Determination of Scandium, Yttrium and Rare Earth Elements in Rocks by High Resolution Inductively...

For many years instrumental neutron activationanalysis (INAA) and isotope dilution mass spectrometry(ID-MS) were the two principal reference techniquesused to determine the rare earth elements (REE) ingeological samples (Balaram 1996). Ion exchangemethods using X-ray fluorescence spectrometry (XRF,

e.g. Robinson et al . 1986) and inductively coupledplasma-atomic emission spectrometry (ICP-AES, e.g.Jarvis and Jarvis 1988) have also been used. However,a recent survey (Potts 1997) has shown a decline in theuse of INAA and ICP-AES from the early-mid 1980’s,w i th a dramat ic inc rease in publ i shed data by

GEOSTANDARDSNEWSLETTERThe Journal of Geostandards and Geoanalysis

Determination of Scandium, Yttrium and Rare EarthElements in Rocks by High Resolution Inductively CoupledPlasma-Mass Spectrometry

Vol. 23 — N°1 p . 3 1 - 4 6

The high sensitivity, minimal oxide formation andsingle internal standard capabili ty of highresolution inductively coupled plasma-massspectrometry (HR-ICP-MS) is demonstrated in thedirect determination of Sc, Y and REE in the interna-tional reference materials: basalts (BCR-1, BHVO-1,BIR-1, DNC-1), andesite (AGV-1) and ultramafics(UB-N, PCC-1 and DTS-1). Time consuming ionexchange separation or preconcentration werefound to be unnecessary. Smooth chondritenormalized plots of the REE in PCC-1 and DTS-1were obtained in the range 0.8-50 ng g-1 (0.01-0.1x chondrite). Method precision was found to bedigestion dependent with an average externalrepeatability of 2-4% for the basalts, AGV-1 andUB-N, and 10% for PCC-1 and DTS-1. The masspeak due to 45Sc was completely resolved from29Si16O and 28Si16O1H spectral interferences usingmedium resolution, which casts doubt on theaccuracy of Sc determinations using quadrupoleICP-MS. Literature values for Y in rock referencematerials were found to be approximately 9% highafter HR-ICP-MS and XRF analysis.

Keywords: Rare earth elements, scandium, yttrium, high resolution inductively coupled plasma-mass spectrometry, silicate rocks, method developments.

La grande sensibilité, la formation minimaled’oxydes et la capacité de normalisation avec unseul standard interne de l'ICP-MS à haute résolutionsont démontrées par l'analyse directe de Sc, Y et desTerres Rares dans les matériaux de référence suivants:basaltes (BCR-1, BHVO-1, BIR-1 et DNC-1), andésite(AGV-1) et ultrabasites (UB-N, PCC-1 et DTS-1). Iln'est plus nécessaire d'effectuer de longues précon-centrations ou séparations sur colonne échangeused’ion. Des spectres de Terres Rares lissés et normalisésaux chondrites sont obtenus pour PCC-1 et DTS-1,pour des concentrations comprises entre 0.8 et 50 ng g-1 (de 0.01 à 0.1 fois les chondrites). La précision de la méthode s'est révélée être dépen-dante de la phase d'attaque et mise en solution,avec une reproductibilité externe moyenne compriseentre 2 et 4% pour les basaltes, AGV-1 et UB-N et10% pour PCC-1 et DTS-1. Le pic de la masse 45 duSc a été complètement séparé des interférencesspectrales dues à 29Si16O et 28Si16O1H en utilisantune résolution moyenne, ce qui laisse planer undoute sur l'exactitude des déterminations de Scfaites par ICP-MS à quadrupôle. Les données de lalittérature pour Y dans les matériaux de référence sesont révelées être 9% plus hautes après analyse parHR-ICP-MS et XRF.

Mots-clés : Terres Rares, scandium, yttrium, ICP-MS haute résolution, roches silicatées, développements de méthodes.

3 1

0699

Philip Robinson (1), Ashley T. Townsend (2), Zongshou Yu (3) and Carsten Münker (1, 3)

(1) School of Earth Sciences, University of Tasmania, Hobart, Tasmania, 7001, Australia. e-mail: [email protected](2) Central Science Laboratory, University of Tasmania, Hobart, Tasmania, 7001, Australia(3) Centre for Ore Deposit Research, University of Tasmania, Hobart, Tasmania, 7001, Australia

Received 06 Aug 98 — Accepted 05 Nov 98

inductively coupled plasma-mass spectrometry (ICP-MS). ICP-MS is now an established technique for thede te rm ina t ion o f t race e lemen t s i n geo log ica lmaterials (Jarvis 1988, Jenner et al. 1990, Eggins et al.1997), offering a true multi-element capability, fastanalysis t ime, low detection l imits , a large l ineardynamic range and the capacity to measure isotoperatios. INAA requires the availabili ty of a nuclearreactor and offers only a limited number of the REE,while ICP-AES and XRF are less sensitive, and alongwi th ID -MS , requ i re a leng thy ion exchangeseparation. ICP-MS also offers the option of using laserablation (e.g. Sylvester and McCandless 1997).

The use of HR-ICP-MS was first reported in 1989(Bradshaw et al. 1989, Morita et al. 1989). Comparedwith traditional ICP-MS instruments which use a qua-drupole mass filter, HR-ICP-MS instruments employ amagnetic sector mass analyser, offering the possibilityto separate analytes of interest from spectral interfe-rences (resolutions to 10,000 are in use, as reportedby Becker and Dietze 1997). This is particularly impor-tant for those elements at masses lower than 80 amu,where polyatomic interferences are commonly encoun-tered (e.g. 45Sc and 29Si16O, 51V and 35Cl16O, 56Feand 40Ar16O etc). HR-ICP-MS instruments have alsobeen found to offer other advantages over quadrupolespectrometers, namely increased sensitivity (i.e. higherion t ransmiss ion) and lower background s ignals(Moens et al. 1995). These advantages may be benefi-cial for the analysis of REE in geological matrices.

It was the aim of this study to demonstrate thecapabil i ty of HR-ICP-MS for the determination ofscandium (atomic mass 45), yttrium (90) and the REE[La (139) to Lu (175)] in rock matrices. The higherresolution available using HR-ICP-MS was appliedto the analysis of Sc (resolution 3000), while Y andthe REE were analysed using low resolution mode(resolution 300). To date, nearly all published workanalysing geological materials has been by qua-drupole instrumentation.

Also included is a study of the yttrium content inrock reference materials. Many ICP-MS workers (e.g.Jenner et al. 1990, Jarvis 1990, Totland et al. 1992,Garbe-Schönberg 1993, Shinotsuka et al . 1996,Münker 1997 and Norman et al. 1998) have foundlow yttrium concentrations by solution ICP-MS compa-red with published values. New synthetic calibrationstandards were prepared for XRF analysis and acomparison of results is reported.

Experimental

Reagents and labware

De-ionised water, purified with a Modulab PureOne system (≥16.7 Mohm cm-1 resistivity), was used forrinsing and solution preparation. Analytical gradeacids (Univar HCl, HF and BDH Analar HNO3) werepurified by double distillation in sub-boiling stills ofsilica or poly(tetrafluorethylene) (PTFE). The HClO4 wasMerck Aristar grade.

Acid digestions were performed in Savillex 7 mlscrew top PTFE vials and PicoTrace PTFE pressurevessels. Before use, polycarbonate sample containerswere soaked in 20% (v/v) HCl for one to three daysfollowed by a similar period in 10% (v/v) HNO3 beforea final rinse with ultra-pure water. All PTFE containerswere heated overnight in each of the acids at 60-80 °Cbefore rinsing.

HF-HNO3 decomposition method

This procedure was used for the basalt samplesand andesi te AGV-1 where resul ts showed goodagreement with HF-HClO4 decomposition. Althoughinsoluble fluorides are a well known problem with REEdeterminations, basalts generally give good resultswith a long (48 hour) HF-HNO3 digestion (e.g. Jenneret al. 1990). Even though this method has been foundto be successful for AGV-1, HF-HNO3 and also HF-HClO4 decomposition are generally inappropriate forfelsic samples since these often contain accessoryzircon which is relatively insoluble. Lithium boratefusion or sodium peroxide sinter are good decomposi-tion methods for these samples. Powdered sample(100 mg) was weighed into a Savillex vial and moiste-ned with ultrapure water. HF (2 ml) and HNO3 (0.5 ml)were added to the sample, the vial sealed and placedon a hotplate for 48 hours at 130 °C. At least twiceduring the first 24 hours, the container was removedfrom the hotplate, cooled and placed in an ultrasonicbath for two minutes. After 48 hours the vials wereopened and evaporated to incipient dryness. Nitricacid (1 ml) was added and further evaporated to inci-pient dryness. The residue was dissolved in 2 mlHNO3 followed by 10-20 ml water, transferred to apolycarbonate container and diluted to 100 ml (1000xdilution of sample). Indium internal standard was alsoadded to give a final concentration of 10 ng ml-1. Atleast two reagent blank solutions were prepared witheach batch of samples.

