Quantitative trace-element analysis of diamond by laser ablation inductively coupled plasma mass spectrometry

Abstract

Laser ablation microprobe ICP-MS has been used to determine quantitatively the trace-element composition of diamond. Experiments with different synthetic multi-element carbon-based standards, various lasers and a range of instrument conditions have shown that a 266 nm UV laser at 10 Hz provided the best sensitivity, and synthetic oil and a doped cellulose proved most suitable as external standards; ¹³C was used as the internal standard. The precision and accuracy of the method, and the homogeneity of the cellulose multi-element standard, were tested by multiple analyses. Artefacts resulting from polyatomic interferences were quantified by analysis of a pure synthetic diamond. Concentrations of 41 elements were determined for two fibrous diamonds from Jwaneng in Botswana (JWA 110 and JWA 115), which have been analysed previously by instrumental neutron-activation analysis (INAA) and proton microprobe (PIXE). A comparison of these three analytical techniques shows that the use of the cellulose standard produces accurate and precise data for most elements. Typical detection limits for the rare earth elements are 5–20 ppb, and for transition elements <500 ppb. Sodium and Fe have higher detection limits (2–3 ppm). The precision (expressed as % rsd) ranges through ∼10% for concentrations between 1–100 ppm, ∼15% for values between 0.1–1 ppm, ∼30% for 0.01–0.1 ppm and ∼25% for values <0.01 ppm, with the accuracy lying in the same range. The trace-element patterns obtained by this technique may be used for the characterisation of diamond in genetic studies. Further analyses are required to test whether reliable identification of the source locality of the diamonds is possible; if so this may have important forensic applications.

---

1. Introduction

Natural diamonds provide valuable records of processes at high temperatures and pressures in the Earth’s mantle. During crystallisation, diamonds commonly trap small inclusions of silicate, oxide and sulfide minerals, fluids and melts, as well as accepting minor and trace elements such as nitrogen and boron into their crystal lattice.¹ The study of these inclusions and trace elements can yield information about the chemical, isotopic and mineralogical composition of the host rocks and the nature of the fluids from which the diamonds formed.

Chrenko et al.² first reported the presence of fluid inclusions in diamonds, using infrared spectroscopy to identify the presence of water and carbonate in the fibrous coating of an octahedral diamond. Elemental impurity concentrations in Russian diamonds were described by Orlov,³ and several studies^4,5 have measured impurity levels in diamonds from sources in South Africa using techniques such as instrumental neutron activation analysis (INAA). Bibby⁶ presented INAA data for 57 elemental impurities in diamonds from worldwide sources in an attempt to correlate diamond types with colour and source. Schrauder et al.⁷ reported trace element data obtained by INAA for a set of fibrous diamonds that were previously analysed by electron microprobe (EMP) and FTIR (Fourier-transform infrared spectroscopy).⁸ Two of these diamonds have also been analysed for trace elements by PIXE (proton induced X-ray emission) microprobe.⁹ More recently, Resano et al.¹⁰ have used laser ablation microprobe (LAM)-ICP-MS for a qualitative study of trace elements in diamond in an attempt to investigate the possibility of diamond fingerprinting.

This study uses the LAM-ICP-MS technique to determine quantitatively the trace-element composition of diamond using carbon-based multi-element external standards (synthetic oil and cellulose). Such a quantitative approach is necessary for comparative and statistical studies, for the development of genetic models, and to allow meaningful inter-laboratory comparisons.

Laser-ablation microprobe ICP-MS is a useful method for the in situ microanalysis of solids because of the spatial resolution (10–100 μm), sub-ppm detection limits and rapid analysis times (typically ≤ 5 min per point analysis) that can be achieved.¹¹ A wide range of elements, including large ion lithophile elements (LILE), high field strength elements (HFSE) and rare earth elements (REE) can be determined. The technique does not require elaborate sample preparation and is relatively non-destructive; it leaves only a small pit in the diamond surface (pit size: d ∼100 μm, depth ∼60 μm). This increases the possibility of the commercial application of this technique to identify the source deposit of a diamond.

Quantitative determination of the trace-element signatures of diamonds can test whether there are systematic differences in composition between diamonds from different localities and allows the comparison of diamonds of different parageneses from single localities. This information will help to constrain the nature of the fluid or agent from which the diamonds crystallise and promote a better understanding of diamond formation in the Earth’s mantle. Statistical analysis of a large body of such trace-element patterns will also test whether identification of the source locality of the diamonds is possible.

2. Experimental methods

The laser-ablation microprobe ICP-MS system consists of two parts: the laser ablation microprobe sampler and the ICP-MS. This technique requires an external standard for calibration: such a standard is a substance that has known concentrations of the elements of interest. An internal standard, which is usually a major element, is also required to correct for differences in ablation yield and instrumental drift; for diamond, the obvious internal standard is carbon. Different synthetic carbon-based standards, various lasers and a range of instrument conditions were tested during the development of this method over a period of three years.

2.1. Instrumentation

2.1.1. Laser-ablation system. A series of experiments was carried out to determine which laser would provide optimum results for the diamond analysis. These experiments involved the ablation of a diamond sample with the different lasers at varying laser frequencies, spot sizes and energies.

The laser operating parameters and their influence on the ablation process were investigated using:

1. Nd:YAG, 213 nm custom built UV (Quantel Brilliant Laser)

2. Nd:YAG, 266 nm custom built UV (Quantel Brilliant laser)

3. Nd:YAG, 266 nm custom built UV (Continuum Surelite I 20 laser)

4. Nd:YAG, Merchantek “LUV266” 266 nm laser ablation system

While no highly systematic comparison of the laser systems was performed, it was found that the 266 nm UV Brilliant laser provided the best signal peak to background ratio for carbon (the internal standard) for diamond analysis. This is presumed to be related to the high energy density attainable with this laser system compared with the 213 nm and the commercial 266 nm laser systems. A high energy density appears to be more important than using shorter wavelengths in producing the large ablation yields required for diamond analysis. This laser was used for the diamond analyses carried out with the cellulose calibration described below. The oil calibration analyses described below were performed prior to obtaining the 266 nm UV Brilliant laser, using the Merchantek LUV 266 nm laser at frequencies of 10 and 20 Hz.

The laser-beam delivery system is similar to that described by Norman et al.¹¹ with the following modifications. The Continuum Surelite I 20 laser has been replaced with a Quantel Brilliant Nd YAG laser with frequency doubling, quadrupling and quintupling harmonic generators, allowing beams of 266 nm or 213 nm, with maximum outputs of ca. 30 mJ and 10 mJ per pulse, respectively. Separation of the required wavelength from residual unconverted wavelengths is achieved via a wavelength separation package consisting of two 45° mirrors in series, with dielectric coating optimised to reflect efficiently a single wavelength. The beam is then passed through an optical attenuator consisting of a half-wave plate and a high power UV polariser consisting of MgF₂ and CaF₂ triangular prisms, as described by Jeffries et al.¹² The MgF₂ polariser, required to accommodate the 213 nm wavelength, replaced a Glan calcite polariser used in the original design. As the optical attenuation system itself produced an ∼35% energy loss, the attenuation optics were removed for some analyses in order to increase energy on the samples. The laser energy, calculated as fluence on the sample ranged from ∼0.19 J cm⁻² (oil) and ∼9.4 J cm⁻² (cellulose) to 38–76.5 J cm⁻² for the diamond analyses.

