GEMOC ARC National Key Centre

This web page is designed to document some of the analytical methods currently in use in the GEMOC Geochemical Analysis Unit, and to provide updates on technological developments (see also the Technology Development section of the GEMOC Annual Report). References are given to published examples where relevant.

Contents:

1. Sulfur analysis in rock samples

2. Trace element analysis of rock samples (ICPMS)

3. In-situ elemental analysis (LAM-ICPMS)

    3.1 Silicates and oxides

    3.2 PGE analysis of sulfide minerals

4. Whole-rock isotopic analysis

    4.1 Sr-Nd-Hf-Pb analysis

4.1.1 Chemistry

4.1.2 Mass spectrometry (MC-ICPMS)

    4.2 Re-Os analysis

4.2.1 Chemistry

4.2.2 Mass spectrometry (MC-ICPMS)

4.2.3 Results

5. In-situ isotopic analysis

    5.1 Re-Os

    5.2 Rb-Sr

    5.3 Sm-Nd

    5.4 Lu-Hf

    5.5 Pb-Pb

    5.6 U-Pb dating of zircons
 
 

1. Sulfur analysis by high-temperature iodo-titration (HTIT)

This description is taken from the PhD thesis of Olivier Alard (2000).

S is a key element in dealing with siderophile and chalcophile elements in the mantle. However, S is not an easy element to analyse, especially in the range below 50 ppm which is the overall detection limit of XRF, ICP-AES and ICPMS. Unfortunately, many mantle samples have S contents below 50 ppm (e.g. Lorand, 1991, Ionov et al., 1992). I therefore used an induction-furnace iodometry method. This method has been used successfully by Dr. J.P. Lorand and M. M. Gros for the last 10 years at the Museum National d'Histoire Naturelle de Paris, France. This method has been directly imported from the MNHN and reproduced at Macquarie University for the needs of this project. Any reference to the work presented here should include proper acknowledgments to M. M. Gros and Dr. J.P. Lorand.

This method is specifically designed for rocks with low S contents, typically between 5 and 500 ppm. More S-rich samples could be analysed but the analytical uncertainty would increase quite rapidly, and the XRF method, which has a detection limit around 50 to 100 ppm, would be more suitable for such high-S samples. The iodometry method is most useful for samples having S contents lower than 100 ppm (e.g. ultrabasic rocks), but is not suitable for rocks having a significant contents of organic matter, for reasons of chemistry and safety.

The accuracy of this method is about 2 ppm. However, in the special case of S analysis in S-depleted rocks, reproducibility is more dependent on the heterogeneous distribution of sulfur than on the analytical technique (see below).

Outline of the method

The method is based on the oxidation of sulfur in the sample to SO₂ and measurement of the quantity of sulfur released by an iodo-titration reaction. Sulfur vaporisation is achieved by first carefully and thoroughly mixing a given amount of sample powder (usually between 0.5 to 2 g) with the same amount of V₂O₅, which acts as a strong oxidant. The mixed powder is then placed in a ceramic or quartz-glass combustion boat (Figure 2.2) and placed in a quartz glass tube, which fits into a tube furnace set to 955±5 °C. The quartz tube is sealed and a constant flow of N₂ gas is introduced. The N₂ is only used as a carrier and does not act as a chemical reagent. At 955±5°C and atmospheric pressure S is volatile. Then under the action of V₂O₅, the S^2- gas released will be oxidised to SO₂ or SO₃ (eq. 2.1).

eq 2.1

A Cu buffer fills the end of the quartz tube, and this reduces any SO₃ to SO₂

eq 2.2  
 

Figure 2.2: Sketch of the sulfur apparatus

A wash bottle linked to this quartz tube allows the SO₂ carried by the N₂ to bubble through a solution of diluted hydrochloric acid and potassium-iodate. At the other end of the wash bottle a piston burette (Titronics 96, SCHOTT) delivers KIO₃ drop by drop. The use of the piston burette allows an accuracy of 0.01 ml and makes the analysis less user-dependant than a "classical" micro-burette. Starch is added to the wash bottle as a coloured indicator of oxido-reduction. The addition of 1 to 2 drops of KIO₃ colours the solution in the wash bottle a pale purple-blue; the reaction is rapid. Therefore as soon as the combustion boat is placed into the tube and the quartz tube is sealed (within 30s), the SO₂ will be released and the solution discoloured (eq 3a and 3b).

eq 2.3a

eq 2.3b

As soon as the blue colour disappears, more KIO₃ is added until no further discolouration occurs. Then the analysis is completed; this usually takes no longer than 30 minutes.

