a beginners guide - Induced coupled plasma mass spectroscopy

38 SPECTROSCOPY 16(4) APRIL 2001 www.spec t roscopyo n l i n e . com

Amazingly, 18 years after the com-mercialization of inductively cou-pled plasma mass spectrometry(ICP-MS), less than 4000 systemshave been installed worldwide. If

you compare this number with anotherrapid multielement technique, inductivelycoupled plasma optical emission spec-trometry (ICP-OES), first commercial-ized in 1974, the difference is quite signif-icant. In 1992, 18 years after ICP-OES wasintroduced, more than 9000 units hadbeen sold, and if you compare it with thesame time period that ICP-MS has beenavailable, the difference is even more dra-matic. From 1983 to the present day,more than 17,000 ICP-OES systems havebeen installed — more than four timesthe number of ICP-MS systems. If thecomparison is made with all atomic spec-troscopy instrumentation (ICP-MS, ICP-OES, graphite furnace atomic absorption[GFAA] and flame atomic absorption[FAA]), the annual turnover for ICP-MSis less than 7% of the total atomic spec-troscopy market — 400 units comparedto approximately 6000 atomic spec-troscopy systems. It’s even more surpris-ing when you consider that ICP-MS of-fers so much more than the othertechniques, including two of its most at-tractive features — the rapid multiele-ment capabilities of ICP-OES, combinedwith the superb detection limits of GFAA.

ICP-MS — ROUTINE OR RESEARCH?Clearly, one of the reasons is price — anICP-MS system typically costs twice asmuch as an ICP-OES system and threetimes more than a GFAA system. But in acompetitive world, the “street price” of anICP-MS system is much closer to a top-of-the-line ICP-OES system fitted with sam-pling accessories or a GFAA system thathas all the bells and whistles on it. So ifICP-MS is not significantly more expen-

sive than ICP-OES and GFAA, why hasn’tit been more widely accepted by the ana-lytical community? I firmly believe thatthe major reason why ICP-MS has notgained the popularity of the other traceelement techniques is that it is still con-sidered a complicated research tech-nique, requiring a very skilled person tooperate it. Manufacturers of ICP-MSequipment are constantly striving tomake the systems easier to operate, thesoftware easier to use, and the hardwareeasier to maintain, but even after 18 yearsit is still not perceived as a mature, rou-tine tool like flame AA or ICP-OES. Thismight be partially true because of the rel-ative complexity of the instrumentation;however, in my opinion, the dominantreason for this misconception is thatthere has not been good literature avail-able explaining the basic principles andbenefits of ICP-MS in a way that is com-pelling and easy to understand for some-one with very little knowledge of thetechnique. Some excellent textbooks (1,2) and numerous journal papers (3–5)are available that describe the fundamen-tals, but they tend to be far too heavy for

a novice reader. There is no question inmy mind that the technique needs to bepresented in a more user-friendly way tomake routine analytical laboratories morecomfortable with it. Unfortunately, thepublishers of the “for Dummies” series ofbooks have not yet found a mass (excusethe pun) market for writing one on ICP-MS. So until that time, we will be present-ing a number of short tutorials on thetechnique, as a follow-up to the posterthat was included in the February 2001issue of Spectroscopy.

During the next few months, we will bediscussing the following topics in greaterdepth:• principles of ion formation • sample introduction • plasma torch/radio frequency genera-

tor • interface region• ion focusing • mass separation • ion detection • sampling accessories• applications.

We hope that by the end of this series,we will have demystified ICP-MS, made it

TUTORIALTUTORIALS P E C T R O S C O P Y

A Beginner’s Guide to ICP-MSPart I

R O B E R T T H O M A S

Figure 1. Generation of positively charged ions in the plasma.

GENERATION OF IONS IN THE PLASMAWe’ll start this series off with a brief de-scription of the fundamental principleused in ICP-MS — the use of a high-temperature plasma discharge to gener-ate positively charged ions. The sample,

typically in liquid form, is pumped intothe sample introduction system, which ismade up of a spray chamber and nebu-lizer. It emerges as an aerosol and eventu-ally finds its way — by way of a sample in-jector — into the base of the plasma. As ittravels through the different heatingzones of the plasma torch it is dried, va-porized, atomized, and ionized. Duringthis time, the sample is transformed froma liquid aerosol to solid particles, theninto a gas. When it finally arrives at theanalytical zone of the plasma, at approxi-mately 6000–7000 K, it exists as excitedatoms and ions, representing the elemen-tal composition of the sample.The excitation of the outer electron ofa ground-state atom, to producewavelength-specific photons of light, isthe fundamental basis of atomic emission.However, there is also enough energy inthe plasma to remove an electron from itsorbital to generate an ion. It is the genera-tion, transportation, and detection of sig-nificant numbers of these positivelycharged ions that give ICP-MS its charac-teristic ultratrace detection capabilities.It is also important to mention that, although ICP-MS is predominantly usedfor the detection of positive ions, negativeions (such as halogens) are also pro-duced in the plasma. However, becausethe extraction and transportation of nega-tive ions is different from that of positiveions, most commercial instruments arenot designed to measure them. Theprocess of the generation of positivelycharged ions in the plasma is shown con-ceptually in greater detail in Figure 1.


a little more compelling to purchase, andultimately opened up its potential as aroutine tool to the vast majority of thetrace element community that has not yetrealized the full benefits of its capabilities.

S P E C T R O S C O P Y T U T O R I A L . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Circle 28

Figure 3. Conversion of a chromium ground-state atom (Cr0) to anion (Cr1).

Figure 2. Simplified schematic of a chromium ground-state atom(Cr0).

Figure 5. Relative abundance of the naturally occurring isotopes of all the elements (6). Reproduced with the permission of PerkinElmerInstruments (Norwalk, CT).

S P E C T R O S C O P Y T U T O R I A L. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Table I. Breakdown of the atomic structure of copper isotopes.

63Cu 65Cu

Number of protons (p ) 29 29Number of electrons (e ) 29 29Number of neutrons (n) 34 36Atomic mass (p n) 63 65Atomic number (p ) 29 29Natural abundance 69.17% 30.83%Nominal atomic weight 63.55*

* Calculated using the formulae 0.6917n 0.3083n p (referenced to theatomic weight of carbon)

Figure 4. Mass spectra of the two copper isotopes — 63Cu and65Cu .

APRIL 2001 16(4) SPECTROSCOPY 41

1 H 99.985 2 H 0.015 3 He 0.000137 4 He 99.999863 5

5.7iL6 5.29iL7

8 9 Be 10010 B 19.911 B 80.1

09.89C2101.1C31

14 N 99.64315 N 0.36616 O 99.76217 O 0.03818 O 0.200

001F9120 Ne 90.4821 Ne 0.2722 Ne 9.2523 Na 100

99.87gM4200.01gM5210.11gM62

27 Al 10028 Si 92.2329 Si 4.6730 Si 3.10

001P1332 S 95.0233 S 0.7534 S 4.2135 Cl 75.7736 S 0.02 Ar 0.33737 Cl 24.23

360.0rA8339 K 93.258140 K 0.0117 Ca 96.941 Ar 99.60041 K 6.730242 Ca 0.64743 Ca 0.13544 Ca 2.086

001cS5446 Ti 8.0 Ca 0.00447 Ti 7.348 Ti 73.8 Ca 0.18749 Ti 5.550 Ti 5.4 V 0.250 Cr 4.34551 V 99.750

987.38rC25105.9rC35

54 Fe 5.8 Cr 2.36555 Mn 10056 Fe 91.7257 Fe 2.258 Fe 0.28 Ni 68.07759 Co 100

322.62iN06

041.1iN16 436.3iN26

63 Cu 69.17 64 Zn 48.6 Ni 0.926 65 Cu 30.83 66 Zn 27.9 67 Zn 4.1 68 Zn 18.8

801.06aG96 70 Ge 21.23 Zn 0.6

298.93aG17 72 Ge 27.66 73 Ge 7.73 74 Ge 35.94 Se 0.89

001sA57 76 Ge 7.44 Se 9.36 77 Se 7.63 78 Kr 0.35 Se 23.78

96.05rB97 80 Kr 2.25 Se 49.61

13.94rB18 82 Kr 11.6 Se 8.73 83 Kr 11.5 84 Kr 57.0 Sr 0.56

561.27bR58 86 Kr 17.3 Sr 9.86 87 Sr 7.00 Rb 27.835 88 Sr 82.58

001Y98 90 Zr 51.45 91 Zr 11.22 92 Zr 17.15 Mo 14.84

001bN39 94 Zr 17.38 Mo 9.25 95 Mo 15.92 96 Zr 2.80 Mo 16.68 Ru 5.52 97 Mo 9.55 98 Mo 24.13 Ru 1.88

7.21uR99 100 Mo 9.63 Ru 12.6

0.71uR101102 Pd 1.02 Ru 31.6103 Rh 100104 Pd 11.14 Ru 18.7105 Pd 22.33106 Pd 27.33 Cd 1.25

938.15gA701108 Pd 26.46 Cd 0.89

161.84gA901110 Pd 11.72 Cd 12.49111 Cd 12.80112 Sn 0.97 Cd 24.13113 Cd 12.22 In 4.3114 Sn 0.65 Cd 28.73115 Sn 0.34 In 95.7116 Sn 14.53 Cd 7.49117 Sn 7.68118 Sn 24.23119 Sn 8.59120 Sn 32.59 Te 0.096

63.75bS121122 Sn 4.63 Te 2.603 123 Te 0.908 Sb 42.64124 Sn 5.79 Te 4.816 Xe 0.10125 Te 7.139126 Te 18.95 Xe 0.09127 I 100128 Te 31.69 Xe 1.91

4.62eX921130 Ba 0.106 Te 33.80 Xe 4.1

2.12eX131132 Ba 0.101 Xe 26.9133 Ce 100134 Ba 2.417 Xe 10.4135 Ba 6.592136 Ba 7.854 Ce 0.19 Xe 8.9137 Ba 11.23138 Ba 71.70 Ce 0.25 La 0.0902

8909.99aL931140 Ce 88.48

001rP141142 Nd 27.13 Ce 11.08143 Nd 12.18144 Nd 23.80 Sm 3.1145 Nd 8.30146 Nd 17.19147 Sm 15.0148 Nd 5.76 Sm 11.3149 Sm 13.8150 Nd 5.64 Sm 7.4

8.74uE151152 Gd 0.20 Sm 26.7

2.25uE351154 Gd 2.18 Sm 22.7155 Gd 14.80156 Gd 20.47 Dy 0.06157 Gd 15.65158 Gd 24.84 Dy 0.10

001bT951160 Gd 21.86 Dy 2.34161 Dy 18.9162 Er 0.14 Dy 25.5163 Dy 24.9164 Er 1.61 Dy 28.2

001oH561166 Er 33.6167 Er 22.95168 Er 26.8 Yb 0.13

001mT961170 Er 14.9 Yb 3.05171 Yb 14.3172 Yb 21.9173 Yb 16.12174 Yb 31.8 Hf 0.162175 Lu 97.41176 Lu 2.59 Yb 12.7 Hf 5.206

606.81fH771792.72fH871926.31fH971

180 Ta 0.012 W 0.13 Hf 35.100

181 Ta 99.988182 W 26.3 183 W 14.3184 Os 0.02 W 30.67

04.73eR581186 Os 1.58 W 28.6187 Os 1.6 Re 62.60188 Os 13.3189 Os 16.1190 Os 26.4 Pt 0.01191 Ir 37.3192 Os 41.0 Pt 0.79193 Ir 62.7

9.23tP4918.33tP591

196 Hg 0.15 Pt 25.3197 Au 100198 Hg 9.97 Pt 7.2199 Hg 16.87200 Hg 23.10201 Hg 13.18202 Hg 29.86

425.92lT302204 Hg 6.87 Pb 1.4

674.07lT502206 Pb 24.1207 Pb 22.1208 Pb 52.4209 Bi 100210211212213214215216217218219220221222223224225226227228229230231 Pa 100232 Th 100233234 U 0.0055235 U 0.7200236237238 U 99.2745

epotosIepotosIepotosIepotosI %%%%%%%%%%%%

Relative Abundance of the Natural Isotopes


ION FORMATIONFigures 2 and 3 show the actual processof conversion of a neutral ground-stateatom to a positively charged ion. Figure 2shows a very simplistic view of thechromium atom Cr0, consisting of a nu-cleus with 24 protons (p1) and 28 neu-trons (n), surrounded by 24 orbiting elec-trons (e2) (It must be emphasized thatthis is not meant to be an accurate repre-sentation of the electrons’ shells and sub-shells, but simply a conceptual explana-tion for the purpose of clarity). From thiswe can say that the atomic number ofchromium is 24 (number of protons), andits atomic mass is 52 (number of protons1 neutrons).

If energy is then applied to thechromium ground-state atom in the formof heat from a plasma discharge, one ofthe orbiting electrons will be stripped offthe outer shell. This will result in only 23electrons left orbiting the nucleus. Be-cause the atom has lost a negative charge(e2) but still has 24 protons (p1) in thenucleus, it is converted into an ion with anet positive charge. It still has an atomicmass of 52 and an atomic number of 24,but is now a positively charged ion andnot a neutral ground-state atom. Thisprocess is shown in Figure 3.

NATURAL ISOTOPESThis is a very basic look at the process,because most elements occur in morethan one form (isotope). In fact,chromium has four naturally occurringisotopes, which means that thechromium atom exists in four differentforms, all with the same atomic numberof 24 (number of protons), but with differ-ent atomic masses (numbers of neu-trons).

To make this a little easier to under-stand, let’s take a closer look at an ele-ment like copper, which has only two dif-ferent isotopes — one with an atomicmass of 63 (63Cu) and the other with anatomic mass of 65 (65Cu). They both havethe same number of protons and elec-trons, but differ in the number of neu-trons in the nucleus. The natural abun-dances of 63Cu and 65Cu are 69.1% and30.9%, respectively, which gives copper anominal atomic mass of 63.55 — thevalue you see for copper in atomic weightreference tables. Details of the atomicstructure of the two copper isotopes areshown in Table I.

When a sample containing naturally oc-curring copper is introduced into the

plasma, two different ions of copper,63Cu1 and 65Cu1, are produced, whichgenerate different mass spectra — one atmass 63 and another at mass 65. This canbe seen in Figure 4, which is an actualICP-MS spectral scan of a sample contain-ing copper. It shows a peak for the 63Cu1

ion on the left, which is 69.17% abundant,and a peak for 65Cu1 at 30.83% abun-dance, on the right. You can also seesmall peaks for two Zn isotopes at mass64 (64Zn) and mass 66 (66Zn) (Zn has a to-tal of five isotopes at masses 64, 66, 67,68, and 70). In fact, most elements haveat least two or three isotopes and manyelements, including zinc and lead, havefour or more isotopes. Figure 5 is a chartthat shows the relative abundance of thenaturally occurring isotopes of all theelements.

During the next few months, we willsystematically take you on a journeythrough the hardware of an ICP massspectrometer, explaining how each majorcomponent works, and finishing the se-ries with an overview of how the tech-nique is being used to solve real-world ap-plication problems. Our goal is to presentboth the basic principles and benefits ofthe technique in a way that is clear, con-cise, and very easy to understand. Wehope that by the end of the series, youand your managers will be in a better po-sition to realize the enormous benefitsthat ICP-MS can bring to your laboratory.

REFERENCES(1) A. Montasser, Inductively Coupled Plasma

Mass Spectrometry (Wiley-VCH, Berlin,1998).

(2) F. Adams, R. Gijbels, and R. Van Grieken,Inorganic Mass Spectrometry (John Wileyand Sons, New York, 1988.).

(3) R.S. Houk, V. A. Fassel, and H.J. Svec, Dy-namic Mass Spectrom. 6, 234 (1981).

(4) A.R. Date and A.L. Gray, Analyst 106,1255 (1981).

(5) D.J. Douglas and J.B. French, Anal.Chem. 53, 37 (1982).

(6) Isotopic Composition of the Elements: PureApplied Chemistry 63(7), 991–1002(1991).

Robert Thomas is the principal of his ownfreelance writing and scientific consultingcompany, Scientific Solutions, based inGaithersburg, MD. He can be contacted byemail at [email protected] or viahis web site at www.scientificsolutions1.com.◆

S P E C T R O S C O P Y T U T O R I A L . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . .

Circle 29

56 SPECTROSCOPY 16(5) MAY 2001 www.spec t roscopyo n l i n e . com

Part II of Robert Thomas’ series on induc-tively coupled plasma mass spectrometrylooks at one of the most critical areas of theinstrument — the sample introduction sys-tem. He discusses the fundamental princi-ples of converting a liquid into a fine-droplet aerosol suitable for ionization inthe plasma, and provides an overview ofthe different types of commercially avail-able nebulizers and spray chambers.

The majority of inductively coupledplasma mass spectrometry (ICP-MS) applications involve the analysisof liquid samples. Even though spec-troscopists adapted the technique

over the years to handle solids, it was de-veloped in the early 1980s primarily to an-alyze solutions. There are many ways ofintroducing a liquid into an ICP massspectrometer, but they all basicallyachieve the same result — they generatea fine aerosol of the sample so it can beefficiently ionized in the plasma dis-charge. The sample-introduction area hasbeen called the Achilles heel of ICP-MSbecause it is considered the weakestcomponent of the instrument, with only1–2% of the sample finding its way intothe plasma (1). Although there has re-cently been much improvement in thisarea, the fundamental design of an ICP-MS sample introduction system has notdramatically changed since the techniquewas first introduced in 1983.

Before discussing the mechanics ofaerosol generation in greater detail, let uslook at the basic components of a sampleintroduction system. Figure 1 shows theproximity of the sample introduction arearelative to the rest of the ICP mass spec-trometer, while Figure 2 represents theindividual components.

The mechanism of introducing a liquidsample into analytical plasma can be con-sidered as two separate events — aerosol

generation using a nebulizer and dropletselection by way of a spray chamber.Sharp carried out a thorough investiga-tion of both processes (2).

AEROSOL GENERATIONAs mentioned previously, the main func-tion of the sample introduction system isto generate a fine aerosol of the sample. Itachieves this purpose with a nebulizerand a spray chamber. The sample is nor-mally pumped at ~1 mL/min via a peri-staltic pump into the nebulizer. A peri-staltic pump is a small pump with lots ofminirollers that rotate at the same speed.The constant motion and pressure of therollers on the pump tubing feed the sam-ple to the nebulizer. The benefit of a peri-staltic pump is that it ensures a constantflow of liquid, irrespective of differencesin viscosity between samples, standards,and blanks. After the sample enters thenebulizer, the liquid is broken up into afine aerosol by the pneumatic action of

gas flow (~1 L/min) smashing the liquidinto tiny droplets, which is very similar tothe spray mechanism of a can of deodor-ant. Although pumping the sample is themost common approach to introducing it,some pneumatic nebulizers, such as theconcentric design, don’t need a pump be-cause they rely on the natural venturi ef-fect of the positive pressure of the nebu-lizer gas to suck the sample through thetubing. Solution nebulization is conceptu-ally represented in Figure 3, which showsaerosol generation using a nebulizer witha crossflow design.

DROPLET SELECTIONBecause the plasma discharge is ineffi-cient at dissociating large droplets, thespray chamber’s function is primarily toallow only the small droplets to enter theplasma. Its secondary purpose is tosmooth out pulses that occur during thenebulization process, due mainly to theperistaltic pump. Several ways exist to en-


A Beginner’s Guide to ICP-MSPart II: The Sample-Introduction System


Figure 1. ICP-MS system diagram showing the location of the sample introduction area.

MAY 2001 16(5) SPECTROSCOPY 57

(4). Therefore, general-purpose ICP-OESnebulizers that are designed to aspirate1–2% dissolved solids, or high-solids neb-ulizers such as the Babbington, V-groove,or cone-spray nebulizers, which are de-signed to handle as much as 20% dis-solved solids, are not ideal for use withICP-MS. The most common of the pneu-matic nebulizers used in commercial ICPmass spectrometers are the concentricand crossflow designs. The concentric de-sign is more suitable for clean samples,while the crossflow is generally more tol-erant to samples containing higher levelsof solids or particulate matter.

Concentric design. In the concentric neb-ulizer, the solution is introduced througha capillary tube to a low-pressure regioncreated by a gas flowing rapidly past theend of the capillary. The low pressure andhigh-speed gas combine to break up thesolution into an aerosol, which forms atthe open end of the nebulizer tip. Figure5 illustrates the concentric design.

Concentric pneumatic nebulizers canprovide excellent sensitivity and stability,particularly with clean solutions. How-ever, the small orifices can be plagued byblockage problems, especially if largenumbers of heavy matrix samples areaspirated.

Crossflow design. For samples that con-tain a heavier matrix or small amounts ofundissolved matter, the crossflow designis probably the best option. With this de-sign the argon gas is directed at right an-gles to the tip of a capillary tube, in con-trast to the concentric design, where thegas flow is parallel to the capillary. Thesolution is either drawn up through thecapillary tube via the pressure created bythe high-speed gas flow or, as is most

common with crossflow nebulizers,forced through the tube with a peristalticpump. In either case, contact between thehigh-speed gas and the liquid streamcauses the liquid to break up into anaerosol. Crossflow nebulizers are gener-ally not as efficient as concentric nebuliz-ers at creating the very small dropletsneeded for ICP-MS analyses. However,the larger diameter liquid capillary andlonger distance between liquid and gasinjectors reduce clogging problems.Many analysts feel that the small penaltypaid in analytical sensitivity and precisionwhen compared with concentric nebuliz-ers is compensated by the fact that thecrossflow design is far more rugged forroutine use. Figure 6 shows a cross sec-tion of a crossflow nebulizer.

Microflow design. A new breed of nebu-lizers is being developed for ICP-MScalled microflow nebulizers, which aredesigned to operate at much lower sam-ple flows. While conventional nebulizershave a sample uptake rate of about1 mL/min, microflow nebulizers typicallyrun at less than 0.1 mL/min. They arebased on the concentric principle, but

sure only the small droplets get through,but the most common way is to use adouble-pass spray chamber where theaerosol emerges from the nebulizer andis directed into a central tube running thewhole length of the chamber. Thedroplets travel the length of this tube,where the large droplets (greater than~10 µm in diameter) fall out by gravityand exit through the drain tube at the endof the spray chamber. The fine droplets(~5–10 µm in diameter) then pass be-tween the outer wall and the central tube,where they eventually emerge from thespray chamber and are transported intothe sample injector of the plasma torch(3). Although many different designs areavailable, the spray chamber’s main func-tion is to allow only the smallest dropletsinto the plasma for dissociation, atomiza-tion, and finally ionization of the sample’selemental components. Figure 4 presentsa simplified schematic of this process.

Let us now look at the different nebu-lizer and spray chamber designs that aremost commonly used in ICP-MS. This ar-ticle cannot cover every type available be-cause a huge market has developed overthe past few years for application-specificcustomized sample introduction compo-nents. This market created an industry ofsmall OEM (original equipment manufac-turers) companies that manufacture partsfor instrument companies as well as sell-ing directly to ICP-MS users.

NEBULIZERSBy far the most common design used forICP-MS is the pneumatic nebulizer,which uses mechanical forces of a gasflow (normally argon at a pressure of20–30 psi) to generate the sampleaerosol. The most popular designs ofpneumatic nebulizers include concentric,microconcentric, microflow, and cross-flow. They are usually made from glass,but other nebulizer materials, such asvarious kinds of polymers, are becomingmore popular, particularly for highly cor-rosive samples and specialized applica-tions. I want to emphasize at this pointthat nebulizers designed for use with ICP-optical emission spectroscopy (OES) arenot recommended for ICP-MS. This factresults from a limitation in total dissolvedsolids (TDS) that can be put into the ICP-MS interface area. Because the orificesizes of the sampler and skimmer conesused in ICP-MS are so small (~0.6–1.2mm), the concentration of matrix compo-nents must generally be kept below 0.2%


Figure 2. Diagram of the ICP-MS sampleintroduction area.

Figure 3. Conceptual representation ofaerosol generation with an ICP-MSnebulizer.

Figure 4. Simplified representation of theseparation of large and fine droplets in thespray chamber.

www.spec t roscopyo n l i n e . com

they usually operate at higher gas pres-sure to accommodate the lower sampleflow rates. The extremely low uptake ratemakes them ideal for applications withlimited sample volume or where the sam-ple or analyte is prone to sample intro-duction memory effects. These nebuliz-ers and their components are typically

constructed from polymer materials suchas polytetrafluoroethylene (PTFE), per-fluoroalkoxy (PFA), or polyvinylidene flu-oride (PVDF). In fact, their excellent cor-rosion resistance means that they havenaturally low blank levels. This character-istic, together with their ability to handlesmall sample volumes such as vapor-phase decomposition (VPD) applications,makes them an ideal choice for semicon-ductor labs that are carrying out ultra-trace element analysis (5). A typical mi-croflow nebulizer made from PFA isshown in Figure 7.

SPRAY CHAMBERSLet us now turn our attention to spraychambers. Basically two designs are usedin commercial ICP-MS instrumentation— double pass and cyclonic spray cham-bers. The double pass is by far the most

common, with the cyclonic type gainingin popularity. Another type of spray cham-ber based on the impact bead design(first developed for flame AA and thenadapted for ICP-OES) was tried on theearly ICP-MS systems with limited suc-cess, but is not generally used today. Asmentioned earlier, the function of thespray chamber is to reject the largeraerosol droplets and also to smooth outpulses produced by the peristaltic pump.In addition, some ICP-MS spray cham-bers are externally cooled (typically to2–5 °C) for thermal stability of the sam-ple and to minimize the amount of solventgoing into the plasma. This can have anumber of beneficial effects, dependingon the application, but the main benefitsare reduction of oxide species and theability to aspirate volatile organicsolvents.


Figure 5. Diagram of a typical concentric nebulizer. Figure 6. Schematic of a crossflow nebulizer.

Figure 7. A typical concentric microflownebulizer. Printed with permission from Ele-mental Scientific (Omaha, NE).

Figure 8. Schematic of a Scott double-pass spray chamber (shownwith crossflow nebulizer). Printed with permission of PerkinElmerInstruments (Norwalk, CT).

Figure 9. Schematic of a cyclonic spray chamber (shown withconcentric nebulizer).

58 SPECTROSCOPY 16(5) MAY 2001

60 SPECTROSCOPY 16(5) MAY 2001 www.spec t roscopyo n l i n e . com

Double pass. By far the most commondesign of double-pass spray chamber isthe Scott design, which selects the smalldroplets by directing the aerosol into acentral tube. The larger droplets emergefrom the tube and, by gravity, exit thespray chamber via a drain tube. The liq-uid in the drain tube is kept at positivepressure (usually by way of a loop),which forces the small droplets back be-tween the outer wall and the central tube,where they emerge from the spray cham-ber into the sample injector of the plasmatorch. Scott double-pass spray chamberscome in a variety of shapes, sizes, andmaterials, but are generally consideredthe most rugged design for routine use.Figure 8 shows a Scott spray chambermade of a polysulfide-type material, cou-pled to a crossflow nebulizer.

Cyclonic spray chamber. The cyclonicspray chamber operates by centrifugalforce. Droplets are discriminated accord-ing to their size by means of a vortex pro-duced by the tangential flow of the sam-ple aerosol and argon gas inside thechamber. Smaller droplets are carried

with the gas stream into the ICP-MS,while the larger droplets impinge on thewalls and fall out through the drain. It isgenerally accepted that a cyclonic spraychamber has a higher sampling effi-ciency, which, for clean samples, trans-lates into higher sensitivity and lower de-tection limits. However, the droplet sizedistribution appears to be different from adouble-pass design, and for certain typesof samples, can give slightly inferior pre-cision. An excellent evaluation of the ca-pabilities of a cyclonic spray chamber wasmade by Beres and co-workers (6). Fig-ure 9 shows a cyclonic spray chamberconnected to a concentric nebulizer.

Many other nonstandard sample intro-duction devices are available that are notdescribed in this particular tutorial, suchas ultrasonic nebulization, membrane de-solvation, flow injection, direct injection,electrothermal vaporization, and laser ab-lation. However, they are becoming moreand more important, particularly as ICP-MS users are demanding higher perfor-mance and more flexibility. For that rea-son, they will be addressed in a separate

Circle 51

tutorial at the end of this series.

REFERENCES(1) R. A. Browner and A.W. Boorn, Anal.

Chem. 56, 786–798A (1984).(2) B.L. Sharp, Analytical Atomic Spectrome-

try 3, 613 (1980).(3) L.C. Bates and J.W. Olesik, Journal of An-

alytical Atomic Spectrometry 5(3), 239(1990).

(4) R.S. Houk, Anal. Chem. 56, 97A (1986).(5) E. Debrah, S. A. Beres, T.J. Gluodennis,

R.J. Thomas, and E.R. Denoyer, AtomicSpectroscopy, 197–202 (September 1995).

(6) S. A. Beres, P. H. Bruckner, and E.R. De-noyer, Atomic Spectroscopy, 96–99(March/April 1994).

Robert Thomas is the principal of his ownfreelance writing and scientific marketingconsulting company, Scientific Solutions,based in Gaithersburg, MD. He specializesin trace element analysis and can be con-tacted by e-mail at [email protected] or via his web site at www.scientificsolutions1.com. ◆

TRAINING COURSESThermo Nicolet (Madison, WI) is offering afree Spring 2001 Spectroscopic SolutionsSeminar Series. The seminars will coverbasic FT-IR spectroscopy, microspec-troscopy, dispersive Raman microscopy,Raman spectroscopy, and specializedsampling techniques. This year’s sched-ule includes the following seminars:

May 22 at the Syracuse Sheraton,Syracuse, NY; May 24 at the Wynd-ham Westborough Hotel, Westbor-ough, MA; May 24 at the MarriottOak Brook, Oak Brook, IL; June 5 atthe East Lansing Marriott, East Lans-ing, MI; June 5 at the Embassy Suites,Overland Park, KS; June 7 at the Sher-aton Indianapolis, Indianapolis, IN;June 12 at the Delta Meadowvale Conference Center, Mississauga, Ontario, Canada; June 12 at the Em-bassy Suites, Brookfield, WI; July 10at the Coast Terrace Inn, Edmonton,Alberta, Canada; and July 26 at the AlaMoana Hotel, Honolulu, HI.For more information, contact Thermo

Nicolet, (800) 201-8132 fax: (608) 273-5046, e-mail: [email protected], web site: www.thermonicolet.com. ◆

“News Spectrum” continued from page 13


26 SPECTROSCOPY 16(6) JUNE 2001 www.spec t roscopyo n l i n e . com

Part III of Robert Thomas’ series on induc-tively coupled plasma–mass spectroscopy(ICP-MS) looks at the area where the ionsare generated — the plasma discharge. Hegives a brief historical perspective of someof the common analytical plasmas usedover the years and discusses the compo-nents that are used to create the ICP. Hefinishes by explaining the fundamentalprinciples of formation of a plasma dis-charge and how it is used to convert thesample aerosol into a stream of positivelycharged ions.

Inductively coupled plasmas are by farthe most common type of plasmasources used in today’s commercialICP–optical emission spectrometry(OES) and ICP-MS instrumentation.

However, it wasn’t always that way. In theearly days, when researchers were at-tempting to find the ideal plasma sourceto use for spectrometric studies, it wasunclear which approach would prove tobe the most successful. In addition toICPs, some of the other novel plasmasources developed were direct currentplasmas (DCP) and microwave-inducedplasmas (MIP). A DCP is formed when agas (usually argon) is introduced into ahigh current flowing between two orthree electrodes. Ionization of the gasproduces an inverted Y-shaped plasma.Unfortunately, early DCP instrumenta-tion was prone to interference effects andalso had some usability and reliabilityproblems. For these reasons, the tech-nique never became widely accepted bythe analytical community (1). However,its one major benefit was that it could as-pirate high levels of dissolved or sus-pended solids, because there was no re-strictive sample injector for the solidmaterial to block. This feature alonemade it attractive for some laboratories,and once the initial limitations of DCPs

were better understood, the techniquebecame more accepted. In fact, for thosewho want a DCP excitation source cou-pled with an optical emission instrumenttoday, an Echelle-based grating using asolid-state detector is commerciallyavailable (2).

