Anion-exchange chromatography on short reversed-phase columns modified with amphoteric (N-dodecyl-N,N-dimethylammonio)alcanoates

Page 1

A

sdiibra©

K

1

emprrtso[caoc

0d

Available online at www.sciencedirect.com

Journal of Chromatography A, 1178 (2008) 60–70

Anion-exchange chromatography on short reversed-phase columnsmodified with amphoteric (N-dodecyl-N,N-dimethylammonio)alcanoates

Ekaterina P. Nesterenko a, Pavel N. Nesterenko b, Brett Paull a,∗a National Centre for Sensor Research, School of Chemical Sciences, Dublin City University, Glasnevin, Dublin 9, Ireland

b Australian Centre for Research on Separation Science (ACROSS), School of Chemistry,University of Tasmania, Private Bag 75, Hobart, TAS 7001, Australia

Received 10 August 2007; accepted 14 November 2007Available online 19 November 2007

bstract

Short reversed-phase columns (50 mm × 4.6 mm Gemini C18) were dynamically coated with carboxybetaines of the generaltructure, C12H25N+(CH3)2(CH2)nCOOH, namely (N-dodecyl-N,N-dimethylammonio)undecanoate, DDMAU (n = 10) and (N-dodecyl-N,N-imethylammonio)butyrate, DDMAB (n = 3), and investigated for the separation of inorganic anions in ion chromatography. The role of theonic strength of coating surfactant solutions on their adsorption and resultant column capacity was studied. The retention of inorganic anions wasnvestigated with different eluents at various concentrations and pH. Interestingly, no retention for anions was found with pure water as the eluent,

ut the addition of small amounts of electrolytes, up to 0.1 mM, caused a sharp increase in the retention of analytes. The effect of increasing anionetention with an increase in eluent cation charge was also observed. Based on this effect a new cation charge gradient concept was proposed andpplied to the separation of a standard mixture of anions. 2007 Elsevier B.V. All rights reserved.

echa

Hsriasfcire

a

eywords: Anion chromatography; Cation charge gradient elution; Retention m

. Introduction

Zwitterionic ion chromatography (ZIC), earlier known aslectrostatic ion chromatography (EIC), is a mode of ion chro-atography, which uses zwitterionic or amphoteric stationary

hases for high-performance anion- or cation-exchange sepa-ations. In the recent past this has mainly been based uponeversed-phase substrates dynamically coated with sulphobe-aines, which provide the zwitterionic stationary phase exchangeites, in place of traditional individual strong or weak cationr anion sites utilised in simple ion-exchange chromatography1,2]. Zwitterionic phases contain both positive and negativeharges, in various configurations, such that analyte ions areble to interact with the stationary phase through a combination

f attractive and repulsive forces as they migrate through theolumn bed.

∗ Corresponding author. Tel.: +353 1 7005060; fax: +353 1 7005503.E-mail address: [email protected] (B. Paull).

ezttise

021-9673/$ – see front matter © 2007 Elsevier B.V. All rights reserved.oi:10.1016/j.chroma.2007.11.038

nism; Carboxybetaines

Zwitterionic stationary phases were introduced into practicalPLC by Yu et al. [3] who used amino acid bound silica for the

eparation of nucleotides. The first application of similar zwitte-ion exchangers for separation of inorganic anions was reportedn 1991 by Nesterenko [4]. These stationary phases could be cre-ted by either chemical modification of the surface of a suitableubstrate [5,6] or by adsorption of a zwitterionic/amphoteric sur-actant onto a hydrophobic support material [7,8]. In the case ofoated phases, as mentioned above, the surfactant used for coat-ng has generally been of a sufobetaine type [6,9–13], althoughecently carboxybetaine type surfactants have been shown toxhibit equally unusual and variable selectivity [14–18].

There are a number of basic advantages of zwitterionic andmphoteric stationary phases, which have resulted in the inter-st shown in such materials by ion chromatographers. Firstly,witterionic stationary phases often exhibit very different selec-ivities in comparison with simple ion exchangers, and have

herefore been applied to the analysis of complex matrices,ncluding highly saline samples. Secondly, the combination ofimultaneous attractive and repulsive ion interactions can bexploited for anion and/or cation separations using only pure

Page 2

roma

wlahstosau[

oZiazcoozbtfhmn[lmdsarapc

spTia(craaoo

2

2

s

3WwACf(ddt(fDmiOU

2

pbcwpa((ft(pAnau

2

fbCDnpatdou

E.P. Nesterenko et al. / J. Ch

ater or very diluted solutions as the eluents [10,19–21]. In theatter case, the obvious advantages of simplified methodologynd potential increases in sensitivity of conductivity detectionave been shown to be partially off-set by problems of poortability of the zwitterionic coatings, and the formation of addi-ional peaks on the chromatogram for each possible combinationf ion pair formed by analyte cations and anions within theample [7,10,19]. The latter problem significantly reduced thenalytical value of this approach, and eventually necessitated these of traditional IC eluents in ZIC to overcome these problems7,22].

