Non-spectral interference effects in inductively coupled plasma mass spectrometry using direct injection high efficiency and microconcentric nebulisation

Abstract

Non-spectral interference effects in inductively coupled plasma mass spectrometry (ICP-MS) were investigated for the direct injection high efficiency nebuliser (DIHEN) the large bore DIHEN (LB-DIHEN) and a microconcentric nebuliser (MCN) with a cyclone spray chamber. Interference effects from nitric and sulfuric acid, methanol and sodium nitrate solutions were studied. Eight elements with different ionisation potentials (IP) and atomic masses were monitored at various nebuliser gas flow rates. The processes giving rise to interference effects observed with the DIHEN were investigated by monitoring relative analyte sensitivities at different radial plasma positions. It was found that the average magnitude of interference effects decreased in the order MCN ≈ DIHEN > LB-DIHEN, showing that analytical performance is improved by removing the spray chamber. It is suggested that remaining interference effects observed for the DIHEN are caused by a matrix induced spatial redistribution of the aerosol in the plasma due to differences in droplet size distribution and density between water and the matrix solutions. In addition, the high plasma solvent load with the DIHEN results in a pronounced negative correlation between relative analyte signal magnitude and IP. For the LB-DIHEN, these effects were smaller, and other element specific interference effects dominated. For the eight elements monitored, element specific interference effects increased in the order MCN < DIHEN < LB-DIHEN. Transient acid effects were also investigated and found to be virtually eliminated by the use of a DIHEN instead of the MCN system. For the latter, stabilisation times of 3–10 min were necessary when changing the nitric acid concentration between 0.22 and 2.2 mol l⁻¹.

---

Introduction

Inductively coupled plasma atomic emission and mass spectrometry (ICP-AES and ICP-MS, respectively) are powerful and widely used techniques for trace element determinations. An important research area in ICP spectrometry concerns non-spectral interference effects, which are summarised in books^1,2 and reviews.^3–5 Interference effects may occur during sample introduction, in the plasma during atomisation, ionisation and excitation and, in addition for mass spectrometry, during extraction, focusing and transport of ions from the plasma to the detector.

Samples are most commonly introduced to the plasma as finely dispersed aerosols generated from aqueous solutions by a pneumatic nebuliser combined with a spray chamber. In self-aspirating sample introduction systems, solution uptake rate⁶ and aerosol characteristics^7,8 depend on density and viscosity of the sample solution and hence on type and concentration of matrix compounds. By using a pump, matrix effects on the sample uptake rate can be virtually eliminated. However, the drop size distribution of the primary aerosol generated by the nebuliser will still be dependent on solution properties. For solutions with low viscosity, i.e., solutions containing organic solvents, the Sauter mean diameter of the primary aerosol will decrease,⁸ increasing aerosol transport efficiency through the spray chamber. Normally, inorganic acids with higher viscosity give rise to a coarser primary aerosol compared with pure water.⁷ In contrast, the tertiary aerosol exiting the spray chamber often has a smaller mean droplet diameter. This is suggested to be a result of higher solution density for inorganic acids, increasing the gravitational settling in the spray chamber.^9,10 In addition, a depletion of analyte in the smaller droplets in the presence of ionic matrix species has been observed. This is believed to be caused by aerosol ionic redistribution processes.^9,11,12 It should be mentioned that exceptions to the above described effects have been observed (see references in ref. 5). Salt matrices affect tertiary aerosols in similar ways to inorganic acids. If introducing volatile solvents, excessive vapour from the spray chamber may cause a large solvent load of the plasma,¹³ which decreases atomisation and ionisation efficiency and may even cause the plasma to extinguish.

The processes in the spray chamber are also susceptible to transient matrix interference effects. These occur when the concentration of matrix, e.g., acids, is different in two successive samples. It has been proposed¹⁴ that transient effects are caused by matrix induced changes in the sample aerosol evaporation rate in the spray chamber, affecting the analyte transport efficiency to the ICP.

The magnitude of matrix effects in the plasma is strongly dependent on whether the plasma is operated under robust (high r.f. power, >1.2 kW, and low nebuliser gas flow rate, <0.7 l min⁻¹ for a conventional system), or weak (low r.f. power and high nebuliser gas flow rate) conditions.^15,16 When operating the plasma under weak conditions, matrix effects can be severe and a negative correlation between relative analyte sensitivity and the sum of elemental ionisation and excitation energies has been found in ICP-AES.^16–18 When operating the plasma under robust conditions, these effects from acids and other inorganic compounds are normally significantly reduced. Also, analyte signal enhancements have been observed in matrices containing an easily ionised element. These effects have been suggested to be a result of enhanced excitation efficiency due to an increased number of collisions between analyte and charged matrix species.^19,20 In ICP-AES, the Mg II (280 nm) to Mg I (285 nm) intensity ratio is often used to monitor the ionisation/excitation capability of the plasma.²¹ For ICP-MS no equivalent, generally accepted, procedure for monitoring species exists. The Ar₂⁺ species has been used in ICP-MS as a plasma diagnostic tool²² and as an internal standard.²³

