816 Volume 55, Number 7, 2001 APPLIED SPECTROSCOPY0003-7028 / 01 / 5507-0816$2.00 / 0q 2001 Society for Applied Spectroscopy

Direct Determination of Metals in Soils and Sediments byInduction Heating-Electrothermal Vaporization (IH-ETV)Inductively Coupled Plasma-Optical Emission Spectrometry(ICP-OES)*

MICHAEL E. RYBAK† and ERIC D. SALIN‡Department of Chemistry, McGill University, 801 Sherbrooke St. W., Montreal, Quebec, Canada H3A 2K6

The application of an induction heating (IH) electrothermal vapor-ization (ETV) sample introduction arrangement for the determi-nation of As, Cd, Cu, Mn, Pb, and Zn in soils and sediments byinductively coupled plasma-optical emission spectrometry (ICP-OES) is presented. Samples were deposited either directly as a solidor by means of slurry sampling into graphite cups that were thenpositioned in a radio-frequency (RF)-� eld and vaporized in a car-rier � ow of 15% (v/v) SF6–Ar. Four certi� ed reference materials(CRMs) were examined: two soil samples—SRM 2710 and SRM2711 (NIST); and two marine sediments—MESS-2 and PACS-2(NRC Canada). In general, sample delivery was simpler and ob-served signal precision was better with slurry sampling when com-pared to the analysis of the solid directly, with peak area RSDsranging from 4–16% (n 5 6). Plots of intensity vs. certi� ed concen-tration for the four CRMs were linear with log-log slopes of 0.98–1.02 and r 2 values $ 0.995 for As, Cu, Pb, and Zn. Recoveries of80–105% were achieved for the above elements in SRM 2711 byusing an external standards curve constructed from the 3 remainingCRMs. Aqueous standard solutions were used for the analysis of all4 CRMs by standard additions, resulting in recoveries ranging from54–139% (average recovery of (101 6 15)%) across all six deter-mined elements in all four samples.

Index Headings: Soil analysis; Sediment analysis; Trace metals;Electrothermal vaporization; Inductively coupled plasma-opticalemission spectrometry; Induction heating; Sample introduction.

INTRODUCTION

The release of metals from various human activitiesinevitably results in their distribution throughout the en-vironment, with soils and sediments functioning as animportant sink for their accumulation. Common examplesof anthropogenic activities that introduce trace elementsinto soils include soil-application practices (e.g., the ap-plication of fertilizers, agrochemicals, sewage sludgesand ef� uents, and manures) as well as irrigation activitiesand general atmospheric deposition.1 Increased frequencyand magnitude of these activities naturally increases thetrace-element circulation in the environment and resultsnot only in an increased accumulation of trace elementsin soils, but ultimately in a build-up of metals in thehuman food chain. With the effects on human health from

Received 22 July 2000; accepted 14 March 2001.* Portions of this paper were presented at the 1999 Annual Meeting of

the Federation of Analytical Chemistry and Spectroscopy Societies(FACSS 99) in Vancouver, BC, Canada.

† Present address: United States Department of Agriculture, Agricultur-al Research Service, Beltsville Human Nutrition Research Center,Food Composition Laboratory, Building 161, BARC-East, Beltsville,Maryland 20705, USA.

‡ Author to whom correspondence should be sent.

the consumption of these elements often being vague andretrospective, the monitoring of soils serves as a pro-spective indicator of trace element consumption, as wellas a retrospective indicator of anthropogenic activity.1Since the chemical form of a trace element often bears adirect consequence on its bioavailability and toxicity, an-alytical methodologies that are capable of speciation, i.e.,isolating and identifying the chemical form of an analyte,are of considerable importance in the determination ofmetals in environmental samples.2 Nonetheless, the de-termination of the total trace element concentration insoils and sediments is still quite valuable in an absolutesense for establishing background levels and identifyingoccurrences of metal contamination.3

