External furnace-fusion digestion for the direct determination of lead in rock samples by inductively coupled plasma mass spectrometry (ICP-MS) using the tungsten boat furnace–sample cuvette technique

Abstract

A newly conceived electrothermal vaporisation system using a tungsten boat furnace (TBF)–sample cuvette technique was applied for the direct analysis of rock samples with detection by inductively coupled plasma mass spectrometry (ICP-MS). Into this sample cuvette made of tungsten, the powdered rock sample and an aliquot of ammonium fluoride solution were placed and the cuvette was heated on a hot-plate for digestion. After the decomposition had been completed, the cuvette was positioned on the TBF. The analyte in the cuvette was vaporised and introduced into the ICP mass spectrometer. Since the solid samples could be decomposed to ash completely in the cuvette and the majority of siliceous components were expelled prior to measurement, the sensitivity was the same as that of aqueous standards. The most remarkable feature is that the small sample cuvette was utilised as a weighing dish, sample carrier, decomposition vessel and electrothermal vaporiser. Thus, the technique overcomes problems such as weighing small amounts of powdered samples, decomposition of solid materials, introduction of the samples into the TBF device and the easy removal of the residue. The method was successfully applied to the direct determination of lead in several standard rock samples.

---

Introduction

Direct solid introduction schemes are very attractive for atomic spectrometry. The advantages of the direct analysis of solid samples by inductively coupled plasma mass spectrometry (ICP-MS) are the time saved by avoiding dissolution and dilution steps, as well as the high sensitivity. Other advantages are that a few milligram amounts of samples are required and contaminations inherent in sample preparation are avoided. Because of the memory effects, however, the solid sampling method for introduction into ICP-MS has not always found wide application, although several approaches have been described, including the application of a direct solid insertion (DSI) device,¹ a laser ablation (LA) system,^2–8 and an electrothermal vaporisation (ETV) device,^9–13 as well as its modification to a furnace fusion (FF)–ETV device.^14–17 Among them, only a few papers have been reported concerning powdered or particulate matter. For soil and sediment samples, Baker et al.⁸ presented a method in which LA-ICP-MS was used to analyse pelletized sand samples. Tanaka et al.,⁴ Wang et al.⁶ and Chin et al.¹³ analysed airborne particulate matter using LA-ICP-MS. The particulates were collected on membrane filters prior to ablation. The LA methods mentioned were only applicable to solid blocks, pelletized samples and membranes on which particulate matters were loaded before measurements, although it is an effective technique for the direct solid determination by ICP-MS.

Considering ETV techniques combined with ICP-MS, a further advantage of increased selective vaporisation of each analyte from matrix compounds is observed. Moreover, permissible amounts of the matrix are significantly improved by applying chemical modification. Moor et al.¹² described a unique ETV-ICP-MS method based on one-particle sampling and in-situ digestion (or micro-digestion) with nitric acid in a graphite tube furnace. When powdered samples are analysed by the ETV method, several difficulties still remain, such as introducing pre-weighed small amounts of finely powdered samples into the ETV device, removing the residues from it easily and finding an accurate calibration method. In order to overcome these, a newly conceived method has been presented, being abbreviated as the furnace-fusion (FF) method.^14–17 The most remarkable feature of the FF method is that not only the organic matrix but also the excess of the reagents could be expelled from the device during the ashing procedures. Another merit is that the sample weighing procedures are facilitated by utilising the sample cuvettes as weighing dishes, sample carriers, crucibles for fusion and electrothermal vaporisers. In previous papers, various biological samples were decomposed with tetramethylammonium hydroxide (TMAH) and diammonium hydrogenphosphate by the FF technique followed by determination with an ICP atomic emission spectrometer^14–16 and an ICP mass spectrometer.¹⁷ However, the combined use of TMAH and diammonium hydrogenphosphate is not suitable for the analysis of rock samples, since no siliceous matrixes are digested and fused by the reagents. Hence, this paper will demonstrate the application of the FF technique to the analysis of rock samples by applying an alternative decomposition reagent.