Compared with ICP-AES, additional interference effects may occur in ICP-MS during focusing and transport of ions from the plasma to the mass spectrometer. If the plasma gas kinetic temperature is altered by the matrix, the sampled analyte ions will attain a different average kinetic energy and thereby the optimum focusing lens voltage will be shifted.²⁴ Further, space-charge effects may increase with matrix concentration, which decreases the analyte ion transport efficiency from the plasma to the detector.²⁵

In ICPs operated under robust conditions, the major part of non-spectral interference effects is believed to originate from the aerosol filtering processes in the spray chamber.²⁶ Recently, at least three different types of direct injection nebulisers that omit the spray chamber have been introduced. These are the direct injection nebuliser (DIN),²⁷ the direct injection high efficiency nebuliser (DIHEN)²⁸ and the large bore DIHEN (LB-DIHEN).²⁹ Several papers have described both fundamental aspects and applications for the DIN and the DIHEN, but no systematic investigation of non-spectral interference effects observed with these devices has been reported.

By omitting the spray chamber there is a possibility of significantly reducing non-spectral interferences. However, compared with conventional spray chamber systems, the plasma solvent load with DIHENs is high since 100% of the generated aerosol is introduced to the plasma. For example, at optimum sensitivity 85 µl of sample solution per minute are introduced to the plasma with the DIHEN, whereas for a cross-flow nebuliser combined with a double pass spray chamber, only ∼20 µl min⁻¹ are introduced (at a 1 ml min⁻¹ liquid flow rate and 2% transport efficiency). The higher solvent load when using direct injection nebulisation may increase interference effects during atomisation and ionisation in the plasma if robust conditions cannot be maintained.

This paper presents an investigation of non-spectral interference effects obtained in ICP-MS using a DIHEN, LB-DIHEN and, for comparison, a microconcentric nebuliser (MCN) with a cyclone spray chamber. The last system was chosen since the MCN is operated at similar liquid flow rates to the DIHENs, and the cyclone spray chamber is known for its relatively robust performance with respect to matrix composition.³⁰ In this work, interference effects from nitric and sulfuric acid, sodium nitrate and methanol matrices were investigated for the different sample introduction systems. Test elements were chosen to cover large ranges of isotope masses and first ionisation potentials. This approach was used to be able to distinguish between non-specific and mass and ionisation potential dependent interference effects.

Experimental

Instrumentation

A Perkin-Elmer Elan 6000 ICP-MS system (Perkin-Elmer SCIEX, Thornhill, ON, Canada) was used. The isotopes monitored were ⁷Li⁺, ⁹Be⁺, ⁶⁸Zn⁺, ⁶⁹Ga⁺, ¹¹⁴Cd⁺, ¹¹⁵In⁺, ²⁰²Hg⁺, ²⁰³Tl⁺ and ⁷⁶(Ar₂)⁺ with a dwell time of 200 ms per isotope in the peak hopping mode. The number of sweeps, readings and replicates were set to 2, 2 and 5, respectively, giving a total measurement time of 36.5 s. The sample introduction systems studied were a DIHEN (Model DIHEN-170-AA, J. E. Meinhard Associates, Santa Ana, CA, USA), an LB-DIHEN (Model DIHEN-30-AA, J. E. Meinhard Associates) and an MCN (MCN-100, Model M-2, CETAC Technologies, Omaha, NE, USA) combined with a quartz cyclone spray chamber (Model 409–23, AHF Analysentechnik, Tübingen, Germany). The aerosol from the cyclone spray chamber was introduced to the plasma via a 2 mm id injector tube. Standard nickel or platinum sampler and skimmer cones with orifice diameters of 1.1 and 0.9 mm, respectively, were used. The sampling depth was kept at 11 mm above the load coil.

The ICP-MS was tuned and the auto lens optimised with water standards of the eight analyte elements, containing 1% w/w nitric acid. Two separate tunings of the mass spectrometer were made, one with the DIHEN and one with the MCN. For the LB-DIHEN, the tuning performed with the DIHEN was used. The auto lens voltages of the Elan 6000 and the X–Y adjustment were optimised for each sample introduction system. The rf power was set to 1500 W to achieve robust plasma conditions and optimum sensitivity for the DIHENs. The liquid flow rate was kept at 85 µl min⁻¹, which is the optimum flow rate for the DIHEN.²⁸ Sample solutions were introduced by a peristaltic pump (Model RS 422, Gilson, Middleton, WI, USA). The nebuliser gas flow rate was varied between 0.2 and 1.2 l min⁻¹ and the back-pressure was kept at 10 bar for the DIHEN and 5.2 bar for the LB-DIHEN and MCN. If not stated otherwise the dead volumes of the DIHEN and LB-DIHEN were reduced by inserting a 30 cm × 250 µm id × 360 µm od silica capillary (Part 2000024, Polymicro Technologies, Phoenix, AZ, USA) into each nebulizer to the position where the capillary tapers. Extreme caution should be exercised when inserting the silica capillary so as not to break the sample capillary of the DIHEN.