With the existence of well-established instrumentaltechnologies such as atomic absorption spectrometry(AAS), X-ray � uoresence (XRF), inductively coupledplasma-optical emission spectrometry (ICP-OES), ICP-mass spectrometry (ICP-MS), glow discharge-MS (GD-MS), and neutron activation analysis (NAA), there is cer-tainly no shortage of techniques that are capable of de-termining the total concentration of metals in samplessuch as soils and sediments. What often precludes theirroutine application are shortcomings in terms of meth-odological requirements. Examples of these include: lackof simultaneous multi-e lement analysis capabilities(AAS); poor analyte sensitivity (XRF); low user preva-lence (GD-MS, NAA); and/or high operating costs(NAA). These issues considered, with the ability to per-form simultaneous or rapid sequential multi-elementanalyses at low ppb range detection limits with relativelyfew interferences, and moderate capital and operatingcosts, ICP-OES presents itself as a well-suited candidatefor the routine analysis of many environmental sampletypes including soils and sediments.4

The most common technique for introducing samplesin ICP-OES is solution nebulization (SN), and though theSN interface is convenient for samples that can be pre-sented in either an existing or an easily generated liquidform, solid samples must typically be subjected to a sam-ple dissolution/digestion step5 to facilitate their introduc-tion. These digestion techniques are well-characterized,widely-used, and recognized by environmental agencies,6

and the advent of microwave-assisted digestion meth-ods7,8 has improved the ease and reliability with whichthey can be conducted and has reduced concerns regard-ing contamination. The conversion of a solid sample toliquid format, however, does present an inconvenience,provides the potential for error, and inevitably results in

APPLIED SPECTROSCOPY 817

FIG. 1. Schematic depiction of the IH-ETV sample introduction sys-tem (not to scale).

sample dilution. As an alternative approach, a thermalsample introduction method such as direct sample inser-tion (DSI)9 or electrothermal vaporization (ETV)10 can beused as a means of analyzing samples such as soils orsediments directly with no or minimal prior sample treat-ment. With these approaches, a solid sample is placed ona surface that is subsequently heated, allowing the va-porized analytes to be swept into the plasma as a vaporor dry aerosol plume, resulting in a transient emissionsignal that is proportional to the amount of vaporizedanalyte. Granted, the analysis of a relatively smallamount (on the order of mg) of a solid sample does raiseconcerns in terms of sample homogeneity, however, ju-dicious sampling practices and rigorous processing (i.e.,pulverizing, mixing) procedures can effectively mitigatethese concerns. Furthermore, the development and appli-cation of slurry sampling, in which a solid is preparedsimply as a � ne, uniform particulate suspension in liquidmedia and can be volumetrically procured and delivered,has presented a means by which solid samples can besystematically and reproducibly delivered into thermalsample introduction systems.11

There exist several examples where soils and sedi-ments have been analyzed directly by using ETV.12–14 Re-cently, a variant of ETV was described in which a radio-frequency (RF) � eld-induced current was used to heat anelectrically conductive vaporization surface15–17 ratherthan an applied direct current (DC), as is commonly thecase in conventional ETV systems. This induction heat-ed-ETV (IH-ETV) arrangement15 is interesting in that itmelds a highly controllable, ex situ thermal vaporizationevent analogous to that achieved in conventional ETVwith an interchangeable graphite cup sample holder in acontact-free heating environment not unlike that encoun-tered in DSI. This type of arrangement lends itself wellto processes that would bene� t from batch sample pro-cessing and high sample throughput, as samples can beprepared in individual probes, subjected to pretreatmentoff-line, and then be vaporized in rapid succession. An-other characteristic of IH-ETV sample introduction is theroutine use of mixed carrier gases, such as SF6–Ar, as ameans towards improving analyte volatilization andtransport while at the same time preventing electrical dis-charging in the RF-� eld due to thermionic electron emis-sion from the graphite sample probe.

The objective of this study was to evaluate the utilityof IH-ETV sample introduction with ICP-OES as a gen-eral survey technique for metals in soil and sedimentsamples. Design aspects and operational considerationsthat are unique to IH-ETV sample introduction were ex-amined in terms of their bearing on overall experimentaldesign and implementation, as well as overall analyticalperformance using different calibration methodologies.

