μFlow-injection–ICP collision cell MS determination of molybdenum, nickel and vanadium in petroleum samples using a modified total consumption micronebulizer

Abstract

An interface based on a total consumption micronebulizer was developed for the introduction of xylene solutions into ICP MS. The increase in the nebulizer capillary diameter and the elimination of the internal connections reduced the problem of clogging, pressure instability and memory effects. The xylene carrier could be introduced for several hours at a rate of 30 μl min⁻¹. The sample (2.5 μl) was injected into the carrier flow to produce peaks of 5 s at half-height (20 s at the base) which allowed a throughput of ca. 100 h⁻¹ for the simultaneous determination of Mo, Ni and V. Calibration curves with good linearity (R² > 0.999) over at least three orders of magnitude and detection limits at the sub-ng ml⁻¹ levels were obtained. The method was validated by the analysis of a sample by an independent (ICP AES) method and by the analysis of a NIST CRM 1085c lubricating oil material. The use of a helium-pressurized collision cell was essential to obtain good accuracy for Ni and V but was not required for Mo.

---

Introduction

Information on trace element concentrations and isotopic ratios in petrochemical samples is increasingly demanded in view of the growing interest in the geochemical characterization of source rocks and basins and the need to control catalyst poisoning and environmental pollution.^1,2 Nickel and vanadium, present in geoporphyrins, have been the most popular elements studied, but the interest is rapidly extending towards a number of other elements, such as Cu, Mo, Ag, Sn, Ba and Pb.³ In terms of analytical techniques used ICP MS is rapidly replacing AAS and ICP AES owing to its lower detection limits, multielement and isotopic-ratio measurement capacities, and the possibility to avoid sample digestion.⁴

The introduction of organic oxygen-free solvents, lubricating oils, and other petroleum products into ICP MS is a challenging task. The most successful approach has consisted of limiting the amount of solvent arriving at the plasma by lowering the uptake rate, desolvation and operating the plasma at higher power levels. The use of conventional micronebulizers (100–300 μl min⁻¹) equipped with a spray chamber in the continuous mode bears the risk of clogging, memory effects and relatively low sample throughput.³ The cooling of the spray chamber rapidly results in condensation, requires considerable dilution factors, and can cause analyte fractionation. A typical recent protocol for one sample analysis takes about 15 min: 2–3 min necessary for signal stabilization, 10 min for the signal acquisition and additional for rinsing.³ The use of an ultrasonic nebulizer combines an efficient nebulization with desolvation of the organic aerosol, but the risk of losses and fouling of the desolvation membrane prevent routine applications to crude oils and their fractions.^5,6 An interesting alternative is the use of direct injection nebulizers such as the DIN^7,8 (no longer commercially available) and the DIHEN (direct injection high efficiency nebulizer)^9–13 because of a low internal dead volume, rapid response times, reduced memory effects, no analyte loss and lack of fractionation in the spray chamber. However, a critical effect of the position of the tip in the torch on the signal intensity and extent of the mass bias were recently shown by Heilmann et al.¹⁴ and Kahen et al.¹⁵

This study aims at the development of a method for a high-throughput simultaneous determination of Ni, V and Mo in petroleum samples. For this purpose, a total consumption micronebulizer successfully applied elsewhere for the introduction of methanol–water solutions was modified to handle crude oils diluted in xylene at higher flow rates (30 μl min⁻¹).

Experimental

Apparatus

The ICP MS instrument was an Agilent 7500ce (Agilent Technologies, Tokyo, Japan). A dual syringe pump (Model 140C, Applied Biosystems, Foster City, CA) was used to assure a stable delivery of xylene into the ICP at 30 μl min⁻¹ flow rate without the use of a splitter to limit the solvent consumption. Injections were made using a low port-to-port dead volume (30 nl) micro-injection valve (Valco Instruments, Houston, TX, USA) fitted with a 2.5 μl sample loop made of 180 μm i.d. (375 μm o.d.) fused silica capillary. The connection between the pump and the valve was made of 75 μm i.d. (375 μm o.d.) fused silica capillary.

The sample introduction system was based on a micro-flow total consumption nebulizer (Fig. 1), fitted with a low dead-volume (8 cm³) spray chamber without drain. In comparison with the instrumental setup described earlier for the flow-injection–ICP MS work^16,17 the internal capillary diameter was increased (75 μm i.d., 150 μm o.d.) and a unique piece of capillary was used to connect the valve exit to the tip of the nebulizer. The nebulizer capillary tip was centred in a sapphire orifice of 254 μm i.d. allowing the generation of a fine aerosol, which could be optimized by adjusting the position of the capillary tip in the nebulizer orifice. Oxygen flow, controlled by a mass flow controller, was mixed with the carrier gas (argon) via a T connection prior to nebulization.

