elemental analysis by inductively coupled plasma mass spectrometry with sector field instruments: a...

JOURNAL OF MASS SPECTROMETRYJ. Mass Spectrom. 33, 579È590 (1998)

SPECIAL FEATURE:PERSPECTIVE

Elemental Analysis by Inductively Coupled PlasmaMass Spectrometry with Sector Field Instruments :a Progress Report

Dietmar Stuewer* and Norbert JakubowskiInstitut fu� r Spektrochemie und angewandte Spektroskopie, Postfach 10 13 52, D-44013 Dortmund, Germany

Inductively coupled plasma mass spectrometry (ICP-MS) for elemental analysis has been dominated since itsintroduction by instruments with quadrupole Ðlters for mass separation. Although considerably impeded by isobaricion interferences, ICP-MS with quadrupoles has nevertheless matured to become the most prominent MS techniquefor element analysis with widespread applications, providing extremely low detection limits in combination withtrue multi-element capabilities. Some years ago, a second generation of double-focusing ICP-MS instruments wasintroduced, o†ering the chance to overcome interference problems by high mass resolution. The special features ofthese instruments for elemental analysis are discussed and the progress achieved in analytical performance isdemonstrated by selected examples, emphasizing speciation as a Ðeld in which organic and inorganic aspects comeclose together. 1998 John Wiley & Sons Ltd.(

KEYWORDS: elemental analysis ; inductively coupled plasma mass spectrometry ; sector Ðeld ; quadrupole

INTRODUCTION

Elemental analysis today

With the rapidly increasing industrial development inthe second half of this century, elemental analysis hasreceived more attention and the responsibility forsecuring high quality standards for industrial productsin particular as well as for quality of life in general. Themain driving force in this development is the contin-uously increasing demand for quantitation with lowerdetection limits, down to previously inaccessible levelsin both trace analysis and micro analysis. On the onehand this requirement arose from the general experi-ence, mainly in material sciences, that physical proper-ties may be changed drastically by extremely lowconcentrations of certain elements in many matrices oftechnical interest. Therefore, elemental analysis gained asigniÐcant role in research and development as well asin support of manufacturing in the branches of industryconcerned. With the increasing social awareness of theproblems, risks and hazards connected with newmaterials, fabrication processes and products, inorganicanalytical chemistry on the other hand began to play amajor role in risk assessment and environmentalcontrol. Thus elemental analysis is now simultaneously

* Correspondence to : D. Stuewer, Institut fu� r Spektrochemie undangewandte Spektroskopie, Postfach 10 13 52, D-44013 Dortmund,Germany

applied for technical innovation and development ofnew products as well as for control of their risks andconsequences.

In real applications the most challenging require-ments for analytical detection power arise in the semi-conductor industry and refer not only to metal andsemiconductor starting materials but even more to thechemicals applied in the fabrication processes foretching, rinsing, etc. A special problem in this Ðeld is thedetermination of Ra, Th and U, which may be dele-terious for electronic circuits by inducing soft errors dueto a-radiation.

The analytical determination of radioactive elements,as just mentioned, is an intriguing special task, becausetheir presence in nuclear waste of any kind is of greatenvironmental signiÐcance, even at very low concentra-tions. Here we are leaving general trace element deter-mination for quality control and crossing over toenvironmental risks. In this Ðeld, elemental analysis isan indispensable requirement in many applications suchas pollution studies for water, soil and air. The medicalinterest in trace element studies, concerning both toxictrace elements and the essential elements, is obvious.Striking examples are the investigations of the conse-quences of the increasing environmental pollution by Ptfrom automotive exhaust catalysts, medical waste fromcancer therapy and artiÐcial fertilizers in agriculture.

The determination of Pt in environmental samples ischallenging owing to the low concentrations involved,and is further aggravated by the requirement for specia-tion, because the real environmental risks can only beestimated when taking into account the bioavailability

CCC 1076È5174/98/070579È12 $17.50 Received 30 March 1998( 1998 John Wiley & Sons, Ltd. Accepted 13 May 1998

580 D. STUEWER AND N. JAKUBOWSKI

of the di†erent species in which the element may bepresent. In the speciation of organic compoundsÈareally demanding task with a very wide Ðeld ofapplicationsÈthe frontiers between inorganic andorganic analysis are increasingly becoming obliterated,because coupling of element detection with chromato-graphic separation of the molecular species involvedmay lead to registration of analytical signals withrespect to selected elements for inorganic species and fororganics, at least when real samples are under investiga-tion.

The demand for lower detection limits has stimulatedthe development of many techniques for elementalanalysis, among which electrochemistry without doubtexcels as far as detection limits are concerned. Lowerdetection limits, however, lead, more or less uninten-tionally, to the consequence of further intricacy owingto an increase in the number of elements that have to beconsidered. This in turn leads to increased interest inanalytical techniques that permit true multi-elementanalysis while simultaneously retaining the maximumdetection power. This special demand can be satisÐedby instrumental techniques, among which mass spec-trometry (MS), after a long period of stagnation, hasnow become a widespread high-performance tool.

Development of elemental MS

The long predominance of atomic emission spectrom-etry (AES) over MS in elemental analysis was due to theformerly unsolved basic dilemma in elemental MS. Onthe one hand the method has unique advantages : incomparison with AES the mass spectra exhibit only alimited number of lines, which can easily be identiÐedand contain relevant analytical information on all ele-ments which may be constituents of the sample underinvestigation. On the other hand, there is the problemof ablation from the sample of material representativeof the composition of the sample and of transformingthe ablated material into an ion beam which can bemass separated and detected by a mass spectrometer.For volatile organic materials, this is usually not aproblem: evaporation with subsequent electron ioniza-tion with low energies is in most cases sufficient and caneasily be performed by simple means. In contrast toorganic substrates, however, this is a difficult obstacle toovercome for most inorganic solids, in particularmetals, for which the relevant thermodynamic proper-ties are much more demanding as far as the develop-ment of suitable ionization sources is concerned.

