1441 h 1441. - Universidad De Antioquiaquimica.udea.edu.co/~carlopez/cromatoion/capillary_ionchroma_2005.pdf · the differences in ion exchange ... sup-pressed capillary IC was

J. Sep. Sci. 2004, 27, 1441–1457 www.jss-journal.de i 2004WILEY-VCH Verlag GmbH&Co. KGaA,Weinheim

Spec

ialF

ocu

sRev

iew

Kuban, Dasgupta 1441

Petr KubanPurnendu K. Dasgupta

Department of Chemistryand Biochemistry,Texas Tech University,Lubbock, TX 79409-1061,USA

Capillary ion chromatography

This review summarizes progress in capillary ion chromatography. Theoreticalaspects and practical limitations of packed and open tubular capillary columns areconsidered. Applications of packed and open tubular capillary IC are described.Emerging technologies such chip-scale IC and the use of monolithic columns are dis-cussed.

KeyWords:Capillary ion chromatography; Capillary separations; Inorganic ions;

Received: April 18, 2004; revised: June 7, 2004; accepted: June 7, 2004

DOI 10.1002/jssc.200401824

1 Introduction

Ion chromatography (IC) is an analytical technique for theseparation of ionic and ionizable compounds. IC exploitsthe differences in ion exchange affinities of different ana-lytes. Lucy [1] has suggested that the documented use ofion exchange goes back to biblical times: when Mosesarrived in Marah after crossing the Red Sea he used atree to “sweeten” brackish water [2]. While any biologicalmaterial undoubtedly contains capillaries, we shall confineourselves here to a less ambitious time frame. The clockin this review begins with the publication of the seminalpaper by Small, Stevens, and Baumann on IC in 1975 [3];this paper has since been designated one of the milestonepapers in Analytical Chemistry [4]. For the first time, theyused a chromatographic system with two columns, a sep-aration column filled with a low capacity ion exchanger, asecond, high capacity ion exchange column for eluentcounterion exchange to reduce the conductivity of thebackground, and an electrical conductivity detector. Thekey new concept was the use of the second column that“neutralized” the eluent ions resulting in “conductivity sup-pression” so that a flow-through conductometric detectorcould monitor the second column effluent for eluting ana-lyte ions with high sensitivity. A detailed discussion of allsignificant developments in IC in the last three decades isbeyond the scope of this review. Some of the more impor-tant milestones include the introduction of membrane sup-pressors [5–7], the electrodialytic generation of elu-ents [8–10], single column or “non-suppressed” IC [11–13], in which only one low capacity column is utilized inconjunction with an eluent of high elution strength but lowconductivity, ion exclusion chromatography [14–16], inwhich weak acids or bases are separated in theiruncharged form, and more recently the use of electricallypolarized ion exchange resin beds that can be used for

generation, suppression, and even purification of elu-ents [17–19]. These and other innovations, notably in thedevelopment of novel ion exchanger phases, have con-tributed to the popularity of IC such that anion analysistoday is unthinkable without IC. The present worldwidemarket for IC equipment and related consumables is esti-mated to be ca. US $ 220million.

At the present time, capillary IC equipment or columns arenot commercially available. Researchers have exploredcapillary scale IC, however, since the early years. Theintroduction of capillary LC in late seventies, generallyattributed to Horvath et al. [20, 21], and related key publi-cations in the area [22–25] had an early impact on carry-ing out IC on a miniature scale. As early as 1983, sup-pressed capillary IC was reported by Rokushika etal. [26]. Due both to theoretical and to practical limitations,including the lack of commercially available instrumenta-tion, capillary IC has not yet emerged beyond researchlaboratories. In this review some theoretical aspects ofcapillary format chromatography are discussed from theperspective of performing IC, and then practical achieve-ments and applications thus far are discussed.

2 Capillary (ion) chromatographyHerein we consider a column to be a capillary column ifthe inner diameter is less than 1 mm. We consider here (i)packed, (ii) open tubular (OT), and (iii) monolithic capillarycolumns.

Packed capillary columns are analogous to conventionallarge bore columns and are typically made by the sameslurry packing methods as their large bore counterparts. Afrit is made at one end of the capillary that retains thepacking material. The capillary columns are typicallypacked under pressure with particles a15 lm in diameter.Drawn packed capillaries [27] (drawing a large bore glasscapillary packed with particles), explored in the early daysof capillary LC have never been used in capillary IC; most

Correspondence: Purnendu K. Dasgupta, Department ofChemistry and Biochemistry, Texas Tech University, Lubbock,TX 79409-1061, USA. Phone: +1 806 742-3064.Fax: +1 806 742-1289. E-mail: [email protected].

1442 Kuban, Dasgupta

ion exchange packings are unlikely to survive such ther-mal conditions.

Open tubular capillary columns are usually fused silicacapillaries of inner diameter in the range of a100 lm. Afew practitioners of capillary LC have ventured to columnsof a10 lm ID: the multiple problems of reliably pumpingvery small flow rates, reproducibly injecting minute quanti-ties of samples, and sensitively detecting the very smallamounts of analyte are formidable. The stationary phaseis attached to the capillary walls by electrostatic interac-tions or by covalent bonding. The wall can also be coatedwith a polymer – the coating thickness can be optimizedto achieve a compromise between acceptably high masstransfer rates and high loading capacity. Extraordinaryefficiencies (over a million theoretical plates) have beendemonstrated in capillary LC; such efficiencies have notbeen reported in IC. While many ionic analytes are smalland thus have relatively high diffusion coefficients, ionexchange kinetics can sometimes be rate-limiting relativeto nonionic adsorption-desorption equilibria. As the hori-zons of capillary IC broaden, it will therefore be interestingto observe how the attainable efficiencies compare in LCvs. IC in the capillary domain. Nevertheless, achievingappropriately small injection and detector volumes whilemaintaining detection sensitivity is likely to remain a chal-lenging task. In this context, the mainstay of IC detection,conductometry, enjoys an advantage. Unlike opticaldetection methods, conductometric detection can bedownscaled without penalties.

Monolithic columns consist of a rigid polymeric structureinside the capillary produced by thermally initiated orphoto-initiated polymerization of monomers with desiredfunctional groups, giving the polymer matrix a variety oftailored properties. Sol-gel processes are typically usedfor preparation of silica-gel monoliths. Relative to packedcolumns, they provide comparable separation efficiency,relatively simple fabrication, and lower flow resistance.

While there are a number of publications on packed capil-lary IC and fewer (but still several) reports on OT IC,monolithic column capillary IC still appears to be in itsinfancy.

3 Theory of chromatography:Consequences for capillary IC

All processes contributing to the final zone width areimportant in capillary IC. These include resistance tomass transfer, eddy diffusion (in packed columns andmonoliths), longitudinal diffusion, and extra-columneffects that include broadening in the suppressor and thedetector. The overall peak variance is the sum of the indi-vidual variances:

r2tot ¼ r2

res þ r2eddy þ r2

long þ r2sup þ r2

det (1)

The final zone width, r2tot, is the measure of column effi-

ciency and for ideal (Gaussian) analyte peaks, the widthof the peak being directly related to the variance, r2.

Some important questions have to be answered: Whatare the advantages and disadvantages of the capillary for-mat vs. the conventional one? Under what conditions willcapillary IC (in any column format) be competitive with itsmacroscale counterpart? Capillary separation systemscan generally produce greater column efficiencies per unitlength but what other limitations are posed by the capillaryformat?

Knox and Gilbert [28] theoretically studied the efficiencyof capillary columns in LC. They derived the optimum par-ticle diameter (dp,opt) for packed capillary columns and theoptimum inner diameter of an open tube capillary column(dc,opt), for a given number of total theoretical plates N,mobile phase viscosity g, analyte diffusion coefficient Dm,and the total pressure drop Dp:

dp,opt = (4000 NgDm/Dp)1/2 (2)

dc,opt = (128NgDm /Dp)1/2 (3)

For typical conditions, such as N = 104, g = 10– 3 N m/s2,Dm = 10–9 m2/s, Dp = 200 bar (26107 N/m2), Knox andGilbert compute dp,opt to be 1.4 lm. Although the use ofvery small particles in the 1–1.5 lm particle size rangehas been in vogue for some time in reverse phase liquidchromatography, this is not the case for IC. The smallestavailable particle size for commercially available columnsis 3 lm [29]. However, comparison of chromatogramsfrom columns from another manufacturer that uses a lar-ger particle size packing clearly indicates that particle sizealone is not the only factor in determining column effi-ciency. Moreover, it is not clear that the same theoreticalconsideration for optimum packing diameter strictlyapplies to IC columns which more often than not consist ofsurface-agglomerated packing. While pumping systemscapable of pumping at very high pressures have becomecommercially available, the maximum pressure toleranceof polymeric particles typically used in IC must also beconsidered.

The optimum inner diameter for open tubular capillaries(dc,opt) is similarly calculated from Eq. (3) to be 0.26 lm.Capillaries with this diameter are not at this time commer-cially available and even if they were, injection (the smal-lest commercially available injector at the time of this writ-ing has a loop volume of 10 nL), detection, and columnplugging problemsmay be nearly insurmountable.