Several mechanisms for the retention and separation of anionsn sulfobetaine-type surfactant-modified stationary phases inIC have been proposed [7,9,10,23–26]. These mechanisms

nvolve concepts such as the simultaneous electrostatic inter-ction of analytes with oppositely charged sites within thewitterion, the formation of “ion pairs” between oppositelyharged ions in solution [10] and the formation of a zwitteri-nic electrical double layer arising from the accumulation ofppositely charged mobile phase ions around the charges of thewitterion [26]. In addition, specific retention mechanisms haveeen investigated using Poisson–Boltzman theory, concludinghat small well-hydrated ions interact with the zwitterionic sur-ace by means of a partition mechanism, while large poorlyydrated ions interact via an ion-pair mechanism [23]. Theost recent mechanism proposed was based on two simulta-

eous effects, namely ion-exclusion and chaotropic interaction24]. The ion-exclusion effect comes from the repulsion of ana-yte anions by the outer negative charges (adsorbed sulfobetaine

olecules). This repulsion effect could be either increased orecreased depending on the nature of the analyte ion and thehielding of the charges of the zwitterionic molecule by anionsnd cations added to the eluent as a buffer or ionic strengthegulator. Retention is therefore determined by the ability ofn analyte to directly (chaotropically) interact with the innerositive quaternary ammonium group, and to compete with theorresponding eluent anion for that interaction.

However, despite the availability of the above studies, nouch systematic investigation has been carried out for reversed-hase substrates modified with carboxybetaine type surfactants.herefore, the main focus of this work was an investigation

nto the retention mechanism and selectivity for anions onreversed-phase stationary phase, dynamically coated with

N-dodecyl-N,N-dimethylammonio)alcanoates of varying inter-harge chain lengths. The paper evaluates the difference inetention of inorganic anions, ion-exchange selectivity and sep-ration efficiency between two such carboxybetaine surfactants,nd introduces a new cation charge gradient concept, based onbserved decrease of retention of anions with decreasing chargef the eluent cation.

. Experimental

.1. Instrumentation

The experimental work was performed with a ProStar HPLCystem consisting of a 230 pump, 410 autosampler, and a

Ftwe

togr. A 1178 (2008) 60–70 61

30 UV detector (all from Varian Chromatographic Systems,alnut Creek, CA, USA). The wavelength for UV detectionas 210 nm and the injection volume for all work was 10 �l.ll anion-exchange separations were performed on a Gemini18 column (50 mm × 4.6 mm I.D., particle size 5 �m, sur-

ace area 375 m2/g, carbon load 14%, pore diameter 11 nm)Phenomenex, Torrance, CA, USA) coated with either (N-odecyl-N,N-dimethylammonio)undecanoate (DDMAU) or (N-odecyl-N,N-dimethylammonio)butyrate (DDMAB), for whichhe general formula is given as C12H25N+(CH3)2(CH2)nCOOHn = 10 or n = 3). The columns were coated with each sur-actant as described in Section 2.3. For data acquisition, aell Optiplex GX-1 PC was used with Varian Star Chro-atography Workstation V5.52 data acquisition software

nstalled. The pH measurements were performed on Thermorion 420 pH meter (Thermo Scientific, Waltham, MA,SA).

.2. Reagents

All chemicals used were of reagent or analytical gradeurity. The sodium or potassium salts of nitrate, nitrite,enzoate, iodate, iodide, bromate, bromide, thiosulphate, thio-yanate, acetate, dichloroacetate, trichloroacetate, phthalate, asell as sodium and cerium chlorides and chlorates, sodiumerchlorate and sulphate, mono-, di-, trisodium phosphatend phosphoric acid were all obtained from Sigma–AldrichGillingham, UK). Acetonitrile was obtained from LabScanDublin, Ireland). DDMAU and DDMAB were obtainedrom Calbiochem (Nottingham, UK). All reagent solu-ions were prepared using distilled deionised Milli-Q waterMillipore, Bedford, MA, USA) from a Millipore waterurification system with a specific resistance of 18.3 m� cm.ll mobile phases were filtered with a 0.47 �m Nylafloylon membrane filter (LifeSciences, Steinheim, Germany)nd degassed in an ultrasonic bath for 20 min prior tose.

.3. Column coating

The amphoteric surfactants (DDMAU and DDMAB) usedor coating the stationary phase are both long-chain carboxy-etaine type surfactants, with a C12 hydrophobic tail and a10 intercharge arm for DDMAU and C3 intercharge arm forDMAB between the inner quaternary ammonium and termi-al carboxylic acids groups. The relative stability of the coatingrovided by these surfactants on various C18-modified station-ry phases is described elsewhere [16]. The required amount ofhe surfactant was weighed out for a stock solution of 5 mM,issolved in Milli-Q water and then filtered. A 100 ml aliquotf this solution was then passed through the Gemini C18 col-mn at a flow rate of 0.5 ml/min until breakthrough occurred.

or DDMAU this occurred after 8 ml of the solution passed

hrough the column. Following coating, the column was washedith Milli-Q water prior to equilibrium with the desired elu-

nt.