Fig. 1 Design of the modified DS-5 nebulizer.

Samples and solutions

Xylene (Baker Analysed ACS grade), purchased from Baker (Mallinckrodt Baker, Phillipsburg, NJ, USA), was used as solvent. Conostan® Mo, Ni, V, Ba, In, and Y monoelemental standards in oil (100 mg l⁻¹) (ConocoPhillips, Houston, TX, USA) were used for sample calibration. A standard reference material (SRM) from NIST (National Institute for Standards and Technology, Gaithersburg, MD, USA) 1085b lubricating oil was used for method validation. The crude oil sample was provided by the French Institute of Petroleum (IFP).

Procedure

The ICP MS sensitivity was optimized daily using Ba, In and Y added to the carrier at 50 ng ml⁻¹. The most important parameter values were the oxygen flow rate (9.2 ml min⁻¹, 8% of the nebulizer gas), and the helium collision gas flow rate (3.2 ml min⁻¹) The complete set of the optimum instrumental parameters is given in Table 1 in the ESI.† Standard solutions and samples (crude oil or NIST SRM) were injected, after appropriate dilution, into the stream of carrier xylene. The calibration was carried out by the method of standard additions with three replicates per point.

Results and discussion

Optimisation of the nebulizer and its performance characteristics

The original DS-5 nebulizer (CETAC, Omaha, NE) operates at an optimum flow rate of 5 μl min⁻¹. This allows an injection of a maximum 200 nl without producing a flat top peak. The construction of the nebulizer was revised in order to maximize the sensitivity through an increase in the injected sample volume and the flow rate while maintaining the total consumption characteristics.

The modified nebulizer is shown in Fig. 1. The major difference in comparison with the original DS-5 nebulizer is an increase in the internal diameter of the capillary from 50 μm to 75 μm i.d. and the use of the same piece of the capillary to connect the exit of the valve and the tip of the nebulizer. This increase in the diameter virtually eliminated the problem of clogging, pressure variations and memory effect.

The effect of the sample uptake rate on the signal intensity is shown in Fig. 2. It is characteristic of a total consumption nebulizer up to a sample uptake rate of 30 μl min⁻¹. Above this value condensation of xylene in the spray chamber is observed. In comparison with water (upper uptake limit 7 μl min⁻¹) the higher possible uptake rate is due to the higher volatility of xylene than water. The nebulizer could work for a period of 2 days without interruption which proves the total sample consumption. The nebulisation was found to be remarkably stable (RSD of 1.5%) in the range 20–30 μl min⁻¹.

Fig. 2 Effect of the xylene flow rate on the performance of the interface. Solid line: intensity. Dashed line: relative standard deviation. (¹³⁸Ba, ¹¹⁵In, ¹³⁸Ba RSD, ¹¹⁵In RSD).

The optimisation of the working parameters was refined by total factorial experiments regarding the tip position, sampling depth and carrier gas flow. The most critical is the optimization of the nebulizer gas flow rate. As it can be seen from Fig. 3 the signal intensity maximum is achieved in a fairly narrow nebuliser gas flow rate, around 1.15 l min⁻¹. Note that although Ba and In only are depicted in Fig. 2 and 3 the effects are similar of all the elements analysed (Mo, Ni, V).

Fig. 3 Effect of the nebulizer gas on the performance of the interface. Solid line: intensity. Dashed line: relative standard deviation. (¹³⁸Ba, ¹¹⁵In, ¹³⁸Ba RSD, ¹¹⁵In RSD).

The oxygen flow rate was optimized at 8% in order to avoid the formation of the carbon deposits on the cones. Higher amounts of oxygen resulted in an increase of the oxide formation rate, increase in the plasma reflected power, and a decrease of the lifetime of the cones. The oxide formation rate measured as the BaO⁺/Ba⁺ intensity ratio was found to be between 1.2–1.3% for xylene flow rates in the range 15–30 μl min⁻¹.

Effect of the helium collision cell on the determination of Ni, V, and Mo

Table 2 in the ESI† summarises the possible interferences on the isotopes of Ni, V and Mo.¹⁸ The presence of chlorine, inherent to petroleum samples, is responsible for the most severe interferences.

For the determination of nickel the use of helium as collision gas was found to be necessary to eliminate the (NaCl)⁺ interference on ⁵⁸Ni. An addition of at least 3.2 ml min⁻¹ of helium was necessary to allow the measurement of a correct (0.3852) ⁶⁰Ni/⁵⁸Ni isotope ratio. Furthermore, the use of a collision cell (with helium at 3.2 ml min⁻¹) resulted in a complete suppression of the background noise at m/z 51 enabling an interference-free determination of vanadium. Indeed, in the absence of the collision cell the noise was evaluated at as much as 1000 cps. In contrast, no effect of the collision cell was observed for ⁹⁸Mo.