Reagents

Working standard solutions containing 10 ng ml⁻¹ of the eight analyte elements were prepared by diluting 1000 µg ml⁻¹ stock standard solutions containing 1% nitric acid with 18 MΩ doubly de-ionised water. From this, standard solutions containing ∼1.2 × 10⁻⁶ mol l⁻¹ nitric acid were obtained. These will hereafter be referred to as water solutions. Matrix solutions of methanol (HPLC gradient grade, J. T. Baker, Deventer, The Netherlands), sodium nitrate (pro analysi, Merck, Darmstadt, Germany), nitric acid (sub-boiled distilled, 65% J. T. Baker) and sulfuric acid (95–97%, J. T. Baker) were also prepared.

Procedure

For the interference studies, four pairs of elements, covering the atomic mass range 7–203 u, were selected. To be able to distinguish between non-specific and mass and ionisation potential dependent interference effects, the elements in each pair had similar atomic masses but very different first ionisation potentials (IP). One group of elements with low IP (eV), Li (5.39), Ga (6), In (5.785) and Tl (6.106), and one group with high IP, Be (9.32), Zn (9.391), Cd (8.991) and Hg (10.43) were selected. IPs < 7 eV were considered as "low" and the elements were chosen so that the difference in IP within each pair was >3 eV. In addition, Ar⁺₂was monitored to diagnose plasma ion-isation conditions.^22,23

Matrix solutions of 0.9 and 3.6 mol l⁻¹ nitric acid, 0.9 mol l⁻¹ sulfuric acid, 100 and 1000 µg ml⁻¹ sodium as sodium nitrate and 5 and 20% v/v methanol were used and contained 10 ng ml⁻¹ of the analyte elements. De-ionised water standards and blanks were also used. Owing to potential problems with analyte instability for aqueous standard solutions, the analytes were added immediately before analysis. The analyte intensities for each solution were measured as a function of nebuliser gas flow rate, and for the DIHEN also as a function of the radial plasma position. All signals were blank corrected with corresponding matrix blank solution for each nebuliser gas flow rate and radial position setting. Before the start of analysis, the instrument was stabilised for at least 30 min. When changing matrix solutions, the system was flushed with matrix blank for 5 min. Between each change of nebuliser gas flow setting (increments of 0.1 l min⁻¹), the system was allowed to stabilise for 45 s.

Results and discussion

Table 1 presents an overview of the magnitude of steady state interference effects observed at optimum nebuliser gas flow rates for each sample introduction system, 0.2 and 1.1 l min⁻¹ for the DIHENs and the MCN, respectively. Table 2 presents the variance for the observed interference effects within each group of four elements (with similar IP) in each matrix. A higher variance should correspond to an increased presence of element specific interference effect. It should be observed that the influence of IP on the variance of interference effects is not taken into account in Table 2.

Table 1 Summary of steady state interference effects observed for elements with low and high IP at nebuliser gas flow rates optimised for water solutions. The values represent mean relative signals compared with water

Nebuliser	IP	Nitric acid (0.9 mol l^−1)	Sulfuric acid (0.9 mol l^−1)	Methanol (5% v/v)	Sodium (100 µg ml^−1)	Absolute average deviation^a
a Absolute average magnitude of observed interference effects.
DIHEN	Low	1.44	1.07	0.84	1.52	0.30
	High	0.39	0.28	0.69	1.37	0.50
LB-DIHEN	Low	1.23	1.14	0.88	0.68	0.20
	High	0.78	0.77	1.13	0.70	0.22
MCN	Low	0.93	0.55	0.28	1.00	0.31
	High	0.59	0.39	0.28	0.86	0.47

---

Table 2 Variance of observed interference effects for each group of four elements obtained with the three sample introduction systems. The highest variance obtained for each group in each matrix is given in bold. Variance calculated as

Nebuliser	IP	Nitric acid (0.9 mol l^−1)	Sulfuric acid (0.9 mol l^−1)	Methanol (5% v/v)	Sodium (100 µg ml^−1)
DIHEN	Low	0.113	0.091	0.040	0.032
	High	0.022	0.005	0.008	0.016
LB-DIHEN	Low	0.197	0.295	0.002	0.007
	High	0.044	0.285	0.029	0.057
MCN	Low	0.005	0.013	0.012	0.003
	High	0.014	0.008	0.031	0.038

---

Steady state matrix effects with the DIHEN

Acid and methanol effects. For the DIHEN, operated under optimum conditions with respect to sensitivity²⁸ (85 µl min⁻¹ liquid flow rate and 0.2 l min⁻¹ nebuliser gas flow rate), the relative analyte signals were lower for sulfuric acid (0.9 mol l⁻¹) than nitric acid (0.9 mol l⁻¹) (see Fig. 1). These results agree qualitatively with earlier studies, using a conventional nebuliser spray chamber, where for a given acid concentration the interference effects increased in the order nitric ≈ hydrochloric ≈ perchloric < phosphoric ≈ sulfuric acid.¹⁰

Fig. 1 Relative analyte signals in acids compared with water solutions at 0.2 min⁻¹ nebuliser gas flow rate for the DIHEN. First ionisation potential (eV) is given for each element.