3 2


HF-HClO4 decomposition method

Th is procedure was used for u l t ramaf ic andbasal t ic samples (Heinr ichs and Herrman 1990,Münker 1998). HClO4 is more efficient at removingMg rich fluorides, although decomposition of spinelscan still be difficult. HF (3 ml) and HClO4 (3 ml) wereadded to 100 mg sample in a PicoTrace PTFE vessel.After closing the vessel and stepwise heating forthree hours, the sample was digested for fifteen hoursunder pressure at 180 °C. After evaporation, thesample was dissolved in 2 ml HNO3 and 1 ml HClfollowed by a few ml water and diluted to 100 ml totalvolume. Indium was also added as internal standard.

Instrumentation

Measurements were carried out using an ELEMENTHR-ICP-MS (Finnigan MAT, Bremen, Germany). This

i n s t r umen t ha s p rede f i ned re so l u t i on se t t i ng s(m/∆m at 10% va l ley de f in i t ion ) o f 300 ( low) ,3000 (med i um ) and 7500 ( h i gh ) . T he s i gna lintensity drops by a factor of eight to twelve whenchanging from resolution 300 to 3000 (Moens eta l . 1995) . Ins t rument se t t ings employed in th i ss t u d y a r e o u t l i n e d i n Ta b l e 1. A s t a n d a r dMeinhard nebuliser and Scott double pass watercoo led sp ray chamber we re emp loyed in t h i swork. Isotopes of interest (Table 2) were analysedusing electr ic scanning, with the magnet held atf i xed mass . The seconda r y e lec t ron mu l t i p l i e rd e t e c t o r w a s o p e r a t e d i n c o u n t i n g m o d e .I n s t r u m e n t t u n i n g a n d o p t i m i s a t i o n w e r eperformed daily using a 10 ng ml-1 multi-elements o l u t i o n c on t a i n i ng t h e e l emen t s o f i n t e re s t .Fu r t he r de ta i l s conce rn ing the ELEMENT havebeen reported in the l i terature (Feldmann et al .1994, Moens et al. 1995).


3 3

Table1.Typical instrument settings

ICP-MS

Instrument: ELEMENT (Finnigan MAT)

Resolutions (m/∆m): Low = 300, Medium = 3000 and High = 7500

Rf power: 1250 W

Gas flows: Plasma gas: 12.5 l min-1

Auxiliary: 0.95-1.1 l min-1

Sample gas: 0.95-1.1 l min-1 (optimized daily)

Torch: Fassel type

Nebuliser: Meinhard

Spray chamber: Scott-type (double pass), cooled to 3.5-5 °CCones: Ni sampler (1.1 mm orifice i.d.) and skimmer (0.8 mm orifice i.d.)

Sample uptake: Pumping via a Spetec peristaltic pump

Instrument tuning: Performed daily using a 10 ng ml-1 multi-element solution

Ion transmission: ~ 100,000 - 150,000 counts s-1 per ng ml-1 In

Scan type: Magnetic jump with electric scan over small mass range

Number of sample scans: 40

Ion sampling depth: Adjusted to obtain maximum signal intensity

Ion lens settings: Adjusted to obtain maximum signal intensity and optimum resolution

XRF (Y Kα analysis)

Instrument: Philips PW1480 X-Ray Spectrometer

X-Ray tube: 3 kW max. Sc Mo anode, operated at 90 kV, 30 mA

Crystal: LiF 200

Background angles: +0.60 deg., -0.30 deg.

Collimator: Primary -fine (0.3 mm) with auxiliary (0.14 mm)

Detector type: Scintillation counter

Total counting time: 200 seconds

Detection limit: 0.7 µg g-1 in a quartz matrix (3σ confidence)

Counting precision: 10.0 ± 0.2 µg g-1, 50.0 ± 0.2 µg g-1 in quartz

General calibration

Aqueous standard solutions covering the concentrationrange 0-50 ng ml-1 were used for external calibrationand were prepared from 10 and 100 µg ml-1 multi-element solutions (Perkin Elmer Atomic Spectroscopyand QCD Analysts-Environmental Science solutions).Two further single element yttrium solutions (1000 µgml-1) were prepared f rom both Spex high pur i ty(99.999%) and BDH (99.9%) Y2O3, ignited at 900 °Cand dissolved in HNO3 before final dilution. All stan-dard solutions were acidified with 2% (v/v) HNO3 andprepared before each analytical sequence using de-ionised water, twice purified in a quartz sub-boilingstill. Typically, six standard solutions were used for each

element, providing calibration lines having correlationcoefficients in excess of 0.995. A drift monitor solutionof 10 ng ml-1 followed by a blank were analysedevery f ive to eight samples during each analysissequence. Uptake time for each sample solution wastypically three minutes, followed by an analysis time oftwo minutes, with a final rinse with 5% (v/v) HNO3 forthree minutes.

Internal standards

Unless otherwise stated, In was used as an internals tandard and was added to each sample a t aconcentration of 10 ng ml-1 prior to ICP-MS analysis.Indium solutions were prepared from a 1000 µg ml-1

3 4


Table 2.Isotopes, abundances and some potential interferences

Element m/z % abundance Possible interferences References

*Sc 45 100 SiO, SiOH, ArLi, COOH C,D,E,G,J*Y 89 100 B,C,E,G,J

Ba 135 6.59 SnO, BaH A,B,E,F,H*Ba 137 11.23 SbO, SnO, BaH C,E,F,G,JBa 138 71.7 Ce, La, BaH D*La 139 99.91 SbO, BaH A,B,C,D,E,F,G,H,J,K*Ce 140 88.48 TeO, SnO A,B,C,D,E,F,G,H,J,K*Pr 141 100 TeO A,B,C,D,E,F,G,H,J,KNd 143 12.18 KNd 145 8.3 C,E,F,K

*Nd 146 17.19 BaO, TeO A,B,D,E,F,G,H,J,K*Sm 147 15 BaO, BaOH B,C,D,E,F,H,KSm 149 13.8 BaO, BaOH, CsO F,H,J,KSm 152 26.7 BaO, CeO, Gd A,G*Eu 151 47.8 BaO, BaOH, CsO A,C,D,E,F,G,H,I,J,KEu 153 52.2 BaO, BaOH B,E,F,G,K

Gd 155 14.8 LaO, BaO, BaOH KGd 156 20.47 LaO, BaO, BaOH, CeO, Dy K

*Gd 157 15.65 CeOH, PrO, BaF, BaOH, ArSn A,B,C,D,E,H,I,KGd 158 24.84 NdO, CeO, Dy, ArSn, BaOH GGd 160 21.86 CeO, CeOH, PrOH, NdO, NdOH, SmO, Dy, ArSn F,J*Tb 159 100 NdO, NdOH, CeO, CeOH, PrO, PrOH, ArSn A,B,C,D,E,F,G,H,I,J,KDy 161 18.9 NdO, NdOH, SmO, SmOH F,J,KDy 162 25.5 NdO, SmO, ArSn, Er K

*Dy 163 24.9 NdO, NdOH, SmO, SmOH A,B,C,D,E,F,H,I,K*Ho 165 100 NdO, NdOH, SmO, SmOH, BaCl A,B,C,D,E,F,G,H,I,J,K

Er 166 33.6 NdO, NdOH, SmO, SmOH B,E,F,G,K*Er 167 22.95 NdO, NdOH, SmO, SmOH, EuO, BaCl C,D,F,H,I,J,KEr 168 26.8 SmO, Yb, CsCl, GdO A

*Tm 169 100 EuO, NdOH, SmO, SmOH, EuOH, GdO, GdOH, BaCl A,B,C,D,E,F,G,H,I,J,KYb 171 14.3 SmO, SmOH, EuO, EuOH, GdO, GdOH, BaCl F,K

*Yb 172 21.9 SmO, SmOH, EuOH, GdO, GdOH, DyO, BaCl B,E,F,KYb 173 16.12 GdO, Hf, BaCl C,I,J,KYb 174 31.8 GdO, Hf, BaCl, ArBa A,D,G,H*Lu 175 97.41 GdO, GdOH, TbO, DyO, DyOH, BaCl, ArBa A,B,C,D,E,F,G,H,I,J,K

* Isotopes used in this study Significant interferences are highlighted

(A) Lichte et al . (1987), (B) Jarvis (1988), (C) Jenner et al. (1990), (D) Ionov et al . (1992), (E) Garbe-Schönberg (1993),(F) Dulski (1994), (G) Hollocher et al. (1995), (H) Barrat et al. (1996), (I) Shinotsuka et al. (1996), (J) Eggins et al . (1997), (K) Pin and Joannon (1997).

single element solution (High Purity Standards). Thulium(Spex High Purity Tm2O3), Lu (High Purity Standardssolution), Re (BDH 99.9%), Bi (Spex High Purity) andenriched 84Sr (83.17% 84Sr, University of Adelaide)were also tested as internal standards.