2.1.2. ICP-MS. The oil calibration analyses discussed below were performed using an ELAN 6000 ICP-MS. The diamond analyses using the cellulose calibration were carried out on an Agilent 7500 ICP-MS. Details of the operating conditions are given in Table 1.

Table 1 Instrument operating conditions

Instrument	Agilent 7500	Elan 6000
Calibration	Cellulose method	Oil method
Laser source	Quantel Brilliant	Merchantek LUV
Wavelength	266 nm	266 nm
Q-switch delay	176 μs	200 μs
Pulse width	3.75 ns	3–5 ns
Frequency	10 Hz	10 Hz
		20 Hz
Power
Energy	1.0–8.0 mJ	0.25–4.0 mJ
Energy density	∼13–102 J cm^−2	∼3–51 J cm^−2
Spot size	∼100 μm	∼50–150 μm
ICP-MS
Plasma conditions
RF Power	1440 W	1300 W
Plasma gas flow	16 L min^−1	16 L min^−1
Auxiliary gas flow	1.0 L min^−1	1.0 L min^−1
Nebuliser gas flow	Ar/He = 1.5	Ar/He = 1.5
	Ar gas ∼0.98 L min^−1	Ar gas ∼0.98 L min^−1
	He gas ∼0.64 L min^−1	He gas ∼0.64 L min^−1
Cones	Pt cones	Pt cones
Data Collection
Scanning mode	Time resolved	Time resolved
Dwell time	10–30 ms	10 ms

---

2.2. Data reduction

Data reduction was carried out using the in-house GLITTER software (www.els.mq.edu.au/GEMOC/), which displays time-resolved signals for each element in a way that highlights spikes, inclusions and other heterogeneities, and allows selection of the most stable part of the signal for integration. It also enables the user to filter any “spikes” from the portion of the signal selected for integration.

2.3. Calibration and standards

Obtaining quantitative results using laser ablation microprobe ICP-MS requires reproducible inter-element sensitivity, allowing the calibration of relative element sensitivity against an external standard and normalisation of each analysis to an internal standard to correct for differences in the amount of material that is ablated and transported during an individual analysis.

2.3.1. Internal standard. For most common rock-forming minerals major elements such as Ca, Mg, or Al are used as internal standards and these are determined independently (e.g., by electron microprobe). For diamond analysis, the total trace-element budget is significantly <0.1 wt%; hence, 100% C (analysed as ¹³C) was used as the internal standard.

2.3.2. External standards. There are no certified values for carbon in any of the conventionally used and commercially available solid microbeam standards such as the NIST 600 series glasses. Therefore, a new standardisation technique has been developed, involving the use of a synthetic carbon-based standard; both synthetic oil and synthetic cellulose-based standards have been used.
2.3.2.1. Oil calibration. The synthetic oil standard used is the “Conostan Custom Blend Multi-Element standard S-21 (+ Bi + Y + Sr)”, which is a hydrocarbon oil, doped at a variety of concentrations with each of 24 elements (Ag, Al, B, Ba, Bi, Ca, Cd, Cr, Cu, Fe, Mg, Mn, Mo, Na, Ni, P, Pb, Si, Sn, Sr, Ti, V, Y, Zn). A 300 ppm oil was used in this study. The oil contains approximately 85% carbon,¹³ and can be diluted to lower trace element concentrations by mixing with “Conostan 75 Base oil”, which is the same oil used in the multi-element standard except that it contains <0.02 ppm of the above-mentioned elements.

The oil was used undiluted and sealed in glass capillary tubes of approximately 1.5 mm bore, which were pre-ablated with a hole of diameter slightly larger than the selected ablation spot size. The oil was then directly ablated through this hole using very low pulse energy (∼0.3 mJ).

Analyses were performed in two stages: first, the oil was used as an external standard with carbon as the internal standard to measure Mg and other trace elements. Then the measured concentration of Mg was used as the internal standard to determine elements not present in the oil, with the NIST SRM 612 or NIST SRM 610 glass as the external standard. Molecular interferences related to the measurement of Mg are discussed in section 2.4.4.

The accuracy and efficiency of this calibration technique was tested by analysing the NIST SRM 1632b. This is a powdered bituminous coal, which has known concentrations (certified values for selected elements and information values for the remainder, http://www.nist.gov/) of up to 40 minor and trace elements. The powder was pressed into a pellet for use in laser ablation. A single-path continuous raster was performed on the bitumen pellet by moving the sample stage to ensure a steady, constant signal for about 120 s. Raster (line scanning) analysis of NIST 1632B bitumen produced significantly more accurate analyses than spot analyses; this result reflects the heterogeneous composition of the bitumen. Based on correlation of signals during time-resolved analysis, this material appears to be composed of several mineral phases, including bitumen, one or more aluminosilicate phases and an iron rich sulfide. Rastering the sample during analysis produces more accurate analyses for two reasons: (1) a larger sample volume is analysed, reducing the effects of sample heterogeneity; (2) marked elemental fractionation occurs during spot analysis. This is believed to be related to differentially changing ablation efficiencies of the mineral phases with decreasing energy density as the ablation pit propagates downward. Integration of a 10 s signal interval from the beginning of a single spot analysis produces concentrations that are up to more than 3 times greater than the concentrations generated using the final 10 s of a 90 s spot ablation. Element concentrations were extracted using C as the internal standard and the synthetic oil as the external standard. The SRM NIST 612 glass was also used as external standard with Mg (determined from the oil calibration) as the internal standard. Significant heterogeneity was observed for most elements from area to area on the pressed pellet: the % rsd for the SRM 1632b using C as the internal standard ranges from 35% to 55% for Na, Si, Ca, Fe, Zn and Pb, and from 18% to 27% for the rest of the elements shown in Fig. 1A. The discrepancies from certified values for some of the first set of elements are attributed to sample heterogeneity and the different ablation efficiencies of the mineral components of the ‘polyphase’ bitumen. The certified values for this type of heterogeneous, ‘poly-phase’ material represent the bulk average values, and hence the SRM1632b is not considered appropriate as a primary external microbeam standard.

Despite this heterogeneity, Fig. 1A shows a good correlation between the analyses and the recommended values for most elements, and Fig. 1B shows a good correlation between analyses of the SRM 1632b performed using the SRM NIST 612 glass (²⁵Mg as internal standard) and using the oil calibration (¹³C as internal standard).

Fig. 1 (A) Comparison of trace-element data for NIST SRM 1632b Bituminous coal (n = 19) analysed using the oil standard (C as the internal standard), with the reference values (information values only for Si, Cr, V and Mo, the values for the rest of the elements are certified). (B) Comparison of trace-element data for NIST SRM 1632b Bituminous Coal (n = 19) analysed using the oil standard (C as the internal standard) and using NIST 612 SRM as external standard, with Mg (determined from the oil calibration) as the internal standard.