Limitations in the DCP approach led tothe development of electrodeless plasma,of which the MIP was the simplest form.In this system, microwave energy (typi-cally 100–200 W) is supplied to the plasmagas from an excitation cavity around aglass or quartz tube. The plasma dis-charge in the form of a ring is generatedinside the tube. Unfortunately, eventhough the discharge achieves a veryhigh power density, the high excitationtemperatures exist only along a central fil-ament. The bulk of the MIP never getshotter than 2000–3000 K, which means itis prone to very severe matrix effects. Inaddition, they are easily extinguished dur-

ing aspiration of liquid samples. For thesereasons, they have had limited success asan emission source, because they are notconsidered robust enough for the analysisof real-world, solution-based samples.However, they have gained acceptance asan ion source for mass spectrometry (3)and also as emission-based detectors forgas chromatography.

Because of the limitations of the DCPand MIP approaches, ICPs became thedominant focus of research for both opti-cal emission and mass spectrometricstudies. As early as 1964, Greenfield andco-workers reported that an atmospheric-pressure ICP coupled with OES could beused for elemental analysis (4). Althoughcrude by today’s standards, the systemshowed the enormous possibilities of theICP as an excitation source and most defi-nitely opened the door in the early 1980sto the even more exciting potential of us-ing the ICP to generate ions (5).


A Beginner’s Guide to ICP-MSPart III: The Plasma Source


Iondetector

MSinterface

ICP torchIon optics

Spraychamber

Nebulizer

Mass separationdevice

Turbomolecularpump

Turbomolecularpump

Mechanicalpump

Radio frequencypower supply

Figure 1. Schematic of an ICP-MS system showing the location of the plasma torch and radiofrequency (RF) power supply.


nents that are used to generate thesource: a plasma torch, a radio frequency(RF) coil, and RF power supply. Figure 1shows their proximity to the rest of theinstrument; Figure 2 is a more detailedview of the plasma torch and RF coil rela-tive to the MS interface.

The plasma torch consists of three con-centric tubes, which are usually madefrom quartz. In Figure 2, these are shownas the outer tube, middle tube, and sam-ple injector. The torch can either be one-piece with all three tubes connected, or itcan be a demountable design in whichthe tubes and the sample injector are sep-arate. The gas (usually argon) used toform the plasma (plasma gas) is passedbetween the outer and middle tubes at aflow rate of ;12–17 L/min. A second gasflow, the auxiliary gas, passes betweenthe middle tube and the sample injectorat ;1 L/min and is used to change theposition of the base of the plasma relativeto the tube and the injector. A third gasflow, the nebulizer gas, also flowing at;1 L/min carries the sample, in the formof a fine-droplet aerosol, from the sampleintroduction system (for details, see PartII of this series: Spectroscopy 16[5],56–60 [2001]) and physically punches achannel through the center of the plasma.The sample injector is often made frommaterials other than quartz, such as alu-mina, platinum, and sapphire, if highlycorrosive materials need to be analyzed.It is worth mentioning that although ar-gon is the most suitable gas to use for allthree flows, there are analytical benefitsin using other gas mixtures, especially inthe nebulizer flow (6). The plasma torch

THE PLASMA TORCHBefore we take a look at the fundamentalprinciples behind the creation of an in-ductively coupled plasma used in ICP-MS, let us take a look at the basic compo-


Plasma

Nebulizer gas

Interface Outertube

RFcoil

Sample injector

RF power

Middletube

Plasmagas

Auxiliarygas

Figure 2. Detailed view of a plasma torchand RF coil relative to the ICP-MS interface.

Quartztorch

Collision-inducedionization of argon

Formation of inductivelycoupled plasma

Sample introducedthrough sample injector

Loadcoil

Tangential flowof argon gas

Electromagneticfield High voltage

spark

(a) (b) (c)

(d) (e)

Figure 3. (right) Schematic of an ICP torchand load coil showing how the inductivelycoupled plasma is formed. (a) A tangentialflow of argon gas is passed between theouter and middle tube of the quartz torch.(b) RF power is applied to the load coil,producing an intense electromagnetic field.(c) A high-voltage spark produces free elec-trons. (d) Free electrons are accelerated bythe RF field, causing collisions and ioniza-tion of the argon gas. (e) The ICP is formedat the open end of the quartz torch. Thesample is introduced into the plasma viathe sample injector.

Circle 21

JUNE 2001 16(6) SPECTROSCOPY 29

Initialradiationzone

Normalanalyticalzone

10,000 K

Preheatingzone

6500 K

6000 K

7500 K 8000 K

Figure 4. Different temperature zones in the plasma.

Droplet (Desolvation) Solid (Vaporization) Gas (Atomization) Atom (Ionization) Ion

From sample injector To mass spectrometer

M(H2O)1 X2 (MX)n MX M M1

Figure 5. Mechanism of conversion of a droplet to a positive ion in the ICP.

is mounted horizontally and positionedcentrally in the RF coil, approximately10–20 mm from the interface. It must beemphasized that the coil used in an ICP-MS plasma is slightly different from theone used in ICP-OES. In all plasmas,there is a potential difference of a fewhundred volts produced by capacitivecoupling between the RF coil and theplasma. In an ICP mass spectrometer,this would result in a secondary dis-charge between the plasma and the inter-face cone, which could negatively affectthe performance of the instrument. Tocompensate for this, the coil must begrounded to keep the interface region asclose to zero potential as possible. I willdiscuss the full implications of this ingreater detail in Part IV of this series.

FORMATION OF AN ICP DISCHARGELet us now discuss the mechanism of for-mation of the plasma discharge. First, atangential (spiral) flow of argon gas is di-rected between the outer and middle tubeof a quartz torch. A load coil, usually cop-per, surrounds the top end of the torchand is connected to a radio frequencygenerator. When RF power (typically750–1500 W, depending on the sample) isapplied to the load coil, an alternatingcurrent oscillates within the coil at a ratecorresponding to the frequency of thegenerator. In most ICP generators thisfrequency is either 27 or 40 MHz. ThisRF oscillation of the current in the coilcauses an intense electromagnetic field tobe created in the area at the top of thetorch. With argon gas flowing throughthe torch, a high-voltage spark is appliedto the gas, which causes some electronsto be stripped from their argon atoms.These electrons, which are caught up andaccelerated in the magnetic field, thencollide with other argon atoms, strippingoff still more electrons. This collision-

induced ionization of the argon continuesin a chain reaction, breaking down thegas into argon atoms, argon ions, andelectrons, forming what is known as aninductively coupled plasma discharge.The ICP discharge is then sustainedwithin the torch and load coil as RF en-ergy is continually transferred to itthrough the inductive coupling process.The sample aerosol is then introducedinto the plasma through a third tubecalled the sample injector. This wholeprocess is conceptionally shown inFigure 3.

THE FUNCTION OF THE RF GENERATORAlthough the principles of an RF powersupply have not changed since the workof Greenfield (4), the components havebecome significantly smaller. Some of theearly generators that used nitrogen or airrequired 5–10 kW of power to sustain theplasma discharge — and literally took uphalf the room. Most of today’s generatorsuse solid-state electronic components,which means that vacuum power ampli-fier tubes are no longer required. Thismakes modern instruments significantlysmaller and, because vacuum tubes werenotoriously unreliable and unstable, farmore suitable for routine operation.

As mentioned previously, two frequen-cies have typically been used for ICP RFgenerators: 27 and 40 MHz. These fre-quencies have been set aside specificallyfor RF applications of this kind, so theywill not interfere with other communica-tion-based frequencies. The early RF gen-erators used 27 MHz, while the more re-cent designs favor 40 MHz. Thereappears to be no significant analytical ad-vantage of one type over the other. How-ever, it is worth mentioning that the 40-MHz design typically runs at lower powerlevels, which produces lower signal inten-sity and reduced background levels. Be-

cause it uses slightly lower power, thismight be considered advantageous when it comes to long-term use of thegenerator.

The more important consideration isthe coupling efficiency of the RF genera-tor to the coil. The majority of modernsolid-state RF generators are on the orderof 70–75% efficient, meaning that 70–75%of the delivered power actually makes itinto the plasma. This wasn’t always thecase, and some of the older vacuumtube–designed generators were notori-ously inefficient; some of them experi-enced more than a 50% power loss. An-other important criterion to consider isthe way the matching network compen-sates for changes in impedance (a mater-ial’s resistance to the flow of an electriccurrent) produced by the sample’s matrixcomponents or differences in solventvolatility. In older crystal-controlled gen-erators, this was usually done with servo-driven capacitors. They worked very wellwith most sample types, but because theywere mechanical devices, they struggledto compensate for very rapid impedancechanges produced by some samples. As aresult, the plasma was easily extin-guished, particularly during aspiration ofvolatile organic solvents.

These problems were partially over-come by the use of free-running RF gen-erators, in which the matching networkwas based on electronic tuning of smallchanges in frequency brought about bythe sample solvent or matrix components.The major benefit of this approach wasthat compensation for impedancechanges was virtually instantaneous be-cause there were no moving parts. Thisallowed for the successful analysis ofmany sample types that would probablyhave extinguished the plasma of acrystal-controlled generator.



IONIZATION OF THE SAMPLETo better understand what happens to thesample on its journey through the plasmasource, it is important to understand thedifferent heating zones within the dis-charge. Figure 4 shows a cross-sectionalrepresentation of the discharge alongwith the approximate temperatures fordifferent regions of the plasma.

zone before it eventually becomes a posi-tively charged ion in the analytical zone.To explain this in a very simplistic way,let’s assume that the element exists as atrace metal salt in solution. The first stepthat takes place is desolvation of thedroplet. With the water moleculesstripped away, it then becomes a verysmall solid particle. As the sample movesfurther into the plasma, the solid particlechanges first into a gaseous form andthen into a ground-state atom. The finalprocess of conversion of an atom to anion is achieved mainly by collisions of en-ergetic argon electrons (and to a lesserextent by argon ions) with the ground-state atom (7). The ion then emergesfrom the plasma and is directed into theinterface of the mass spectrometer (fordetails on the mechanisms of ion genera-tion, please refer to Part I of this series:Spectroscopy 16[4], 38–42 [2001]). Thisprocess of conversion of droplets intoions is represented in Figure 5.

The next installment of this series willfocus on probably the most crucial areaof an ICP mass spectrometer — the inter-face region — where the ions generatedin the atmospheric plasma have to besampled with consistency and electricalintegrity by the mass spectrometer,which is under extremely high vacuum.

REFERENCES(1) A.L. Gray, Analyst 100, 289–299 (1975).(2) G.N. Coleman, D.E. Miller, and R.W.

Stark, Am. Lab. 30(4), 33R (1998).(3) D.J. Douglas and J.B. French, Anal.

Chem. 53, 37-41 (1981).(4) S. Greenfield, I.L. Jones, and C.T. Berry,

Analyst 89, 713–720 (1964).(5) R.S. Houk, V. A. Fassel, and H.J. Svec,

Dyn. Mass Spectrom. 6, 234 (1981).(6) J.W. Lam and J.W. McLaren, J. Anal.

Atom. Spectom. 5, 419–424 (1990).(7) T. Hasegawa and H. Haraguchi, ICPs in

Analytical Atomic Spectrometry, A.Montaser and D.W. Golightly, Eds., 2d ed.(VCH, New York, 1992).

Robert Thomas is the principal of his ownfreelance writing and scientific marketingconsulting company, Scientific Solutions,based in Gaithersburg, MD. He specializesin trace-element analysis and can be con-tacted by e-mail at [email protected] or via his web site at www.scientificsolutions1.com. ◆


As mentioned previously, the sampleaerosol enters the injector via the spraychamber. When it exits the sample injec-tor, it is moving at such a velocity that itphysically punches a hole through thecenter of the plasma discharge. It thengoes through a number of physicalchanges, starting at the preheating zoneand continuing through the radiation

Circle 22

www.spec t roscopyo n l i n e . com

The interface region is probably themost critical area of the whole induc-tively coupled plasma mass spec-trometry (ICP-MS) system. It cer-tainly gave the early pioneers of the

technique the most problems to over-come. Although we take all the benefitsof ICP-MS for granted, the process of tak-ing a liquid sample, generating an aerosolthat is suitable for ionization in theplasma, and then sampling a representa-tive number of analyte ions, transportingthem through the interface, focusingthem via the ion optics into the massspectrometer, finally ending up with de-tection and conversion to an electronicsignal, are not trivial tasks. Each part ofthe journey has its own unique problemsto overcome but probably the most chal-lenging is the movement of the ions fromthe plasma to the mass spectrometer.Let’s begin by explaining how the ion-

sampling process works, which will givereaders an insight into the many prob-lems faced by the early researchers.

SAMPLING THE IONSFigure 1 shows the proximity of the inter-face region to the rest of the instrument.The role of the interface is to transportthe ions efficiently, consistently, and withelectrical integrity from the plasma,which is at atmospheric pressure (760Torr), to the mass spectrometer analyzerregion, which is at approximately 1026

Torr. One first achieves this by directingthe ions into the interface region. The in-terface consists of two metallic coneswith very small orifices, which are main-tained at a vacuum of ;2 Torr with a me-chanical roughing pump. After the ionsare generated in the plasma, they passthrough the first cone, known as the sam-pler cone, which has an orifice diameter

of 0.8–1.2 mm. From there they travel ashort distance to the skimmer cone,which is generally sharper than the sam-pler cone and has a much smaller orifice(0.4–0.8 mm i.d.). Both cones are usuallymade of nickel, but they can be made ofmaterials such as platinum that are farmore tolerant to corrosive liquids. To re-duce the effects of the high-temperatureplasma on the cones, the interface hous-ing is water-cooled and made from a ma-terial that dissipates heat easily, such ascopper or aluminum. The ions thenemerge from the skimmer cone, wherethey are directed through the ion optics,and finally are guided into the mass sepa-ration device. Figure 2 shows the inter-face region in greater detail; Figure 3shows a close-up of the sampler andskimmer cones.

CAPACITIVE COUPLING This process sounds fairly straight-forward but proved very problematic dur-ing the early development of ICP-MS be-cause of an undesired electrostatic(capacitive) coupling between the loadcoil and the plasma discharge, producinga potential difference of 100–200 V. Al-though this potential is a physical charac-teristic of all inductively coupled plasmadischarges, it is particularly serious in anICP mass spectrometer because the ca-pacitive coupling creates an electrical dis-charge between the plasma and the sam-pler cone. This discharge, commonlycalled the pinch effect or secondary dis-charge, shows itself as arcing in the re-gion where the plasma is in contact withthe sampler cone (1). This process isshown very simplistically in Figure 4.

If not taken care of, this arcing cancause all kinds of problems, including anincrease in doubly charged interferingspecies, a wide kinetic energy spread ofsampled ions, formation of ions gener-


A Beginner’s Guide to ICP-MSPart IV: The Interface Region


Iondetector

MSinterface

ICP torchIon optics

Spraychamber

Nebulizer


Turbomolecularpump

Turbomolecularpump

Mechanicalpump

Radio power supply

Figure 1. Schematic of an inductively coupled plasma mass spectrometry (ICP-MS) system,showing the proximity of the interface region.

26 SPECTROSCOPY 16(7) JULY 2001

ated from the sampler cone, and a de-creased orifice lifetime. These problemswere reported by many of the early re-searchers of the technique (2, 3). In fact,because the arcing increased with sam-pler cone orifice size, the source of thesecondary discharge was originallythought to be the result of an electro-gas-dynamic effect, which produced an in-crease in electron density at the orifice(4). After many experiments it was even-tually realized that the secondary dis-charge was a result of electrostatic cou-pling of the load coil to the plasma. Theproblem was first eliminated by ground-ing the induction coil at the center, whichhad the effect of reducing the radio fre-quency (RF) potential to a few volts. Thiseffect can be seen in Figure 5, taken fromone of the early papers, which shows thereduction in plasma potential as the coil isgrounded at different positions (turns)along its length.

Originally, the grounding was imple-mented by attaching a physical ground-ing strap from the center turn of the coilto the interface housing. In today’s instru-mentation the grounding is achieved in anumber of different ways, depending onthe design of the interface. Some of themost popular designs include balancingthe oscillator inside the circuitry of theRF generator (5); positioning a groundedshield or plate between the coil and theplasma torch (6); or using two interlacedcoils where the RF fields go in opposingdirections (7). They all work differentlybut achieve a similar result of reducing oreliminating the secondary discharge.

ION KINETIC ENERGYThe impact of a secondary discharge can-

not be overestimated with respect to its ef-fect on the kinetic energy of the ions beingsampled. It is well documented that the en-ergy spread of the ions entering the massspectrometer must be as low as possible toensure that they can all be focused effi-ciently and with full electrical integrity bythe ion optics and the mass separation de-vice. When the ions emerge from the ar-gon plasma, they will all have different ki-netic energies based on their mass-to-change ratio. Their velocities should all besimilar because they are controlled byrapid expansion of the bulk plasma, whichwill be neutral as long as it is maintained atzero potential. As the ion beam passesthrough the sampler cone into the skim-mer cone, expansion will take place, but itscomposition and integrity will be main-tained, assuming the plasma is neutral.This can be seen in Figure 6.

Electrodynamic forces do not play arole as the ions enter the sampler or theskimmer because the distance overwhich the ions exert an influence on eachother (known as the Debye length) issmall (typically 1023–1024 mm) com-pared with the diameter of the orifice(0.5–1.0 mm) (8), as Figure 7 shows.

It is therefore clear that maintaining aneutral plasma is of paramount impor-tance to guarantee electrical integrity ofthe ion beam as it passes through the in-terface region. If a secondary dischargeis present, it changes the electrical char-acteristics of the plasma, which will affectthe kinetic energy of the ions differently,depending on their mass. If the plasma isat zero potential, the ion energy spread isin the order of 5–10 eV. However, if a sec-ondary discharge is present, it results ina much wider spread of ion energies en-

tering the mass spectrometer (typically20–40 eV), which makes ion focusing farmore complicated (8).

BENEFITS OF A WELL-DESIGNEDINTERFACEThe benefits of a well-designed interfaceare not readily obvious if simple aqueoussamples are analyzed using only one setof operating conditions. However, it be-comes more apparent when many differ-ent sample types are being analyzed, re-quiring different operating parameters. Atrue test of the design of the interface oc-curs when plasma conditions need to bechanged, when the sample matrixchanges, or when a dry sample aerosol isbeing introduced into the ICP-MS. Ana-lytical scenarios like these have the po-tential to induce a secondary discharge,change the kinetic energy of the ions en-tering the mass spectrometer, and affectthe tuning of the ion optics. It is therefore


Plasmatorch

Ionoptics

~2 Torrvacuum

SampleconeSkimmer

cone

RF coil

Figure 2. Detailed view of the interfaceregion.

Interfacehousing

Samplercone

Skimmercone

Figure 3. Close-up view of the sampler andskimmer cones. (Courtesy PerkinElmer Instruments, Norwalk, CT.)

Secondarydischarge

Figure 4. Interface area affected by sec-ondary discharge.

100

80

60

40

20

00.0 0.5 1.0 1.5

Load coil turns

Plas

ma

pote

ntia

l

2.0 2.5 3.0

n

n

n

n

n

n

n

Figure 5. Reduction in plasma potential asthe load coil is grounded at different posi-tions (turns) along its length.

JULY 2001 16(7) SPECTROSCOPY 27

28 SPECTROSCOPY 16(7) JULY 2001 www.spec t roscopyo n l i n e . com

critical that the interface groundingmechanism can handle these types ofreal-world applications, of which typicalexamples include• The use of cool-plasma conditions. It isstandard practice today to use cool-plasma conditions (500–700 W power and1.0–1.3 L/min nebulizer gas flow) tolower the plasma temperature and reduceargon-based polyatomic interferences

such as 40Ar16O, 40Ar, and 38ArH, in thedetermination of difficult elements like56Fe, 40Ca, and 39K. Such dramaticchanges from normal operating condi-tions (1000 W, 0.8 L/min) will affect theelectrical characteristics of the plasma. • Running volatile organic solvents. Ana-lyzing oil or organic-based samples re-quires a chilled spray chamber (typically220 °C) or a membrane desolvation sys-

tem to reduce the solvent loading on theplasma. In addition, higher RF power(1300–1500 W) and lower nebulizer gasflow (0.4–0.8 L/min) are required to dis-sociate the organic components in thesample. A reduction in the amount of sol-vent entering the plasma combined withhigher power and lower nebulizer gasflow translate into a hotter plasma and achange in its ionization mechanism. • Reducing oxides. The formation of ox-ide species can be problematic in somesample types. For example, in geochemi-cal applications it is quite common to sac-rifice sensitivity by lowering the nebulizergas flow and increasing the RF power toreduce the formation of rare earth ox-ides, which can interfere spectrally withthe determination of other analytes. Un-fortunately these conditions have the po-tential to induce a secondary discharge.• Running a “dry” plasma. Samplingaccessories such as membrane desolva-tors, laser ablation systems, and elec-trothermal vaporization devices are beingused more routinely to enhance the flexi-bility of ICP-MS. The major difference be-tween these sampling devices and a con-ventional liquid sample introductionsystem, is that they generate a “dry” sam-ple aerosol, which requires totally differ-ent operating conditions compared with aconventional “wet” plasma. An aerosolcontaining no solvent can have a dramaticeffect on the ionization conditions in theplasma.

Even though most modern ICP-MS in-terfaces have been designed to minimizethe effects of the secondary discharge, it


Ion optics

Skimmer Sampler

760 Torr3–4 TorrExpansion ofion beam

Ion beamSampleaerosol

Figure 6. The composition of the ion beam is maintained, assuming a neutral plasma.

Interfacecone

Debyelength

Orificediameter

Figure 7. Electrodynamic forces do not af-fect the composition of the ion beam enter-ing the sampler or the skimmer cone.

Circle 17

“Tutorial” continued on page 34

34 SPECTROSCOPY 16(7) JULY 2001 www.spec t roscopyo n l i n e . com

of the vendor) have been checked to en-sure compliance with the regulations.The audit should be planned and shouldcover items such as the design and pro-gramming phases, product testing and re-lease, documentation, and support; a re-port of the audit should be produced afterthe visit. Two published articles have cov-ered vendor audits in more detail (3, 4).

The minimum audit is a remote vendoraudit using a checklist that the vendorcompletes and returns to you. This isusually easy to complete, but the writer ofthe checklist must ensure that the ques-tions are written in a way that can be un-derstood by the recipient, because lan-guage and cultural issues could affect aremote checklist. Moreover, there is littleway of checking the answers you receive.However, for smaller software systems —and some spectrometers fall into this cat-egory — a remote audit is a cost-effectiveway of getting information on how a ven-dor carries out its development process,so long as you know and understand itslimitations.

SYSTEM SELECTION: PART TWOIf the vendor audit, price quote, instru-ment, and software are all acceptable,you’ll be raising a capital expenditurerequest (or whatever it is called in yourorganization) and then generating a pur-chase order. The quote and the purchaseorder are a link in the validation chain;they provide a link into the next phase ofthe validation life cycle: qualification. Thepurchase order is the first stage in defin-ing the initial configuration of the system,as we’ll discover in the next article in thisseries.

REFERENCES(1) R.D. McDowall, Spectroscopy 16(2),

32–43 (2001).(2) IEEE Standard 1012-1986, “Software Vali-

dation and Verification Plans,” Institute ofElectronic and Electrical Engineers, Pis-cataway, NJ, USA.

(3) R.D. McDowall, Sci. Data Mgmt. 2(2), 8(1998).

(4) R.D. McDowall, Sci. Data Mgmt. 2(3), 8(1998). ◆

F O C U S O N Q U A L I T Y . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . .

shouldn’t be taken for granted that theycan all handle changes in operating con-ditions and matrix components with thesame amount of ease. The most notice-able problems that have been reportedinclude spectral peaks of the cone mater-ial appearing in the blank (9); erosion ordiscoloration of the sampling cones;widely different optimum plasma condi-tions for different masses (10); and in-creased frequency of tuning the ion op-tics (8). Of all these, probably the mostinconvenient problem is regular opti-mization of the lens voltages, becauseslight changes in plasma conditions canproduce significant changes in ion ener-gies, which require regular retuning ofthe ion optics. Even though most instru-ments have computer-controlled ion op-tics, it becomes another variable thatmust be optimized. This isn’t a majorproblem but might be considered an in-convenience for a high–sample through-put lab. There is no question that theplasma discharge, interface region, andion optics all have to be designed in con-cert to ensure that the instrument canhandle a wide range of operating condi-tions and sample types. The role of theion optics will be discussed in greater de-tail in the next installment of this series.

“Tutorial” continued from page 34

Circle 22, 23

REFERENCES(1) A.L. Gray and A.R. Date, Analyst 108,

1033 (1983).(2) R.S. Houk, V. A. Fassel, and H.J. Svec, Dy-

namic Mass Spectrosc. 6, 234 (1981).(3) A.R. Date and A.L. Gray, Analyst 106,

1255 (1981).(4) A.L. Gray and A.R. Date, Dynamic Mass

Spectrosc. 6, 252 (1981).(5) S.D. Tanner, J. Anal. At. Spectrom. 10,

905 (1995).(6) K. Sakata and K. Kawabata, Spectrochim.

Acta 49B, 1027 (1994).(7) S. Georgitus, M. Plantz, poster paper pre-

sented at the Winter Conference onPlasma Spectrochemistry, FP4, FortLauderdale, FL (1996).

(8) D.J. Douglas and J.B. French, Spec-trochim. Acta 41B, 3, 197 (1986).

(9) D.J. Douglas, Can. J. Spectrosc. 34, 2(1089).

(10) J.E. Fulford and D.J. Douglas, Appl. Spec-trosc. 40, 7 (1986).

Robert Thomas has more than 30 years ex-perience in trace element analysis. He is theprincipal of his own freelance writing andscientific consulting company, Scientific So-lutions, based in Gaithersburg, MD. He canbe contacted by e-mail at [email protected] or via his web site atwww.scientificsolutions1.com.◆

38 SPECTROSCOPY 16(9) SEPTEMBER 2001 www.spec t roscopyo n l i n e . com

Part V of the series on inductively coupledplasma–mass spectrometry (ICP-MS) takesa detailed look at the ion focusing system— a crucial area of the ICP mass spec-trometer where the ion beam is focused be-fore it enters the mass analyzer. Sometimesknown as the ion optics, it comprises one ormore ion lens components, which electro-statically steer the analyte ions from the in-terface region into the mass separation de-vice. The strength of a well-designed ionfocusing system is its ability to produce lowbackground levels, good detection limits,and stable signals in real-world samplematrices.

Although the detection capability ofICP-MS is generally recognized asbeing superior to any of the otheratomic spectroscopic techniques, itis probably most susceptible to the

sample’s matrix components. The inher-ent problem lies in the fact that ICP-MS isrelatively inefficient; out of every millionions generated in the plasma, only one ac-tually reaches the detector. One of themain contributing factors to the low effi-ciency is the higher concentration of ma-trix elements compared with the analyte,which has the effect of defocusing theions and altering the transmission charac-teristics of the ion beam. This is some-times referred to as a space charge ef-fect, and it can be particularly severewhen the matrix ions have a heavier massthan the analyte ions (1). The role of theion focusing system is therefore to trans-port the maximum number of analyteions from the interface region to themass separation device, while rejecting asmany of the matrix components and non-analyte-based species as possible. Let usnow discuss this process in greater detail.

ROLE OF THE ION OPTICSFigure 1 shows the position of the ion op-tics relative to the plasma torch and inter-face region; Figure 2 represents a moredetailed look at a typical ion focusingsystem.

The ion optics are positioned betweenthe skimmer cone and the mass separa-tion device, and consist of one or moreelectrostatically controlled lens compo-nents. They are not traditional optics thatwe associate with ICP emission or atomicabsorption but are made up of a series ofmetallic plates, barrels, or cylinders thathave a voltage placed on them. The func-tion of the ion optic system is to take ionsfrom the hostile environment of theplasma at atmospheric pressure via theinterface cones and steer them into themass analyzer, which is under high vac-uum. As mentioned in Part IV of the se-ries, the plasma discharge, interface re-gion, and ion optics have to be designed

in concert with each other. It is absolutelycritical that the composition and electricalintegrity of the ion beam is maintained asit enters the ion optics. For this reason itis essential that the plasma is at zero po-tential to ensure that the magnitude andspread of ion energies are as low aspossible (2).

A secondary but also very importantrole of the ion optic system is to stop par-ticulates, neutral species, and photonsfrom getting through to the mass ana-lyzer and the detector. These speciescause signal instability and contribute tobackground levels, which ultimately af-fect the performance of the system. Forexample, if photons or neutral speciesreach the detector, they will elevate thebackground noise and therefore degradedetection capability. In addition, if particu-lates from the matrix penetrate fartherinto the mass spectrometer region, theyhave the potential to deposit on lens com-

A Beginner’s Guide to ICP-MSPart V: The Ion Focusing System


Figure 1. Position of ion optics relative to the plasma torch and interface region.


making it into the mass spectrometer.The first method is to place a groundedmetal stop (disk) behind the skimmercone. This stop allows the ion beam tomove around it but physically blocks theparticulates, photons, and neutral speciesfrom traveling downstream (3). The other

approach is to set the ion lens or mass an-alyzer slightly off axis. The positivelycharged ions are then steered by the lenssystem into the mass analyzer, while thephotons and neutral and nonionic speciesare ejected out of the ion beam (4).

It is also worth mentioning that somelens systems incorporate an extractionlens after the skimmer cone to electro-statically pull the ions from the interfaceregion. This has the benefit of improvingthe transmission and detection limits ofthe low-mass elements (which tend to bepushed out of the ion beam by the heav-ier elements), resulting in a more uni-form response across the full mass rangeof 0–300 amu. In an attempt to reducethese space charge effects, some olderdesigns have used lens components to ac-celerate the ions downstream. Unfortu-nately this can have the effect of degrad-ing the resolving power and abundancesensitivity (ability to differentiate an ana-lyte peak from the wing of an interfer-ence) of the instrument because of themuch higher kinetic energy of the accel-erated ions as they enter the massanalyzer (5).

DYNAMICS OF ION FLOWTo fully understand the role of the ion op-tics in ICP-MS, it is important to have anappreciation of the dynamics of ion flowfrom the plasma through the interface re-gion into the mass spectrometer. Whenthe ions generated in the plasma emergefrom the skimmer cone, there is a rapidexpansion of the ion beam as the pres-sure is reduced from 760 Torr (atmos-pheric pressure) to approximately 1023 to1024 Torr in the lens chamber with aturbomolecular pump. The compositionof the ion beam immediately behind thecone is the same as the composition infront of the cone because the expansionat this stage is controlled by normal gasdynamics and not by electrodynamics.One of the main reasons for this is that,in the ion sampling process, the Debyelength (the distance over which ions ex-ert influence on each other) is small com-pared with the orifice diameter of thesampler or skimmer cone. Consequentlythere is little electrical interaction be-tween the ion beam and the cone and rel-atively little interaction between the indi-vidual ions in the beam. In this way,compositional integrity of the ion beam ismaintained throughout the interface re-gion (6). With this rapid drop in pressurein the lens chamber, electrons diffuse out

ponents and, in extreme cases, get intothe mass analyzer. In the short term thiswill cause signal instability and, in thelong term, increase the frequency ofcleaning and routine maintenance.

Basically two approaches will reducethe chances of these undesirable species

Figure 2. A generic ion focusing system, showing position of ion optics relative to the inter-face cones and mass analyzer.

Figure 3. Extreme pressure drop in the ion optic chamber produces diffusion of electrons, re-sulting in a positively charged ion beam.


of the ion beam. Because of the small sizeof the electrons relative to the positivelycharged ions, the electrons diffuse far-ther from the beam than the ions, result-ing in an ion beam with a net positivecharge. This is represented schematicallyin Figure 3.