Basically there are two di†erent approaches for thegeneration of the required ion beam. The straightfor-ward approach is its direct production from a solidsample by means of electric discharges, sputtering bysurface bombardment or any other means involvingelectrical, thermal and mechanical processes in a moreor less complex interaction which still eludes any satis-factory theoretical description of the frequency distribu-tion of the ionization states involved. Hence absolutequantitation is unsatisfactory and can only be per-formed by using sensitivity factors derived from approp-rate standards, which in many cases are not available.

The alternative approach corresponds in some way tothe famous Gordian knot : the solid sample is disinte-grated by solution, but at the severe risk of introducingcontamination from the chemicals used in the necessarysteps which may include digestion and extraction.During the stagnation phase of elemental MS, there wasno suitable ionization technique available for solid orliquid analysis, except the sophisticated technique ofthermal ionization which, however, is restricted to aminority of elements.

Any consideration of the preferred approach in ananalytical task has to take into account also the ques-tion of quantitation. Solution analysis o†ers the advan-tage of external calibration by means of standards aswell as by standard addition techniques. Additionally,solution avoids the problem of local inhomogeneitieswhich are always crucial in direct analysis of solids.External calibration by standards is much moredemanding for the analysis of solids, because thecomplex interactions in the atomization and ion forma-tion processes mean that standards must be similar tothe sample under investigation in physical and crystal-lographic properties and surface structure as well aschemical composition. The latter is a general problemfor solids, in particular when concentrations down tothe ultratrace level must be determined, but not so forsolutions.

The duality of the approaches of direct solid analysisor solution analysis as just discussed is reÑected indevelopments since the early 1980s, when a renaissanceof MS in elemental analysis occurred. New ion sourcetechniques were introduced : (i) glow discharge massspectrometry (GDMS) for direct bulk analysis and addi-tionally for surface and depth analysis in the lm region,(ii) sputtered neutral mass spectrometry (SNMS) fordirect surface and depth analysis in the nm region and(iii) inductively coupled plasma mass spectrometry(ICP-MS), a procedure which uses the alternative ofsolution analysis. The last method has meanwhile foundwidespread application, not only in the Ðeld of solutionanalysis but also for the direct analysis of solids bymicro-sampling with laser ablation.

The development of GDMS and ICP-MS, althoughproceeding in parallel, showed a remarkable di†erence.The Ðrst commercial instruments in the Ðeld of GDMSwere double-focusing devices based on magnetic sectorÐeld mass analyzers, whereas the regime of ICP-MS wasdominated for a long time by low-resolution instru-ments which used quadrupole Ðlters for mass analysis(the acronym ICP-QMS will be used henceforth when-ever the discussion is restricted to ICP-MS with quad-rupole Ðlters as mass analyzers ; correspondingly, theacronym ICP-SFMS will be used when sector Ðeldinstruments are intended). Of course, from the begin-ning there were the idea, expectation and hope for anICP-MS instrument with high mass resolution in a rea-sonable price range not too far above that of the quad-rupole instruments. However, neglecting instruments ofa Ðrst generation in a nearly exclusive price class, ittook until the mid-1990s for the second generation ofcommercial high mass resolution ICP-MS instruments,speciÐcally designed for elemental MS, to appear. Thisdevelopment provided a new, strong impetus to theapplication of ICP-MS, and the consequences will be

( 1998 John Wiley & Sons, Ltd. J. Mass Spectrom. 33, 579È590 (1998)

ICP-MS WITH SECTOR FIELD INSTRUMENT 581

outlined in this paper after a short survey of someimportant aspects of conventional ICP-QMS from ananalytical point of view, but neglecting the physicaldetails.

ANALYTICAL QUADRUPOLE ICP-MS: ANASSESSMENT

Instrumentation

In the inductively coupled plasma (ICP) as used forexcitation in AES and for ionization in MS, an electricalÑame, the plasma, is produced by inductive coupling ofhigh-frequency energy to a working gas Ñow, usuallyargon. An aerosol of a nebulized analyte solution isinjected into the center of the plasma where the aerosoldroplets are vaporized and ionized. The plasma is oper-ated at atmospheric pressure ; however, for mass separa-tion by electromagnetic forces it is, of course, necessaryto increase the mean free path lengths of the ions up tothe geometrical dimensions of the housing, which meansthat the pressure must be reduced to high vacuum con-ditions. This is achieved by di†erential pumping in aninterface region as illustrated in Fig. 1. In the Ðrst stageof the interface, ions are sampled from the plasma Ñowby an entrance oriÐce, the so-called “sampler,Ï and trans-ferred to the mass analyzer volume by a second oriÐce,the so-called “skimmer.Ï This is the crucial region of thewhole system. An increase in the oriÐce diameter pro-vides higher ion yields but also a higher gas load on thepumping system; smaller diameters, on the other hand,

reduce the demands for the pumping system, but only atthe price of lower ion yields while simultaneouslyincreasing the risk of inÑuences from cool boundarylayers around the aperture on the composition of theplasma and the resulting ion species. Further problemsin the interface include the risk of secondary discharges,ablation of the oriÐce materials and clogging of thesampler or skimmer by salt or carbon deposits. Acertain problem area in the QMS arrangement is alsothe ion optical system. This lens system usually includesa Bessel box with center beam stop which is employedfor energy Ðltering and also serves to shield the detec-tion system from photons and high-energy particles,while shaping the ion beam for optimum acceptance bythe subsequent quadrupole Ðlter. In most cases analogand digital detection by a Faraday collector and anelectron multiplier device, respectively, are provided,which in combination provide a dynamic range ofabout nine orders of magnitude for detection.