Experimental

Instruments

A Seiko II (Chiba, Japan) Model SPQ9000 ICP mass spectrometer was utilised. A Seiko II Model EV-300 metal furnace vaporiser was used after modification. This vaporisation head of TBF device has been described in detail previously.^14–20 A PTFE tube (4 mm id × 50 cm long) was used for the connection of the outlet port of the TBF vaporiser and the inlet port of the ICP torch. The small sample cuvettes (10 mm × 20 mm) were shaped by cutting both edges of the tungsten boats. Solid samples were ground manually with agate mortars. Gilson Medical Electronics (Villiers-le-Bel, France) Model Pipetman P-20 and Model Microman M-50 digital pipettes were used for standards and reagent injections, respectively. An Advantec Toyo (Tokyo, Japan) Model TH-350 hot-plate, whose maximum heating temperature was 350 °C, was used to decompose powdered rock samples. Details of the vaporiser head, the TBF and the small sample cuvettes are illustrated in refs. 15 and 16.

Solid standards and reagents

Standard rock samples were obtained from the Geological Survey of Japan (Tsukuba, Japan) as GSJ series. Aqueous standards for lead(II) were prepared by diluting a 1000 mg dm⁻³ lead(II) stock solution (Atomic Absorption Standard Solution, Hayashi Pure Chemical, Osaka, Japan) with 0.14 mol dm⁻³ nitric acid. A 1000 mg dm⁻³ silicate(IV) solution (Kanto Chemical, Tokyo) was used as received. A chemical modifier solution was prepared by dissolving 25 g of ammonium fluoride (Pro Analysi grade, Merck, Darmstadt, Germany) with 100 cm³ of water.

Recommended procedure

A rock sample was ground to a fine powder in the agate mortar. The resulting average particle size was found to be approximately 10 µm by using a particle analyser. For routine analysis, 1 g of the sample was taken. By means of the microbalance, 0.2 mg of the prepared sample was weighed accurately into the sample cuvette, which was preconditioned by heating to expel lead impurities (confirmed by a blank signal level). If higher detectabilities are needed, up to 10 mg of the prepared sample can be used. In order to prepare the standard batch used for construction of a calibration curve, silicon(IV) standard solution was placed into the cuvette instead of silicate powder, followed by addition of an aliquot of the lead(II) standard solution. An aliquot of the ammonium fluoride solution was added to each cuvette with the Microman digital pipette. The cuvettes were placed for typically about 20 min on the hot-plate kept at 150 °C until white fumes containing silicon tetrafluoride and excess ammonium fluoride were almost expelled. Then, one of the cuvettes was placed on the TBF through the sample insertion port of the glass dome with a pair of tweezers. The TBF was ramped through two steps for the ashing procedure followed by a vaporisation and introduction step. During the ashing-1 step, the remaining silicon tetrafluoride and ammonium fluoride were removed completely via the open sample insertion port. After the insertion port had been closed with a stopper, the temperature was ramped up to pyrolyse the contents of the cuvette. Finally, the analyte on the cuvette was vaporised and introduced into the plasma ion source through the port to the torch. The transient signal of ²⁰⁸Pb was integrated and the peak area was estimated using the software attached to the spectrometer, which ran synchronously with the vaporiser. The operating conditions used in this work are summarised in Table 1.

Table 1 Instrument operating conditions

ICP mass spectrometer (Seiko II SPQ9000)
Rf incident power	1.2 kW
Argon gas flow rate
Plasma gas	16 dm^3 min^−1
Auxiliary gas	1.0 dm^3 min^−1
Analytical mass	^208Pb
Sampling depth	8.0 mm
Metal furnace vaporiser (Seiko II EV-300)
Powdered sample	ca. 0.5 mg
NH4F	12.5 mg
Heating programme
Ashing-1	220 °C for 30 s (ramp10 s)
Ashing-2	600 °C for 90 s (ramp10 s)
Vaporisation	1500 °C for 11 s (ramp14 s)
Carrier gas flow rate
Argon gas	550 cm^3 min^−1
Hydrogen gas	20 cm^3 min^−1