X-Ray fluorescence spectrometry

Yttrium was also determined by XRF using a PhilipsPW1480 X-ray spectrometer. Instrument settings arel i s ted in Table 1. Pressed powder pel le ts wi th adiameter of 32 mm were prepared at 3.5 tonnes cm-2

using 6 g of sample powder, polyvinyl alcohol (PVA) asa binder, and a backing of boric acid. Two sets ofcalibration standards were prepared in duplicate fromSpex h igh pur i t y Y2O3, ign i ted a t 900 °C, andCominex pure quartz. Set A (2000 µg g-1 Y) wasprepared by mixing Y2O3 with 20 g quartz for twominutes in a Rocklabs ring mill using a chrome steelhead. Pressed powder pellets were prepared. A quartzblank passed through the same procedure yielded< 1 µg g-1 Y. Set B, consisting of 0, 50, 100 and 200µg g-1 Y, was prepared using a method similar to Pottset al. (1990). Aliquots of 1 mg ml-1 Y solution werethoroughly mixed with 6 g of finely ground quartz inan agate pestle and mortar to give a slurry, dried at105 °C, mixed again and made into pressed powderpellets with PVA as a binder.

Corrections for mass absorption from the majorelements were calculated using Philips X40 softwarewith De Jongh’s calibration model and Philips alphacoefficients. The Rb Kβ1,3 doublet, overlapping withY Kα, is a major interference and must be corrected forcarefully. This was achieved using a 1000 µg g-1

Rb/SiO2 mixture prepared from Spex high puri tychemicals. Other interferences such as Pb Lγ and weakTh lines must be corrected for when Pb and Th levelsare very high. This was unnecessary for the referencematerials analysed in this study.

Calibration and analysis of low abundance samples

For the mafic and ultramafic samples of very lowREE abundances analysed here (PCC-1, DTS-1, UB-Nand BIR-1), precautions were taken in order to avoidmemory effects. Before each analytical sequence, ins-trument cones, torch, spray chamber and nebuliserwere c leaned and PVC pump tub ing rep laced .Memory effects for Sc, Y and the REE have been repor-ted to be minimal when compared with high field

strength elements (HFSE), in particular Ta and Nb (e.g.McGinnis et al. 1997, Münker 1998). Nevertheless, rin-sing time and uptake time were extended here to3.5 min and 4 min, respectively. A rinsing solution similarto the sample solutions was used (e.g. 2% v/v HNO3

in the case of the HF/HNO3 digestions). This minimisespotential element exchange between the walls of theinjection system and the sample solution which mightbe triggered by a change in acidity from sample torinsing solution. Low abundance samples were analy-sed in order of increasing REE content to avoid memoryfrom previous samples. Likewise, calibration solutions oflower concentration (0-1 ng ml-1) were analysed at thebeginning of each analytical session. Following theinitial calibration standards, a 1 ng ml-1 drift monitor,standard blank and two sample blanks were analysedbefore the first rock sample. Calibration solutions ofhigher abundance (necessary for high Y and Sc) wereanalysed after the low abundance rock samples. Eachlow abundance digestion was performed sometimes induplicate, mostly in triplicate. This procedure allowsmonitoring of possible memory effects because in thecase of memory, the measured concentrations shoulddecrease within the three repetitions of one sample. Alllow abundance values reported here are means ofreplicates which are identical within internal error.Although the above procedure and precautions wereregularly applied, memory effects were rarely found forthe REE in this study.

Results and discussion

Interferences, isotopes and choice of resolution

It has been well documented that the oxides (andto a lesser extent hydroxides) of Ba and the light rareearth elements (LREE) can cause interference problemswith the heavier REE (e.g. Dulski 1994). Possible oxide,hydroxide and isobaric interferences reported in thel i terature are detai led in Table 2. However, wi thmodern instruments and with careful optimisation ofinstrument conditions, these interferences can often beminimised. Specific oxide formation rates found in thiswork were ~ 0.2% for BaO+/Ba+ and ~ 0.4% forPrO+/Pr+, while hydroxide levels were measured as~ 0 .05% fo r BaOH +/Ba+ and ~ 0 .03% fo rCeOH+/Ce+. Barium, Pr and Ce interfere to a smallextent in the determination of Eu and Gd. Oxide for-mation for the majority of the REE was 0.15-0.25%,which is generally much lower than that reportedfrom quadrupole based ICP-MS instruments (Dulski


3 5

1994, Yoshida et al. 1996, Pin and Joannon 1997,Mak i sh ima and Nakamura 1997, Egg ins e t a l .1997). This could be a function of plasma condi-t ions, sampling depth and the nebulisation systemcharacteristic of the Finnigan instrument.

Some REE isotopes could be separated from oxideand hydrox ide in te r fe rences us ing the h ighes tresolution available with our ICP-MS instrument (nomi-nally m/∆m = 7500). For example, 157Gd could beseparated from 141Pr16O using 7500 (resolution of~ 7300 required) , whi le 151Eu and 159Tb requireresolutions of 7800 and 7700, respectively, for completeseparation from 135Ba16O and 143Nd16O. There is,however, a marked drop in sensit ivi ty when usinghigher resolution, which for example, would render Gdundetectable using resolution 7500 in most rock types.As low REE concentrations were expected in the rocksamples, and few interferences could be overcomeusing higher resolution, low resolution mode offeringmaximum sensitivity was employed for the analysis ofREE, with subsequent oxide correction.

Table 3 shows the apparent REE concentrationsresulting from oxide interferences from individual 100ng ml-1 REE solutions and 1000 ng ml-1 Ba preparedfrom Spex pure reagents (expressed as ng ml-1). Thevalues shown were obtained using resolution 300,with In as the internal standard (10 ng ml-1), underroutine operating conditions. Most interferences werefound to be minor, with corrections usually only beingnecessary for BaO on Eu, PrO and CeOH on Gd, andoccasionally NdO on Tb. Daily variations of up to

10-15% in the correction factors necessitate their mea-surement with each calibration. Isotopes determined inthis study were chosen based on (a) the oxide formationlevels measured on our instrument (Table 3) and (b)the isotopes used in other studies (Table 2).

Medium resolution (m/∆m=3000) was found to beessential for Sc analysis, as higher than expectedresu l t s we re found us ing low reso lu t ion mode .Examples of the mass spectra for Sc in BHVO-1, diges-ted using HF/HNO3 and measured using resolutionsof 300 and 3000, are shown in Figures 1 and 2. Thenarrower mass range and reduced intensity in resolu-tion 3000 are clearly shown in Figure 2. The presenceof polyatomic interferences 29Si16O and 28Si16O1H arealso noted. Scandium is unresolved from these interfe-rences in low resolution mode of HR-ICP-MS (Figure 1),and similarly, using quadrupole ICP-MS instruments.L i t t le ev idence could be found for 13C16O2 and12C16O2

1H using resolution 3000, although thesepotential interferences have previously been reportedin the determination of 45Sc in rock samples (Longerichet al. 1990, Garbe-Schönberg 1993, Eggins et al.1997). The effect of these interferences on the quantifi-ca t ion o f Sc i s demons t ra ted by the fo l low ingexamples: Sc in BHVO-1, measured using resolution300 was found to be 35.4 µg g-1, while that measuredusing resolution 3000 was 31.9 µg g-1; Tafahi: resolution300: 51.1 µg g-1 and resolution 3000: 45.2 µg g-1;TASBAS (in house standard): resolution 300: 19.0 µg g-1

and resolution 3000: 14.1 µg g-1. We have found thatafter dissolution of sample using both the HF/HNO3

attack and Na2O2 sinter (Robinson et al. 1986), trace

3 6


Table 3.Summary of oxide formation for selected 100 ng ml-1 REE and 1000 ng ml-1 Ba solutions

Apparent REE concentration (ng ml-1) when the following individual solutions were aspirated

Ba La Ce Pr Nd Sm Eu Gd Tb Dy1000 ng ml-1 100 ng ml-1 100 ng ml-1 100 ng ml-1 100 ng ml-1 100 ng ml-1 100 ng ml-1 100 ng ml-1 100 ng ml-1 100 ng ml-1

139La 0.024 - - - - - - - - -140Ce 0.003 - - - - - - - - -

141Pr - - 0.004 - - - - - - -146Nd 0.003 - 0.005 0.003 - - - - - -147Sm - - - - 0.008 - - - - -151Eu 0.130 - - - - - - - - -

157Gd 0.004 0.022 0.160 3.060 0.040 - - - - -159Tb - - 0.003 - 0.072 - - - - -163Dy - - - - 0.057 0.114 - - - -165Ho - - - - - 0.030 - - - -

167Er - - - - 0.005 0.004 0.124 - - -169Tm - - - - - 0.003 0.034 - - -172Yb - - - - - - - 0.279 - 0.065175Lu - - - - - - - 0.005 0.336 0.028

Major interferences are highlighted.

silica still remains when none might be expected, inthe first case owing to volatilisation of SiF4 and in thesecond, washing away of sodium silicates. In contrast,no silica peak was observed after HF/HClO4 digestionof BIR-1 and DNC-1. This may be due to the extravolume of acids used in the HF/HClO4 digestionprocedure, along with the higher efficiency of HClO4

in removing fluorides (Bock 1979).

