AN

ALYST

FULL PA

PER

THE

www.rsc.org/analyst

Microchip separations of transition metal ions via LEDabsorbance detection of their PAR complexes

Qin Lua and Greg E. Collins*b

a GeoCenters, Inc., 1801 Rockville Pike, Suite 405, Rockville, MD 20852, USAb Naval Research Laboratory, 4555 Overlook Ave., S.W., Chemistry Division, Code 6116,

Washington, D.C. 20375-5342, USA

Received 20th October 2000, Accepted 5th February 2001First published as an Advance Article on the web 14th March 2001

Micellar electrokinetic chromatography was utilized in the electrophoretic separation of seven transition metalions, colorimetrically complexed by 4-(2-pyridylazo)resorcinol (PAR) on a glass capillary electrophoresismicrochip. Detection of the PAR metal chelates was demonstrated using a green light emitting diode (540 nm) anda miniature photomultiplier tube. Parameters investigated included the effect of buffer type, pH and surfactantconcentration (sodium dodecyl sulfate, SDS) on the separation efficiency. The optimally determined backgroundelectrolyte contained 10 mM ammonium phosphate buffer (pH 7.5), 1 mM PAR to prevent kinetic labilityproblems and 75 mM SDS for enhanced resolution. The separation of seven transition metal ions, Co2+, V3+, Ni2+,Cu2+, Fe2+, Mn2+ and Cd2+, was achievable in under 65 s, with the resolution of each metal ion in excess of 1.60.Detection limits obtained ranged from 400 ppb for Ni2+ to 1.2 ppm for Mn2+.

Introduction

Since the first introduction of microfabricated separationdevices,1,2 numerous applications have arisen in order tocapitalize on the potential advantages afforded the microchipwhen compared with conventional ‘benchtop’ separation sys-tems. These advantages include improvements in speed, cost,portability, automation and solvent/sample consumption.3,4

Bioanalytical applications utilizing microchip based deviceshave expanded rapidly to include DNA restriction digestion andsubsequent size-based separations,5 polymerase chain reactions(PCRs) with amplicon sizing,6 cell sorting and membrane lysisof selected cells,7 amino acid chirality,8 neurotransmitterdeterminations9 and enzyme assays,10 to name only a few.

In contrast to the vast number of bio-technological applica-tions being applied to microfabricated capillary electrophoresis(CE) devices, there are only two examples in the literature ofmetal ion separations being applied to a CE microchip, and bothrely on the fluorescent metal complexation behavior of8-hydroxyquinoline-5-sulfonic acid (HQS) for enabling detec-tion.11,12 This is surprising when we consider the impact thattraditional CE has had on the development of new techniquesfor the analytical separation of complex metal ion mixtures.Ramsey et al.’s investigations demonstrated the possibilitiesinherent in performing sensitive (ppb), high resolution separa-tions of Zn2+, Cd2+ and Al3+ (46, 57 and 30 ppb detectionlimits)11 and Mg2+ and Ca2+ ( 0.5 and 18 ppb detection limits)12

on extremely rapid timescales ( < 20 s) with a microchip baseddevice.

From the standpoint of developing an inexpensive, fieldportable monitor for metal ions, the chelating ligand, HQS, isless than ideal, however, owing to its UV excitation profile,which requires the use of an argon ion laser (351.1–363.8 nm).Alternatively, we have been interested in metal complexationdyes with significantly red-shifted absorbance wavelengths( > 500 nm), which will allow the incorporation of compact andinexpensive light sources, such as light emitting diodes (LEDs)and laser diodes, and detectors, such as photodiodes, into thesensor design.

Desirable parameters for a colorimetric metal chelate appli-cable to a microchip platform include a large bathochromic shift

by the metal chelate away from the ligand’s absorbancespectrum in order to minimize spectral overlap, maximumabsorbance by the metal chelate at a wavelength > 500 nm forreduced interferences and the inclusion of inexpensive lightsources, high molar absorptivity for enhanced sensitivity, broadmetal complexation capabilities for separating a range ofdifferent metal ions and, finally, a mechanism for allowing theCE separation of similarly charged metal complexes. The metalchelate 4-(2-pyridylazo)resorcinol (PAR) combines each ofthese different characteristics, including the sensitive andselective detection of various transition metal ions by micellarelectrokinetic chromatography (MEKC).13

Because the absorbance of its metal chelates directly overlapsthe emission from an inexpensive, compact green LED lightsource (l = 540 nm), we pursued the feasibility of utilizingPAR for transition metal ion separation and detection on a CEmicrochip that is comprised of truly portable, compact andinexpensive components. The benefits of adapting PAR to a CEmicrochip platform for the detection of metal cations areseveralfold: significantly improved sample analysis times, verysmall sample size requirements with minimal waste generation,sensitivities comparable to those of commercial instruments andthe potential for portable sensing.

