ICP-MS Measurement

This section will focus upon considerations in developing an ICP-MS measurement procedure for those using a spectrometer employing a quadrupole mass filter. These instruments are relatively easy to operate, have good stability, and are the most common instruments used by trace elemental analytical laboratories. They also have a resolution of something less than 1 amu (atomic mass unit). These instruments are typically referred to as low-resolution instruments.

Some other types of spectrometers will not be addressed in this guide. These include: (a) spectrometers using magnetic fields to disperse the ion beams, which can operate up to a resolving power of 1 part in 10,000; (b) time-of-flight spectrometers; and (c) spectrometers utilizing ion-trap principles. For more detail on the quadrupole, as well as the other types of ICP-MS spectrometers, I encourage you to refer to the following references:

The single quadrupole remains the workhorse of the trace elemental laboratory, with two changes to how it is configured. Collision and reaction cells are now standard equipment rather than an option, and tandem instruments, in which a first quadrupole filters the beam before the cell, are increasingly common where the harder interferences have to be resolved. Everything below applies to those instruments unchanged. A cell removes an interference, it does not remove the need to understand where the interference came from.

Plasma Source Mass Spectrometry - Developments and Applications; Holland, G., Tanner, S. D., Eds.; The Royal Society of Chemistry: Cambridge, U.K., 1997.

Inductively Coupled Plasma Mass Spectrometry; Montaser, A., Ed.; Wiley-VCH: New York, 1998.

Taylor, H. E. Inductively Coupled Plasma Mass-Spectrometry, Practices and Techniques; Academic Press: New York, 2001.

Resolution is a property that should be understood. The peaks are considered to be resolved if the magnitude of the valley between two adjacent peaks is less than 10% of the mean of the magnitude of the peaks:

Most commercial quadrupole mass spectrometers are capable of 0.8 amu mass resolution (at 10% of the valley definition and having equal adjacent peak intensities).

Having adjacent peaks at the same intensity is not a realistic or typical situation. Therefore, the ability to measure an m/z peak at a low concentration adjacent to a high-concentration m/z peak is a situation that must be considered. This very important consideration is referenced as Abundance Sensitivity. The concern is that tailing from the larger peak into the smaller peak will occur, giving false high results for the smaller peak. I personally experience this problem on a daily basis while attempting to measure ppb to ppt levels of impurities in our products that are typically diluted to 100-200 µg/g for ICP-MS trace impurity analysis.

For the quadrupole mass filter, the abundance sensitivities for adjacent peaks on the low and high mass are not equal. This is because the peaks are asymmetric and tend to tail more on the low-mass side. If we have a high-concentration element with a peak intensity of I_h at mass M, and want to measure a low-concentration element with a peak intensity of I_L-1 on the low (M-1) mass side or I_L+1 on the high (M+1) mass side, the ratios I_L-1/I_h and I_L+1/I_h are referred to as the low-mass abundance sensitivity and high-mass abundance sensitivity, respectively. For a peak width of 0.8 amu, expected values of 1 × 10⁻⁵ and 1 × 10⁻⁶ for the low-mass and high-mass abundance sensitivities are not unreasonable. This means the concentrations will be measured to be ~1 ng/g on the low-mass side and ~0.1 ng/g on the high-mass side, adjacent to a mass with a concentration of 100 µg/g. If you are working at dilution factors of 10 to 400, this translates into significant errors. In these situations, I prefer to use axial view ICP-OES for the low- or high-mass elements.

• Identify the risk before the run. A semi-quantitative scan shows where the major elements sit, and any analyte one mass unit away from a major element should be treated as suspect until proven otherwise.
• Prefer the high-mass side. Because peaks tail toward low mass, an analyte at M+1 is roughly an order of magnitude safer than one at M−1.
• Dilute the major element rather than the analyte where the method allows it, since abundance sensitivity is a ratio.

Detection limits and what they describe

A measurement procedure is only defensible if the laboratory can separate instrument performance from method performance, and the three limits below do exactly that.

• The gap between the IDL and the MDL is a direct measure of what the sample preparation is costing. A large gap points at the digestion, the dilution or the blank rather than at the instrument.

• Every dilution multiplies the reporting limit. A detection limit of 0.1 ug/L in the measured solution becomes 40 ug/L in the original sample at a dilution factor of 400, so where dilution is being used to control dissolved solids that trade against sensitivity should be made explicitly.

• ICP-MS is an ion counting technique, so precision is limited by the number of ions counted. Increasing dwell time on the low level analytes and reducing it on the majors is the most effective way to improve precision without extending the run.

This next section presents the most common interferences encountered in ICP-MS using the quadrupole mass filter (hereafter referred to as low-resolution ICP-MS). You can also find the major interferences for the popular isotopes in our interactive periodic table.