However, open tubular columns can and often are oper-ated at conditions far from theoretical optima and can stillgenerate enough plates for the desired separation. Withinjection and detection volumes of 1 nL and a pressure dif-ference of 100 bar (1500 psi), a performance comparable


Capillary ion chromatography 1443

with conventional packed columns can be theoreticallyachieved atN = 30,000, using a 5 m long column of 10 lmID at a flow rate of 30 nL/min and diffusion coefficient Dm =10–9 m2 s– 1. With a 100-fold larger injector/detector acomparable performance at 10-fold greater plate count(N = 30,000) can be achieved.

Often the column performance is measured in terms ofnumber of theoretical plates per unit time based onGolay’s performance index [30] that measures the effi-ciency with which a given column achieves a certain reso-lution with a given pressure differential at a given time.This may not be satisfactory because efficiency can begenerally increased by increasing the pressure or bydecreasing the eluent viscosity. For general comparisonof various columns types, Knox [31] introduced a termseparation impedance, E :

E 3t0NDpN

1g¼ h2

b ð4Þ

where t0 is the void retention time, Dp is the pressure dif-ference across the column, g is the eluent viscosity, h isthe reduced plate height, and b is the column flow resis-tance factor.

E is dimensionless; the lower the value of E, the better isthe column performance. If two columns are to be com-pared, they can be compared based on the value ofE.

Based on the calculation of E under optimum conditions,Knox and Gilbert concluded that (i) very small detection/injection volumes are essential for packed capillary col-umns to be competitive with large bore packed columns inperformance, (ii) OT capillaries will be particularly attrac-tive at very high values of N and will necessitate low injec-tion/detection volumes and very small diameters: suchconditionsmay be feasible in the future in on-chip chroma-tographic systems.

At room temperature, the low diffusion coefficient of theanalyte in themobile phase results in a lowmass transportrate in OTLC, resulting in poor plate heights. Some havebelieved that hydrodynamic means to improve masstransfer to the wall will be useful, i.e., mass transfer to thewall should be changed from a strictly diffusive regime (asin laminar flow) to systems where there is active convec-tive transport to the wall. Coiling the separation capillary(as in Helical Open Tubular columns) enhances the masstransfer by radial convection due to the secondary flow.Tijssen [32, 33] (who has called such capillaries HOT col-umns) and Tsuda and Novotny [27] have demonstratedthat radial convection indeed improves the plate height inrelatively large bore (typically 200–300 lm ID) tubes.However, noticeable improvement is observed primarilyat high flow rates, and the separation column must thenbe sufficiently long to allow sufficient time for the interac-

tion of the analyte with the stationary phase. Thus far, theexperimental data have been observed to agree well withtheory for unretained analytes but actual chromatographicperformance falls far short for retained analytes. Becausediffusion coefficients in the stationary and mobile phasesin LC/IC are not as different as in GC, there are, compara-tively speaking, less limitations on the permissible thick-ness of the stationary phase. For reasonably permeablecoatings, relatively little loss of efficiency should occur forwall coating thicknesses up to 25% of the tube diameter.Thus, a thick layer of porous or otherwise permeable ionexchanger phase will improve column capacities in OTIC.The astute reader will observe that the boundary betweenthis and highly porous monolithic columns in a capillaryscale will become blurry.

4 Applications of packed capillaries in ICWhile packed capillary chromatography is completelyanalogous to conventional column chromatography, pre-paration of capillary columns requires very small amountsof packing material. As such, stationary phases that areexpensive to prepare, are more easily used in the capillaryformat. Kryptand-based stationary phases that have beenintroduced for variable capacity gradients [34, 35] can, forexample, profit from a capillary column adaptation.Further, the consumption of eluent and sample solutionsis decreased and this becomes invaluable when samplesare of the order of lL or sub-lL and cannot easily be ana-lyzed by conventional scale LC or IC.

4.1 The introduction of capillary IC

The evolution of capillary IC has closely followed that ofcapillary LC. The first publication on suppressed capillaryIC appeared in 1983 [26]. The development of IC withconductivity suppression on the macroscale inspiredRokushika et al. to use capillary IC columns of 0.19 mmdiameter and couple these to a 0.2610 mm Nafionm per-fluorosulfonate hollow fiber suppressor immersed in alarge reservoir of a low-bleed, large counterion acidregenerant, dodecylbenzenesulfonic acid. Using a sur-face agglomerated anion exchanger of 10 lm base parti-cle size, sodium carbonate-bicarbonate as eluent, theynot only demonstrated the separation and detection ofseveral inorganic anions and organic acids at low ppmlevels, they showed the applicability of the method to realsamples ranging from river water to several types of fruitjuice. Figure 1 shows the separation of a standard mix-ture. The plate heights ranged from 0.15–0.28 mm forNO2

– and NO3– , respectively, at a linear velocity of

0.67 mm/s and increased to 0.25–0.45 mm as the velo-city was increased to 2.4 mm/s. Later in the same year,using the same column and carbonate or hydroxide basedeluents with or without added methanol, this group illu-strated the use of UV detection in capillary IC for the sep-



aration and analysis of UV-absorbing anions, manyorganic acids including aromatic acids such as aminoben-zoic acids, nucleotides and nucleobases [36]. Detectionwavelengths ranged from 195 nm upwards and subse-quently the applicability of UV detection was also demon-strated for polyallylamine-coated silica gel columns (3 and10 lm particle size) to separate and quantitate hydroxy-benzoic acids and nitrophenols with carbonate-bicarbo-nate eluents [37].

4.2 Contributions from the Takeuchi Laboratory

The use of capillary chromatography techniques for theanalysis of ionic species has been extensively investi-gated by Takeuchi et al. beginning in 1988 and continuingto date. They have explored a variety of separation anddetection methods in packed fused silica capillary col-umns. In 1988, they demonstrated the analysis of inor-ganic anions by indirect UV detection on a 0.326100 mmcolumn packed with 3 lm octadecyl silica (ODS) particlescoated with hexadecyltrimethylammonium bromide using1 mM sodium salicylate containing 5% acetonitrile elu-ent [38]. The basic technique of using an adsorbed lipo-philic cation was modeled after the earlier work of Cassidyand Elchuk [39] in macro systems. In the same year, sep-aration and photometric detection of monovalent cationswas demonstrated in a cation exchange system(0.35650 mm, packed with 10 lm particles) with HNO3

as eluent, a capillary scale OH–-form packed column sup-pressor (0.35635–40 mm) followed by an iodide- ornaphthalenesulfonate-form anion exchange column forion-replacement to improve detectability [40]. While theion replacement strategy was inspired by earlier macroscale studies by others [41–43], it is doubtful that ionreplacement here led to better detectability than conduc-tometric detection. Iodide, which displays a much loweroptical absorption than naphthalenesulfonate, led in factto better detection limits and more uniform response, dueboth to the higher background response and to the diffi-culty of replacing an anion as tightly held as naphthalene-sulfonate. Suitably small volume conductometric cellswere used by the authors only later. In this subsequentwork, for reasons not fully understood by the presentreviewers, the authors went to the complexity of using a 4-electrode cell and abandoned the use of suppression.They used a 0.35650 mm column with 0.1 mM potas-sium hydrogen phthalate (KHP) containing 5% acetonitrileand noted that detection limits from chloride to sulfate var-ied from 0.33 to 1.8 ppm [44]. A column of the samedimension but packed with a different 5 lm size anionexchanger was subsequently used by them in conjunctionwith 0.25 mM 2,6-anthraquinonedisulfonate (2,6-AQDS)as eluent, a highly absorbing ion with considerable elutingpower, to achieve an order of magnitude better detectionlimits [45] than their previous efforts involving a dynami-cally coated column and salicylate eluent [38].

4.2.1 Stationary phases coatedwith immobilizedbovine serum albumin

In 1996, Takeuchi et al. published three separateaccounts of ionic determination on columns containingbovine serum albumin (BSA) immobilized on 5 lm ODSparticles. As the eluent pH decreases, more of the BSA isprotonated and anionic analyte retention generallyincreases. Using 0.35675–150 mm columns and an elu-ent containing NaI and tartaric acid, the retention of chlor-ide, nitrite, and bromide was studied by indirect photo-metric detection. The method was also applied to tapwater [46]. Direct UV detection was subsequently usedwith these columns with various acid eluents in thepH range 3–4 for the separation and detection of iodate,bromide, nitrate, iodide, and thiocyanate. The latter threeions were determined in the saliva of smokers vs. non-smokers [47]. In a final publication of this series [48], theycompared NaI, KHP, Na-salicylate, and 2,6-AQDS as elu-ents for indirect photometric detection on a BSA-basedcapillary column under weakly acidic to neutral conditions;the latter eluent ion had the highest absorptivity and pro-vided the best detection limits. In principle, BSA-coatedcolumns can be used as cation exchangers at high pH. Inpractice, the column packing degrades under these condi-tions and cation exchange chromatography cannot be


Figure 1. First capillary ion chromatogram of inorganicanions. Column: 190 lm647 cm, YEW AX-I resin. Eluent:4 mM Na2CO3 and 4 mM NaHCO3 (pH 10.2). Temperature:408C. Flow-rate: 1.9 lL/min. Pressure: 48 kg/cm2. Splittingratio: 70:l. Sample size: 20 lL. Peaks: 1 = F– (1.4 ng); 2 =Cl – (2.8 ng); 3 = NO2

– , (4.2 ng); 4 = PO43 – (8.4 ng); 5 = Br –

(2.8 ng); 6 = NO3– (8.4 ng); 7 = SO4

2 – (11.2 ng). From Ref.[36]. Reprinted by permission from Elsevier Scientific.


carried out on BSA-based columns. The authors earliercarried out cation exchange chromatography on0.35650 mm silica gel columns for the separation of alkalimetal cations using a solution of benzyltrimethylammo-nium chloride (BTMAC, typically 2 mM in 30% acetonitrile)as the eluent [49]. Applicability of the method to real sam-ples such as whisky and ketchup was shown. Subse-quently, the effort towards cation determination by indirectphotometric detection was extended to include alkalineearth metals Mg2+ and Ca2+ on a column of similar dimen-sions using a more strongly absorbing cation 1,19-dimethyl-4,49-bipyridinium as the eluent (in the dichlorideform) [50]; the applicability of the method was demon-strated for beverage samples, including red wine.