One of the disadvantages of using the oil is that it volatilises even when it is not being ablated, which adds to the overall carbon background. In addition, preparation of the oil-filled capillaries is time-consuming. The oil was also found to degrade over time, resulting in some of the elements precipitating out of the oil.

2.3.2.2. Cellulose calibration. Cellulose (–C₆H₁₂O₅–) is a natural polymer of β-D-glucose. The sugar units are linked when water is eliminated, by combining the –OH group and H. It contains approximately 44% carbon by stoichiometric calculation. The cellulose standard used here was prepared from cellulose powder that had been doped with a multi-element solution, a technique adapted from the method outlined by Witte.¹⁴
Preparation. Approximately 10 g of Whatman cellulose powder CF11 was weighed out and washed with an excess of 25% HNO₃ to remove inorganic contaminants. The cellulose was put into a Teflon beaker with the nitric acid and stirred with a magnetic stirrer for three hours. The cellulose was then filtered through a 0.45 μm cellulose acetate filter paper with a Büchner funnel under suction, washed thoroughly and repeatedly with Milli-Q water, and air-dried in a fume cupboard (this took up to 15 hours). The dry cellulose was then crushed in an agate mortar and pestle, until the fibres were broken and separated, and transferred to a 50 mL centrifuge tube.

The multi-element solution was made up to contain 49 elements at a concentration of ∼20 ppm (except for: Na, Mg, Cr ∼200 ppm; B ∼50 ppm; Si, Fe, K ∼100 ppm; S, Ca ∼1000 ppm) and a known amount was then added to 2 g of the washed and crushed cellulose powder. 3–5 ml of Milli-Q water was added to wash the cellulose powder from the sides of the beaker and to facilitate stirring. Care was taken not to add a large amount of water as excess solution over that required to saturate the cellulose may cause problems of grain boundary chemical heterogeneity when the cellulose is dried. The solution was stirred for 3 h with a magnetic stirring rod in a 100 ml Teflon beaker, and then allowed to air dry in a fume cupboard at least overnight (the solution may also be placed on a hot plate at 40 °C or put under a heat lamp to aid drying). When the resulting ‘cake’ was thoroughly dried, it was crushed in an agate mortar and pestle to produce a homogeneous powder. The ‘doped’ cellulose powder was then pressed into a pellet using a stainless steel press, similar to that used in the preparation of XRF pellets. Weighing paper was used on the piston to prevent contamination from the stainless steel.

Solution analysis. Metal concentrations in the spiked cellulose were determined through digestion of a 100 mg piece of the pellet. The cellulose was digested in 0.2 ml H₂O₂ and 2 ml HNO₃ in a Savillex screw-top Teflon beaker at 100–120 °C overnight. The beakers were allowed to cool for a few hours before approximately 5 ml of 2% HNO₃ were added. The samples were ultrasonicated for 30 min to 1 hr before being placed on a hotplate at 100–120 °C for up to 1 hr. Once cool, they were diluted to 100 ml using 2% HNO₃.

Duplicate digestions were performed on both spiked and unspiked cellulose and internal standards ⁶Li, ¹⁰³Rh, ¹⁶⁹Tm, ²⁰⁹Bi were added. The samples were analysed by ICP-MS against an independent multi-element standard solution. The calibration standards were run between every 5–8 unknowns. Internal standardisation was used to make corrections for drift and matrix effects for each element, using linear interpolations with mass between the bracketting internal standards. Drift corrections were also applied with time between the calibration standards.

The measured results were within 10% of the calculated values for the cellulose (actual element concentrations added to the cellulose powder). Elements such as Si and Ca show a larger error, possibly due to loss or contamination, respectively, during the preparation process. The mean values for each element were accepted as the concentrations of the elements in the cellulose standard (Table S1) for laser calibration purposes.

Unlike the oil calibration method, using cellulose to calibrate laser analyses involves only one stage. Any number or combination of elements may be added to the cellulose, whereas the oil standard is restricted to the elements added by the manufacturer. The cellulose standard gives better external precision (e.g., % rsd for Na by the oil method = 20.3, and for the cellulose method = 8.5; for Al = 60.9 and 8.8, respectively, etc.) and lower carbon backgrounds than the oil standard. It is also much simpler to use since the cellulose pellet, once prepared, is ready to ablate.

As both the cellulose and diamond have predominantly carbon-based matrices, any resulting matrix-related interferences are considered to be similar for both materials.

Reproducibility/precision. Multiple analyses of the cellulose multi-element standard were carried out to test the homogeneity of the standard. A spot size of approximately 100 μm was used in each case and the analyses were performed using the SRM NIST 612 glass as the external standard and Sr as the internal standard, treating the cellulose standard as an unknown (Table S2). Strontium was chosen for several reasons: its concentration (in the cellulose and the diamonds) is well above the detection limit, it is homogeneously distributed in the cellulose and it does not show large fractionation relative to carbon. The cross-analysis of the cellulose using the NIST 612 glass shows good precision (1σ for 317 analyses/spots is <2 ppm) and accuracy (compared with a solution analysis of the cellulose standard) for most elements. However, despite the elaborate homogenisation procedures, several elements (Si, K, Ca, Cr, Fe, Zn, Ba, W) still show significant lateral heterogeneity in the cellulose pellets (typically 15–50%, up to 60% for Ba). The element column in Table S2 indicates which isotope was selected for determination of each element, based on its susceptibility to mass interferences.

2.4. Operating conditions

Table 1 shows the typical laser and ICP-MS instrument operating conditions. The torch position, gas flows and ion lens settings of the Agilent 7500 ICP-MS were tuned daily on signals produced by ablation of NIST SRM 612 (or 610) glass, paying particular attention to achieving comparable sensitivity at both ends of the mass range, while maintaining low oxide levels (ThO/Th <1%). Then, by ablation of diamond, the C peak/background ratio, which was variable from session to session, was checked. A minimum ratio of 2.5 was accepted. Some details of the operating conditions are explained below.

2.4.1. Gas configuration. Ar and/or He gases were flushed through the sample cell to carry the ablated sample to the ICP-MS. The two gases can be used either separately or as a mixture with the help of a two-way gas valve and a T junction. All ablations were performed with only He gas flowing through the sample cell and Ar mixed in before the ablated material entered the ICP-MS. Helium reduces the amount of ablated material condensed back onto the sample surface and thus improves signal intensities.^15,16

2.4.2. Laser energy. Under normal circumstances it is best for the external calibration standard and the unknown sample to be ablated under as similar conditions as possible. In the case of the oil standard, ablation of the liquid at high energy may produce splattering of the oil, resulting in contamination of all surfaces inside the sample cell. Therefore, very low energies between 0.20–0.30 mJ were used for the oil analyses.

It is not possible to use the same laser energy for the cellulose standard and the diamond sample, because the physical properties of the two materials are very different. Diamond has a much higher ablation-energy threshold than the cellulose. Typically, the cellulose was ablated at an energy of approximately 0.8 mJ per pulse, whereas the diamonds were ablated at energies ranging from 2.7 to 8 mJ per pulse. The use of an internal standard, in this case carbon, corrects for fluctuations in the pulse energy during the ablation event and differences in the ablation yield between sample and standard without degradation of the accuracy of the analytical results.¹⁷

Using larger spot sizes and/or higher energies did not produce significantly greater build-up of deposits on the cones than we experience during routine silicate analysis under conditions similar to those used here for the SRM NIST glass.