Compared with water, the relative Ar⁺₂ signals were 87 and 38% in the nitric and sulfuric acid matrices, respectively. This indicates that analyte ionisation will be more suppressed for sulfuric than for nitric acid.

There is a pronounced negative correlation between the analytes' IPs and their relative signals in acid matrices, showing that robust plasma conditions are not obtained at the optimum liquid flow rate.

As can be seen in Fig. 2(a) and (b), the signals for the analytes, exemplified here by lithium, and Ar⁺₂ decreased in both water and sulfuric acid for increased nebuliser gas flow rates. It is interesting that above a gas flow rate of 0.5 l min⁻¹, the Li⁺ signal was higher in the presence of sulfuric acid than in water [see Fig. 2(a)]. This trend was not observed for Ar⁺₂ [see Fig. 2(b)]. The Ar⁺₂ signal was generally less affected than the analyte signals by increases in the nebuliser gas flow rate, since the formation of Ar⁺₂ ions will not be affected by decreased aerosol residence time in the plasma, but only by plasma cooling effects. The reasons for the increase in relative analyte signals at higher nebuliser gas flow rates will be discussed below in connection with spatially resolved measurements.

Fig. 2 Lithium (a) and argon dimer (b) signals for the DIHEN as a function of nebuliser gas flow rate.

With the DIHEN, except for indium, relative analyte signals of 65–80% were observed for solutions containing 5% v/v methanol at the optimum nebuliser gas flow rate (see Fig. 3). For 20% v/v methanol, all relative analyte signals decreased further. Compared with water solutions, relative Ar⁺₂ signals of 30 and 5% were obtained at a 0.2 l min⁻¹ nebuliser gas flow rate for samples containing 5 and 20% methanol, respectively. This indicates severe absorption of plasma energy by the methanol, decreasing atomisation and ionisation efficiency. As the nebuliser gas flow rate was increased from 0.2 to 0.6 l min⁻¹, relative analyte signals of 75–94% were obtained (except for indium) for 5% v/v methanol. At even higher gas flow rates up to 0.8 l min⁻¹, the relative signals remained unchanged. The interference effects in methanol solutions were not correlated with isotope mass or ionisation potential. The residence time of analyte species in the plasma is a particularly critical parameter when samples containing organic solvents are analysed.³¹ As the dissociation of the solvent molecules absorbs plasma energy,³² the required residence time to obtain a given degree of atomisation and ionisation of analyte species will be increased. This requirement is approached by decreasing the nebuliser gas flow rate. In addition, at low nebuliser gas flow rates higher temperatures will exist in the central channel of the plasma that will increase the degree of analyte ionisation. As mentioned above, the relative analyte signals in 5% methanol solutions increased with increase in nebuliser gas flow rate, in contrast to what could be expected from the above discussion. Obviously, other factors, such as droplet size distribution as a function of nebuliser gas flow rate in the presence and absence of methanol, discussed below, are of importance for the interference effects.

Fig. 3 Relative analyte signals for methanol solutions compared with water at 0.2 l min⁻¹ nebuliser gas flow rate for the DIHEN.

Radially resolved measurements. To understand better the mechanisms of the previously described interference effects, measurements were performed at different radial plasma positions. Although the lateral movement of the mass spectrometer, and hence the sampler cone, relative to the plasma torch is limited to about ±2 mm, the distribution of ions over the corresponding cross-sectional area might still provide valuable information when dealing with DIHENs. For the results discussed here, measurements were performed at 0, 1 and 2 mm off the central axis of the torch. Fig. 4(a)–(d) shows sensitivity as a function of the radial torch position for lithium and beryllium in water, 0.9 mol l⁻¹ nitric acid and 5% v/v methanol solutions at two different nebuliser gas flow rates. The trends observed for lithium (low IP) and beryllium (high IP) were representative, with respect to ionisation potential, for all elements, although minor differences between different elements were found.

Fig. 4 Lithium and beryllium signals at different radial plasma positions for the DIHEN for 0.2 (a, b) and 0.6 (c, d) l min⁻¹ nebuliser gas flow rates. (a, c) Lithium; (b, d) beryllium.