The generation of a positively chargedion beam is the first stage in the chargeseparation process. Unfortunately, the net

positive charge of the ion beam meansthat there is now a natural tendency forthe ions to repel each other. If nothing isdone to compensate for this, ions with ahigher mass-to-charge ratio will dominatethe center of the ion beam and force thelighter ions to the outside. The degree ofloss will depend on the kinetic energy ofthe ions: those with high kinetic energy(high mass elements) will be transmitted

in preference to ions with medium (mid-mass elements) or low kinetic energy(low-mass elements). This is shown inFigure 4. The second stage of charge sep-aration is therefore to electrostaticallysteer the ions of interest back into thecenter of the ion beam by placing volt-ages on one or more ion lens compo-nents. Remember, however, that this ispossible only if the interface is kept atzero potential, which ensures a neutralgas-dynamic flow through the interfaceand maintains the compositional integrityof the ion beam. It also guarantees thatthe average ion energy and energyspread of each ion entering the lens sys-tems are at levels optimum for mass sepa-ration. If the interface region is notgrounded correctly, stray capacitance willgenerate a discharge between the plasmaand sampler cone and increase the ki-netic energy of the ion beam, making itvery difficult to optimize the ion lenssystem (7).

COMMERCIAL ION OPTIC DESIGNSOver the years, there have been many dif-ferent ion optic designs. Although they allhave their own characteristics, they per-form the same basic function: to discrimi-nate undesirable matrix- or solvent-basedions so that only the analyte ions aretransmitted to the mass analyzer. Themost common ion optics design used to-day consists of several lens components,which all have a specific role to play inthe transmission of the analyte ions (8).With these multicomponent lens sys-tems, the voltage can be optimized onevery lens of the ion optics to achieve thedesired ion specificity. Over the years thistype of lens configuration has proven tobe very durable and has been shown toproduce very low background levels, par-ticularly when combined with an off-axismass analyzer. However, because of theinteractive nature of parameters that af-fect the signal response, the more com-plex the lens system the more variableshave to be optimized, particularly if manydifferent sample types are being ana-lyzed. This isn’t such a major problem be-cause the lens voltages are all computer-controlled, and methods can be stored forevery new sample scenario. Figure 5 is acommercially available multicomponentlens system, with an extraction lens andoff-axis quadrupole mass analyzer, show-ing how the ion beam is deflected into themass analyzer, while the photons and

Figure 4. The degree of ion repulsion will depend on kinetic energy of the ions: those withhigh kinetic energy (green with red +) will be transmitted in preference to ions with medium(yellow with red +) or low kinetic energy (blue with red +).

Figure 5. Schematic of a multicomponent lens system with extraction lens and off-axisquadrupole mass analyzer (courtesy of Agilent Technologies [Wilmington, DE]).


neutral species travel in a straight lineand strike a metal plate.

Another, more novel approach is to usejust one cylindrical ion lens, combinedwith a grounded stop positioned just in-side the skimmer cone as shown inFigure 6 (9).

With this design, the voltage is dynam-ically ramped on-the-fly, in concert withthe mass scan of the analyzer (typically aquadrupole). The benefit of this approachis that the optimum lens voltage is placedon every mass in a multielement run toallow the maximum number of analyteions through, while keeping the matrixions to an absolute minimum. This is rep-resented in Figure 7, which shows a lensvoltage scan of six elements: lithium,cobalt, yttrium, indium, lead, and ura-nium, at 7, 59, 89, 115, 208, and 238 amu,respectively. We can see that each ele-ment has its own optimum value, which isthen used to calibrate the system, so thelens can be ramp-scanned across the fullmass range. This type of approach is typi-cally used in conjunction with a groundedstop to act as a physical barrier to reduceparticulates, neutral species, and photonsfrom reaching the mass analyzer and de-tector. Although this design producesslightly higher background levels, it of-fers excellent long-term stability withreal-world samples. It works well formany sample types but is most effectivewhen low mass elements are being deter-mined in the presence of high-mass–matrix elements.

It is also worth emphasizing that anumber of ICP-MS systems offer what isknown as a high-sensitivity interface.

These all work slightly differently butshare similar components. By using acombination of slightly different conegeometry, higher vacuum at the inter-face, one or more extraction lenses, andslightly modified ion optic design, they of-fer as much as 10 times the sensitivity ofa traditional interface (10). However, thisincreased sensitivity is usually combinedwith inferior stability and an increase inbackground levels, particularly for sam-ples with a heavy matrix. To get aroundthis degradation in performance onemust usually dilute the samples beforeanalysis, which limits the systems’ applic-ability for real-world samples (11). How-ever, they have found a use in non-liquid-based applications in which highsensitivity is crucial, for example in theanalysis of small spots on the surface of ageological specimen using laser ablationICP-MS. For this application, the instru-ment must offer high sensitivity becausea single laser pulse is used to ablate thesample and sweep a tiny amount of thedry sample aerosol into the ICP-MS (12).

The role of the ion focusing systemcannot be overestimated. It affects thebackground noise level of the instrument.It has a huge impact on both long- andshort-term signal stability, especially inreal-world samples, and it also dictatesthe number of ions that find their way tothe mass analyzer. However, it must beemphasized that the ion optics are only asgood as the ions that feed it, and for thisreason it must be designed in concertwith both the plasma source and the in-terface region. There is no question thatthis area is crucial to the design of thewhole ICP mass spectrometer. In the

next part of the series we will discuss theheart of the ICP mass spectrometer: themass analyzer.

REFERENCES(1) J. A. Olivares and R.S. Houk, Anal. Chem.

58, 20 (1986).(2) D.J. Douglas and J.B. French, Spec-

trochim. Acta 41B(3), 197 (1986).(3) E.R. Denoyer, D. Jacques, E. Debrah, and

S.D. Tanner, At. Spectrosc. 16(1), 1(1995).

(4) D. Potter, American Lab (July 1994).(5) P. Turner, paper presented at Second In-

ternational Conference on Plasma SourceMass Spec, Durham, UK, 1990.

(6) S.D. Tanner, D.J. Douglas, and J.B.French, Appl. Spectrosc. 48, 1373 (1994).

(7) R. Thomas, Spectroscopy 16(7), 26–34(2001).

(8) Y. Kishi, Agilent Technologies ApplicationJournal, August (1997).

(9) S.D. Tanner, L.M. Cousins, and D.J. Dou-glas, Appl. Spectrosc. 48, 1367 (1994).

(10) I.B. Brenner, M. Liezers, J. Godfrey, S.Nelms, and J. Cantle, Spectrochim. ActaPart 53B(6–8), 1087 (1998).

(11) B.C. Gibson, presented at Surrey Interna-tional Conference on ICP-MS, London,UK, 1994.

(12) T. Howe, J. Shkolnik, and R.J. Thomas,Spectroscopy 16(2), 54 (2001).


Figure 6. Schematic of a single ion lens andgrounded stop system (not to scale[courtesy of PerkinElmer Instruments{Norwalk, CT}]).

Figure 7. A calibration of optimum lens volt-ages is used to ramp-scan the ion lens inconcert with the mass scan of the analyzer.The signals have been normalized for com-parison purposes.

44 SPECTROSCOPY 16(10) OCTOBER 2001 www.spec t roscopyo n l i n e . com

Part VI of the series on inductively coupledplasma–mass spectroscopy (ICP-MS) fun-damentals deals with the heart of the sys-tem — the mass separation device. Some-times called the mass analyzer, it is theregion of the ICP mass spectrometer thatseparates the ions according to their mass-to-charge ratio (m/z). This selectionprocess is achieved in a number of differentways, depending on the mass separationdevice, but they all have one common goal:to separate the ions of interest from all theother nonanalyte, matrix, solvent, and argon-based ions.

Although inductively coupledplasma–mass spectroscopy (ICP-MS) was commercialized in 1983,the first 10 years of its developmentprimarily used traditional quadru-

pole mass filter technology to separatethe ions of interest. These worked excep-tionally well for most applications butproved to have limitations in determiningdifficult elements or dealing with more-complex sample matrices. This led to the

development of alternative mass separa-tion devices that pushed the capabilitiesof ICP-MS so it could be used for morechallenging applications. Before we dis-cuss these different mass spectrometersin greater detail, let’s take a look at the lo-cation of the mass analyzer in relation tothe ion optics and the detector. Figure 1shows this in greater detail.

As we can see, the mass analyzer is po-sitioned between the ion optics and thedetector and is maintained at a vacuum ofapproximately 1026 Torr with a secondturbomolecular pump. Assuming the ionsare emerging from the ion optics at theoptimum kinetic energy (1), they areready to be separated according to theirmass-to-charge ratio by the mass ana-lyzer. There are basically four kinds ofcommercially available mass analyzers:quadrupole mass filters, double focusingmagnetic sector, time-of-flight, andcollision–reaction cell technology. Theyall have their own strengths and weak-nesses, which we will discuss in greaterdetail in the next few installments of this

column. Let’s first begin with the mostcommon of the mass separation devicesused in ICP-MS, the quadrupole mass filter.

QUADRUPOLE MASS FILTERTECHNOLOGYDeveloped in the early 1980s,quadrupole-based systems represent ap-proximately 90% of all ICP mass spec-trometers used today. This design wasthe first to be commercialized; as a result,today’s quadrupole ICP-MS technology isconsidered a very mature, routine, high-throughput, trace-element technique. Aquadrupole consists of four cylindrical orhyperbolic metallic rods of the samelength and diameter. They are typicallymade of stainless steel or molybdenum,and sometimes have a ceramic coatingfor corrosion resistance. Quadrupolesused in ICP-MS are typically 15–20 cm inlength and about 1 cm in diameter andoperate at a frequency of 2–3 MHz.

BASIC PRINCIPLES OF OPERATION By placing a direct current (dc) field onone pair of rods and a radio frequency(rf) field on the opposite pair, ions of a se-lected mass are allowed to pass throughthe rods to the detector, while the othersare ejected from the quadrupole. Figure 2shows this in greater detail.

In this simplified example, the analyteion (black) and four other ions (colored)have arrived at the entrance to the fourrods of the quadrupole. When a particu-lar rf-dc voltage is applied to the rods, thepositive or negative bias on the rods willelectrostatically steer the analyte ion ofinterest down the middle of the four rodsto the end, where it will emerge and beconverted to an electrical pulse by the de-tector. The other ions of different mass-to-charge ratios will pass through thespaces between the rods and be ejected


Beginner’s Guide to ICP-MSPart VI — The Mass Analyzer


Iondetector

MSinterface

ICP torchIon optics

Spraychamber

Nebulizer


Turbomolecularpump

Turbomolecularpump

Mechanicalpump

RFpowersupply

Figure 1. Schematic of an ICP-MS system showing the location of the mass separation device.

OCTOBER 2001 16(10) SPECTROSCOPY 45

from the quadrupole. This scanningprocess is then repeated for another ana-lyte at a completely different mass-to-charge ratio until all the analytes in a mul-tielement analysis have been measured.The process for the detection of one par-ticular mass in a multielement run is rep-resented in Figure 3. It shows a 63Cu ionemerging from the quadrupole and beingconverted to an electrical pulse by the de-tector. As the rf-dc voltage of the quadru-pole — corresponding to 63Cu — is re-peatedly scanned, the ions as electricalpulses are stored and counted by a multi-channel analyzer. This multichannel data-acquisition system typically has 20 chan-nels per mass, and as the electrical pulsesare counted in each channel, a profile ofthe mass is built up over the 20 channels,

corresponding to the spectral peak of63Cu. In a multielement run, repeatedscans are made over the entire suite ofanalyte masses, as opposed to just onemass represented in this example.

Quadrupole scan rates are typically onthe order of 2500 atomic mass units(amu) per second and can cover the en-tire mass range of 0–300 amu in about0.1 s. However, real-world analysis speedsare much slower than this, and in practice25 elements can be determined in dupli-cate with good precision in 1–2 min.

QUADRUPOLE PERFORMANCE CRITERIATwo very important performance specifi-cations of a mass analyzer govern its abil-ity to separate an analyte peak from aspectral interference. The first is resolv-

S P E C T R O S C O P Y T U T O R I A L . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . .

+

+

+

+

++

+Quadrupole rods

-

+

Ionflow

Figure 2. Schematic showing principles of a quadrupole mass filter.

Quadrupole

Quadrupolemass scancontrollerMultichannel data

acquisition system

Ch l

Ions (electricalpulses)

Mass scan (amu)

63

1n

Detector

63Cu

63Cu scan

Copper ion — m/z 63

1 2 3 4 5 6 7 ............ 20

Figure 3. Profiles of different masses are built up using a multichannel data acquisition system.Circle 33


ing power (R), which in traditional massspectrometry is represented by the fol-lowing equation: R 5 m/Dm, where m isthe nominal mass at which the peak oc-curs and Dm is the mass difference be-tween two resolved peaks (2). However,for quadrupole technology, the term reso-lution is more commonly used, and isnormally defined as the width of a peak at10% of its height. The second specifica-tion is abundance sensitivity, which is the

signal contribution of the tail of an adja-cent peak at one mass lower and onemass higher than the analyte peak (3).Even though they are somewhat relatedand both define the quality of a quadru-pole, the abundance sensitivity is proba-bly the most critical. If a quadrupole hasgood resolution but poor abundance sen-sitivity, it will often prohibit the measure-ment of an ultratrace analyte peak next toa major interfering mass.

RESOLUTIONLet us now discuss this area in greater de-tail. The ability to separate differentmasses with a quadrupole is determinedby a combination of factors includingshape, diameter, and length of the rods, fre-quency of quadrupole power supply, oper-ating vacuum, applied rf-dc voltages, andthe motion and kinetic energy of the ionsentering and exiting the quadrupole. Allthese factors will have a direct impact onthe stability of the ions as they travel downthe middle of the rods and thus thequadrupole’s ability to separate ions of dif-fering mass-to-charge ratios. This is repre-sented in Figure 4, which shows a simpli-fied version of the Mathieu mass stabilityplot of two separate masses (A and B) en-tering the quadrupole at the same time (4).

Any of the rf-dc conditions shown un-der the light blue plot will allow onlymass A to pass through the quadrupole,while any combination of rf-dc voltagesunder the yellow plot will allow only massB to pass through the quadrupole. If theslope of the rf-dc scan rate is steep, repre-sented by the light blue line (high resolu-tion), the spectral peaks will be narrow,and masses A and B will be well sepa-rated (equivalent to the distance betweenthe two blue arrows). However, if theslope of the scan is shallow, representedby the red line ( low resolution), the spec-tral peaks will be wide, and masses A andB will not be so well separated (equiva-lent to the distance between the two redarrows). On the other hand, if the slopeof the scan is too shallow, represented bythe gray line (inadequate resolution), thepeaks will overlap each other (shown bythe green area of the plot) and themasses will pass through the quadrupolewithout being separated. In theory, theresolution of a quadrupole mass filter canbe varied between 0.3 and 3.0 amu. How-

S P E C T R O S C O P Y T U T O R I A L . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

F(rf)

F(dc

)

High resolutionA

BLow

resolution

Inadequateresolution

Overlap

Figure 4. Simplified Mathieu stability dia-gram of a quadrupole mass filter, showingseparation of two different masses, A (lightblue plot) and B (yellow plot).

Resolution

Inte

nsity

High resolution(0.3 amu)

Normal resolution (1.0 amu)

Low resolution (3.0 amu)

Figure 5. Sensitivity comparison of aquadrupole operated at 3.0, 1.0, and 0.3amu resolution (measured at 10% of itspeak height).

Resolution 0.70 amuResolution 0.50 amu

30,000

20,000

10,000

062 63 64 65 66

Mass

Puls

e in

tens

ity

Figure 6. Sensitivity comparison of two cop-per isotopes, 63Cu and 65Cu, at resolutionsettings of 0.70 and 0.50 amu.

Circle 34

OCTOBER 2001 16(10) SPECTROSCOPY 47

ever, improved resolution is always ac-companied by a sacrifice in sensitivity, asseen in Figure 5, which shows a compari-son of the same mass at a resolution of3.0, 1.0, and 0.3 amu.

We can see that the peak height at 3.0amu is much larger than the peak heightat 0.3 amu but, as expected, it is alsomuch wider. This would prohibit using aresolution of 3.0 amu with spectrally com-plex samples. Conversely, the peak widthat 0.3 amu is very narrow, but the sensi-tivity is low. For this reason, a compro-mise between peak width and sensitivityusually has to be reached, depending onthe application. This can clearly be seenin Figure 6, which shows a spectral over-lay of two copper isotopes — 63Cu and65Cu — at resolution settings of 0.70 and0.50 amu. In practice, the quadrupole isnormally operated at a resolution of0.7–1.0 amu for most applications.

It is worth mentioning that mostquadrupoles are operated in the first sta-bility region, where resolving power istypically ;400. If the quadrupole is oper-ated in the second or third stability re-gions, resolving powers of 4000 (5) and9000 (6), respectively, have beenachieved. However, improving resolutionusing this approach has resulted in a sig-nificant loss of signal. Although there areways of improving sensitivity, other prob-lems have been encountered, and as a re-sult, to date there are no commercial in-struments available based on this design.

Some instruments can vary the peakwidth on-the-fly, which means that the res-olution can be changed between 3.0 and0.3 amu for every analyte in a multiele-ment run. For some challenging applica-tions this can be beneficial, but in realitythey are rare. So, even thoughquadrupoles can be operated at higherresolution (in the first stability region),

until now the slight improvement has notbecome a practical benefit for most rou-tine applications.

ABUNDANCE SENSITIVITYWe can see in Figure 6 that the tails of

the spectral peaks drop off more rapidlyat the high mass end of the peak com-pared with the low mass end. The overallpeak shape, particularly its low mass andhigh mass tail, is determined by the abun-dance sensitivity of the quadrupole,


Mass (amu)

Inte

nsity

Quadrupole

Mass (M)

M 1 M 1

Ion exitingIon entering

Figure 7. Ions entering the quadrupole areslowed down by the filtering process andproduce peaks with a pronounced tail orshoulder at the low-mass end.

1 ppm of monoisotopic arsenic (75As)

75 75

Doubly charged 151Eu (75.5 amu)

(a) (b)

Figure 8. A low abundance sensitivity specification is critical to minimize spectral interfer-ences, as shown by (a) a spectral scan of 50 ppm of 151Eu++ at 75.5 amu and (b) an ex-panded view, which shows how the tail of the 151Eu++ elevates the spectral background of1 ppb of As at mass 75.

Circle 35


which is affected by a combination of fac-tors including design of the rods, fre-quency of the power supply, and operat-ing vacuum (7). Even though they are allimportant, probably the biggest impactson abundance sensitivity are the motionand kinetic energy of the ions as they en-ter and exit the quadrupole. If one looksat the Mathieu stability plot in Figure 3, itcan be seen that the stability boundariesof each mass are less defined (not sosharp) on the low mass side than they areon the high mass side (4). As a result, thecharacteristics of ion motion at the lowmass boundary is different from the highmass boundary and is therefore reflectedin poorer abundance sensitivity at the lowmass side compared with the high massside. In addition, the velocity (and there-fore the kinetic energy) of the ions enter-ing the quadrupole will affect the ion mo-tion and, as a result, will have a directimpact on the abundance sensitivity. Forthat reason, factors that affect the kineticenergy of the ions, like high plasma po-tential and the use of lens components toaccelerate the ion beam, will degrade theinstrument’s abundance sensitivity (8).

These are the fundamental reasonswhy the peak shape is not symmetricalwith a quadrupole and explains why thereis always a pronounced shoulder at thelow mass side of the peak compared tothe high mass side — as represented inFigure 7, which shows the theoreticalpeak shape of a nominal mass M. We cansee that the shape of the peak at onemass lower (M 2 1) is slightly differentfrom the other side of the peak at onemass higher (M 1 1) than the mass M.For this reason, the abundance sensitivityspecification for all quadrupoles is alwaysworse on the low mass side than on thehigh mass side and is typically 1 3 1026

at M 2 1 and 1 3 1027 at M 1 1. In otherwords, an interfering peak of 1 millioncounts per second (cps) at M 2 1 wouldproduce a background of 1 cps at M,while it would take an interference of 107

cps at M 1 1 to produce a background of1 cps at M.

BENEFITS OF GOOD ABUNDANCESENSITIVITYFigure 8 shows an example of the impor-tance of abundance sensitivity. Figure 8ais a spectral scan of 50 ppm of the doublycharged europium ion — 151Eu11 at 75.5amu (a doubly charged ion is one withtwo positive charges, as opposed to a nor-mal singly charged positive ion, and ex-hibits a m/z peak at half its mass). We cansee that the intensity of the peak is sogreat that its tail overlaps the adjacentmass at 75 amu, which is the only avail-able mass for the determination of ar-senic. This is highlighted in Figure 8b,which shows an expanded view of the tailof the 151Eu11, together with a scan of 1ppb of As at mass 75. We can see veryclearly that the 75As signal lies on thesloping tail of the 151Eu11 peak. Mea-surement on a sloping background likethis would result in a significant degrada-tion in the arsenic detection limit, particu-larly as the element is monoisotopic andno alternative mass is available. This ex-ample shows the importance of a lowabundance sensitivity specification inICP-MS.

DIFFERENT QUARDUPOLE DESIGNSMany different designs of quadrupole areused in ICP-MS, all made from differentmaterials with various dimensions,shapes, and physical characteristics. Inaddition, they are all maintained atslightly different vacuum chamber pres-sures and operate at different frequen-

cies. Theory tells us that hyperbolic rodsshould generate a better hyperbolic (el-liptical) field than cylindrical rods, result-ing in higher transmission of ions athigher resolution. It also tells us that ahigher operating frequency means ahigher rate of oscillation — and thereforeseparation — of the ions as they traveldown the quadrupole. Finally, it is verywell accepted that a higher vacuum pro-duces fewer collisions between gas mole-cules and ions, resulting in a narrowerspread in kinetic energy of the ions andtherefore less of a tail at the low massside of a peak. However, given all thesespecification differences, in practice theperformance of most modern quadrupoleICP-MS instrumentation is very similar.

So even though these differences willmainly be transparent to users, there aresome subtle variations in each instru-ment’s measurement protocol and thesoftware’s approach to peak quantitation.This is a very important area that we willdiscuss it in greater detail in a future col-umn. The next part of the series will con-tinue with describing the fundamentalprinciples of other types of mass analyz-ers used in ICP-MS.

REFERENCES(1) R. Thomas, Spectroscopy 16(9), 38–44

(2001).(2) F. Adams, R. Gijbels, and R. Van Grieken,

Inorganic Mass Spectrometry (John Wileyand Sons, New York, 1988).

(3) A. Montasser, Ed. Inductively CoupledPlasma Mass Spectrometry (Wiley-VCH,Berlin, 1998).

(4) P.H. Dawson, Ed., Quadrupole Mass Spec-trometry and its Applications (Elsevier,Amsterdam, 1976; reissued by AIP Press,Woodbury, NY, 1995).

(5) Z. Du, T.N. Olney, and D.J. Douglas, J.Am. Soc. Mass Spectrom. 8, 1230–1236(1997).

(6) P.H. Dawson and Y. Binqi, Int. J. MassSpectrom., Ion Proc. 56, 25 (1984).

(7) D. Potter, Agilent Technologies ApplicationNote, 228–349 (January, 1996).

(8) E.R. Denoyer, D. Jacques, E. Debrah, andS.D. Tanner, At. Spectrosc. 16(1), 1(1995).


S P E C T R O S C O P Y T U T O R I A L . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Circle 36

22 SPECTROSCOPY 16(11) NOVEMBER 2001 www.spec t roscopyon l i ne . com

Although quadrupole mass analyzers repre-sent more than 90% of all inductively cou-pled plasma mass spectrometry (ICP-MS)systems installed worldwide, limitations intheir resolving power has led to the develop-ment of high-resolution spectrometers basedon the double-focusing magnetic-sector de-sign. Part VII of this series on ICP-MStakes a detailed look at this very powerfulmass separation device, which has foundits niche in solving challenging applicationproblems that require excellent detectioncapability, exceptional resolving power, orvery high precision.

As discussed in Part VI of this series(1), a quadrupole-based ICP-MSsystem typically offers a resolutionof 0.7–1.0 amu. This is quite ade-quate for most routine applications,

but has proved to be inadequate for manyelements that are prone to argon-,solvent-, or sample-based spectral inter-ferences. These limitations inquadrupoles drove researchers in the di-rection of traditional high-resolution,magnetic-sector technology to improvequantitation by resolving the analytemass away from the spectral interference(2). These ICP-MS instruments, whichwere first commercialized in the late1980s, offered resolving power as high as10,000, compared with a quadrupole,which had a resolving power of approxi-mately 300. This dramatic improvementin resolving power allowed difficult ele-ments like Fe, K, As, V, and Cr to be de-termined with relative ease, even in com-plex sample matrices.

TRADITIONAL MAGNETIC-SECTORINSTRUMENTSThe magnetic-sector design was first usedin molecular spectroscopy for the struc-tural analysis of complex organic com-pounds. Unfortunately, it was initially

found to be unsuitable as a separation de-vice for an ICP system because it requireda few thousand volts of potential at theplasma interface area to accelerate theions into the mass analyzer. For this rea-son, basic changes had to be made to theion acceleration mechanism to optimize itas an ICP-MS separation device. This wasa significant challenge when magnetic-sector systems were first developed in thelate 1980s. However, by the early 1990s,instrument designers solved this problemby moving the high-voltage componentsaway from the plasma and interface andcloser to the mass spectrometer. Today’sinstrumentation is based on two differentapproaches, commonly referred to asstandard or reverse Nier-Johnson geome-try. Both these designs, which use thesame basic principles, consist of two ana-

lyzers — a traditional electromagnet andan electrostatic analyzer (ESA). In thestandard (sometimes called forward) de-sign, the ESA is positioned before themagnet, and in the reverse design it is po-sitioned after the magnet. A schematic ofa reverse Nier-Johnson spectrometer isshown in Figure 1.

PRINCIPLES OF OPERATION With this approach, ions are sampledfrom the plasma in a conventional man-ner and then accelerated in the ion opticregion to a few kilovolts before they enterthe mass analyzer. The magnetic field,which is dispersive with respect to ion en-ergy and mass, focuses all the ions withdiverging angles of motion from the en-trance slit. The ESA, which is only disper-sive with respect to ion energy, then fo-


A Beginner’s Guide to ICP-MSPart VII: Mass Separation Devices — Double-Focusing Magnetic-Sector Technology


Electrostaticanalyzer

Electromagnet

Detector

Plasma torch

Entranceslit

Exit slit

Accelerationoptics

Focusingoptics

Figure 1. Schematic of a reverse Nier-Johnson double-focusing magnetic-sector mass spectrometer (Courtesy of Thermo Finnigan [San Jose, CA]).

cuses the ions onto the exit slit, wherethe detector is positioned. If the energydispersion of the magnet and ESA areequal in magnitude but opposite in direc-tion, they will focus both ion angles (firstfocusing) and ion energies (second ordouble focusing), when combined to-gether. Changing the electrical field inthe opposite direction during the cycletime of the magnet (in terms of the masspassing the exit slit) has the effect offreezing the mass for detection. Then assoon as a certain magnetic field strength

is passed, the electric field is set to itsoriginal value and the next mass isfrozen. The voltage is varied on a per-mass basis, allowing the operator to scanonly the mass peaks of interest ratherthan the full mass range (3, 4).

Because traditional magnetic-sectortechnology was initially developed for thestructural or qualitative identification oforganic compounds, there wasn’t a realnecessity for rapid quantitation of spectralpeaks required for trace element analysis.They functioned by scanning over a largemass range by varying the magnetic fieldover time with a fixed acceleration volt-age. During a small window in time,which was dependent on the resolutionchosen, ions of a particular mass-to-charge are swept past the exit slit to pro-duce the characteristic flat top peaks. Asthe resolution of a magnetic-sector instru-ment is independent of mass, ion signals,particularly at low mass, are far apart.The result was that a large amount oftime was spent scanning and settling the

magnet. This was not such a major prob-lem for qualitative analysis, but proved tobe impractical for routine trace elementanalysis. This concept is shown in greaterdetail in Figure 2, which is a plot of fourparameters — magnetic field strength,accelerating voltage, mass, and signal in-tensity — against time for four separatemasses (M1–M4). Scanning the magnetfrom point A to point B (accelerating volt-age is fixed) results in a scan across themass range, generating spectral peaks forthe four different masses. It can be seenthat this increased scanning and settlingoverhead time (often referred to as deadtime) would result in valuable measure-ment time being lost, particularly for highsample throughput that required ultra-trace detection levels.

Changing the electric field in the oppo-site direction to the field strength of themagnet during the cycle time of the mag-net has the effect of “stopping” the massthat passes through the analyzer. Then,as soon as the magnetic field strength ispassed, the electrical field is set to itsoriginal value and the next mass“stopped” in the same manner. The accel-erating voltage, as well as its rate ofchange, has to be varied depending onthe mass, but the benefit of this methodis that only the mass peaks of interest areregistered. This process is seen in Figure3, which shows the same four massesscanned. The only difference this time isthat as well as scanning the magnet frompoint A to point B, the accelerating volt-age is also changed, resulting in a step-wise jump from one mass to the next.This means that the full mass range iscovered much faster than just by scan-ning the magnet alone (because of the in-creased speed involved in electricallyjumping from one mass to another) (5).Once the magnet has been scanned to aparticular point, an electric scan is usedto cover an area of � 10–30% of the mass,either to measure the analyte peak ormonitor other masses of interest. Peakquantitation is typically performed by tak-ing multiple data points over a presetmass window and integrating over a fixedperiod of time.

It should be pointed out that althoughthis approach represents enormous timesavings over older, single-focusingmagnetic-sector technology, it is still sig-nificantly slower than quadrupole-basedinstruments. The inherent problem lies inthe fact that a quadrupole can be elec-tronically scanned much faster than a


A

Measurement timeScanning and

settling (dead) time

B

M1

M4

M2 M3

Inte

nsity

Time

Mas

sAc

cele

ratio

n po

tent

ial

Mag

netic

fiel

d

Figure 2. A plot of magnetic field strength,accelerating voltage (fixed), mass, and sig-nal intensity over time for four separatemasses (M1–M4). Note that only the mag-net is scanned, while the accelerating volt-age is fixed — resulting in long scan timesbetween the masses.

Changing the electric

field in the opposite

direction to the field

strength of the magnet

during the cycle time of

the magnet has the

effect of “stopping” the

mass that passes

through the analyzer.

A

Measurement time

Jump time

B

M1M4

M2M3

Inte

nsity

Time

Mas

sAc

cele

ratio

n po

tent

ial

Mag

netic

fiel

d

Figure 3. A plot of magnetic field strength,accelerating voltage (changed), mass, andsignal intensity over time for the same fourmasses (M1–M4). This time, in addition tothe magnet being scanned, the acceleratingvoltage is also changed, resulting in rapidelectric jumps between the masses.

NOVEMBER 2001 16(11) SPECTROSCOPY 23

magnet. Typical speeds for a full massscan (0–250 amu) of a magnet are in theorder of 400–500 ms, compared with 100ms for a quadrupole. In addition, it takesmuch longer for magnets to slow downand settle to take measurements — typi-cally 30–50 ms compared to 1–2 ms for aquadrupole. So, even though in practice,the electric scan dramatically reduces theoverall analysis time, modern double-focusing magnetic-sector ICP-MS sys-tems, especially when multiple resolutionsettings are used, are significantly slowerthan quadrupole instruments. Thismakes them less than ideal for routine,high-throughput applications or for sam-ples that require multielement determina-tions on rapid transient signals.