Note: NaN₃ (a powerful reducing agent) can be added to the HCl + KI + starch solution in order to prevent any "small" explosion related to organic matter in the samples.

Following equations 3a and 3b, the quantity of S in the samples can be calculated as follows:

• 1 mole of KIO₃ (M(KIO₃)=214g) reacts with 3 moles of S (M(S)=32.06), which means that 214g of KIO₃ reacts with 96.18 g of sulfur. 6 electrons are exchanged during this oxidation-reduction reaction as iodine goes from the +4 state in IO₃^- to the -1 state in I^- when the reaction is completed.
• Given q the normality of the KIO₃ solution, then:

1 mil of KIO₃ is equal to µg of S in x grams of sample (eq 2.4)

Assessment of the accuracy of the iodo-titration method

The suitability of this method and its accuracy have been assessed by Lorand and Gros (unpublished report) by comparing S analyses obtained by iodotitration and by XRF (Figure 2.3).

Figure 2.3: Comparison between XRF and the Iodotitration method. Data from Lorand and Gros (unpublished data).

The accuracy of the method is heavily dependent on the accuracy with which the KIO₃ solution is delivered. The accuracy of the method itself is better than ±0.5 ppm, but a more realistic estimate is about 1 to 2 ppm. The detection limit of this method can not be calculated in the usual way (3s of the blank), because the blank is usually less than the smallest quantity analysable (4 to 5 ppm). However the blank is checked at the beginning of each analytical session by running 1 or 2 g of vanadium penta-oxide (the fusing oxidising reagents). Seven analysis as such demonstrate a blank level below 6 ppm.

Replicate analyses of the USGS standard SY-2 (Syenite) gave an average of 121 ± 4 ppm (2s error) over a 4-year period, compared to a recommended value equal to 110±9 ppm (Figure 2.4).

Figure 2.4: Replicate analyses of the rock standard SY-2 (Syenite, CCRMP) using the HTIT facility at the Muséum National d'Histoire Naturelle de Paris (France). Data are from Lorand (1991), Pattou (1995), Luguet (2000), and this study. The XRF value of 110±9 ppm is provided by the CCRMP (Canmet). 

Figure 2.5: Replicate analyses of the rock standard JP-1 (Orogenic Peridotite, Horoman, Japan, JGS)using the GEMOC HTIT analytical facility. * denotes information value rather than certified value.

More recently, replicate analyses at Macquarie University of JP1 (peridotite, JGS) gave an average of 27±1 ppm over a 1 year period (Table 2.4, Figure 2.5). Proposed or information values (Govidaraju, 1994; Ando et al., 1989; Japanese Geological Survey Report) of 30 and 26.9 ppm agree well with our results. However Burnham (1995) obtained an extreme value of 0±8 ppm using a LECO analyser. This report argued that nugget effects on the oxidation of S may produce high variability between analyses. Despite some variations, which may be ascribed to the "nugget effect", we do not observe such discrepancies.

To further document the suitability of this method we have analysed various reference materials having certified or recommended values for S (Table 2.4, Figure 2.6). These reference materials have been chosen to cover the S range reported for mantle peridotites (<5 to 500 ppm).

Table 2.4. S values (± standard deviation) for several international rock standards using HTIT method.

N#: number of analyses performed. All but SY-2 were performed using the GEMOC HTIT facility. SY-2 analyses from the MNHN Lorand and Gros, unpublished data. 1, CCRMP (Canmet) values; 2, Japanese Geological Survey value; 3, Ando et al., 1987; 4, Govindaraju, 1994; 5, USGS value; n.a., not available. V.T, Value Type: R, recommended; P, Proposed; I, information; IR, information range.