The results for lithium at the optimum nebuliser gas flow rate [Fig. 4(a)] indicate that for elements with low IP, the observed acid and methanol interference effects are mainly a result of a matrix induced spatial redistribution of the aerosol in the plasma. The nitric acid gives rise to increased mean droplet diameter and droplet density compared with water.¹⁰ As a result, the nitric acid aerosol is more confined to the central channel of the aerosol beam and the plasma where analyte ion concentrations are increased. In line with this, on-axis the signals were enhanced and off-axis were suppressed for nitric acid compared with water solutions. For the methanol solutions the situation is the opposite, and smaller mean droplets with lower density compared with water are obtained⁸ and the aerosol is therefore spread over a larger cross-sectional area. As a result, compared with water, on-axis suppression and off-axis enhancement of signals were obtained.

For elements with high ionisation potential, exemplified here by beryllium, there was an additional ionisation interference in the acid matrices as discussed above, giving on-axis signal suppression at the optimum nebuliser gas flow rate [see Fig. 4(b)].

At higher nebuliser gas flow rates, the aerosol is likely to be less confined, and therefore more evenly distributed over the cross-sectional area of the aerosol beam. Therefore, for all solutions, analyte sensitivity is less dependent on the radial position of the plasma. As can be seen in Fig. 4(c), the spatial signal profiles in water and methanol matrices levelled out much more than for the nitric acid matrix, which explains why larger analyte signal enhancements were observed in the acid at higher nebuliser gas flow rates. As also can be seen in Fig. 4(c) and (d), at a nebuliser gas flow rate of 0.6 l min⁻¹, the signal profiles are very similar for water and methanol matrices. For elements with high IP the absolute analyte signals in water and 5% v/v methanol solutions were higher at 1 mm off-axis than on-axis. This is probably a result of the flat distribution of analyte species and the higher temperature off-axis,^32,33 increasing the ionisation efficiency.

Fig. 5(a) and (b) shows the normalised sensitivities for lithium and beryllium in sulfuric acid for 0.2 and 0.6 l min⁻¹ nebuliser gas flow rates as a function of the radial plasma position. For comparison, the signals for nitric acid are also included in Fig. 5. The figure shows that a relatively flat spatial signal profile was obtained for sulfuric acid at the optimum nebuliser gas flow rate (0.2 l min⁻¹), resulting in a smaller off-axis signal decrease compared with water and nitric acid solutions [see Fig. 4(a) and (b)]. This cannot be explained by the hypothesis of radial aerosol redistribution since sulfuric acid is known to give rise to aerosols with larger mean droplets of higher density compared with water and nitric acid.¹⁰ For the 0.6 l min⁻¹ nebuliser gas flow rate, the radial signal profile in sulfuric acid was almost the same as for 0.2 l min⁻¹. As a consequence, the signal profile in sulfuric acid was, at the higher nebuliser gas flow rate, more confined to the central axis than for the other matrices (see Fig. 5), in line with the spatial redistribution of aerosol hypothesis presented above. It was also found that the spatial signal profile in sulfuric acid was much steeper for elements with low IP than for those with high IP. Further work is necessary to understand fully the mechanisms leading to interference effects in the presence of sulfuric acid.

Fig. 5 Normalised analyte signals for lithium (a) and beryllium (b) at different radial plasma positions for the DIHEN in 0.9 mol l⁻¹ sulfuric and nitric acid.

The matrix induced spatial aerosol redistribution effects observed for the DIHEN are likely to be more pronounced for direct injection nebulisation than spray chamber systems. With the latter, only droplets with a diameter smaller than the spray chamber cut-off diameter are introduced into the plasma via an injector tube of typically 2 mm id. With the DIHEN, the entire primary aerosol is introduced to the plasma via a 100 µm id sample capillary. As a result, for the DIHEN, even a small spatial aerosol redistribution will greatly affect analyte sensitivity. With a conventional system, the aerosol is distributed over a larger cross-sectional area, and the above discussed effects should be small compared with other types of interference effects generated in the spray chamber.

Effect of sodium nitrate. When introducing solutions containing 100 µg ml⁻¹ sodium with the DIHEN, at the optimum nebuliser gas flow rate, the analyte sensitivity increased 1–2-fold compared with water solutions (see Fig. 6). For the highest nebuliser gas flow rate, 0.8 l min⁻¹, the signals increased 3–5-fold. It should be noted that by increasing the nebuliser gas flow rate from 0.2 to 0.8 l min⁻¹, the analyte sensitivity in water standards decreased on average by two orders of magnitude, hence the absolute analyte signals in sodium nitrate solutions are much smaller at the higher nebuliser gas flow rate. For 1000 µg ml⁻¹ sodium, lower absolute and relative analyte signals, compared with 100 µg ml⁻¹ sodium, were obtained, but the same trend of increased relative signals with nebuliser gas flow rates was observed.

Fig. 6 Relative analyte signals for 100 µg ml⁻¹ sodium solution compared with water as a function of nebuliser gas flow rate for the DIHEN.