Matrix effects

Matrix effects were examined by analysing an inhouse basalt standard TASBAS, at dilutions of 1000x,

2000x and 5000x. No discernible difference wasobserved between each dilution. High resolution ICP-MS should be no different from quadrupole ICP-MS inthis regard, as dilution dependent matrix problems aretypically associated with the plasma source.

Internal standard

It is common practice in ICP-MS analysis to use aninternal standard to compensate for matrix effects andinstrumental drift (Evans and Giglio 1993). An internals tandard element of s imi lar mass and ionisat ionenergy to the analyte is often chosen, with more than


3 7

m/z

Inte

nsity

(cou

nts

s-1)

Figure 1. Mass spectrum of 45Sc in BHVO-1 using low

resolution (300).

Inte

nsity

(cou

nts

s-1)

m/z

Figure 2. Mass spectrum of 45Sc in BHVO-1 using medium

resolution (3000) showing separation of Sc from nearby

interferences 29Si16O, 28Si16O1H, 13C16O2 and 12C16O21H.

Table 4.Analysis of BHVO-1 using different internal standards (µg g-1)

84Sr 115In 169Tm 175Lu 185Re 209Bi Reference (A) Reference (B)

45Sc 32.1 31.4 - - 31.9 - 31.8 31.889Y 25.2 24.0 24.5 24.8 24.5 - 27.6 28

137Ba - 133.0 133.0 138.8 135.0 - 139 133139La - 15.5 15.5 16.0 15.9 - 15.8 15.5

140Ce - 37.4 37.5 37.7 38.5 - 39 38141Pr - 5.48 5.48 5.68 5.69 - 5.7 5.45

146Nd - 24.1 23.5 25.3 24.4 - 25.2 24.7147Sm - 6.01 5.93 6.47 6.11 - 6.2 6.17151Eu - 2.09 2.08 2.22 2.14 - 2.06 2.06

157Gd - 6.22 6.40 6.56 6.48 - 6.4 6.22159Tb - 0.92 0.95 0.96 0.97 - 0.96 0.95163Dy - 5.15 5.14 5.53 5.24 - 5.2 5.25165Ho - 0.99 1.01 1.03 1.03 - 0.99 1

167Er - 2.50 2.48 2.69 2.53 - 2.4 2.56169Tm - 0.33 - 0.35 0.34 - 0.33 0.33172Yb - 1.95 1.91 2.12 1.96 - 2.02 1.98175Lu - 0.26 0.27 - 0.28 - 0.291 0.278

208Pb - 2.12 2.09 2.37 2.09 2.06 2.6 2.1232Th - 1.25 1.28 1.30 1.32 1.17 1.08 1.26238U - 0.44 0.42 0.44 0.46 0.44 0.42 0.42

Mean of five HF/HNO3 digestions.

(A) Govindaraju (1994), (B) Eggins et al. (1997) “preferred values”.

one typically being used for each elemental suite. Forexample , in te rna l s tandard combina t ions haveincluded Ru and Re (Dulski 1994); Rh, In and Re(Münker 1998); In and Bi (Ionov et al. 1992); In andRe (Pin and Joannon 1997); Rh and Bi (Balaram et al.1996); Be, In and Re (Garbe-Schönberg 1993); Rhand/or In (Yoshida et al. 1996). These studies aretyp ica l l y charac ter ized by two to th ree in te rnalstandards of high and low mass, respectively. Recently,Eggins et al . (1997) found one internal standardunsat i s fac tory for geochemical analy s i s us ing aquadrupole ICP-MS instrument, and so developed amethod using four enriched and five natural isotopeswhich gave very precise results (1-2% RSD).

Preliminary results from our HR-ICP-MS indicatedthat only one internal standard, indium, may suffice forthe analysis of rock solutions of 1000x dilution. Toconfirm this, a series of five basaltic (BHVO-1) solutionswere prepared by HF/HNO3 dissolution and spikedwith 10 ng ml-1 enriched 84Sr, natural 115In, 169Tm,175Lu, 185Re and 209Bi to determine the most suitablechoice of internal standard(s). Corrections were madefor the Sr, Tm and Lu already in the reference material.Scandium, Y, Ba, REE, Pb, Th, and U were measuredand results are shown in Table 4.

Compared with 115In, there is little difference in theresults when using 84Sr as an internal standard for the

analysis of the lighter mass elements 45Sc and 89Y.Likewise, using both 115In and 209Bi as internal stan-dards showed similar results for the heavier masses208Pb, 232Th and 238U. The middle masses showed noimprovement using 169Tm, 175Lu or 185Re as internalstandards instead of 115In alone. Based on theseresults, indium was used as a single internal standardfor isotopes from 45Sc to 238U. A similar outcome wasfound by Townsend et al. (1998) when analysing urinesamples with the same HR-ICP-MS instrument. In thatwork, little difference was found for the analysis of Cu,Zn, Cd and Pb when using Sc, In and Bi as an internalstandard combination over 115In alone.

Detection limits

Although instrument sensitivity is a major factor inachieving good detection limits, a high backgroundcaused by memory effects and sample contaminationcan also have a large influence. Table 5 lists thedetec t ion l imi ts measured whi le aspirat ing threesolutions: (a) ultra pure water (doubly distilled in asub-boiling quartz still), (b) a 2% v/v HNO3 solution,and (c) an ordinary sample blank solution (HF/HClO4

digestion followed by final dilution with 2% HNO3).Detection limits shown were defined as three times thestandard deviation of (at least) ten consecutive blankmeasurements (3σ) . Resul ts were obtained underroutine operating conditions with no special instrument

3 8


Table 5.ICP-MS detection limits

*This Study Ionov et al. 1992 Pin and Joannon 1997 Makishima and Nakamura 1997

*sample blank sample blankHF/HClO4 digestion micro-conc. nebuliser flow injection

Element *pure water *2% v/v HNO3 (2% v/v HNO3 final soln.) 0.5 mol l-1 HNO3

pg ml-1 in solution pg ml-1 in solution pg ml-1 in solution pg ml-1 in solution pg ml-1 in solution pg ml-1 in solution low dilution (113x)(ng g-1 in rock, (ng g-1 in rock, (ng g-1 in rock, (ng g-1 in rock)

1000 x dilution) 1000 x dilution) 1000 x dilution)

Sc 17.9 13.9 13.3 10 - - -Y 0.38 0.80 2.54 - - 34 4La 0.15 0.35 1.74 1.5 1.6 7 0.8Ce 0.33 0.88 1.37 1.5 6.8 8 0.9Pr 0.09 0.16 0.34 1 1.4 4 0.5Nd 1.06 1.21 1.31 2 0.6 20 2Sm 0.50 1.27 1.28 1.5 0.7 15 2Eu 0.35 0.37 0.55 0.4 0.4 4 0.4Gd 0.97 3.02 1.44 1 0.5 26 3Tb 0.09 0.11 0.21 0.2 0.6 4 0.4Dy 0.16 0.78 0.58 1 0.4 21 2Ho 0.04 0.39 0.22 0.25 0.4 3 0.4Er 0.10 0.28 0.44 0.6 0.5 23 3Tm 0.05 0.41 0.19 0.2 0.5 3 0.3Yb 0.12 0.37 0.71 0.4 0.5 6 0.6Lu 0.05 0.63 0.22 0.2 0.5 4 0.4

* Reported detection limits are three times the standard deviation of 10-15 blank measurements.