2.4.3. Backgrounds. As ¹³C is used as an internal standard, the data quality depends on the peak to background ratio of carbon: a better peak to background ratio generally improves the detection limits and accuracy. The peak to background ratio for carbon can be enhanced by increasing the carbon signal, i.e., increasing the amount of ablated material by increasing spot size or energy, or decreasing the background signal. Some of the factors influencing the background are discussed below.
Air entrainment. For “dry” plasmas, the elemental material that is contributed to the plasma from air entrainment becomes significant. Gray¹⁸ has reported the use of a flared torch bonnet, which fits over the ICP torch in a manner similar to traditional bonnets. This bonnet was flared to match the angle of the sampler and had an outer diameter equal to that of the sampler cone. Ince et al.¹⁹ showed that a significant reduction in ICP-MS signal noise was obtained when using this bonnet. The noise reduction was related to a decrease in the interaction of the plasma with atmospheric gases, i.e., the reduction of air entrainment into the plasma. Use of the flared bonnet reduces air entrainment and results in lower polyatomic backgrounds and enhanced sensitivity for the lighter elements.

Experiments were carried out using a similar flared bonnet in an ongoing effort to reduce the effect of the air entrainment. Kr gas was introduced externally into the torch box at an approximate rate of 1.2 L min⁻¹. Kr was chosen as it is easy to detect, is not a major component of air, and the background levels are low. Kr levels were measured using both a conventional bonnet and a flared bonnet. It was found the Kr gas entrainment was reduced by approximately 80% with the flared bonnet (4.4 × 10⁶ signal cps for ⁸³Kr using the normal bonnet, 0.64 × 10⁶ signal cps for ⁸³Kr using the flared bonnet). Unfortunately, the C background showed a reduction of only ∼15%. Further experiments are required to improve this result: tests with variable length of the flared bonnet are needed to reduce the background for carbon and other air species.

Tubing. Experiments were carried out to determine the effect of using varying lengths of sample transfer tubing and of by-passing the ablation cell (i.e., no tubing except that inside the instrument) on the C background. Removing the sample transfer tubing and sample cell resulted in a drop from ca. 600 000 cps to 300 000 cps in the ¹³C background signal. This result suggests that there may be a build-up of ablated material on the inner walls of the tubing used to transport material from the ablation cell to the plasma. Plasticisers used in the preparation of certain kinds of tubings also may contribute to the carbon background. Numerous trials involving close monitoring of the carbon (¹³C) signal/background and the signals for other elements such as I, Br and Cl were carried out using different types of tubing. Tygon 2075, which is free of plasticisers, has extremely low outgassing properties and inhibits particulate build-up due to its smoother inner surface, was found to be most suitable. Changing to Tygon tubing resulted in ca. a 10% decrease in ¹³C background, 25% decrease in ³⁵Cl background and 50% decrease in ¹²⁷I background. No significant change was observed in the ⁷⁹Br background signal. Care was also taken to limit the use of carbon-emitting substances (such as epoxy, plasticine and ethanol) for cleaning or mounting of samples in the sample cell.

2.4.4. Signal selection. A typical analysis lasts for 150 s, including gas backgrounds collected for the initial 50 s prior to beginning ablation of the sample. The first ca. 10 s of signal after ablation started was not integrated because some elements (e.g., ³⁵Cl, ³⁴S, ²⁴Mg) show fractionation relative to the internal standard (¹³C) during this interval. This procedure also avoids including any signal due to surface contaminants that might still be present. The subsequent portion of the signal, which is flatter and more stable, was integrated for concentration calculations.

2.4.5. Elemental fractionation. Fractionation is defined as a time-dependent variation in intensity ratios for different elements during ablation of a homogeneous sample. Non-stoichiometric ablation or elemental fractionation describes the non-sample-related change in the element intensities with time.²⁰ Analytical results are affected if the sample and the reference material show significantly different fractionation behaviour.²¹ It has been widely accepted that fractionation can occur during the ablation process, i.e., at the ablation site, and can be influenced by laser parameters,^22,23 the transportation process and the excitation process.²⁴

Fryer et al.²² concluded that lithophile elements generally behave similarly to each other and many chalcophile elements show similar behaviour during ablation, whereas siderophile elements show intermediate fractionation behaviour relative to Ca. In the present case, C has been used as the internal standard and the behaviour of elements relative to carbon may not necessarily be the same as those relative to Ca or other elements commonly used as internal standards.

Fryer et al.²² also noted that experiments with a variety of natural minerals produce relative fractionations similar to those of the NIST glass, suggesting that fractionation between elements is substantially independent of sample matrix. No single physical or chemical property of the elements correlates with, or explains, the pattern of fractionation factors.

An index of time-dependent fractionation can be calculated,²² by dividing a time-resolved signal into two equal time segments and comparing the element intensity ratios between these. This calculation was performed on the cellulose with the fractionation index (FI) calculated relative to C, with a repetition rate of 10 Hz and ablation time of 30 s (Fig. S1). The average FI is ca. 0.93; any calibration errors associated with this degree of fractionation are within the range of error (ca. ± 15%) for most diamond analyses, and we conclude that fractionation of elements with respect to ¹³C during ablation is not a major source of error. Zn and Cd show large negatively anomalous fractionation factors, which is in contrast to large positive fractionation factors in silicates.²² The negative anomalies reported in this study are not understood, but are presumed to be related to different behaviour of these elements during ablation of carbon-rich matrices.

2.4.6. Mass interferences. Apart from isobaric overlap, the most common interferences in ICP-MS are due to recombination of ions. There are different types of interferences.

1. The argon plasma: Ar⁺, Ar₂⁺

2. Polyatomic species: formed by incomplete dissociation of the sample matrix or recombination in the plasma tail, usually in the form of oxides MO⁺ (or MO²⁺, MO³⁺).

3. Air entrainment and gas impurity (N⁺, O²⁺, NO²⁺, etc…)

4. Material eroded from the cones (isotopes of Ni, Cu, Mo, etc…)

5. Various carbide interferences (e.g. ¹²C + ¹³C = ²⁵Mg) formed due to the large amount of carbon introduced into the system.

Measures taken to deal with air entrainment have been discussed above. Special care is taken to thoroughly clean all the torch components, cones and extraction lenses prior to every analytical session.