RESOLVING POWERAs mentioned previously, most commer-

cial magnetic-sector ICP-MS systems of-fer as much as 10,000 resolving power(5% peak height/10% valley definition),which is high enough to resolve the ma-jority of spectral interferences. It’s worthemphasizing that resolving power (R) isrepresented by the equation: R � m/�m,where m is the nominal mass at whichthe peak occurs and �m is the mass dif-ference between two resolved peaks (6).In a quadrupole, the resolution is se-lected by changing the ratio of the rf/dcvoltages on the quadrupole rods. How-ever, because a double-focusingmagnetic-sector instrument involves fo-cusing ion angles and ion energies, massresolution is achieved by using two me-chanical slits — one at the entrance tothe mass spectrometer and another at theexit, before the detector. Varying resolu-tion is achieved by scanning the magnetic

field under differ-ent entrance- andexit-slit width con-ditions. Similar tooptical systems,low resolution isachieved by usingwide slits,whereas high res-olution is achievedwith narrow slits.Varying the widthof both the en-trance and exitslits effectivelychanges the oper-ating resolution.This can be seen

in Figure 4, which shows two slit widthscenarios. Figure 4a shows an example ofa wide exit slit producing relatively lowresolution and a characteristic flat-toppedpeak. Figure 4b shows the same size en-trance slit, but a narrower exit slit, pro-ducing higher resolution with a charac-teristic triangular peak. The lowestpractical resolution achievable with a dou-ble-focusing magnetic-sector instrument,using the widest entrance and exit slits, isapproximately 300–400, whereas thehighest practical resolution, using thenarrowest entrance and exit slits, is ap-proximately 10,000. Most commercialsystems operate at fixed resolution set-tings — for example, low is typically300–400; medium is typically 3000–4000,and high is typically 8000–10,000 (thechoice of settings will vary depending onthe instrumentation).

However, it should be emphasized that,similar to optical spectrometry, as theresolution is increased, the transmissiondecreases. So even though extremelyhigh resolution is available, detection lim-its will be compromised under these con-ditions. This can be seen in Figure 5,which shows a plot of resolution againstion transmission. Figure 5 shows that aresolving power of 400 produces 100%transmission, but at a resolving power of10,000, only �2% is achievable. This dra-matic loss in sensitivity could be an issueif low detection limits are required inspectrally complex samples that requirethe highest possible resolution; however,spectral demands of this nature are notvery common. Table I shows the resolu-


Ion beam

Entrance slit Analyzer Exit slit

Wide slit Lower resolution

Higher resolution

Detector

Inte

nsity

(a)

(b)

Ion beam

Entrance slit Analyzer Exit slit

Narrow slit

Detector

Inte

nsity

Figure 4. Resolution obtained using (a) wide and (b) narrow exit slit widths as the magnetic field is scanned. The entrance slit widths are thesame in (a) and (b).

Table I. Resolution required to resolve some common polyatomic interferences from a selected group of isotopes.

Isotope Matrix Interference Resolution Transmission

39K H2O 38ArH 5570 6%40Ca H2O 40Ar 199,800 0%44Ca HNO3

14N14N 16O 970 80%56Fe H2O 40Ar16O 2504 18%31P H2O 15N16O 1460 53%34S H2O 16O18O 1300 65%

75As HCl 40Ar35Cl 7725 2%51V HCl 35Cl16O 2572 18%

64Zn H2SO432S16O16O 1950 42%

24Mg Organics 12C12C 1600 50%52Cr Organics 40Ar12C 2370 20%

55Mn HNO340Ar15N 2300 20%

www.spec t roscopyon l i ne . com24 SPECTROSCOPY 16(11) NOVEMBER 2001

26 SPECTROSCOPY 16(11) NOVEMBER 2001

tion required to resolve fairly commonpolyatomic interferences from a selectedgroup of elemental isotopes, togetherwith the achievable ion transmission.

Figure 6 is a comparison between aquadrupole instrument and a magnetic-sector instrument with one of the mostcommon polyatomic interferences —40Ar16O on 56Fe, which requires a resolu-tion of 2504 to separate the peaks. Figure6a shows a spectral scan of 56Fe using aquadrupole instrument. What it doesn’tshow is the massive polyatomic interfer-ence 40Ar16O (produced by oxygen ionsfrom the water combining with argonions from the plasma) completely over-lapping the 56Fe. It shows very clearlythat these two masses are unresolvablewith a quadrupole. If that same spectralscan is performed on a magnetic-sectorinstrument, the result is the scan shownin Figure 6b. To see the spectral scan onthe same scale, it was necessary to exam-ine a much smaller range. For this rea-son, a 0.100-amu window was taken, as in-dicated by the dotted lines.

OTHER BENEFITSBesides high resolving power, another at-tractive feature of magnetic-sector instru-ments is their very high sensitivity com-bined with extremely low backgroundlevels. High ion transmission in low-resolution mode translates into sensitivityspecifications of typically 100–200 millioncounts per second (mcps) per ppm, whilebackground levels resulting from ex-tremely low dark current noise are typi-cally 0.1–0.2 cps. This compares with sen-sitivity of 10–50 mcps and backgroundlevels of �10 cps for a quadrupole instru-ment. For this reason detection limits, es-

pecially for high-mass elements like ura-nium where high resolution is generallynot required, are typically an order ofmagnitude better than those provided bya quadrupole-based instrument.

Besides good detection capability, an-other of the recognized benefits of themagnetic-sector approach is its ability toquantitate with excellent precision. Mea-surement of the characteristically flat-topped spectral peaks translates directlyinto high-precision data. As a result, inthe low-resolution mode, relative stan-dard deviation (RSD) values of0.01–0.05% are fairly common, whichmakes magnetic-sector instruments anideal tool for carrying out high-precision

isotope ratio work (7). Although preci-sion is usually degraded as resolution isincreased (because the peak shape getsworse), modern instrumentation withhigh-speed electronics and low mass biasis still capable of precision values of�0.1% RSD in medium- or high-resolutionmode (8).

The demand for ultrahigh-precisiondata, particularly in the field of geochem-istry, has led to the development of in-struments dedicated to isotope ratioanalysis. These are based on the double-focusing magnetic-sector design, but in-stead of using just one detector, these in-struments use multiple detectors. Oftenreferred to as multicollector systems,


100

80

60

40

20

0400 2000 4000 6000 8000 10,000

Resolution

Tran

smis

sion

(%)

Figure 5. Ion transmission with a magnetic-sector instrument decreases as the resolu-tion increases.

55 56 57

0.100 amu scan

56Fe 40Ar 16O

56Fe

(a)

m/z

Inte

nsity

55.935 55.957

40Ar 16O

(b)

m/z

Inte

nsity

Figure 6. Comparison of resolution between (a) a quadrupole and (b) a magnetic-sector in-strument for the polyatomic interference of 40Ar16O on 56Fe.

www.spec t roscopyon l i ne . com

NOVEMBER 2001 16(11) SPECTROSCOPY 27

they offer the capability of detecting andmeasuring multiple ion signals at exactlythe same time. As a result of this simulta-neous measurement approach, they arerecognized as producing the ultimate inisotope ratio precision (9).

There is no question that double-focusing magnetic-sector ICP-MS sys-tems are no longer a novel analyticaltechnique. They have proved themselvesto be a valuable addition to the trace ele-ment toolkit, particularly for challengingapplications that require good detectioncapability, exceptional resolving power,and very high precision. They do havetheir limitations, however, and perhapsshould not be considered a competitor forquadrupole instruments when it comes torapid, high-sample-throughput applica-tions or when performing multielementdeterminations on fast transient peaks,using sampling accessories such as elec-trothermal vaporization (10) or laserablation (11).

REFERENCES(1) R. Thomas, Spectroscopy 16(10), 44–48

(2001).(2) N. Bradshaw, E.F.H. Hall, and N.E.

Sanderson, J. Anal. At. Spectrom. 4,801–803 (1989).

(3) R. Hutton, A. Walsh, D. Milton, and J.Cantle, ChemSA 17, 213–215 (1991).

(4) U. Geismann and U. Greb, Fresnius’ J.Anal. Chem. 350, 186–193 (1994).

(5) U. Geismann and U. Greb, Poster Presen-tation at Second Regensburg Symposiumon Elemental Mass Spectrometry (1993).

(6) F. Adams, R. Gijbels, and R. Van Grieken,Inorganic Mass Spectrometry (John Wileyand Sons, New York, 1988).

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

(8) M. Hamester, D. Wiederin, J. Willis, W.Keri, and C.B. Douthitt, Fresnius’ J. Anal.Chem. 364, 495–497 (1999).

(9) J. Walder and P. A. Freeman, J. Anal. At.Spectrom. 7, 571 (1992).

(10) S. Beres, E. Denoyer, R. Thomas, and P.Bruckner, Spectroscopy 9(1), 20–26(1994).

(11) S. Shuttleworth and D. Kremser, J. Anal.At. Spectrom. 13, 697–699 (1998).

Robert Thomas has more than 30 years ofexperience in trace element analysis. He isthe principal of his own freelance writingand scientific consulting company, ScientificSolutions, based in Gaithersburg, MD. Hecan be contacted by e-mail at [email protected] or via his web site atwww.scientificsolutions1.com.�

. . . . . . . . . . . . . . . . . . . . . . . . .

Circle 21

Circle 20

36 SPECTROSCOPY 17(1) JANUARY 2002 www.spec t roscopyon l i ne . com

Continuing with the discussion onmass analyzers used in inductivelycoupled plasma–mass spectrometry(ICP-MS), let’s now turn our atten-tion to the most recent mass separa-

tion device to be commercialized — time-of-flight (TOF) technology. Although thefirst TOF mass spectrometer was first de-scribed in the literature in the late 1940s(1), it has taken more than 50 years toadapt it for use in an ICP-MS system. Therecent growth in TOF ICP-MS sales is inresponse to the technology’s unique abil-ity to sample all ions generated in theplasma at exactly the same time, which isadvantageous in three major areas:• Multielement determinations of rapid

transient signals generated by sam-pling accessories such as laser ablationand electrothermal vaporizationdevices

• High-precision, ratioing techniquessuch as internal standardization andisotope ratio analysis

• Rapid multielement measurements,especially where sample volume islimited. TOF’s simultaneous nature of sampling

ions offers distinct advantages over tradi-tional scanning (sequential) quadrupoletechnology for ICP-MS applicationswhere large amounts of data need to becaptured in a short amount of time. Be-fore we go on to discuss this in greaterdetail, let’s go through the basic princi-ples of TOF analyzers.

BASIC PRINCIPLES OF TOFAll TOF-MS instruments are based onthe same fundamental principle that thekinetic energy (Ek) of an ion is directlyproportional to its mass (m) and velocity(v), represented by equation 1

Ek � 1⁄2mv2 [1]

Therefore, if a population of ions — allhaving different masses — are given thesame kinetic energy by an acceleratingvoltage (U ), the velocities of the ions willall be different, based on their masses.This principle is then used to separateions of different mass-to-charge ratios(m/z) in the time (t) domain, over a fixedflight path distance (D) — represented byequation 2

m/z = 2Ut2/D2 [2]

This is shown schematically in Figure1, with three ions of different mass-to-charge ratios being accelerated into aflight tube and arriving at the detector atdifferent times. It can be seen that, basedon their velocities, the lightest ion arrivesfirst, followed by the medium mass ion,and finally the heaviest one. Using flighttubes of 1 m in length, even the heaviestions typically take less than 50 �s toreach the detector. This translates intoapproximately 20,000 mass spectra/s —approximately 2–3 orders of magnitudefaster than the sequential scanning modeof a quadrupole system.

DIFFERENT SAMPLING APPROACHESEven though this process sounds fairlystraightforward, sampling the ions in asimultaneous manner from a continuoussource of ions being generated in theplasma discharge is not a trivial task. Ba-sically two sampling approaches are usedin commercial TOF mass analyzers. Theyare the orthogonal design (2), where theflight tube is positioned at right angles tothe sampled ion beam, and the axial de-sign (3), where the flight tube is in thesame axis as the ion beam. In both de-signs, all ions that contribute to the massspectrum are sampled through the inter-face cones, but instead of being focused

into the mass filter in the conventionalway, packets (groups) of ions are electro-statically injected into the flight tube atexactly the same time. With the orthogo-nal approach, an accelerating potential isapplied at right angles to the continuousion beam from the plasma source. Theion beam is then chopped by using apulsed voltage supply coupled to the or-thogonal accelerator to provide repetitivevoltage slices at a frequency of a few kilo-hertz. The sliced packets of ions, whichare typically long and thin in cross sec-tion (in the vertical plane), are then al-lowed to drift into the flight tube wherethe ions are temporally resolved accord-ing to their velocities. Figure 2 shows thisprocess schematically.

With the axial approach, an accelerat-ing potential is applied axially (in thesame axis) to the incoming ion beam as itenters the extraction region. Because theions are in the same plane as the detec-tor, the beam has to be modulated usingan electrode grid to repel the gatedpacket of ions into the flight tube. Thiskind of modulation generates an ionpacket that is long and thin in cross sec-tion (in the horizontal plane). The differ-


A Beginner’s Guide to ICP-MSPart VIII — Mass Analyzers: Time-of-Flight Technology


Acceleratingvoltage

Flighttube

Flight path distance

Detector

Figure 1. Principles of ion detection usingTOF technology, showing separation ofthree masses in the time domain.

JANUARY 2002 17(1) SPECTROSCOPY 37

ent masses are then resolved in the timedomain in a similar manner to the orthog-onal design. The layout of an on-axisTOF system is shown schematically inFigure 3.

Figures 2 and 3 represent a rather sim-plistic explanation of TOF principles ofoperation. In practice, the many complexion focusing components in a TOF massanalyzer ensure that a maximum numberof analyte ions reach the detector andalso that undesired photons, neutralspecies, and interferences are ejectedfrom the ion beam. Some of these compo-nents are shown in Figure 4, whichshows a more detailed view of a typicalorthogonal system. This design showsthat an injector plate is used to injectpackets of ions at right angles from theion beam emerging from the MS inter-face. These packets of ions are then di-rected toward a deflection–steering platewhere pulsed voltages steer the ions (orthrow out unwanted species) in the direc-tion of a reflectron. The packets of ionsare then deflected back 180º, where theyare detected by a channel electron multi-plier or discrete dynode detector. Thereflectron is a type of ion mirror and func-tions as an energy compensation device,so that different ions of the same mass ar-rive at the detector at the same time.Even though the on-axis design mightuse slightly different components, theprinciples are very similar.

DIFFERENCES BETWEEN ORTHOGONALAND ON-AXIS TOF TECHNOLOGYAlthough there are real benefits of usingTOF over quadrupole technology forsome ICP-MS applications, each type ofTOF design also has subtle differences inits capabilities. (However, it is not the in-tent of this tutorial to make any personaljudgement about the benefits or disad-vantages of either design.) Let’s take alook at some of these differences ingreater detail (4, 5).

Sensitivity. The axial approach tends toproduce higher ion transmission becausethe steering components are in the sameplane as the ion generation system(plasma) and the detector. This meansthat the direction and magnitude of great-est energy dispersion is along the axis ofthe flight tube. In addition, when ions areextracted orthogonally, the energy dis-persion can produce angular divergenceof the ion beam resulting in poor trans-mission efficiency. However, the sensitiv-ity of either TOF design is still generally

lower than the latest commercial quadru-pole instruments.

Background levels. The on-axis designtends to generate higher background lev-els because neutral species and photonsstand a greater chance of reaching thedetector. This results in background lev-els in the order of 20–50 counts/s — approximately 5–10 times higher than theorthogonal design. However, because theion beam in the axial design has a smallercross section, a smaller detector can beused, which generally has better noisecharacteristics. In comparison, most com-mercial quadrupole instruments offerbackground levels of 1–10 counts/s,depending on the design.

Duty cycle. Duty cycle is usually definedas the fraction (percentage) of extractedions that actually make it into the massanalyzer. Unfortunately, with a TOF ICP-MS system that has to use pulsed ionpackets from a continuous source of ionsgenerated in the plasma, this process isnot very efficient. It should be empha-sized that even though the ions are sam-pled at the same time, detection is not si-multaneous because different massesarrive at the detector at different times.The difference between the samplingmechanisms of orthogonal and axial TOFdesigns translates into subtle differencesin their duty cycles.


Orthogonalaccelerator Flight tube DetectorIon

packets

Ion beam

Plasma torch

Figure 2. Schematic of an orthogonal acceleration TOF analyzer.

Flight tube Detector

Ionpacket

Ion beammodulation

Ion beam

Plasma torch

Figure 3. Schematic of an on-axis acceleration TOF analyzer.


With the orthogonal design, duty cycleis defined by the width of the extractedion packets, which are typically long andthin in cross section, as shown in Figure2. In comparison, the duty cycle of the ax-ial design is defined by the length of theextracted ion packets, which are typicallywide and thin in cross section, as shownin Figure 3. Duty cycle can be improvedby changing the cross-sectional area ofthe ion packet but, depending on the de-sign, is generally improved at the ex-pense of resolution. In practice, the duty

cycles for both orthogonal and axial de-signs are in the order of 15–20%.

Resolution. The resolution of theorthogonal approach is slightly betterbecause of its two-stage extraction/acceleration mechanism. Because a pulseof voltage pushes the ions from the ex-traction area into the acceleration region,the major energy dispersion lies alongthe axis of ion generation. For this rea-son, the energy spread is relatively smallin the direction of extraction compared tothe axial approach, resulting in better res-

olution. However, the resolving power ofboth commercial TOF ICP-MS systems istypically in the order of 500–2000 (4), de-pending on the mass region, whichmakes them inadequate to resolve manyof the problematic polyatomic species en-countered in ICP-MS (6). In comparison,commercial high-resolution systemsbased on the double-focusing magnetic-sector design offer resolving power ashigh as 10,000, while commercialquadrupoles typically achieve 300–400.

Mass bias. This is the degree to whichion transport efficiency varies with mass.All instruments show some degree ofmass bias, which is usually compensatedfor by measuring the difference betweenthe theoretical and observed ratio of twoisotopes of the same element. In TOF, thevelocity (energy) of the initial ion beamwill affect the instrument’s mass biascharacteristics. In theory, mass biasshould be less with the axial design be-cause the extracted ion packets don’thave any velocity in a direction perpen-dicular to the axis of the flight tube,which could potentially impact theirtransport efficiency.

BENEFITS OF TOF TECHNOLOGYFOR ICP-MSIt should be emphasized that these per-formance differences between the twodesigns are subtle and should not detractfrom the overall benefits of the TOF ap-proach for ICP-MS. As mentioned earlier,a scanning device such as a quadrupolecan only detect one mass at a time, whichmeans that a compromise always existsbetween number of elements, detectionlimits, precision, and the overall measure-ment time. However, with the TOF ap-proach, the ions are sampled at exactlythe same moment in time, which meansthat multielement data can be collectedwith no significant deterioration in qual-ity. The ability of a TOF system to capturea full mass spectrum, significantly fasterthan a quadrupole, translates into threemajor benefits.

RAPID TRANSIENT PEAK ANALYSISProbably the most exciting potential forTOF ICP-MS is in the multielement analy-sis of a rapid transient signal generatedby sampling accessories such as laser ab-lation (7), electrothermal vaporization,and flow injection systems (4). Eventhough a scanning quadrupole can beused for this type of analysis, it struggles


Deflection–steering plate

Reflectron

Interface

Detector

Injectorplate

Figure 4. A more detailed view of a typical orthogonal TOF analyzer, showing some of the ionsteering components.

2500

2000

1500

1000

500 3.251.75

0.25020 29 43 50 76 138 181 207

Mass (amu)

Cou

nts

Tim

e (s

)

Figure 5. A full mass scan of a transient signal generated by 10 �L of a 5-ppb multielementsolution using an electrothermal vaporization sampling accessory coupled to a TOF ICP-MSsystem (courtesy of GBC Scientific Equipment [Arlington Heights, IL]).


to produce high-quality, multielementdata when the transient peak lasts only afew seconds. The simultaneous nature ofTOF instrumentation makes it ideallysuited for this type of analysis, becausethe entire mass range can be collected in

less than 50 �s. Figure 5 shows a fullmass scan of a transient peak generatedby an electrothermal vaporization sam-pling accessory coupled to a TOF ICP-MS system. The technique has generateda healthy signal for 10 �L of a 5-ppb

multielement solution in less than 10 s.TOF technology is probably better suitedthan any other design of ICP-MS for thistype of application.

IMPROVED PRECISIONTo better understand how TOF technol-ogy can help improve precision in ICP-MS, it is important to know the majorsources of instability. The most commonsource of noise in ICP-MS is flicker noiseassociated with the sample introductionprocess (from peristaltic pump pulsa-tions, nebulization mechanisms, andplasma fluctuations) and shot noise de-rived from photons, electrons, and ionshitting the detector. Shot noise is basedon counting statistics and is directly pro-portional to the square root of the signal.It therefore follows that as the signal in-tensity gets larger, the shot noise has lessof an impact on the precision (% RSD) ofthe signal. At high ion counts the mostdominant source of imprecision in ICP-MS is derived from flicker noise gener-ated in the sample introduction area.

One of the most effective ways to re-duce instability produced by flicker noiseis to use a technique called internal stan-dardization, where the analyte signal iscompared and ratioed to the signal of aninternal standard element (usually of sim-ilar mass and ionization characteristics)that is spiked into the sample. Eventhough a quadrupole-based system cando an adequate job of compensating forthese signal fluctuations, it is ultimatelylimited by its inability to measure the in-ternal standard at exactly the same timeas the analyte isotope. So to compensate


In reference to Part VII of my tutorial series, “A Beginner’sGuide to ICP-MS,” which was published in the November

2001 issue of Spectroscopy, I would like to make a numberof corrections. Even though the intent of the article was togive a general overview of double-focusing magnetic-sectormass analyzers for beginners, Thermo Finnigan contactedSpectroscopy to inform the editors and me that the columncontained errors, and that it did not reflect the current per-formance of their instrument, the ELEMENT2. For that rea-son, I wish to make the following amendments to the article.

• My statement that double-focusing magnetic sectorICP-MS instruments are signif icantly slower than quadrupole technology does not hold true today. Recent improvements in the scan rate of the ELEMENT2 translatesinto speeds approaching that of quadrupole-based instruments.

• My statement that typical scan speeds for a full massscan were 400–500 ms is reflective of older magnetic sec-tor technology. This is not representative of the ELEMENT2,which has a scan speed in the order of 150–200 ms.

• My statement that typical sensitivity was in the order of100–200 mil l ion cps/ppm is not reflective of the ELEMENT2, which has a specification for 115In of 1 billioncps/ppm.

• My conclusion should therefore be modified to say thatif transient peak analysis is a requirement, modern double-focusing magnetic sector technology such as the ELEMENT2, with its improved scan speeds, should be con-sidered a viable option to quadrupole technology.

I wish to apologize for any inconvenience caused bythese statements.Robert Thomas

Correction

Circle 33

JANUARY 2002 17(1) SPECTROSCOPY 41

for sample introduction– and plasma-based noise and achieve high precision,the analyte and internal standard isotopesneed to be sampled and measured simul-taneously. For this reason, the design of aTOF mass analyzer is perfect for truesimultaneous internal standardization re-quired for high-precision work. It follows,therefore, that TOF is also well suited forhigh-precision isotope ratio analysiswhere its simultaneous nature of mea-surement is capable of achieving preci-sion values close to the theoretical limitsof counting statistics. And unlike a scan-ning quadrupole-based system, it canmeasure ratios for as many isotopes orisotopic pairs as needed — all with excel-lent precision (8).

ANALYSIS TIMEAs with a scanning ICP–optical emissionspectroscopy system, the speed of aquadrupole ICP mass spectrometer is lim-ited by its scanning rate. To determine 10elements in duplicate with good precisionand detection limits, an integration time of3 s/mass is normally required. Whenoverhead scanning and settling times areadded for each mass and each replicate,this translates to approximately 2min/sample. With a TOF system, thesame analysis would take significantlyless time because all the data are capturedsimultaneously. In fact, detection limit lev-els in a TOF instrument are typicallyachieved using a 10–30 s integration time,which translates into a 5–10-fold improve-ment in measurement time over a quadru-pole instrument. The added benefit of aTOF instrument is that the speed of analy-sis is not impacted by the number of ana-lytes being determined. It wouldn’t matterif the suite of elements in the method was10 or 70 — the measurement time wouldbe approximately the same. However, onepoint must be stressed: A large portion ofthe overall analysis time is taken up withflushing an old sample out and pumping anew sample into the sample introductionsystem. This can be as much as 2min/sample for real-world matrices. Sowhen this time is taken into account, thedifference between the sample through-put of a quadrupole system and a TOFICP-MS system is not so evident.

TOF ICP-MS, with its rapid, simultane-ous mode of measurement, excels at mul-tielement applications that generate fasttransient signals. It offers excellent preci-sion, particularly for isotope-ratioing

S P E C T R O S C O P Y T U T O R I A L . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Circle 34

techniques, and also has the capability forhigh speeds of analysis. However, eventhough it has enormous potential, TOFwas only commercialized in 1998, so it isrelatively immature compared withquadrupole ICP-MS technology, which isalmost 20 years old. For that reason,there is currently only a small number ofTOF instruments carrying out high-throughput, routine applications.

REFERENCES(1) A.E. Cameron and D.F. Eggers, The Re-

view of Scientific Instruments 19(9), 605(1948).

(2) D.P. Myers, G. Li, P. Yang, and G.M.Hieftje, J. Am. Soc. Mass Spectrom. 5,1008–1016 (1994).

(3) D.P. Myers, paper presented at 12th Asilo-mar Conference on Mass Spectrometry,Pacific Grove, CA, Sept. 20–24, (1996).

(4) R.E. Sturgeon, J.W.H. Lam, and A. Saint,J. Anal. At. Spectrom. 15, 607–616,(2000).

(5) F. Vanhaecke, L. Moens, R. Dams, L.Allen, and S. Georgitis, Anal. Chem. 71,3297 (1999).

(6) N. Bradshaw, E.F. Hall, and N.E. Sander-son, J. Anal. At. Spectrom. 4, 801–803(1989).

(7) P. Mahoney, G. Li, and G.M. Hieftje, J.Am. Soc. Mass Spectrom. 11, 401–406(1996).

(8) K.G. Heumann, S.M. Gallus, G.Radlinger, and J. Vogl, J. Anal. At. Spec-trom. 13, 1001 (1998).

Robert Thomas has more than 30 years ofexperience in trace element analysis. He isthe principal of his own freelance writingand scientific consulting company, ScientificSolutions, based in Gaithersburg, MD. Hecan be contacted by e-mail at [email protected] or via his web site atwww.scientificsolutions1.com. �

small number ofelements are recog-nized as having poordetection limits by

inductively coupled plasmamass spectrometry (ICP-MS).These elements are predomi-nantly ones that suffer frommajor spectral interferencesgenerated by ions derived fromthe plasma gas, matrix compo-nents, or the solvent–acid usedto get the sample into solution.Examples of these interferencesinclude:• 40Ar16O on the determination

of 56Fe • 38ArH on the determination

of 39K • 40Ar on the determination of

40Ca • 40Ar40Ar on the determination

of 80Se • 40Ar35Cl on the determination

of 75As • 40Ar12C on the determination

of 52Cr • 35Cl16O on the determination

of 51V.

AA The cold/cool plasma ap-proach, which uses a lower tem-perature to reduce the forma-tion of the argon-basedinterferences, is a very effectiveway to get around some of theseproblems (1); however, it issometimes difficult to optimize,it is only suitable for a few of theinterferences, it is susceptible tomore severe matrix effects, andit can be time consuming tochange back and forth betweennormal- and cool-plasma con-ditions. These limitations andthe desire to improve perform-ance led to the development ofcollision/reaction cells in thelate 1990s. Designed originallyfor organic MS to generatedaughter species to confirmidentification of the structure ofthe parent molecule (2), theywere used in ICP-MS to stop theappearance of many argon-based spectral interferences.

Basic Principles of Collision/Reaction CellsWith this approach, ions enter

the interface in the normalmanner, where they are ex-tracted under vacuum into acollision/reaction cell that is po-sitioned before the analyzerquadrupole. A collision/reactiongas such as hydrogen or heliumis then bled into the cell, whichconsists of a multipole (aquadrupole, hexapole, or octa-pole), usually operated in theradio frequency (rf)-only mode.The rf-only field does not sepa-rate the masses like a traditionalquadrupole, but instead has theeffect of focusing the ions,which then collide and reactwith molecules of the collision/reaction gas. By a number ofdifferent ion-molecule collisionand reaction mechanisms, poly-atomic interfering ions like 40Ar,40Ar16O, and 38ArH, will either beconverted to harmless noninter-fering species, or the analyte willbe converted to another ionwhich is not interfered with.This is exemplified by the reac-tion [1], which shows the use ofhydrogen gas to reduce the38ArH polyatomic interference inthe determination of 39K. Hy-drogen gas converts 38ArH to theharmless H3

� ion and atomicargon, but does not react withthe potassium. The 39K analyteions, free of the interference,then emerge from thecollision/reaction cell, wherethey are directed toward thequadrupole analyzer for normalmass separation.

www.spectroscopyonl ine.com42 Spectroscopy 17(2) February 2002

A Beginner‘s Guide to ICP-MSPart IX — Mass Analyzers: Collision/Reaction Cell TechnologyRobert Thomas

T U T O R I A LT U T O R I A L

RobertThomashas more than30 years ofexperience intrace elementanalysis. He isthe principal ofhis ownfreelance writingand consultingcompany,ScientificSolutions, basedin Gaithersburg,MD. He can becontacted by e-mail [email protected] via his website at www.scientificsolutions1.com.

The detection capability of traditional quadrupole mass analyzers forsome critical elements is severely compromised by the formation ofpolyatomic spectral interferences generated by either argon, solvent, orsample-based ionic species. Although there are ways to minimize theseinterferences — including correction equations, cool plasma technology,and matrix separation — they cannot be completely eliminated. How-ever, a new approach called collision/reaction cell technology has re-cently been developed that virtually stops the formation of many ofthese harmful species before they enter the mass analyzer. Part IX ofthis series takes a detailed look at this innovative new technique andthe exciting potential it has to offer.

February 2002 17(2) Spectroscopy 43

Tutorial

38ArH� � H2 � H3� � Ar

39K� � H2 � 39K� � H2 (no reaction) [1]

The layout of a typical collision/reaction cell instrument is shown inFigure 1.

Different Collision/Reaction ApproachesThe previous example is a very simplis-tic explanation of how a collision/reaction cell works. In practice, com-plex secondary reactions and collisionstake place, which generate many unde-sirable interfering species. If thesespecies were not eliminated or rejected,they could potentially lead to additionalspectral interferences. Basically two ap-proaches are used to reject the productsof these unwanted interactions:• Discrimination by kinetic energy • Discrimination by mass.

The major differences between thetwo approaches are in the types of mul-tipoles used and their basic mechanismfor rejection of the interferences. Let’stake a closer look at how they differ.

Discrimination by Kinetic Energy The first commercial collision cells forICP-MS were based on hexapole tech-nology (3), which was originally de-signed for the study of organic mole-cules using tandem MS. The morecollision-induced daughter species thatwere generated, the better the chance ofidentifying the structure of the parentmolecule; however, this very desirablecharacteristic for liquid chromatogra-

phy or electrospray MS studies was adisadvantage in inorganic MS, wheresecondary reaction–product ions aresomething to be avoided. There wereways to minimize this problem, butthey were still limited by the type ofcollision gas that could be used. Unfor-tunately, highly reactive gases — such asammonia and methane, which are moreefficient at interference reduction —could not be used because of the limita-tions of a non-scan-ning hexapole (in rf-only mode) toadequately control thesecondary reactions.The fundamental rea-son is that hexapolesdo not provide ade-quate mass discrimi-nation capabilities tosuppress the un-wanted secondary re-actions, which necessi-tates the need forkinetic energy dis-crimination to distin-guish the collisionproduct ions from theanalyte ions. This istypically achieved bysetting the collisioncell bias slightly lesspositive than the massfilter bias. This meansthat the collision-product ions, whichhave the same energyas the cell bias, are dis-criminated against

and rejected, while the analyte ions,which have a higher energy than the cellbias, are transmitted.