Nebulization

A special part of the ICP-MS instrument is the nebu-lization system, which is required to produce an aerosolfrom the analyte solution for sample introduction intothe ICP. In order to permit maximum atomization effi-ciency and minimum loss by droplets which are toolarge to be vapourized completely, the size distributionshould be restricted to droplet diameters less than(about) 5 lm. Pneumatic nebulizers are standard andoccur in many varieties, among which the Meinhardtype is probably the most prominent. However, itsu†ers from a low overall efficiency of about 1È2% for atypical sample uptake rate of 1 ml min~1, so much

Figure 1. Schematic illustration of the plasma and interface region. 1 ¼Aerosol ; 2 ¼plasma torch; 3 ¼induction coil ; 4 ¼vaporizationregion; 5 ¼induction region; 6 ¼ionization region; 7 ¼boundary layer ; 8 ¼sampler orifice ; 9 ¼skimmer orifice ; 10 ¼differential pumping.



research e†ort has been aimed at high-efficiency, low-cost devices which maximize sensitivity and power ofdetection. Among di†erent high-efficiency nebulizationtechniques, ultrasonic nebulization (USN)1 and hydrau-lic high-pressure nebulization (HHPN)2 excel in per-formance. As an additional advantage, HHPN is moretolerant of a high salt load or high analyte viscosity.However, the high water load to the plasma often leadsto the additional requirement for a desolvation systemwhich combines heating and cooling stages forreduction of the solvent load. The higher efficiency ofthese nebulization systems is, of course, an advantagethat improves sensitivity and detection limits by morethan an order of magnitude for a majority of elements,so that the choice of the nebulizer system strongly inÑu-ences the Ðnal performance of the analytical procedure.

Interferences

The basic problem in elemental MS is not only pro-duction of an ion beam which represents the composi-tion of the sample but also ensuring that the recordedspectrum represents, as far as possible, nothing else butthe sample under investigation. The appearance of anycontamination from the sample preparation, in particu-lar chemical pretreatment, or from residual gases andproducts of any plasma reactions may be deleterious forthe reliability of an analytical procedure and the accu-racy of its results. This is valid for direct analysis ofsolids by GDMS and for solution analysis by ICP-MS,but is of considerably greater signiÐcance in solutionanalysis owing to higher risks during the more com-prehensive sample preparation. Furthermore, sourceoperation in the open atmosphere may lead to a greatervariety of spectroscopic interferences. We can, of course,never totally get rid of these interferences ; therefore,care is always required in order to avoid misinterpreta-tion and erroneous results.

Strictly, “spectroscopic interferenceÏ means the coin-cidence of contributions from di†erent species in onesignal, which is to say in one unresolved spectral line inthe case of MS. In ICP-QMS we are dealing withinstruments that usually exhibit unit resolution. There-fore, it has become common practice to speak of inter-ferences more generally in the case of di†erent specieswith the same nominal mass, whether or not the signalsare resolved in a spectrum.

Spectroscopic interferences may be subdivided intodi†erent types. First we have the isobaric interferencesfrom atoms with a di†erent number of protons but thesame total of nucleons. This is not too great an impedi-ment, because for all elements, except In, there is atleast one isotope free of any isobaric interference. Inmost cases, however, this will not be the isotope of thehighest abundance, so that the detection limit will beworsened by the selection of a minor isotope. The mostannoying example of an isobaric interference is thecoincidence of 40Ar and 40Ca. A similar type of quasi-isobaric interference is caused by ions of mass m withmultiple (z) charges which appear in the spectrumaccording to their m/z ratios, with both quadrupolesand magnetic sector Ðelds. A prominent example of thistype is the couple 28Si` and 56Fe2`.

The most signiÐcant type of interference, in bothnumber and as in intensity, is caused by a variety ofmolecular ions. These are hydrides, oxides, etc., mol-ecules of the argide type and clusters of any abundantcomponents including organic species, a particularproblem when organic solvents are involved. Recog-nizing and eliminating these moleculars interferencesmay become crucial to the success of the analysis.Important examples are, to give only a few, the inter-ference of 15N 16O` and 14N 16O 1H` with 31P`, of40Ar 16O` with 56Fe` and of 40Ar 35Cl` with 75As`.The appearance of a majority of these molecular inter-ferences not only depends on the sample itself and theworking gas of the discharge, but also on the chemicalsused in sample pretreatment, in particular the solvents.Table 1 compiles selected examples indicating also pos-sible origins of the constituents. It should be mentionedthat mass resolution values in this case are theoreticalones calculated from the mass di†erences, whereas inpractical examples the 10% valley deÐnition is alwaysused.

Finally, it should also be mentioned that highlyabundant neighboring signals, mainly caused by thematrix or the working gas of the discharge, may alsocontribute to a disturbance of the true analytical signalwhich, particularly with the low-resolution instruments,may become a problem. When operated at low massresolution, the “abundance sensitivityÏ of ICP-SFMSinstruments is about an order of magnitude lower thanthat of ICP-QMS instruments, but can be considerablyimproved by an increase in the resolution.

The signiÐcance of the interference problem varies, ofcourse, with the matrix under investigation. Simpleaqueous solutions with low salt loads may be easy toanalyze. Analysis of metals after solution by acids hasits special problems. Quantitative multi-elementanalysis of complex materials such as blood or ceramicsmay lead to a variety of unforeseen interferences.Special attention must also be paid to the interference

Table 1. Some typical interferences in elemental analysis byICP-MS and the theoretically required resolution

Atomic Molecular Required

ion ion resolution Origin

24Mg½ 12C2½ 1605 C (including organic)

28Si½ 14N2

958 HNO3

28Si½ 12C 16O½ 1557 C (including organic)

31P½ 14N 16O 1H½ 968 H2O, Ar

31P½ 15N 16O½ 1458 HNO3

32S½ 16O2½ 1801 H

2O

44Ca½ 12C 16O2½ 1281 C (including organic)

48Ti½ 32S 16O½ 2519 H2SO

451V½ 35Cl 16O½ 2572 HCl

52Cr½ 35Cl 16O 1H½ 1671 HCl

52Cr½ 40Ar 12C½ 2375 C (including organic), Ar

54Cr½ 40Ar 14N½ 2054 HNO3, Ar

54Fe½ 40Ar 14N½ 2088 HNO3, Ar

56Fe½ 40Ar 16O½ 2502 H2O, Ar

64Zn½ 32S 16O2½ 1952 H

2SO

464Zn½ 32S

2½ 4264 H

2SO

475As½ 40Ar 35Cl½ 7775 HCl, Ar

96Mo½ 40Ar 56Fe½ 13043 Ar



problem when organic compounds are involved as, e.g.in ICP-MS analysis of chromatographic fractions. Atypical example of this kind is the interference of40Ar 12C` with 52Cr`, the latter being the main isotopeof an element which is of great interest in the Ðeld ofelement speciation. This example is chosen here to showwhat spectra recorded by QMS and SFMS look like.Figure 2 shows a mass spectrum obtained by ICP-QMScovering the region in which Cr appears and also amoderate-resolution SFMS spectrum at 52 u thatallows the isobaric ions to be separated.