Page 3

6 hromatogr. A 1178 (2008) 60–70

2

0stwsatw(IwCtaaf

q

wbat

3

aciwptriepr

3

utoPtonpGnp

Fa

rca

vraalos

a

q

woa

f

f

TSoa3FTgt

2 E.P. Nesterenko et al. / J. C

.4. Adsorption isotherms

Two series of DDMAU solutions with concentrations from.0 to 10.0 mM (n = 23) were prepared in pure water (DDMAUolution measured pH 6.6) and in a 2 mM phosphoric acid solu-ion (pH 3.0). Ten-milligram samples of Gemini C18 sorbentere placed into test tubes and 2 ml of the appropriate DDMAU

olution was added to each. The solutions were shaken every 8 hnd left for sedimentation of the sorbent. After 72 h the concen-rations of DDMAU in the solutions above the sorbent samplesere determined using RP-HPLC with UV detection at 192 nm

conditions: column—Supelcosil LC-18-DB, 33 mm × 4.6 mm.D., 3 �m particle size; mobile phase = 60% acetonitrile–40%ater, F = 1.0 ml/min (retention time of DDMAU was 2.67 min).oncentrations of DDMAU were calculated from a calibra-

ion graph, built using chromatographic peak areas. From initialnd final solution concentrations of DDMAU, the amount ofdsorbed surfactant q (mM/g) was calculated according to theormula:

(v)i = (c0

i − ci)v

1000ms(1)

here c0i (mM) is the concentration of the surfactant solution

efore adsorption; ci (mM), the concentration of the surfactantfter adsorption; ms (g), the weight of the sorbent sample; v (ml),he volume of the surfactant solution added [27].

. Results and discussion

In the current work, a Gemini C18 stationary phase was useds the reversed-phase substrate, which belongs to the group of so-alled “hybrid” or “twin” technology phases. This type of phases based upon a combined silica and polymer matrix (hybrid),ith a silica core and a grafted polymer surface (twin). The latterhase exhibits mechanical stability and efficiencies equivalento pure silica phases with additional pH stability over a widerange [28]. The hydrophobicity of the Gemini C18 phase andts suitability for coating with carboxybetaines was evaluatedarlier, and it was found that retention on this type of stationaryhase was similar to, or greater than, that observed alternativeeversed-phase substrates [15].

.1. Adsorption isotherms of DDMAU

As long as dynamic modification or physical adsorption issed to obtain the zwitterionic ion exchangers, it is very impor-ant to establish the structure of coating (mono-layer, poly-layer,r adsorbed micelles) formed on the surface of the sorbent.reviously, this was mainly investigated for zwitterionic surfac-

ants adsorbed on polar adsorbents [29–32] and in a less degreen hydrophobic substrates [33–35]. In order to understand theature of the DDMAU coating on the surface of the reversed-

hase substrate, adsorption isotherms of this surfactant on theemini C18 phase at pH 3.0 and 6.6 were obtained. Given theature of the terminal weak carboxylic acid group (for DDMAUKa = 5.04, DDMAB pKa = 4.89, determined from potentiomet-

siao

ig. 1. Adsorption isotherms of DDMAU on a Gemini C18 surface at (a) pH 6.6nd (b) pH 3.0 at 20 ◦C.

ic acid–base titration curves) it was expected that pH of theoating solution would result in different adsorption isothermsnd resultant surface arrangements.

A known mass of adsorbent solid was shaken with a knownolume of DDMAU solution at a given temperature until equilib-ium. The concentration of surfactant in solution was determinednd resultant adsorption isotherms were compared with knowndsorption isotherms to elucidate a particular type of adsorptionayer on the surface. The carboxybetaine adsorption isothermsnto the Gemini C18 surface obtained by this procedure arehown in Fig. 1.

Adsorption isotherms obtained at both pH 6.6 and 3.0 (Fig. 1)ppear to obey the following Langmuir equation:

= q∞Kc

1 + Kc= θ

θm

(2)

here q, adsorption; K, equilibrium constant; c, concentrationf the surfactant; q∞, limit of adsorption; θ, number of occupieddsorption centers; θm, total number of adsorption centers [27].

Both absorption isotherm curves can be described by theollowing general formula, which is equivalent to Eq. (2):

= bx

1 + ax(3)

he parameters within Eq. (3) were calculated (usingigmaplot 2001 for Windows v.7.101 software) for thebtained adsorption isotherm experimental data. At pH 6.6,= 0.240 ± 0.027, b = 0.430 ± 0.026, R = 0.995, n = 14. At pH.0, a = 0.094 ± 0.011, b = 0.202 ± 0.008, R = 0.997, n = 14.ig. 1 shows the fitted curves to the two data sets obtained.he curve shape indicates that under the conditions investi-ated DDMAU molecules form a mono-layered structure upo 8 mM concentration in aqueous solutions, as for example

chematically represented as Fig. 2a. Some observed increasen adsorption at higher concentrations of the surfactant can bettributed to the beginning of a multilayer adsorption (Fig. 2b)r formation of different aggregates at the hydrophobic surface.