Results from spatially resolved measurements with 100 µg ml⁻¹ sodium solutions could in part explain the observed effects. The radial signal profiles for the sodium matrix were similar to those for nitric acid, resulting in on-axis analyte signal enhancements as a result of a more confined analyte aerosol. This effect was, however, much smaller for the sodium matrix, although the analyte signal increased much more at higher nebuliser gas flow rates than for nitric acid solutions. In addition to aerosol redistribution, we found that the Ar⁺₂ signal increased significantly in the presence of sodium, especially at high nebuliser gas flow rates and at higher sodium concentrations (see Fig. 7). This phenomenon was not observed for the other matrices investigated, and indicates that sodium has a significant positive effect on the ionisation efficiency of the plasma when solutions are introduced with the DIHEN.

Fig. 7 Relative argon dimer signal for sodium solutions compared with water solution as a function of nebuliser gas flow rate with the DIHEN.

A positive correlation between isotope mass and relative signal in sodium nitrate solutions was also found, especially for 1000 µg ml⁻¹ sodium. This could be explained by smaller space-charge interference effects for elements with higher masses. We conclude that sodium affects the analyte sensitivity in a complex way. It seems to increase the ionisation efficiency, it increases the analyte transport efficiency to the sampler cone by a redistribution of the aerosol but, as a result of space-charge effects, it also decreases the ion transport efficiency to the detector. The relative importance of these effects probably depends on sodium concentration, isotope mass and IP of the analyte element and operating conditions of the instrument.

Steady state matrix effects with the LB-DIHEN

The design and analytical performance of the DIHEN and LB-DIHEN have been described and compared previously.²⁹ The most important physical differences are larger dimensions of the sample capillary and gas annulus area of the LB-DIHEN. As a result, at an optimum nebuliser gas flow rate, the LB-DIHEN generates an aerosol with droplets of larger Sauter mean diameter and a wider size distribution. Also, the mean velocity of droplets is lower for the LB-DIHEN. As the nebuliser gas flow rate is increased, the properties of the aerosols generated by the two DIHENs become more similar. As can be seen in Table 1 and Fig. 8(a), smaller IP correlated signal suppressions were observed in acid matrices with the LB-DIHEN compared with the DIHEN. These improvements might be explained by a different aerosol droplet size distribution and longer analyte residence time in the plasma due to lower mean axial velocity of the droplets compared with the DIHEN.²⁹

Fig. 8 Relative analyte signals in (a) acid and (b) methanol solutions compared with water at 0.2 l min⁻¹ nebuliser gas flow rate for the LB-DIHEN.

The effects of 5% v/v methanol were moderate at optimum nebuliser gas flow rate [see Table 1 and Fig. 8(b)] and the relative signals were generally higher compared with the DIHEN. Compared with water, the signals were slightly higher for elements with high IP. These results indicate that, compared with the DIHEN, the spatial aerosol redistribution in the presence of methanol is smaller for the LB-DIHEN. It should be mentioned that the inner diameter of the sample capillary is about three times wider for the LB-DIHEN, which might reduce matrix induced aerosol redistribution effects. Also, the longer analyte residence time might increase atomisation and ionisation efficiency. For the 20% methanol solution, relative signals of 40–90% compared with water solution were obtained.

Table 1 shows that the average magnitudes of interference effects observed in this work were smallest for the LB-DIHEN compared with the other systems. However, from Table 2 it can be seen that generally larger variations in interference effects for the different elements were obtained with the LB-DIHEN. These results are most likely due to the lack of spray chamber effects and the relatively small spatial aerosol redistribution effects. As a consequence, the absolute magnitudes of interference effects are reduced, and the remaining effects are more dependent on the physical-chemical properties of the elements.

Steady state matrix effects with the MCN system

A description of the design and performance of the microconcentric nebuliser, MCN-100, used here can be found elsewhere.³⁴

With the MCN, a moderate negative correlation between relative analyte signals and IP was observed for acid solutions (see Fig. 9). Again, the analyte signals were more severely suppressed for 0.9 mol l⁻¹ sulfuric acid than for the same concentration of nitric acid. Analogously to the DIHENs, the relative signals in the acid matrices increased with increase in nebuliser gas flow rate, but this effect was much smaller for the MCN system.

Fig. 9 Relative analyte signals in acids compared with water solutions at 1.1 l min⁻¹ nebuliser gas flow rate with the MCN system.

The results for methanol demonstrate the importance of analyte plasma residence time in the presence of an organic matrix. For methanol solutions, optimum sensitivity is shifted towards lower nebuliser gas flow rates. As this was increased from 1.0 to 1.2 l min⁻¹, for 5% v/v methanol, the average relative analyte signals decreased from 135 to 18% and the Ar⁺₂ signal decreased from 97 to 9%. The lower relative Ar⁺₂ signal is probably caused by increased methanol solvent load at higher nebuliser gas flow rates.³⁵

In contrast to the DIHEN, the sodium matrix had a relatively element uniform effect on analyte signals, and no statistically significant correlation with analyte IP or isotope mass was found. The interference caused by 100 µg ml⁻¹ sodium was generally small at the optimum nebuliser gas flow rate, 1.1 l min⁻¹, and relative signals between 90 and 110% were obtained. For 1000 µg ml⁻¹ sodium, relative signals of 45–70% compared with pure water were obtained. For zinc, relative signals of 60 and 30% were obtained for 100 and 1000 µg ml⁻¹ sodium, respectively. At a higher nebuliser gas flow rate, 1.2 l min⁻¹, relative average signal enhancements of about 40% compared with water solutions were observed for 100 µg ml⁻¹ sodium (except for zinc). The Ar⁺₂ signal decreased to 80% in the presence of sodium nitrate, indicating that sodium has no positive effect on plasma ionisation efficiency in this case.