Many attempts have been made to cope with inter-ference problems, as reviewed recently.3 Mixed gasplasmas were investigated, collision or reaction cellswere additionally employed with di†erent gases andmore recently the cold plasma approach has foundsome attraction.4 However, none of these techniquesrealizes a general solution to the problem; at most theyo†er a chance to get rid of several problems from agroup of interferences or for selected elements in specialcases so that, in conclusion, the interference problem isstill the most signiÐcant weakness of ICP-QMS.

For reasons of completeness, it should be mentionedthat in addition to these spectral interferences, non-spectral interferences, generally addressed as matrixe†ects, can also be an obstacle for ICP-MS. Forexample, a concentration of 50 ppm of a heavy elementcan cause a 10È20% depression of the analytical signalof a light element, and at 1000 ppm the loss of sensi-tivity may even amount to a factor of 50. This pheno-menon, which nowadays is often interpreted in terms ofspace charge e†ects in the interface region, results incalibration non-linearities with increasing concentrationof major sample constituents.5 The general experiencewith such e†ects can be summarized in the statementthat elements of greater mass usually exert stronger

inÑuences of this type whereas light elements are usuallya†ected more strongly. A common technique to copewith such non-linearities is internal standardization byselected elements such as Be, Sc, Co, Y, Rh, In, Hf andBi. Monoisotopic elements not present in the sampleare preferable as standards ; similarity to the elements ofmain analytical interest in various properties such asmass, ionization energy and oxide formation is desir-able.

In conclusion, the unique combination of advantagesof ICP-MS for elemental analysisÈextremely low detec-tion limits for (nearly) all elements down to the pptregion and true multi-element capabilities in a quasi-simultaneous procedureÈshould be recalled. On thebasis of permanent “learning by doing,Ï the analyticalcommunity has learned to live with the above-mentioned deÐciencies. The unique combination ofadvantageous features has had convincing analyticalbeneÐts in many Ðelds of application. This success,together with the awareness that considerable improve-ments can be expected from high-resolution instru-ments, is the reason why the appearance of suchinstruments has been awaited for many years.

ANALYTICAL SECTOR FIELD ICP-MS:GENERAL ASPECTS

Enhancement of the mass resolution means reduction ofthe interference problem, but it cannot be a panaceaagainst spectroscopic interferences in general, becausethe mass resolution will always be limited. As anexample, consider the determination of As in the pres-ence of the nearly inevitable 40Ar 35Cl` ion as illus-trated in Fig. 3. This is probably the most signiÐcantexample in which a resolution of about 104 is required.

Figure 2. Uncovering an interference: mass region of Cr measured by ICP-QMS and measurement at 52 u by ICP-SFMS (R ¼3000)showing the strong interference from 40Ar 12C½ to 52Cr½ (Cr concentration 0.2 ng mlÉ1). Registration of 56Fe was suppressed for overloadprotection.



Figure 3. Interference of 40Ar 35Cl½ with 75As½ measured with a resolution R ¼13 000, from a determination of 10 ng mlÉ1 As in 1.0% HCl(courtesy of Finnigan MAT, Bremen, Germany).

However, with R\ 104 the separation of 40Ar 56Fe`from 96Mo` remains a problem, and there are somecases in which even a resolution of R\ 105 may nothelp, e.g. the interference of 40Ar 51V` with 91Zr`. Inconclusion, a great majority of the interferences of ana-lytical signiÐcance can be overcome by a resolution ofRP 104.

In performance evaluation we have to take intoaccount a further inÑuential parameter. In practice, anincrease in the mass resolution is achieved by areduction in the widths of beam deÐning slits, and thisin turn leads to a decrease in ion transmission, whichfor instance amounts to about one order of magnitudewhen the resolution is increased from 300 to 3000. In

spite of the general familiarity with the basic trade-o† ofmass resolution for ion transmission, its signiÐcance forelemental analysis will nevertheless be illustrated here.The example chosen is the interference of 40Ar 16O`with 56Fe`, the major isotope of this importantelement. Figure 4 shows ICP-SFMS signals for theregion of 56 u recorded with resolution R varying from300 up to 2500. A small Fe signal can be observed evenin the highest resolution scan, but the drastic intensityreduction due to the resolution increase is also obvious.One should be aware that there are other argide inter-ferences also.

Higher mass resolution always has to be paid for by adecrease in sensitivity and detection limits. Hence the

Figure 4. ICP-SFMS measurement of the region of 56 u with varying mass resolution. An Fe concentration of 20 ng mlÉ1 can finally bedetermined on the low-mass tail of the extremely abundant 40Ar 16O½ molecule. R ¼(1) 300; (2) 1500; (3) 2000; (4) 2500.



dilemma arises of whether higher mass resolution orhigher ion transmission is of greater priority for an ana-lytical task. Ultimate detection limits can only be rea-lized with low or moderate resolution using sector Ðeldinstruments, and these instruments are therefore oftenoperated under conditions which come close toICP-QMS as far as the resolution is concerned. This isthe reason why it is generally preferable to characterizeICP-MS with high-resolution instruments as ICP-SFMS (sector Ðeld) and not as ICP-HRMS (highresolution).