Page 4

E.P. Nesterenko et al. / J. Chromatogr. A 1178 (2008) 60–70 63

F ce, ina MAU

obfaoiof4ttb[a(

ts

pac

Tob

S

w

a

ig. 2. Possible structures of adsorbed DDMAU on a C18 reversed-phase surfadsorbed micelles, (d) internal salt interactions and (e) scheme of adsorbed DD

Depending upon surface hydrophobicity and the propertiesf the zwitterionic surfactant itself, the formation of a num-er of different aggregates is possible [34]. For example, theormation of spherical micelles on hydrophilic silicon nitridend mica has been reported, and hemi-cylindrical micellesf dodecyldimethylammoniopropanesulfonate [DDAPS, crit-cal micellar concentration (CMC) 3.0 mM in water] werebserved on hydrophobic (graphite) surfaces. DDAPS alsoorms globular aggregates of defined periodicity in the range.3 ± 0.3 nm on the flat hydrophobic surface of silicone wafersreated with trimethylchlorosilane, at concentrations 4.5 higherhan its known CMC value [35]. Therefore, it is also possi-le that under these conditions both DDMAU (CMC 0.13 mM36]) and DDMAB (CMC 4.3 mM [36]) could form similarggregates on the surface of the Gemini reversed-phase material

Fig. 2c).

Using Langmuir isotherm Eq. (2) it is possible to find outhe limit of adsorption (q∞) of DDMAU on the reversed-phaseubstrate, the adsorption equilibrium constant (K) and the area

KDra

cluding (a) mono-layer at pH < 5, (b) double-layer at neutral pH, (c) possiblemolecules in electrolyte-containing solution.

er molecule at the solid phase-solution interface. For this thedsorption isotherm was plotted in linear coordinates C/q versususing Eq. (4)

c

q= 1

q∞c + 1

q∞K(4)

he slope gave the value of 1/q∞ and its intercept gave the valuef 1/q∞K. With the known limit of adsorption an area occupiedy the adsorbed molecule was calculated using Eq. (5)

0 = 1

q∞Na(5)

here Na is Avogadro’s number.The adsorption equilibrium constants values for DDMAU

dsorption at pH 3.0 and 6.6 were K = 0.383 ± 0.016 and

= 0.100 ± 0.004, respectively. The limit of adsorption ofDMAU on Gemini C18 was found to be higher at pH 6.6

ather than one for pH 3.0, being q∞ = 2.024 ± 0.162 �M/m2

nd q∞ = 1.616 ± 0.065 �M/m2, respectively. The calculated

Page 5

6 hromatogr. A 1178 (2008) 60–70

aewCcc

gob(nataansr

3

cstudsedfrracD

tsoatIwctedbfioctri1

Fm

au

3a

Dsspartccrf

piwalitcsctre

4 E.P. Nesterenko et al. / J. C

reas occupied by the DDMAU molecule at the surface werestimated as 102 ± 5 A2 at pH 3.0 and 82 ± 4 A2 for pH 6.0,hich is in very good agreement with the results obtained byhevalier et al. [37]. The higher area per molecule in acidiconditions can be as explained by the repulsion of uniformlyharged molecules of surfactant adsorbed at the surface.

Obviously, under acidic conditions at pH < 4.0 the carboxylicroups of DDMAU molecules are protonated, so the formationf a mono-layer on the surface of the sorbent is more proba-le, due to the repulsion between adsorbed charged moleculesFig. 2a) [37]. At pH 6.6 DDMAU molecules can form inter-al salts or interact with oppositely charged groups of otherdsorbed surfactant molecules (Fig. 2b and d), with the potentialo form multi-layered structures, resulting in the higher observeddsorption limits. However, the formation of internal salts couldlso result in the decrease of apparent ion-exchange capacity ateutral pH. In either case, using eluents of relatively high ionictrength and/or of low pH, should result in the expected relativelyapid conversion of the surface to a uniform mono-layer.

.2. Effect of coating solution ionic strength on capacity

Increasing the ionic strength of the coating solution mayause either an enhancement in adsorption of zwitterionicurfactants due to a ‘salting out’ effect, based upon reduced elec-rostatic repulsion and closer packing of the surfactant moleculespon the surface, or a possible reduction in relative adsorptionue to a decrease in the monomolecular concentration of theurfactant in solution resulting from lower CMC values. Forxample, Patil and Okada noted the drop in CMC values for 3-(N-odecyl-N,N-dimethylammonio)propane-1-sulfonate (DDAPS)rom 3.0 mM in pure water to 2.0 mM in 0.05 M sodium perchlo-ate or in 0.4 M sodium chloride [38]. However, for the combinedeasons mentioned above, the observed cumulative effect of theddition of sodium chloride to the coating solution at 0.1 Moncentration caused less than 5% reduction in the amount ofDAPS adsorbed on octadecylsilica surface.Here, to study the effect of ionic strength on the adsorp-

ion of DDMAU, the reversed-phase column was coated using aeries of coating solutions containing increasing concentrationsf NaCl. To evaluate the resultant amount of adsorbed DDMAUn acid–base on-column titration procedure was applied withhe pH of the effluent monitored using a flow through pH meter.nitially the column coated with DDMAU was equilibratedith 5 mM sodium phosphate buffer (pH 3.0) to ensure that all

arboxylic groups of adsorbed surfactant molecules were pro-onated. The column outlet pH was recorded to ensure columnquilibration at pH 3.0 was complete. Then, a pH gradient waseveloped by washing the column with 5 mM sodium phosphateuffer solution at pH 6.0. The pH gradient was carried out twice:rst with an unmodified column in order to exclude influencef equipment dead volume and then again with the DDMAUoated column. The resulting capacity was estimated by sub-

raction of the first gradient from the second one. The obtainedesults (Fig. 3) showed significant increases in capacity withncreasing ionic strength of the DDMAU coating solution, up to0 mM NaCl, followed by a more gradual increase between 10

misc

ig. 3. Determined column exchange capacities (determined using H+ titrationethod) resulting from increasing ionic strength of column coating solution.

nd 100 mM, resulting in an effective capacity range from 4.5p to ∼12 �mol.