---

Results and discussion

Optimisation of operating conditions

To optimise the operating conditions for the lead vaporisation, the amount of ammonium fluoride and the heating programme were investigated by recording signals from aliquots of GSJ JR-1 (Rhyolite). In the common fusion technique for the decomposition of rock samples, sodium carbonate, sodium hydroxide, potassium carbonate, potassium hydroxide, or a mixture of these reagents are used as flux materials. However, these reagents are not usually recommended for atomic spectrometry because of the high salt content in the resulting solutions and their spectral interferences during the measuring procedure.

Ammonium fluoride is corrosive enough to decompose the siliceous matrixes. Moreover, it is completely removable by heating at approximately 430 °C or above. As Fig. 1 shows, when no ammonium fluoride was added to the sample an almost negligible signal was observed. The sensitivity increased with the addition of increasing amounts of ammonium fluoride, and remained almost constant over the range of 5 mg or more. Therefore 12.5 mg was selected, i.e., each 50 mm³ aliquot of the 25% ammonium fluoride aqueous solution was added to the cuvette, in which approximately 0.5 mg of a rock sample had previously been weighed out. The decomposition process was carried out on the hot-plate prior to the introduction. During the process, ammonium fluoride was gradually degraded to produce ammonia and hydrofluoric acid, which mainly corroded the siliceous matrixes. As a result, silicon dioxide, the major component of the rock samples, was decomposed to silicon tetrafluoride, the product being removed subsequently as a vapour. Regarding the degradation temperature, the temperature of the hot-plate was set to 150 °C, and was sufficient to digest these samples using ammonium fluoride and to remove any excess as a vapour. The use of hydrofluoric acid should be avoided, because the fast reactions with the finely powdered rock samples cause poor analytical reproducibility. The FF method, by which the sample is digested on the TBF, is not recommended in this experiment, since the generated hydrofluoric acid vapour tends to corrode the inside of the glass dome. Owing to the fluorine etching, the transparency of the dome is gradually decreased.

Fig. 1 Effect of ammonium fluoride on the signal intensity of ²⁰⁸Pb. Sample, Rhyolite (GSJ JR-1). Operating conditions are as in Table 1.

After the sample cuvette containing the pretreated sample was placed on the TBF and the insertion port was closed with the stopper, the cuvette was heated at the ashing-1 temperature to remove the excess ammonium fluoride completely and to substitute argon gas for air in the glass dome. During the ashing-2 stage, the matrix was pyrolysed and the lead reacted with fluoride ions to form PbF₂, a thermally stable compound. The lead was retained up to a high temperature (700 °C), then a loss was observed. Therefore, 600 °C was selected as an optimum ashing-2 temperature. A 1500 °C vaporisation temperature was suitable for the determination of lead.

Determination of lead in several rock samples using aqueous standard solutions

In the proposed method, each solid sample was chemically predigested with ammonium fluoride in the sample cuvette during the external decomposition stage. Therefore, the chemical species of the analyte in the digested solid samples were the same as those in the respective aqueous standards. For calibrating solids by the calibration curve method using aqueous standard solutions, lead was determined in several standard rock samples. The analytical results are listed in Table 2. These results are in satisfactory agreement with the reference values of the Geographical Survey of Japan.