4.2.2 Ionically coated ion exchangers

In 1997, taking a cue from the success of Dionex Corpora-tion in modifying surface-sulfonated resins with ionexchange particles of opposite charge, Takeuchi et al.began experimenting with modifying ion exchangers withpolysaccharides that contain ionizable groups [51]. Acoating of chondroitin sulfate (the sugar skeleton containsa –COOH group, a –CH2OSO3H group, and a –NHCO-CH3 group, potentially providing both anion and cationexchange sites) was tested on two different stationaryphases, TSKgel IC-Anion-SW and TSKgel IC-Anion-SW,each in a 0.326100 mm column format. These resins arerespectively silica- and polymer based, differ greatly intheir ion exchange capacities (the former having an orderof magnitude greater ion exchange capacity than the lat-ter) and their pore size distribution such that the chondroi-tin sulfate (estimated average MW 4.1–6.56104) wascompletely excluded from the pores of the first resin butnot of the second. Coating led to quite different effects onretention. For the silica based high capacity low pore-sizeexchanger, absolute retention (using nitrate as a probeanalyte) decreased substantially upon coating. With 1–100 mM Na2SO4 as eluent, the retention of nitratedecreased with increasing concentration of the eluentbefore coating the column, asmay be expected. However,after coating the column, the retention increased withincreasing eluent concentration. This most unusual beha-vior clearly indicated that the mode of analyte retentionwas not simple ion exchange. In contrast to the silicabased exchanger, coating the polymer based stationaryphase led to increased retention compared to theuncoated column and predictable retention behavior as afunction of eluent concentration. Further work was con-ducted only with the silica column. Separation and detec-tion of common UV absorbing anions such as NO3

– , I – ,and SCN– was readily possible with low concentrations(10 mM) of Na2SO4 as eluent and direct UV detection. Thesame conditions could be readily used in detecting theseions in saliva. That the mechanism was not ion exchange

was indeed apparent in that pure water could be used asthe eluent and these three ions could be separated astheir sodium salts. This ability of pure water-based elutionis known for other zwitterionic stationary phases [52].Finally, the authors showed the conductometric detect-ability of a number of anions, using very dilute tartaric acidas eluent.

They followed up this work with heparin-coated col-umns [53]. Like chondroitin, heparin contains an amido-sulfonic acid group aside from carboxylate and sulfonategroups. Retention behavior of both anions and cationswas investigated. Retention of anions on the anionexchanger was remarkably reduced after heparin-modifi-cation. Eluent cations were found to affect analyte anionretention behavior. With Na2SO4 and MgSO4 eluents,retention factor of anions increased with increasing eluentconcentration but opposite (more normal ion exchangelike) behavior was observed with eluents containingAl2(SO4)3, CuSO4, and H2SO4. With CuSO4 as the eluent,the retention and elution of both cations and anions on themodified stationary phase could be indirectly visualized bymonitoring the absorption due to copper ions. With justwater as eluent, the heparin-modified stationary phaseshowed no anion retention. The general area of chromato-graphy on such mucopolysaccharide modified anionexchanger stationary phases in microcolumn format wasthen summarized by the authors [54].

In 1999, Takeuchi et al. began studying dextran sulfate(DS). This anionic polysaccharide is available in pure formin different molecular weight ranges [55]. Using DS ofdifferent molecular weight range and anion exchangeresins of different pore size, they conclusively proved thatwhen the DS is large compared with the pore size of theresin, the negatively charged DS merely attaches to theexposed cationic surface sites of the resin, resulting indecreased anion exchange capacity and normal elutionbehavior (log k 9 of an analyte anion linearly decreasingwith increasing log eluent concentration) is maintained.When the DS is smaller than the pore size of the resin, theDS can get into the pores of the resin and can attach aswell to the cationic sites inside the resin. The greater thepore size relative to the size of the DS, the more DS isadsorbed. Under these conditions, the retention of theanalyte anion does not show ion exchange like behavior:log k 9 increases linearly with increasing log eluent concen-tration. The authors suggest in a qualitative fashion thatrepulsion from the cation exchange sites (i.e., excessunbound sulfate groups in DS) may be increasing withdecreasing eluent concentration and overrides the normalion exchange retention that is affected by the eluent anion.Experiments involving a series of eluent cations andanions of different ion exchange affinities at the sameionic strength will be necessary to better understand the



behavior of such systems in light of the Stern-Guoy-Chap-man double layer theory [56, 57].

Using low MW DS-coated columns, separation of alkaliand alkaline earth cations was demonstrated using bothCuSO4 and the much more strongly absorbing rutheniumtris(bipyridyl)3+ as the eluent; anion separation perform-ance was not particularly attractive. Using the same col-umn dimensions (0.326100 mm), the technique was sub-sequently demonstrated for analysis of blood serum sam-ples [58]. The general behavior of positively chargedanion exchange columns coated with substances that arenegatively or zwitterionically charged are similar to nonpo-lar columns coated with zwitterionic substances; Takeuchiet al. have also contributed to this area; this is discussedlater.

4.2.3 Recent efforts

The fluorimetric detection of Mg and Al as their oxine com-plex was studied in the microcolumn format (0.326100 mm, ODS packing), with (a) oxine present in the elu-ent (comprising a buffered micellar sodium dodecylsulfatemedium to effect the solubilization of oxine) or (b) oxineadded postcolumn [59]. The incorporation of oxine in theeluent was found to produce substantially superior per-formance relative to postcolumn addition. The LOD for Mgwas 18 lM. The method was significantly more sensitivefor Al but blanks could not be reduced below 0.80 lM.There is great potential of these type of approaches fordetecting a variety of metal ions. The sulfonated derivativeof oxine is water soluble and has been extensively studiedfor the numerous fluorescent metal complexes itforms [60]. Attractive detection of many metal ions hasbeen demonstrated in macroscale systems [60–62]. Highintensity light emitting diodes that are ideally suitable toexcite metal-sulfoxine complexes have become inexpen-sively commercially available [63]. In combination withcapillary scale liquid core waveguides based on TeflonAF-coated fused silica tubes can form the basis of dedi-cated, monochromatorless, highly sensitive fluorescencedetectors [64].

The environmental nitrogen budget, in particular the sup-ply of nitrogen to bodies of water, is often a topic of consid-erable interest. The dominant forms of soluble nitrogenconsist of nitrate-N and ammonium-N, with small amountsof nitrite-N being occasionally present. The Takeuchigroup developed a method for determining NO2

– and NO3–

by direct UV detection at 206 nm using a 0.326100 mmcolumn packed with a commercial anion exchanger and a5 mM Na2SO4 eluent. Ammonium is not significantlyretained by the column and comes out first. The UV detec-tor effluent is mixed with o-phthaldialdeyde-mercap-toethanol reagent in a tee and heated to temperatures upto 658C in a 0.5–2 m long 50 lm ID reaction coil to forman isoindole derivative that is detected by its fluorescence.

The UV detector itself responds to eluted NO2– and NO3

– .Application to analysis of river water was demon-strated [65]. The Takeuchi group has also studied largevolume injections in a microcolumn format; this is dis-cussed in a later section.

Takeuchi et al. have obviously made impressive contribu-tions to capillary chromatography in general and capillaryIC in particular, especially in reference to exploring novelstationary phases and retention mechanisms. The aver-age IC practitioner still needs to be sold on performance,however, especially in comparison with conventional sys-tems.

4.3 Capillary IC on bile salt and other zwitterionicmicelle coated columns

The work of Takeuchi et al. described above provides anappropriate backdrop to next consider zwitterionic micel-lar bile-salt coated reverse phase columns that have beenused for IC, both on the capillary scale and otherwise. Thefirst reported application of such a stationary phase for ICwas by Hu, Takeuchi, and Haraguchi in 1992 in which theseparation of alkali metal ions was achieved by indirectphotometric detection using CuSO4 [66]; the detectionlimits were modest. The same technique was used by theauthors more sensitively and effectively with a CeCl3 elu-ent and it was possible to separate both alkali and alkalineearth metals [67] and applied extensively to differenttypes of beverages. Hu and Haraguchi then applied themethod for the determination of UV-absorbing ions, nota-bly NO2

– , NO3– , I – , and SCN– , in saliva of subjects of

different age, sex, and smoking habits [68]. The LODsranged from 0.5 lM for I – to 2.3 lM for SCN– with a phos-phate buffer as eluent and UV detection at 230 nm. Insmokers [SCN– ] was significantly elevated and alsodepended on the age and sex, while the other anion con-centrations did nor show any particular pattern.