.3. The effect of eluent concentration on retention ofnions

The anion-exchange properties of the columns coated withDMAU and DDMAB (coated at pH 6.6, no salts added) were

tudied. The sodium salts of chloride, chlorate perchlorate andulphate were chosen as the eluents, due to their UV trans-arency. The retention data for nitrite, nitrate, bromide, iodatend iodide were collected for a series of eluents of concentrationsanging from 0 to 15 mM. The effect of altering the concentra-ion of each of the above eluents upon retention for the DDMAUoated column is shown in Fig. 4 and in Fig. 5 for the DDMABoated column. Interestingly, a very clear point of maximumetention for eluent concentrations of ∼0.1 mM was recordedor all eluent salts and was observed for both coatings.

In previous studies investigating DDAPS coated reversed-hase columns, analyte retention times were seen to eithernitially increase (as compared to the retention obtained in pureater) when a weakly retained anion (e.g., SO4

2−, Cl−) wasdded to the eluent, or decrease when a strongly retained anionike ClO4

− was used, reaching a point where further increasesn eluent strength showed little effect upon retention [25]. Inhis study, the sharp increase in retention of anions with thehange from pure water to the low strength eluents, and the sub-equent decrease in retention with further increases in eluentoncentration can be explained as follows. It is predicted thathe carboxybetaine molecules adsorbed at the surface remainelatively mobile and flexible. This is assumed as it was shownarlier that for a completely coated surface with a homogeneous

ono-layer distribution of the adsorbed molecules, that the spac-

ng between them is approximately 10.2 A (for DDMAU). Thispacing is approximately the same as the length of the inter-harge arm between –N+(CH3)2– group and –COO− group,

Page 6

E.P. Nesterenko et al. / J. Chromatogr. A 1178 (2008) 60–70 65

retent

wfltF

i[miabisthbwczwb

btsbbFtd

3

iwosz

Fig. 4. Effect of concentration of sodium salts within the eluent on the

hich is 9.9 A [23]. This makes the molecules on the surfaceexible enough for the formation of both internal salts and for

he other possible changes in molecular configuration shown inig. 2d.

In the case of DDMAB, the length of the intercharge arms only 5.2 A, while the distance between molecules is 11.4 A37]. Both intermolecular and intramolecular (as long as theolecule intercharge arm is flexible enough to form stable

nternal salts) interactions could take place between positivend negative groups within the surfactants molecules, but theseecome significantly weaker with increasing salt concentrationn the eluent. This means that the dipolar layer is missing or sub-tantially reduced in pure water and appears immediately withhe increase of eluent ions. Similar changes in the dipolar layerave been reported by Chevalier et al. [39–41], who found thatecause of stronger anion binding with the zwitterion than seenith cations, the adsorbed surfactant betaine molecules became

harged, causing intermicellar repulsive effects and swelling ofwitterionic layer. They also showed that this effect increasesith an increase of intercharge arm length and explained thisy electrostatically driven ‘reorganisation’. Based on this, it can

ltba

ion of different anions on a Gemini C18 column coated with DDMAB.

e assumed that the nature of the eluent anion is significant ashe stronger the anion binds to the zwitterionic molecule, thetronger intermicellar repulsion is and as a result, interactionsetween positive and negative parts of the surfactant moleculeecome weaker making formation of internal salts less probable.ollowing this initial surface reorganisation effect, with a fur-

her increase of eluent concentration the expected ion exchangeominates and retention continually decreases.

.4. Effect of eluent cation charge on analyte retention

According to the results of Cook et al. [25], who studied theon-exchange mechanism for octadecylsilica substrates coatedith sulfobetaine surfactants, the retention mechanism consistsf two parts: ion-exclusion and chaotropic interactions. In thistudy, the carboxylic acid groups on the external part of thewitterionic layer possess a negative charge, which repels ana-

yte anions by acting as a Donnan membrane. The magnitude ofhis negative charge depends on the strength of the interactionetween cations from the eluent and this weak acid group, as wells on the strength of interaction between anions and the inter-

Page 7

66 E.P. Nesterenko et al. / J. Chromatogr. A 1178 (2008) 60–70

tentio

ncaSwqltowbr

ammi6o

ocaeawbfiidvlwl

Fig. 5. Effect of concentration of sodium salts within the eluent on re

al quaternary ammonium groups. Strong interaction with eluentations decreases the surface negative charge, while strong inter-ction with eluent anions increases the surface negative charge.o the outer negative charge forms an effective barrier throughhich the analyte anions must penetrate in order to reach theuaternary ammonium functional group (the interaction of ana-yte anions with the quaternary ammonium functional group ofhe carboxybetaines depends also on the chaotropic characterf the anion). Ignoring the obvious significance of eluent pH,hich will be discussed later, this charge barrier can be loweredy partial neutralisation using 2+ or 3+ cations within the eluent,esulting in an observed increase in retention for analyte anions.