The technical demands and the economic risks in thedevelopment of an instrument specially designed forICP-SFMS are the reasons why progress was so slow inthis Ðeld. The Plasmatrace I (VG/Fisons, Winsford,UK), was the only instrument of the Ðrst generation,and the introduction of ICP-SFMS with a speciÐcallydesigned double-focusing MS analyzer as a high-performance extension of the conventional ICP-QMShad to wait until the early 1990s. Four manufacturersnow o†er a selection of such instruments : the Element(Finnigan MAT, Bremen, Germany), the Plasmatrace II(Micromass, Manchester, UK), the JMS-Plasmax 2(Jeol USA, Peabody, MA, USA) and the Plasma 54 (VGElemental, Winsford, UK; Micromass), and two newinstruments with improved performance features havebeen announced : the Axiom (VG Elemental) and theIsoprobe (Micromass). Technical details are beyond thescope of this paper and can be found in the manufac-turersÏ brochures. It should merely be pointed out thatthe Plasma 54, the Isoprobe and also the Axiom wereespecially developed for high-precision isotope ratiomeasurement with a multi-collector system. Further-more, it may be of interest in this context that only inthe Element instrument is the acceleration voltageapplied to the analyzer system, so that operation of theICP and interface is possible without any high-voltagehazards.

Figures of merit

The primary aims of elemental analysis are (i) quantitat-ive determination of the elemental composition of asample and (ii) calculation of a detection limit whichprovides an estimate of the upper limit possible whenthe presence of an element is not established. Elementalconcentrations are determined from a calibration line,the slope of which is the sensitivity. Determination of adetection limit involves, in addition, repeated measure-ment of a blank sample at the spectral position of theanalytical signal. Concerning sensitivity, it is an advan-tage of SFMS that the ion transmission generally doesnot depend on the mass whereas QMS exhibits anincreasing sensitivity loss above 100 u. The sensitivity ofICM-SFMS instruments is in the region of 107È109 cpsper lg ml~1 (cps \ counts per second). A sensitivity of108 cps per lg ml~1 has also been achieved with anICP-QMS instrument of the most recent generation.6Generally, however, SFMS instruments are superiorwhen operated at low resolution and are comparableoverall to QMS when operated at moderate resolution.

Extremely low background noise below 1 count s~1,which is an order of magnitude lower than with quadru-

pole instruments, is a further major advantage of ICP-SFMS instruments. This generally o†ers the chance toachieve detection limits below 1 pg ml~1, but which inmany practical applications cannot be realized owing totoo high blank values. A typical example is the appear-ance of a high blank value for Ni, which is inevitablewhen the interface cones are made of this metal. The useof Pt cones is a frequently applied alternative. In con-clusion, the higher transmission and lower backgroundof SFMS permit an improvement in the detection limitsfor low-resolution operation that may amount to up tothree orders of magnitude in comparison with QMS aslong as blank values do not prohibit this.

Precision is particularly relevant for isotope ratiomeasurement. The most important limiting factors arisefrom Ñuctuations in aerosol generation, nebulizationand plasma instabilities. These e†ects can only be beovercome with real simultaneous operation in multi-collector arrangements. The limitations due to sourcenoise, however, remain the most serious.

APPLICATIONS

Reduction of interferences

Basically, two types of applications may be discernedthat su†er severely from interferences. In the analysis ofpure and ultrapure metals, in particular as startingmaterials for “high-techÏ products, the analytical per-formance is limited for a variety of elements by spectraloverlap at least for the most abundant isotope of theelement in question. This is particularly valid for thesemiconductor industry. On the other hand, in complexmatrices, the wide variety of components may give riseto unforeseen interferences as a source of systematicerrors ; typical cases of this type are, for instance, theanalysis of sea water, serum or urine and also ofceramics of refractory oxide mixtures.

For steel analysis, ICP-MS o†ers a particular chanceto analyze non-metallic elements such as P and Stogether with the metal components. These elementscan have a strong inÑuence to the properties of steeleven at low concentration levels (lg g~1). Owing tointense interferences, detection limits are poor in QMS.With SFMS, however, detection limits for the solid inthe ng g~1 range are generally achieved. Even for phos-phorus, which is not a desired element in steel, the con-centration has to be determined according to increasingquality demands down to the low ng g~1 range. Thedetermination of P by ICP-MS requires a resolution ofabout 1500 owing to interferences from 15N 16O` and14N 16O1H, as demonstrated in Fig. 5. In comparisonwith the conventional determination of P by absorptionspectrophotometry, which is laborious and time con-suming, the determination by ICP-SFMS is a fast,multi-element alternative, as has been shown in recentwork.7

In a study concerning the analysis of high-puritystandard reference metals (Cd, Cu, Ga, Zn), mean detec-tion limits of about 15 pg ml~1 were obtained for 30elements ; the corresponding mean value for the solidswas 3.5 ng g~1.8 An interesting Ðnding was the obser-vation of a matrix-induced interference, which in



Figure 5. Determination of 31P in the presence of abundant interferences using ICP-SFMS with standard addition for calibration. R ¼2100(inset). 1 ¼Blank; 2 ¼sample ; 3 ¼sample ½1.63 ng mlÉ1 ; 4 ¼sample ½3.26 ng mlÉ1.

ICP-QMS would have resulted in a matrix e†ect. Theisotope concerned was 151Eu`, for which an inter-ference from 40Ar 111Cd` became apparent with Cd asmatrix. This demonstrates that ICP-SFMS is a usefultool for securing accuracy, as in particular required inthe analysis of standard reference materials, by the elu-cidation of contributions from interferences, even ifunforeseen.

Human serum, as well as whole blood, urine andtissue, represents the type of complex matrix for whichanalysis by ICP-QMS su†ers from more interferencesthan usual. In addition to the contributions from morecommon species such as argides and oxides, the pres-ence of a variety of elements in serum (C, Na, P, S, Cl,K, Ca) gives rise to many unusual interferences, and theapparent concentration, i.e. the element concentrationcorresponding to the signal intensity of the interferent,of these species can be as high as the lg ml~1 level. Ithas been shown that for some important trace elements(e.g. Fe, Cu, Zn), a reliable determination in their typicalconcentration range is nevertheless possible, whereas anaccurate determination with ICP-QMS is more or lessimpossible for others (Ti, V, Cr, Ni).9

Really difficult examples of complex matrices areceramics, soils and rocks. Digestion of such samplesusually requires a mixture of acids, so that the risk ofinterferences from contamination is multiplied accord-ing to the number of components. To consider one casein more detail, the signal of 48Ti` often su†ers consider-ably from the interference of 32S 16O` owing to theapplication of in the mixture used for digestionH2SO4of powdered materials such as SiC, whereas the minorabundant isotopes of Ti are also more or less obscured.Figure 6 shows results from the calibration for Ti in2.5% solution. With respect to the mass di†er-H2SO4ence, a resolution of 2518 should be sufficient for signalseparation ; in practice, however, a resolution of nearly

5000 is required, owing to the high abundance of theinterfering molecular ion.