For the MCN system, the average magnitude of interferences was larger than for the DIHENs (see Table 1). However, generally more element uniform effects were observed (see Table 2). The dominating interferences are likely to be caused by matrix induced changes of aerosol transport efficiency through the spray chamber, which are not element specific.

Transient acid effects

Transient matrix effects will occur when a change in matrix composition or concentration for two successive samples affects the analyte response. If data acquisition for a new sample (with a different matrix composition) is started before steady state conditions have been established, the precision and accuracy of the corresponding measurement will deteriorate. Normally, the design of the sample introduction system dictates the time necessary to reach steady state conditions. A spray chamber with a large dead volume and surface area covered with a relatively large volume of liquid is susceptible to large transient matrix effects. These have been thoroughly studied in ICP spectrometry for both double pass¹⁴ and cyclone spray chambers.³⁶

Fig. 10 shows beryllium signals for the MCN system and the DIHEN with changes in nitric acid concentration from 0.22 to 2.2 and from 2.2 to 0.22 mol l⁻¹. On increasing the acid concentration the signals were within the fluctuations of the steady state signal after first 3.5 min for the MCN and after 0.3 min with the DIHEN. With the MCN, a stabilisation time of more than 10 min was needed when the acid concentration was decreased. For the DIHEN, no transient effects were observed, but a linear drift of the signal was observed during 2 min.

Fig. 10 Beryllium signals when changing nitric acid concentrations. Solid traces, 0.22 to 2.2 mol l⁻¹ nitric acid; dashed traces 2.2 to 0.22 mol l⁻¹ nitric acid. (a) MCN system; (b) DIHEN.

The much smaller transient matrix effects observed with the DIHEN would permit the use of shorter analysis times for successive samples with large differences in matrix concentration. It should be observed that these results for the DIHEN were obtained without a reduced nebuliser dead volume (see Experimental). The fact that transient acid effects were absent for the DIHEN is in line with the theory that transient matrix effects originate from changes in tertiary aerosol volume flux.¹⁴

Conclusions

For the matrices and analytes investigated, large interference effects were obtained with all three sample introduction systems. However, at optimum nebuliser gas flow rates, the absolute average signal changes brought about by the matrices were smallest with the LB-DIHEN and virtually the same with the DIHEN and the MCN system (see Table 1). Generally, the relative signals were lower for elements with high IP.

The results show that direct injection nebulisation improves analytical performance since spray chamber related effects are eliminated. At the same time, with the DIHENs, interference effects caused by matrix induced redistribution of the aerosol in the plasma increase. Depending on the drop size distribution and droplet density of the aerosols from the matrix compared with water solution, the aerosols will be more (for large droplets with high density) or less (small droplets with low density) confined to the central axis, giving signal enhancements or suppressions, respectively.

Relatively large differences in interference effects were observed between the two types of DIHENs, showing the crucial importance of mean values and distributions of the aerosol droplets' diameter and velocity.

With the DIHEN, the dominating interference effects were caused by matrix induced aerosol redistribution and signal suppressions correlated with the analytes' IP. With the LB-DIHEN, these effects were smaller and other, element specific, interference effects were of major importance. As a consequence, selection of internal standards for correction for non-spectral interference effects should be more critical with the DIHENs, and especially the LB-DIHEN, compared with conventional systems. Therefore, if possible, isotope dilution calibration is advisable when using either of the two DIHENs.

The Ar⁺₂ species is not generally applicable for the prediction or correction of matrix induced changes in analyte sensitivity. Clearly, a simple and reliable procedure to diagnose the plasma's ionisation efficiency in ICP-MS would be very useful.

Transient matrix effects are significantly reduced by utilising direct injection nebulisation. As a consequence, wash-out and read delay times can be significantly reduced when analysing samples that vary in matrix composition.

It should be noted that all interference effects investigated in this work can be corrected for by using isotope dilution and some by using internal standardisation. Although not investigated in this work, the use of a DIHEN eliminates potential interferences of species specific responses on total determinations of analytes.³⁷ Such interferences cannot be easily corrected for.

Acknowledgements

The Swedish Natural Science Research Council and Knut and Alice Wallenberg Foundation financed this research. The authors thank J. E. Meinhard Associates for the loan of a LB-DIHEN, Professor Akbar Montaser of George Washington University for technical support, Johanna Qvarnström of Umeå University for valuable discussions and James Snell of Umeå University for linguistic revision of the manuscript.