Simultaneous determination of V, Ni and Mo by μflow-injection–ICP MS

Fig. 4 shows a typical recording of data, obtained in the collision cell mode, for the simultaneous determination of V, Ni and Mo in a petroleum sample with calibration carried out by the method of standard additions at four levels. The average peak width at half height was 5–6 s, and at the base of 20–25 s. The memory effects were minimal; the background was achieved within 25–30 s from the moment of injection (cf. inset to Fig. 4). This led to a sample throughput exceeding 100 h⁻¹. This sample throughput allowed a complete calibration to be performed within 5 min which was distinctly faster than using methods employing a micronebulizer.³ The method of standard additions was found more suitable than the external calibration for quantification as for some, especially non-diluted samples, the matrix effect was sample-dependent.

Fig. 4 A typical recording of the standard addition calibration data during the simultaneous determination of V, Ni, and Mo. (1): NIST 1085b lubricating oil standard reference material, 10⁴ times diluted. (2–5): Standard addition of V, Ni, and Mo; (2): +10 ng ml⁻¹, (3): +20 ng ml⁻¹, (4): +40 ng ml⁻¹, (5): +80 ng ml⁻¹. The inset shows a zoom of a typical peak (marked with an asterisk).

The precision, RSD on ten replicate injections of a 25 ng ml⁻¹ multielemental solution, was 1.3, 1.6 and 1.0% for ⁵¹V, ⁶⁰Ni and ⁹⁸Mo, respectively, in the peak area mode, and 3.0, 2.9 and 2.1% for ⁵¹V, ⁶⁰Ni and ⁹⁸Mo, respectively, in the peak height mode. The worse precision in the peak height mode is likely to be due to difficulties with the definition of the peak apex because of narrow peaks.

The detection limits (without collision cell) were 0.9, 0.6, and 0.1 ng mL⁻¹ for ⁵¹V, ⁶⁰Ni and ⁹⁸Mo, respectively. This corresponds to the possibility of detection of absolute amounts of 2.25 pg of ⁵¹V, 1.5 pg of ⁶⁰Ni and 0.25 pg ⁹⁸Mo. When the collision cell was used the detection limits were 0.3, 0.6 and 0.8 ng ml⁻¹ for ⁵¹V, ⁶⁰Ni and ⁹⁸Mo, respectively. This represents an improvement for V but degradation for Mo. The detection limits were calculated for the standard solutions in xylene as the concentration producing a signal equivalent to three times the standard deviation of the noise.

The developed μflow-injection–ICP MS method allowed the analysis of a 1 : 10 diluted crude oil sample without any memory effect or a clogging of the nebulizer. This was mainly achieved by a small sample injection volume of 2.5 μl and a low carrier flow rate (30 μl min⁻¹) in combination with a total consumption interface. In contrast, conventional systems with a continuous sample introduction of 100 μl min⁻¹ required at least a dilution factor of 100 in order to avoid sample deposition in the spray chamber and to limit the amount of carbon entering the plasma.³

Validation of the μflow-injection–ICP MS method developed

The analytical setup developed was applied to the analysis of one crude oil sample. The results are shown in Table 1. The difference between the results obtained with and without collision cell is significant, except for molybdenum. The results without collision cell are lower for Ni and V. The comparison with the results obtained by a reference method (ICP AES) shows an agreement within at least 10% between the ICP AES results and ICP collision cell MS results for Ni and V.

Table 1 Results for the analysis of a crude oil sample (a: slope, b: intercept of the standard addition calibration). Three independent measurements

Element	ICP AES/ng ml^−1	Without collision cell
		a × 10^−4	b × 10^−4	Measured concentration/ng ml^−1	RSD (%)	R ^2
Mo		0.79 ± 0.15	1.76 ± 0.17	221 ± 56	25.3	0.991
Ni	9857	86.9 ± 84.9	4.82 ± 0.27	8850 ± 1154	13	0.995
V	6594	1822.3 ± 153.3	161.28 ± 4.91	5541 ± 572	10.3	0.999

---

		With collision cell
		a × 10^−5	b × 10^−4	Measured concentration/ng ml^−1	RSD (%)	R ^2
Mo		0.19 ± 0.11	0.49 ± 0.01	194 ± 13	6.7	0.993
Ni	9857	17.0 ± 2.4	0.76 ± 0.01	10837 ± 320	2.9	0.991
V	6594	17.9 ± 3.09	1.31 ± 0.02	6671 ± 183	1.7	0.995

---

The method was validated by the analysis of a NIST 1085b lubricating oil standard reference material. Table 2 demonstrates a considerable improvement of the recoveries (in comparison with the certified values) of Ni and V owing to the collision cell. Regarding molybdenum, the results obtained both without and with the collision cell are in good agreement with the certified value. These results confirm the need for the collision cell in the determination of nickel and vanadium by ICP MS.