3 9

n=2 na=4This work RSD Ref. (A)

Element µg g-1 % µg g-1

AGV-1Sc 12.4 0.8 12.2Y 18.7 0.2 20La 39.2 2.1 38Ce 72.1 3.6 67Pr 8.79 2.5 7.6Nd 32.5 2.2 33Sm 6.02 3.8 5.9Eu 1.68 4.2 1.64Gd 4.94 3.3 5Tb 0.69 3.8 0.7Dy 3.5 1.9 3.6Ho 0.69 2.5 0.67Er 1.81 4.3 1.7Tm 0.26 4.7 0.34Yb 1.68 0.9 1.72Lu 0.251 4 0.27Average RSD (%), all elements 2.8

n=2 na=4This work RSD Ref. (A)

Element µg g-1 % µg g-1

BCR-1Sc 32.3 1.7 32.6Y 33.6 0.6 38La 25.2 2 24.9Ce 53 6.3 53.7Pr 6.87 3.9 6.8Nd 27.9 4.2 28.8Sm 6.57 7.1 6.59Eu 1.96 2.1 1.95Gd 6.62 3.3 6.68Tb 1.05 5.1 1.05Dy 6.05 2.3 6.34Ho 1.26 0.9 1.26Er 3.52 1.8 3.63Tm 0.52 3 0.56Yb 3.29 3.2 3.38Lu 0.5 2.9 0.51Average RSD (%), all elements 3.2

n=2 na=4 n=19 na=1This work RSD This work RSD Ref. (A)

19 digestions over 18 months

Element µg g-1 % µg g-1 % µg g-1

BHVO-1Sc 31.9 1.8 31 3.5 31.8Y 24.9 1.9 24 4.2 27.6La 16 1 15.5 3.6 15.8Ce 39 3 38 3.2 39Pr 5.65 1.7 5.5 3.7 5.7Nd 25.1 2.2 25 4.6 25.2Sm 6.26 2.2 6.23 4.9 6.2Eu 2.12 2 2.14 3.4 2.06Gd 6.26 3.1 6.35 6.6 6.4Tb 0.97 3.8 0.94 5.2 0.96Dy 5.34 2.6 5.28 5.3 5.2Ho 1.02 4 1.01 6.4 0.99Er 2.59 2.4 2.57 6.4 2.4Tm 0.34 3.2 0.34 5.4 0.33Yb 1.99 3.5 2 5.7 2.02Lu 0.28 3.7 0.28 6.8 0.291Average RSD (%), all elements 2.6 4.9

Table 6.Trace element concentrations (µg g-1), precision (%RSD) and published values for rock reference materials

n=7 na=1 n=6 n=3This work RSD Ref. (A) Ref. (C) Ref. (E)

Element µg g-1 % µg g-1 µg g-1 µg g-1

DNC-1Sc 30.5 2.7 31 31.1 -Y 15.9 1.5 18 18.03 -La 3.66 3.2 3.8 3.68 3.91Ce 8.04 3.4 10.6 8.17 8.46Pr 1.11 1.4 1.3 1.113 1.11Nd 4.89 2.2 4.9 4.95 4.8Sm 1.47 2.3 1.38 1.44 1.3Eu 0.61 2.7 0.59 0.592 0.515Gd 2.04 1.9 2 2.02 1.79Tb 0.39 1.8 0.41 0.39 0.33Dy 2.76 2.3 2.7 2.71 2.35Ho 0.64 1.7 0.62 0.638 0.537Er 2 2.7 2 1.945 1.63Tm 0.294 1.9 [0.33] - 0.271Yb 2.01 3 2.01 1.915 1.8Lu 0.308 2.3 0.32 0.292 0.3Average RSD (%), all elements 2.3 0.9 1

n=13 na=1This work RSD Ref. (A) Ref. (B)

Element µg g-1 % µg g-1 µg g-1

BIR-1Sc 42 3.1 44 -Y 14.1 1.8 16 15.5La 0.63 5.5 0.62 0.61Ce 1.89 5.3 1.95 1.95Pr 0.38 4.6 0.38 0.38Nd 2.31 3.8 2.5 2.34Sm 1.07 2.6 1.1 1.1Eu 0.52 3.4 0.54 0.52Gd 1.77 2.5 1.85 1.84Tb 0.35 3.5 0.36 0.36Dy 2.43 2.8 2.5 2.51Ho 0.55 3.5 0.57 0.56Er 1.64 3.1 1.7 1.66Tm 0.24 4.7 0.26 0.25Yb 1.61 2.7 1.65 1.63Lu 0.24 3.9 0.26 0.25Average RSD (%), all elements 3.6

n=10 Resolution n=10 Resolution n=12na=1 300 na=1 3000

This work RSD This work RSD Ref. (C)Element µg g-1 % µg g-1 % µg g-1

TafahiSc Si interference - 45.24 2.1 45.5Y 7.76 1.7 8.24 2.1 9.11La 0.93 1.6 0.95 4.8 0.938Ce 2.22 2 2.168 3.8 2.22Pr 0.368 1.6 0.331 3.9 0.361Nd 1.91 1.5 1.887 3.4 1.93Sm 0.72 1.5 0.711 6.7 0.722Eu 0.309 2.2 0.288 6.4 0.305Gd 1.04 2.1 1.046 5.9 1.069Tb 0.188 1.4 0.192 6.4 0.207Dy 1.35 1.5 1.394 5.3 1.384Ho 0.32 1.3 0.311 6 0.322Er 0.951 0.9 0.965 5.5 0.98Tm 0.147 2 0.141 4.5 -Yb 0.987 0.9 1.029 4.5 0.992Lu 0.156 1.7 0.147 4 0.153Average RSD (%), all elements 1.6 4.7 1.5

preparation or maintenance, other than cleaning ofcones, torch and spray chamber.

Comparing detection limits with other laboratoriescan be a di f f icu l t exerc ise , s ince fu l l detai l s o ft he conditions and solutions used are not alwaysgiven. Table 5 lists results from three studies usingquadrupole instruments, each of which focused on theanalysis of low level (ng g-1) samples. Ionov et al.1992 reported exceptional ly low detec t ion l imitsconsidering the sample blank was used for thesemeasurements . “Special measures” were taken toreduce memory effects, blanks were prepared in aclean room environment, while the ICP-MS instrumentwas solely dedicated to peridotite analysis. Pin andJoannon (1997) gained increased sensitivity by usingion exchange separation as a sample pretreatment.The blank used to determine detection limits was notspecified. Makishima and Nakamura (1997) used flowinjection analysis, increased plasma power and lowsample dilution (113x) to achieve good detection limits(which were measured in an “ideal solution” of 0.5mol l-1 HNO3).

In this study, the detection limits found under stan-dard conditions with a mixed acid sample blank arecomparable to those reported from other laboratorieswho have specialised in low level determinations.However, with an extra clean blank (pure water only)the detection l imits were improved by a factor of(approximately) ten. We regard the detection limitsobtained from the mixed acid blank as more realisticsince they reflect those during the routine analyticalprocedure. Except for Sm, Eu, Gd and Tb in the refe-rence materials PCC-1 and DTS-1, detection limitsobtained here from the mixed acid blanks always liemore than a factor of ten below the abundance in therock solution.

Precision and accuracy

Results for a range of international rock referencematerials are compared with published values in Table 6.Initial work using AGV-1, BCR-1 and BHVO-1 producedresults derived by averaging measurements from twodecompositions, with each determination being theaverage of four ICP-MS measurements taken separatelyover several weeks. Further work over an eighteenmonth period resulted in separate digestions and ana-lyses for BHVO-1 (19x), BIR-1 (13x), Tafahi [20x, abasalt with low REE (Eggins et al. 1997)], DNC-1 (7x),UB-N (6x), PCC-1 (8x) and DTS-1 (7x).

4 0


n=8 na=2-3 n=12 n=7 n=4This work RSD Ref. (A) Ref. (C) Ref. (D) Ref. (F)

Element µg g-1 % µg g-1 µg g-1 µg g-1 µg g-1

PCC-1Sc 8 4.7 8.4 9 7 -Y 0.079 3.4 [0.1] 0.087 - 0.079La 0.046 10.7 0.052 0.029 0.039 0.034Ce 0.0528 1.8 0.1 0.053 0.057 0.061Pr 0.0076 9 0.013 0.0068 0.0085 0.0091Nd 0.026 2.5 0.042 0.025 0.03 0.035Sm 0.007 19.5 0.0066 0.005 0.008 0.0095Eu 0.0009 22.7 0.0018 0.0011 0.0018 0.0024Gd 0.0059 16.9 [0.014] 0.0061 0.008 0.013Tb 0.0011 16.7 0.0015 0.0012 0.0015 0.0014Dy 0.011 13.9 0.01 0.0087 [0.013] 0.016Ho 0.003 7.6 0.0025 0.0027 0.0038 0.0034Er 0.0117 8.3 [0.012] 0.0113 0.0123 0.016Tm 0.0028 7.4 0.0027 - 0.0025 0.0032Yb 0.0227 4.6 0.024 0.0213 0.0215 0.028Lu 0.0047 4.9 0.0057 0.0046 0.0049 0.0054Average RSD (%), all elements 9.7 17.4 10.8 13.2

n=6 na=2 n=10 n=3This work RSD Ref. (A) Ref. (D) Ref. (E)