Table 2 Determination of lead in standard rock samples

Sample	Lead/µg g^−1
	Found^a^,^b	Reference
a Calibration curve method (Pb aqueous standards). b Mean ± average deviation, n = 5.
Basalt (GSJ JB-1a)	7.33 ± 0.44	6.76
Basalt (GSJ JB-2)	5.33 ± 0.55	5.33
Granodiorite (GSJ JG-1a)	27. 7± 2.7	26.4
Granodiorite (GSJ JG-3)	11. 4± 0.6	11.7
Rhyolite (GSJ JR-1)	18. 7± 0.9	19.3

---

Basic analytical performance characteristics

Since there are no realistic rock samples containing zero lead, the conventional definition of the detection limit is inappropriate in direct solid sampling. Therefore, the detection limit was estimated using aqueous standards containing potassium silicate under optimum conditions. The detection limit was estimated to be 30 pg of lead, which was defined as the absolute amount of lead required to yield a net peak area of three times the standard deviation of the blank in the presence of the ammonium fluoride. This value corresponds to 60 ng g⁻¹ of the lead concentration in solid samples, when 0.5 mg of a rock sample is placed on the sample cuvette. With a larger sample amount, a proportionally lower detection limit could be attained. Less than 10 mg of rock sample powder can be placed into the cuvette. A linear calibration graph for lead intersecting almost the origin of the coordinate axis and covering absolute amounts of at least 10 ng of lead was established. The relative standard deviation of eight replicate measurements obtained with 500 pg of lead was 4.5%. By using a number of exchangeable small sample cuvettes, the sample throughput can be increased. Approximately 20 batches could be vaporised per hour.

ack

The authors express thanks to the Geological Survey of Japan for supplying standard rock samples.

References

1. L. Blain, E. D. Salin and D. W. Boomer, J. Anal. At. Spectrom., 1989, 4, 721.
2. C. D. Garbe-Schönberg, C. Reimann and V. A. Pavlov, Environ. Geol., 1997, 32, 9.
3. S. A. Watmough, T. C. Hutchinson and R. D. Evans, Environ. Sci. Technol., 1998, 32, 2185.
4. S. Tanaka, N. Yasushi, N. Sato, T. Fukasawa, S. J. Santosa, K. Yamanaka and T. Ootoshi, J. Anal. At. Spectrom., 1998, 13, 135.
5. E. V. Putten, F. Dehairs, L. André and W. Baeyens, Anal. Chim. Acta, 1999, 378, 261.
6. C. F. Wang, C. J. Chin, S. K. Lvo and L. C. Men, Anal. Chim. Acta, 1999, 389, 257.
7. K. M. Lee, J. Appleton, M. Cooke, F. Keenan and K. Sawicka-Kapusta, Anal. Chim. Acta, 1999, 395, 179.
8. S. A. Baker, M. Bi, R. Q. Aucelio, B. W. Smith and J. D. Winefordner, J. Anal. At. Spectrom., 1999, 14, 19.
9. F. Vanhaecke, S. Boonen, L. Moens and R. Dams, J. Anal. At. Spectrom., 1995, 10, 81.
10. L. Moens, P. Verrept, S. Boonen, F. Vanhaecke and R. Dams, Spectrochim. Acta, Part B, 1995, 50, 463.
11. S. Boonen, F. Vanhaecke, L. Moens and R. Dams, Spectrochim. Acta, Part B, 1996, 51, 271.
12. C. Moor, P. Boll and S. Wiget, Fresenius' J. Anal. Chem., 1997, 359, 404.
13. C. J. Chin, C. F. Wang and S. L. Jeng, J. Anal. At. Spectrom., 1999, 14, 663.
14. Y. Okamoto, N. Takagaki, T. Fujiwara and T. Kumamaru, Anal. Commun., 1997, 34, 283.
15. Y. Okamoto, J. Anal. At. Spectrom., 1999, 14, 1767.
16. Y. Okamoto, Fresenius' J. Anal. Chem., 2000, 367, 295.
17. Y. Okamoto, Fresenius' J. Anal. Chem., 2000, 367, 300.
18. K. Fujiwara, Y. Okamoto, M. Ohno and T. Kumamaru, Anal. Sci., 1995, 11, 829.
19. Y. Okamoto, K. Sugawa and T. Kumamaru, J. Anal. At. Spectrom., 1994, 9, 89.
20. Y. Okamoto, J. Anal. At. Spectrom., 1999, 14, 1631.

---