To study the above cation charge effect upon the retention ofnions on the carboxybetaine coated columns, the use of sodium,agnesium and cerium chlorides and perchlorates as sources of

ono-, di- and triply charged cations within the eluent were

nvestigated. Ten-millimolar eluents with an approximate pH.0–6.6 were prepared, ensuring that the carboxylic acid groupsf the carboxybetaines were fully dissociated. Again column

ca

a

n of different anions on a Gemini C18 column coated with DDMAU.

utlet pH was monitored to ensure column equilibration wasomplete. In the case of eluents prepared from both chloridend perchlorate salts, the increase in the cation charge within theluent caused an increase in the retention for all of the analytenions investigated, bar iodide, for which maximum retentionas observed with Mg2+ salts. This effect was observed foroth carboxybetaine coated columns (see Fig. 6). The reasonor the atypical behavior of iodide may be the formation of anodide complex with cerium [42], hence causing a reductionn retention. For both DDMAU and DDMAB columns similarependences were observed, the only difference being lower kalues in case of DDMAB coated column. This was due to of theower hydrophobicity of DDMAB as compared with DDMAU,hich resulted in a smaller amount of adsorbed DDMAB and a

ower ion-exchange capacity of the corresponding column. The

orresponding values of log P are 1.46 and 6.89 for DDMABnd DDMAU, respectively [43].

There was a weak interaction between Na+ and the carboxyliccid group of adsorbed surfactant and a strong interaction

Page 8

E.P. Nesterenko et al. / J. Chromatogr. A 1178 (2008) 60–70 67

F n a Gs e of D

bwf(

Fm

ig. 6. Effect of the mobile phase cation charge on the retention of analytes oolutions of (a and c) perchlorates and (b and d) chlorides as the eluents. In cas

etween ClO4− and the internal quaternary ammonium group

hen investigating the 10 mM NaClO4 eluent. This led to theormation of a strong negatively charged Donnan membraneFig. 7b), which exerts a strong repulsive effect on analyte

aaip

ig. 7. Proposed schematic representation of retention mechanism. (a) Establishmeobile phase, (d) use of low pH eluent. This schematic diagram is adapted from [25]

emini C18 column coated with DDMAU (a and b) and DDMAB (c and d) inDMAB all eluents also contained 0.2 mM DDMAB.

nions. When the 10 mM CeCl3 eluent was used, a strong inter-ction between Ce3+ and the carboxylic acid group and a weaknteraction between Cl− and quaternary ammonium group tooklace (Fig. 7c). In this case, the Donnan membrane becomes

nt of Donnan membrane, (b) use of NaClO4 mobile phase, (c) use of CeCl3.

Page 9

68 E.P. Nesterenko et al. / J. Chromatogr. A 1178 (2008) 60–70

F oatedb t UV

wo

sipDgaacnat

t(Tott

cecf>Drt[is

3

ot

TC

I

IBNBNIT

n

ig. 8. Separation of a standard mixture of anions on a Gemini C18 column cromide; 4, nitrate; 5, iodide. Eluent: 10 mM MgCl2, pH 6.4, 1.0 ml/min. Direc

eakly positively charged, which results in stronger retentionf analyte anions.

This mechanism also explains the strong anion retentioneen in acidic eluents. It has been shown previously for sim-lar carboxybetaine type surfactants, the pH of the mobilehase strongly affected the retention of anions [8,16]. As theDMAU and DDMAB molecules contain a weak acid terminalroup, a decrease of eluent pH reduces the number of dissoci-ted carboxylic acid groups and hence significantly increasesnion-exchange capacity. As hydronium ions interact with thearboxylic acid group more strongly than any other cation, theeutralised negative barrier can be overcome by analyte anionsnd they can readily interact with quaternary ammonium func-ional groups (Fig. 7d).

To compare this effect for the two carboxybetaine surfactants,wo separations were obtained using a 10 mM MgCl2 eluentpH 6.4) on DDMAU (Fig. 8a) and DDMAB columns (Fig. 8b).

he DDMAU coated column exhibited better efficiency and res-lution compared with DDMAB column, predominantly dueo its higher overall column capacity. The efficiency, resolu-ion and capacity factor values for DDMAU and DDMAB

eoaa

able 1hromatographic characteristics of (N-dodecyl-N,N-dimethylammonio)alcanoate coa

on DDMAU DDM

k N theor.pl./column

Rs α k

odate 1.03 ± 0.05 196 ± 6 – – 0.5romate 1.22 ± 0.07 548 ± 5 0.33 ± 0.12 1.12 ± 0.05 0.7itrite 3.02 ± 0.15 733 ± 41 3.05 ± 0.22 2.50 ± 0.16 1.4romide 4.00 ± 0.29 762 ± 80 1.33 ± 0.18 1.32 ± 0.04 4.8itrate 6.22 ± 0.14 598 ± 36 1.92 ± 0.10 1.58 ± 0.07 8.1

odide 50.2 ± 2.2 2651 ± 172 11.0 ± 0.40 8.34 ± 0.24 17.4hiocyanate 185.0 ± 6.1 2936 ± 155 9.82 ± 0.32 3.69 ± 0.07 49.8