Table 2 Results for the analysis of a NIST 1085b SRM lubricating oil (a: slope, b: intercept of the standard addition calibration). Three independent measurements

Element	Certified/ng ml^−1	Without collision cell
		a × 10^−5	b × 10^−4	Measured concentration/ng ml^−1	RSD (%)	R ^2	Recovery
Mo	300.6 ± 3.2	7.73 ± 0.06	2.59 ± 0.01	297 ± 3	1	0.9996	98.8
Ni	295.9 ± 7.4	1.01 ± 0.46	5.08 ± 0.11	198 ± 11	5.6	0.998	66.9
V	297.8 ± 4.6	43.98 ± 0.19	17.02 ± 0.04	258 ± 2	0.6	0.9999	86.6

---

		With collision cell
		a × 10^−5	b × 10^−4	Measured concentration/ng ml^−1	RSD (%)	R ^2	Recovery
Mo	300.6 ± 3.2	1.68 ± 0.12	5.58 ± 0.04	300 ± 3	0.9	0.9999	99.8
Ni	295.9 ± 7.4	2.39 ± 0.59	8.50 ± 0.01	281 ± 10	3.4	0.999	95
V	297.8 ± 4.6	3.83 ± 0.66	13.60 ± 0.01	281 ± 7	2.4	0.9997	94.4

---

Conclusions

The modification of the DS-5 micronebulizer allowed the development of a total consumption ICP MS sample introduction system capable of taking up xylene at a rate of 30 μl min⁻¹ without noticeable problems. The method developed allows the simultaneous determination of Mo, Ni and V at the sub-ng ml⁻¹ levels in petroleum-related samples with sample throughput an order of magnitude higher than the methods based on continuous sample uptake with comparable sensitivity. The use of a helium collision cell is essential for accurate measurements of nickel and vanadium.

Acknowledgements

The authors thank Dr H. Preud’homme for assistance with ICP MS measurements. The support of Agilent for the ICP MS platform is greatly acknowledged. The authors thanks Dr S. Dreyfus and Dr C. Pécheyran for valuable discussions. This research was partly funded by the IFP.

References

1. L. Lopez and S. Lo Monaco, Fuel, 2004, 83, 365–374.
2. L. Lopez, S. Lo Monaco and M. Richardson, Org. Geochem., 1998, 29, 613–629.
3. S. Dreyfus, C. Pecheyran, C. Magnier, A. Prinzhofer, C. P. Lienemann and O. F. X. Donard, J. ASTM Int., 2005, 2, 51–58.
4. M. C. Williams, J. ASTM Int., 1991, 96–104.
5. R. I. Botto and J. J. Zhu, J. Anal. At. Spectrom., 1994, 9, 905–912.
6. C. Duyck, N. Miekeley, C. L. Porto da Silveira and P. Szatmari, Spectrochim. Acta, Part B, 2002, 57, 1979–1990.
7. T. W. Avery, C. Chakrabarty and J. J. Thompson, Appl. Spectrosc., 1990, 44, 1690–1698.
8. R. I. Botto, Can. J. Anal. Sci. Spectrosc., 2002, 47, 1–13.
9. E. Björn and W. Frech, J. Anal. At. Spectrom., 2001, 16, 4–11.
10. E. Björn and W. Frech, Anal. Bioanal. Chem., 2003, 376, 274–278.
11. E. Björn, T. Jonsson and D. Goitom, J. Anal. At. Spectrom., 2002, 17, 1257–1263.
12. J.-L. Todolí and J.-M. Mermet, J. Anal. At. Spectrom., 2001, 16, 514–520.
13. M. Wind, A. Eisenmengerb and W. D. Lehmann, J. Anal. At. Spectrom., 2002, 17, 21–26.
14. J. Heilmann, S. F. Boulyga and K. G. Heumann, Anal. Bioanal. Chem., 2004, 380, 190–197.
15. K. Kahen, A. Strubinger, J. R. Chirinos and A. Montaser, Spectrochim. Acta, Part B, 2003, 58, 397–413.
16. D. Schaumlöffel, J. R. Encinar and R. Lobinski, Anal. Chem., 2003, 75, 6837–6842.
17. D. Schaumlöffel, P. Giusti, M. V. Zoriy, C. Pickhardt, J. Szpunar, R. Lobinski and J. S. Becker, J. Anal. At. Spectrom., 2005, 20, 17–21.
18. Finnigan MAT, ICP MS Interference-Mass Table, 1995.

---

Footnote

† Electronic supplementary information (ESI) available: Table 1—optimum instrumental parameters and Table 2—possible interferences in ICP MS for isotopes of interest. See DOI: 10.1039/b611542j

---