Element µg g-1 % µg g-1 µg g-1 µg g-1

UB-NSc 11.4 1.6 13 14 -Y 2.45 3.3 2.5 - -La 0.323 1.3 0.35 0.33 0.319Ce 0.76 2.3 0.8 0.8 0.77Pr 0.12 2.7 0.12 0.123 0.113Nd 0.6 2 0.6 0.61 0.594Sm 0.226 2.8 0.2 0.216 0.204Eu 0.084 1.1 0.08 0.081 0.0726Gd 0.326 2.7 0.3 0.32 0.294Tb 0.061 2 0.06 0.06 0.054Dy 0.433 1.3 0.38 0.42 0.377Ho 0.099 0.6 0.09 0.097 0.0843Er 0.308 0.6 0.28 0.282 0.251Tm 0.046 1.9 0.045 0.0434 0.0415Yb 0.315 2.6 0.28 0.283 0.272Lu 0.049 2.5 0.045 0.046 0.0453Average RSD (%), all elements 2 2.8 14.9

n=7 na=2-3 n=2 n=3 n=4This work RSD Ref. (A) Ref. (C) Ref. (D) Ref. (F)

Element µg g-1 % µg g-1 µg g-1 µg g-1 µg g-1

DTS-1Sc 3.16 2.2 3.5 5.3 3.5 -Y 0.037 8.4 0.04 0.038 - 0.042La 0.041 6.7 0.029 0.0246 0.025 0.023Ce 0.05 3.4 0.072 0.1 0.05 0.052Pr 0.0068 5 0.0063 0.0063 0.0074 0.0066Nd 0.022 5.2 0.029 0.0234 0.027 0.025Sm 0.0063 16.5 0.0046 0.0031 [0.007] 0.009Eu 0.0008 12.4 0.0012 0.0013 0.0013 0.0019Gd 0.0042 11.7 [0.0038] 0.0044 0.0063 0.0081Tb 0.0007 18.1 0.0008 0.0007 0.001 0.0012Dy 0.006 25.1 [0.003] 0.0038 0.0085 0.0082Ho 0.0017 12.6 0.0013 0.0014 0.0016 0.0024Er 0.0055 7.8 [0.004] 0.005 0.0074 0.0061Tm 0.001 14.2 0.0014 - 0.0013 0.0017Yb 0.0097 3.5 0.01 0.009 0.01 0.011Lu 0.0021 5 0.0024 0.0019 0.0022 0.0029Average RSD (%), all elements 9.9 13

Information values are shown in brackets. n number of digestions.

na number of ICP-MS analyses performed on each digestion.

(A) Govindaraju (1994), (B) Jochum et al . (1994), (C) Eggins et al .(1997), (D) Ionov et al . (1992), (E) Pin and Joannon (1997), (F) Makishima and Nakamura (1997).

Table 6 (continued).Trace element concentrations (µg g-1), precision (%RSD) and published values for rock reference materials


4 1

Average precision on the duplicate digestions ofAGV-1, BCR-1 and BHVO-1 (eight analyses for eachreference material) was 2.6%-3.2%, whereas that forBHVO-1 was 4.9% from nineteen digestions, and3.6% from thirteen digestions for BIR-1. The poorer ave-rage precision in the latter results reflects a variety offactors over time, including different digestion methodsby various workers and varied instrument conditions(e.g. plasma stability, cone conditions, inclusion ofmedium and low resolution results). Excellent averageprecision of 1.6% was obtained for Tafahi in low reso-lution (300) when ten samples were digested andanalysed simultaneously. A second set of ten digestionswere analysed independently using medium (3000)resolution. While the final concentration values weresimilar for both cases, the average precision usingresolution 3000 was only 4.7% because of the lowersignal intensities associated with this resolution mode(Moens et al. 1995). Ultramafics with very low elementconcentrations gave 2.0% average precision for UB-N,9.7% for PCC-1 and 9.9% for DTS-1, which compareswell with results from other workers (2.8%-17.4%)where special analytical conditions or modificationswere often employed. (e.g. Ionov et al. 1992, Eggins etal. 1997, Makishima and Nakamura 1997).

Internal (instrumental) precisions of 1-2% wereobtained using HR-ICP-MS. Much higher externalrepeatabilities, however, were found when multiplesample digestions were considered (Table 6). Whiledissolution is relatively straightforward for mafic rocksas in this study, samples with refractory minerals suchas zircon, sphene, spinel and garnet can be extremelydifficult. Often the HREE, Th, U and Zr-Hf reside insuch minerals . Preparat ion techniques such as a

sodium peroxide sinter (Robinson et al. 1986) or alithium borate fusion can be employed (e.g. Jarvis1992, Chao and Sanzolone 1992). However, problemsfound with the fusion method include contaminationfrom fusion reagents and possible loss of volati leelements such as Pb, Tl and Sb (Totland et al. 1992).As a result, acid digestion research work is preferredin our laboratory using a pressure digestion systemsuch as PicoTrace.

Accuracy can be judged by comparing resultsfound in this work with those from other studies, andplotting chondrite normalized data (Figures 3-7). Withfew exceptions, our results are in agreement with thosefound for the well characterized reference materialsAGV-1, BCR-1, BHVO-1, BIR-1 and DNC-1. Exceptionsare Y for all reference materials, as well as Ce, Pr andTm in AGV-1 and DNC-1. Our Y values are approxi-mately 9% lower than published values, and arediscussed further in the next section. Cerium is approxi-mately 8% higher in AGV-1 and 24% lower in DNC-1compared with Govindaraju’s 1994 compilation. Pr is16% higher in AGV-1 and 15% lower in DNC-1, Tm is24% lower in AGV-1 and 11% lower in DNC-1. Ourchondrite normalized data plot smoothly for AGV-1and DNC-1, suggesting published results may be inerror. Results for low level reference materials Tafahi,UB-N, PCC-1 and DTS-1 are in general agreementwith other workers, although PCC-1 and DTS-1 are solow tha t the re i s l i t t l e re l iab le da ta pub l i shed .Chondrite normalized REE patterns for DTS-1 and PCC-1(Figures 6 and 7) show good agreement with thosefrom available literature references. The consistentV-shape observed by the d i f fe ren t labora tor iessuggests that ICP-MS can be a reliable method at

Figure 3. Chondrite normalized

plots for AGV-1, BCR-1 and BHVO-1.

Rock

/cho

ndri

te

Rock

/cho

ndri

te

Figure 4. Chondrite normalized

plots for BIR-1, UB-N and DTS-1.

such low abundance. This steep, linear, positive HREE-MREE gradient and upward-inflected LREE patternseems to be characteristic of peridotite from subductionenvironments (Parkinson and Pearce 1998).

There has been some concern noted in the literatureon the possible interference of La and Pr by CaClO4,formed during HF/HClO4 digestions (Longerich 1993).This interference was not apparent in this work, asdemonstrated by the fact that similar La and Pr valueswere obtained for BIR-1 when prepared by bothHF/HClO4 and HF/HNO3 (C l - f ree ) d iges t ion .Occasionally, however, spurious high results wereobtained for La in BIR-1 (e.g. 0.6-0.8 µg g-1 comparedwith an expected value of ~ 0.62 µg g-1) found fromboth acid mixtures. As all sample blanks were consis-tently low, this was attributed to accidental contamina-tion in the BIR-1 sample, possibly from the use of La asa heavy element absorber in XRF flux in our laboratory.Lanthanum values (though not Pr) for PCC-1 (0.046 µg g-1)and DTS-1 (0.041 µg g-1) were found to be slightlyhigher than some published results (e.g. Ionov et al.1992 and Makishima and Nakamura 1997, whoreported La values of 0.023-0.039 µg g-1) also usingHF/HClO4 digestion. Lanthanum in these referencematerials is well above our detection limit (25x) andgave good precision (~ 7-10%) in multiple digestionsover three months, which suggests there is no contami-nation in these samples. Another ultramafic rock, UB-N(0.32 µg g-1 La), and Tafahi basalt (0.94 µg g-1 La)were in good agreement with published values.

Yttrium in rock reference materials

Work in our laboratory consistent ly found lowyttrium results in rock reference materials in comparisonwith published values. Poor dissolution was initially sus-pected as the cause of this discrepancy. However,good agreement was found for many other elements(such as Zr, Hf, Th, U and the HREE) that often reside inminerals which are the main hosts to Y. Low yttriumvalues have been noted in other studies (e.g. Totlandet al. 1992, Shinotsuka et al. 1996). From the analysisof approximately one hundred samples (basal tsthrough to rhyolites) for yttrium by both ICP-MS andXRF, Münker (1997) found ICP-MS results to be consis-tently 10-15% lower than XRF results. It is relevant tonote that rock reference materials were used for theXRF calibration.

Yttrium in a range of rock samples was measuredby XRF using pure chemical reagents as calibration

4 2


Figure 5. Chondrite normalized plots for DNC-1,

Tafahi (resolution 300 and 3000) and PCC-1.

Rock

/cho

ndri

teRo

ck/c

hond

rite

Rock

/cho

ndri

te

Figure 6. Chondrite normalized plot for PCC-1.

Comparison with other laboratories.

Figure 7. Chondrite normalized plot for DTS-1.