Experimental

Microchip fabrication

The Borofloat glass microchips utilized in this study wereobtained from Micralyne, Inc. (Alberta, Canada). The chipformat consisted of a 16 3 95 3 2.2 mm substrate with channelplate and cover plate thicknesses of 1.1 mm. A simple crosspattern was utilized for the injector design, with a separationchannel length of 85.0 mm (80.0 mm from the intersection) andan injection channel length of 4.0 mm. Four access holes(2.0 mm diameter) were drilled through the cover plate in orderto define the different solution reservoirs. The microchips had atrapezoidal channel cross-section of 50 mm wide 3 20 mm deep.Micropipette tips (200 mL) were inserted into the buffer, sample

This journal is © The Royal Society of Chemistry 2001

DOI: 10.1039/b008595m Analyst, 2001, 126, 429–432 429

Publ

ishe

d on

14

Mar

ch 2

001.

Dow

nloa

ded

by C

olor

ado

Col

lege

on

30/1

0/20

14 2

3:18

:55.

View Article Online / Journal Homepage / Table of Contents for this issue

and sample waste reservoirs, while an HPLC fitting wasepoxied to the buffer waste reservoir to permit simpleattachment of the microchannels to a pressurized argon tank forhydrostatic conditioning and preparation of the microchip for agiven separation. The glass microchips were sequentiallypreconditioned at the beginning of each day with 0.1 M NaOH,de-ionized water (Milli-Q system; Millipore, Bedford, MA,USA) and then the appropriately buffered reagent solution.

High voltage switching hardware

The high voltage switching apparatus consisted of two 8 kVhigh voltage power supplies (Bertan, Model PMT-75C) and anarray of high voltage reed relays (Crydom, Model DAT71210)which were configured so as to permit remote switching via aLABView interface (National Instruments) between one ofthree different modes of microfluidic operation: separation,pinch load or floating load. Descriptions of these modes ofoperation have been delineated elsewhere.14

Optical microchip instrumentation

Colorimetric detection of the PAR metal chelates was accom-plished using the instrumental set-up shown in Fig. 1. A greenLED light source (540 nm, AND, Model AND520HG) wasoriented directly above the separation microchannel, as close asphysically possible to the buffer waste reservoir’s micropipettetip and the microchannel itself. Light passing through themicrochannel was collected using a microscope objective(Newport, 203) and directed through a perpendicularly ori-ented, 200 mm 3 3 mm rectangular slit on to a miniature, red-shifted photomultiplier tube (Hamamatsu, Model H5783-01).The PMT current was monitored using a Keithley Model 617programmable electrometer. PAR measurements in the electro-pherograms are given in terms of the PMT current and areproportional to the transmittance of the green LED light throughthe microchip separation channel.

Solutions and reagents

PAR and the surfactant sodium dodecyl sulfate (SDS) werepurchased from Aldrich (Milwaukee, WI, USA). Standardsolutions of metal ions (2.0 mM) were prepared from either1000 mg L21 atomic absorption standards or their nitrate salts(Sigma and Aldrich). Buffers were prepared from sodiumtetraborate (borax, Na2B4O7), sodium phosphate monobasic(NaH2PO4), sodium phosphate dibasic (Na2HPO4), ammoniumdihydrogenphosphate [(NH4)H2PO4], ammonium hydrogen-phosphate [(NH4)2HPO4], tris(hydroxymethyl)aminomethane

(TRIS) or N-[tris(hydroxymethyl)methyl]glycine (Tricine).Pre-capillary ligand–metal complexation was achieved byadding the appropriate concentration of metal ion to a buffered1 mM PAR solution. NaOH or HCl was used to readjust the pHof the sample solutions, after which the solutions were filteredthrough a 2 mm membrane filter (ChromTech, Apple Valley,MN, USA), sonicated for 5 min and then introduced into thesample reservoir of the microchip for analysis.