EXPERIMENTAL

IH-ETV and ICP-OES Instruments. The IH-ETV ar-rangement used in this study was a system designed andconstructed in-house and has been described previously.15

The system is depicted schematically in Fig. 1. In brief,a long, undercut graphite cup sample probe is positionedaxially in a quartz cylinder partially encircled by an RF-generator load coil such that the cup-portion of the probe

is centered in the coil. Application of forward power tothe load coil causes the cup portion to heat inductivelywith the temperature achieved being proportional to thepower applied. The heating of the surface permits thevaporization of analytes from the cup, and a carrier-gas� ow from the lower end of the cylinder sweeps the va-porized analyte plume through a transport-tubing networkinto the injector of the ICP torch. The vaporized analytecloud is passed through the center channel of the plasmadischarge where element-speci� c transient emission sig-nals are generated, optically isolated, and recorded by anICP-OES instrument. The temperature of the vaporizationsurface is monitored by using an infrared thermocouplepositioned above the quartz chamber at an angle of 308off the probe axis.

In this study, the IH-ETV was operated with a carriergas mixture of 15% (v/v) SF6–Ar. Sulfur hexa� uoride(SF 6) has been demonstrated as an effective means ofimproving the volatilization and transport of analytes inIH–ETV sample introduction through the formation ofvolatile metal � uorides by reaction with its thermal de-composition products (e.g., SF4, F·).15 In addition to in-creased analyte volatility, SF6 has also been shown toimprove the overall transport ef� ciency of analytes dueto a mixture of chemical and physical transport phenom-ena.15

Samples in this study were vaporized from probes ma-chined from cylindrical (2 in. (508 mm) length 3 3/8 in.(9.525 mm) diameter) high-density graphite rod stock(HLM grade, SGL Carbon Group, Speer Canada Inc., St.-Laurent, QC, Canada) to the dimensions appearing in thecutaway diagram in Fig. 2. The volume of the cup portionof the sample probe is approximately 480 mL, and canbe heated to a maximum temperature of approximately2000 8C (time constant of (2.0 6 0.2) s) with the currentinduction furnace arrangement.15 A bench-top lathe(Emco Compact 5, Emco Maier, Columbus, OH) andhigh-speed stainless-steel tools were used for the ma-chining process (described in detail elsewhere).18 Vapor-ized samples were introduced into a radially-viewed ICP-OES with echelle optics and charge-injection device(CID) detection (TJA IRIS, TJA Solutions, Franklin,

818 Volume 55, Number 7, 2001

FIG. 2. Dimensions of the graphite cup sample probe used. Imperialunits denote the dimensions of the tools used for machining (except forthe base of the probe).

TABLE I. Instrumental settings.

Sample vaporization step (IH-ETV)-Carrier gasCarrier gas � ow rateDrying stepPre-vaporization � ush timeVaporization temperatureVaporization time

15% (v/v) SF6–Ar0.50 L min21 a

Ex-situ30 sø1900 8C60 s

ICP-OES (TJA IRIS—radial view, Echelle spectrometer, CID detec-tion)Plasma forward powerAr plasma gas � ow rateAr auxiliary � ow rateAr injector � ow rateWavelengths monitored (nm)

1150 W15.0 L min21

1.0 L min21

0.50 L min21 a

As: 189.042, Cd: 214.438,Cu: 324.754, Mn:257.610, Pb: 220.353, Zn:213.856

Signal increment (% full well capacity)Number of time slicesTime per slice

75%3000.2 s

a Total � ow for gaseous mixture.

MA). Transient analyte emission signals were processedby using the IRIS operating software (ThermoSPEC/CIDv2.10, TJA Solutions). The area of the transient emissionsignal was used throughout the study as the dependentcalibration variable. Details regarding the vaporizationsequence and instrumental settings for both the IH-ETVand ICP-OES appear in Table I.