Comparison with other laboratories.

standards. Results are reported in Table 7 and arecompared with ICP-MS values calibrated using stan-dard solut ions from four manufacturers (solut ionsfrom Perkin Elmer, QCD, Spex and BDH Y2O3 wereall similar). The XRF results listed in this table werederived using two sets of synthetic Y standards pre-pared via different methods. Both XRF and ICP-MSvalues were found to be in agreement and were (onaverage) 9% lower than published results, sugges-ting that the literature values may be in error. Thereare a number of possible reasons for this. Yttrium is amono-isotopic element and cannot be determinedby isotope dilution mass spectrometry. INAA is alsoinappropriate as there is no emission of γ rays from90Y, which is the only nuclide produced by (n,γ) reac-t ion on Y (e.g. Shinotsuka et al . 1996). Thus, themajority of reported values for yttr ium have beenobtained using XRF where rock reference materialsof uncertain value are often used for calibration.Yttrium determinations are not straightforward by XRFas there is a large Rb correction involved. Yttriuma l so con t r ibu tes a spec t ra l i n te r fe rence to theCompton line if a Mo X-ray tube is used, affectingabso rp t i on co r rec t i on s , pa r t i cu la r l y on h igh Ysamples such as is likely to be found in calibrationstandards. The use of Compton scattering (where theCompton peak on the 2000 µg g-1 calibration stan-dards was inadequately corrected for Y overlap)gave resul ts on the rock reference mater ials 3%lower than l i terature values, instead of 9% usingalpha coefficients.

In summary, the reason for many recommended Yvalues in rock reference materials being high may bepoor quality XRF data. The danger of using a rockre fe rence mate r ia l fo r ca l ib ra t ion e .g . BHVO-1(Eggins et al. 1997) is illustrated here, as althoughsuch results agree with recommended values, theyare not necessarily correct.

The possibility of Y results being 9% high in thel i terature has signif icant geological implicat ions.Yittrium shows a geochemical behaviour similar tothat of Ho, a heavy rare earth element (HREE). Thisis manifested by virtually constant Y/Ho ratios inmafic rocks and chondrites (Jochum et al . 1986).Y t t r i u m i s , t h e r e f o r e , u s e d t o u n d e r s t a n dpetrogenet ic processes in igneous rocks . Recentwork (Bau 1996) showed that the Y/Ho ratio canbe substantially changed in felsic magmatic andhydrothermal systems. Yt tr ium in conjunc tion withHo m igh t , t h e r e f o r e , b e u s ed f o r mon i t o r i n ghyd ro t he rma l a l t e ra t i on p roce s se s . I t may beprudent to reassess the Y content in chondrites byICP-MS and improved XRF methods using syntheticY calibration standards for both.

Advantages of HR-ICP-MS

High resolution magnetic sector ICP-MS offers anumber of advantages over quadrupole-based instru-ments. High ion transmission combined with a lowbackground signal provides superior detection limits.


4 3

Table 7.Yttrium concentration (µg g-1) in rock reference materials by HR-ICP-MS and XRF

Published ICPMS/ XRF/HR-ICP-MS XRF (a) XRF (b) mean XRF values published published

BCR-1 33.6 33.9 33 33.5 38 0.884 0.88

BHVO-1 24.5 24.9 24.3 24.6 27.6 0.888 0.891

AGV-1 18.7 17.9 17.6 17.8 20 0.935 0.888

DNC-1 15.9 16.6 16.3 16.5 18 0.883 0.914

BIR-1 14.1 14.4 14.2 14.3 15.5 0.912 0.923

Tafahi 8 7.9 8 8 9.11 0.878 0.873

UB-N 2.4 2.4 2.6 2.5 2.5 0.96 1

PCC-1 0.077 < 1 < 1 < 1 0.087 0.885

DTS-1 0.036 < 1 < 1 < 1 0.04 0.900

mean = 0.903 0.91

RSD = 0.022 0.031

XRF (a) 2000 µg g-1 Y calibration standards (mixed powders, Y2O3 with pure quartz).

XRF (b) 0, 50, 100, 200 µg g-1 Y calibration standards (mixed Y solution with pure quartz).

HR-ICP-MS sensitivity is typically 100,000 - 150,000counts s-1 per ng ml-1 In. Four quadrupole ICP-MSgroups (Pin and Joannon 1997, Shinotsuka et al. 1996,Eggins et al . 1997, Niu and Batiza 1997) quotesensitivities ranging from 8,000 to 35,000 counts s-1

per ng ml-1 In. It should be noted, however, that sensiti-vity for modern HR-ICP-MS and quadrupole systemshas increased substantially since this paper was initiallyprepared. With HR-ICP-MS, reagent blanks are oftenthe limiting factor with regard to low level analysis.Although excellent detection limits have often beenobtained by quadrupole ICP-MS (Ionov et al. 1992,Eggins et al. 1997), such values are more easily obtainedby HR-ICP-MS. Without a clean room or specific dedi-cation of the instrument to continued low level work,the increased sensitivity of HR-ICP-MS enabled abun-dances of REE in the ultramafic rocks PCC-1 and DTS-1,which have REE contents in the range of 0.8-50 ng g-1

(0.01-0.1x chondrite), to be accurately and preciselyquantified. No sample preconcentration was necessary,reducing the risk of contamination and analyte loss,while saving preparation time and associated costs.

Low oxide formation was also found when using thisHR-ICP-MS instrument. Oxide corrections were generallyonly required for Pr and Ce on Gd, and Ba on Eu.When observed, oxide and hydroxide interferencescould often not be separated from the analytes ofinterest with the higher resolutions available usingHR-ICP-MS. Medium resolution (m/∆m = 3000), however,was found to be necessary for the accurate quantifica-tion of Sc. Using this resolution, 29Si16O and 28Si16O1Hinterferences could be identified as the major causes forimprecise Sc results using quadrupole instruments. TheseSi-based interferences were completely resolved from45Sc when using high resolution. Depending on thedissolution procedure, Sc determinations using conven-tional ICP-MS with pure element calibration solutionsmay be expected to give too high Sc results in rocksamples, while the use of rock reference materials forcalibration may give, apparently, more accurate results.

The successful use of synthet ic (pure element)calibration solutions and a single internal standard(indium) was demonstrated using HR-ICP-MS, providedrock samples were dissolved and diluted by at leasta factor of 1000. Indium (115In) was found to be asatisfactory internal standard in comparison with a mix of84Sr, 169Tm, 175Lu, 185Re and 209Bi for the analysis of 45Scthrough to 238U, providing a true simplified procedurewithout the need to resort to more complicated internalstandard combinations (e.g. Eggins et al. 1997).

Although this study only reports Sc, Y and REE data,typically thirty two elements are often determined byHR-ICP-MS during each analysis in our laboratory. It isplanned to report findings on other elements in duecourse.

Acknowledgements

The fo l low ing are thanked fo r the i r va r iouscontributions to this study: Joe Stolz for early adviceand preparat ion of a number of rock referencema t e r i a l s o l u t i o n s , J ohn Foden (Un i v e r s i t y o fAdelaide) for prov iding 84Sr solu t ion and SteveEggins (Australian National University, Canberra) forthe donation of Tafahi basalt powder. The facilitieso f the Cent ra l Sc ience Laboratory, Univers i t y o fTasmania, are gratefully acknowledged. This projectwas suppor ted by Aus t ra l ian Research Counc i lgrants and industry research funding to the Schoolof Earth Sciences and Centre for Ore Deposit andExploration Studies, University of Tasmania.

References

Balaram V. (1996)Recent trends in the instrumental analysis of rare earthelements in geological and industrial materials.Trends in Analytical Chemistry, 15, 475-485.

Balaram V., Ramesh S.L. and Anjaiah K.V. (1996)New trace element and REE data in thirteen GSF referencesamples by ICP-MS. Geostandards Newsletter, 20, 71-78.

Barrat J.A., Keller F., Amossé J., Taylor R.N., Nesbitt R.W. and Hirata T. (1996)Determination of rare earth elements in sixteen silicatereference samples by ICP-MS after Tm addition and ionexchange separation. Geostandards Newsletter, 20,133-139.

Bau M. (1996)Controls on the fractionation of isovalent trace elementsin magmatic and aqueous systems: Evidence from Y/Ho,Zr/Hf, and lanthanide tetrad effect. Contributions toMineralogy and Petrology, 123, 323-333.

Becker J.S. and Dietze H.-J. (1997)Double focusing sector field inductively coupled plasma-mass spectrometry for highly sensitive multi-element andisotopic analysis. Journal of Analytical AtomicSpectrometry, 12, 881-889.

Bock R. (1979)A handbook of decomposition methods in analytical chemistry. Blackie (Glasgow and London), 60-61.

Bradshaw N., Hall E.F.H. and Sanderson N.E. (1989)Inductively coupled plasma as an ion source for high-resolution mass spectrometry. Journal of AnalyticalAtomic Spectrometry, 4, 801-803.