Results and discussion

Fig. 2 shows the spectra for PAR and its metallochromiccomplexes formed with various transition metal ions. From anoptical detection standpoint, PAR benefits from the capabilityfor chelating a number of different transition and heavy metalcations with high molar absorptivities (e.g., Co–PAR, e = 4.13 104 L mol21 cm21 at 540 nm) and large bathochromic shifts(e.g., Co–PAR, D = 95 nm). The large red shifts evident in theabsorbtion spectrum of the metal chelates of PAR are a criticalelement to its successful implementation on the microchip,owing to the importance that PAR be added directly to thebackground electrolyte (BGE) in addition to the analyte samplesolution. The kinetic lability of many strong metal complexa-tion agents such as PAR is a common problem encountered inthe CE analysis of metals.15 Despite large thermodynamicstability constants for metal complexation, many metal com-plexes will dissociate during the course of the separation,rendering their detection impossible when the ligand and metalion are isolated. This problem is typically dealt with via theaddition of the chelating ligand to the background electrolyte ofthe buffer and also the sample pre-complexing solution,adjusting the equilibrium in favor of the metal complex. Theoptimum concentration of chelating ligand in the buffernecessary to attain the highest level of resolution and sensitivitywas concluded to be 1 mM. In order to prevent high backgroundlevels associated with absorption of the uncomplexed PARligand, a 540 nm LED light source was utilized, as indicated inFig. 2.

From pH 6 to 11, the o-phenol of PAR deprotonates to givea univalent anion (HL2) which, in general, complexes divalentmetal cations in a terdentate fashion with a 2+1 ligand to metalratio.16 This behavior leaves the metal complexes uncharged

Fig. 1 CE microchip instrumental set-up for monitoring the separation ofmetal PAR chelates.

Fig. 2 Absorbance versus wavelength spectra for (a) 0.075 mM PAR, (b)0.024 mM PAR–Ni2+, (c) PAR–Co2+ and (d) PAR–Cu2+. The emissionwavelength maximum of the green LED light source is indicated by thedotted line at 540 nm.

430 Analyst, 2001, 126, 429–432

Publ

ishe

d on

14

Mar

ch 2

001.

Dow

nloa

ded

by C

olor

ado

Col

lege

on

30/1

0/20

14 2

3:18

:55.

View Article Online

and seemingly inseparable by CE. Iki et al., however,demonstrated the separation of four metal ions using PAR, andpostulated that the pKa of the p-phenolic group of PAR isinfluenced by the central metal ion, thereby allowing theirseparation by CE.17 The CE microchip showed similar results tothose of Iki et al. [Fig. 3(a)], with the exception that theseparation time for successfully resolving the four metal ions(Co2+, Cu2+, V3+ and Cd2+) was reduced from 6 min to justunder 20 s. Unfortunately, the introduction of additional metalions (Ni2+, Fe2+ and Mn2+), resulted in overlap and poorlyresolved peaks.

Improvements in the number of metal cations simultaneouslydetermined have been achieved in many cases through theapplication of MEKC.18 MEKC permits the separation of bothneutral and charged species via the addition of surfactants to theBGE, resulting in the formation of charged micelles that can beseparated by the applied field. MEKC, for example, has beensuccessfully applied to PAR CE separations, increasing thenumber of metal cations simultaneously determined from fourto nine.13 In an effort to improve the resolution on the CEmicrochip, the MEKC method of Timerbaev et al. wasattempted.13

Fig. 3 shows the effect of increasing the surfactant (SDS)concentration on PAR’s capability for resolving seven transi-tion metal ions on a CE microchip. Continual enhancements inthe resolution were evident as the concentration of SDS wasincreased from 0 to 75 mM. Fig. 3(d) demonstrates the completeseparation of seven metal ions, albeit at slightly longer retentiontimes compared with the BGE containing no surfactant. There issome broadening of the peak shapes evident at higher levels ofSDS, but the peak separations are sufficiently large to make75 mM SDS the optimum concentration. Higher concentrationsof SDS were not pursued in order to prevent the current withinthe microchannels from rising above 30 mA, where Jouleheating effects can seriously impair the resolution.

The type of buffer utilized was found to play a critical role indetermining the quality of the electropherogram obtained.Because of the high concentration of SDS necessary in the BGEto improve the separation efficiency (75 mM), buffer selectionwas limited to larger, minimally charged ions with lowermobility and, hence, minimal current generation in the bufferedpH range optimal for PAR colorimetric metal chelation (pH 7–

9). Initial investigations of the biological buffers TRIS andTricine in the absence of SDS gave electropherograms withbroad, overlapping peak shapes and poor resolution of the sevenmetal ion mixture.