Figure 2.6: S contents obtained by the HTIT method during this study vs published values for international rock standards

Data are reported in Table 2.4. Dashed line (1:1 line) represents perfect agreement; light and dark grey fields are within 5% and 10%, respectively, of this line.

Figure 2.6 and Table 2.4 show that our values are in excellent agreement with previously published values obtained by alternative methods. Unfortunately, errors and methods used are difficult to obtain. Therefore our precision cannot be truly compared to other methods. The data for RGM-1 (Rhyolite, USGS) and BHVO-2 (basalt, USGS) show discrepancies of more than 10% from previously reported values. For RGM-1 our value is significantly lower (10±3, N#=2) than the literature value of about 54 ppm. However, this RGM-1 value is only a proposed value (Govindaraju, 1994), and its validity unfortunately cannot be assessed. The BHVO-2 (basalt, USGS) does not yet have a recommended S value; I therefore assumed that it is roughly similar to BHVO-1 (S=102, proposed value, Govindaraju, 1994, Information value, USGS). However, I measured a S content of ca. 132 ppm (±10%, N#=1) for BHVO-2.

The otherwise excellent agreement between HTIT values and published values as well as the long term accuracy demonstrated by replicate analyses of SY-2 and JP-1 indicate that this method produces reliable results.

References

Burnham, M.O., 1995. The geochemistry of Re and Os in ultramafic rocks from the Pyrenées and Massif Central, France. Ph.D. Thesis, Open university, 248 pp.

Govidaraju, K., 1994. 1994 compilation of working values and samples description for 383 geostandards. Geostandards Newsletter, 18((special issue)): 158 pp,.

Ionov, D.A., Hoefs, J., Wedepohl, K.H. and Wiechert, U., 1992a. Content and isotopic composition of sulphur in ultramafic xenoliths from central Asia. Earth and Planetary Science Letters, 111: 269-286

Lorand, J.-P., 1991. Sulphide Petrology and Sulphur Geochemistry of Orogenic Lherzolites: A Comparative Study of the Pyrenean Bodies (France) and the Lanzo Massif (Italy). J Pet, "Orogenic Lherzolites" Spec. Issue: 56-77.

Luguet, A., 2000. Pétrologie des sulfures de Fe-Ni-Cu et Géochimie des éléments fortement siédrophiles: etude couplée dans les péridotites abyssales de la zone de fracture Kane (zone MARK, 20-24, Ride médio-atlantique et du site EDUL (49-70°E, ride sud-ouest indienne). Ph.D Thesis, Museum National d'Histoire naturelle de Paris, Paris, France, 355 pp.

2. ICPMS Analysis of rock samples

The description given here is taken from the PhD thesis of Olivier Alard (2000). It is particularly concerned with dissolution methods and their effects on accuracy and precision, and with inter-laboratory comparisons.

Solution ICPMS work carried out in the GEMOC Geochemical Analysis Unit at Macquarie University during this work used two ICPMS instruments: a Perkin Elmer Sciex Elan 6000 and an Agilent HP4500 featuring a shield torch. An Agilent 7500 ICPMS is currently in operation and details of its performance will be posted here later.

Method

REE, HFSE (Nb, Ta, Zr, Hf), LILE ( Cs, Rb , Ba, Th, U, Sr) as well as minor and transition elements (Ni, Cr, Mo, Zn, Cu, V, Ti, Ga) were analysed. Several rock digestion procedures have been used, all performed in teflon beakers using 100 mg of sample. For each batch, USGS and JGS rock standards have been included in order to check the accuracy of the analyses (Table 2.2).

