Optimization of an Analytical Method for the Determination of Trace Metals in Urine by ICP-MS

Urinary metal concentrations may be effective indicators of particulate metal fume exposure and provide a better understanding of specific exposure-response relationships if an analytical method can be developed that is sufficiently sensitive to measure levels at or near background concentrations. Analyzing urine samples by inductively coupled plasma–mass spectrometry (ICP-MS) poses several distinct challenges as the matrix contains components likely to cause false-positives and reduced analytical sensitivity. This study will demonstrate a method that mitigates these problems and produces quantifiable results even at low parts-per-trillion concentrations.

When metals are rapidly liquefied by electric arc forced-gas techniques, such as industrial electric arc induction furnaces and gas metal arc welding, they frequently volatilize into an aerosol of metal particles that can be <1 µm in diameter; such aerosols are sometimes referred to as particulate metal fumes. Previous investigations established that the formation rate and the precise chemical composition of particulate metal fumes are highly dependent on the parameters of specific processes (for example, temperature, gas flow rate, and raw materials) and can contain various metals, including aluminum, cadmium, chromium, copper, iron, lead, magnesium, manganese, molybdenum, nickel, titanium, vanadium, and zinc (1).

For workers that are frequently exposed to the complex metal oxides found in particulate metal fumes (for example, welders, boilermakers, and steel mill workers), the occupational respiratory health hazards and potentially long-term adverse health effects of exposure are of considerable concern. The United States Occupational Safety and Health Administration (Washington, D.C.) recognizes that symptoms such as eye, nose, and throat irritation, fever, chills, headache, nausea, shortness of breath, and muscle pain can result from even brief acute exposure to particulate metal fumes (an illness commonly referred to as metal fume fever). It is suspected that chronic occupational exposure is related to several cardiopulmonary diseases and cancer (2). Although the characteristics of particulate metal fumes and the symptoms of exposure have already been investigated, few studies have identified precise exposure-response relationships between specific metals and adverse health effects. An understanding of these relationships is essential for risk assessment and the development of effective exposure prevention strategies.

Depending on the circumstances, various measures are employed to minimize occupational exposure to particulate metal fumes, such as ventilation and personal respirators, but the effectiveness of these protective measures is typically determined by testing the resulting air for concentrations of metals. However, information regarding actual biological exposure to particulate metal fumes is required to understand exposure–response relationships. Biomarkers are often used in epidemiological and toxicological studies to establish biological exposure to various substances in the environment. In contrast to air measurements, information regarding the concentrations of soluble metals in the urine of workers exposed to particulate metal fumes should provide a better understanding of actual biological exposure, exposure–response relationships, and the effectiveness of protective measures.

As part of a study funded by the National Institute of Environmental Health Sciences (Research Triangle Park, North Carolina) and conducted by the Department of Environmental Health at the Harvard School of Public Health (Boston, Massachusetts), urine samples were collected from nearly 350 individuals known to be occupationally exposed to particulate metal fumes (see Acknowledgments). One phase of this study included analyses of the samples for urinary metal concentrations to investigate the relationship between particulate metal fume exposure and urinary metal concentrations; validate the use of this biomarker as an indicator of exposure; and to contribute further to the investigation of specific exposure-response relationships. Brooks Rand Labs (Seattle, Washington) was contracted to perform analyses of the collected urine samples for the determination of Al, Cd, Cr, Cu, Fe, Mn, Ni, Pb, V, and Zn. For the results of these analyses to have adequate significance to this study, an analytical method was required that would be sufficiently sensitive to achieve quantifiable results even at low parts-per-trillion concentrations with a high degree of accuracy and precision.

Analysis by inductively coupled plasma–mass spectrometry (ICP-MS) with a quadrupole MS system is widely recognized as one of the most advanced techniques currently available for the determination of trace element concentrations. However, analysis by ICP-MS also has traditionally been challenging in two aspects: the potential formation of polyatomic spectral interferences and an intolerance for elevated concentrations of dissolved solids.

