006) 1263.15] J. Hou, J. Zheng, S.A.A. Rizvi, S.A. Shamsi, Electrophoresis 28 (2007) 1352.16] G.W. Somsen, R. Mol, G.J. de Jong, J. Chromatogr. A 1000 (2003) 953.17] R. Mol, E. Kragt, I. Jimidar, G.J. de Jong, G.W. Somsen, J. Chromatogr. B 843 (2006)

283.18] H.L. Cheng, M.-C. Tseng, P.-L. Tsai, G.R. Her, Rapid Commun. Mass Spectrom. 15

(2001) 1473.19] M.-C. Tseng, Y.-R. Chen, G.R. Her, Anal. Chem. 76 (2004) 6306.20] Y. Takada, M. Sakairi, H. Koizumi, Rapid Commun. Mass Spectrom. 9 (1995) 488.21] R. Mol, G.J. de Jong, G.W. Somsen, Anal. Chem. 77 (2005) 5277.22] J. Schappler, D. Guillarme, J. Prat, J.-L. Veuthey, S. Rudaz, Electrophoresis 28

(2007) 3078.23] P. Hommerson, A.M. Khan, G.J. de Jong, G.W. Somsen, Electrophoresis 28 (2007)

1444.24] K.A. Cox, C.D. Cleven, R.G. Cooks, Int. J. Mass Spectrom. Ion Process. 144 (1995)

47.25] R. Mol, G.J. de Jong, G.W. Somsen, Electrophoresis 26 (2005) 146.26] A. Raffaelli, A. Saba, Mass Spectrom. Rev. 22 (2003) 318.

Anal. Chem. 79 (2007) 5351.28] N.B. Cech, C.G. Enke, Mass Spectrom. Rev. 20 (2001) 362.29] A. Giuliani, D. Debois, O. Laprevote, Eur. J. Mass Spectrom. 12 (2006) 189.30] R. King, R. Bonfiglio, C. Fernadez-Metzler, C. Miller-Stein, T. Olha, J. Am. Soc.

Mass Spectrom. 11 (2000) 942.