The best known publication in this area, by Hu, Takeuchi,and Haraguchi, appeared in 1993; “Electrostatic Ion Chro-matography” was performed in this report on both the con-ventional and capillary scales [69]. An illustrative separa-tion using pure water as eluent is shown in Figure 2. Thelure of using pure water as eluent is considerable. How-ever, the intervening time has proven that this quest maybe elusive. On such a system, the actual separation isconsiderably dependent on the ionic strength of the sam-ple itself (as would be predicted from the Stern-Guoy-Chapman theory of the double layer) and the possibilitythat the exact elution time of an anion may be dependenton the counterion present. The exact mechanism of sep-aration that was proposed in this paper has also sincebeen revisited. Hu and Haraguchi also revisited determi-nation of analytes in saliva [70] and proceeded to showthat simultaneous determination of cations and anions ispossible on such a column using CuSO4 as the eluent to



permit optical detection [71]. An illustrative separation isshown in Figure 3. Both CuSO4 and CeCl3 were used aseluents for the separation and determination of Na, K, Mg,and Ca in human saliva on 0.356150 mm columns; theresults agreed very well with values determined byplasma-atomic emission methods [72].

4.4 Unique applicability of capillary IC

There have been few occasions where the scale of theseparation/detection system has been vital to the successof the mission. B�chmann et al. [73] wanted to analyzesize-classified raindrops for both their cation and anioncomposition. According to their relative sensitivity anal-ysis, the optimum sample volumes for conventional scaleIC, capillary IC (0.5–0.75 mm ID), and capillary zoneelectrophoresis (CZE, 75 lm ID) are: 50, 0.5, and 0.01 lL(these translate to spherical drop diameters of 4.6, 1.0,and 0.25 mm), respectively. Depending on windspeed,raindrops can span these size ranges. Accordingly, theresearchers used all three techniques. (Interestingly,while chromatograms for actual raindrops were presentedfor both scales, only standard electropherograms wereshown. Perhaps it is not surprising that CZE is no longerconsidered a viable competitor to IC for ion analysis!)Cation chromatography involved 110–250 mm long0.75 mm ID capillary columns with a Ce(NO3)3 eluent andindirect photometric detection via fluorescence. Injectionvolumes ranged from 0.5 to 2.5 lL. Anion chromatogra-phy was conducted with a 0.56150 mm reversed phasecolumn coated with dodecylamine with indirect absor-bance detection at 205 nm using K3[Fe(CN)6] as the elu-ent. For all of the common anions and cations measured(Na+, NH4

+, Ca2+, Mg2+, Cl – , NO3– , SO4

2 – ), the concentra-tion in the raindrops increased dramatically with decreas-ing size.

The second application involves for the same reason thatcapillary LC is finally beginning to enjoy commercial suc-cess – it is the optimum scale for coupling with a massspectrometer. Buchberger and Haider first demonstratedthe advantages of capillary scale IC-MS for the analysis ofinorganic anions and aminopolycarboxylic acids [74]. TheS/N = 3 LOD for a variety of anions were in the 20–80 ngrange.

4.5 Large volume injections in capillary IC:On-column enrichment

In liquid chromatography, direct large volume injectionsare often impractical because the sample matrix cannotoften be made weak enough to prevent premature elutionof sample components. In ion exchange chromatography,water as an eluent has near-zero eluent strength. As such,on-column enrichment by large volume injections of lowionic strength samples are straightforward. Kourilova etal. [75] used sample volumes ranging from 200 nL to 1 mLon 0.7–0.96150 mm columns packed with 7.5 lm cationexchange particles. Using eluents containing ethyl-enediamine and tartaric acid and conductometric detec-tion, they were able to obtain 5–50 nM detection limits for1 mL injection volumes for a variety of metals. This is parti-cularly impressive in view of the equipment used and therelatively poor column efficiencies attained. Although at


Figure 2. Chromatogram of inorganic anions. Column,Develosil ODS-5 coated with CHAPSO micelles, mobilephase: pure water, flow rate: 2.8 lL/min, UV detection at230 nm, analytes concentration: 0.1 mM each. Reprintedfrom Ref. [69] by permission from the American ChemicalSociety.

Figure 3. Simultaneous separation of inorganic cations andanions: column, 0.35 mm ID6150 mm packed with DevelosilODS-5 coated with NaTDC micelles; mobile phase: 5 mMcopper(II) sulfate aqueous solution, UV detection at 210 nm,analytes concentration: 0.5 mM each. Reprinted from Ref.[71] by permission from the American Chemical Society


1 mm column ID, it is on the border of what can be calledcapillary scale, Ito’swork [76] on the direct determination ofiodide in seawater by using large volume (2 mL) injectionandUVdetection at 226 nm isworthyofmentionbecause itsolves a practical need using a very short column(1635 mm). With a high capacity ion exchange packing,an eluent containing 30 mM NaClO4, 500 mM NaCl, and5 mM Na-Phosphate buffer (pH 6.0), the author was ableto achieve an LOD of 200 ng/L, a2 nM, practical for iodidedetermination in both oceanic and estuarine waters. Lim etal. have also recently reported on the separation anddetection of ppb levels of different anions using a 10 lLinjection on a 0.326100 mm column (5 lm high capacityion exchanger) using an NaCl eluent and UV detection at210 nm [77]; the separation efficiency is, however, notcompetitivewith commercialmacroscale counterparts.

4.6 Suppressed IC in capillary format

Since the original work of Rokushika et al., few research-ers have ventured into membrane suppressed capillaryIC, attempting to miniaturize precisely what in macroscalehas proved to be a considerable success. Sj�gren etal. [10] constructed the first capillary IC complete with aelectrodialytic generator (EDG) for the eluent. Indeed, thecapillary scale greatly facilitated fabrication of an EDG onthe high pressure side which had not previously beenachieved in macroscale systems. The system diagram isshown in Figure 4. A commercially available low coststepper motor driven syringe pump containing a 500 lLglass syringe is used with a 3-port header, where a mildlypressurized water reservoir is connected through an inletcheck valve. The syringe itself and the downstream com-ponents are connected through the other two ports. Tobegin operation, the syringe begins to aspirate, whichopens the inlet check valve and allows water to come in tothe syringe. During the dispensing process, a rapid initialmovement builds up pressure and seals the inlet check

valve and water flows out via the pressure sensor tee tothe high pressure EDG. The EDG consists of a lower flowpath cut into a stainless steel (SS) block through whichthe syringe output flows. This is separated by a thickcation exchange membrane from an upper flow path cut ina polymer insert. A donor NaOH solution is made to flowat a low flow rate by modest pneumatic pressure throughthe upper flow channel to waste. A platinum wire is alsopresent in the upper flow path. Application of voltagebetween the SS block (–) and the Pt wire (+) causes Na+

to migrate across the membrane and form NaOH and H2

gas on the cathode side, the amount being faradaicallyproportional to the microampere level current flowingthrough the system. The NaOH output from the SS chan-nel is connected by a PTFE tube (under the operatingpressure, the hydrogen gas present in the stream isremoved by permeation through the polymer tube wall) toa low-volume miniature in-line conductivity cell to mea-sure the generated NaOH concentration. The eluent con-centration can be programmed to change by changing theEDG current. The eluent flows through a 100 nL volumeloop injector, a 0.186560 mm capillary column packedwith AS–11 packing material from Dionex, a 80–100 lmID Nafion tube based suppressor (L11 mm active length),and a capillary conductivity cell. The small column diam-eter led to high permeability and allowed the use of a long,high plate number column at a flow rate of 2 lL/min(1.3 mm/s, equivalent to the flow rate of 1 mL/min in a4 mm ID column) at a pressure drop of a800 psi. In iso-cratic operation, the column efficiencies per unit length forall the tested ions were significantly better than a 2 mm IDcolumn packed with the same packing. The differencewas as large as 2.8 times for an early eluting ion like fluor-ide to a point of no difference for late eluting phosphate(suggesting that suppressor/detector contributions to lossof plates in conventional bore systems could have beensignificant). A gradient separation on this system is shownin Figure 5. Experience with these systems suggests that


Figure 4. Schematic of the experimen-tal setup for electrodialytic eluent gen-eration suppressed capillary IC usedby Sj�gren et al. [10]. Reprinted bypermission from the American Chemi-cal Society.


the high permeability of capillary systems may allowlonger columns operated at higher than optimum flowrates to achieve same separations faster than a conven-tional bore system.

In the following year, Boring et al. reported a field-portableversion of this system that fits into a custom briefcase(28643615 cm, 10 kg) [78]. Several novel features wereintroduced: a trap column consisting of a chelating resinand a mixed bed resin was used ahead of the EDG toremove impurities, a polystyrene capillary (80 lm ID,250 lm OD, 10 cm long) was used downstream of theEDG for more efficient H2 gas removal with a lower resi-dence volume, a 0.05619 mm radiation grafted sulfo-nated PTFE ion exchange fiber provided suppression anda novel conductivity cell where the electrode spacing wasreadily changeable was used for detection along withbipolar pulse electronics, a 10 cm long 180 lm ID precon-centration column was used for preconcentration experi-ments. The total power requirement of the system, notincluding that of the laptop computer used for instrumentcontrol and data acquisition was 17.5 W.

Boring et al. [79] then coupled a preconcentration columnequipped capillary IC system described above to a minia-ture parallel plate denuder used for collecting solublegases. They showed that low parts per trillion level detect-ability of trace ionogenic gases is readily attainable(Figure 6).