Samples, Standards, and Reagents. For the prepa-ration of samples and standards, distilled deionized water(DDW) (Milli-Q Water System, Millipore Corp., Bed-ford, MA) and trace metal grade acids (Instra-Analyzed,JT Baker, Phillipsburg, NJ) were used throughout, withACS reagent grade acids (Baker-Analyzed, JT Baker) andDDW being used for the cleaning and conditioning ofsample containers. Four certi� ed reference materials(CRMs) were used: two soil samples (SRM 2710 (Mon-tana Soil–Highly Elevated Traces) and SRM 2711 (Mon-tana Soil–Moderately Elevated Traces), National Instituteof Standards and Technology (NIST), Gaithersburg,MD); and two marine sediment samples (MESS-2 (Beau-fort Sea Marine Sediment) and PACS-2 (Esquimalt, BC(harbor) Marine Sediment), Institute for National Mea-surement Standards, National Research Council of Can-ada (NRC), Ottawa, ON, Canada). The soil and marinesediment CRMs have maximum particle diameters of 74mm and 125 mm, respectively, and their certi� ed valuesand corresponding uncertainties (expressed as 95% con-� dence limit) are valid for sub-samples $ 250 mg. Slur-ries were prepared by mixing an accurately (gravimetri-cally) determined amount of CRM with an accuratelyknown amount of DDW. The proportion of CRM toDDW in the slurries ranged from 1:10 to 1:100 (m/m)

and were prepared in volumes of DDW of at least 30 ml.Slurries were stored in polypropylene containers (Nal-gene) that had been preconditioned for 24 h with 1:1 (v/v) HNO3 in DDW and rinsed thoroughly with DDW. Toensure the homogeneity of the sample being drawn, slur-ries were actively stirred during the sampling step by us-ing a cleaned and preconditioned Te� ony stirring bar anda magnetic stirrer. Slurries were sampled with the use ofa 10–100 mL micropipette (Eppendorf) with trace-metalgrade disposable tips (Fisher Scienti� c, Nepean, ON,Canada). Weighed multi-element standard solutions wereprepared by mixture and serial dilution of 1000 ppm sin-gle-element standards (SCP Science, Baie-d’Urfe, QC,Canada) into a � nal volume of DDW containing 0.5%trace metal grade HNO3. Standard solutions were alsostored in preconditioned polypropylene containers.

RESULTS AND DISCUSSION

General Observations. The temporal signal behaviorobserved with IH-ETV sample introduction is identicalto that observed with conventional ETV in that a tran-sient peak results. Figure 3 shows the elemental emissionsignals typically observed for the vaporization of a soilor sediment sample, using SRM 2711 as an example.Notwithstanding differences in peak appearance (i.e.,height, half-width, area) that are directly related to theamount of analyte being vaporized, analogous temporalsignal behavior was observed for all elements except ar-senic. The broader, tailing signals observed with As werea consequence of increased analyte diffusion resultingfrom the formation of volatile AsF5 due to � uorinationby the SF6–Ar carrier.15 In ETV, analyte is usually trans-ported to the plasma as a � ne aerosol of solid particlesthat form soon after vaporization. Presuming that thismechanism of transport was taking place for the remain-ing elements, one can assume that the larger and heavieraerosol particles would not diffuse as rapidly duringtransport in the carrier gas as smaller, lighter gas mole-cules. While this may result in a broader transient signal,the formation of a non-condensing AsF5 species does im-

APPLIED SPECTROSCOPY 819

FIG. 3. Typical IH-ETV-ICP-OES analyte emission signals for a soilsample (SRM 2711). Approximately 1 mg of sample, deposited as anaqueous slurry (100 mL), was vaporized with 50 mL of concentratedHCl.

FIG. 4. Log-log calibration plot for the CRMs studied (n 5 6). In eachcase, 1–10 mg of sample, deposited as an aqueous slurry (100 mL), wasvaporized with 50 mL of concentrated HCl.

part some bene� ts, such as high transport ef� ciency (onthe order of 90%) and low detection limits.15