To identify and quantify the various isotopic interferences arising from the re-combination of C, Ar, N, O, H, Cl etc. a synthetic chemical vapour deposited (CVD) diamond has been repeatedly analysed. The synthetic diamond was grown from a vapour of CH₂ in the presence of H₂ and is assumed to have no elemental impurities, i.e., it consists of pure carbon. Hence, any signals obtained by ablation of this diamond are assumed to be artefacts due to molecular interferences. The concentrations calculated from these signals can be considered as ‘blank’ values representing the contribution of the interferences on that mass. Analyses were performed on this synthetic diamond using multiple isotopes for various elements to determine the average “concentration” for different elements and to select the isotopes with least interference. Table S3 shows the ‘blank’ concentrations obtained for 28 analyses of the synthetic diamond. The range in values appears to reflect variations in instrument conditions. Elements that have ‘blank’ concentrations lower than the mean MDL are considered to have negligible interference contribution to the overall signal. For polyatomic species formed by the combination of C atoms, e.g. ²⁵Mg = ¹²C + ¹³C, ⁵³Cr = ¹³C + ⁴⁰Ar, count rates are proportional to carbon signal, but the apparent or ‘blank’ concentrations are independent of the C signal (Fig. S2 (a and b)). This shows that the ratio of the carbide interferences to the total carbon signal remains constant. The calculated blank concentrations for interferences related to combinations of Ar, N and O, but not C, decline slightly with increasing C signal. With due caution, the measured blank concentrations can thus be subtracted from the concentrations obtained for the unknown diamonds. Trace-element concentrations measured well above this blank level will have negligible interference contribution.

2.4.7. Sample preparation. The diamonds are first ultrasonically cleaned in acetone and/or isopropyl alcohol for 30–60 min to remove any organic contaminants and then put in concentrated HNO₃ for 30 min. The cleaned diamonds are mounted on a Teflon block using double-sided tape and loaded into the laser cell.

3. Results

The methods were tested by the analysis of two fibrous diamonds from the Jwaneng diamond mine (Botswana), which have been characterised previously by proton microprobe (PIXE⁹). Schrauder et al.⁷ report trace element data obtained by INAA for a set of 13 diamonds, which include JWA-115 and JWA-110, that were previously examined by EMP and FTIR.⁸

Fibrous diamonds, unlike most other diamonds, are considered to have grown rapidly, possibly in the kimberlite magma during or shortly before its ascent to the surface, and their fibrous structure has trapped submicroscopic inclusions. When cut, diamonds of this type show a radiating fibrous texture²⁵ and cloudy zones with microinclusions. These inclusions contain syngenetic mantle fluids under pressure,^26,27 as well as a range of solid phases. Analyses of fluid inclusions in fibrous diamonds from Botswana suggest that the bulk composition of the fluids within individual diamonds is uniform, but the fluids in different diamonds range in composition from carbonatitic to hydrous endmembers, with intermediate compositions.⁸

The trace-element concentration in the diamond depends on the abundance and the size of these inclusions trapped between the diamond fibres. As a result, the fibrous diamonds are heterogeneous in their trace element contents, and concentrations may vary from point to point by factors of 3–10 for individual elements.

3.1. Oil calibration

The fibrous diamonds JWA 115 and JWA 110 were analysed first using the oil calibration method and the results were compared with the data obtained by INAA⁷ and the proton microprobe.⁹

The comparison between the data from the proton microprobe and the oil calibration (Fig. 2(a), Table 2(A)) shows an overall good correlation for both diamonds except for elements such as Ca and Ba. The discrepancy for Ca is believed to be due to a combination of possible factors, including heterogeneity of Ca in this diamond, as indicated by large ranges in the LAM-ICP-MS and PIXE Ca measurements, and uncorrected C-based interferences (e.g., C₁₂N₁₄O₁₆⁺) (see section 2.4.6). The inaccuracy of the PIXE analyses may be the reason for discrepancies in the Ba values (the PIXE data have a large uncertainty). The wide range in values demonstrates the heterogeneity of the diamonds. Similarly, the comparison between the INAA data and the oil calibration shows a good correlation, except for the Si content in JWA 110, which again may reflect the inhomogeneous nature of the diamond (Table 2(B), Fig. 2(b)).

Fig. 2 (a) Comparison of trace-element data for JWA 110 and JWA 115 analysed by LAM-ICP-MS (oil calibration) and by proton microprobe (PIXE) and (b) by neutron activation (INAA).

Table 2 (A) Comparison of trace-element data for JWA 110 and JWA 115 analysed by LAM-ICP-MS (oil calibration) and by proton micro-probe (PMP). (B) Comparison of trace-element data for JWA 115 and JWA 110 analysed by LAM-ICP-MS (oil calibration) and by instrumental neutron activation analysis (INAA)

(A) JWA 110
	Griffin et al.^9				LAM-ICP-MS analysis
Element	Average (ppm)	Max (ppm)	Min (ppm)	1σ	Average (ppm) n = 4	1σ	Difference (%)	Max (ppm)	Min (ppm)
Ca	24.6	68.6	4.40	19.6	96.0	15.6	291	114	76.3
Ti	7.46	12.7	0.20	4.90	10.9	4.60	45.6	14.9	5.80
Fe	33.4	73.4	5.94	22.7	24.4	15.2	−26.9	45.3	10.0
Ni	0.94	1.40	0.42	0.34	0.54	0.049	−42.2	0.61	0.50
Cu	1.20	2.30	0.15	0.77	0.39	0.13	−67.7	0.54	0.30
Sr	1.22	3.84	<DL	1.20	0.92	0.43	−24.6	1.52	0.58
Ba	48.4	87.4	25.7	19.5	1.65	0.69	−96.6	2.56	1.00

---

JWA 115
	Griffin et al.^9				LAM-ICP-MS analysis
Element	Average (ppm)	Max (ppm)	Min (ppm)	1σ	Average (ppm) n = 4	1σ	Difference (%)	Max (ppm)	Min (ppm)
Ca	28.5	38.4	19.7	6.39	96.0	13.7	237	111	77.7
Ti	30.8	46.3	20.9	8.03	19.5	4.70	−36.7	26.2	15.3
Fe	104	147	70.9	28.0	47.5	8.20	−54.1	59.5	41.3
Ni	0.85	1.12	0.42	0.26	0.60	0.20	−33.0	0.85	0.44
Sr	2.79	3.85	1.91	0.60	1.70	0.30	−37.8	1.98	1.37
Ba	25.9	79.0	< DL	27.8	2.50	0.80	−90.2	3.61	1.83

---

(B) JWA 110
	Schrauder et al.^7,8	LAM-ICP-MS analysis
Element	Average (ppm)	Average (ppm) n = 4	1σ	Difference (%)	Max (ppm)	Min (ppm)
Na	6.00	4.87	1.47	−19	6.71	3.21
Mg	12.1	10.7	3.98	−12	15.6	7.32
Al	10.2	12.0	4.95	18	17.1	6.29
Si	76.2	211	15.9	177	230	195
Ti	10.9	10.9	4.64	−0.3	14.9	5.80
Fe	31.2	24.4	15.2	−22	45.3	9.99
Ni	0.54	0.54	0.049	0.1	0.61	0.50
Rb	0.22	0.18	0.050	−18	0.25	0.13
Sr	1.47	0.92	0.43	−38	1.52	0.58
Ba	1.70	1.65	0.69	−3	2.56	1.00

---

JWA 115
	Schrauder et al.^7,8			LAM-ICP-MS analysis
Element	Average (ppm)	Max (ppm)	Min (ppm)	Average (ppm) n = 4	1σ	Difference (%)	Max (ppm)	Min (ppm)
Na	16.5	21.4	11.7	10.1	2.05	−39	12.4	8.60
Mg	26.0	33.8	18.2	11.9	0.76	−54	12.7	11.3
Al	27.5	35.7	19.3	30.2	18.4	10	51.2	16.7
Si	230	300	161	246	24.6	7	266	215
Ti	27.0	35.1	18.9	19.5	4.65	−28	26.2	15.3
Fe	69.1	88.4	49.9	47.5	8.18	−31	59.5	41.3
Ni	1.22	1.32	1.12	0.57	0.19	−53	0.85	0.44
Rb	0.50	0.59	0.40	0.36	0.057	−28	0.43	0.31
Sr	2.18	2.40	1.95	1.73	0.26	−20	1.98	1.37
Ba	2.81	3.71	1.90	2.54	0.76	−10	3.61	1.83

---

The results of this comparison show that the oil calibration provides good results for a range of elements within an acceptable accuracy, considering the heterogeneous nature of the diamond samples.