The inability to adequately controlthe secondary reactions meant that lowreactivity gases like He, H2, and Xe werethe only option. The result was thation-molecule collisional fragmentation(and not reactions) was thought to bethe dominant mechanism of interfer-ence reduction. So even though the iontransmission characteristics of a hexa-pole were considered very good (withrespect to the range of energies andmasses transmitted), background levelswere still relatively high because the in-terference rejection process was notvery efficient. For this reason, its detec-tion capability — particularly for someof the more difficult elements, like Fe,K, and Ca — offered little improvementover the cool plasma approach. Table Ishows some typical detection limits inppb achievable with a hexapole-basedcollision cell ICP-MS system (4).

Recent modifications to the hexapoledesign have significantly improved its

Analyzingquadrupole

Multipolereaction–

collision cell

Ion lenssystem

ICP-MSinterface

ICP

Detector

Gas inlet

Figure 1. Layout of a typical collision/reaction cell instrument.

Table I. Typical detection limits (in ppb) achievable with a hexapole-based collision cell ICP-MS system (4).

Elemental DetectionSensitivity Limit

Element Isotope (cps/[�g/mL]) (ppb)Be 9 6.9 � 107 0.0077Mg 24 1.3 � 108 0.028Ca 40 2.8 � 108 0.07V 51 1.7 � 108 0.0009Cr 52 2.4 � 108 0.0007Mn 55 3.4 � 108 0.0017Fe 56 3.0 � 108 0.017Co 59 2.7 � 108 0.0007Ni 60 2.1 � 108 0.016Cu 63 1.9 � 108 0.003Zn 68 1.1 � 108 0.008Sr 88 4.9 � 108 0.0003Ag 107 3.5 � 108 0.0003Cd 114 2.4 � 108 0.0004Te 128 1.3 � 108 0.009Ba 138 5.9 � 108 0.0002Tl 205 4.0 � 108 0.0002Pb 208 3.7 � 108 0.0007Bi 209 3.4 � 108 0.0005U 238 2.3 � 108 0.0001


Tutorial

collision/reaction characteristics. In ad-dition to offering good transmissioncharacteristics and kinetic energy dis-crimination, they now appear to offerbasic mass-dependent discriminationcapabilities. This means that the kineticenergy discrimination barrier can beadjusted with analytical mass, which of-fers the capability of using smallamounts of highly reactive gases. Figure2 shows an example of the reduction ofboth 40Ar12C and 37Cl16O using heliumwith a small amount of ammonia, inthe isotopic ratio determination of

52Cr/53Cr (52Cr is 83.789% and 53Cr is9.401% abundant). It can be seen thatthe 52Cr/53Cr ratio is virtually the samein the chloride and carbon matrices asit is with no matrix present when theoptimum flow of collision/reaction gasis used (5).

Another way to discriminate by ki-netic energy is to use an octapole in thecollision/reaction cell instead of a hexa-pole. The benefit of using a higherorder design is that its transmissioncharacteristics, particularly at the lowmass end, are slightly higher than lower

order multipoles. Similar in design tothe hexapole, collisional fragmentationand energy discrimination are the pre-dominant mechanisms for interferencereduction, which means that lower re-activity gases like hydrogen and heliumare preferred. By careful design of theinterface and the entrance to the cell,the collision/reaction capabilities can beimproved, by reducing the number ofsample/solvent/plasma-based ions en-tering the cell. This enables the collisiongas to be more effective at reducing theinterferences. An example of this is theuse of H2 as the cell gas to reduce theargon dimer (40Ar2

�) interference in thedetermination of the major isotope ofselenium at mass 80 (80Se). Figure 3shows an example of a dramatic reduc-tion in the 40Ar2

� background at mass 80using an ICP-MS fitted with an octa-pole reaction cell. By using the opti-mum flow of H2, the spectral back-ground is reduced by about six ordersof magnitude, from 10,000,000 cps to10 cps, producing a background equiva-lent concentration of approximately 1 ppt for 80Se (6).

Discrimination by MassThe other way to reject the products ofthe secondary reactions/collisions is todiscriminate them by mass. Unfortu-nately, higher order multipoles cannotbe used for efficient mass discrimina-tion because the stability boundariesare diffuse, and sequential secondary

14

12

10

8

6

4

20 1 2 3 4 5 6

Collision gas flow rate (mL/min)

52Cr

/53Cr

No matrix0.2% Cl matrix0.06% C matrix

Figure 2. The use of a helium/ammonia mixture with a hexapole-basedcollision cell for the successful determination of 52Cr/53Cr isotopic ratios(courtesy of Thermo Elemental, Franklin, MA).

0 0.7 1.1 1.5 1.9 2.3 2.7 3.1 3.5 3.9 4.3 4.7 5.0

Hydrogen flow (mL/min)

Sign

al a

t mas

s 80

(cps

)

BEC

(ppb

)

107

106

105

104

103

102

101

100

104

103

102

101

100

10�1

10�2

10�3

1ppb Se (cps)

Blank (cps)

BEC (ppb)

Figure 3. Background reduction of the argon dimer (40Ar2�) with

hydrogen gas using an octapole reaction cell (Courtesy of AgilentTechnologies, Wilmington, DE).

Quadrupolemass filter

Analyzerquadrupole

Ion–moleculereactions and collisions

Gas inlet

Reactive gas (NH3)

++ ++

++

++++

++ ++++

++++++++++

++ ++

++

++++

++

40Ar16O

ArO� � NH3 O � Ar � NH3�

56Fe

56Fe

NH3�

O

Ar

Figure 4. Elimination of the ArO interference with a dynamic reaction cell.


pole electrical fields, unwanted reac-tions between the gas and the samplematrix or solvent (which could poten-tially lead to new interferences) are pre-vented. Therefore, every time an analyteand interfering ions enter the dynamicreaction cell, the bandpass of thequadrupole can be optimized for thatspecific problem and then changed on-the-fly for the next one. Figure 4 showsa schematic of an analyte ion 56Fe andan isobaric interference 40Ar16O enteringthe dynamic reaction cell. The reactiongas NH3 reacts with the ArO� to formatomic oxygen and argon together witha positive NH3 ion. The quadrupole’selectrical field is then set to allow thetransmission of the analyte ion 56Fe tothe analyzer quadrupole, free of theproblematic isobaric interference,40Ar16O. In addition, the NH3

� is pre-

vented from reacting further to producea new interfering ion. The advantage ofthis approach is that highly reactivegases can be used, which increases thenumber of ion–molecule reactions tak-ing place and therefore more efficientremoval of the interfering species. Ofcourse, this also potentially generatesmore side reactions between the gasand the sample matrix and solvent;however, by dynamically scanning thebandpass of the quadrupole in the reac-tion cell, these reaction by-products arerejected before they can react to formnew interfering ions.

The benefit of the dynamic reactioncell is that by careful selection of the re-action gas, the user takes advantage ofthe different rates of reaction of the an-alyte and the interfering species. Thisprocess is exemplified by the elimina-tion of 40Ar� by NH3 gas in the determi-nation of 40Ca. The reaction between

Tutorial

reactions cannot be easily intercepted.The way around this problem is to use aquadrupole (instead of a hexapole oroctapole) inside the reaction/collisioncell, and use it as a selective bandpassfilter. The benefit of this approach isthat highly reactive gases can be used,which tend to be more efficient at inter-ference reduction. One such develop-ment that uses this approach is calleddynamic reaction cell technology (7, 8).Similar in appearance to the hexapoleand octapole collision/reaction cells, the

dynamic reaction cell is a pressurizedmultipole positioned before the ana-lyzer quadrupole. However, the similar-ity ends there. In dynamic reaction celltechnology, a quadrupole is used in-stead of a hexapole or octapole. Ahighly reactive gas, such as ammonia ormethane, is bled into the cell, which is acatalyst for ion molecule chemistry totake place. By a number of different re-action mechanisms, the gaseous mole-cules react with the interfering ions toconvert them either into an innocuous

species different from theanalyte mass or a harmlessneutral species. The analytemass then emerges from thedynamic reaction cell free ofits interference and steeredinto the analyzer quadru-pole for conventional massseparation. The advantagesof using a quadrupole inthe reaction cell is that thestability regions are muchbetter defined than a hexa-pole or an octapole, so it isrelatively straightforward tooperate the quadrupole in-side the reaction cell as amass or bandpass filter, andnot just an ion-focusingguide. Therefore, by carefuloptimization of the quadru-

��

�

�

�

�

Ca�

Ar�

Ar� � NH3

(Exothermic reaction)

NH3� � Ar

Ca� � NH3

(Endothermic reaction)

no reaction

NH3

Figure 5. The reaction between NH3 and Ar� is exothermic and fast, while there is no reactionbetween NH3 and Ca� in the dynamic reaction cell.

Table II. Typical detection limits in ppt of an ICP-MS system fitted with a dynamicreaction cell (9).

Detection DetectionAnalyte Limit (ppt) Analyte Limit (ppt)

Li 0.08 Co 0.07Be 0.6 60Ni 0.4B 1.1 Zn 1Na 0.3 As 1.2Mg 0.6 Se* 5Al 0.07 Sr 0.02K* 1 Rh 0.0140Ca* 1 In 0.01V* 0.3 Sb 0.06Cr* 0.25 Cs 0.03Mn* 0.09 Pb 0.0356Fe* 0.15 U 0.01

* Indicates elements determined in dynamic reaction cell mode.

Collision/reaction cells

have given a new lease on

life to quadrupole mass

analyzers used in ICP-MS.


Tutorial

NH3 gas and the 40Ar� interference,which is predominantly a charge ex-change, occurs because the ionizationpotential of NH3 (10.2 eV) is low com-pared with that of Ar (15.8 eV). There-fore, the reaction is exothermic and fast;

however, the ionizationpotential of Ca (6.1 eV)is significantly less thanthat of NH3, so the reac-tion, which is endother-mic, is not allowed toproceed (8). Figure 5shows this process ingreater detail.

This highly efficientreaction mechanismtranslates into a dra-matic reduction of thespectral background atmass 40, which isshown graphically inFigure 6. At the opti-mum NH3 flow, a re-duction in the 40Ar

background signal of about eight ordersof magnitude is achieved, resulting in adetection limit of 0.5–1.0 ppt for 40Ca.

Table II shows some typical detectionlimits in parts per trillion (ppt) of anICP-MS system fitted with a dynamic

reaction cell. The elements with an as-terisk were determined using ammoniaor methane as the reaction gas, whilethe other elements were determined inthe standard mode (no reaction gas).

Collision/reaction cells have given anew lease on life to quadrupole massanalyzers used in ICP-MS. They haveenhanced its performance and flexibil-ity, and most definitely opened up thetechnique to more-demanding applica-tions that were previously beyond itscapabilities. However, it must be em-phasized that even though differencesexist between commercially availableinstruments, they all perform very well.The intent of this tutorial is to presentthe benefits of the technology to begin-ners and give an overview of the differ-ent approaches available. If it has cre-ated an interest, I strongly suggest that aperformance evaluation is made basedon your own sample matrices.

References1. K. Sakata and K. Kawabata, Spec-

trochimica Acta 49B, 1027 (1994).2. B.A. Thomson, D.J. Douglas, J.J. Corr,

J.W. Hager, and C.A. Joliffe, Anal. Chem.67, 1696–1704 (1995).

3. P. Turner, T. Merren, J. Speakman, andC. Haines, in Plasma Source MassSpectrometry: Developments and Ap-plications, eds. G. Holland and S. Tan-ner (Royal Society of Chemistry, Cam-bridge, UK, 1996).

4. I. Feldmann, N. Jakubowski, C. Thomas,and D. Stuewer, Fresenius‘ J. Anal.Chem. 365, 422–428 (1999).

5. ”Collision Cell Technology with EnergyDiscrimination,“ Thermo Elemental Ap-plication Note (Thermo Elemental,Franklin, MA, September 2001).

6. E. McCurdy and D. Potter, Agilent Tech-nologies ICP-MS Journal 10, October2001.

7. Covered by U.S. Patent No. 6140638.8. S.D. Tanner and V.I. Baranov, At. Spectr.

20(2), 45–52, (1999).9. U. Voellkopf, K. Klemm, and M. Pfluger,

At. Spectr. 20(2), 53–59, (1999). �

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

NH3 flow (mL/min)

Ion

sign

al (c

ps)

109

108

107

106

105

104

103

102

101

100 ppt Ca

Water blank

Figure 6. A reduction of eight orders of magnitude in the 40Arbackground signal is achievable with the dynamic reaction cell —resulting in �1 ppt detection limit for 40Ca.

Circle 28

more secondary electrons form.The potential gradient insidethe tube varies based on posi-tion, so the secondary electronsmove farther down the tube. Asthese electrons strike new areasof the coating, more secondaryelectrons are emitted. Thisprocess is repeated many times.The result is a discrete pulse thatcontains many millions of elec-trons generated from an ionthat first hits the cone of the de-tector (1). This process is shownsimplistically in Figure 1.

This pulse isthen sensedand detectedby a very fastpreamplifier.The outputpulse from thepreamplifierthen goes to adigital dis-criminator andcounting cir-cuitry, whichcounts onlypulses above a

certain threshold value. Thisthreshold level needs to be highenough to discriminate againstpulses caused by spurious emis-sion inside the tube, stray pho-tons from the plasma itself, orphotons generated from fastmoving ions striking thequadrupole rods.

Sometimes the rate of ionshitting the detector is too highfor the measurement circuitryto handle in an efficient man-

emission spectroscopy (ICP-OES); however, instead of usingindividual dynodes to convertphotons to electrons, the chan-neltron is an open glass cone —

coated with a semiconductor-type material — that generateselectrons from ions impingingon its surface. For the detectionof positive ions, the front of thecone is biased at a negative po-tential and the far end, nearestthe collector, is kept at ground.When the ion emerges from thequadrupole mass analyzer, it isattracted to the high negativepotential of the cone. When theion hits this surface, one or

For some applications where

ultratrace detection limits are

not required, the ion beam from

the mass analyzer is directed

into a simple metal electrode,

or Faraday cup.

www.spectroscopyonl ine.com34 Spectroscopy 17(4) April 2002

ince inductively coupledplasma–mass spectrom-etry (ICP-MS) was firstintroduced in the early

1980s, a number of different iondetection designs have beenused, the most popular beingelectron multipliers for low ion-count rates, and Faraday collec-tors for high-count rates. Today,the majority of ICP-MS systemsused for ultratrace analysis usedetectors that are based on theactive film or discrete dynodeelectron multiplier. They arevery sophisticated pieces ofequipment compared with ear-lier designs and are very effi-cient at converting ion currentsinto electrical signals. Before wedescribe these detectors ingreater detail, it is worth lookingat two of the earlier designs —the channel electron multiplier(channeltron) and the Faradaycup — to get a basic under-standing of how the ICP-MS iondetection process works.

Channel electron multiplier. Theoperating principles of thechannel electron multiplier aresimilar to those of a photomul-tiplier tube used in ICP–optical

A Beginner‘s Guide to ICP-MSPart X — DetectorsRobert Thomas



Part X of this series on ICP-MS discusses the detection system — animportant area of the mass spectrometer that counts the number of ionsemerging from the mass analyzer. The detector converts the ions intoelectrical pulses, which are then counted by its integrated measurementcircuitry. The magnitude of the electrical pulses corresponds to thenumber of analyte ions present in the sample. Trace element quantita-tion in an unknown sample is then carried out by comparing the ionsignal with known calibration or reference standards.

SSFor some applications where

ultratrace detection limits are

not required, the ion beam from

the mass analyzer is directed

into a simple metal electrode,

or Faraday cup.

way the measurement circuitry handleslow and high ion-count rates. WhenICP-MS was first commercialized, itcould only handle as many as five or-ders of dynamic range; however, whenattempts were made to extend the dy-namic range, certain problems were en-countered. Before we discuss how mod-ern detectors deal with this issue, let’sfirst take a look at how it was addressedin earlier instrumentation.

Extending the Dynamic Range Traditionally, ICP-MS using the pulsecounting measurement is capable ofabout five orders of linear dynamicrange. This means that ICP-MS calibra-tion curves, generally speaking, are lin-ear from ppt levels to as much as a fewhundred parts-per-billion. However, anumber of ways exist to extend the dy-namic range of ICP-MS another threeto four orders of magnitude to workfrom sub-part-per-trillion levels, to asmuch as 100 ppm. Following is a briefoverview of some of the different ap-proaches that have been used.

Filtering the ion beam. One of the firstapproaches to extend the dynamicrange in ICP-MS was to filter the ionbeam by putting a non-optimum volt-age on one of the ion lens componentsor the quadrupole itself to limit thenumber of ions reaching the detector.This voltage offset, which was set on anindividual mass basis, acted as an en-ergy filter to electronically screen theion beam and reduce the subsequention signal to within a range covered bypulse-counting ion detection. The maindisadvantage with this approach wasthat the operator had to have priorknowledge of the sample to know whatvoltage to the apply to the high concen-tration masses.

in the early 1990s to develop an ICP-MS system using a Faraday cup detectorfor environmental applications, but itssensitivity was compromised and, as aresult, it was considered more suitablefor applications requiring ICP-OES de-tection capability. However, Faradaycup technology is still used in somemagnetic sector instruments, particu-larly where high ion signals areencountered in the determination of

high-precision isotope ratios using amulticollector detection system.

Discrete dynode electron multiplier. Thesedetectors, which are often called activefilm multipliers, work in a similar wayto the channeltron, but use discretedynodes to carry out the electron mul-tiplication (2). Figure 2 illustrates theprinciples of operation of this device.The detector is usually positioned off-axis to minimize the background fromstray radiation and neutral species com-ing from the ion source. When an ionemerges from the quadrupole, it sweepsthrough a curved path before it strikesthe first dynode. On striking the firstdynode, it liberates secondary electrons.The electron-optic design of thedynode produces acceleration of thesesecondary electrons to the next dynode,where they generate more electrons.This process is repeated at each dynode,generating a pulse of electrons that is fi-nally captured by the multiplier collec-tor or anode. Because of the materialsused in the discrete dynode detectorand the difference in the way electronsare generated, it is typically more sensi-tive than channeltron technology.

Although most discrete dynode de-tectors are very similar in the way theywork, there are subtle differences in the

When ICP-MS was first commercialized, it could only

handle as many as five orders of dynamic range;

when attempts were made to extend the dynamic

range, certain problems were encountered.

April 2002 17(4) Spectroscopy 35

Tutorial

ner. This situation is caused by ions ar-riving at the detector during the outputpulse of the preceding ion and notbeing detected by the counting system.This “dead time,” as it is known, is afundamental limitation of the multi-plier detector and is typically 30–50 ns,depending on the detection system.Compensation in the measurement cir-cuitry has to be made for this dead timein order to count the maximum num-ber of ions hitting the detector.

Faraday cup. For some applicationswhere ultratrace detection limits arenot required, the ion beam from themass analyzer is directed into a simplemetal electrode, or Faraday cup (1).With this approach, there is no controlover the applied voltage (gain), so aFaraday cup can only be used for highion currents. Their lower working rangeis in the order of 104 counts/s, whichmeans that if a Faraday cup is to beused as the only detector, the sensitivityof the ICP mass spectrometer will beseverely compromised. For this reason,Faraday cups are normally used in con-junction with a channeltron or discretedynode detector to extend the dynamicrange of the instrument. An additionalproblem with the Faraday cup is that,because of the time constant used in thedc amplification process to measure theion current, it is limited to relatively lowscan rates. This limitation makes it un-suitable for the rapid scan rates re-quired for traditional pulse countingused in ICP-MS and also limits its abil-ity to handle fast transient peaks.

The Faraday cup never became pop-ular with quadrupole ICP-MS systemsbecause it wasn’t suitable for very lowion-count rates. An attempt was made

Preamplifier

�3 kV

( )

Secondaryelectrons

Ions frommass analyzer

Figure 1. The path of an ion through a channelelectron multiplier.

When ICP-MS was first commercialized, it could only

handle as many as five orders of dynamic range;

when attempts were made to extend the dynamic

range, certain problems were encountered.

Figure 3 (above). Dual-stage discrete dynode detector measurement circuitry. (Figures 3, 4, and 5are courtesy of PerkinElmer Instruments, Shelton, CT.)


Tutorial

Using two detectors. Another techniquethat was implemented in some of theearly quadrupole ICP-MS instrumenta-tion was to use two different detectors,such as a channel electron multiplier tomeasure low current signals, and aFaraday cup to measure high ion cur-rents. This process worked reasonablywell, but struggled with some applica-tions because it required rapid switch-ing between the two detectors. Theproblem was that the ion beam had tobe physically deflected to select the op-timum detector. Not only did this de-grade the measurement duty cycle, butdetector switching and stabilizationtimes of several seconds also precludedfast transient signal detection.

The more modern approach is to usejust one detector to extend the dynamic

range. By using the detector in both thepulse-counting and analog modes, highand low concentrations can be deter-mined in the same sample. Three ap-proaches use this type of detection sys-tem; two of them involve carrying outtwo scans of the sample, while the thirduses only one scan.

Using two scans with one detector. Thefirst approach uses an electron multi-plier operated in both digital and ana-log modes (3). Digital counting pro-vides the highest sensitivity, whileoperation in the analog mode (achievedby reducing the high voltage applied tothe detector) is used to reduce the sen-sitivity of the detector, thus extendingthe concentration range for which ionsignals can be measured. The system isimplemented by scanning the spec-

trometer twice for each sample. A firstscan, in which the detector is operatedin the analog mode, provides signals forelements present at high concentra-tions. A second scan, in which the de-tector voltage is switched to digital-pulse counting mode, provides highsensitivity detection for elements pres-ent at low levels. A major advantage ofthis technology is that users do notneed to know in advance whether to useanalog or digital detection because thesystem automatically scans all elementsin both modes. However, its disadvan-tage is that two independent mass scansare required to gather data across an ex-tended signal range. This not only re-sults in degraded measurement effi-ciency and slower analyses, but it alsomeans that the system cannot be usedfor fast transient signal analysis of un-known samples because mode switch-ing is generally too slow.

The second way of extending the dy-namic range is similar to the first ap-proach, except that the first scan is usedas an investigative tool to examine thesample spectrum before analysis (4).This first prescan establishes the masspositions at which the analog and pulsemodes will be used for subsequently col-lecting the spectral signal. The secondanalytical scan is then used for data col-lection; the system switches the detectorback and forth rapidly between pulseand analog mode depending on theconcentration of each analytical mass.

The main disadvantage of these twoapproaches is that two separate scansare required to measure high and lowlevels. With conventional nebulization,this isn't such a major problem exceptthat it can impact sample throughput.However, it does become a concernwhen it comes to working with tran-sient peaks found in electrothermal va-porization, flow injection, or laser sam-pling ICP-MS. Because these transientpeaks often last only a few seconds, allthe available time must be spent meas-uring the masses of interest to get thebest detection limits. When two scanshave to be made, valuable time is

Ion path

Noise

Individual dynodesQuadrupole rods

Generationof electrons

Figure 2 (left). Schematic of a discrete dynodeelectron multiplier.

Incoming ion

DetectorMCA

Analog signal

Pulse signal

Counter1

Counter2

To quadrupole

Scancontroller

Midpointdynode

Data system

Figure 4. The pulse-counting mode covers rates as high as 106 counts/s, and the analog circuitry issuitable from 104 to 109 counts/s with a dual-mode discrete dynode detector.


Tutorial

wasted, which is not contributing toquality of the analytical signal.

Using one scan with one detector. Theselimitations of using two scans led tothe development of a third approachusing a dual-stage discrete dynode de-tector (5). This technology uses meas-urement circuitry that allows bothhigh and low concentrations to be de-termined in one scan. This is achievedby measuring the ion signal as an ana-log signal at the midpoint dynode.When more than a threshold numberof ions are detected, the signal isprocessed through the analog cir-cuitry. When fewer than the thresholdnumber of ions are detected, the signalcascades through the rest of the dyn-odes and is measured as a pulse signalin the conventional way. This process,

which is shown in Figure 3, is com-pletely automatic and means that boththe analog and the pulse signals are col-lected simultaneously in one scan (6).

The pulse-counting mode is typicallylinear from zero to about 106 counts/s,while the analog circuitry is suitablefrom 104 to 109 counts/s. To normalizeboth ranges, a cross calibration is per-formed to cover concentration levels,which could generate a pulse and ananalog signal. This is possible becausethe analog and pulse outputs can be de-fined in identical terms of incomingpulse counts per second, based onknowing the voltage at the first analogstage, the output current, and a conver-sion factor defined by the detection cir-cuitry electronics. By performing across calibration across the mass range,a dual-mode detector of this type is ca-pable of achieving approximately eightto nine orders of dynamic range in onesimultaneous scan. Figure 4 shows thepulse-counting calibration curve (yel-low) is linear up to 106 cps, and the ana-log calibration curve (blue) is linearfrom 104 to 109 cps. Figure 5 shows thatafter cross calibration, the two curvesare normalized, which means the detec-tor is suitable for concentration levelsbetween 0.1 ppt and 100 ppm — typi-cally eight to nine orders of magnitudefor most elements.

There are subtle variations of thistype of detection system, but its majorbenefit is that it requires only one scan

106

00.1 ppt 100 ppm

Analyte concentration

Pulse

(cps

)

109

104

Anal

og (c

ps)

Circle 20

Figure 5. Using cross calibration of the pulse and analog modes, quantitation from sub-part-per-trillion to high parts-per-million levels is possible.

April 2002 17(4) Spectroscopy 39

Tutorial

109

00.1 ppt 100 ppm

Analyte concentration

Cros

s-ca

libra

ted

coun

ts (c

ps)

Circle 21 Circle 22

to determine both high and low con-centrations. Therefore, it not only offersthe potential to improve samplethroughput, it also means that the max-imum data can be collected on a tran-sient signal that only lasts a few sec-onds. This process will be described ingreater detail in the next installment ofthis series, in which I will discuss differ-ent measurement protocols and peakintegration routines.

References1. Inductively Coupled Plasma Mass

Spectrometry, Ed. A. Montasser (Wiley-VCH, Berlin, 1998).

2. K. Hunter, Atomic Spectroscopy 15(1),17–20 (1994).

3. R.C. Hutton, A.N. Eaton, and R.M.Gosland, Applied Spectroscopy 44(2),238–242 (1990).

4. Y. Kishi, Agilent Technologies Applica-tion Journal (August 1997).

5. E.R. Denoyer, R.J. Thomas, and L.Cousins, Spectroscopy 12(2), 56–61,(1997).

6. Covered by U.S. Patent No. 5,463,219. �

odern ICP-MS mustbe very flexible tomeet such diverseapplication needs

and keep up with the increasingdemands of its users. Nowhereis this more important than inthe area of peak integration andmeasurement protocol. The waythe analytical signal is managedin ICP-MS directly impacts itsmultielement capability, detec-tion limits, dynamic range, andsample throughput — the fourmajor strengths that attractedthe trace element community tothe technique almost 20 yearsago. To understand signal man-agement and its implications ondata quality in greater detail,this installment of this serieswill discuss how measurementprotocol is optimized based onthe application’s analytical re-quirements. I will discuss itsimpact on both continuous sig-nals generated by traditionalnebulization devices and tran-sient signals produced by alter-native sample introduction

M

www.spectroscopyonl ine.com28 Spectroscopy 17(7) July 2002

• the detection limits required• the precision and accuracy

expected• the dynamic range needed• the integration time used• the peak quantitation

routines.Before discussing these fac-

tors in greater detail, and howthey affect data quality, it is im-portant to remember how ascanning device such as aquadrupole mass analyzerworks (1). Although we willfocus on quadrupole technol-ogy, the fundamental principlesof measurement protocol will bevery similar for all types of massspectrometers that use a scan-ning approach for multielementpeak quantitation.

Measurement Protocol Figure 1 shows the principles ofscanning with a quadrupolemass analyzer. In this simplifiedexample, the analyte ion (black)and four other ions (colored)have arrived at the entrance tothe four rods of the quadrupole.When a particular rf/dc voltageis applied to the rods, the posi-tive or negative bias on the rodswill electrostatically steer theanalyte ion of interest down themiddle of the four rods to theend, where it will emerge and beconverted to an electrical pulseby the detector. The other ionsof different mass-to-charge ra-tios will pass through the spacesbetween the rods and be ejectedfrom the quadrupole. This scan-ning process is then repeated foranother analyte at a completelydifferent mass-to-charge ratio

techniques such as flow injec-tion and laser ablation.

Measurement VariablesMany variables affect the qualityof the analytical signal in ICP-MS. The analytical requirementsof the application will often dic-tate this factor, but instrumentaldetection and measurement pa-rameters can have a significantimpact on the quality of data inICP-MS. Some of the variablesthat can affect the quality ofyour data, particularly when car-rying out multielement analysis,include • whether the signal is continu-

ous or transient• the temporal length of the

sampling event• the volume of sample

available• the number of samples being

analyzed• the number of replicates per

sample• the number of elements being

determined

With its multielement capability, superb detection limits, wide dynamicrange, and high sample throughput, inductively coupled plasma–massspectrometry (ICP-MS) is proving to be a compelling technique for moreand more diverse application areas. However, no two application areashave the same analytical requirements. For example, environmentaland clinical contract laboratories — although requiring reasonably lowdetection limits — do not really push the technique to its extremedetection capability. Their main requirement is usually high samplethroughput because the number of samples these laboratories cananalyze in a day directly impacts their revenue. On the other hand, asemiconductor fabrication plant or a supplier of high-purity chemicalsto the electronics industry is interested in the lowest detection limits thetechnique can offer because of the contamination problems associatedwith manufacturing high performance electronic devices.


A Beginner‘s Guide to ICP-MSPart XI — Peak Measurement ProtocolRobert Thomas

M


July 2002 17(7) Spectroscopy 29

Tutorial

on the masses of interest. Full peak pro-filing is not normally used for doingrapid quantitative analysis becausevaluable analytical time is wasted takingdata on the wings and valleys of thepeak, where the signal-to-noise ratio ispoorest.

When the best possible detectionlimits are required, the peak-hoppingapproach is best. It is important to un-derstand that, to get the full benefit ofpeak hopping, the best detection limitsare achieved when single-point peakhopping at the peak maximum is cho-sen. However, to carry out single-pointpeak hopping, it is essential that the

mass stability is good enough to repro-ducibly go to the same mass point everytime. If good mass stability can be guar-anteed (usually by thermostating thequadrupole power supply), measuringthe signal at the peak maximum will al-ways give the best detection limits for agiven integration time. It is well docu-mented that there is no benefit tospreading the chosen integration timeover more than one measurement pointper mass. If time is a major considera-tion in the analysis, then using multiplepoints is wasting valuable time on thewings and valleys of the peak, whichcontribute less to the analytical signal

Figure 2. Detection and measurement protocol using a quadrupole mass analyzer.

Figure 1. Principles of mass selection with a quadrupole mass filter.

until all the analytes in a multielementanalysis have been measured.

The process for detecting one partic-ular mass in a multielement run is rep-resented in Figure 2, which shows a 63Cuion emerging from the quadrupole andbeing converted to an electrical pulse bythe detector. As the rf/dc voltage of thequadrupole — corresponding to 63Cu— is repeatedly scanned, the ions aselectrical pulses are stored and countedby a multichannel analyzer. This multi-channel data-acquisition system typi-cally has 20 channels per mass and asthe electrical pulses are counted in eachchannel, a profile of the mass is built-up over the 20 channels, correspondingto the spectral peak of 63Cu. In a multi-element run, repeated scans are madeover the entire suite of analyte masses,as opposed to just one mass representedin this example.

The principles of multielement peakacquisition are shown in Figure 3. Inthis example (showing two masses), sig-nal pulses are continually collected asthe quadrupole is swept across the massspectrum (in this case three times).After a given number sweeps, the totalnumber of signal pulses in each channelare counted.

When it comes to quantifying an iso-topic signal in ICP-MS, there are basi-cally two approaches to consider (2).One is the multichannel ramp scanningapproach, which uses a continuoussmooth ramp of 1 to n channels (wheren is typically 20) per mass across thepeak profile. This approach is shown inFigure 4.