References

1. Inductively Coupled Plasmas in Analytical Atomic Spectrometry, ed. A. Montaser, D. W. Golightly, Wiley-VCH, New York, 1992..
2. Inductively Coupled Plasma Mass Spectrometry, ed. A. Montaser, Wiley, New York, 1998..
3. E. H. Evans and J. J. Giglio, J. Anal. At. Spectrom., 1993, 8, 1.
4. H. Vanhoe, J. Trace. Elem. Elect. Electrolytes Health Dis., 1993, 7, 131.
5. J. L. Todolí and J. M. Mermet, Spectrochim. Acta, Part B, 1999, 54, 895.
6. A. W. Boorn and R. F. Browner, in Inductively Coupled Plasma Emission Spectroscopy. Part II, ed. P. W. J. M. Boumans, Wiley, New York, 1987, p. 244..
7. S. Nukiyama and Y. Tanasawa, Trans. Soc. Mech. Eng. Jpn., 1939, 5, 68.
8. A. Canals, V. Hernandis and R. F. Browner, J. Anal. At. Spectrom., 1990, 5, 61.
9. M. Marichy, M. Mermet and J. M. Mermet, Spectrochim. Acta, Part B, 1990, 45, 1195.
10. A. Canals, V. Hernandis, J. L. Todolí and R. F. Browner, Spectrochim. Acta, Part B, 1995, 50, 305.
11. R. F. Browner, A. W. Boorn and D. D. Smith, Anal. Chem., 1982, 54, 1411.
12. J. L. Todolí, J. M. Mermet, A. Canals and V. Hernandis, J. Anal. At. Spectrom., 1998, 13, 55.
13. F. J. M. J. Maessen, P. J. H. Seevesens and G. Kreuning, Spectrochim. Acta, Part B, 1984, 39, 1171.
14. I. I. Stewart and J. W. Olesik, J. Anal. At. Spectrom., 1998, 13, 843.
15. C. Dubuisson, E. Poussel, J. M. Mermet and J. L. Todolí, J. Anal. At. Spectrom., 1998, 13, 63.
16. A. Fernández, M. Murillo, N. Carrión and J. M. Mermet, J. Anal. At. Spectrom., 1994, 9, 217.
17. F. J. M. J. Maessen, J. Blake and J. L. M. De Boer, Spectrochim. Acta, Part B, 1982, 37, 517.
18. E. Yoshimura, H. Suzuki, S. Yamazaki and S. Toda, Analyst, 1990, 115, 167.
19. C. Veillon and M. Margoshes, Spectrochim. Acta, Part B, 1968, 23, 503.
20. M. W. Blades and G. Horlick, Spectrochim. Acta, Part B, 1981, 36, 881.
21. J. M. Mermet, Anal. Chim. Acta., 1991, 250, 85.
22. F. Vanhaecke, G. Galbacs, S. Boonen, L. Moens and R. Dams, J. Anal. At. Spectrom., 1995, 10, 1047.
23. D. Beauchemin, J. W. McLaren and S. S. Berman, Spectrochim. Acta, Part B, 1987, 42, 467.
24. I. I. Stewart and J. W. Olesik, J. Anal. At. Spectrom., 1998, 13, 1313.
25. G. R. Gillson, D. J. Douglas, J. E. Fulford, K. W. Halligan and S. D. Tanner, Anal. Chem., 1988, 60, 1472.
26. J. M. Mermet, J. Anal. At. Spectrom., 1998, 13, 419.
27. K. E. Lawrence, G. W. Rice and V. A. Fassel, Anal. Chem., 1984, 56, 289.
28. J. A. McLean, H. Zhang and A. Montaser, Anal. Chem., 1998, 70, 1012.
29. B. W. Acon, J. A. McLean and A. Montaser, Anal. Chem., 2000, 72, 1885.
30. S. Maestre, J. Mora, J. L. Todolí and A. Canals, J. Anal. At. Spectrom., 1999, 14, 61.
31. D. J. Kalnicky, R. N. Kniseley and V. A. Fassel, Spectrochim. Acta, Part B, 1975, 30, 511.
32. D. G. Weir and M. W. Blades, J. Anal. At. Spectrom., 1994, 9, 1311.
33. T. Hasegawa and H. Haraguchi, Spectrochim. Acta, Part B, 1985, 40, 1505.
34. F. Vanhaecke, M. Van Holderbeke, L. Moens and R. Dams, J. Anal. At. Spectrom., 1996, 11, 543.
35. P. Barrett and E. Pruszkowska, Anal. Chem., 1984, 56, 1927.
36. J. L. Todolí and J. M. Mermet, J. Anal. At. Spectrom., 2000, 15, 863.
37. E. Björn, J. P. Snell and W. Frech, 1999 FACSS Conference, Vancouver, BC, Canada, October 24–29, 1999, Abstract 139.

---