Isobaric Interferences

Isobaric interference is a result of equal-mass isotopes of different elements present in the sample solution. Low-resolution instruments cannot distinguish between the isotopes. There are many examples in the intermediate mass regions where the second and third row transitions and the rare earths appear. Fortunately, there are no elemental singly charged isotopes that overlap with monoisotopic elements (⁹Be, ²³Na, ²⁷Al, ⁴⁵Sc, ⁵⁵Mn, ⁷⁵As, ⁸⁹Y, ¹⁰³Rh, ¹²⁷I, ¹³³Cs, ¹⁴¹Pr, ¹⁵⁹Tb, ¹⁶⁵Ho, ¹⁶⁹Tm, ¹⁹⁷Au, and ²³²Th). However, for the monoisotopic elements, be aware of the other interferences to be discussed later. For elements having more than one isotope, the quickest fix may be to use another isotope of that element ("may" is used because other interferences could be encountered). If the interference is from an isotope with roughly the same or lower peak intensity, it is possible to perform a correction by measuring the intensity of another isotope of the interfering element and subtracting the appropriate correction factor from the intensity of the interfered isotope. If you are working with an unknown sample composition, a semi-quantitative analysis is suggested with low-resolution instruments, using a quick scan of the sample and the rather sophisticated semi-quantitative programs available on current instrumentation.

Polyatomic (Molecular) Interferences

Molecular interferences are due to the recombination of sample and matrix ions with Ar and other matrix components, such as O, N, H, C, Cl, S, and F. The light elements (Li, Be, B) are not affected due to their small masses.

Starting with ³⁹K, this type of interference becomes a significant issue. For example, ³⁹K is interfered with by ³⁸ArH and ²³Na¹⁶O. Some polyatomic interferences can be avoided by eliminating certain matrix elements, such as the classic ⁴⁰Ar³⁵Cl interference upon the monoisotopic element ⁷⁵As, where the use of HCl in the sample preparation is to be avoided. The isotopes ⁵⁶Fe, ³⁹K, and ⁴⁴Ca or ⁴⁰Ca are all interfered with by combinations of the Ar, O, and N isotopes.

As we go to the heavier elements, the major polyatomic interferences come from isotopes that are 16 atomic mass units lower than the analyte isotope, through molecular oxide (MO) interference. The lanthanide element isotopes are especially prone to molecular oxide formation.

The use of cool plasma techniques, reaction/collision cells, desolvation, and chromatographic separations — to name a few approaches — has resulted in reduction and, in some cases, complete elimination of many polyatomic interferences. The severity of the MO interference can be reduced through reduction of the sample argon gas flow rate. Mass corrections may be an option in cases where the use of an alternate isotope is not an option. Polyatomic interferences are particularly troublesome in the determination of first-row periodic table elements (K through Se), due to the vast number of combinations of Ar with matrix components.

Doubly Charged Ion Interferences

Doubly charged ion interference is due to doubly charged element isotopes with twice the mass of the analyte isotope. For example, interference from ²⁰⁶Pb²⁺ (m/e = 103) upon ¹⁰³Rh is likely at high Pb concentration levels. Reduction in the sample Ar will minimize this interference. Fortunately, this type of interference is not as prominent in Ar plasmas, but care should be exercised in matrices containing high levels of mid- to heavy-mass element isotopes. The alkaline and rare earth elements form doubly charged ions to an extent that is greater, relative to the other elements.

A cell between the ion optics and the mass filter offers three distinct tools, and choosing between them is a question of how specific the problem is.

• Helium collision mode with kinetic energy discrimination. The polyatomic ion has the larger cross section, so it undergoes more collisions, loses more energy, and is rejected at an energy barrier. It is non-selective, which is its strength: one gas handles most argon- and chloride-based interferences across the mass range, at some cost in sensitivity.
• Reaction mode on mass. A reactive gas removes the interfering ion while the analyte passes through unchanged. More efficient where it applies, but the gas has to suit the specific interference.
• Reaction mode with mass shift. The analyte itself is reacted, usually with oxygen, and measured at a product mass where the interference does not exist. This is the most powerful option and the main reason tandem instruments are chosen.

• Calibrate in the cell mode in which the samples are measured. A cell mode changes sensitivity, background and interference profile, so a calibration acquired in one mode does not transfer to another.

• A correction equation subtracts an estimate, not a measurement, and it carries the noise of the correcting mass into the result. Verify every active correction against an interference check solution containing the interferent without the analyte, and against the same solution spiked with analyte.