Although element- and sample-speci� c trends in theobserved precision values were not easily discernible, thepeak area reproducibility did appear to generally improvewhen either the amount of analyte deposited approached1 mg or the amount of CRM deposited approached 5 mg,after which no signi� cant improvement in precision wasobserved. The addition of an aliquot (50 mL) of concen-trated HCl to the sample prior to vaporization was foundto generally improve the reproducibility of the transientemission signals, although no signi� cant increase in sig-nal intensity was observed. Under these conditions andslurry sampling (100 mL of sample containing 10–100mg ml21 of sample), the precision of the transient emis-sion signal peak area ranged from 4 to 15% RSD (n 56) depending on the element and the CRM being ana-lyzed. In terms of general sampling strategies, both theanalysis of the solid directly and as an aqueous slurrywas attempted. It was immediately apparent, however,that analyzing the solid directly was quite labor-intensiveand prone to errors, as it involved the practice of havingto accurately and reproducibly deposit a mass of solidsample on the order of 1–10 mg directly into the sampleprobe for each vaporization event. This was re� ected inthe fact that the reproducibility of the transient signalpeak areas was considerably poorer (7–37% RSD, n 56) when direct analysis of the solid was performed. Bycomparison, slurry sampling provided a simple and re-producible means by which such small quantities of sam-ple could be transferred into the sample probe with noweighing step involved, and the use of a relatively largeamount of sample in preparing the slurry helped reducesample homogeneity and weighing errors.

Calibration and Determination. For matters of con-venience, calibration was � rst attempted with aqueous ex-ternal standards. The relatively clean matrix of these stan-dards, however, did not approximate well the complex

composition of the soil and sediment samples and re-sulted in cases where equivalent standard and sample an-alyte concentrations would yield signals with markedlydifferent peak areas, thus degrading the calibration meth-od. As an alternative approach, an attempt was made atusing the CRMs as matrix standards. Figure 4 shows thelog-log curve that results when peak area (blank subtract-ed) is plotted as a function of certi� ed concentration forthe CRMs studied. For As, Cu, Pb, and Zn, linear curveswith log-log slopes of 0.98–1.02 were achieved with r 2

values $ 0.995. The curves for Cd and Mn were alsoreproducible; however, their log-log slopes deviated no-ticeably from unity (0.81 and 0.93, respectively). Theconsiderable deviation in the log-log slope for Cd issomewhat surprising; however, it should be noted that theCd levels in the CRMs were the lowest among the ele-ments studied in each case, and the transient emissionsignals for Cd in some cases had the poorest reproduc-ibility among all of the elements and CRM studied, fac-tors that together may have contributed to the deviationin the log-log slope observed.

To demonstrate the approach of using the CRMs as anexternal standards curve, a calibration was performed todetermine the concentration of As, Cu, Pb, and Zn inSRM 2711 using a linear regression based on the peakarea vs. certi� ed concentration values for MESS-2,PACS-2, and SRM 2710. The selection of SRM 2711 asthe unknown for this analysis was made because its cer-ti� ed analyte concentrations fell towards the middle ofthe concentration ranges for the CRMs studied. In eachcase, 100 mL of slurry and 50 mL of concentrated HClwere deposited in a sample probe, dried off-line, and thenvaporized under full power in the IH-ETV. The grossamount of sample used in preparing the slurries for thiscalibration was adjusted such that the net amount depos-ited in 100 mL covered a concentration range of approx-imately one decade with the concentrations of the se-lected analytes in the unknown (SRM 2711) falling asclose as possible to the median of this range. The resultsof this calibration appear in Table II, reported at the 95%

820 Volume 55, Number 7, 2001

TABLE II. Concentrations of metals determined in SRM 2711(NIST) using CRMsa as external standards (slurry sampling).

Element

Concentration found/mg g21

IH-ETV-ICP-OES Certi� ed value % recovery

AsCuPbZn

83.6 6 6113 6 18968 6 44368 6 11

105 6 8114 6 2

1162 6 31350.4 6 4.8

809983

105

% recovery(average)

92 6 19

a CRMs used in calibration curve: SRM 2710, MESS-2, PACS-2.

TABLE III. Concentrations of metals determined in CRMs usingstandard additions with aqueous standards (slurry sampling).