3.2. Cellulose calibration

Multiple analyses were performed on the fibrous diamonds JWA 110 and JWA 115 using the cellulose calibration method.

Table 3 (A and B) and Fig. 3 (a and b) show the results of 72 analyses of JWA 115 and 28 analyses of JWA 110 compared with data from INAA⁷ and PIXE.⁹ For the calculation of averages, values below the detection limit were replaced by zeroes. The tables show two columns for the standard deviation; the ‘calculated’ 1σ for the sample set reflects the heterogeneity of the diamond. The ‘analytical’ 1σ gives the mean 1σ for individual analyses calculated from counting statistics, and reflects the intrinsic precision of the analytical technique. Table 3 contains data for additional elements which have not been determined either by INAA or by PIXE.

Fig. 3 (a) Comparison of trace-element data for JWA 110 and JWA 115 analysed by LAM-ICP-MS (cellulose calibration) and by proton microprobe (PIXE) and (b) by neutron activation (INAA).

Table 3 (A) Comparison of trace-element data for JWA 110 and JWA 115 analysed by LAM-ICP-MS (cellulose calibration) and by proton micro-probe (PMP). (B) Comparison of trace-element data for JWA 110 and JWA 115 analysed by LAM-ICP-MS (cellulose calibration) and by instrumental neutron activation analysis (INAA), and additional data obtained by LAM-ICP-MS

(A) JWA 110
	Griffin et al.^9				LAM-ICP-MS analysis
Element	Average (ppm)	Max (ppm)	Min (ppm)	1σ	Average (ppm) n = 28	1σ (calculated)	1σ (analytical)	Avg MDL	Difference (%)	Max (ppm)	Min (ppm)
Ca	24.6	68.6	4.36	19.6	25.8	18.0	6.12	11.5	5	54.0	<DL
Ti	7.46	12.7	0.20	4.88	10.6	6.84	0.82	0.19	42	24.7	0.20
Cr	0.31	1.47	<DL	0.46	<DL	—	—	0.22	—	3.53	<DL
Fe	33.4	73.4	5.94	22.7	26.4	17.7	2.19	1.53	−21	63.4	<DL
Ni	0.94	1.42	0.42	0.34	0.32	0.13	0.038	0.035	−66	0.53	<DL
Cu	1.20	2.30	0.15	0.77	0.062	0.088	0.037	0.083	−95	0.36	<DL
Zn	1.48	3.08	0.52	0.80	0.21	0.36	0.090	0.10	−86	1.06	<DL
Sr	1.22	3.84	<DL	1.18	0.89	0.59	0.069	0.002	−27	2.14	0.004
Zr	1.09	1.63	0.43	0.45	0.57	0.38	0.045	0.004	−48	1.40	<DL
Ba	48.4	87.4	25.7	19.5	1.99	1.33	0.27	0.027	−96	5.23	<DL

---

JWA 115
	Griffin et al.^9				LAM-ICP-MS analysis
Element	Average (ppm)	Max (ppm)	Min (ppm)	1σ	Average (ppm) n = 72	1σ (calculated)	1σ (analytical)	Avg MDL	Difference (%)	Max (ppm)	Min (ppm)
Ca	28.5	38.4	19.7	6.39	35.7	17.1	8.92	17.4	25	80.2	9.5
Ti	30.8	46.3	20.9	8.03	28.8	8.13	2.33	0.27	−7	62.6	15.3
Cr	0.20	0.32	0.044	0.086	<DL	—	—	0.30	—	3.40	<DL
Fe	104	147	70.9	28.0	79.8	20.9	6.57	2.28	−23	142	40.6
Ni	0.85	1.12	0.42	0.26	0.36	0.10	0.048	0.050	−58	0.79	0.23
Cu	0.37	0.89	<DL	0.28	0.26	0.12	0.059	0.11	−30	0.57	<DL
Zn	1.74	2.87	0.67	0.81	0.68	0.87	0.16	0.15	−61	2.07	<DL
Sr	2.79	3.85	1.91	0.60	2.12	0.51	0.18	0.003	−24	3.71	1.16
Zr	0.95	2.00	0.16	0.53	0.42	0.12	0.038	0.005	−56	0.96	0.21
Ba	25.9	79.0	<DL	27.8	4.48	1.17	0.76	0.038	−83	8.00	2.33

---

(B) JWA 110
	Schrauder et al.^7,8	LAM-ICP-MS analysis
Element	Average (ppm)	Average (ppm) n = 28	1σ (calculated)	1σ (analytical)	Avg MDL	Difference (%)	Max (ppm)	Min (ppm)
Na	6.00	4.29	3.63	0.50	0.73	−28	13.0	<DL
Mg	12.1	14.6	9.77	1.05	0.24	21	36.0	0.18
Al	10.2	12.8	8.24	1.30	0.09	25	31.2	0.04
K	61.3	56.2	38.6	3.91	0.57	−8	136	<DL
Ca	—	25.8	18.0	6.12	11.5	—	54.0	<DL
Ti	10.9	10.6	6.84	0.82	0.19	−3	24.7	0.20
V	—	0.036	0.030	0.010	0.019	—	0.101	<DL
Cr	1.63	—	—	—	0.22	—	3.53	<DL
Mn	—	0.28	0.19	0.034	0.053	—	0.65	<DL
Fe	31.2	26.4	17.7	2.19	1.53	−15	63.4	<DL
Co	—	0.026	0.017	0.006	0.010	—	0.060	<DL
Ni	0.54	0.32	0.13	0.038	0.035	−41	0.53	<DL
Cu	—	0.062	0.088	0.037	0.083	—	0.36	<DL
Zn	0.17	0.21	0.36	0.090	0.10	22	1.06	<DL
Ga	—	0.062	0.07	0.011	0.017	—	0.30	<DL
Rb	0.22	0.22	0.14	0.020	0.014	−3	0.48	<DL
Sr	1.47	0.89	0.59	0.069	0.002	−39	2.14	0.004
Y	—	0.025	0.018	0.003	0.002	—	0.060	<DL
Zr	1.12	0.57	0.38	0.045	0.004	−49	1.40	<DL
Nb	—	0.057	0.039	0.007	0.002	—	0.14	0.001
Mo	—	0.004	0.005	0.006	0.015	—	0.015	<DL
Cs	0.004	0.002	0.004	0.004	0.009	−59	0.016	<DL
Ba	1.70	1.99	1.33	0.27	0.027	17	5.23	<DL
La	0.09	0.089	0.059	0.008	0.003	−5	0.21	0.002
Ce	0.19	0.15	0.10	0.012	0.002	−21	0.34	0.002
Pr	—	0.022	0.015	0.003	0.002	—	0.051	<DL
Nd	—	0.12	0.17	0.029	0.033	—	0.75	<DL
Sm	0.018	0.028	0.081	0.010	0.018	57	0.43	<DL
Eu	0.007	0.006	0.006	0.003	0.007	−17	0.020	<DL
Gd	0.007	0.017	0.032	0.010	0.019	139	0.13	<DL
Dy	—	0.003	0.007	0.006	0.013	—	0.020	<DL
Ho	—	0.002	0.002	0.001	0.002	—	0.007	<DL
Er	—	0.003	0.005	0.005	0.012	—	0.015	<DL
Yb	0.001	0.004	0.014	0.007	0.017	245	0.062	<DL
Lu	—	0.001	0.002	0.001	0.003	—	0.006	<DL
Hf	0.015	0.012	0.012	0.006	0.011	−20	0.047	<DL
Ta	—	0.002	0.003	0.002	0.004	—	0.009	<DL
Pb	—	0.037	0.024	0.009	0.011	—	0.089	<DL
Th	0.016	0.012	0.009	0.003	0.004	−23	0.033	<DL
U	0.004	0.002	0.004	0.002	0.005	−40	0.015	<DL