Ultimate detection limits

The performance of ICP-SFMS in recent work on thedetermination of Pd and Pt in urine illustrates its appli-cation to environmental e†ects of automotive catalystexhaust.10 A detection limit of about 0.05 pg ml~1could be obtained, which is comparable to those of anextremely sensitive adsorptive voltammetric procedureand deÐnitely below the results obtained with any othertechnique. Further improvement may be achieved byeven more careful control of blanks and removal ofinterferences by using suitable sample preparation tech-niques. Based on this work, the authors recommend theapplication of the standard addition procedure forquantitation in order to correct for non-spectroscopicinterferences associated with high matrix concentra-tions.

The excellent performance of ICP-SFMS as a multi-element technique becomes particulary obvious inpurity analysis of the type required in the semicon-ductor industry. Recently a comparative study of thedetermination of ultratrace components in GaAs hasshown that solution analysis by ICP-SFMS permitsdetection limits for the solid in the low ng g~1 range.This value is as good as in GDMS, the alternative directtechnique, in spite of the dilution in the liquid analyte.11This result, however, could only be achieved bydemanding e†orts to avoid contamination. A furtherexample of a typical application in this Ðeld has alreadybeen presented in Fig. 3. In this case, the analytical taskwas the determination of 10 ng ml~1 of As in 1.0% HCl.It may be estimated from Fig. 3 that detection limits inthe sub-ppb region could be realized. The most impor-



Figure 6. Determination of 48Ti½ in 2.5% solution resulting in a strong interference from 32S 16O½. Owing to the highly abundantH2SO

4interference together with low mass tailing of the molecular peak, a resolution of R ¼5000 is necessary instead of 2520 as calculated fromthe mass difference. 1 ¼Blank; (2) ¼1 ng mlÉ1 ; 3 ¼2 ng mlÉ1 ; 4 ¼5 ng mlÉ1.

tant application of ICP-SFMS in the semiconductorindustry, one in which it really excels, is the multi-element analysis of solvents and other liquid reagents.As an example, Fig. 7 shows detection limits from theanalysis of semiconductor grade sulfuric acid obtainedusing a pneumatic nebulizer and various resolvingpowers. The values are mainly concentrated within twoorders of magnitude with a mean of about 5 pg ml~1.

An impressive example of ultimate detection limitscan be taken from early work in this Ðeld by Yamasakiet al.,12 who investigated the determination of ultratraceelements in terrestrial water. In comparison with the

preceding example, one has to take into account thatthis work was performed utilizing the higher efficiencyof USN instead of pneumatic nebulization. Detectionlimits (DL) calculated from three times the scatter of theblank signal were presented for 50 elements with fourdi†erent mass resolution settings. A compilation of theresults is given in Fig. 8. Generally the DL values of theelements are in the sub-pg ml~1 region with the excep-tion of only a few. For the transition elements whichwere measured with R\ 3000, the DL values are at alevel of 100 fg ml~1. Arsenic was the only element torequire a mass resolution of R[ 7500, and a DL of 10

Figure 7. Detection limits from an ICP-SFMS analysis of semiconductor-grade sulfuric acid with various settings of the mass resolution Rusing a pneumatic nebulizer (courtesy of VG Elemental, Winsford, UK).



Figure 8. Representation of detection limits in ICP-SFMS determinations with various settings of the mass resolution R using an ultrasonicnebulizer, as reported by Yamasaki et al .12

pg ml~1 resulted. Presumably unrivalled is the determi-nation of a DL for 235U: a value of 0.07 fg ml~1 wasachieved in a 1 h measurement of a blank solution. Inspite of extreme precautions, limitations arise mainlyfrom blank values.

Another application in which ultimate detectionlimits are sought is the determination of radioactivenuclides. Reliable monitoring is necessary owing totheir high environmental risk potential. In the work ofKim et al.,13 ICP-SFMS was successfully employed forthe determination of some long-lived radionuclides(99Tc, 226Ra, 232Th, 237Np, 238U, 239Pu, 240Pu). Withapplication of USN, detection limits in the region of1È10 fg ml~1 were obtained, corresponding to activitiesin the low nBq region. For reasons of comparison, thisrequired D2.5 min using ICP-SFMS, whereas with con-ventional radiometric techniques at least 1 day is neces-sary to register about 100 counts per gram of samplematerial.

Isotope ratio measurement

Notwithstanding the basic superiority of thermal ion-ization mass spectrometry (TIMS), ICP-MS instrumentsalso provide isotope ratio measurements and thus appli-cation of the isotope dilution technique. The capabilitiesare sufficient for many applications and, in particular,for tracer studies in biological matrices. ICP-MS is cer-tainly superior to TIMS as far as sample preparationdemands and analysis time are concerned. In compari-son with ICP-QMS, the performance is improved bySFMS, which must be attributed mainly to improvedcounting statistics. A relative standard deviation (RSD)of 0.04% with low resolution has been demonstrated,for instance, for Mg and Pb.14 Lead contents of 0.02È9.7 ng g~1 were determined with respect to their isotoperatios with an average RSD of only 0.14%. The

improvement achieved by ICP-SFMS has also beendemonstrated in ratio measurements of the radio-isotopes 240Pu and 239Pu, which were performed in a 21pg ml~1 Pu solution with a precision of 2% and a devi-ation of 0.8% from the certiÐed value.15 With aresolution of 3000 an RSD of \0.1% was obtained forthe determination of the 63Cu/65Cu ratio in 10 replicatemeasurements.16 ICP-SFMS may furthermore beattractive for studies of the uptake and excretion ofstable tracers in the human body in order to avoid radi-ation hazards from radiotracers on the one hand andcost and time consumption associated with TIMS oreven accelerator MS on the other.17 The performance ofICP-SFMS may become competitive with TIMS byutilization of multicollector arrangements with trulysimultaneous detection.