= 3.a (N-dodecyl-N,N-dimethylammonio)acetate coated Supelcosil C18 column.

with DDMAU (a) and DDMAB (b). Elution order: 1, bromate; 2, nitrite; 3,detection at 210 nm.

oated columns are presented in Table 1 and compared toarlier obtained results by O’Riordain et al. [16] on a (dode-ylamino)acetic acid (DDAA) coated stationary phase. It wasound that for the DDMAU coated column the efficiency wasfour times more efficient (∼12,000–49,000 N/m) than for theDMAB coated column (∼2000–13,000 N/m), with improved

esolution values for most anion pairs. Comparing these datao results obtained for DDAA coated reversed-phase columns44], DDMAU and DDMAB columns showed higher selectiv-ty factors for all analytes, while DDMAU showed very highelectivity for iodide and thiocyanate anions.

.5. Cation charge gradient concept

An obvious way to reduce analysis time is the applicationf gradient elution using one or more eluent variables. Moreraditional methods alter some component concentration of the

luent per unit time. It has been reported previously that the usef a cation concentration gradient for the separation of anions onreversed-phase substrate with attached macrocycles has been

pplied. As the attached macrocycles can complex eluent cations

ted columns

AB DDMAAa [44]

N theor.pl./column

Rs α k α

9 ± 0.03 112 ± 17 – – – –0 ± 0.05 94 ± 11 0.15 ± 0.03 1.18 ± 0.14 – –1 ± 0.21 158 ± 18 1.06 ± 0.44 1.99 ± 0.13 0.35 –2 ± 0.28 1194 ± 86 5.01 ± 0.37 3.46 ± 0.19 0.57 1.626 ± 0.19 277 ± 27 1.53 ± 0.13 1.69 ± 0.13 0.84 1.473 ± 2.0 732 ± 90 3.40 ± 0.35 2.14 ± 0.22 6.20 7.399 ± 4.1 350 ± 19 2.98 ± 0.12 2.86 ± 0.22 – –

Page 10

E.P. Nesterenko et al. / J. Chromatogr. A 1178 (2008) 60–70 69

F n cha5 yanatm DDM

tod[

ttasaaoeowtMtef

TPc

A

ANBNTDIPBT

tbadtp

4

bwbafis

ig. 9. Separation of a standard mixture of anions without (a) and with (b) catio, thiosulphate; 6, dichloroacetate; 7, iodide; 8, phthalate; 9, benzoate; 10, thiocaintained 20 mM NaCl for next 70 min (b). Column: Gemini C18, coated with

hus generating new anion-exchange sites, gradient separationsf anions can be performed due to the change of column capacityuring the separation, achieved by changing the eluent cation44].

Here, a very similar result can be achieved. As the retentionimes decrease with a decrease in cation charge, it was decidedo investigate a “cation charge gradient” for the separation ofnions. For the purpose of concept illustration, magnesium andodium chloride based eluents were selected, as they do notbsorb in the UV spectral range. To prove the concept, two sep-rations (with and without cation charge gradient applied) werebtained, whilst keeping the nature and concentration of theluting anion constant. Without such a gradient applied, usingnly 10 mM MgCl2 as an eluent, separation of ten test anionsas achieved in an excessive 120 min (Fig. 9a). The same mix-

ure was then run with a cation charge gradient from 10 mMgCl2 to 20 mM NaCl (different concentrations of salts were

aken in order to keep the same ionic strength I = 4 × 10−4 andxclude its influence) for 10 min and maintained at 20 mM NaClor next 70 min. When the gradient profile was applied, the run

able 2eaks asymmetries for separations on DDMAU coated column with and withoutation gradient

nion Peak asymmetry;no gradient

Peak asymmetry; cation gradient

cetate 6.25 3.24itrite 3.70 1.67romide 1.67 1.53itrate 2.00 1.80hiosulphate 1.28 1.12ichloroacetate 2.50 1.0

odide 1.29 1.25hthalate 1.30 1.0enzoate 1.50 1.50hiocyanate 1.16 1.57

naeta

R

[[[

rge gradient applied. Elution order: 1, acetate; 2, nitrite; 3, bromide; 4, nitrate;e. Eluent: 10 mM MgCl2 (a) and 10 mM MgCl2 to 20 mM NaCl for 10 min and

AU, flow rate 1.0 ml/min, UV detection at 210 nm.

ime was roughly halved (Fig. 9b). Furthermore, peak shape andaseline resolution of some peaks improved markedly. Visu-lly resolution improved between bromide, nitrate, thiosulphate,ichloroacetate and iodide peaks. Peak asymmetry values forhe separation without the gradient are presented in Table 2. Theeak asymmetry factors were calculated at 10% of peak height.