Although sodium phosphate buffers were eliminated owingto the generation of excess current within the microchannels ofthe microchip, ammonium phosphate gave narrow peaks in theabsence of SDS [despite overlapping retention times for severalmetal ions, see Fig. 3(a)], and low, microchannel currentsamenable to the addition of 75 mM SDS. Sodium tetraborategave similar peak quality and current generation properties, andso the two buffer systems were compared on the microchip withrespect to the separation of seven transition metal ions in a BGEcontaining 75 mM SDS (pH 8.0). Although the sodiumtetraborate BGE was successful in separating six of the sevenmetal ions, vanadium and cobalt were poorly resolved and thecopper and iron peaks directly overlapped. The ammoniumphosphate BGE, on the other hand, gave reasonable separationefficiencies for all seven transition metal ions, making this ourbuffer system of choice. Improved ion pairing between themicellar PAR metal chelates and the cationic component of theelectrophoretic buffer may explain the resolution enhancementsseen in the electropherogram.

The pH of the electrophoretic buffer must be carefullycontrolled, as it not only influences the electroosmotic flow(EOF), but also the acid dissociation equilibria of the PARmetal complexes, ultimately affecting the sensitivity andresolution. Efforts to optimize the separation BGE includedvarying the pH of a 10 mM ammonium phosphate buffercontaining 75 mM SDS and 1 mM PAR. Fig. 4 shows threeelectropherograms recorded at pH 7.0, 7.5 and 8.0. At pH valuesof 7.0 and lower, the quality of the electropherograms wasseriously impaired owing to overlapping and broad peaks. Twoobservations can be made concerning the pH 7.0 electrophero-gram [Fig. 4(a)]: (1) the Ni(II)–, Cu(II)– and Mn(II)–PARcomplexes exhibited significantly longer migration times thanmight be expected from the drop in EOF which occurs withdecreasing pH, and (2) the Ni(II)– and Fe(II)–PAR complexesexhibited a significant decrease in signal intensity. Examinationof the pKa values determined for various PAR metal chelates17

indicates that both of the aforementioned effects can probably

Fig. 3 Effect of SDS on the separation efficiency of seven PAR metalchelates: 1 = 1.8 ppm Co2+; 2 = 3.0 ppm V3+; 3 = 1.8 ppm Ni2+; 4 = 3.0ppm Cu2+; 5 = 3.0 ppm Fe2+; 6 = 3.0 ppm Mn2+; and 7 = 1.8 ppm Cd2+.(a) 0; (b) 25; (c) 50; and (d) 75 mM SDS. BGE consisted of 10 mM(NH4)2HPO4–NH4H2PO4 (pH 7.5) and 1 mM PAR; separation voltage,+8410 V/85 cm; injection, 40 s floating load.

Fig. 4 Effect of pH on the separation efficiency of seven PAR metalchelates: 1 = 1.8 ppm Co2+; 2 = 3.0 ppm V3+; 3 = 1.8 ppm Ni2+; 4 = 3.0ppm Cu2+; 5 = 3.0 ppm Fe2+; 6 = 3.0 ppm Mn2+; and 7 = 1.8 ppm Cd2+.pH = (a) 7.0; (b) 7.5; and (c) 8.0. The BGE consisted of 10 mM(NH4)2HPO4–NH4H2PO4 (pH 7.5) and 1 mM PAR; separation voltage,+8410 V/85 cm; injection, 40 s floating load.

Analyst, 2001, 126, 429–432 431

Publ

ishe

d on

14

Mar

ch 2

001.

Dow

nloa

ded

by C

olor

ado

Col

lege

on

30/1

0/20

14 2

3:18

:55.

View Article Online

be attributed to protonation of either the o- or p-hydroxy groupof the PAR chelate, giving weaker metal–PAR complexes and/or causing significant changes to the electrophoretic mobilitiesof the various metal complexes. As the pH was increased to!8.0, the increase in EOF resulted in faster retention times forall of the metal ions examined, a factor ultimately limiting theresolution due to insufficient time within the microchannel foran efficient separation to take place. The optimum pH selectedwas pH 7.5.

Fig. 4(b) demonstrates the separation of seven metal cations(Co2+, V3+, Ni2+, Cu2+, Fe2+, Mn2+ and Cd2+) on a CEmicrochip utilizing the optimized BGE of 10 mM ammoniumphosphate (pH 7.5), 75 mM SDS and 1 mM PAR. Comparingthe retention times with those obtained by Timerbaev et al.13 ona commercial CE instrument, the metal cations were detected 20times faster, with the first metal cation complex, cobalt, elutingafter only 26 s in comparison with the retention time of 6 minobtained by Timerbaev et al.13 Table 1 summarizes theseparation characteristics attained on the microchip. Despite theshort retention times, the resolution of each metal ion pair was> 1.60 for all cases, indicative of the separation efficienciesachievable on these small microchip devices. Theoreticaldetection limits for the seven metal cations examined rangedfrom 400 ppb for nickel to 1.2 ppm for manganese, based on asignal-to-noise ratio of 3+1. Of the five metal ions separatedpreviously by Ramsey et al. using a capillary electrophoresismicrochip and HQS fluorescence detection,11,12 only Cd(II) canbe compared directly, giving a detection limit of 57 ppb, a factorof 10 better than the detection limit reported here via LEDabsorbance detection.

