Section 1

Sample Preparation and Introduction

SAMPLE INTRODUCTION, PLASMA-SAMPLE INTERACTIONS, ION TRANSPORT AND ION-MOLECULE REACTIONS: FUNDAMENTAL UNDERSTANDING AND PRACTICAL IMPROVEMENTS IN ICP-MS

John W. Olesik, Carl Hensman, Savelas Rabb and Deanna Rago

Laboratory for Plasma Spectrochemistry, Laser Spectroscopy and Mass Spectrometry, Department of Geological Sciences, Ohio State University, Columbus, OH 43210, USA

1 INTRODUCTION

Improved fundamental understanding of the processes that control ICP-MS signals can lead to practical improvements in analytical capabilities. Here we discuss high efficiency analyte transport at sample uptake rates as high as 0.8 mL/min and the unique behavior of selenium which when present at high concentrations is vaporized before other elements. We also explore the use of ion-molecule reactions in a dynamic reaction cell to overcome spectral overlaps and examine the ability of the quadrupole-based cell to prevent formation of undesired product ions.

Droplet-droplet collisions and coagulation in the spray chamber appear to be the main processes that limit analyte transport efficiency. When the sample uptake rate is increased, transport efficiency naturally decreases. By promoting evaporation within the spray chamber, it is possible to attain virtually 100% sample transport efficiency. The deleterious effects of solvent vapor loading prevent use of high analyte transport rates unless most of the sample solvent vapor is removed prior to entering the plasma. Even water vapor loads greater than about 60 mg/min can cause deleterious effects in the plasma. However, if solvent vapor is efficiently removed, analyte from up to 0.8 mL/min of sample can be introduced into the ICP and ionized.

Most of the mass dependent matrix effects that occur when the sample contains high concentrations of dissolved solids to be due to space charge induced loss of ions. However, Farnsworth has reported measurements of ion concentrations between the sampler and skimmer orifices that suggest that significant matrix effects are occurring during sampling of ions in contrast to previous assumptions.

Within the ICP itself, ions are generated earlier closer to the load coil) when the sample contains high concentrations of dissolved solids. This is consistent with shifts in emission and laser induced fluorescence profiles Lazar and Farnsworth recently proposed that the earlier appearance of ions is due to an increase in the size of the desolvated particle, so that the time required for desolvation of a fixed size aerosol droplet is smaller when the dissolved solids concentration is high. Here we confirm that the initial appearance of ions in the plasma is consistent with the proposed explanation for a variety of different dissolved solids. Previously, we have seen no evidence of element dependent initial appearance of sample ions in the plasma; all analyte ions appeared at the same time. However, the behavior due to the presence of high concentrations of selenium oxide is very different. We show that this is due to vaporization of the selenium prior to vaporization of analyte species in the particle.

Spectral overlaps have been a major problem in ICP-MS, particularly for quadrupole mass spectrometers, since inception of the technique. Several approaches have been used to overcome potential errors due to spectral overlaps. Sector based mass spectrometers can provide resolution of 10,000 or more to separate many, but not all, spectral overlaps. Mathematical corrections can be used if the ratio of the spectral overlap ion to the analyte ion signals is not too large. Removal of the solvent before the sample is introduced into the plasma and mixed gas plasmas have also been used to reduce spectral overlaps. "Cold" or "cool" plasmas have also been used although this approach can suffer from much more severe chemical matrix effects than normal or "hot" plasmas.

Attempts to use collisionally induced dissociation (CID) to break molecular ions after sampling from the ICP in reaction or collision cells were not successful because such high energies are required to break most molecular ion bonds that scattering losses are too large. However, Rowan and Houk described charge transfer reactions that likely occurred due to low-energy collisions while Douglas reacted O2 with Ce+ to form Ceo+.

Ion-molecule reactions, promoted in a pressurized cell after sample ions pass through the skimmer and ion optics, can very efficiently remove many ions that would result in spectral overlaps in ICP-MS. Barinaga, Eiden and Koppenaal described the use of ion-molecule reactions (using H2 or H2O) in an ion trap or reaction cell to overcome argide ions. Turner et al described an rf-only hexapole "collision cell" originally proposed to remove argide ions due to collisions with He but later attributed to reactions with H2.

Tanner and co-workers have described the design and use of a quadrupole reaction cell, as was used in this work. The most effective reaction gas and pressure will depend on the chemistry of the element of interest and the spectral overlap ion. We will show some examples in this chapter. Signals due to spectral overlap ions can be reduced by over eight orders of magnitude in some cases. However, it is important to prevent the formation of product ions that will produce new spectral overlaps. We will discuss how the bandpass of a quadrupole reaction cell can be used to prevent the production of new spectral overlap ions by making one of the reactants unstable in the cell.

2 RESULTS AND DISCUSSION

2.1 High Efficiency Sample Introduction

If a pneumatic, concentric nebulizer is used in a conventional spray chamber, analyte transport efficiency will naturally decrease from better than 50% at a sample uptake rate of 20 :L/min to less than 2% at a sample uptake rate of 1 mL/min. As the sample uptake rate is increased the droplet number density in the spray chamber increases, droplet-droplet collisions become more likely and drop-drop coagulation leads to loss of small droplets into large ones. The large droplets have a low probably of survival through the spray chamber and into the ICP.