Polyatomic Spectral Interferences

The accurate determination of a small number of elements using ICP-MS can be severely compromised because of spectral interferences. These interferences are caused by various elements found in the plasma gas, solvents, or sample matrix forming polyatomic ions that have the same atomic masses as the isotopes of the analytes of interest. The elements As, Cr, Fe, Ni, Se, and V are all known to be frequently affected by these spectral interferences, resulting in false positives and elevated detection limits (3). Although some techniques to minimize these interferences have been developed, including sample dilution, correction equations, matrix separation, and cold plasma technology, these solutions can in turn lead to an unacceptable increase in detection limits and a greater degree of uncertainty in the data. In some instances, less abundant isotopes that may not be affected as strongly by these interferences can be targeted for detection, but not without the potential for a significant loss in analytical sensitivity.

Because this study required quantifiable results for several of the metals known to be affected, an investigation into the potential formation of polyatomic spectral interferences was undertaken. The primary sources of potential interferences when analyzing urine samples occurs when components in the urine matrix (that is, dissolved salts or urea) combine in the argon plasma to form polyatomic ions with the same atomic masses as the major isotopes of Cr, Fe, Ni, and V (4). Examples of the effect of polyatomic spectral interferences on the most abundant isotopes of each element are listed in Table I.

Table I: Example interferences for urine analyses by ICP-MS

Relatively recent advances have resulted in several applications with the potential for overcoming the challenges associated with these interferences. One approach in particular, using a collision– reaction cell, provides conventional detection by ICP-MS and can be configured using a combination of focusing and filtering mechanisms to eliminate spectral interferences. We explored the optimal configuration of the collision–reaction cell to resolve many of the polyatomic spectral interferences that were anticipated to contribute to elevated method detection limits for Cr, Fe, Ni, and V in samples that contain large amounts of calcium, carbon, and chlorine.

Located between the ion optics and the instrument's mass analyzer, the dynamic collision–reaction cell is an enclosed and positively pressurized chamber with an rf/dc quadrupole independent of the mass analyzer quadrupole. After exiting the plasma, the ion beam enters the interface in the normal manner — passing through the sampler or skimmer cones and the ion lens before entering the collision–reaction cell chamber (see Figure 1). A highly reactive gas is injected into the chamber, resulting in gas-phase chemical reactions that breakdown polyatomic ions into neutral elemental species and the rf/dc quadrupole ejects potential sources of new interferences, focusing the analyte ions through to the mass analyzer quadrupole for detection.

Figure 1: Location of the dynamic reaction cell.

In a process described as chemical resolution, reactive gasses are introduced into the chamber as the ion beam passes through and can result in numerous ion-molecule reactions. Under the correct conditions, ions entering the collision–reaction cell will interact with the polar molecules of the reaction gas and effectively transfer their charge. According to the specifics of thermodynamic reaction kinetics, this charge transfer reaction depends on its exothermicity (5). The difference in the ionization potential (eV) of the ions and the polar reaction gas molecules determines whether the reaction will be exothermic and proceed to occur. Table II lists the ionization potentials for some of the elements and reactive gases that are relevant to this investigation.

Table II: Ionization potentials

It can be predicted that charge transfer reactions between ions and polar molecules will happen exothermically if the ionization potential of the reaction gas molecule is significantly less than the analyte. This process is equally applicable to argide interferences and polyatomic ions; that is to say, when they have a greater ionization potential than the reaction gas molecules they will undergo charge transfer reactions and breakdown into neutral elemental atoms. Conversely, if the ionization potential of the ion is less than the reaction gas molecules, an endothermic reaction will generally not occur. Examples of this process also can be expressed as

Ar⁺ + NH₃→ Ar + NH₃⁺

ArO⁺ + NH₃→ Ar + O + NH₃⁺

V⁺ + NH₃→ no charge transfer

The reaction gas that is ultimately selected is based on its predictability to undergo a controlled and specific gas-phase chemical reaction with the polyatomic ions responsible for spectral interferences at a rate considerably faster than it will react with the analyte ion. Ammonia is commonly used as a reaction gas because it has an intermediate ionization potential that is between those of many analyte elements and most polyatomic ions, making it highly effective at chemical resolution.