In addition to the matrix effects encountered for ICP-OES that are discussed in Section 15, ICP-MS suffers from space charge effects and salt buildup on the orifice of the interface sampler cone.

Space Charge Effects

These effects are thought to occur at the MS interface (the region between the skimmer tip and ion optics) and in the ion optics region. The net result is a suppression of the signal in high concentrations of a matrix element. The kinetic energy of the ion element matrix affects the degree of suppression, with larger masses (higher kinetic energy) causing more depression than lower masses. Due to differences between instruments in interface and ion optic designs, it is difficult to predict the conditions under which the effect is minimal. In my case, working at matrix element concentrations of 100-200 µg/g will cause only a slight reduction in the signal. Under "cool plasma" conditions, I have found this suppression effect to be more pronounced. The approach taken in our laboratory is to attempt to keep the matrix element concentration at or below the 100 µg/g concentration level.

Salt Buildup

The buildup of salts/oxides in samples containing high levels of matrix elements (sea water is a common example) will result in partial or total clogging of the sampler cone. Techniques used to help control this effect include dilution to below 0.1% total solids, flow injection analysis, or ion exchange removal of the matrix component(s).

Dissolved Solids, Dilution and Drift Control

• The 0.1% total dissolved solids figure above is the practical ceiling for extended unattended operation. Short runs tolerate more, but cone deposition begins immediately, and the drift it produces is progressive rather than random.
• Aerosol dilution, in which argon is added downstream of the spray chamber, reduces the solids reaching the plasma without raising the reporting limits by the same factor as a liquid dilution, and produces a drier plasma that suppresses oxide formation.
• The internal standard response is the most useful diagnostic in the run, reporting transport efficiency, plasma condition, cone condition, and space charge suppression in a single number. Regulatory methods commonly require it to fall within 60 to 125% of the calibration blank response, and many laboratories apply tighter internal limits.
• Read the pattern, not only the pass or fail. A gradual decline across the run points to cone deposition, a sudden drop for one sample points to that matrix, and a drop affecting the high-mass internal standard more than the low-mass one points to space charge suppression. Where an internal standard fails, dilute and reanalyze rather than accepting a heavily corrected result.
• Carbon in solution enhances the signal for elements with high ionization potentials, notably arsenic and selenium, by tens of percent. The effect can be used deliberately, but a variable carbon content across the run cannot be tolerated.

An element that is slow to reach a steady signal is usually also slow to leave the system, because the same surface interaction is responsible for both. Mercury, boron, iodine, gold, and the platinum group metals cause the most persistent memory in routine work.

• Rinse chemistry matters more than rinse duration. Extending a dilute nitric acid rinse will not remove an element that is retained through complexation or adsorption. Mercury requires a stabilizer such as gold, iodine is rinsed in dilute alkali, and the platinum group metals need hydrochloric acid in the rinse at a concentration comparable to the sample matrix.
• Alternate an aggressive rinse with a matrix-matched blank rather than relying on a single rinse solution. The aggressive rinse strips the surface, and the matrix-matched blank confirms that baseline has genuinely been reached.
• Shorten the sample path and sequence the run in the laboratory's favor. Every unnecessary fitting and dead volume adds surface area, and analyzing low-level samples before high-level ones removes most carryover without any change to the method.

What follows are some of the more popular techniques used for quantitative analysis using ICP-MS.

External Calibration

This is the calibration technique that is most popular. Many analysts use this approach for matrices that are known and can be matched. The use of internal standards is helpful in accounting for drift. The choice of the internal standard/isotope mass combination is reasonably well understood. Finally, the use of spike recoveries on a split portion of the sample allows the analyst to determine if space charge effects are significant. The following are some brief notes on this topic:

• Know or learn about the sample composition. A semi-quantitative analysis using a scanning approach for the entire mass range allows the analyst to predict interferences and select internal standards and analyte isotopic masses.
• Perform interference check analysis. Prepare for the variations in the matrix and analyte composition, and determine if corrections that have been built into the procedure are capable of providing the required accuracy.
• Use internal standards to help correct for drift. Follow these basic guidelines for internal standard selection:
• Avoid M²⁺ interferences.
• Avoid MO and other molecular interferences.
• Any naturally occurring internal standard element in your sample must be insignificant in comparison to the amount added.
• Use internal standard elements as close as possible to the masses of the analyte elements.
• Make sure the matrix does not react with the internal standard to alter (lower) its concentration.
• Common internal standard elements are ⁶Li, Be, Sc, Ga, Ge, Y, Rh, In, Cs, Pr, Tb, Ho, Re, Bi, and Th. Note that many are monoisotopic.
• Use peak hopping rather than scanning for the final analysis. Peak hopping will save time, and this capability is one of the major advantages of low-resolution systems.
• At the beginning of the analytical day, optimize the instrument using "optimization" standards. I prefer to use a 10 ppb combination of Mg, U, Ce, and Rh. In addition, I like to optimize the instrument to obtain ¹⁴⁰CeO/¹⁴⁰Ce and ¹⁴⁰Ce²⁺/¹⁴⁰Ce currents of less than 0.5% relative. I routinely obtain a "time scan" of ²⁴Mg, ³⁶Ar, ⁷⁰Ce²⁺, ¹⁰³Rh, ¹⁴⁰Ce, ¹⁵⁶CeO, ²³⁰BKG, and ²³⁸U at the beginning of each analytical day. These scans are saved and accompany the following analytical data. Torch alignment, sample argon (nebulizer) flow, and ion optics settings are the parameters I most often change (in the order listed) in the optimization process.
• Make sure the introduction system is clean. I prefer using glass concentric nebulizers and cyclonic spray chambers. I use dilute nitric acid for cleaning. It is often advantageous to change the entire introduction system, sipper to torch, each analytical day. In addition, the sample interface cones need to be rotated each analytical day with cleaned cones. Cleaning the cones in a 1% solution of nitric acid using an ultrasonic bath for 1-2 minutes is typically all that is required. Carefully dry the cones in a drying oven before reusing.
• Periodically check the performance of the ICP-MS during the analytical "run." I prefer to split the sample and spike half of the sample with a known low ppb addition of an assortment of analytes ranging from Mg to U. After confirming the calibration by analyzing the standards, I like to use an analysis sequence of blank, sample, and sample + spike. The spike recovery allows me to determine if space charge effects from the matrix element(s) have significantly lowered the analyte signal.
• Know the stability of your standard. For guidance, consult our Part-Per-Billion Stability Study.

• Add the internal standard online through a mixing tee where possible, and use more than one across a wide analyte range, since space charge suppression is mass dependent.

Standard Additions

This approach is common with ICP-OES, but it may give the analyst a false sense of security when using ICP-MS. It is a concern that this technique has earned such a "good reputation," in view of the fact that it does not guarantee anything except a perfect matrix match. ICP-MS has many more potentially serious problems than matrix matching. The same interference issues discussed above must come into consideration if you choose to use standard additions. For example, if you have a molecular MO interference before the addition and do not use an alternate mass or perform a correction, you will still obtain a false high result. Spend the time to learn about the matrix and identify potential interference issues. After you reach a high level of confidence in the identification and correction for and/or elimination of interferences, then the standard additions approach is a convenient way to "match" a complicated matrix.

Isotope Dilution

The technique of isotope dilution ICP mass spectrometry (ID-ICP-MS) provides the analyst with the possibility of using a primary (definitive) analytical method for the determination of trace metals in a variety of sample types. Examples of primary analytical methods are isotope dilution mass spectrometry (IDMS), ID-ICP-MS, gravimetry, titrimetry, coulometry, differential scanning calorimetry, and nuclear magnetic resonance spectroscopy. ID-ICP-MS is of particular interest to the reference material producer of materials for trace metals content. Unfortunately, there are several elements that are monoisotopic (⁹Be, ²³Na, ²⁷Al, ⁴⁵Sc, ⁵⁵Mn, ⁷⁵As, ⁸⁹Y, ¹⁰³Rh, ¹²⁷I, ¹³³Cs, ¹⁴¹Pr, ¹⁵⁹Tb, ¹⁶⁵Ho, ¹⁶⁹Tm, ¹⁹⁷Au, and ²³²Th), making ID-ICP-MS useless for these elements. Our laboratory has been studying ID-ICP-MS along with the execution of accurate isotopic abundance ratio measurements (another possible primary method), and will publish these studies in the months to come.

The sequence in which solutions are presented to the instrument does as much for data quality as the tuning does.

• Tune and performance check, recorded and retained with the data, followed by the calibration blank and standards acquired in the cell mode that will be used for the samples.
• An initial calibration verification from an independent second source, followed by a calibration blank. A second source that agrees confirms the calibration standard itself, which a repeat of the same standard cannot do.
• Interference check solutions, one containing the interferents alone and one containing the interferents with the analytes, to confirm the corrections and the cell mode.
• Samples in batches of ten, with a continuing calibration verification and blank after each batch and at the end of the run, and a fortified blank, matrix spike, and duplicate in each preparation batch.
• A serial dilution test on a representative sample. Where the concentration is well above the reporting limit, a five-fold dilution should agree with the undiluted result within about 10%, and disagreement is strong evidence of a matrix effect. Where recovery is in question, a post-digestion spike outside roughly 75 to 125% indicates suppression or enhancement.