If droplet evaporation can be made rapid enough in the spray chamber, desolvation of the droplets will be virtually complete before droplets impact the walls of the spray chamber. The nebulizer gas jet entrains a huge amount of gas into it. Therefore, if heated gas can be fed into the nebulizer gas and aerosol jet, it can be efficiently mixed with the aerosol to promote rapid evaporation of the droplets in the spray chamber. This is the basis of the spray chamber design described by Debrah and Legere, and shown in Figure 1. A Meinhard HEN nebulizer was used with a nebulizer gas flow rate of approximately 0.3 L/min. A make-up gas (approximately 0.6 L/min) was added to the spray chamber and heated as it flowed along the spray chamber wall, which was heated to a temperature of approximately 180E C. Previously, we showed that the ICP-MS signal increases linearly and proportionally as the sample uptake rate is increased from 25 to 250 :L/min, using this system. This will produce an improvement in ICP-MS sensitivity by a factor of 5 to 10 compared to a conventional concentric nebulizer/spray chamber.

In order to obtain virtually 100% analyte transport rate with sample uptake rates greater than about 50 :L/min, desolvation of the sample aerosol and removal of the solvent vapor produced is essential 5. Solvent vapor loading has severe effects on the plasma (even extinguishing it) at water vapor loads of 50 to 60 mg/min.

If desolvation and solvent vapor removal are efficient enough, sample uptake rates up to 0.8 mL/min can be used without distinguishing the plasma (Figure 2). Sensitivities greater than 1.6 x 109 cps/ppm can be obtained for U+ at a sample uptake rate of 0.6 mL/min. With a conventional nebulizer/spray chamber this instrument typically has a sensitivity of approximately 0.04 x 109 cps/ppm. Therefore, a factor of 40 increase in The Rh+ sensitivity was 0.6 x 109 cps/ppm at a sample uptake rate of 0.6 mL/min, compared to a typical sensitivity of 0.03 x 109 cps/ppm for a conventional nebulizer/spray chamber with an uptake rate of 1 mL/min, a factor of 20 increase in sensitivity can be obtained using the High Efficiency Sample Introduction System (HESIS).

In order to operate at high sample uptake rates, the temperature of the Ar in the spray chamber must be sufficiently high to efficiently desolvate the aerosol before it hits the wall. Furthermore, the temperature in the membrane dryer must be kept above the dew point so that recondensation does not occur. High sweep gas flow rates must be used to remove the solvent vapor.

There is some nonlinearity in the sensitivity as a function of sample uptake rate. The sampling depth and total center gas flow rate (nebulizer + make up) was fixed as the sample uptake rate was varied. The amount of solvent vapor entering the ICP likely changed. The optimum sampling depth or total center gas flow rate may vary as a function of sample uptake rate, causing the nonlinear response. This requires further investigation.

2.2 Desolvation and Chemical Matrix Effects in the ICP

Desolvation of aerosol droplets, vaporization of desolvated particles, atomization and ionization in the ICP as well as ion transport from the ICP to the MS detector can be studied with previously unattainable clarity by introducing the sample as isolated, monodisperse droplets. The monodisperse dried microparticulate injector (MDMI) device provides such a sample introduction system.

When the sample contains a high concentration of dissolved salts, several of the processes that occur in the plasma are affected. Atoms and ions appear earlier (closer to the load coil) when the sample contains a high concentration of dissolved solids. Lazar and Farnsworth proposed that for two identically sized aerosol droplets, the atoms and ions appear earlier because less solvent must be evaporated from the droplet to obtain a desolvated particle. Laser induced fluorescence and emission profiles also shift closer to the load coil when a conventional nebulizer/spray chamber sample introduction system is used.

The square of the aerosol droplet diameter should decrease linearly with time as droplets desolvate in the plasma. The size of the desolvated particle that results after the solvent completely evaporates from the droplet in the plasma can be predicted from the density of the original solid sample, assuming a spherical particle.

If the model of Lazar and Farnsworth is correct, the shift to earlier appearance times should therefore be linearly dependent on the difference in the square of estimated desolvated particle diameters with an without the addition of matrix. Figure 3 shows the earlier appearance of Sr+ emission as the concentrations of YCl3 added to the sample is increased from 0 to 3 mM.

If the proposed model is correct, the shift in initial appearance of analyte signals should depend only on the estimated desolvated particle diameter, which is a function of the concentration and density of the matrix compound. Figure 4 shows the shift in appearance times for a variety of elements (Se, Sr, Pb) with various concentrations of several different solids (including SeO2, YCl3, PbCl2,) dissolved in the sample. The good linear correlation is entirely consistent with the model of Farnsworth et al. Furthermore, a desolvation rate can be estimated from the slope of the line fit to the data. The desolvation rate calculated from the slope is similar to the rate measured by Kinzer and Olesik using resonance Mie scattering from desolvating aerosol droplets.