IC equipment manufacturers may finally have realized themerit of the capillary scale as evidenced from a recent pre-

sentation from Dionex Corporation [80]. Although manydetails are still proprietary, attractive performance hasalready been demonstrated. The impressive repeatabilityof an isocratic chromatogram is shown in Figure 7.


Figure 5. Gradient chromatogram:The right ordinate shows the gradi-ent (dashed trace) in mM NaOH.Peak identities: 1, Fluoride; 2, For-mate; 3, Monochloroacetate; 4,Bromate; 5, Chloride; 6, Nitrite; 7,Trifluoroacetate, 8, Dichloroace-tate; 9, Bromide; 10, Nitrate; 11,Chlorate; 12, Selenite; 13, Tar-trate; 14, Sulfate; 15, Selenate; 16,Phthalate; 17, Phosphate; 18,Arsenate; 19, Citrate. All com-pounds injected were 50 lMexcept selenate and sulfate thatwere 25 lM. The baselineincreased by L500 nS/cm duringthe run and a baseline subtractionwas performed using a blank gradi-ent run. The fluoride peak co-elutes with an unidentified impurity.0.186560 mm capillary columnpacked with Dionex AS–11 pack-ing, a 80–100 lm ID Nafion tubebased suppressor (L11 mm activelength). Flow rate of 2 lL/min,pressure drop a800 psi. Reprintedfrom Ref. [10] by permission fromthe American Chemical Society.

Figure 6. Chromatogram resulting from sampling blank air(top) and 80 pptv SO2 (middle) (both at 500 sccm) andlaboratory air (bottom, 250 sccm). Peak identities: 1, chlor-ide; 2, nitrite; 3, carbonate; 4, sulfate. Reproduced by per-mission from Elsevier Scientific


5 Open tubular IC

Open tubular capillary IC provides the advantage overpacked columns of low flow resistance, permitting the useof low pressure pumping systems. This may lead to thenatural progression on IC separations on a chip, wherethe limited number of attempts thus far have used eitherelectroosmotically generated flow or off-chip HPLCpumps and split flow. The possibilities of using low pres-sure, pneumatic or even gravity-induced flow for chroma-tographic separations makes the OT approach attractivenot only with respect to microfabricated systems but withfield-portable chromatographs in general. The open tubu-lar approach however has also some inherent disadvan-tages that stem from the relatively small diffusion coeffi-cients of analytes in the liquid phase at room temperature.While open tubular capillaries up to 530 lm in ID are rou-tinely used in capillary gas chromatography with goodseparation efficiencies, the diffusion coefficients of theanalytes, especially at elevated temperatures, are typi-cally L104 times larger than those encountered in LC. Thecharacteristic time t for diffusive travel of distance d is pro-portional to d 2/D where D is the diffusion coefficient. It fol-lows that ideally capillary internal diameters for use in LCshould be about 102 times smaller than in GC. However,the Knox and Gilbert [28] optimum ID of a0.3 lm is unli-kely to be practical in the near future. Going down in col-umn diameter also places corresponding burdens oninjection and detection techniques. The injector demandscan be bypassed by splitting the flow after the injector butthere is no way to bypass the small detection volumerequirement. As a result, progress with attractive perform-ance has not been plentiful.

As early as in 1981, Ishii and Takeuchi demonstratedopen tubular cation-exchange chromatography [81]. Intheir work, 30–60 lm soda lime glass capillaries werecoated with phenyltriethoxysilane or 2-mercaptoethyl-triethoxysilane, and subsequently ion exchange groupswere introduced by treatment with concentrated H2SO4 at908C for 4 hours or by oxidation with permanganate. Sep-aration of four nucleosides was demonstrated. Elevatedtemperature operation was shown to be beneficial due tothe increase of the diffusion coefficients of the analytes.Further, surface treatment of the soda lime capillary with1N NaOH over a prolonged period prior to coating resultedin 9–18 times greater surface area and correspondinglyhigher column capacity.

In 1983, Manz and Simon [82] illustrated the use of apotentiometric microelectrode with a demonstrated detec-tion volume a500 pL, with the best estimate of the effec-tive volume being 10 fL with a 25 lm ID column. A 1 lmtip diameter ion selective electrode was inserted into theopen end of a 25–30 lm diameter column. The detectorperformed admirably but the separation system showedpoor efficiency for significantly retained analytes. By 1989this problem was solved and the group was able to coatthe interior of columns 4.6 to 9.5 lm in ID with polybuta-dienesulfonic and polybutadiene-maleic acid (PBMA) pre-polymers and polymerize them in situ. For the 4.6 lm IDPBMA column (estimated coating thickness of 0.1 lm),100% and 89% of the theoretically expected efficiencywas achieved for a 90 cm long column, with 680,000 and246,000 plates for analytes with k 9 values of 0 and 0.16,respectively. The authors were able to separate alkali andalkaline earth metals in about 4 min and alkali metals as agroup in under 90 s, reaching LODs of 0.2–20 fmol [83].They also showed separations of various amino acids.Using coatings of sulfopropylhydroxysilane and amino-alkylalkoxysilane in 4.6 lm ID columns, 0.33–0.37 m inrespective length, the group was then able to show sep-arations of alkali metal cations (10 pL injection, 20 mMHCOOH, 10 nL/min) in under 25 s and common anions(10 pL injection, 10 mM HCOOH, 26 nL/min) in under35 s [84]. These are shown in Figure 8. They alsoshowed that, on operation at an alkaline pH (NH3/NH4Clbuffer), it is possible to separate alkali cations in a 2.3 lmID bare silica column, by virtue of ionization of the silanolgroups.

Rather than going to smaller capillaries to solve the pro-blem of slow mass transfer to the wall, Pyo et al. [85]attempted a different and in some sense a more directapproach: increasing the diffusion coefficient of the ana-lyte. They accomplished this by raising the column tem-perature as high as 1508C. Relative to analytes in LC, ICis already better off in that most analytes of interest aresmall high mobility ions. According to the Stokes-Einstein


Figure 7. Prototype 380 lm6250 mm column, capillaryelectrogenerated KOH eluent, flow rate: 12 lL/min, tempera-ture: 358C. Tubular Capillary Suppressor, suppressed con-ductivity detection. 20 separate runs overlaid on oneanother. Courtesy Yan Liu, Dionex Corporation.


equation, the analyte diffusion coefficient D changes withtemperature according to:

D = kT/(4 p g r) (5)

where k is Boltzmann’s constant, T is the absolute tem-perature, g is themobile phase viscosity, and r is the sphe-rical equivalent radius of the analyte. As it happens, for alargely aqueous eluent, the T/g ratio increases by a factorof four from 25 to 1008C. They demonstrated that up to a6-fold increase in the number of theoretical plates isobserved in anion separations in 50 lm capillaries fromroom temperature to 120–1508C. However, at least insome cases, this increase in efficiency is not due to theincrease in the diffusion coefficient of the analyte alone;the retention itself also decreases significantly at highertemperatures, particularly when adsorbed ion exchangersare used (vide infra). One way in which these experimentswere conducted was through the use of positively charged(anion exchanger) latex particles electrostatically boundon the negatively charged walls of the fused silica column.This approach provides relatively good retention but qua-ternary ammonium functional groups degrade rapidly atincreasing temperature due to Hofmann degradation.They experimented therefore with dynamically adsorbedion exchanger coatings (e.g., with eluents such as hexa-decyltrimethylammonium chloride). This was also notwithout problems: the adsorption of the surfactant on thecolumn walls decreased markedly at temperatures above1008C. (The eluent was maintained in the liquid state attemperatures over its normal boiling point by using an exitrestriction capillary.) Pyo et al. also studied the relationbetween the optimum flow rate (the Van Deemter mini-mum) as a function of the column temperature. They

showed that the optimum velocity moves to higher flowvelocities at increased temperature [86, 87]. Further, therelative rate of loss of plates with increasing flow velocitypast the optimum is much less pronounced at higher tem-peratures than at lower temperatures. Fast separationstherefore should be more easily attained at higher tem-peratures.

Functionalizing the interior walls of a small capillary is anontrivial task. Especially when used in IC, the siloxanebonds obtained with alkoxysilane chemistries [82, 83] arenot acceptable for practical use because of their hydrolyticinstability. Recently, Pohl et al. have pioneered anapproach where diepoxide-amine chemistry can be usedto build up a unique hydroxide selective anion exchangerlayer by layer [88]. In unpublished work, Kuban et al. [89]have fabricated open tubular capillaries using a 75 lm IDcolumn and 25 alternating layers of 1,4-butanediol diglyci-dyl ether and methylamine. Electron microscopy showsthat the interior coating is only 0.7 lm thick (Figure 9).Using such a column, suppressed open tubular IC wascarried out for the first time (Figure 10) using a 11 mmlong 80 lm diam. membrane suppressor. At the presenttime, such approaches are not practical because the sep-aration still takes too long and the column exchange capa-city is still rather inadequate. However, it definitely pro-vides a path forward for future experiments with smallerdiameter columns.