The peak-hopping approach is wherethe quadrupole power supply is drivento a discrete position on the peak (nor-mally the peak point) and allowed tosettle; a measurement is then taken for afixed amount of time. This approach isrepresented in Figure 5.

The multipoint scanning approach isbest for accumulating spectral and peakshape information when doing massscans. It is normally used for doingmass calibration and resolution checks,and as a classical qualitative method de-velopment tool to find out what ele-ments are present in the sample, as wellas to assess their spectral implications


and more to the background noise. Fig-ure 6 shows the degradation in signal-to-background noise ratio of 10 ppb Rhwith an increase in the number of pointsper peak, spread over the same total inte-gration time. Detection limit improve-ment for a selected group of elementsusing 1 point/peak, rather than 20points/peak, is shown in Figure 7.

Optimization of MeasurementProtocolNow that the fundamentals of thequadrupole measuring electronics havebeen described, let us now go into moredetail on the impact of optimizing themeasurement protocol based on the re-

quirement of the application. Whenmultielement analysis is being carriedout by ICP-MS, a number of decisionsneed to be made. First, we need to knowif we are dealing with a continuous sig-nal from a nebulizer or a transient sig-nal from an alternative sampling acces-sory. If it is a transient event, how longwill the signal last? Another questionthat needs to be addressed is, how manyelements are going to be determined?With a continuous signal, this isn’t sucha major problem, but it could be anissue if we are dealing with a transientsignal that lasts a few seconds. We alsoneed to be aware of the level of detec-tion capability required. This is a major

consideration with a single-shot laserpulse that lasts 5–10 s. Also with a con-tinuous signal produced by a concentricnebulizer, we might have to accept acompromise of detection limit based onthe speed of analysis requirements oramount of sample available. What ana-lytical precision is expected? If it’s iso-tope ratio/dilution work, how many ions do wehave to count to guarantee good preci-sion? Does increasing the integrationtime of the measurement help the pre-cision? Finally, is there a time constrainton the analysis? A high-throughput lab-oratory might not be able to afford touse the optimum sampling time to getthe ultimate in detection limit. In otherwords, what compromises need to bemade between detection limit, preci-sion, and sample throughput? Clearly,before the measurement protocol canbe optimized, the major analytical re-quirements of the application need tobe defined. Let’s take a look at thisprocess in greater detail.

Multielement Data QualityObjectivesBecause multielement detection capa-bility is probably the major reason why

most laboratories invest in ICP-MS, it isimportant to understand the impact ofmeasurement criteria on detection lim-its. We know that in a multielementanalysis, the quadrupole’s rf/dc ratio isscanned to mass regions or driven,which represent the elements of inter-est. The electronics are allowed to settle

Figure 3 (above left). A profile of the peak is built up by continually sweeping the quadrupoleacross the mass spectrum.

Figure 4 (above right). Multichannel ramp scanning approach using 20 channels per amu.

Table I. Precision of Pb isotope ratio measurement as a function of dwell time using a total integration time of 5.5 s.Dwelltime %RSD, %RSD,(ms) 207Pb/206Pb 208Pb/206Pb

2 0.40 0.365 0.38 0.3610 0.23 0.2225 0.24 0.2550 0.38 0.33100 0.41 0.38

Figure 5 (below right). Peak-hopping approach.

Tutorial


Tutorial

and then dwell on the peak, or sit, andtake measurements for a fixed period oftime. This step is usually performed anumber of times until the total integra-tion time is fulfilled. For example, if adwell time of 50 ms is selected for allmasses and the total integration time is1 s, then the quadrupole will carry out20 complete sweeps per mass, per repli-cate. It will then repeat the same rou-tine for as many replicates that havebeen built into the method. Thisprocess is illustrated very simplisticallyin Figure 8, which shows the scanningprotocol of a multielement scan ofthree different masses.

In this example, the quadrupole isscanned to mass A. The electronics areallowed to settle (settling time) and leftto dwell for a fixed period of time atone or multiple points on the peak(dwell time); intensity measurementsare then taken (based on the dwelltime). The quadrupole is then scannedto masses B and C and the measure-ment protocol repeated. The completemultielement measurement cycle(sweep) is repeated as many times asneeded to make up the total integrationper peak. It should be emphasized thatthis example is a generalization of themeasurement routine — managementof peak integration by the software will vary slightly, based on different instrumentation.

It is clear from this information that,during a multielement analysis, a sig-

nificant amount of time is spent scan-ning and settling the quadrupole, whichdoesn’t contribute to the quality of theanalytical signal. Therefore, if the meas-urement routine is not optimized care-fully, it can have a negative impact ondata quality. The dwell time can usuallybe selected on an individual mass basis,but the scanning and settling times arenormally fixed because they are a func-tion of the quadrupole and detectorelectronics. For this reason, it is essen-tial that the dwell time — which ulti-mately affects detection limit and preci-sion — must dominate the totalmeasurement time, compared with thescanning and settling times. It follows,therefore, that the measurement dutycycle (percentage of actual measuringtime compared with total integrationtime) is maximized when the quadru-pole and detector electronics settlingtimes are kept to an absolute minimum.Figure 9 shows a plot of percentage ofmeasurement efficiency against dwelltime for four different quadrupole set-tling times — 0.2, 1.0, 3.0, and 5.0 msfor one replicate of a multielement scanof five masses, using one point perpeak. In this example, the total integra-tion time for each mass was 1 s, withthe number of sweeps varying, depend-ing on the dwell time used. For this ex-ercise, the percentage of measurementefficiency is defined by the followingcalculation:

So to achieve the highest measure-ment efficiency, the nonanalytical timemust be kept to an absolute minimum.This leads to more time being spentcounting ions and less time scanningand settling, which does not contributeto the quality of the analytical signal.This factor becomes critically impor-tant when a rapid transient peak isbeing quantified, because the availablemeasuring time is that much shorter(3). Generally speaking, peak quantita-tion using multiple points per peak andlong settling times should be avoided inICP-MS because it ultimately degradesthe quality of the data for a given inte-gration time.

Figure 9 also shows that shorter dwelltimes translate into a lower measure-ment efficiency. For this reason, it isprobably desirable, for normal quanti-tative analysis work, to carry out multi-ple sweeps with longer dwell times(typically 50 ms) to get the best detec-tion limits. So if an integration time of1 s is used for each element, this wouldtranslate into 20 sweeps of 50 ms dwelltime per mass. Although 1 s is longenough to achieve reasonably good de-tection limits, longer integration timesgenerally have to be used to reach the

Figure 6. Signal-to-background noise ratio degrades when more thanone point, spread over the same integration time, is used for peakquantitation.

Figure 7. Detection limit improvement using 1 point/peak rather than 20points/peak over the mass range.


Tutorial

lowest possible detection limits. Figure10 shows detection limit improvementas a function of integration time for238U. As would be expected, there is afairly predictable improvement in thedetection limit as the integration time isincreased because more ions are beingcounted without an increase in thebackground noise. However, this onlyholds true up to the point where thepulse-counting detection system be-comes saturated and no more ions canbe counted. In the case of 238U, this oc-curs around 25 s, because there is noobvious improvement in detection limitat a higher integration time. So fromthese data, we can say that there appearsto be no real benefit in using an inte-gration time longer than 7 s. When de-ciding the length of the integrationtime in ICP-MS, you have to weigh thedetection limit improvement againstthe time taken to achieve that improve-ment. Is it worth spending 25 s measur-ing each mass to get a 0.02 ppt detec-tion limit if 0.03 ppt can be achievedusing a 7-s integration time? Alterna-tively, is it worth measuring for 7 swhen 1 s will only degrade the perform-ance by a factor of 3? It really dependson your data quality objectives.

For applications such as isotopedilution/ratio studies, high precision isalso a very important data quality ob-jective (4). However, to understandwhat is realistically achievable, we mustbe aware of the practical limitations ofmeasuring a signal and counting ions inICP-MS. Counting statistics tells us thatthe standard deviation of the ion signalis proportional to the square root of thesignal. It follows, therefore, that the rel-ative standard deviation (RSD), or pre-cision, should improve with an increasein the number (N) of ions counted asshown by the following equation:

In practice this holds up very well, asshown in Figure 11. In this plot of stan-dard deviation as a function of signalintensity for 208Pb, the dots represent thetheoretical relationship as predicted bycounting statistics. It can be seen thatthe measured standard deviation (bars)follows theory very well up to about100,000 cps. At that point, additionalsources of noise (for example, sample

introduction pulsations or plasma fluc-tuations) dominate the signal, whichleads to poorer standard deviation values.

So based on counting statistics, it islogical to assume that the more ionsthat are counted, the better the preci-sion will be. To put this in perspective,it means that at least 1 million ionsneed to be counted to achieve an RSDof 0.1%. In practice, of course, thesekinds of precision values are very diffi-cult to achieve with a scanning quadru-

Table II. Impact of integration time on the overall analysis time for Pb isotope ratios.Dwell time Number of Integration time %RSD, %RSD, Analysis time

(ms) sweeps (s)/mass 207Pb/206Pb 207Pb/206Pb for 9 reps25 220 5.5 0.24 0.25 2 min 29 s25 500 12.5 0.21 0.19 6 min 12 s25 700 17.5 0.20 0.17 8 min 29 s

Figure 8 (left).Multielement scanning andmeasurement protocol of aquadrupole.

Figure 9 (below). Percentof measurement efficiencyas a function of dwell timewith varying scanning/settling times.


Tutorial

pole system because of the additionalsources of noise. If this information iscombined with our knowledge of howthe quadrupole is scanned, we begin tounderstand what is required to get thebest precision. This is confirmed by thespectral scan in Figure 12, which showsthe predicted precision at all 20 chan-nels of a 5 ppb 208Pb peak (2).

Therefore, the best precision is ob-tained at the channels where the signalis highest, which as we can see are the

ones at or near the center of the peak.For this reason, if good precision is afundamental requirement of your dataquality objectives, it is best to usesingle-point peak hopping with integra-tion times in the order of 5–10 s. Onthe other hand, if high-precision iso-tope ratio or isotope dilution work isbeing done — in which analysts wouldlike to achieve precision values ap-proaching counting statistics — thenmuch longer measuring times are re-

quired. That is why integration times of5–10 min are commonly used for deter-mining isotope ratios with a quadru-pole ICP-MS system (5, 6). For this typeof analysis, when two or more isotopesare being measured and ratioed to eachother, it follows that the more simulta-neous the measurement, the better theprecision becomes. Therefore, the abil-ity to make the measurement as simul-taneous as possible is considered moredesirable than any other aspect of themeasurement. This is supported by thefact that the best isotope ratio precisiondata are obtained with time-of-flight ormulticollector, magnetic sector ICP-MSsystems, which are both considered si-multaneous in nature. So the best wayto approximate simultaneous measure-ment with a rapid scanning device, suchas a quadrupole, is to use shorter dwelland scanning/settling times, resulting inmore sweeps for a given integrationtime. Table I shows precision of Pb iso-tope ratios at different dwell times car-ried out by researchers at the GeologicalSurvey of Israel (7). The data are basedon nine replicates of a NIST SRM-981(75 ppb Pb) solution, using 5.5 s of in-tegration time per isotope.

From these data, the researchers con-cluded that a dwell time of 10 or 25 msoffered the best isotope ratio precisionmeasurement (quadrupole settling timewas fixed at 0.2 ms). They also foundthat they could achieve slightly betterprecision by using a 17.5-s integrationtime (700 sweeps at a 25-ms dwelltime), but felt the marginal improve-ment in precision for nine replicateswas not worth spending the approxi-mately 3.5-times-longer analysis time,as shown in Table II.

This work shows the benefit of opti-mizing the dwell time, settling time,and the number of sweeps to get thebest isotope ratio precision data. Theresearchers were also very fortunate tobe dealing with relatively healthy signalsfor the three Pb isotopes, 206Pb, 207Pb,and 208Pb (24.1%, 22.1%, and 52.4%abundance, respectively). If the isotopicsignals were dramatically different likein 235U to 238U (0.72 % and 99.2745%abundance, respectively), then the abil-ity to optimize the measurement proto-

Figure 10. Plot of detection limit against integration time for 238U.

Figure 11. Comparison of measured standard deviation of a 208Pb signal against that predicted bycounting statistics.


Conference Preview

col for individual isotopes becomes ofeven greater importance to guaranteeprecise data.

It is clear that the analytical demandsput on ICP-MS are probably higherthan any other trace element techniquebecause it is continually being asked tosolve a wide variety of applicationproblems. However, by optimizing themeasurement protocol to fit the analyt-ical requirement, ICP-MS has shownthat it has the capability to carry outrapid trace element analysis, with su-perb detection limits and good preci-sion on both continuous and transientsignals, and still meet the most strin-gent data quality objectives.

References1. R. Thomas, Spectroscopy 16(10),

44–48 (2001).2. E.R. Denoyer, At. Spectroscopy 13(3),

93–98 (1992).3. E.R. Denoyer and Q.H. Lu, At. Spec-

troscopy 14(6), 162–169 (1993).4. T. Catterick, H. Handley, and S. Merson,

At. Spectroscopy 16(10), 229–234(1995).

5. T.A. Hinners, E.M. Heithmar, T.M. Spit-tler, and J.M. Henshaw, Anal. Chem. 59,2658–2662 (1987).

6. M. Janghorbani, B.T.G. Ting, and N.E.Lynch, Microchemica Acta 3, 315–328,(1989).

7. L. Halicz, Y. Erel, and A. Veron, At. Spec-troscopy 17(5), 186–189 (1996). ■

applications and provides everythingnecessary for a typical x-ray detector,incorporating one DXP spectrometerchannel, preamplifier power, and detec-tor HV bias in one compact chassis. Itsinput is compatible with a wide rangeof common detectors, including pulsedoptical reset, transistor reset, and RCfeedback preamplifiers. The Saturn of-fers complete computer control over allamplifier and spectrometer functionsincluding gain, filter peaking time, andpileup inspection criteria. Its DXP digi-tal filters significantly increase through-put compared to typical analog systems.

The new X-Beam x-ray source fromX-Ray Optical Systems (Albany, NY)delivers an intense, micrometer-sizedfocal spot. Designed for OEM use inmicro-XRF instruments, the compactunit uses polycapillary focusing optics

Figure 12. Comparison of % RSD with signalintensity across the mass profile of a 208Pb peak.

and 50 W of power to generate an ex-tremely high flux-density gain, the com-pany reports. Increased spatial resolutionand beam stability are also promised. Anintegrated cooling system eliminates theneed for a separate cooling unit.

Attendees can see many of theseproducts, along with others not men-tioned, at the 2002 Denver X-ray Con-ference — sponsored by the Interna-tional Centre for Diffraction Data — atAntlers Adam’s Mark Hotel (formerlyAntlers Doubletree Hotel), ColoradoSprings, Colorado, July 29–August 2,2002. For more information, contactDenise Flaherty, DXC Conference Co-ordinator, 12 Campus Boulevard, New-town Square, PA 19073-3273, (610)325-9814, fax: (610) 325-9823,e-mail: [email protected], web site:www.dxcicdd.com. ■

Circle 25

“Conference Preview” continued frompage 27

We have previously covered themajor instrumental componentsof an ICP mass spectrometer; nowlet’s turn our attention to thetechnique’s most common inter-ferences and what methods areused to compensate for them. Al-though interferences are reason-ably well understood in induc-tively coupled plasma–massspectrometry (ICP-MS), it canoften be difficult and time-consuming to compensate forthem, particularly in complexsample matrices. Having priorknowledge of the interferences as-sociated with a particular set ofsamples will often dictate thesample preparation steps and theinstrumental methodology used toanalyze them.

nterferences in ICP-MS aregenerally classified intothree major groups — spec-tral, matrix, and physical.

Each of them has the potential tobe problematic in its own right,but modern instrumentation andgood software, combined withoptimized analytical methodolo-gies, has minimized their negativeimpact on trace element determi-nations by ICP-MS. Let us take alook at these interferences ingreater detail and describe thedifferent approaches used tocompensate for them.

Spectral InterferencesSpectral overlaps are probablythe most serious types of inter-

I

www.spectroscopyonl ine.com24 Spectroscopy 17(10) October 2002

40, whereas the combination ofargon and oxygen in an aqueoussample generates the 40Ar16O in-terference, which has a signifi-cant impact on the major iso-tope of Fe at mass 56. Thecomplexity of these kinds ofspectral problems can be seen inFigure 1, which shows a massspectrum of deionized waterfrom mass 40 to mass 90.

Additionally, argon can alsoform polyatomic interferenceswith elements found in the acidsused to dissolve the sample. Forexample in a hydrochloric acidmedium, 40Ar combines with themost abundant chlorine isotopeat 35 amu to form 40Ar35Cl,which interferes with the onlyisotope of arsenic at mass 75,while in an organic solvent ma-

ferences seen in ICP-MS. Themost common type is known asa polyatomic or molecular spec-tral interference, which is pro-duced by the combination oftwo or more atomic ions. Theyare caused by a variety of factors,but are usually associated witheither the plasma and nebulizergas used, matrix components inthe solvent and sample, otheranalyte elements, or entrainedoxygen or nitrogen from the sur-rounding air. For example, in theargon plasma, spectral overlapscaused by argon ions and combi-nations of argon ions with otherspecies are very common. Themost abundant isotope of argonis at mass 40, which dramaticallyinterferes with the most abun-dant isotope of calcium at mass


A Beginner’s Guide to ICP-MSPart XII — A Review of Interferences

Robert Thomas

I


Figure 1. Mass spectrum of deionized water from mass 40 to mass 85.

3

2

1

40 50 60 70 80 90

Mass (amu)

Inte

nsity

(cp

s �

105 )

40Ar40ArH

12C16O16O14N16O16O

40Ar16O

40Ar12C

CO2H

40Ar16O

40Ar18O

40Ar40Ar

40Ar40ArH

40Ar38Ar

40Ar36Ar

40Ar14N

October 2002 17(10) Spectroscopy 25

Tutorial

Figure 2. Relative isotopic abundances of the naturally occurring elements, showing all the potential isobaric interferences.

1 H 99.985 2 H 0.015 3 He 0.000137 4 He 99.999863 5 6 Li 7.5 7 Li 92.5 8 9 Be 10010 B 19.911 B 80.112 C 98.9013 C 1.1014 N 99.64315 N 0.36616 O 99.76217 O 0.03818 O 0.20019 F 10020 Ne 90.4821 Ne 0.2722 Ne 9.2523 Na 10024 Mg 78.9925 Mg 10.0026 Mg 11.0127 Al 10028 Si 92.2329 Si 4.6730 Si 3.1031 P 10032 S 95.0233 S 0.7534 S 4.2135 Cl 75.7736 S 0.02 Ar 0.33737 Cl 24.2338 Ar 0.06339 K 93.258140 K 0.0117 Ca 96.941 Ar 99.60041 K 6.730242 Ca 0.64743 Ca 0.13544 Ca 2.08645 Sc 10046 Ti 8.0 Ca 0.00447 Ti 7.348 Ti 73.8 Ca 0.18749 Ti 5.550 Ti 5.4 V 0.250 Cr 4.34551 V 99.75052 Cr 83.78953 Cr 9.50154 Fe 5.8 Cr 2.36555 Mn 10056 Fe 91.7257 Fe 2.258 Fe 0.28 Ni 68.07759 Co 10060 Ni 26.223

61 Ni 1.140 62 Ni 3.634 63 Cu 69.17 64 Zn 48.6 Ni 0.926 65 Cu 30.83 66 Zn 27.9 67 Zn 4.1 68 Zn 18.8 69 Ga 60.108 70 Ge 21.23 Zn 0.6 71 Ga 39.892 72 Ge 27.66 73 Ge 7.73 74 Ge 35.94 Se 0.89 75 As 100 76 Ge 7.44 Se 9.36 77 Se 7.63 78 Kr 0.35 Se 23.78 79 Br 50.69 80 Kr 2.25 Se 49.61 81 Br 49.31 82 Kr 11.6 Se 8.73 83 Kr 11.5 84 Kr 57.0 Sr 0.56 85 Rb 72.165 86 Kr 17.3 Sr 9.86 87 Sr 7.00 Rb 27.835 88 Sr 82.58 89 Y 100 90 Zr 51.45 91 Zr 11.22 92 Zr 17.15 Mo 14.84 93 Nb 100 94 Zr 17.38 Mo 9.25 95 Mo 15.92 96 Zr 2.80 Mo 16.68 Ru 5.52 97 Mo 9.55 98 Mo 24.13 Ru 1.88 99 Ru 12.7100 Mo 9.63 Ru 12.6101 Ru 17.0102 Pd 1.02 Ru 31.6103 Rh 100104 Pd 11.14 Ru 18.7105 Pd 22.33106 Pd 27.33 Cd 1.25107 Ag 51.839108 Pd 26.46 Cd 0.89109 Ag 48.161110 Pd 11.72 Cd 12.49111 Cd 12.80112 Sn 0.97 Cd 24.13113 Cd 12.22 In 4.3114 Sn 0.65 Cd 28.73115 Sn 0.34 In 95.7116 Sn 14.53 Cd 7.49117 Sn 7.68118 Sn 24.23119 Sn 8.59120 Sn 32.59 Te 0.096

121 Sb 57.36122 Sn 4.63 Te 2.603 123 Te 0.908 Sb 42.64124 Sn 5.79 Te 4.816 Xe 0.10125 Te 7.139126 Te 18.95 Xe 0.09127 I 100128 Te 31.69 Xe 1.91129 Xe 26.4130 Ba 0.106 Te 33.80 Xe 4.1131 Xe 21.2132 Ba 0.101 Xe 26.9133 Ce 100134 Ba 2.417 Xe 10.4135 Ba 6.592136 Ba 7.854 Ce 0.19 Xe 8.9137 Ba 11.23138 Ba 71.70 Ce 0.25 La 0.0902139 La 99.9098140 Ce 88.48141 Pr 100142 Nd 27.13 Ce 11.08143 Nd 12.18144 Nd 23.80 Sm 3.1145 Nd 8.30146 Nd 17.19147 Sm 15.0148 Nd 5.76 Sm 11.3149 Sm 13.8150 Nd 5.64 Sm 7.4151 Eu 47.8152 Gd 0.20 Sm 26.7153 Eu 52.2154 Gd 2.18 Sm 22.7155 Gd 14.80156 Gd 20.47 Dy 0.06157 Gd 15.65158 Gd 24.84 Dy 0.10159 Tb 100160 Gd 21.86 Dy 2.34161 Dy 18.9162 Er 0.14 Dy 25.5163 Dy 24.9164 Er 1.61 Dy 28.2165 Ho 100166 Er 33.6167 Er 22.95168 Er 26.8 Yb 0.13169 Tm 100170 Er 14.9 Yb 3.05171 Yb 14.3172 Yb 21.9173 Yb 16.12174 Yb 31.8 Hf 0.162175 Lu 97.41176 Lu 2.59 Yb 12.7 Hf 5.206177 Hf 18.606178 Hf 27.297179 Hf 13.629180 Ta 0.012 W 0.13 Hf 35.100

181 Ta 99.988182 W 26.3 183 W 14.3184 Os 0.02 W 30.67185 Re 37.40186 Os 1.58 W 28.6187 Os 1.6 Re 62.60188 Os 13.3189 Os 16.1190 Os 26.4 Pt 0.01191 Ir 37.3192 Os 41.0 Pt 0.79193 Ir 62.7194 Pt 32.9195 Pt 33.8196 Hg 0.15 Pt 25.3197 Au 100198 Hg 9.97 Pt 7.2199 Hg 16.87200 Hg 23.10201 Hg 13.18202 Hg 29.86203 Tl 29.524204 Hg 6.87 Pb 1.4205 Tl 70.476206 Pb 24.1207 Pb 22.1208 Pb 52.4209 Bi 100210211212213214215216217218219220221222223224225226227228229230231 Pa 100232 Th 100233234 U 0.0055235 U 0.7200236237238 U 99.2745

Isotope Isotope Isotope Isotope% % % % % % % % % % % %

Relative Abundance of the Natural Isotopes

Oxides, Hydroxides, Hydrides, andDoubly Charged SpeciesAnother type of spectral interference isproduced by elements in the samplecombining with H, 16O, or 16OH (eitherfrom water or air) to form molecularhydride (H), oxide (16O), and hydroxide(16OH) ions, which occur at 1, 16, and17 mass units higher than its mass (2).These interferences are typically pro-duced in the cooler zones of the plasma,immediately before the interface region.They are usually more serious whenrare earth or refractory-type elementsare present in the sample, because manyof them readily form molecular species(particularly oxides), which create spec-tral overlap problems on other elementsin the same group. Associated withoxide-based spectral overlaps are dou-bly charged spectral interferences.These are species that are formed when

an ion is generated with a double posi-tive charge, as opposed to a normal sin-gle charge, and produces a peak at halfits mass. Like the formation of oxides,the level of doubly charged species is re-lated to the ionization conditions in theplasma and can usually be minimizedby careful optimization of the nebulizergas flow, rf power, and sampling posi-tion within the plasma. It can also beimpacted by the severity of the second-ary discharge present at the interface(3), which was described in greater de-tail in Part IV of the series (4). Table IIshows a selected group of elements, thatreadily form oxides, hydroxides, hy-drides, and doubly charged species, to-gether with the analytes that are af-fected by them.

trix, argon and carbon combine to form40Ar12C, which interferes with 52Cr, themost abundant isotope of chromium.Sometimes, matrix or solvent speciesneed no help from argon ions and com-bine to form spectral interferences oftheir own. A good example is in a sam-ple that contains sulfuric acid. Thedominant sulfur isotope 32S combineswith two oxygen ions to form a32S16O16O molecular ion, which inter-feres with the major isotope of Zn atmass 64. In the analysis of samples con-taining high concentrations of sodium,such as seawater, the most abundantisotope of Cu at mass 63 cannot be usedbecause of interference from the40Ar23Na molecular ion. There are manymore examples of these kinds of poly-atomic and molecular interferences (1).Table I represents some of the mostcommon ones seen in ICP-MS.


Tutorial

Isobaric InterferencesThe final classification of spectral inter-ferences is called “isobaric overlaps,”produced mainly by different isotopesof other elements in the sample thatcreate spectral interferences at the samemass as the analyte. For example, vana-dium has two isotopes at 50 and 51amu. However, mass 50 is the onlypractical isotope to use in the presence

of a chloride matrix, because of thelarge contribution from the 16O35Cl in-terference at mass 51. Unfortunatelymass 50 amu, which is only 0.25%abundant, also coincides with isotopesof titanium and chromium, which are5.4% and 4.3% abundant, respectively.This makes the determination of vana-dium in the presence of titanium andchromium very difficult unless mathe-matical corrections are made. Figure 2— the relative abundance of the natu-rally occurring isotopes — shows all thenaturally occuring isobaric spectraloverlaps possible in ICP-MS (5).

Ways to Compensate for SpectralInterferencesLet us look at the different approachesused to compensate for spectral interfer-ences. One of the very first ways used toget around severe matrix-derived spectralinterferences was to remove the matrixsomehow. In the early days, this involvedprecipitating the matrix with a complex-ing agent and then filtering off the pre-cipitate. However, this has been more re-cently carried out by automated matrixremoval and analyte preconcentrationtechniques using chromatography-typeequipment. In fact, this method is pre-ferred for carrying out trace metal deter-minations in seawater because of the ma-trix and spectral problems associatedwith such high concentrations of sodiumand magnesium chloride (6).

Mathematical Correction EquationsAnother method that has been success-fully used to compensate for isobaricinterferences and some less severe poly-atomic overlaps (when no alternativeisotopes are available for quantitation)is to use mathematical interference cor-rection equations. Similar to inter-element corrections (IECs) in ICP–optical emission spectroscopy, thismethod works on the principle ofmeasuring the intensity of the interfer-ing isotope or interfering species at an-other mass, which ideally is free of anyinterferences. A correction is then ap-plied by knowing the ratio of the inten-sity of the interfering species at the ana-lyte mass to its intensity at the alternatemass.

Let’s take a look at a real-world ex-ample of this type of correction. Themost sensitive isotope for cadmium isat mass 114. However, there is also aminor isotope of tin at mass 114. Thismeans that if there is any tin in thesample, quantitation using 114Cd canonly be carried out if a correction ismade for 114Sn. Fortunately Sn has atotal of 10 isotopes, which means thatat least one of them will probably befree of a spectral interference. There-fore, by measuring the intensity of Sn atone of its most abundant isotopes (typ-ically 118Sn) and ratioing it to 114Sn, acorrection is made in the method soft-ware in the following manner:

Table I. Some common plasma, matrix, and solvent-related polyatomic spectral interferences seen in ICP-MS.

Element/ Matrix/ Isotope solvent Interference

39K H2O 38ArH40Ca H2O 40Ar56Fe H2O 40Ar16O80Se H2O 40Ar40Ar51V HCl 35Cl16O

75As HCl 40Ar35Ci28Si HNO3

14N14N44Ca HNO3

14N14N16O55Mn HNO3

40Ar15N48Ti H2SO4

32S16O52Cr H2SO4

34S18O64Zn H2SO4

32S16O16O63Cu H3PO4

31P16O16O24Mg Organics 12C12C52Cr Organics 40Ar12C65Cu Minerals 48Ca16OH64Zn Minerals 48Ca16O63Cu Seawater 40Ar23Na

Figure 3. Spectral scan of 100 ppt 56Fe and deionized water using coolplasma conditions.

54 55 56 57 58

Mass (amu)

Ion

sign

al

Blank H2O

100 ppt Fe

Figure 4. Reduction of the 40Ar35Cl interference makes it possible todetermine low ppt levels of monoisotopic 75As in a high chloride matrixusing dynamic reaction cell technology.

30

20

10

073 74 75 76 77

Mass (amu)

1000 ppm NaCl

50 ppt As in 1000 ppm NaCl


Tutorial

Total counts at mass 114 � 114Cd � 114Sn

Therefore 114Cd � total counts at mass114 � 114Sn

To find out the contribution from 114Sn,it is measured at the interference-freeisotope of 118Sn and a correction of theratio of 114Sn/118Sn is applied:

Which means 114Cd � counts at mass114 � (114Sn/118Sn) � (118Sn)

Now the ratio (114Sn/118Sn) is the ratio ofthe natural abundances of these twoisotopes (0.65%/24.23%) and is alwaysconstant

Therefore 114Cd � mass 114 �(0.65%/24.23%) � (118Sn)

or 114Cd � mass 114 � (0.0268) �(118Sn)

An interference correction for 114Cdwould then be entered in the software as:

�(0.0268)*(118Sn).

This is a relatively simple example,but explains the basic principles of theprocess. In practice, especially in spec-trally complex samples, correctionsoften have to be made to the isotopebeing used for the correction — thesecorrections are in addition to the ana-lyte mass, which makes the mathemati-cal equation far more complex.

This approach can also be used forsome less severe polyatomic-type spec-tral interferences. For example, in thedetermination of V at mass 51 in di-luted brine (typically 1000 ppm NaCl),there is a substantial spectral interfer-ence from 35Cl16O at mass 51. By meas-uring the intensity of the 37Cl16O at mass53, which is free of any interference, acorrection can be applied in a similarway to the previous example.

Cool/Cold Plasma TechnologyIf the intensity of the interference islarge, and analyte intensity is extremely

Circle 17

Table II. Some elements that readily form oxides, hydroxides,or hydrides and doubly charged species in the plasma and the analyte affected by the potential interference.

Oxide/hydroxide/hydride doubly charged species Analyte

40Ca16O 56Fe48Ti16O 64Zn

98Mo16O 114Cd138Ba16O 154Sm, 154Gd139La16O 155Gd140Ce16O 156Gd, 156Dy

40Ca16OH 57Fe31P18O16OH 66Zn

79BrH 80Se31P16O2H 64Zn

138Ba2� 69Ga139La2� 69Ga140Ce2� 70Ge, 70Zn


Tutorial

low, mathematical equations are notideally suited as a correction method.For that reason, alternative approacheshave to be considered to compensatefor the interference. One such ap-proach, which has helped to reducesome of the severe polyatomic overlaps,

Circle 18

is to use cold/cool plasma conditions.This technology, which was reported inthe literature in the late 1980s, uses alow-temperature plasma to minimizethe formation of certain argon-basedpolyatomic species (7).