However, the resulting products of this chemical resolution process can maintain the potential to undergo secondary reactions and form further polyatomic interferences if they are not effectively removed from the chamber. In addition to focusing the analyte ions through the collision–reaction cell, the electrical fields of the rf/dc quadrupole can be optimized to behave as a selective bandpass mass filter. In a process called dynamic bandpass tuning, the amplitude and frequency of the quadrupole are adjusted to establish precisely defined stability regions that act as a low-mass or high-mass cut off. Ions that do not fall within this window will be ejected from the chamber, preventing the secondary reactions that can lead to new interferences (6).

By configuring the collision–reaction cell with the appropriate reaction gas for chemical resolution and sweeping the bandpass rf/dc quadrupole of the collision–reaction cell in tandem with the mass analyzer quadrupole, polyatomic spectral interferences are largely eliminated and the optimal transmission of the analyte ions into the mass analyzer quadrupole for conventional mass separation and detection can occur.

Dissolved Solids

Traditionally, analysis by ICP-MS also is prone to additional limitations because of the instrument's intolerance for dissolved solids. If samples with elevated concentrations of total dissolved solids (TDS) are introduced to the system, material can begin to deposit on the tip of the sample injector, the small orifices of the sampler and skimmer cones, the ion optics, and so forth. These deposits can quickly build up, leading to blockages that cause a degradation of signal stability and a loss of analytical sensitivity. Therefore, it is generally recommended that samples contain no more than 0.2% TDS for optimal instrument performance and signal stability (7).

Samples containing concentrations of TDS that exceed the recommended level can be diluted with a suitable diluent, solvent, or aerosol gas to reduce the concentrations to a level that the instrument can tolerate better. As might be expected, urine samples can contain more than 5% suspended and dissolved solids, including elevated concentrations of dissolved calcium, sodium, chlorine, proteins, and other organic matter. Therefore, these samples typically require significant dilution before the concentrations of TDS fall to recommended levels. These dilutions can lead to elevated method detection limits that may exceed the concentrations of the analytes.

Because this study required quantifiable results even at background levels, we investigated a means to prevent the elevated TDS concentrations in urine from leading to significantly elevated method detection limits. Recent innovations offer solutions to address this issue, primarily by reducing the length of time that the instrument is exposed to the samples and reducing the volume of sample that is introduced to the system, thereby minimizing the amount of solids that can accumulate and affect instrument performance.

The precise configuration of the sample introduction system that delivers samples to the ICP-MS for analyses can vary widely. Conventional designs are typically configured to deliver a relatively constant flow of sample to the instrument until the required measurements have been confirmed and then perform a thorough rinse cycle to prevent any contamination of the subsequent sample. This process can take as long as 10 min depending on the application, during which the instrument may be exposed to the sample matrix for a significant amount of time.

Other innovations in sample introduction configurations, in conjunction with appropriate nebulizers, now allow for a dramatic reduction in the sample volume that must be injected for accurate measurements and the time that each analysis requires. High-throughput sample injection systems allow for precisely determined amounts of sample, depending on the required analyses, to be rapidly pumped into a short sample loop located in close proximity to the nebulizer. The sample is then injected into the nebulizer along with a constant flow of carrier solution, ensuring that no air is introduced into the sample line. This process minimizes the amount of time required for signal stabilization to occur and the volume of sample that must be introduced.

For efficient vaporization and ionization, liquid samples are typically injected into the argon plasma of an ICP-MS instrument as fine aerosols using a concentric nebulizer that shears the liquid stream into small droplets with argon gas and a spray chamber that rejects all but the smallest droplets. The length of time it takes for the instrument's signal to stabilize after sample injection begins, and therefore the amount of sample that the instrument is effectively exposed to, depends in part on how efficiently this aerosol generation process occurs.