6Monolithic columns in ICMonolithic columns occupy an intermediate positionbetween packed and open tubular columns albeit they aremore closely related to the former. Flow resistance in


Figure 8. Fast separation of alkalimetal cations. Reprinted from Ref.[84] by permission from Wiley Inter-science.


monolithic columns is lower than that in packed columns.In properly designed monoliths, column efficiencies aremaintained at relatively high flow velocities, enabling highvelocity rapid separations. The use of commercially avail-able porous silica-based monolithic rods in a conventionalsize (i.e., 4.6 mm610 mm) [90] for high speed IC separa-tions using ion interaction chromatography and indirect

conductometric detection was recently demonstrated byHatsis and Lucy [91, 92]. They were able to accomplishthe separation of 8 common inorganic anions in an amaz-ing 15 s at a flow rate of 16 mL/min. The eluent consump-tion in this mode is obviously very high; however, in acapillary format this will not be an issue. It should benoted, however, that unlike the case of transition betweenconventionally sized packed columns to packed capil-laries where column permeability generally increasesmarkedly, an increase in permeability in going from con-ventional to capillary sized monoliths may not be realized.The pressure drop on monolithic columns will still be highenough to require high pressure pumping systems.

Monolithic columns have been created inside a microborecapillary by polymerization or by a sol-gel process. Inwhat is thus far the only report on use of monolithic silicacolumns in capillary format IC, Motokawa et al. [93] pre-pared porous silica monoliths in 50 to 200 lm fused silicacapillaries with various skeleton and through-pore sizesand used it for high speed ion chromatographic determi-nation of acidity [94].

7 Ion chromatography on a chipLiquid chromatography on a chip was first demonstratedbyManz et al. [95] in 1990. Although capillary electrophor-esis has been widely implemented in a chip format,reports on chip scale LC/IC are scarce, due to the obviousdifficulties of integrating pumping, injection, and detectionsystems. If all else is located off-chip, it is still necessaryto fabricate efficient chip-scale microcolumns. He et al.[96] have used a chromatographic structure fabricated on


Figure 9. High magnificationimage of the multilayer (7 layers)coating inside a capillary madeby diepoxide-amine chemistry.The coating is estimated to be200 nm thick. Courtesy RobertAnderson, Texas Tech Univer-sity.

Figure 10. Separation of inorganic anions in OT ion chroma-tography. Eluent: 1 mM KOH flow rate 1.5 mL/min,5 m675 lm ID column, membrane-suppressed, contactlesson-capillary conductivity detection; 25 pmol each analyte,a20 psi pressure drop.


a quartz wafer for LC on a chip with flow being electroos-motically generated. The fabrication of a monolithic col-umn inside the microfluidic structure on a chip wasreported recently by Ericson et al. [97]. They have usedboth electroosmotic flow and pressure driven flow for on-chip LC. The integration of a chromatographic pumpingsystem on a chip, capable of providing high pressure elu-ent flow, has not yet been achieved.

Thus far, there are only two reports on chip-scale IC.Kang et al. [98] described a microfabricated device withmultiple parallel channels for conducting IC. The station-ary phase was prepared in situ and each channel con-tained its own dedicated conductivity detection elec-trodes. The limit of detection (LOD) of injected analyteswas stated to be about 50 mM, which may have been wellabove the concentration (unspecified) of the biphthalateor carbonate eluent used. This performance level under-scores the importance of the need to develop somemeans of carrying out continuous ion exchange basedsuppression. Murrihy et al. [99] described an IC system inwhich separation takes place in a latex-coated channelmicrofabricated on silicon (as well as in non-negligiblelengths of connecting tubing) with off-chip optical detec-tion. In both reports, the pumping was achieved by an off-chip conventional HPLC pump.

8 Evolution of suppressorsAt least in the conventional format, most find ion chroma-tography synonymous with suppressed conductometricdetection. Suppressed conductometry is both uniquelyapplicable to ions and provides sensitive detection that isnot degraded by scaling down. Naturally, efforts are madeto continue to perform both on the capillary scale; how-ever, this is a challenging task. The issue of miniaturiza-

tion of conductometric detectors is discussed in the nextsection. Presently we discuss the issue of suppression.Although packed column based suppression has beenused in the capillary scale [40], this is a dead end; it haseven more disadvantages on the capillary scale than inmacroscale IC. Early on Rokushika et al. [26] used a shortlength of a thermally stretched Nafion membrane as asuppressor and small diameter (75–80 lm ID) Nafionmembranes have since been extruded. However, thismay be the limit of extrusion technology. Such mem-branes and radiation grafted tubular membranes havebeen used both in suppressed capillary IC [10, 78–80])and suppressed conductometric capillary electrophoresis[100–104]. Low dispersion interconnections between themembrane and the suppressor and the suppressor andthe detector on this scale are not a trivial task. Tubularmembranes of even smaller diameter can be custom-fab-ricated. In unpublished work, Kuban and Dasgupta havedemonstrated that very small diameter membrane tubescan be cast with Nafion solutions, using a platinum wireinserted in a capillary as a mandrel. A photograph of sucha membrane is shown in Figure 11. In principle, this tech-nology can obviate the need for interconnects. When min-iaturizations proceed to the chip-scale, the construction ofsuppressors may need to be rethought. Polymeric chipsof appropriate construction may lend themselves to con-version of a portion of the chip into a membrane. Withglass/quartz/silicon wafers, a liquid or suspension basedion exchange suppression will be practical and can bescaled to almost any dimension. Microscale continuousion exchange has recently been demonstrated by Kubanet al. in both chip scale flows occurring in horizontal side-by-side streams [105] and vertical, layered streams [106],using both an ion exchanger dissolved in an immisciblesolvent or a colloidal size ion exchange resin dispersed inwater.

9 Miniaturizing conductivity detectionEarly attempts at miniaturizing conductivity detectors lar-gely involved putting together stainless steel tubular elec-trodes face to face in close proximity. Although notcoupled to capillary systems, a variety of electrode/celldesigns using small diameter metallic capillaries/needleswere described in the literature early on [107]. Thin-walledstainless steel tubes are readily available down to 50 lmin diameter and readily permit making such detectors ordetectors of the wall-jet geometry (such as that deployedby Boring et al. [78]). Conductivity detectors have beenextensively used in capillary electrophoresis as well. Thefirst commercial conductivity-based capillary electrophor-esis detector is also essentially of a shielded wall-jetdesign [108]. Huang et al. [109] described an on-columnconductivity detector which involved laser-drilling 40-mmholes on opposite sides of the capillary walls, followed by


Figure 11. Photograph of a 50 lm ID Nafion membrane castdirectly on capillaries.


the insertion of two 25-mm diameter Pt wires inside theholes to act as electrodes. Subsequently, a simplergrounded end-column arrangement was reported for bothconductometric and amperometric detection. The sensingelectrode (Pt wire, 50-mm in diameter) in this set up wasplaced at the outlet of the separation capillary [110].Some further refinements of this end column detectionarrangement were later reported [111]. The suppressedconductometric detectors used by the Dasguptagroup [100, 102–104] were constructed by inserting two100-lm-diameter Pt wires through the wall of a 190-lm-IDpoly(vinyl chloride) (PVC) capillary in parallel and as closeto each other as possible. Basically the same geometrywas later modified to be a reproducible detection systemby using a bifilar wire [112]. The group at Dionex [101]used disk-shaped metallic electrodes with 75 lm holes inthem separated by a similar but insulating spacer disc asthe conductivity detection cell for the same purpose.Given current microfabrication technology, conductivitydetectors are perhaps best fabricated on-chip [98]; theremay be merits to such detectors even when separationsare carried out in conventional capillaries. Very low disper-sion chip-capillary connectors have been designed [113].Meanwhile, nongalvanic contact conductivity detection byusing high frequency ac excitation voltages has beendeveloped. Such techniques have some obvious advan-tages in that detection can be conducted directly on theseparation capillary, without having to resort to a separatecell. Such detectors are described in the next section.However, unlike galvanic contact conductivity detectors,signal/noise in a nongalvanic contact conductometricdetector (hereinafter called capacitively coupled contact-less conductometric detector, C4D) may be somewhatscale-dependent. This has not yet been systematicallystudied but it is believed that conductivity of the doublelayer near the capillary wall dominantly governs theobserved signal [114].

9.1 C4D and its potential for capillary IC

Contactless conductometric detection based on oscillo-metry [115] was used as early as 1983 by Pungor et al.[116], who constructed a flow-through conductivity detec-tor without galvanic contact. The detector cell was con-nected to a commercial oscillotitrator and consisted of twoconcentric electrodes coated with either methylsilicone orTeflon. It was used for flow injection basedmeasurementsof conductivity and permittivity. In 1988 they used thisdetector in an end-column format for IC [117]. The electro-nic circuitry was modified from the previous version and ahigh frequency (42 MHz) crystal oscillator was used forexcitation. Detector performance was evaluated in bothsuppressed (carbonate/bicarbonate) and non-sup-pressed (salicylate/salicylic acid or benzoate/benzoicacid) eluent systems: the performance of the detector was

shown to be comparable with state-of-art commercialdetectors using galvanic contact.