Under normal plasma conditions(typically 1000–1400 W rf power and0.8–1.0 L/min of nebulizer gas flow),argon ions combine with matrix andsolvent components to generate prob-lematic spectral interferences such as38ArH, 40Ar, and 40Ar16O, which impact

Figure 5 (above). Separation of 75As from 40Ar35Cl using the high resolvingpower (10,000) of a double-focusing magnetic sector instrument(Courtesy of Thermo Finnigan).

74.905 74.915 74.925 74.935 74.945

Mass

10 ppb As in 1% HCI40Ar35Cl

75As

Figure 6 (upper right). The transmission characteristics of a magneticsector ICP mass spectrometer decreases as the resolving power increases.

100

80

60

40

20

0400 2000 4000 6000 8000 10,000

Resolution

Tran

smis

sion

(%

)


Tutorial

the detection limits of a small numberof elements including K, Ca, and Fe. Byusing cool plasma conditions (500–800W rf power and 1.5–1.8 L/min nebu-lizer gas flow), the ionization condi-tions in the plasma are changed so thatmany of these interferences are dramat-ically reduced. The result is that detec-tion limits for this group of elementsare significantly enhanced (8).

An example of this improvement isseen in Figure 3, which shows a spectralscan of 100 ppt of 56Fe (its most sensi-tive isotope) using cool plasma condi-tions. It can be clearly seen that there isvirtually no contribution from 40Ar16O,as indicated by the extremely low back-ground for deionized water, resulting insingle-figure parts-per-trillion (ppt) de-tection limits for iron. Under normalplasma conditions, the 40Ar16O intensityis so large that it would completelyoverlap the 56Fe peak.

Cool plasma conditions are limitedto a small group of elements in simpleaqueous solutions that are prone toargon-based spectral interferences. Itoffers very little benefit for the majorityof the other elements, because its ion-ization temperature is significantlylower than a normal plasma. In addi-tion, it is often impractical for theanalysis of complex samples, because ofsevere signal suppression caused by thematrix.

Collision/Reaction CellsThese limitations have led to the devel-opment of collision and reaction cells,

which use ion–molecule collisions andreactions to cleanse the ion beam ofharmful polyatomic and molecular in-terferences before they enter the massanalyzer. Collision/reaction cells areshowing enormous potential to elimi-nate spectral interferences and makeavailable isotopes that were previouslyunavailable for quantitation. For exam-ple, Figure 4 shows a spectral scan of 50ppt As in 1000 ppm NaCl, together with1000 ppm NaCl at mass 75, using a dy-namic reaction cell with hydrogen/argonmixture as the reaction gas. It can beseen that there is insignificant contribu-tion from the 40Ar35Cl interference, as in-dicated by the NaCl baseline. The capa-bility of this type of reaction cell tovirtually eliminate the 40Ar35Cl interfer-ence now makes it possible to determinelow ppt levels of mono-isotopic 75As in ahigh chloride matrix — previously notachievable by conventional interferencecorrection methods (9). For a completereview of the benefits of collision/reac-tion cells for ICP-MS, refer to part 9 ofthis series (10).

High Resolution Mass AnalyzersThe best and probably most efficientway to remove spectral overlaps is to re-solve them away using a high resolutionmass spectrometer (11). During thepast 10 years this approach, particularlywith double-focusing magnetic sectormass analyzers, has proved to be invalu-able for separating many of the prob-lematic polyatomic and molecular in-terferences seen in ICP-MS, without the

need to use cool plasma conditions orcollision/reaction cells. Figure 5 shows10 ppb of 75As resolved from the 40Ar35Clinterference in a 1% hydrochloric acidmatrix, using normal, hot plasma con-ditions and a resolution setting of10,000.

However, even though their resolvingcapability is far more powerful thanquadrupole-based instruments, there is asacrifice in sensitivity if extremely highresolution is used, as shown in Figure 6.This can often translate into a degrada-tion in detection capability for some ele-ments, compared to other spectral inter-ference correction approaches. You willfind an overview of the benefits of mag-netic sector technology for ICP-MS inpart VII of this series (12).

Matrix InterferencesLet’s now take a look at the other classof interference in ICP-MS — suppres-sion of the signal by the matrix itself.There are basically two types of matrix-induced interferences. The first andsimplest to overcome is often called asample transport effect and is a physicalsuppression of the analyte signal,brought on by the matrix components.It is caused by the sample’s impact ondroplet formation in the nebulizer ordroplet-size selection in the spraychamber. In the case of organic matri-ces, it is usually caused by a differencein sample viscosities of the solventsbeing aspirated. In some matrices, sig-nal suppression is caused not so much

Figure 8. The analyte response curve is updated across the full massrange, based on the intensities of low, medium, and high mass internalstandards.

Figure 7. Matrix suppression caused by increasing concentrations ofHNO3, using cool plasma conditions (rf power: 800 W, nebulizer gas: 1.5L/min).

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.00 % 1 % 5 %

Concentration of HNO3

NaAlKCaFeCuZn

0 250

Mass (amu)

Sens

itivi

ty

Updated response curve

Original response curveLow mass

internal standard

Medium massinternal standard

High massinternal standard


Tutorial

by sample transport effects, but by itsimpact on the ionization temperatureof the plasma discharge. This is exem-plified when different concentrations ofacids are aspirated into a cool plasma.The ionization conditions in the plasmaare so fragile that higher concentrationsof acid result in severe suppression ofthe analyte signal. Figure 7 shows thesensitivity for a selected group of ele-ments in varying concentrations of ni-tric acid in a cool plasma (13).

Internal StandardizationThe classic way to compensate for aphysical interference is to use internalstandardization. With this method ofcorrection, a small group of elements(usually at the parts-per-billion level)are spiked into the samples, calibrationstandards, and blank to correct for anyvariations in the response of the ele-ments caused by the matrix. As the in-tensity of the internal standards change,the element responses are updatedevery time a sample is analyzed. Thefollowing criteria are typically used forselecting the internal standards:● They are not present in the sample ● The sample matrix or analyte ele-

ments do not spectrally interferewith them

● They do not spectrally interfere withthe analyte masses

● They should not be elements thatare considered environmentalcontaminants

● They are usually grouped with ana-lyte elements of a similar mass range.For example, a low mass internalstandard is grouped with the lowmass analyte elements and so on upthe mass range

● They should be of a similar ioniza-tion potential to the groups of ana-lyte elements so they behave in a sim-ilar manner in the plasma

● Some of the common ones reportedto be good candidates include 9Be,45Sc, 59Co, 74Ge, 89Y, 103Rh, 115In, 169Tm,175Lu, 187Re, and 232Th.A simplified representation of inter-

nal standardization is seen in Figure 8,which shows updating the analyte re-sponse curve across the full mass range,based on the intensities of low,

medium, and high mass internal stan-dards. It should also be noted that in-ternal standardization is also used tocompensate for long-term signal driftproduced by matrix components slowlyblocking the sampler and skimmer coneorifices. Even though total dissolvedsolids are kept below �0.2% in ICP-MS, this can still produce instability ofthe analyte signal over time with somesample matrices.

Space-Charge InterferencesMany of the early researchers reportedthat the magnitude of signal suppres-sion in ICP-MS increased with decreas-ing atomic mass of the analyte ion (14).More recently it has been suggested thatthe major cause of this kind of suppres-sion is the result of poor transmissionof ions through the ion optics due tomatrix-induced space charge effects(15). This has the effect of defocusingthe ion beam, which leads to poor sen-sitivity and detection limits, especiallywhen trace levels of low mass elementsare being determined in the presence oflarge concentrations of high mass ma-trices. Unless any compensation ismade, the high-mass matrix elementwill dominate the ion beam, pushingthe lighter elements out of the way.Figure 9 shows the classic space chargeeffects of a uranium (major isotope238U) matrix on the determination of7Li, 9Be, 24Mg, 55Mn, 85Rb, 115In, 133Cs, 205Tl,and 208Pb. The suppression of low masselements such as Li and Be is signifi-cantly higher than with high mass ele-ments such as Tl and Pb in the presenceof 1000 ppm uranium.

There are a number of ways to com-pensate for space charge matrix sup-pression in ICP-MS. Internal standardi-zation has been used, but unfortunatelydoesn’t address the fundamental causeof the problem. The most common ap-proach used to alleviate or at least re-duce space charge effects is to applyvoltages to the individual ion lens com-ponents. This is achieved in a numberof ways but, irrespective of the design ofthe ion focusing system, its main func-tion is to reduce matrix-based suppres-sion effects by steering as many of theanalyte ions through to the mass ana-

lyzer while rejecting the maximumnumber of matrix ions. Space charge ef-fects and different designs of ion opticswere described in greater detail in partV of this series (16).

References1. G. Horlick and S.N. Tan, Appl. Spectrosc.

40, 4 (1986).2. G. Horlick and S.N. Tan, Appl. Spectrosc.

40, 4 (1986).3. D.J. Douglas and J.B. French, Spec-

trochim. Acta 41B(3), 197 (1986).4. R. Thomas, Spectroscopy 16(7), 26–34,

(2001).5. ”Isotopic Composition of the Ele-

ments,“ Pure Applied Chemistry 63(7),991–1002, (IUPAC, 1991).

6. S.N. Willie, Y. Iida, and J.W. McLaren,Atom. Spectrosc. 19(3), 67 (1998).

7. S.J. Jiang, R.S. Houk, and M.A. Stevens,Anal. Chem. 60, 1217 (1988).

8. S.D. Tanner, M. Paul, S.A. Beres, andE.R. Denoyer, Atom. Spectrosc. 16(1),16 (1995).

9. K.R. Neubauer and R.E. Wolf, ”Determi-nation of Arsenic in Chloride Matrices,“PerkinElmer Instruments ApplicationNote (PerkinElmer Instruments, Shel-ton, CT, 2000).

10. R. Thomas, Spectroscopy 17(2), 42–48,(2002).

11. R. Hutton, A. Walsh, D. Milton, and J.Cantle, ChemSA 17, 213–215 (1992).

12. R. Thomas, Spectroscopy 16(11),22–27, (2001).

13. J.M. Collard, K. Kawabata, Y. Kishi, andR. Thomas, Micro, January, 2002.

14. J.A. Olivares and R.S Houk, Anal. Chem.58, 20 (1986).

15. S.D. Tanner, D.J. Douglas, and J.B.French, Appl. Spectrosc. 48, 1373,(1994).

16. R. Thomas, Spectroscopy 16(9), 38–46,(2001). ■

Figure 9. Space charge matrix suppressioncaused by 1000 ppm uranium is significantlyhigher on low mass elements Li and Be than it iswith the high mass elements Tl and Pb.

120100806040200

0 50 100 150 200 250

7Li, 9Be

205Tl, 208Pb

Mass (amu)

Reco

very

tandard ICP-MS instrumentation using atraditional sample in-troduction system com-

prising a spray chamber andnebulizer has certain limitations,particularly when it comes to theanalysis of complex samples.Some of these known limitationsinclude● Total dissolved solids must be

kept below 0.2%● Long washout times are re-

quired for samples with aheavy matrix

● Sample throughput is limitedby the sample introductionprocess

● Contamination issues canoccur with samples requiringmultiple sample preparationsteps

● Dilutions and addition of in-ternal standards can be labor-intensive and time-consuming

● Matrix has traditionally beendone off-line

● Matrix suppression can bequite severe with somesamples

S

www.spectroscopyonl ine.com26 Spectroscopy 17(11) November 2002

cialized sample introductiontools — not only by the instru-ment manufacturers themselves,but also by companies specializ-ing in these kinds of accessories.The most common ones usedtoday include● Laser ablation/sampling ● Flow injection analysis ● Electrothermal vaporization ● Desolvation systems ● Chromatography separation

techniques.Let’s now take a closer look at

each of these techniques to un-derstand their basic principlesand what benefits they bring toICP-MS. In the first part of thistutorial, we will focus on laserablation and flow injection,whereas in the next tutorial onsampling accessories, we will ex-amine electrothermal vaporiza-tion, desolvation systems, andchromatography separationdevices.

Laser Ablation/SamplingThe limitation of ICP-MS to di-rectly analyze solid materials orpowders led to the developmentof high-powered laser systemsto ablate the surface of a solidand sweep the sample aerosolinto the ICP mass spectrometerfor analysis in the conventionalway (1). Before we describesome typical applications suitedto laser ablation ICP-MS, let’sfirst take a brief look at the his-tory of analytical lasers and how

● Matrix components can gen-erate severe spectral overlapson analytes

● Organic solvents can presentunique problems

● The analysis of solids, pow-ders, and slurries is verydifficult

● It is not suitable for the deter-mination of elemental speciesor oxidation states.Such were the demands of

real-world users to overcomethese kinds of problem areas thatinstrument companies deviseddifferent strategies based on thetype of samples being analyzed.Some of these strategies involvedparameter optimization or themodification of instrumentcomponents, but it was clear thatthis approach alone was notgoing to solve every conceivableproblem. For this reason, theyturned their attention to the de-velopment of sampling acces-sories, which were optimized fora particular application problemor sample type. During the past10–15 years, this demand has ledto the commercialization of spe-


Beginner’s Guide to ICP-MSPart XIII — Sampling Accessories

Robert Thomas


Today, sampling tools — such as laser ablation, flow injection, elec-trothermal vaporization, desolvation systems, and chromatographydevices — are considered absolutely critical to enhance the practicalcapabilities of inductively coupled plasma–mass spectrometry (ICP-MS) for real-world samples. Since their development more than 10years ago, these kinds of alternate sampling accessories have provedto be invaluable for certain applications that are considered problem-atic for ICP-MS.

S

November 2002 17(11) Spectroscopy 27

Tutorial

mentation, where the sampling step wascompletely separated from the excita-tion or ionization step. The major bene-fit was that each step could be inde-pendently controlled and optimized.These early devices used a high-energylaser to ablate the surface of a solidsample, and the resulting aerosol wasswept into some kind of atomic spec-

trometer for analysis. Although initiallyused with atomic absorption (5, 6) andplasma-based emission techniques (7,8), it wasn’t until the mid-1980s, whenlasers were coupled with ICP-MS, thatthe analytical community stood up andtook notice (9). For the first time, re-searchers were showing evidence thatvirtually any type of solid could be va-porized, irrespective of electrical char-acteristics, surface topography, size, orshape, and transported into the ICP foranalysis by atomic emission or MS. Thiswas an exciting breakthrough for ICP-MS, because it meant the techniquecould be used for the bulk sampling ofsolids or for the analysis of small spotsor microinclusions, in addition to beingused for the analysis of solutions.

The first laser ablation systems devel-oped for ICP instrumentation werebased on solid-state ruby lasers, operat-ing at 694 nm. These were developed inthe early 1980s but did not prove to besuccessful for a number of reasons —including poor stability, low power den-sity, low repetition rate, and large beamdiameter — which made them limitedin their scope and flexibility as a sampleintroduction device for trace elementanalysis. It was at least another fiveyears before any commercial instru-mentation became available. Theseearly commercial laser ablation systems,which were specifically developed for

they eventually became such usefulsampling tools.

The use of lasers as vaporization de-vices was first investigated in the early1960s. When light energy with an ex-tremely high power density (typically1012 W/cm2) interacts with a solid mate-rial, the photon-induced energy is con-verted into thermal energy, resulting invaporization and removal of the mate-rial from the surface of the solid (2).Some of the early researchers used rubylasers to induce a plasma discharge onthe surface of the sample and measuredthe emitted light with an atomic emis-sion spectrometer (3). Although thisproved useful for certain applications,the technique suffered from low sensi-tivity, poor precision, and severe matrixeffects caused by nonreproducible exci-tation characteristics. Over the years,various improvements were made tothis basic design with very little success(4), because the sampling process andthe ionization/excitation process (bothunder vacuum) were still intimatelyconnected and highly interactive witheach other.

This limitation led to the develop-ment of laser ablation as a sampling de-vice for atomic spectroscopy instru-

Figure 1. Optical layout of the laser ablation system used in this study (courtesy of CETACTechnologies [Omaha, NE]).

High-resolutioncolor monitor

Color CCDcamera

Motorized zoom

Pentaprism

Binocular

Translationstage

To ICP-MSNebulizer

gas in

Polarized lens withlight source

Solid sample

Energyprobe

Polarizedlens

Compensatinglens

Aperturesystem

Nd:YAG UV 266-nm laserflat-top

beam profile

Objective lens

Ringilluminator

Table I. Typical detection limits (DL) and sensitivities achievable using laser ablation for NIST 612 glass (using an Agilent Technologies 4500 ICP-MS system).

3� DL Sensitivity 3� DL SensitivityElement (ppb) (cps/ppb) Element (ppb) (cps/ppb)

B 3.0 293 Ce 0.053 131Sc 3.4 135 Pr 0.047 121Ti 9.1 5.5 Nd 0.54 12.1V 0.40 84 Sm 0.09 44.7Fe 13.6 5.2 Eu 0.10 46.7Co 0.05 90.1 Gd 1.5 4.3Ni 0.70 39.0 Dy 0.45 11.1Ga 0.18 135.3 Ho 0.01 98.1Rb 0.10 218 Er 0.21 13.8Sr 0.07 90.0 Yb 0.40 10.4Y 0.04 64.0 Lu 0.04 98.1Zr 0.20 29.6 Hf 0.40 15.2Nb 0.47 20.3 Ta 0.09 65.8Cs 0.17 274.5 Th 0.02 96.0Ba 0.036 16.0 U 0.02 294La 0.045 185

Energy � 5.0 mJ without beam attenuationPulse rate � 20 HzScan rate � 10 �m/s


Tutorial

ICP-MS, used the Nd:YAG design, op-erating at the primary wavelength of1064 nm in the infrared (10). They ini-tially showed a great deal of promisebecause analysts could finally determinetrace levels directly in the solid withoutsample dissolution. However, it soonbecame apparent that they didn’t meet

the expectations of the analytical com-munity for many reasons, includingtheir complex ablation characteristics,poor precision, lack of optimization formicroanalysis and, because of poorlaser coupling, their unsuitability formany types of solids. By the early 1990s,most of the laser ablation systems pur-

chased were viewed as novel and inter-esting, but not suited to solve real-world application problems.

These basic limitations in IR lasertechnology led researchers to investigatethe benefits of shorter wavelengths. Sys-tems were developed that were based onNd:YAG technology at the 1064-nmprimary wavelength, but using opticalcomponents to double (532 nm) andquadruple (266 nm) the frequency. In-novations in lasing materials and elec-tronic design, together with better ther-

Figure 2. (right) Imagetaken by a

petrographicmicroscope of the

surface of a thinsection of a garnet

sample, showing 10-�m spot sizes (red

circles).

Circle 22 Circle 23

Figure 3. (far right)10-�m-diameter spot

showing the flat-top energy

distribution.

Lasercrater

10- m diameter circle

Temperatureprofile of crater

Flat top of energydistribution of beam


Tutorial

mal characteristics, produced higherenergy with higher pulse-to-pulse sta-bility. These more advanced UV lasersshowed significant improvements, par-ticularly in the area of coupling effi-ciency, making them more suitable for awider array of sample types. In addi-tion, the use of more comprehensiveoptics allowed for a more homogeneouslaser beam profile, which provided theoptimum energy density to couple withthe sample matrix. This resulted in theability to make spots much smaller andwith more controlled ablations irre-spective of sample material, which wascritical for the analysis of surface de-fects, spots, and microinclusions.

The successful trend toward shorterwavelengths and the improvements inthe quality of optical components alsodrove the development of UV gas-filledlasers, such as XeCl (308 nm), KrF (248nm), and ArF (193 nm) excimer lasers.These showed great promise, especiallythe ones operated at shorter wave-lengths, which were specifically de-signed for ICP-MS. Unfortunately, theynecessitated a more sophisticated beamdelivery system, which tended to makethem more expensive. In addition, thecomplex nature of the optics and thefact that gases had to be changed on aroutine basis made them a little moredifficult to use and maintain and, as aresult, required a more skilled operatorto run them. However, their complexitywas far outweighed by their better ab-

Figure 4. Signal response of five single-shotablations for 138La and 153Eu in NIST 612 glass(using a PerkinElmer Instruments [Shelton, CT]SCIEX ELAN 6000 ICP-MS system).

3000

2000

1000

0120 130 140 150 160 170 180 190

Time (s)

Inte

nsity

(cp

s)

138La 153Eu

Circle 24

sorption capabilities for UV-transpar-ent materials like calcites, fluorites, andsilicates; smaller particle size; andhigher flow of ablated material. Evi-dence also suggested that the shorter-wavelength excimer laser exhibited better elemental fractionation charac-teristics (preferential ablation of someelements over others based on their

volatility) than the longer-wavelengthNd:YAG design.

Today there are a number of laser ab-lation designs on the market of varyingwavelengths, output energy, power den-sity, and beam profile. Even thougheach one has slightly different ablationcharacteristics, they all work extremelywell depending on the types of samples

30 Spectroscopy 17(11) November 2002

Tutorial

being analyzed. Laser ablation is nowconsidered a very reliable samplingtechnique for ICP-MS, which is capableof producing data of the very highestquality directly on solid samples andpowders. Some of the many benefits of-fered by this technique include● Direct analysis of solids without

dissolution● Ability to analyze virtually any kind

of solid material including rocks,minerals, metals, ceramics, polymers,plant material, and biologicalspecimens

● Ability to analyze a wide variety ofpowders by pelletizing with a bindingagent

● No requirement for sample to beelectrically conductive

● Sensitivity in the parts-per-billion toparts-per-trillion range, directly inthe solid

● Labor-intensive sample preparationsteps are eliminated

● Contamination is minimized becausethere are no digestion/dilution steps

● Reduced polyatomic spectral interfer-ences compared to solutionnebulization

● Examination of small spots, inclu-sions, defects, or microfeatures onsurface of sample

● Elemental mapping across the surfaceof a mineral

● Depth profiling to characterize thinfilms or coatings.Let us now exemplify some of these

benefits with some application workcarried out on a commercially-availablelaser ablation system (LSX 200 Plus,CETAC Technologies, Omaha, NE). Theoptical layout of the LSX 200 system isshown in Figure 1. (Note: Since thiswork was carried out, CETAC Tech-nologies has developed a more ad-vanced laser system called the Clarus266, which operates at the same wave-length, but includes modifications thatproduce a more homogeneous beamprofile. The major benefit of this im-proved optical design is that the high-density, flat-top beam reduces elemen-tal fractionation effects and createsmore uniform and reproducible spotsizes across a wide variety of complexmaterials.)

There is no question that geo-chemists and mineralogists have driventhe development of laser ablation forICP-MS because of their desire for ul-tratrace analysis of optically challeng-ing materials such as calcite, quartz,glass, and fluorite, combined with thecapability to characterize small spotsand microinclusions on the surface ofthe sample. For that reason, the abilityto view the structure of a thin section ofa mineral sample with a petrographicmicroscope is crucial to examine andselect an area for analysis. Figure 2shows the digital image of a garnet

Figure 5. Elemental mapping across the surface of an andradite garnet (using Agilent Technologies[Palo Alto, CA] HP 4500 ICP-MS system).

1000

800

600

400

200

01 2 3 4 5 6 7 8 9 10 11

55Mn 69Ga 95Mo 238U

Site number

Sens

itivit

y

Figure 6. Schematic of a flow injection system used for the process of microsampling.

Sample flow Autosampler

Sample loop

FIA valve

Interface Plasma torch

Intensity

Time

Waste

Pump 1

Pump 2

Carrier

Transient signalprofile

Table II. Analytical results for NASS-4 open-ocean seawater certified reference material, using flow injection ICP-MS methodology.

NASS-4 (ppb)Isotope LOD (ppt) Determined Certified

51V 4.3 1.20 � 0.04 Not certified63Cu 1.2 0.210 � 0.008 0.228 � 0.01160Ni 5 0.227 � 0.027 0.228 � 0.00966Zn 9 0.139 � 0.017 0.115 � 0.018

55Mn Not reported 0.338 � 0.023 0.380 � 0.02359Co 0.5 0.0086 � 0.0011 0.009 � 0.001208Pb 1.2 0.0090 � 0.0014 0.013 � 0.005114Cd 0.7 0.0149 � 0.0014 0.016 � 0.003

www.spectroscopyonl ine.com


Tutorial

thin-section sample (500� magnifica-tion) using a petrographic microscope.This image was generated by illuminat-ing the sample from the bottom withtransmitted light and adjustment of thepolarizers underneath the sample com-partment. On closer examination ofFigure 2, small red circles can be seen,which are 10 �m in diameter. The abla-tion sites are located in each of thesered circles. Spot sizes in this range areconsidered optimal to characterize theelemental composition across complexgrain boundaries of most minerals andstill have adequate sensitivity usingICP-MS.

To achieve this, it is desirable that thelaser beam have a homogenous profilethat produces the optimal energy den-sity to couple with the sample matrix.This results in small craters �10 �m indiameter with a flat-top energy distri-bution, as shown in Figure 3. Most ap-plications of this type will benefit fromusing higher laser energy at the samplesurface (5–6 mJ). Under these idealconditions, the flat, uniform beam isimaged directly onto the sample surfaceproviding the optimal condition forlaser coupling. The primary advantageis that the energy density on the samplesurface is uniform, constant, and inde-pendent of spot size, so that good preci-sion is achieved even at a high laserenergy.

When working with such small cratersizes, such as in the elemental charac-terization of minute inclusions, grains,or nodules, the laser system is typicallyused in the single-shot mode. In thismode of analysis, where the laser beamis fired just once, it is imperative to havespatial control and ablation cell stabilityto efficiently transport such smallamounts of ablated aerosol into theplasma. This results in good shot-to-shot precision and sensitivity as illus-trated in Figure 4, which shows signalsfrom five separate single laser shots ofthe rare earth elements 138La and 153Eu inNIST 612 glass. It must be emphasizedthat the ICP-MS instrument must offergood sensitivity on a fast transient peakfor this type of analysis. This meansthat, depending on the type of massspectrometer used, the instrumentscanning and settling times must be op-

timized for transient peak analysis.Table I shows some typical 3� detectionlimits and sensitivities achievable bylaser ablation coupled to a quadrupoleICP-MS. The laser conditions for thisexperiment are shown at the bottom ofthe table.

The final example of using laser abla-tion as a geochemical analytical tool isin its ability to map the surface of amineral. Figure 5 shows an elementalmap of an andradite garnet, which hadshown heavy zoning when viewed

through a petrographic microscopeusing transmitted polarized light. Thegarnet was sampled at 20 sites acrossthe section, using a 20-s acquisitiontime per site, a laser power of 3 mJ, anda pulse repetition rate of 5 Hz. By ex-amining data collected in this manner,differences in site mineralization behav-ior could be predicted and applied,to complement other geologicalmeasurements.

The applications described here are avery small subset of what is being done

Figure 7. A 3-D plot of analyte intensity versus mass in the time domain for the determination of agroup of elements in a transient peak.

Time Mass

Intensity

Mass spectral peak

Transient signal profile Reading10

t10

t20

t30

t40

t50

m1

m5

m10

m15

Reading20 Reading

30 Reading40 Reading

50

Figure 8. Analyte and blank spectral scans of (a) Co, (b) Cu, (c) Cd, and (d) Pb in NASS-4 open-ocean seawater certified reference material, using flow injection coupled to an ICP-MS system.

(a) (b)200

150

100

50

0

2000

1500

1000

500

00 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40

Inte

nsity

(cou

nts)

(c) 120

100

80

60

40

20

00 5 10 15 20 25 30 35 40

Time (s)

120

100

80

60

40

20

00 5 10 15 20 25 30 35 40

Time (s)

Inte

nsity

(cou

nts)

Inte

nsity

(cou

nts)

(d)

Inte

nsity

(cou

nts)


Tutorial

in the real world. A large number of ref-erences in the public domain describethe analysis of metals, ceramics, poly-mer, rocks, minerals, biological tissue,paper, and many other sample types(11–15). These references should be in-vestigated further to better understandthe suitability of laser sampling ICP-MSfor your application.

Flow Injection AnalysisFlow injection is a powerful front-endsampling accessory for ICP-MS that canbe used for preparation, pretreatment,and delivery of the sample. Originallydescribed by Ruzicka and Hansen (16),flow injection involves the introductionof a discrete sample aliquot into a flow-ing carrier stream. Using a series of au-tomated pumps and valves, procedurescan be carried out online to physicallyor chemically change the sample or an-alyte before introduction into the massspectrometer for detection. There aremany benefits of coupling flow injec-tion procedures to ICP-MS, including● Automation of on-line sampling pro-

cedures, including dilution and addi-tions of reagents

● Minimum sample handling translatesinto less chance of samplecontamination

● Ability to introduce low sample orreagent volumes

● Improved stability with harshmatrices

● Extremely high sample throughputusing multiple loops.In its simplest form, flow injection

ICP-MS consists of a series of pumpsand an injection valve preceding thesample introduction system of the ICPmass spectrometer. A typical manifoldused for microsampling is shown inFigure 6.

In the fill position, the valve is filledwith the sample (orange). In the injectposition, the sample is swept from thevalve and carried to the ICP by meansof a carrier stream (green). The meas-urement is usually a transient profile ofsignal versus time, as shown by thegreen peak in Figure 6. The area of thesignal profile measured is greater forlarger injection volumes, but for vol-umes of 500 �L or greater, the signal

peak height reaches a maximum equalto that observed using continuous solu-tion aspiration. The length of a tran-sient peak in flow injection is typically20–60 s, depending on the size of theloop. This means if multielement deter-minations are a requirement, all thedata quality objectives for the analysis,including detection limits, precision,dynamic range, number of elements,and so forth, must be achieved in thistime frame. Similar to laser ablation, ifa sequential mass analyzer such as aquadrupole or single collector magneticsector system is used, the electronicscanning, dwelling, and settling timesmust be optimized to capture the maxi-mum amount of multielement data inthe duration of the transient event (17),as seen in Figure 7, which shows athree-dimensional transient plot of in-tensity versus mass in the time domainfor the determination of a group ofelements.

Some of the many online proceduresthat are applicable to flow injectionICP-MS include● Microsampling for improved stability

with heavy matrices (18)● Automatic dilution of samples and

standards (19)● Standards addition (20)● Cold vapor and hydride generation

for enhanced detection capability forelements such as Hg, As, Sb, Bi, Te,and Se (21)

● Matrix separation and analyte pre-concentration using ion-exchangeprocedures (22)

● Elemental speciation (23).Flow injection coupled to ICP-MS

has shown itself to be very diverse andflexible in meeting the demands pre-sented by complex samples as indicatedin the above references. However, one ofthe most exciting areas of research atthe moment is in the direct analysis ofseawater by flow injection ICP-MS. Tra-ditionally the analysis of seawater usingICP-MS is very difficult because of twomajor problems. First, the high NaClcontent will block the sampler cone ori-fice over time, unless a 10–20-fold dilu-tion of the sample is made. This isn’tsuch a major problem with coastal wa-ters, because the levels are high enough.

However, if the sample is open-oceanseawater, this isn’t an option becausethe trace metals are at a much lowerlevel. The other difficulty associatedwith the analysis of seawater is that ionsfrom the water, chloride matrix, and theplasma gas can combine to generatepolyatomic spectral interferences,which are a problem, particularly forthe first-row transition metals.