CURRENT TRENDS

Instrumental developments

As can be seen from the preceding selection of applica-tion examples, the appearance of ICP-SFMS instru-ments has provided a strong impetus for new anddemanding analytical applications. Beyond this, therapid increase in new and sometimes exciting analyticalexperiences has a strongly stimulating e†ect for thewhole Ðeld of ICP-MS. The performance features of thenew instruments and their acceptance in the analyticalcommunity are the basis for a rapid increase in applica-tions. The commercial success in turn is stimulatingcommercial competition, as demonstrated by theappearance of instruments of the second generation inrapid succession.

The progress in elemental analysis which has beenachieved with the appearance of ICP-SFMS instru-ments has even resulted in a certain interest in still



higher mass resolution. In pioneering work, an ICP-MSinstrument achieving R\ 120 000 has been described.18A further promising approach to SFMS instruments ispossible by a return to the old MattauchÈHerzoggeometry as the basis for truly simultaneous detec-tion.19 Once this was utilized mainly with photoplatesin spark source instruments for the analysis of solids,and a revival now seems possible with the ICP assource by application of an array detector. An arrange-ment of this type can be expected to o†er the highestsensitivity for simultaneous multi-element operationand also convincing precision for isotope ratio measure-ments.

Considerable activity in instrumental and method-ological endeavors can also be observed in the Ðeld ofICP-QMS, probably as a result of mutual challenge. Anexample of a new technique is the “cold plasmaÏapproach, which is especially suited to improve in“cleanÏ solutions and semiconductor-grade chemicals thedetection limits for K, Ca and Fe.20 Another approachused with increasing frequency is the inclusion of colli-sion cells that are useful in the reduction of argide-typeinterferences.21

Recent work with the aim of developing high-resolution quadrupoles has shown promising results.22With a newly developed system, it was possible toresolve 40Ar 16O` from 56Fe`. The quadrupoleanalyser was operated in the second stability regionwhich minimized the sensitivity losses normally appear-ing with high-resolution settings of quadrupoles. Iontrap analyzers o†er another alternative to conventionalquadrupole mass Ðlters, and high mass resolution canbe obtained as is well known in organic MS. Ions canalso be stored in a commercially available ion trapoperated as a reaction cell, which has been successfullyexploited for the destruction of polyatomic ions.23

As a further MS technique, time-of-Ñight massanalysis is also under investigation for ICP-MS. It hasbeen demonstrated that a mass resolution R[ 1000 canbe obtained. Additionally, the high time resolution ofthese instruments makes them well suited to follow fasttransient signals, as has been demonstrated in an appli-cation with electrothermal evaporation.24

In conclusion, in the early days of ICP-MS the devel-opment was mainly inÑuenced by unsatisÐed demandsfrom the application of AES. Typical users of thismethod were the Ðrst to become interested in ICP-MSand its analytical applications and to develop it tothe maturity it has now reached. The current trends,however, demonstrate that typical tools of organic MSsuch as collision cells, ion traps and time-of-Ñightinstruments now are increasingly being considered asnew and promising components for further develop-ment.

Hyphenated methods

The increase in performance and in the Ðeld of applica-tions is a reason to consider ICP-SFMS also for ele-mental detection in “hyphenatedÏ techniques. Inparticular, this involves coupling with various chro-matographic procedures utilizing the on-line capabil-ities of modern high-efficiency nebulization techniques.

Basically the settling time of the magnet limits scanspeed and repetition frequency of ICP-SFMS instru-ments. With the improved performance of modern lami-nated magnets, however, ICP-SFMS instruments areÐnding increasing interest as detection systems to becoupled not only with di†erent sample introductiontechniques for direct analysis of solids, mainly electro-thermal vaporization and laser ablation, but also forliquid chromatography and gas chromatography. Inresponse to the increasing demand for speciation, thecapabilities of ICP-SFMS in combination with di†erentchromatographic techniques are under investigation inseveral research groups. The capabilities for measure-ment of transient signals do not reach the excellent timeresolution of ICP-QMS; however, they will be sufficientfor many applications of ICP-SFMS as a detectionmethod in analytical procedures. In speciation, ultimatedetection limits are in many cases of greater priority, sothat research in this Ðeld will increasingly be under-taken with ICP-SFMS instruments for element-speciÐcdetection.

As an early example of processing of transient signalsfor only a few seconds duration by ICP-SFMS, thecoupling of an ICP-MS instrument for element detec-tion with a capillary gas chromatograph has beenreported.25 The aim of this investigation was the simul-taneous multi-element speciation of organic pollutantsin water from a harbor basin. Xe was added as internalstandard and detected continuously together with Sn,Hg and Pb. The widest mass span to be covered bypeak hopping in the multiple ion detection procedurewas from 120Sn up to 208Pb. All isotopes taken intoaccount could be measured three times a second, andthe results obtained were satisfactory with respect toprecision and accuracy.

The basic advantage of ICP-SFMS of exhibiting truemulti-element capabilities is speciÐcally applied to thetask of multi-species analysis, which opens up a newÐeld because more conventional speciation techniquesare usually restricted to one element only. In an investi-gation by ICP-MS of Pt-treated plants, on-line couplingwith size-exclusion chromatography gave multi-elementdeterminations of indicator elements and enabled con-clusions to be drawn about the nature of the fractionsobserved in combination with size-exclusion chroma-tography.26 As an example from this investigationwhich was performed with both low and high massresolution, a strong correlation was observed betweenthe elements Pt and S for one of the organic fractions.As far as resolution is concerned, the measurementsdemonstrate a background reduction and an improve-ment in signal-to-noise ratio by SFMS in comparisonwith QMS when applied for recording a mass-speciÐcchromatogram. As far as the interrelation betweenorganic and inorganic MS is concerned, the correlationof the signals for Pt and S is a clear hint that we aredealing here with a protein fraction, probably a phyto-chelatine. In another fraction, the correlation betweenPt and Ca indicates an exchange mechanism for Ca andPt in polygalacturonic acids as a consequence of the Ptstress. In a further recent investigation of DNA adducts,the quantitative determination of all observed specieswas possible by evaluation of the phosphorus signaldetected with ICP-SFMS.27 These are Ðrst preliminary



examples that multi-element analysis by ICP-MS can bea useful tool for the interpretation and quantitativedetermination of complex organic species by measure-ment of indicator elements. Simple quantitation bystandard addition allows the lack of suitable organicstandards to be overcome. It may be foreseen that ele-mental analysis by ICP-SFMS will become a versatiletool, even for many typical organic applications.