Almost a decade later, C4D was revived for capillary for-mat separation techniques, notably capillary electrophor-esis (CE). Independently, Zemann et al. [118] and daSilva and do Lago [114] refined C4D. In the on-columndesign, the C4D detector cell consists of two cylindricalelectrodes that are positioned directly on the outer wall ofthe separation capillary (typically made of an insulatingmaterial, commonly silica) and are not in galvanic contactwith the solution inside the capillary. They are separatedby a small gap, typically 1–2 mm, that defines the lengthof the solution inside the separation capillary, the conduc-tivity of which is measured. See Figure 12. The electro-des are painted directly on the capillary outer wall using asilver conductive epoxy or alternatively, the separationcapillary may be inserted inside two independent syringecannulae. The latter design is more flexible and is thuspreferred. High frequency alternating voltage (typically20–600 kHz sine wave) is applied to the first electrode.The alternating current output at the second electrode isrectified and amplified using relatively simple circuitry.Excitation amplitude was typically 10–20 V; the schemehas been widely used in CE [114, 118–123]. Tanyanyiwaet al. [124] have subsequently demonstrated that if theexcitation voltage amplitude is increased (in the range of25 to 250 V), the sensitivity increases nearly linearly withthe voltage applied while the noise remains almost thesame. In their detector configuration, the noise waslargely generated by the detector electronics and not thedetection cell.

Since its resurgence in 1998, the number of C4D publica-tions has crossed the half century mark at the time of thiswriting. A recent review by Zemann [125] indicates thatC4D has been used primarily for electrophoresis, both incapillary and chip formats. The only exceptions are thework of Hilder et al. [119] and Kuban et al. [126] where thedetector was used in IC. Kuban et al. used a high voltage


Figure 12. A typical on-capillary C4D cell, schematicallyshown.


C4D on a 150 lm ID fused silica capillary in series with acommercial galvanic contact based conductometricdetector while monitoring the effluent from a conventionalscale ion chromatograph. The performance of the twodetectors in terms of S/N, etc. was essentially the same,confirming the conclusion of Pal et al. [117] from two de-cades ago.

The on-column C4D design offers several advantages.The detector cell can be positioned at any location alongthe separation capillary (unlike the detectors with galvaniccontact or the original oscillometric detector used by Pun-gor et al.). While there may be some dependence of thedetection sensitivity and/or the S/N ratio on the capillaryID, this does not appear to be nearly as great as in opticaldetection. Mayrhofer et al. [127] have shown that C4Ddetection of inorganic cations, separated by electrophor-esis in capillaries of 10 lm ID is feasible. Hilder et al. [119]used C4D in anion-exchange capillary electrochromato-graphy; detection was performed directly across a 75 lmID packed bed. Obviously, the concept should be of utilityin both packed and open tubular capillary IC systems. It isinexpensive to build C4D electronics (L US $ 200 in com-ponents) and it is compact so that it is particularly attrac-tive in portable instrumentation [128]. Multiple C4D sys-tems can be used on a single separation capillary, and incombination with other detection techniques, such as UVdetection [129].

Capillary IC is an evolving field that is still very much in themaking. Whether it will finally emerge as a viable entity inthe form of open, packed, or monolithic columns, in acapillary form or on a chip, will require a sharp enoughvision to see inside a capillary!

References[1] C.A. Lucy, J. Chromatogr. A 1996, 739, 3.

[2] Exodus, Chapter 15, verses 22–25. Text from the KingJames version.

[3] H. Small, T.S. Stevens, W.C. Bauman, Anal. Chem. 1975,47, 1801.

[4] American Chemical Society,Milestones in Analytical Chem-istry,Washington DC, 1994.

[5] T.S. Stevens, J.C. Davis, H. Small, Anal. Chem. 1981, 53,1488.

[6] T.S. Stevens, G.L. Jewett, R.A. Bredeweg, Anal. Chem.1982, 54, 1206.

[7] Y. Hanaoka, T. Myrayama, S. Muramoto, T. Matsuura, A.Nanba, J. Chromatogr. 1982, 239, 537.

[8] D.L. Strong, P.K. Dasgupta, K. Friedman, J.R. Stillian, Anal.Chem. 1991, 63, 480.

[9] D.L. Strong, C.U. Joung, P.K. Dasgupta, J. Chromatogr.1991, 546, 159.

[10] A.Sjogren, C.B. Boring, P.K. Dasgupta, J.N. Alexander, IV,Anal. Chem. 1997, 69, 1385.

[11] D.T. Gjerde, J.S. Fritz, G. Schmuckler, J. Chromatogr.1979, 186, 519.

[12] D.T. Gjerde, G. Schmuckler, J.S. Fritz, J. Chromatogr.1980, 187, 35.

[13] J.S. Fritz, D.T. Gjerde, R.M. Becker, Anal. Chem. 1980, 52,1519.

[14] R.M. Wheaton, W.C. Bauman, Ind. Eng. Chem. 1953, 45,22.

[15] G.A. Harlow, D.H. Morman,Anal. Chem. 1964, 36, 2438.[16] K. Tanaka, T. Ishisuka, J. Chromatogr. 1979, 174, 153.[17] H. Small, J. Riviello, Anal. Chem. 1998, 70, 2205.[18] H. Small, Y. Liu, N. Avdalovic,Anal. Chem. 1998, 70, 3629.[19] K. Srinivasan, R. Lin, S. Saini, C. Pohl, N. Avdalovic,

Abstract 2020–2, Pittcon 2003, Orlando, FL, March 13,2003.

[20] C.G. Horvath, B.A. Preiss, S.R. Lipsky, Anal. Chem. 1967,39, 1422.

[21] C.G. Horvath, S.R. Lipsky,Anal. Chem. 1969, 41, 1227.[22] D. Ishii, K. Asai, K. Hibi, T. Jonokuchi, M. Nagaya, J. Chro-

matogr. 1977, 144, 157.[23] T. Tsuda, M. Novotny,Anal. Chem. 1979, 50, 271.[24] T. Tsuda, M. Novotny,Anal. Chem. 1979, 50, 623.[25] R.P.W. Scott, P. Kucera, J. Chromatogr. 1979, 169, 51.[26] S. Rokushika, Z.Y. Qiu, H. Hatano, J. Chromatogr. 1983,

260, 81.

[27] T. Tsuda, M. Novotny,Anal. Chem. 1978, 50, 271.[28] J.H. Knox, M.T. Gilbert, J. Chromatogr. 1979, 186, 405.[29] http://www.alltechweb.com/literature/brochure/B484p8.pdf.

[30] M. Golay,Gas chromatography 1958, D.H. Desty ed. Butter-worths, London, 1959, p. 36.

[31] J.H. Knox, J. Chromatogr. Sci. 1980, 18, 453.[32] R. Tijssen,Chromatographia 1970, 3, 525.[33] R. Tijssen,Sep. Sci. Technol. 1978, 13, 681.[34] A. Woodruff, C.A. Pohl, A. Bordunov, N. Avdalovic, J. Chro-

matogr. A 2002, 956, 35.[35] A. Woodruff, C.A. Pohl, A. Bordunov, N. Avdalovic, J. Chro-

matogr. A 2003, 997, 33.[36] S. Rokushika, Z.Y. Qiu, Z.L. Sun, H. Hatano, J. Chromatogr.

1983, 280, 69.[37] S. Rokushika, D.Y. Huang, Z.Y. Qiu, H. Hatano, J. Chroma-

togr. 1985, 332, 15.[38] T. Takeuchi, T. Miwa, A. Hemmi, T. Maeda, Chromatogra-

phia 1993, 37, 173.[39] R.M. Cassidy, S. Elchuk,Anal. Chem. 1982, 54, 1558.[40] T. Takeuchi, E. Suzuki, D. Ishii, Chromatographia 1988, 25,

582.[41] S.W. Downey, G.M. Hieftje,Anal. Chim. Acta 1983, 153, 1.[42] K. Tanaka, J.S. Fritz,Anal. Chem. 1987, 59, 708.[43] H. Shintani, P.K. Dasgupta, Anal. Chem. 1987, 59, 1963.[44] T. Takeuchi, T. Miwa, A. Hemmi, T. Maeda, Chromatogra-

phia 1993, 37, 173.[45] J. Chu, R. Hu, T. Miwa, T. Takeuchi, Chromatographia

1995, 40, 379.[46] T. Takeuchi, R. Zein, E. Munaf, T. Miwa, J. Chromatogr. A

1996, 755, 37.[47] R. Zein, E. Munaf, T. Takeuchi, T. Miwa, Anal. Chim. Acta

1996, 335, 261.[48] E. Munaf, R. Zein, T. Takeuchi, T. Miwa, Chromatographia

1996, 43, 304.[49] E. Munaf, R. Zein, T. Takeuchi, T. Miwa, Anal. Chim. Acta

1996, 334, 39.[50] E. Munaf, R. Zein, T. Takeuchi, T. Miwa, Anal. Chim. Acta

1999, 379, 33.



[51] T. Takeuchi, Safni, T. Miwa, J. Chromatogr. A 1997, 789,201.

[52] W. Hu, H. Tao, H. Haraguchi,Anal. Chem. 1994, 66, 2514.

[53] T. Takeuchi, Safni, T. Miwa, Y. Hashimoto, H. Moriyama,Analusis 1998, 26, 61.

[54] T. Takeuchi, Safni, T. Miwa, Y. Hashimoto, H. Moriyama, J.Chromatogr. A 1998, 804, 79.

[55] T. Takeuchi, Safni, T. Miwa, Y. Hashimoto, H. Moriyama,Chromatographia 1999, 50, 70.

[56] F.F. Cantwell, S. Puon,Anal. Chem. 1979, 51, 623.

[57] P.K. Dasgupta, in Ion Chromatography, J.G. Tarter, Ed.,Marcel Dekker, 1987, pp 191–367.