---

JWA 115
	Schrauder et al.^7,8	LAM-ICP-MS analysis
Element	Average (ppm)	Max (ppm)	Min (ppm)	Average (ppm) n = 72	1σ (calculated)	1σ (analytical)	Avg MDL	Difference (%)	Max (ppm)	Min (ppm)
Na	16.5	21.4	11.7	18.1	6.77	1.55	1.15	10	33.5	3.95
Mg	26.0	33.8	18.2	33.6	8.69	2.61	0.36	29	60.3	6.29
Al	27.5	35.7	19.3	37.2	9.85	3.28	0.14	35	61.7	6.81
K	98.5	128	69.0	105	26.9	8.12	0.65	6.7	187	15.8
Ca	—	—	—	35.7	17.1	8.92	17.4	—	80.2	1.96
Ti	27.0	35.1	18.9	28.8	8.13	2.33	0.27	6.4	62.6	5.45
V	—	—	—	0.06	0.03	0.014	0.027	—	0.230	0.018
Cr	0.24	0.23	0.24	<DL	—	—	0.30	—	3.40	<DL
Mn	—	—	—	0.75	0.20	0.072	0.074	—	1.38	0.18
Fe	69.1	88.4	49.9	79.8	20.9	6.57	2.28	15	142.3	11.9
Co	—	—	—	0.059	0.020	0.012	0.019	—	0.13	0.027
Ni	1.22	1.32	1.12	0.36	0.10	0.048	0.050	−71	0.79	0.21
Cu	—			0.26	0.12	0.059	0.11	—	0.66	<DL
Zn	0.19	0.19	0.18	0.68	0.87	0.16	0.15	263	5.57	<DL
Ga	—	—	—	0.17	0.15	0.023	0.027	—	0.59	0.040
Rb	0.50	0.59	0.40	0.53	0.14	0.047	0.018	7.7	0.96	0.11
Sr	2.18	2.40	1.95	2.12	0.5	0.18	0.003	−2.3	3.71	0.40
Y	—	—	—	0.066	0.018	0.007	0.002	—	0.11	0.014
Zr	0.46	0.50	0.42	0.42	0.12	0.038	0.005	−9.0	0.95	0.072
Nb	—	—	—	0.30	0.10	0.032	0.003	—	0.78	0.06
Mo	—	—	—	0.024	0.020	0.012	0.025	—	0.071	<DL
Cs	0.063	0.091	0.035	0.019	0.039	0.006	0.011	−69	0.24	<DL
Ba	2.81	3.71	1.90	4.48	1.17	0.76	0.038	60	8.29	0.91
La	0.19	0.24	0.13	0.23	0.06	0.021	0.004	23	0.39	0.039
Ce	0.32	0.41	0.22	0.34	0.09	0.030	0.003	7.6	0.61	0.064
Pr	—	—	—	0.057	0.086	0.007	0.003	—	0.76	0.006
Nd	—	—	—	0.20	0.26	0.041	0.042	—	2.02	0.065
Sm	0.023	0.029	0.016	0.042	0.024	0.014	0.022	83	0.13	<DL
Eu	0.009	0.010	0.007	0.014	0.010	0.005	0.009	60	0.049	<DL
Gd	0.015	0.020	0.010	0.033	0.019	0.012	0.021	120	0.097	<DL
Dy	—	—	—	0.024	0.013	0.008	0.014	—	0.070	<DL
Ho	—	—	—	0.010	0.008	0.002	0.004	—	0.045	<DL
Er	—	—	—	0.012	0.008	0.008	0.018	—	0.033	<DL
Yb	0.004	0.005	0.003	0.021	0.022	0.011	0.026	402	0.092	<DL
Lu	—			0.008	0.008	0.002	0.004	—	0.046	<DL
Hf	0.011	0.015	0.008	0.015	0.009	0.007	0.014	37	0.031	<DL
Ta	—	—	—	0.014	0.007	0.004	0.005	—	0.036	<DL
Pb	—	—	—	0.050	0.026	0.013	0.017	—	0.12	<DL
Th	0.027	0.036	0.018	0.037	0.016	0.006	0.006	37	0.094	<DL
U	0.010	0.010	0.010	0.011	0.011	0.004	0.006	17	0.069	<DL

---

Despite the heterogeneity of these stones, the mean values determined by all three methods on JWA 115 agree within one standard deviation for 23 of 28 elements. Ni is difficult to analyse by INAA, and the Ni values derived by ICP-MS agree more closely with the PIXE values; the agreement is good also for JWA 110. After subtraction of the blank, the Cr values were reduced to below the average detection limit. P and As values determined by ICP-MS are clearly too high, and relatively constant from point to point: they may reflect an unidentified artefact.

The same analyses have been processed using the SRM NIST612/610 glass as the external standard, and ⁸⁸Sr, derived from the cellulose calibration, as the internal standard. ⁸⁸Sr was chosen as it gives quite consistent concentrations above the detection limit. The results for JWA 115 (Table S4) show excellent agreement with those derived from the cellulose calibration. This agreement indicates that the matrix effects are much less significant than commonly supposed.