SampleEle-ment

Concentration found/mg g21

IH-ETV-ICP-OES Certi� ed value % recovery

MESS-2 AsCdCuMnPbZn

22.6 6 4.20.13 6 0.0531.1 6 9.3292 6 1424.6 6 3.1190 6 16

20.7 6 0.240.24 6 0.0139.3 6 2365 6 2121.9 6 1.2172 6 16

109547980

112110

PACS-2 AsCdCuMnPbZn

26.7 6 8.12.2 6 0.6240 6 40354 6 5181 6 45391 6 87

26.2 6 1.52.11 6 0.15310 6 12440 6 19183 6 8364 6 23

102104778099

107

SRM 2710 AsCdCuMnPbZn

868 6 7327.2 6 3.53020 6 260

10800 6 15005970 6 5206860 6 710

626 6 3821.8 6 0.22950 6 130

10100 6 4005532 6 806952 6 91

13912510210710899

SRM 2711 AsCdCuMnPbZn

136 6 1152 6 7.3

95.3 6 8.9547 6 82

1120 6 190360 6 37

105 6 841 6 0.25114 6 2638 6 28

1162 6 31350.4 6 4.8

130127848696

103

% recovery(average)

101 6 15

TABLE IV. Method detection limit (MDL) for soil/sediment anal-ysis by ICP-OES using IH-ETV sample introduction (MESS-2, slur-ry sampling).

Element

MDL

Absolute/ng Relative/mg g21 a

AsCdCuMnPbZn

20158

125

20.10.50.81.20.5

a Based on the vaporization of 10 mg of sample.

con� dence level (c. l.), with the agreement between thedetermined and certi� ed values appearing as % recovery(i.e., the concentration determined for a given elementexpressed as a percent of its mean certi� ed value). Thepractice of using CRMs as external standards appears tobe a good approximation, with percent recoveries rangingfrom 80–105% for the elements determined in SRM2711. The errors observed in the determination probablyarose from differences in the matrices of the CRMs, andwere exacerbated by use of varying amounts of samplein preparing the slurries so as to offset the signi� cantlydifferent concentrations present in the CRMs.

For analysis by standard additions, two-point standardaddition curves were constructed by vaporizing 50 mL ofslurry and 50 mL of concentrated HCl with a 50 mL spikeof a weighed aqueous standard for the spiked point, andwith an equivalent aliquot of blank solution (0.5% HNO3)for the point representing the sample alone. In this cali-bration, the gross amount of sample used in preparing theslurries was adjusted such that the net analyte concentra-tions were as close as possible to each other, using thecerti� ed concentration of As as a guide. The results ofthis analysis appear in Table III, reported at 95% c. l. for6 separate calibrations, with the agreement between thedetermined and certi� ed values expressed as % recovery.Although the percent recoveries achieved did cover aconsiderable range (54–139%), the average recovery was101%, with half of the values reported being within 10%of their certi� ed counterparts. Although the analytes pre-sent in both the sample and standard spike are co-vola-tilized with the sample matrix, differences in the chemicalform of the analytes could still play a role in terms oftheir volatilization and transport ef� ciency and in turnwill in� uence the accuracy of the determination. The ad-dition of a second spike did not reveal any irregularitiesin the linearity of the standard additions curves, as r 2

values of at least 0.99 and as high as 0.9999 were ob-tained for the elements studied. Although not attemptedhere, the practice of spiking with a standard that closelymatches the matrix forms of the analytes being deter-mined (e.g., CRM), may further improve the accuracy ofthe results observed here. While the accuracy in the de-termination did not appear to have any dependence onthe sample type, the precision in the determination wasslightly better with the soil CRMs than the marine sedi-ment CRMs (average of 11 vs. 19%, respectively). TableIV lists the method detection limit (MDL) for each of theanalytes determined. These values represent the absoluteand relative (based on a 10 mg sample) concentrationsnecessary to generate a net signal three times (33) the

standard deviation of a low-level analyte signal (usingMESS-2 in this case). Comparison of these values withthe determined concentrations in Tables II and III showsthat the analyte levels present in the CRMs studied areat least an order of magnitude higher than their respectiveMDL, with the concentration of Cd in MESS-2 being theonly possible exception.