[58] Safni, T. Takeuchi, T. Miwa, J. Chromatogr. B 2001, 753,409.

[59] T. Takeuchi, S. Inoue, T. Miwa, J. Microcol. Sep. 2000, 12,450.

[60] K. Soroka, R.S. Vithanage, D.A. Phillips, B. Walker, P.K.Dasgupta, Anal. Chem. 1987, 59, 629.

[61] D.A. Phillips, K. Soroka, R.S. Vithanage, P.K. Dasgupta,Mikrochim. Acta 1986, I, 207.

[62] P.K. Dasgupta, K. Soroka, R.S. Vithanage, J. Liq. Chroma-togr. 1987, 10, 3287.

[63] P.K. Dasgupta, I.-Y. Eom, K.J. Morris, J. Li, Anal. Chim.Acta 2003, 500, 337.

[64] P.K. Dasgupta, Z. Genfa, J. Li, C.B. Boring, S. Jambu-nathan, R. Al-Horr,Anal. Chem. 1999, 71, 1400.

[65] Y. Kitamaki, J.-Y. Jin, T. Takeuchi, J. Chromatogr. A 2003,1003, 197.

[66] W.Z. Hu, T. Takeuchi, H. Haraguchi, Anal. Sci. 1992, 8, 507.

[67] W. Hu, T. Takeuchi, H. Haraguchi, Anal. Chim. Acta 1992,267, 141.

[68] W.Z. Hu, H. Haraguchi, Bull. Chem. Soc. Jpn. 1993, 66,1420.

[69] W. Hu, T. Takeuchi, H. Haraguchi, Anal. Chem. 1993, 65,2204.

[70] W. Hu, H. Haraguchi, Anal. Chim. Acta 1994, 285, 335.

[71] W. Hu, H. Haraguchi, Anal. Chem. 1994, 66, 765.

[72] W. Hu, H. Haraguchi, Anal. Chim. Acta 1994, 289, 231.

[73] K. Bachmann, K.H. Steeg, T. Groh, A. Roder, I. Haumann,J. Boden, Intern. J. Anal. Chem. 1992, 49, 87.

[74] W. Buchberger, K. Haider, J. Chromatogr. A 1997, 770, 59.

[75] D. Kourilova, N.T.P. Thao, M. Krejci, Intern. J. Environ. Anal.Chem. 1987, 31, 183.

[76] K. Ito, J. Chromatogr. A 1999, 830, 211.

[77] L.W. Lim, J.-Y. Jin, T. Takeuchi,Anal. Sci. 2003, 19, 447.

[78] C.B. Boring, P.K. Dasgupta, A. Sjogren, J. Chromatogr. A1998, 804, 45.

[79] C.B. Boring, S.K. Poruthoor, P.K. Dasgupta, Talanta 1999,48, 674.

[80] Y. Liu, N. Avdalovic, V. Barreto, C.A. Pohl, Paper presentedat Pittcon 2004, Chicago, IL. Abstract No. 9200–500,March, 2004.

[81] D. Ishii, T. Takeuchi, J. Chromatogr. 1981, 218, 189.

[82] A. Manz,W. Simon, J. Chromatogr. Sci. 1983, 21, 326.

[83] S.R. Muller, W. Simon, H.M. Widmer, K. Grolimund, G.Schomburg, P. Kolla, Anal. Chem. 1989, 61, 2747.

[84] S.R. Muller, D. Scheidegger, C. Haber, W. Simon, J. HighRes. Chromatogr. 1991, 14, 174.

[85] D. Pyo, P.K. Dasgupta, L.S. Yengoyan,Anal. Sci. 1997, 13,185.

[86] D. Pyo, H. Kim, P.K. Dasgupta, B. Kor. Chem. Soc. 1999,20, 223.

[87] D. Pyo, H. Kim,Microchem. J. 2001, 70, 159.

[88] C.A. Pohl, B. Joyce, R. Kiser, D. Later, C. Saini, N. Shira-kawa, A. Woodruff, A New Hydroxide-Selective Column forthe Analysis of Disinfectant by Product Anions, AbstractNo. 9000–800, Pittcon 2004, Chicago, IL, March 2004.

[89] P. Kuban, P.K. Dasgupta, C.A. Pohl, unpublished work2004.

[90] www.chromolith.com.

[91] P. Hatsis, C.A. Lucy,Analyst 2002, 127, 451.

[92] P. Hatsis, C.A. Lucy,Anal. Chem. 2003, 75, 995.

[93] M. Motokawa, H. Kobayashi, N. Ishizuka, H. Minakuchi, K.Nakanishi, J. Jinnai, K. Hosoya, T. Ikegami, N. Tanaka, J.Chromatogr. A 2002, 961, 53.

[94] Q. Xu, K. Tanaka, M. Mori, M.I.H. Helaleh, W. Hu, K.Hasebe, H. Toada, J. Chromatogr. A 2003, 997, 183.

[95] A. Manz, Y. Miyahara, J. Miura, Y. Watanabe, H. Miyagi, K.Sato,Sens. Actuators B 1990, 1, 249.

[96] B. He, N. Tait, F. Regnier,Anal. Chem. 1998, 70, 3790.

[97] C. Ericson, J. Holm, T. Ericson, S. Hjerten, Anal. Chem.2000, 72, 81.

[98] Q. Kang, N.C. Golubovic, N.G. Pinto, H.T. Henderson,Chem. Eng. Sci. 2001, 56, 3409.

[99] J.P. Murrihy, M.C. Breadmore, A. Tan, M. McEnery, J.Alderman, C. O’Mathuna, A.P. O’Neill, P. O’Brien, N. Avda-lovic, P.R. Haddad, J.D. Glennon, J. Chromatogr. A 2001,924, 233.

[100] P.K. Dasgupta, L. Bao,Anal. Chem. 1993, 65, 1003–1011.

[101] N. Avdalovic, C.A. Pohl, R.D. Rocklin, J.R. Stillian, Anal.Chem. 1993, 65, 1470.

[102] P.K. Dasgupta, S. Kar,Anal. Chem. 1995, 67, 3853.

[103] S. Kar, P.K. Dasgupta, J. Microcol. Sep. 1996, 8, 561.

[104] S. Kar, P.K. Dasgupta, Am. Lab. 1997, 29(16), 17C.

[105] P. Kuban, P.K. Dasgupta, K.J. Morris, Anal. Chem. 2002,74, 5667.

[106] P. Kuban, J. Berg, P.K. Dasgupta, Anal. Chem. 2003, 75,3549.

[107] D. Qi, T. Okada, P.K. Dasgupta, Anal. Chem. 1989, 61,1383.

[108] W.R. Jones, J. Soglia, M. McGlynn, C. Haber, J. Reineck,C. Krstanovic,Am. Lab. 1996, 28(5), 25.

[109] X. Huang, T.-K. Pang, M.J. Gordon, R.N. Zare, Anal.Chem. 1987, 59, 2747.

[110] X. Huang, R.N. Zare, S. Sloss, A.G. Ewing, Anal. Chem.1991, 63, 189.

[111] X. Huang, R.N. Zare,Anal. Chem. 1991, 63, 2193.

[112] S. Kar, P.K. Dasguta, H. Liu, H. Hwang, Anal. Chem. 1994,66, 2537.

[113] S.R. Liu,Electrophoresis 2003, 24, 3755.

[114] J.A.F. da Silva, C.L. Do Lago,Anal. Chem. 1998, 70, 4339.

[115] E. Pungor, Oscillometry and Conductometry. PergamonPress, Oxford 1965.

[116] E. Pungor, F. Pal, K. Toth,Anal. Chem. 1983, 55, 1728.

[117] F. Pal, E. Pungor, E. Kovats,Anal. Chem. 1988, 60, 2254.

[118] A.J. Zemann, E. Schnell, D. Volgger, G.K. Bonn, Anal.Chem. 1998, 70, 563.



[119] E.F. Hilder, A.J. Zemann, M. Macka, P.R. Haddad, Electro-phoresis 2001, 22, 1273.

[120] J. Muzikar, T. van de Goor, B. Gas, P. Coufal, J. Chroma-togr. A 2001, 924, 147.

[121] P. Tuma, F. Opekar, I. Jelinek, Electroanalysis 2001, 13,989.

[122] J.A.F. da Silva, N. Guzman, C.L. do Lago, J. Chromatogr.A 2002, 942, 249.

[123] P. Kuban, B. Karlberg, P. Kuban, V. Kuban, J. Chromatogr.A 2002, 964, 227.

[124] J. Tanyanyiwa, B. Galliker, M.A. Schwarz, P.C. Hauser,Analyst 2002, 127, 214.

[125] A.J. Zemann,Electrophoresis 2003, 24, 2125.

[126] P. Kuban, M.A. Muri, P.C. Hauser,Analyst 2004, 129, 82.

[127] K. Mayrhofer, A.J. Zemann, E. Schnell, G.K. Bonn, Anal.Chem. 1999, 71, 3828.

[128] T. Kappes, B. Galliker, M.A. Schwarz, P.C. Hauser, TrAC2001, 20, 133.

[129] T. Chvojka, I. Jelinek, F. Opekar, K. Stulik, Anal. Chim Acta2001, 433, 13.


Documents

1441 h 1441. - Universidad De Antioquiaquimica.udea.edu.co/~carlopez/cromatoion/capillary_ionchroma_2005.pdf · the differences in ion exchange ... sup-pressed capillary IC was