3.2.1. Detection limits. Detection limits for LAM-ICP-MS analysis are a function of the amount of material presented to the ICP-MS for analysis (pit size), the counting time per element (this is a function of the number of elements determined for a given time of ablation) and the abundance of the isotope being measured.¹⁷ Detection limits improve with lower background levels, better ICP-MS sensitivity (i.e., higher counts ppm⁻¹) and increased ablation rates, and hence, will vary from analysis to analysis. An average of the detection limits obtained for each element during the analyses of JWA 115 and JWA 110 is given in Table 3 (A) and (B). Typical detection limits for most of the rare earth elements are 5–20 ppb. The transition elements have slightly higher detection limits, with most <500 ppb. Na and Fe have higher detection limits (2–3 ppm), which reflect the higher background signal levels for these elements and the low isotopic abundance of ⁵⁷Fe.

3.3. Geological context

Fig. S3 shows the mean chondrite-normalised trace element patterns for JWA 115 and JWA 110 compared with those of average kimberlite²⁸ and average carbonatite.²⁹ These diamonds are enriched in incompatible elements such as the REE, Rb, Sr, Ba, Th and Nb, reflecting the nature of the melts and fluids trapped in their inclusions. The broad similarity of these patterns to those of kimberlites and especially carbonatites has been noted previously.^30,31

Conclusions

LAM-ICP-MS, using the multi-element oil or the cellulose multi-element pellet with ¹³C as the internal standard, is an effective technique for the quantitative determination of trace-element concentrations in the diamonds. This technique produces accurate and precise data for most elements, and the heterogeneity of individual natural diamonds is the major limiting factor on the reproducibility of the results. The cellulose pellet is the preferred standard as it is simpler to use and delivers better accuracy and precision than the oil calibration method.

Acknowledgements

Oded Navon generously provided the Jwaneng diamonds JWA 110 and JWA 115 earlier selected by Jeff Harris. Jim Butler is thanked for providing the synthetic CVD diamond, and Jeff Harris for samples, discussions and encouragement. We also thank the GEMOC laboratory staff (Suzy Elhlou, Peter Wieland, Carol Lawson and Ashwini Sharma) for their continued patience and support, and the two anonymous reviewers for their comments. This is contribution number 407 from the ARC National Key Centre for Geochemical Evolution and Metallogeny of Continents (www.els.mq.edu.au/GEMOC/)

References

1. H. Meyer in Inclusions in Diamond. Mantle Xenoliths, ed. P. H. Nixon, J. Wiley & Sons, Chichester, 1987, pp. 501–522.
2. R. M. Chrenko, R. S. McDonald and K. A. Darrow, Nature, 1967, 213.
3. Y. L. Orlov, The Mineralogy of the Diamond, Wiley, New York, 1977.
4. H. W. Fesq, D. M. Bibby, C. S. Erasmus, E. J. D. Kable and J. F. P. Sellschop, Phys. Chem. Earth, 1975, 9, 817–836.
5. H. W. Fesq, D. M. Bibby, J. F. P. Sellschop and J. I. W. Watterson, J. Radioanal. Chem., 1973, 17, 195–216.
6. D. M. Bibby, Chem. Phys. Carbon, 1982, 18, 3–91.
7. M. Schrauder, C. Koeberl and O. Navon, Geochim. Cosmochim. Acta, 1996, 60, 4711–4724.
8. M. Schrauder and O. Navon., Geochim. Cosmochim. Acta, 1994, 58, 761–771.
9. W. L. Griffin, C. G. Ryan and T. T. Win, CSIRO Restricted Report, 1993, 363R.
10. M. Resano, F. Vanhaecke, D. Hutsebaut, K. De Corte and L. Moens, J. Anal. At. Spectrom., 2003, 18, 1238–1242.
11. M. D. Norman, N. J. Pearson, A. Sharma and W. L. Griffin, Geostand. Newsl., 1996, 20, 247–261.
12. E. T. Jeffries, S. E. Jackson and H. P. Longerich, J. Anal. At. Spectrom., 1998, 13, 935–940.
13. C. Dalpé, personal communication, 2004.
14. K. Witte, Colorado School of Mines, Masters thesis, 2001.
15. S. M. Eggins, L. P. Kinsley and J. M. G. Shelly, Appl. Surf. Sci., 1998, 129, 278–286.
16. D. Gunther and C. A. Heinrich, J. Anal. At. Spectrom., 1999, 14, 1363–1368.
17. S. E. Jackson, H. P. Longerich, G. R. Dunning and B. J. Fryer, Can. Mineral., 1992, 30, 1049–1064.
18. A. Gray, J. Anal. At. Spectrom., 1992, 7, 1151–1153.
19. A. T. Ince, J. G. Williams and A. Gray, J. Anal. At. Spectrom., 1993, 8, 899–903.
20. D. Gunther and B. Hattendorf, in Laser-Ablation- ICPMS in the Earth Sciences. Principles and Applications, ed. P. Sylvester, Mineralogical Association of Canada, St. John’s, Newfoundland, 2001, vol. 29, p. 243.
21. C. A. Heinrich, T. Pettke, W. E. Halter, M. Aigner-Torres, A. Audetat, D. Gunther, B. Hattendorf, D. Bleiner, M. Guillong and I. Horn, Geochim. Cosmochim. Acta, 2003, 67, 3473–3496.
22. B. J. Fryer, S. E. Jackson and H. P. Longerich, Can. Mineral., 1995, 33, 303–312.
23. H. P. Longerich, D. Günther and S. E. Jackson, Fresenius’ J. Anal. Chem., 1996, 355, 358–542.
24. M. Guillong and D. Gunther, J. Anal. At. Spectrom., 2002, 17, 831–837.
25. A. R. Lang, J. Cryst. Growth, 1974, 24/25, 108–115.
26. O. Navon, I. D. Hutcheon, G. R. Rossman and G. J. Wasserburg, Nature, 1988, 335, 784–789.
27. O. Navon, Nature, 1991, 353, 746–748.
28. R. H. Mitchell, Kimberlites, Plenum Press, New York, 1986.
29. A. R. Woolley and D. R. C. Kempe, Carbonatites Genesis and Evolution, ed. K. Bell, 1989, vol. 1, pp. 1–14.
30. S. E. Jackson, R. M. Davies, W. L. Griffin, S. Y. O’Reilly and B. Doyle, 9th Annual V. M. Goldsmidt Conference, Extended abstracts, Cambridge, Massachusetts, 1999.
31. S. Rege, R. M. Davies, W. L. Griffin, S. E. Jackson and S. Y. O’Reilly, 8th International Kimberlite Conference Extended abstracts, Victoria, Canada, 2003.

---

Footnote

† Electronic supplementary information (ESI) available: results of solution analysis of the multi-element cellulose pellet (these values were used as standard values for the LAM-ICP-MS analysis); cellulose standard analysed as an unknown with SRM NIST612 glass as the external standard, and Sr as the internal standard; average ‘blank’ concentrations for a synthetic CVD diamond (n = 28); comparison of trace-element data obtained for JWA 115 by LAM-ICP-MS [cellulose calibration (C-internal standard) versus NIST 612/610 calibration (Sr-internal standard)]; average fractionation index (FI) calculated for 22 analyses of the cellulose; chondrite normalized trace-element patterns for JWA 115 and JWA 110 (LAM-ICP-MS) compared with those of Average Kimberlite and Average Carbonatite. See http://www.rsc.org/suppdata/ja/b5/b501374g/

---