CONCLUSION

By using either CRMs in an external standardizationmethodology or aqueous standards in a standard additionscalibration, selected elements present in soil and sedimentsamples were determined with reasonable accuracy andprecision by ICP-OES using IH-ETV sample introduc-tion. Precision was found to vary from 4–16% RSD(transient signal peak area) for CRMs sampled as aque-ous slurries, and at least half of the determinations witheither methodology were within 10% of their mean cer-ti� ed concentrations. These results are quite good in light

APPLIED SPECTROSCOPY 821

of the fact that no matrix modi� er was added to the sam-ple, as is often the case with soil and sediment analysesusing ETV sample introduction,14 and that only an off-line drying step was incorporated into the vaporizationsequence prior to the high-temperature volatilization step.Nonetheless, future IH-ETV studies should look to theuse of matrix modi� ers and temperature programming asmeans of improving overall performance. Most graphite-ETV techniques also use vaporization substrates withhighly-ordered pyrolytic graphite coatings as a means ofimproving vaporization reproducibility and substrate life-time; the use of a pyrolytically coated graphite cup in theIH–ETV system should similarly improve performancein this regard.

The interchangeable sample probe design of IH-ETVlent itself well to this type of analysis, facilitating thepreparation of multiple samples off-line, and the rapidintroduction and removal of sample probes from the in-duction � eld. If one considers a typical ETV cycle(which, complete with drying, ashing, vaporization,cleaning, and cooling stages, is usually on the order ofseveral minutes), by using a system in which the samplesupports are interchangeable, the complete vaporizationcycle is now appreciably shorter as it consists only of avaporization stage (usually a few seconds in duration)and a stage for probe changeover. The automation of thechangeover and vaporization processes in IH–ETV wouldresult in a sample introduction system capable ofthroughput surpassing that of many existing techniques.

ACKNOWLEDGMENTSThe authors wish to express their sincere thanks to Hai Ying for his

assistance in developing software for the conversion of time scan � les.

Scholarship support provided by the province of Quebec through Fondspour la Formation de Chercheurs et l’Aide a la Recherche (FCAR) isgratefully acknowledged by author MER. This research was also sup-ported through funding from the Natural Sciences and EngineeringCouncil of Canada (NSERC) through an NSERC Operating Grant.

1. G. S. Senesi, G. Baldassarre, N. Senesi, and B. Radina, Chemo-sphere 39, 343 (1999).

2. F. M. G. Tack and M. G. Verloo, Int. J. Environ. Anal. Chem. 59,225 (1995).

3. C. R. Frink, J. Soil Contam. 5, 329 (1996).4. I. Jarvis and K. E. Jarvis, Chem. Geol. 95, 1 (1992).5. I. Novozamsky, H. J. van der Lee, and V. J. G. Houba, Mikrochim.

Acta 119, 183 (1995).6. Method 3050B, United States Environmental Protection Agency

(EPA).7. K. J. Lamble and S. J. Hill, Analyst 123, 103R (1998).8. Method 3051, United States Environmental Protection Agency

(EPA).9. R. Sing, Spectrochim. Acta, Part B 54, 411 (1999).

10. A. M. Gunn, D. L. Millard, and G. F. Kirkbright, Analyst 103, 1066(1978).

11. S. A. Darke and J. F. Tyson, Microchem. J. 50, 310 (1994).12. W. Schron, A. Liebmann, and G. Nimmerfall, Fresenius’ J. Anal.

Chem. 366, 79 (2000).13. K. Florian, J. Hassler, N. Pliesovska, and W. Schroen, Microchem.

J. 54, 375 (1996).14. G. Zaray and T. Kantor, Spectrochim. Acta, Part B 50, 489 (1995).15. M. E. Rybak and E. D. Salin, Spectrochim. Acta, Part B 56, 289

(2001).16. D. M. Goltz, C. D. Skinner, and E. D. Salin, Spectrochim. Acta,

Part B 53, 1139 (1998).17. D. M. Goltz and E. D. Salin, J. Anal. At. Spectrom. 12, 1175

(1997).18. C. D. Skinner and E. D. Salin, J. Anal. At. Spectrom. 12, 725

(1997).