Attempts have been made over theyears to remove the NaCl matrix and topreconcentrate the analytes using vari-ous types of chromatography and ion-exchange column technology. One suchearly approach was to use an HPLC sys-tem coupled to an ICP mass spectrome-ter using a column packed with silica-immobilized 8-hydroxyquinoline (24).This worked reasonably well, but wasnot considered a routine method,because silica-immobilized 8-hydroxy-quinoline was not commer-cially available; also, spectral interfer-ences produced by HCl and HNO3

(pictures used to elute the analytes)precluded determination of elementssuch as Cu, As, and V. More recently,chelating agents based on the iminodi-acetate acid functionality group havegained wider success but are still notconsidered truly routine for a numberof reasons, including the necessity forcalibration using standard additions,the requirement of large volumes ofbuffer to wash the column after loadingthe sample, and the need for condition-ing between samples because some ion-exchange resins swell with changes inpH (25–27).

However, a research group atCanada’s National Research Council hasdeveloped a very practical on-line ap-proach, using a flow injection samplingsystem coupled to an ICP mass spec-trometer (22). Using a special formula-tion of a commercially available imino-diacetate ion-exchange resin (with amacroporus methacrylate backbone),trace elements can be separated fromthe high concentrations of matrix com-ponents in the seawater with a pH 5.2buffered solution. The trace metals aresubsequently eluted into the plasmawith 1 M HNO3, after the column hasbeen washed out with deionized water.


Tutorial

The column material has sufficient se-lectivity and capacity to allow accuratedeterminations at parts-per-trillion lev-els using simple aqueous standards,even for elements such as V and Cu,which are notoriously difficult in achloride matrix. Figure 8 shows spectralscans for a selected group of elementsin a certified reference material open-ocean seawater sample (NASS-4). TableII compares the results for this method-ology with the certified values, togetherwith the limits of detection. Using thison-line method, the turnaround time isless than 4 min per sample, which isconsiderably faster than other high-pressure chelation techniques reportedin the literature.

References1. E.R. Denoyer, K.J. Fredeen, and J.W.

Hager, Anal. Chem. 63(8), 445–457(1991).

2. J.F. Ready, Affects of High Power LaserRadiation (Academic Press, New York,1972).

3. L. Moenke-Blankenburg, Laser Micro-analysis (Wiley, New York, 1989).

4. E.R. Denoyer, R. Van Grieken, F. Adams,and D.F.S. Natusch, Anal. Chem. 54,26A (1982).

5. J.W. Carr and G. Horlick, SpectrochimicaActa 37B, 1 (1982).

6. T. Kantor et al., Talanta 23, 585 (1979).7. H.C.G. Human et al., Analyst 106, 265

(1976).8. M. Thompson, J.E. Goulter, and

F. Seiper, Analyst 106, 32 (1981).9. L. Gray, Analyst 110, 551 (1985).

10. P.A. Arrowsmith and S.K. Hughes, Appl.Spectrosc. 42, 1231–1239 (1988).

11. S.E. Jackson, H.P. Longerich, G.R. Dun-ning, and B.J. Fryer, Canadian Mineral-ogist 30, 1049–1064 (1992).

12. D. Gunther and C.A. Heinrich, J. Anal.At. Spectrom. 14, 1369 (1999).

13. D. Gunther, I. Horn, and B. Hattendorf,Fresenius‘ J. Anal. Chem. 368, 4–14(2000).

14. R.E. Wolf, C. Thomas, and A. Bohlke,Appl. Surf. Sci. 127–129, 299–303(1998).

15. T. Howe, J. Shkolnik, and R. Thomas,Spectroscopy 16(2), 54–66 (2001).

16. J. Ruzicka and E.H. Hansen, Anal. Chim.Acta 78, 145 (1975).

17. R. Thomas, Spectroscopy 17(5), 54–66(2002).

18. A. Stroh, U. Voellkopf, and E. Denoyer,J. Anal. At. Spectrom. 7, 1201 (1992).

19. Y. Israel, A. Lasztity, and R.M. Barnes,Analyst 114, 1259 (1989).

20. Y. Israel, and R.M. Barnes, Analyst 114,843 (1989).

21. M.J. Powell, D.W. Boomer, and R.J.McVicars, Anal. Chem. 58, 2864(1986).

22. S.N. Willie, Y. Iida, and J.W. McLaren, At.Spec. 19(3), 67 (1998).

23. R. Roehl and M.M. Alforque, At. Spec.11(6) 210 (1990).

24. J.W. McLaren, J.W.H. Lam, S.S. Berman,K. Akatsuka, and M.A. Azeredo, J. Anal.At. Spectrom. 8, 279–286 (1993).

25. L. Ebdon, A. Fisher, H. Handley, andP. Jones, J. Anal. At. Spectrom. 8,979–981, (1993).

26. D.B. Taylor, H.M. Kingston, D.J. Nogay,D. Koller, and R. Hutton, J. Anal. At.Spectrom. 11, 187–191 (1996).

27. S.M. Nelms, G.M. Greenway, andD. Koller, J. Anal. At. Spectrom. 11,907–912 (1996). ■

Circle 25 Circle 26

lectrothermal atomiza-tion (ETA) for use withatomic absorption (AA)has proven to be a very

sensitive technique for trace ele-ment analysis during the previ-ous three decades; however, thepossibility of using the atomization/heating device forelectrothermal vaporization(ETV) sample introduction intoan ICP mass spectrometer wasidentified in the late 1980s (2).The ETV sampling process relieson the basic principle that a car-bon furnace or metal filamentcan be used to thermally sepa-rate the analytes from the matrixcomponents and then sweepthem into the ICP mass spec-trometer for analysis. This isachieved by injecting a smallamount of the sample (usually20–50 �L via an autosampler)into a graphite tube or onto ametal filament. After the sampleis introduced, drying, charring,and vaporization are achieved byslowly heating the graphite

EE


be reduced or eliminated. TheETV sampling process consistsof six discrete stages: sample in-troduction, drying, charring(matrix removal), vaporization,condensation, and transport.Once the sample has been intro-duced, the graphite tube isslowly heated to drive off thesolvent. Opposed gas flows, en-tering from each end of thegraphite tube, then purge thesample cell by forcing the evolv-ing vapors out the dosing hole.As the temperature increases,volatile matrix components arevented during the charringsteps. Just before vaporization,the gas flows within the samplecell are changed. The centralchannel (nebulizer) gas then en-ters from one end of the fur-nace, passes through the tube,and exits out the other end. Thesample-dosing hole is then au-tomatically closed, usually bymeans of a graphite tip, to en-sure no analyte vapors escape.After this gas flow pattern hasbeen established, the tempera-ture of the graphite tube isramped up very quickly, vapor-izing the residual componentsof the sample. The vaporizedanalytes either recondense inthe rapidly moving gas streamor remain in the vapor phase.These particulates and vapors

tube/metal filament. The samplematerial is vaporized into a flow-ing stream of carrier gas, whichpasses through the furnace orover the filament during theheating cycle. The analyte vaporrecondenses in the carrier gasand is then swept into theplasma for ionization.

One of the attractive charac-teristics of ETV for ICP-MS isthat the vaporization and ioniza-tion steps are carried out sepa-rately, which allows for the opti-mization of each process. This isparticularly true when a heatedgraphite tube is used as the va-porization device, because theanalyst typically has more con-trol of the heating process and,as a result, can modify the sam-ple by means of a very precisethermal program before it is in-troduced to the ICP for ioniza-tion. By boiling off and sweepingthe solvent and volatile matrixcomponents out of the graphitetube, spectral interferences aris-ing from the sample matrix can

Sampling accessories are considered critical to enhance the practical ca-pabilities of inductively coupled plasma–mass spectrometry (ICP-MS);since their development more than 10 years ago, they have proved to beinvaluable for difficult, real-world applications. In the first part of thetutorial on sampling accessories (1), we looked at laser ablation andflow injection techniques. In this second installment, we will focus onthree other important sampling approaches: electrothermal vaporiza-tion, desolvation systems, and chromatographic separation devices.

RobertThomashas more than30 years ofexperience intrace elementanalysis. He isprincipal ofScientificSolutions, acompany thatserves thetechnical writingand marketneeds of thescientificcommunity. Heis based inGaithersburg,MD, and he canbe contacted by e-mail [email protected] via his website at www.scientificsolutions1.com.

Beginner’s Guide to ICP-MSPart XIV — Sampling Accessories, Part II

Robert Thomas



Tutorial

are then transported to the ICP in the carrier gas where they are ionizedby the ICP for analysis in the mass spectrometer.

Another benefit of decoupling thesampling and ionization processes is theopportunity for chemical modificationof the sample. The graphite furnace it-self can serve as a high-temperature re-action vessel where the chemical natureof compounds within it can be altered.In a manner similar to that used in AA,chemical modifiers can change thevolatility of species to enhance matrixremoval and increase elemental sensi-tivity (3). An alternate gas such as oxy-gen may also be introduced into thesample cell to aid in the charring of thecarbon in organic matrices such as bio-logical or petrochemical samples. Herethe organically bound carbon reactswith the oxygen gas to produce carbondioxide, which is then vented from thesystem. A typical ETV sampling device,showing the two major steps of samplepretreatment (drying and ashing) andvaporization into the plasma, is shownschematically in Figure 1.

During the past 15 years, ETV sam-pling for ICP-MS has mainly been usedfor the analysis of complex matrices in-cluding geological materials (4), biolog-ical fluids (5), seawater (6), and coalslurries (7), which have proven difficultor impossible to analyze by conven-tional nebulization. By removal of thematrix components, the potential forsevere spectral and matrix-induced in-terferences is dramatically reduced.Even though ETV-ICP-MS was initiallyapplied to the analysis of very smallsample volumes, the advent of low-flownebulizers has mainly precluded its usefor this type of work.

An example of the benefits of ETVsampling is the analysis of samples con-taining high concentrations of mineralacids such as hydrochloric, nitric, andsulfuric acids. Besides physically sup-pressing analyte signals, these acids generate massive polyatomic spectraloverlaps that interfere with many ana-lytes, including arsenic, vanadium, iron,potassium, silicon, zinc, and titanium.By carefully removing the matrix com-ponents with the ETV device, the deter-

Figure 1. A graphite furnace ETV sampling device for ICP-MS, showing the two distinct steps of (a)sample pretreatment and (b) vaporization into the plasma. (Courtesy of PerkinElmer Instruments,Shelton, CT.)

(a)

Internalgas

Graphitetube

Internalgas

Carriergas

External gas

Graphite sealing probe

Samplingvalve

ICP-MS ETV

(b) Internalgas

Internalgas

Carriergas

External gasSamplingvalve

ICP-MS ETV

Figure 2. A temporal display of 50 pg of magnesium, antimony, arsenic, iron, vanadium, andmolybdenum in 37% hydrochloric acid by ETV-ICP-MS (8).

24Mg

51V

56Fe

75As

48Mo

121Sb

Sb

Mg

Fe

V

As

Mo

300

250

200

150

100

50

00.0 1.0 2.0 3.0 4.0

Time (s)

Ions

/s (�

103 )

mination of these elements becomesrelatively straightforward. Figure 2shows a spectral display in the time do-main for 50-pg spikes of a selectedgroup of elements in concentrated hy-drochloric acid (37% w/w) using agraphite furnace–based ETV-ICP-MSsystem (8). It can be seen in particularthat good sensitivity is obtained for 51V,56Fe, and 75As, which would have beenvirtually impossible by direct aspirationbecause of spectral overlaps from35Cl16O, 40Ar16O, and 40Ar35Cl, respectively.The removal of the chloride and waterfrom the matrix translates into parts-per-trillion detection limits directly in37% hydrochloric acid, as shown inTable I.

Figure 2 also shows that the elementsare vaporized off the graphite tube inthe order of their boiling points. Inother words, magnesium, which is themost volatile, is driven off first, whilevanadium and molybdenum, which arethe most refractory, come off last. How-ever, even though they emerge at differ-ent times, the complete transient eventlasts <3 s. This physical time limitation,imposed by the duration of the tran-sient signal, makes it imperative that allisotopes of interest be measured underthe highest signal-to-noise conditionsthroughout the entire event. The rapidnature of the transient also limits theusefulness of ETV sampling for routinemultielement analysis because realisti-cally only a small number of elementscan be quantified with good accuracyand precision in <3 s. In addition, thedevelopment of low-flow nebulizers,desolvation devices, and collision celltechnology means that rapid multiele-ment analysis can now be carried outon difficult samples without the needfor ETV sample introduction.


Tutorial

was not such an obvious benefit forICP-MS because more matrix was en-tering the system compared with a con-ventional nebulizer, increasing the po-tential for signal drift, matrixsuppression, and spectral interferences.This was not a problem with simpleaqueous samples, but was problematicfor real-world matrices. The elementsthat showed the most improvementwere the ones that benefited from lowersolvent-based spectral interferences.Unfortunately, many of the other ele-ments exhibited higher background lev-els and, as a result, showed no signifi-cant improvement in detection limit. Inaddition, because of the increasedamount of matrix entering the massspectrometer, it usually necessitated theneed for larger dilutions of the sample,which again negated the benefit ofusing a USN with ICP-MS. This limita-tion led to the development of an ultra-sonic nebulizer fitted with a membranedesolvator — in addition to the con-ventional desolvation system. This de-sign removed virtually all the solventfrom the sample, which dramaticallyimproved detection limits for a largenumber of the problematic elementsand also lowered oxide levels by at leastan order of magnitude (11).

The principle of aerosol generationusing an ultrasonic nebulizer is basedon a sample being pumped onto aquartz plate of a piezo-electric trans-ducer. Electrical energy of 1–2 MHz fre-quency is coupled to the transducer,which causes it to vibrate at high fre-quency. These vibrations disperse thesample into a fine droplet aerosol,which is carried in a stream of argon.With a conventional ultrasonic nebu-lizer, the aerosol is passed through aheating tube and a cooling chamber,where most of the sample solvent is re-moved as a condensate before it entersthe plasma. If a membrane desolvationsystem is fitted to the ultrasonic nebu-lizer, it is positioned after the coolingunit. The sample aerosol enters themembrane desolvator, where the re-maining solvent vapor passes throughthe walls of a tubular microporousPTFE membrane. A flow of argon gasremoves the volatile vapor from the ex-terior of the membrane, while the ana-

Desolvation DevicesDesolvation devices are mainly used inICP-MS to reduce the amount of sol-vent entering the plasma. With organicsamples, desolvation is absolutely criti-cal because most volatile solvents wouldextinguish the plasma if they weren’t re-moved or at least significantly reduced.However, desolvation of all types ofsamples can be very useful because itreduces the severity of the solvent-induced spectral interferences like ox-ides, hydroxides, and argon/solvent-based polyatomics that are common inICP-MS. The most common desolva-tion systems used today include:● Water-cooled spray chambers● Peltier-cooled spray chambers● Ultrasonic nebulizers (USNs) with

water/Peltier coolers● USNs with membrane desolvation● Microconcentric nebulizers (MCNs)

with membrane desolvation.Water- and/or Peltier- (thermo elec-

tric) cooled spray chambers are stan-dard on a number of commercial in-struments. They are usually used withconventional or low-flow pneumaticnebulizers to reduce the amount of sol-vent entering the plasma. This has theeffect of minimizing solvent-basedspectral interferences formed in theplasma, and can also help to reduce theeffects of a nebulizer-flow–induced sec-ondary discharge at the interface of theplasma with the sampler cone. Withsome organic samples, it has proved tobe very beneficial to cool the spraychamber to �10 to �20 °C (with anethylene glycol mix) in addition toadding a small amount of oxygen intothe nebulizer gas flow. This has the ef-fect of reducing the amount of organicsolvent entering the interface, which isbeneficial in eliminating the build-up of carbon deposits on the sampler cone orifice and also minimizing theproblematic carbon-based spectralinterferences (9).

Ultrasonic nebulization was first de-veloped for use with ICP–optical emis-sion spectroscopy (OES) (10). Its majorbenefit was that it offered an approxi-mately 10–20-fold improvement in de-tection limits because of its more effi-cient aerosol generation. However, this

Table I. Detection limits for vanadium, iron, and arsenic in 37% hydrochloric acid by ETV-ICP-MS.Element DL (ppt)

51V 5056Fe 2075As 40


Tutorial

lyte aerosol remains inside the tube andis carried into the plasma for ioniza-tion. Figure 3 shows a schematic of aUSN, and Figure 4 shows the principlesof membrane desolvation.

For ICP-MS, the system is best oper-ated with both desolvation stages work-ing, although for less demanding ICP-OES analysis, the membrane stage canbe bypassed if required. The power ofthe system when coupled to an ICPmass spectrometer can be seen in TableII, which compares the sensitivity

(counts per second [cps]) and signal-to-background of a membrane desolva-tion USN with a conventional crossflownebulizer for three classic solvent-basedpolyatomic interferences — 12C16O2 on44Ca, 40Ar16O on 56Fe, and 40Ar16OH on57Fe — using a quadrupole ICP-MS sys-tem. The sensitivity for the analyte iso-topes are all background subtracted.

It can be seen that for all threeanalyte isotopes, the net signal-to-background ratio is significantly betterwith the membrane ultrasonic nebu-lizer than with the crossflow design,which is a direct impact of the reduc-tion of the solvent-related spectralbackground levels. Even though this ap-proach works equally well and some-times better when analyzing organicsamples, it does not work for analytesthat are bound to an organic molecule.The high volatility of certain types oforganometallic species means that theystand a very good chance of passingthrough the microporous PTFE mem-brane and never making it into the ICP-

MS. For this reason, caution must beused when using a membrane desolva-tion system for the analysis of organicsamples.

A variation of the membrane desol-vation system uses a microconcentricnebulizer in place of the ultrasonic neb-ulizer. A schematic of this design isshown in Figure 5.

The benefit of this approach is notonly the reduction in solvent-relatedspectral interferences with the mem-brane desolvation system, but also ad-vantage can be taken of the microcon-centric nebulizer’s ability to aspiratevery low sample volumes (typically20–100 �L). This can be particularlyuseful when sample volume is limited,as in vapor phase decomposition(VPD) analysis of silicon wafers. Theproblem with this kind of demandingwork is that there is typically only 500 �L of sample available, whichmakes it extremely difficult using a tra-ditional low-flow nebulizer because itrequires the use of both cool and nor-mal plasma conditions to carry out acomplete multielement analysis. Byusing an MCN with a membrane desol-vation system, the full suite of elements— including the notoriously difficultiron, potassium, and calcium — can bedetermined on 500 �L of sample usingone set of normal plasma conditions(12).

Low-flow nebulizers were describedin greater detail in Part II of this series.The most common ones used in ICP-MS are based on the microconcentricdesign and operate at 20–100�L/min.Besides being ideal for small sample

Figure 3. Schematic of an ultrasonic nebulizerfitted with a membrane desolvation system.

Heatertube

MembranedesolvatorSweep

gas out

Condenser

DrainTransducer

Argon sweep gas in

Argon

Sample inDrainQuartz

plate

ICP

Figure 5. Schematic of a microconcentricnebulizer fitted with a membrane desolvationsystem. (Figures 3, 4, and 5 are courtesy ofCETAC Technologies, Omaha, NE.)

Microconcentric nebulizer

Heated membranedesolvator

Nebulizer gas

NitrogenSample

Heated PTFESweep gas in

Sweep gas out

ToICP-MS

Figure 4. Principles of membrane desolvation.

Heated tube

Heatedoutertube

PTFE membraneArgon sweep gas out

Argon sweepgas in

Aerosolin

Solventout

PTFEmembrane

End-on view

ToICP

Table II. Comparison of sensitivity and signal/background ratios for three isotopes(courtesy of Cetac Technologies)

Crossflow MembraneIsotope/ Mass nebulizer Net signal/ Desolvation Net signal/interference (amu) (cps) BG USN (cps) BG

44Ca (25 ppb) 2300 20,80012C16O2 (BG) 44 7640 0.3 1730 1256Fe (10 ppb) 95,400 262,00040Ar16O (BG) 56 868,000 0.1 8200 3257Fe (10 ppb) 2590 6400 3240Ar16OH (BG) 57 5300 0.5 200


Tutorial

volumes, the major benefit is that lessmatrix is entering the mass spectrome-ter, which means that there is lesschance of sample-induced long-termdrift. In addition, most low-flow nebu-lizers use chemically inert plastic capil-laries, which make them well suited forthe analysis of highly corrosive chemi-cals. This kind of flexibility has madelow-flow nebulizers very popular, par-ticularly in the semiconductor industrywhere it is essential to analyze high-purity acids using sample introductionsystems free of sources of contamina-tion (13).

Chromatographic SeparationDevicesICP-MS has gained in popularity,mainly because of its ability to rapidlyquantitate ultratrace metal contamina-tion levels. However, in its basic design,ICP-MS cannot reveal anything aboutthe metal’s oxidation state, alkylatedform, or how it is bound to a bio-molecule. The desire to understand in

what form or species an element existsled researchers to investigate the combi-nation of chromatographic separationdevices with ICP-MS. The ICP massspectrometer becomes a very sensitivedetector for trace element speciationstudies when coupled with a chromato-graphic technique like high perform-ance liquid chromatography (HPLC),ion chromatography (IC), gas chro-matography (GC), or capillary elec-trophoresis (CE). In these hybrid tech-niques, element species are separatedbased on their chromatographic retention/mobility times and theneluted/passed into the ICP mass spec-trometer for detection (14). The inten-sity of the eluted peaks are then dis-played for each isotopic mass of interestin the time domain as shown in Figure6, which shows a typical chromatogramfor a selected group of masses between60 and 75 amu.

There is no question that the ex-tremely low detection capability of ICP-MS has allowed researchers in the envi-

ronmental, biomedical, geochemical,and nutritional fields to gain a muchbetter insight into the impact of differ-ent elemental species on us and our environment — something that wouldnot have been possible 10–15 years ago.The majority of trace element specia-tion studies being carried out today can be broken down into three majorcategories:● Those involving redox systems, wherethe oxidation state of a metal canchange. For example, hexavalentchromium, Cr(VI), is a powerful oxi-dant and extremely toxic, but in soilsand water systems, it reacts with or-ganic matter to form trivalentchromium, Cr(III), which is the morecommon form of the element and is anessential micronutrient for plants andanimals (15).● Another important class is alkylatedforms of the metal. Very often the natu-ral form of an element can be toxic,while its alkylated form is relativelyharmless — or vice versa. A good exam-ple of this is the element arsenic. Inor-ganic forms of the element such asAs(III) and As(V) are toxic, whereasmany of its alkylated forms, such asmonomethylarsonic acid (MMA) anddimethylarsonic acid (DMA), are rela-tively innocuous (16).● An area being investigated more andmore is biomolecules. For example, inanimal studies, activity and mobility ofan innocuous arsenic-based growthpromoter is determined by studying itsmetabolic impact and excretion charac-teristics. Measurement of the biochemi-cal form of arsenic is crucial to know itsgrowth potential (17).

Table III represents a small cross sec-tion of speciation work that has beencarried out by chromatography tech-niques coupled to ICP-MS in thesethree major categories.

As mentioned previously, there is alarge body of application work in thepublic domain that has investigated theuse of different chromatographic sepa-ration devices, such as LC (18,19), IC(20), GC (21,22), and CE (23,24) withICP-MS. The area that is probably get-ting the most attention is the couplingof HPLC with ICP-MS. By using eitheradsorption, ion-exchange, gel perme-

Figure 6. A typical chromatogram generated by a liquid chromatograph coupled to an ICP massspectrometer, showing a temporal display of intensity against mass. (Courtesy of PerkinElmerInstruments.)

80,000

60,000

40,000

20,000

0500 600 700 800

75

70

65

60

Mas

s (m

/z)

Time (s)

Coun

ts

Table III. Some elemental species that have been studied by researchers using chromatographic separation devices coupled toICP-MSRedox systems Alkylated forms BiomoleculesSe(IV)/Se(VI) Methyl — Hg, Ge, Sn, Pb, Organo — As,

As, Sb, Se, Te, Zn, Cd, Cr Se, CdAs(III)/As(V) Ethyl — Pb, Hg Metallo-porphyrinesSn(II)/Sn(IV) Butyl — Sn Metallo-proteinsCr(III)/Cr(VI) Phenyl — Sn Metallo-drugsFe(II)/Fe(III) Cyclohexyl — Sn Metallo-enzymes


Tutorial

ation, or normal- or reversed-phasechromatography configurations, valu-able elemental speciation informationcan be derived from a sample. Let’s takea look at one of these applications —the determination of different forms ofinorganic arsenic in soil, using ion-exchange HPLC coupled to ICP-MS, toget a better understanding of how thetechnique works.

Arsenic toxicity depends directly onthe chemical form of the arsenic. In itsinorganic form, arsenic is highly toxic,while many of its organic forms are rel-atively harmless. Inorganic species ofarsenic that are of toxicological interestare the trivalent form (As[III]), such asarsenious acid, H3AsO3 and its arsenitesalts; the pentavalent form (As[V]),such as arsenic acid, H3AsO5, and its arsenate salts; and arsine (AsH3), a poisonous, unstable gas used in themanufacture of semiconductor devices.Arsenic is introduced into the environ-ment and ecosystems from naturalsources by volcanic activity and theweathering of minerals, and also fromanthropogenic sources, such as oresmelting, coal burning, industrial dis-charge, and pesticide use. The ratio ofnatural arsenic to anthropogenic ar-

senic is approximately 60:40.A recent study investigated a poten-

tial arsenic contamination of the soil inand around an industrial site in Europe.Soil in a field near the factory in ques-tion was sampled, as was soil inside thefactory grounds. The soil was dried,weighed, extracted with water, and fil-tered. This careful, gentle extractionprocedure was used to avoid disturbingthe distribution of arsenic species origi-nally present in the sample — an im-portant consideration in speciationstudies. A 10-mL sample was injectedonto an amine-based, anion-exchangeresin, where the different oxidationstates of arsenic in soil were chromato-graphically extracted from the matrixand separated using a standard LCpump. The matrix components passedstraight through the column, whereasthe arsenic species were retained andthen isocratically eluted into the nebu-lizer of the ICP mass spectrometerusing 5 mM ammonium malonate. Thearsenic species were then detected andquantified by running the instrumentin the single-ion monitoring mode, setat mass 75 — the only isotope for ar-senic. Figure 7 shows that both As(III)

and As(V) were eluted off the columnin less than three minutes using thisHPLC-ICP-MS setup. The chro-matogram also shows that both speciesare approximately three orders of mag-nitude lower in the soil sample from thesurrounding field, compared with thesoil sample inside the factory grounds.Although the arsenic does not exceedaverage global soil levels, it is a clear in-dication that the factory is a source ofarsenic contamination.

It is worth mentioning that for somereversed-phase HPLC separations, gra-dient elution of the analyte species withmixtures of organic solvents likemethanol might have to be used (25). Ifthis is a requirement, considerationmust be given to the fact that largeamounts of organic solvent will extin-guish the plasma, so introduction of theeluent into the ICP mass spectrometercannot be carried out using a conven-tional nebulization. For this reason,special sample introduction systems likerefrigerated spray chambers (26) or de-solvation systems (27) have to be used,in addition to small amounts of oxygenin the sample aerosol flow to stop thebuild-up of carbon deposits on thesampler cone. Other approaches, suchas direct injection nebulization (DIN)(28), have been used to introduce thesample eluent into the ICP-MS, but un-fortunately have not gained widespreadacceptance because of usability issues.DIN was very popular when first devel-oped in the early 1990s because of itsability to handle small sample volumesand its low memory characteristics.However, it has been replaced by othersampling techniques and, as a result,appears to have limited commercialviability.

References1. Spectroscopy 17(11), 26–33 (2002).2. C.J. Park, J.C. Van Loon, P. Arrowsmith,

and J.B. French, Anal. Chem. 59,2191–2196 (1987).

3. R.D. Ediger and S.A. Beres, Spec-trochimica Acta 47B, 907 (1992).

4. C.J. Park and G.E.M. Hall, J. Anal. At.Spectrom. 2, 473-480 (1987).

5. C.J. Park and J.C. Van Loon, Trace Ele-ments in Medicine 7, 103 (1990).

6. G. Chapple and J.P. Byrne, J. Anal. At.Spectrom. 11, 549–553, (1996).

Figure 7. HPLC-ICP-MS chromatogram showing comparison of As(III) and As(V) levels in twodifferent soil samples in and around an industrial site. (Courtesy of CETAC Technologies.)

As(V)

Uncontaminated soil (surrounding fields)

Contaminated soil(factory grounds)

As(III)

As(III) : 0.002 �g/gAs(V) : 0.023 �g/g

0 2 4

Time (min)

Inte

nsity

As(V)

As(III) As(III) : 1.8 �g/gAs(V) : 4.2 �g/g

0 3 6

Time (min)

Inte

nsity


Tutorial

7. U. Voellkopf, M. Paul, and E.R. Denoyer,Fresenius‘ J. Anal. Chem. 342, 917–923(1992).

8. S.A. Beres, E.R. Denoyer, R. Thomas,and P. Bruckner, Spectroscopy 9(1),20–26 (1994).

9. F. McElroy, A. Mennito, E. Debrah, andR. Thomas, Spectroscopy 13(2), 42–53(1998).

10. K.W. Olson, W.J. Haas Jr., and V.A. Fas-sel, Anal. Chem. 49(4), 632–637,(1977).

11. J. Kunze, S. Koelling, M. Reich, and M.A.Wimmer, Atom. Spectrosc. 19, 5(1998).

12. G. Settembre and E. Debrah, Micro(June, 1998).

13. R.A. Aleksejczyk and D. Gibilisco, Micro(September, 1997).

14. R. Lobinski, I.R. Pereiro, H. Chassaigne,A. Wasik, and J. Szpunar, J. Anal. Atom.Spectrom. 13, 860–867 (1998).

15. A.G. Cox and C.W. McLeod, Mikrochim-ica Acta 109, 161–164, (1992).

16. S. Branch, L. Ebdon, and P. O’Neill, J.Anal. Atom. Spectrom. 9, 33–37(1994).

17. J.R. Dean, L. Ebdon, M.E. Foulkes, H.M.Crews, and R.C. Massey, J. Anal. Atom.Spectrom. 9, 615–618 (1994).

18. N.P. Vela and J.A. Caruso, J. Anal. Atom.Spectrom. 8, 787 (1993).

19. S. Caroli, F. La Torre, F. Petrucci, and N.Violante, Environ: Science and PollutionResearch 1(4), 205–208 (1994).

20. J.I. Garcia-Alonso, A. Sanz-Medel, and L.Ebdon, Analytica Chimica Acta 283,261–271 (1993).

21. A.W. Kim, M.E. Foulkes, L. Ebdon, S.J.Hill, R.L. Patience, A.G. Barwise, and S.J.Rowland, J. Anal. Atom. Spectrom. 7,1147–1149 (1992).

22. H. Hintelmann, R.D. Evans, and J.Y. Vil-leneuve, J. Anal. Atom. Spectrom. 10,619–624 (1995).

23. J.W. Olesik, K.K. Thaxton, J.A. Kinzer,and E.J. Grunwald, ”Elemental Specia-tion by Capillary Electrophoresis andIonspray Mass Spectrometry,“ oralpaper given at the Winter Conferenceon Plasma Spectrochemistry (1998),sponsored by ICP Information Newsletter.

24. J. Miller-Ihli, ”ICP-MS for ElementalSpeciation: ETV and/or CZE,“ oral paper

given at the Winter Conference onPlasma Spectrochemistry (1998), spon-sored by ICP Information Newsletter.

25. J. Szpunar, H. Chassaigne, O.F.X.Donard, J. Bettmer, and R. Lobinski, Ap-plications of ICP-MS, ed. G. Hollandand S. Tanner, in Proceedings from FifthInternational Conference on PlasmaSource Mass Spectrometry (Royal Soci-ety of Chemistry, Cambridge, UK, 1997)p.131.

26. A. Al-Rashdan, D. Heitkemper, and J.A.Caruso, J. Chromatogr. Sciences 29, 98(1996).

27. H. Ding, L.K. Olson, and J.A. Caruso,Spectrochimica Acta, Part B, 51, 1801(1996).

28. S.C.K. Shum, R. Nedderden, and R.S.Houk, Analyst 117, 577 (1992). ■

Circle 41

a beginners guide - Induced coupled plasma mass spectroscopy

Documents

Transcript of a beginners guide - Induced coupled plasma mass spectroscopy