Attack A : HNO₃/HF (1:1) at 120ºC for 24 hours in a closed beaker, followed by an evaporation and then a second HNO₃/HF (1:1) digestion. The mixture is evaporated to incipient dryness and 10 ml of 6N HNO₃ is added. The beaker is tightly capped and heated overnight; again the sample is taken to dryness and then 10 ml of 2% HNO₃ is added. The beaker is capped, heated and held in an ultrasonic bath until the residue is fully dissolved. The solution is then transferred to a bottle and the volume is make up to the final weight; e.g to 100 g for a dilution factor of ca. 1000.

Attack B: This is a perchloric/hydrofluoric attack method developed in Montpellier II University, France (Geofluid laboratory, see . The sample is attacked with HF/HCLO₄ (2.5:1) at 160ºC for 48 hours in a closed beaker. This first step is followed by evaporation to dryness and a second HF/HCLO₄ (2:1) attack is done overnight. Then three evaporations are performed with deceasing quantities of HCLO₄ (1, 0.5, 0,25 ml) and increasing temperature up to 180-190ºC, in order to expel all fluoride salts. Once the sample is dry, it is dissolved in 2% HNO₃ and diluted for ICPMS analysis.

The solutions have been analysed either at GEMOC using Perkin Elmer Sciex Elan 6000 and an Agilent HP4500 or at Montpellier II University using a VG 353 Plasma Quad ICPMS. Drift is corrected using doping elements: In and Bi in Montpellier, ⁶Li, Ru, In and Bi at GEMOC. External standard calibration is used in both laboratories.

Reproducibility, accuracy.

Standards

The accuracy of each analytical technique has been checked by carrying out replicate analyses of standard materials having certified values. Good agreement between our values and recommended or previously published working values is observed for most of the elements and for most of the reference materials analysed during this study when analyses obtained after HNO₃/HF digestion are compared. This conclusion is true for both mafic and ultramafic standard materials. Similar good agreement is obtained with HClO₄ attacks for mafic standards. However, we obtained systematically higher contents for rare earth elements in ultramafic standards when using HClO₄ acid digestions. As discussed above, this is ascribed to the higher recovery achieved with HClO₄ procedures. Good agreement between Montpellier and GEMOC data is observed for ultramafic standards analysed after HClO₄-based digestion. This agreement allows a straightforward and confident comparison of the data obtained during this study in the two different laboratories.

Our analyses of JP-1 in both laboratories show systematic discrepancies with the JGS working values. However because of: (1), the overall good agreement between our values and previously published values for other reference materials having similar matrices; (2), the good agreement between our values and other recently published values (Ionov et al., 1992; Yokoyama et al., 1999), we may suggest that the JGS values need to be reassessed.

Reproducibility issues

Recent work has suggested that HNO₃-HF attack and evaporation gives lower recovery, especially for REE, than can be obtained using a HF-HClO₄ digestion technique (Yokoyama et al., 1999). The formation of fluoride complexes may scavenge and precipitate out of solution some of the REEs. The addition of HClO₄ to the digestion mixture and/or during subsequent evaporations has been long recognised as an efficient way to get rid of these fluoride complexes (Ionov et al., 1992; Blichert-Toft et al., 1997; Yokoyama et al., 1999). The amount of REE deficit for HF/HNO₃ attacks has been estimated in the GEMOC laboratory by repetitively analysing international reference materials using the HNO₃ and HClO₄ digestion procedures (Sharma et al., 2000). Such estimates are essential in order to assess to which extent inter-laboratory and inter-technique data can be compared. The results of this study demonstrate that:

• For ultramafic matrices, REE recovery using HF/HClO₄ digestion and evaporation are higher than for HF/HNO₃ procedure. This is especially significant for the MREE; recovery is 20-40% higher for MREE, while LREE and MREE show an increase of about 5-20% (Figure 2.1). Higher recoveries are also observed for LILEs (Å5-25%). Much higher U recovery is also observed (up to 50%).
• For mafic matrices (ie. basalts: BCR-1, BHVO-1) recovery percentages obtained with the two methods are within error of each other (<5%).

This indicates that there is a matrix effect on either fluoride formation or fluoride breakdown (Yokoyama et al., 1999). However, HClO₄ digestions yield higher blank levels, because of the difficulty of efficiently purifying HClO₄, and the longer evaporation time required by the HClO₄
 
 

Figure 2.1:Comparison of recovery percentage between HClO₄ and HNO₃ digestion procedures.