4 4


references

Chao T.T. and Sanzolone R.F. (1992)Decomposition techniques. Journal of GeochemicalExploration, 44, 65-106.

Dulski P. (1994)Interferences of oxide, hydroxide and chloride analytespecies in the determination of rare earth elements ingeological samples by inductively coupled plasma-mass spectrometry. Fresenius’ Journal of AnalyticalChemistry, 350, 194-203.

Eggins S.M., Woodhead J.D., Kinsley L.P.J.,Mortimer G.E., Sylvester P., McCulloch M.T., Hergt J.M. and Handler M.R. (1997)A simple method for the precise determination of ≥ 40trace elements in geological samples by ICP-MS usingenriched isotope internal standardisation. ChemicalGeology, 134, 311-326.

Evans H.E. and Giglio J.J. (1993)Interferences in inductively coupled plasma-mass spectrometry. A review. Journal of Analytical AtomicSpectrometry, 8, 1-18.

Feldmann I., Tittes W., Jakubowski N., Stuewer D.and Giessmann U. (1994)Performance characteristics of inductively coupled plasma-mass spectrometry with high mass resolution.Journal of Analytical Atomic Spectrometry, 9, 1007-1014.

Garbe-Schönberg C.-D. (1993)Simultaneous determination of thirty seven trace elementsin twenty eight international rock standards by ICP-MS.Geostandards Newsletter, 17, 81-97.

Govindaraju K. (1994)1994 Compilation of working values and sample description for 383 geostandards. GeostandardsNewsletter, 18 (Special Issue), 158pp.

Heinrichs H. and Herrman A.G. (1990)Praktikum der analytischen Geochemie. Springer Verlag,Berlin, 669pp.

Hollocher K., Fakhry A. and Ruiz J. (1995)Trace element determinations for USGS basalt BHVO-1and NIST standard reference materials 278, 688 and694 by inductively coupled plasma-mass spectrometry.Geostandards Newsletter, 19, 35-40.

Ionov D.A., Savoyant L. and Dupuy C. (1992)Application of the ICP-MS technique to trace elementanalysis of peridotites and their minerals. GeostandardsNewsletter, 16, 311-315.

Jarvis I. (1992)Sample preparation for ICP-MS. In: Jarvis K.E., Gray A.L.and Houk R.S. (eds.), Handbook of inductively coupledplasma-mass spectrometry. Blackie, (Glasgow) 172-224.

Jarvis K.E. and Jarvis I. (1988)Determination of the rare earth elements and yttrium inthirty seven international silicate reference materials byinductively coupled plasma-atomic emission spectrometry.Geostandards Newsletter, 12, 1-12.

Jarvis K.E. (1988)Inductively coupled plasma-mass spectrometry: a newtechnique for the rapid or ultra-trace level determinationof the rare earth elements in geological materials.Chemical Geology, 68, 31-39.

Jarvis K.E. (1990)A critical evaluation of two sample preparation techniques for low-level determination of some geologically incompatible elements by inductively coupled plasma-mass spectrometry. Chemical Geology,83, 89-103.

Jenner G.A., Longerich H.P., Jackson S.E. and Fryer B.J. (1990)ICP-MS - A powerful tool for high-precision trace elementanalysis in Earth sciences: Evidence from analysis ofselected USGS reference samples. Chemical Geology,83, 133-148.

Jochum K.P., Rehkämper M. and Seufert H.M. (1994)Trace element analysis of basalt BIR-1 by ID-SSMS, HPLCand LIMS. Geostandards Newsletter, 18, 43-51.

Jochum K.P., Seufert H.M., Spettel B. and Palme H. (1986)The solar-system abundances of Nb, Ta and Y, and therelative abundances of refractory lithophile elements indifferentiated planetary bodies. Geochimica etCosmochimica Acta, 50, 1173-1183.

Lichte F.E., Meier A.L. and Crock J.G. (1987)Determination of the rare earth elements in geologicalmaterials by inductively coupled plasma-mass spectrometry. Analytical Chemistry, 59, 1150-1157.

Longerich H.P. (1993)Oxychlorine ions in inductively coupled plasma-massspectrometry. Effect of Cl speciation as Cl- and ClO4

-.Journal of Analytical Atomic Spectrometry, 8, 439-444.

Longerich H.P., Jenner G.A., Fryer B.J. and Jackson S.E. (1990)Inductively coupled plasma-mass spectrometric analysisof geological samples: A critical evaluation based oncase studies. Chemical Geology, 83, 105-118.

Makishima A. and Nakamura E. (1997)Suppression of matrix effects in ICP-MS by high poweroperation of ICP: Application to precise determination ofRb, Sr, Y, Cs, Ba, REE, Pb, Th and U at ng g-1 levels in milligram silicate samples. Geostandards Newsletter: TheJournal of Geostandards and Geoanalysis, 21, 307-319.

McGinnis C.E., Jain J.C. and Neal C.R. (1997)Characterisation of memory effects and development ofan effective wash protocol for the measurement of petrogenetically critical trace elements in geologicalsamples by ICPMS. Geostandards Newsletter: TheJournal of Geostandards and Geoanalysis, 21, 289-305.

Moens L., Vanhaecke F., Riondato J. and Dams R. (1995)Some figures-of-merit of a new double focusing inductively coupled plasma-mass spectrometer. Journalof Analytical Atomic Spectrometry, 10, 569-574.


4 5

references

Morita M., Ito H., Uehiro T. and Otsuka K. (1989)High resolution mass spectrometry with inductively coupled argon plasma ionisation source. AnalyticalSciences, 5, 609-610.

Münker C. (1997)Geochemical and isotopic systematics of the CambrianDevil River Volcanics in the Takaka Terrane, NewZealand. PhD thesis, University of Göttingen, Germany.

Münker C. (1998)Nb/Ta fractionation in a Cambrian arc/back arc system,New Zealand: Source constraints and application of refined ICP-MS techniques. Chemical Geology, 144, 23-45.

Niu Y. and Batiza R. (1997)Trace element evidence from seamounts for recycledocean crust in the Eastern Pacific mantle. Earth andPlanetary Science Letters, 148, 471-483.

Norman M.D., Griffin W.L., Pearson N.J., Garcia M.O. and O’Reilly S.Y. (1998)Quantitative analysis of trace element abundances inglasses and minerals: A comparison of laser ablationinductively coupled plasma-mass spectrometry, solutioninductively coupled plasma-mass spectrometry, protonmicroprobe and electron microprobe data. Journal ofAnalytical Atomic Spectrometry, 13, 477-482.

Parkinson I.J. and Pearce J.A. (1998)Peridotites from the Izu-Bonin-Mariana forearc (ODP Leg125): Evidence for mantle melting and melt-mantle interaction in a supra-subduction zone setting. Journal ofPetrology, 39, 1577-1688.

Pin C. and Joannon S. (1997)Low level analysis of lanthanides in eleven silicate rockreference samples by ICP-MS after group separationusing cation-exchange chromatography. GeostandardsNewsletter: The Journal of Geostandards andGeoanalysis, 21, 43-49.

Potts P.J. (1997)Geoanalysis: Past, Present and Future. The Analyst, 122,1179-1186.

Potts P.J., Webb P.C. and Watson J.S. (1990)Zirconium determination by ED-XRF: A critical evaluationof silicate reference materials as calibration standards.Geostandards Newsletter, 14, 127-136.

Robinson P., Higgins N.C. and Jenner G.A. (1986)Determination of rare earth elements, yttrium and scandium in rocks by an ion exchange-X-ray fluorescencetechnique. Chemical Geology, 55, 121-137.

Shinotsuka K., Hidaka H., Ebihara M. andNakahara H. (1996)ICP-MS analysis of geological standard rocks for yttrium,lanthanoids, thorium and uranium. Analytical Sciences,12, 917-922.

Sylvester P.J. and McCandless T.E. (1997)Guest editors in special issue on LA-ICP-MS.Geostandards Newsletter: The Journal of Geostandardsand Geoanalysis, 21, 173-305.

Totland M., Jarvis I. and Jarvis K.E. (1992)An assessment of dissolution techniques for the analysisof geological samples by plasma spectrometry. ChemicalGeology, 95, 35-62.

Townsend A.T., Miller K.A., McLean S. and Aldous S. (1998)The determination of copper, zinc, cadmium and lead inurine by high resolution ICP-MS. Journal of AnalyticalAtomic Spectrometry, 13, 1213-1219.

Yoshida S., Muramatsu Y., Tagami K. and Uchida S. (1996)Determination of major and trace elements in Japaneserock reference samples by ICP-MS. International Journalof Environmental Analytical Chemistry, 63, 195-206.

4 6


Determination of Scandium, Yttrium and Rare Earth Elements in Rocks by High Resolution Inductively...

Documents

Transcript of Determination of Scandium, Yttrium and Rare Earth Elements in Rocks by High Resolution Inductively...