. Conclusions

The investigation of novel zwitterion ion exchangers preparedy dynamic coating of short reversed-phase Gemini C18 columnsith carboxybetaine surfactants DDMAU and DDMAB haseen carried out. The adsorption of surfactants was investigatednd possible structures of adsorbed layers are proposed. It wasound that coating type (mono- or poly-layer) and column capac-ty depended on pH and ionic strength as these factors have thetrongest affects upon surfactant micelle formation and inter-al salt formation. The effect of mobile phase cation charge onnion retention was also studied. A new cation charge gradientlution concept was proposed and its application demonstratedo reduce separation time by almost 50% for a mixture of testnions.

eferences

[1] P.N. Nesterenko, P.R. Haddad, Anal. Sci. 16 (2000) 565.[2] K. Irgum, W. Jiang, Anal. Chem. 71 (1999) 333.[3] L.W. Yu, T.R. Floyd, R.A. Hartwick, J. Chromatogr. Sci. 24 (1986) 177.[4] P.N. Nesterenko, J. High Resolut. Chromatogr. 14 (1991) 767.[5] P.N. Nesterenko, P.A. Kebets, J. Anal. Chem. 62 (2007) 2.[6] C. Viklund, K. Irgum, Macromolecules 33 (2000) 2539.[7] H.A. Cook, G.W. Dicinoski, P.R. Haddad, J. Chromatogr. A 997 (2003) 13.[8] C. O’Riordain, E. Gillespie, D. Connolly, P.N. Nesterenko, B. Paull, J.

Chromatogr. A 1142 (2007) 185.[9] W. Hu, Langmuir 15 (1999) 7168.10] W. Hu, T. Takeuchi, H. Haraguchi, Anal. Chem. 65 (1993) 2204.11] E. Twohill, B. Paull, J. Chromatogr. A 973 (2002) 103.12] M. Macka, P.R. Haddad, J. Chromatogr. A 884 (2000) 287.

Page 11

7 hrom

[

[

[

[[

[[[

[

[[[[[[

[[

[[[[[[[

[

[[

[

0 E.P. Nesterenko et al. / J. C

13] W. Hu, P.R. Haddad, H. Cook, H. Yamamoto, K. Hasebe, K. Tanaka, J.S.Fritz, J. Chromatogr. A 920 (2001) 95.

14] W. Hu, P.R. Haddad, K. Tanaka, K. Hasebe, Anal. Bioanal. Chem. 375(2003) 259.

15] C. O’Riordain, L. Barron, E. Nesterenko, P.N. Nesterenko, B. Paull, J.Chromatogr. A 1109 (2006) 111.

16] C. O’Riordain, P. Nesterenko, B. Paull, J. Chromatogr. A 1070 (2005) 71.17] E.P. Nesterenko, L. Barron, P.N. Nesterenko, B. Paull, J. Sep. Sci. 29 (2006)

228.18] L. Barron, P.N. Nesterenko, B. Paull, Anal. Chim. Acta 567 (2006) 127.19] P.N. Nesterenko, J. Chromatogr. 605 (1992) 199.20] W. Hu, Sh. Cao, M. Tominaga, A. Miyazaki, Anal. Chim. Acta 322 (1996)

43.21] W. Hu, H. Tao, M. Tominaga, A. Miyazaki, H. Haraguchi, Anal. Chim.

Acta 299 (1994) 249.22] W. Hu, P.R. Haddad, K. Tanaka, K. Hasebe, Analyst 125 (2000) 241.23] T. Okada, J.M. Patil, Langmuir 14 (1998) 6241.

24] K. Iso, T. Okada, Langmuir 16 (2000) 9199.25] H.A. Cook, W. Hu, J.S. Fritz, P.R. Haddad, Anal. Chem. 73 (2001) 3022.26] W. Hu, P.R. Haddad, Trends Anal. Chem. 17 (1998) 73.27] Yu.S. Nikitin, P.S. Petrova, Experimental Methods in Adsorption and

Molecular Chromatography, Moscow University, Moscow, 1991.

[

[[[

atogr. A 1178 (2008) 60–70

28] Gemini Product Brochure, Phenomenex, Torrance, CA, 2005.29] I. Harwigsson, F. Tiberg, Y. Chevalier, J. Colloid Interface Sci. 183 (1996)

380.30] P.F. Brode, Langmuir 4 (1988) 176.31] J. Zajac, C. Chorro, M. Lindheimer, S. Partyka, Langmuir 13 (1997) 1486.32] J. Zajac, J. Colloid Surf. A 167 (2000) 3.33] W. Hu, P.R. Haddad, Chromatographia 52 (2000) 543.34] L.M. Grant, W.A. Ducker, J. Phys. Chem. B 101 (1997) 5337.35] J.L. Wolgemuth, R.K. Workman, S. Manne, Langmuir 16 (2000) 3077.36] C. Brenner, G. Jan, Y. Chevalier, H. Wroblewski, Anal. Biochem. 224

(1995) 515.37] Y. Chevalier, Y. Storet, S. Pourchet, P. Le Perchec, Langmuir 7 (1991)

848.38] J.M. Patil, T. Okada, Anal. Commun. 36 (1999) 9.39] Y. Chevalier, N. Kamenka, M. Chorro, R. Zana, Langmuir 12 (1996)

3225.40] N. Kamenka, Y. Chevalier, R. Zana, Langmuir 11 (1995) 3351.

41] N. Kamenka, Y. Chevalier, M. Chorro, H. Levy, R. Zana, Langmuir 11

(1995) 4234.42] S.W. Mayer, S.D. Schwartz, J. Am. Chem. Soc. 73 (1951) 222.43] KowWin logP calculator, Syracuse Research Corporation, NY, 1999.44] J.D. Lamb, R.G. Smith, Talanta 39 (1992) 923.