Two representative experiments are presented here for mafic and ultramafic matrices.

,

Where:  and is the average concentration of an element i obtained from 4 analyses of 2 different powder aliquots of the same material performed with HClO₄ and HNO₃ digestion, respectively.  is the published value for this element in the same reference material. Data from Sharma et al. (2000).

These results suggest that care must be exercised in comparing whole-rock REE data obtained with various chemical procedures. Only large variations should be considered as "real". If sample and standard do not have similar matrix chemistry, and agreement between the analyses and recommended values for standard materials does not ensure the accuracy of the sample analyses.

Detection Limits

Detection limits are calculated as and are reported in Table 2.3. The detection limit reflects the level and especially the variability of the chemistry blank for each element. This blank is the sum of acid blank, teflon ware blank and contamination during the evaporation and manipulation stage. Detection limits for most of the heavy and medium REE in both the Montpellier and the Macquarie laboratories are below 1 ppb. Light REE show slightly higher detection limits, up to Å 10 ppb for La, reflecting a more variable blank level for those elements. However, elements such as Ba, Sr and Pb show much higher detection limits due to a high blank level. Those elements are unfortunately notorious for being common contaminants in all laboratories as they are very abundant elements.

Table 2.3. Detection limits (in ppb) for solution ICPMS runs

?: detection limits obtained at Geofluides laboratory are calculated chemistry (method) blanks for a 5 year period (1996-2000). All Geofluides laboratory dissolutions involve large quantities of HClO₄.*Geochemical Analyses Unit at GEMOC, Macquarie University

Table 2.2: International rock standard trace element values and reference data for solution ICP-MS work.

All reference data are ICPMS data. R1, Eggins et al., (1997) HF/HNO₃ digestion procedure. R2, JGS working values; R3. Ionov et al. (1992) Montpellier University HF/HClO₄ procedure; R4, Garbe-Shönberg (1993). *using a HF/HNO₃ attack; ? using a HF/HClO₄ attack; ?, analytical technique not reported; ?: denotes value in ppm; n.d., not determined.

References

Ionov, D.A., Savoyant, L. and Dupuy, C., 1992. Application of the ICP-MS technique to trace element analysis of peridotites and their minerals. Geostandards Newsletter, 16: 311-315.

3. In-situ trace element analysis (LAM-ICPMS)

The information presented here is extracted from the PhD thesis of Olivier Alard (2000).

GEMOC, in collaboration with New Wave Research and Agilent Technologies, is currently carrying out detailed tests of different laser systems and their performance with the Agilent 7500 ICPMS. Results will be posted here when the tests are completed.

Laser ablation microprobe ICPMS has several advantages:

• High spatial resolution
• Low detection limit
• Rapid analysis
• Wide range of elements (LILE, HFSE, REE, HSE)
• Variety of target materials (Silicates, Carbonates, Oxides, Metal, Sulfides)

Instruments

Basically, the instrument consists of two parts; a laser ablation system in which the sample is ablated and transferred, by a stream of high purity gas, to the second part of the instrument, the quadrupole ICPMS

Two Laser ablation microprobe systems were used during this study:

• an in-house laser ablation line (built by S. E. Jackson),
• a Merchantek LUV266 laser system.

Three quadrupole ICPMS instruments have been used during this study, the Perkin-Elmer Sciex ELAN 5100, the Perkin-Elmer Sciex ELAN 6000 and the Hewlett Packard 4500. Most of the work presented in this thesis has been done with the PE 6000. The various ICPMS instruments were tuned to produce ²⁴⁸ThO/²³²Th<1%. Typical operating conditions for each ICP-MS are reported in Table 2.12. Note that The HP4500 features a shield torch. This system greatly enhances the sensitivity of the instrument as can be seen in Table 2.13, where typical detection limits for each instrument are reported.

Table 2.12: Typical ICP-MS operating conditions for Laser ablation runs