CONCLUSION

It was a primary aim of this progress report to demon-strate through several examples that ICP-SFMS hasrapidly gained a signiÐcant role in elemental analysis. Itis an important high-perforance tool for many applica-tion Ðelds, such as materials science, environmentalanalysis, the electronics industry, nuclear technology,biology and medicine. As always, success has many

origins : high mass resolution, extremely low noise, ulti-mate detection limits, satisfactory precision and accept-able cost. Increasing acceptance leads to increasingcompetition : further reductions in costs can beexpected, and more research and development shouldresult in innovations leading to improved performanceand versatility. Established concepts and techniquesfrom organic MS, e.g. the application of ion traps andtime-of-Ñight analyzers, are being introduced into inor-ganic MS. The future will probably bring more simi-larities between organic and inorganic MS and also anenhanced interrelation that is fruitful for both. In theÐelds of speciation for metals and non-metals in bio-molecules, the frontiers are already being obliterated.Both Ðelds can complement each other and both standin a certain need of each other to satisfy the permanentdemand to identify and determine more and morecomplex substances in lower concentrations with higherprecision and accuracy.

REFERENCES

1. S. I. Yamasaki, A. Tsumura and Y. Takaku, Microchem. J. 49,305 (1994).

2. N. Jakubowski, C. Thomas, D. Stuewer, I. Dettlaff and J.Schram, J.Anal . At . Spectrom. 11, 1023 (1996).

3. E. H. Evans and J. J. Giglio, J . Anal . At . Spectrom. 8, 1(1993).

4. S. D. Tanner, J . Anal . At . Spectrom. 10, 905 (1995).5. G. R. Gilson, D. J. Douglas, J. E. Fulford, K. W. Halligham and

S. D. Tanner, Anal . Chem. 60, 1472 (1988).6. E. R. Denoyer, D. Jacques, E. Debrah and S. D. Tanner, At .

Spectrom. 16, 3 (1995).7. S. Finkeldei and G. Staats, Fresenius J. Anal . Chem. 359, 357

(1997).8. R. Matschat, M. Czerwensky, M. Hamester and S. Pattberg, S.

Fresenius’ J . Anal . Chem. 359, 418 (1997).9. C. Vandecasteele, H. Vanhoe and R. Dams, J. Anal . At .

Spectrom. 8, 781 (1993).10. J. Begerow, M. Turfeld and L. Dunemann, J. Anal . At .

Spectrom. 12, 1095 (1997).11. J. S. Becker, V. K. Panday, R. S. Soman, M. Jurisch and H.-J.

Dietze, presented at the 1998 Winter Conference on PlasmaSpectrochemistry, Scottsdale, AZ (1998).

12. S. I. Yamasaki, A. Tsumura and Y. Takaku, Microchem. J. 49,305 (1994).

13. C.-K. Kim, R. Seki, S. Morita, S. I. Yamasaki, A. Tsumura, Y.Takaku, Y. Igarashi and M. Yamamoto, J. Anal . At . Spectrom.6, 205 (1991).

14. F. Vanhaecke, L. Moens, R. Dams and P. Taylor, Anal . Chem.68, 567 (1996).

15. C.-K. Kim, R. Seki, S. Morita, S. I. Yamasaki, A. Tsumura, Y.Takaku, Y. Igarashi and M. Yamamoto, J. Anal . At . Spectrom.6, 205 (1991).

16. F. Vanhaecke, L. Moens, R. Dams, I. Papadakis and P. Taylor,Anal . Chem. 69, 268 (1997).

17. S. Stu� rup, M. Hansen and C. J. Anal . At . Spectrom.M‰lgaard,12, 919 (1997).

18. M. Morita, H. Itoh, T. Uehiro and K. T. Otsuka, Anal . Sci . 5,609 (1989).

19. E. F. Cromwell and P. Arrowsmith, J. Am. Soc. MassSpectrom. 7, 458 (1996).

20. S. D. Tanner, J . Anal . At . Spectrom. 10, 905 (1995).21. P. Turner, T. Merren, J. Speakman and C. Haines, in Plasma

Source Spectrometry—Developments and Applications ,edited by G. Holland and S. D. Tanner, p. 28. Special Pub-lication No. 202, Royal Chemical Society, Cambridge (1997).

22. J.-F. Ying and D. J. Douglas, Rapid Commun. MassSpectrom. 10, 649 (1996).

23. D. W. Koppenaal, C. J. Barinaga and M. R. Smith, J. Anal . At .Spectrom. 9, 1053 (1994).

24. P. P. Mahoney, S. J. Ray and G. Hieftje, Appl . Spectrosc. 51,16A (1997).

25. T. De Smaele, L. Moens and R. Dams, in Plasma SourceSpectrometry—Developments and Applications , edited by G.Holland and S. D. Tanner, p. 109. Special Publication No.202, Royal Chemical Society, Cambridge (1997).

26. D. Klueppel, N. Jakubowski, J. Messerschmidt, D. Stuewerand D. Klockow, J.Anal . At . Spectrom. in press.

27. N. Jakubowski, C. Siethoff, M. Linscheid, I. Feldmann and D.Stuewer, presented at the 1998 Winter Conference on PlasmaSpectrochemistry, Scottsdale, AZ (1998).


elemental analysis by inductively coupled plasma mass spectrometry with sector field instruments: a...

Documents