Conclusions

We have demonstrated that extremely rapid, high resolutionseparations of multiple metal ions are achievable on glass

microchips when done in conjunction with MEKC and strong,metal complexing ligands such as PAR. Despite the smallmicrochannel depths of CE microchips, sub-ppm detectionlimits can be obtained for metal ions when utilizing a compact,inexpensive LED light source and a miniature PMT detectoroperable via a 15 V power supply.

More importantly, each of the elements necessary forpackaging a truly portable sensor system for metal ions hasproven viable, and efforts are currently under way to engineer afield portable metal ion sensor. Although the current sensitivityof this microchip device limits its applicability to ppm levelscreening of metal ion contaminants (e.g., hot spots), we areseeking methods for lowering the detection limits through theapplication of deeper microchannels (longer light pathlengths),stabilization of the LED light source and signal averagingsoftware routines for improving the signal-to-noise ratios.

Acknowledgement

The authors gratefully acknowledge the Environmental Man-agement Science Program of the Department of Energy forfinancial support of this program.

References

1 A. Manz, J. C. Fettinger, E. Verpoorte, H. Lüdi, H. M. Widmer andD. J. Harrison, Trends Anal. Chem., 1991, 10, 144.

2 A. Manz, D. J. Harrison, E. Verpoote, J. C. Fettinger, A. Paulus, H.Lüdi and H. M. Widmer, J. Chromatogr., 1992, 593, 253.

3 A. Manz, D. J. Harrison, E. Verpoote and H. M. Widmer, Adv.Chromatogr., 1993, 33, 1.

4 J. M. Ramsey, S. C. Jacobson and M. R. Knapp, Nature Med., 1995,1, 1093.

5 S. C. Jacobson and J. M. Ramsey, Anal. Chem., 1996, 68, 720.6 A. T. Woolley, D. Hadley, P. Landre, A. J. deMello, R. A. Mathies

and M. A. Northrup, Anal. Chem., 1996, 68, 4081.7 P. C. H. Li and D. J. Harrison, Anal. Chem., 1997, 69, 1564.8 L. D. Hutt, D. P. Glavin, J. L. Bada and R. A. Mathies, Anal. Chem.,

1999, 71, 4000.9 A. T. Woolley, K. Lao, A. N. Glazer and R. A. Mathies, Anal. Chem.,

1998, 70, 684.10 A. G. Hadd, D. E. Raymond, J. W. Halliwell, S. C. Jacobson and J. M.

Ramsey, Anal. Chem., 1997, 69, 3407.11 S. C. Jacobson, A. W. Moore and J. M. Ramsey, Anal. Chem., 1995,

67, 2059.12 J. P. Kutter, R. S. Ramsey, S. C. Jacobson and J. M. Ramsey,

J. Microcol. Sep., 1998, 10, 313.13 A. R. Timerbaev, O. P. Semenova, P. Jandik and G. K. Bonn,

J. Chromatogr. A, 1994, 671, 419.14 S. J. Haswell, Analyst, 1997, 122, 1R.15 M. Macka and P. R. Haddad, Electrophoresis, 1997, 18, 2482.16 CRC Handbook of Organic Analytical Reagents, ed. K. L. Cheng, K.

Ueno and T. Imamura, CRC Press, Boca Raton, FL, 1992, p. 195.17 N. Iki, H. Hoshino and T. Yotsuyanagi, Chem. Lett., 1993, 701.18 P. R. Haddad, M. Macka, E. F. Hilder and D. P. Bogan,

J. Chromatogr. A, 1997, 780, 329.

Table 1 Microchip separation characteristics for individual metal ions

Metal ionTheoreticalplates Resolution

Detectionlimit (ppm)

Co 7 900 0.471.99

V 3 500 0.971.62

Ni 4 000 0.402.37

Cu 10 400 0.413.59

Fe 11 600 1.001.92

Mn 4 600 1.154.21

Cd 5 100 0.54

432 Analyst, 2001, 126, 429–432

Publ

ishe

d on

14

Mar

ch 2

001.

Dow

nloa

ded

by C

olor

ado

Col

lege

on

30/1

0/20

14 2

3:18

:55.

View Article Online