Conventional concentric nebulizers are typically constructed of quartz and aspirate samples at flow rates of approximately 1 mL/min. As little as 2% of the aspirated sample may actually achieve the required droplet size and successfully pass through the spray chamber. This relatively low sample transport efficiency requires that the instrument be continuously exposed to the sample until signal stability is eventually achieved, increasing the potential for dissolved solids to deposit in amounts sufficient to cause lowered instrument sensitivity. This issue raises even greater concerns when the analyses of large quantities of samples are required, because it can necessitate increased instrument maintenance and very often reanalysis of the samples.

More recently, specialized high-efficiency nebulizers that use higher gas pressures have been developed that can aspirate samples at flow rates as low as 20 µL/min. With nearly all of the aspirated sample successfully passing through the spray chamber, these microflow nebulizers provide a significantly higher sample transport efficiency than conventional concentric nebulizers. Additionally, microflow nebulizers are typically constructed from fluoropolymers that are extremely resistant to corrosion, which reduces the potential for contamination and makes them highly appropriate for applications that require extremely low detection limits in complex sample matrices.

By configuring a sample introduction system that incorporates a high throughput sample injection device and a microflow nebulizer, the issues associated with elevated concentrations of TDS in sample matrices should be eliminated without compromising the goal of low detection limits.

Experiments

As a major program of the Center for Disease Control and Prevention's (CDC, Atlanta, Georgia) National Center for Health Statistics, the National Health and Nutrition Examination Survey (NHANES) is a large-scale study designed to assess the health of people throughout the United States. As part of the survey, urine samples were collected from participants to measure exposure to environmentally toxic substances, including trace metals. The laboratory method employed by the National Center for Environmental Health's Division of Laboratory Sciences in support of NHANES (CDC Method No: ITU001) was developed to achieve rapid and accurate quantification of As, Ba, Be, Cd, Cs, Co, Mo, Pb, Pt, Sb, Tl, W, and U by ICP-MS in more than 5000 urine samples annually (8).

With one exception, this method was not developed to address any of the specific trace metals analyses requested of Brooks Rand Labs (Al, Cd, Cr, Cu, Fe, Mn, Ni, Pb, V, and Zn), but because of the broad acceptance of the method an effort was made to replicate it to the extent achievable. However, because of the presence of polyatomic spectral interferences and elevated concentrations of dissolved solids, strict adherence to the method led to elevated method detection limits. We attempted to resolve those issues through the optimization of the collision–reaction cell parameters and implementation of various sample introduction configurations.

The CDC method describes the use of an ELAN DRC II MS system (Perkin-Elmer LAS, Shelton, Connecticut), a Cetac ASX 500 series autosampler (Cetac Technologies, Omaha, Nebraska), and a Meinhard Type A quartz concentric nebulizer TQ-30-A3 (Meinhard Glass Products, Golden, Colorado). Additionally, our experiments included an ESI SC-4 autosampler with an ESI SC-FAST valve, an ESI PFA-ST MicroFlow nebulizer, and an ESI PC3 Peltier-cooled spray chamber (Elemental Scientific Inc., Omaha, Nebraska).

The urine samples for this study had been stored frozen (-80 °C) in 15-mL polypropylene centrifuge tubes for more than three years before submission to the laboratory. The sample tubes were placed in a Class 100 laminar flow clean hood and allowed to reach room temperature. An aliquot of each sample was added to an autosampler tube and diluted 1:9 with 1% HNO₃; matrix spikes were added at the same time as the diluent. To dissolve any suspended material, the aliquots were capped and heated to 90 °C for 30 min or until particulates were no longer visible. The vials were allowed to cool to room temperature and then loaded directly into the autosampler for analysis.

In an attempt to overcome the recognized potential for polyatomic spectral interferences, ammonia was selected as the appropriate reaction gas for the collision–reaction cell based on the reasons outlined earlier and a compromise flow rate of 0.8 mL/min was chosen in the interest of throughput efficiency. Optimal RPq values for the electrical fields of the rf/dc quadrupole were selected on an element-specific basis (see Table III). The magnitude of interferences (both polyatomic and isobaric) for the most abundant isotope of Ni were so significant that superior analytical sensitivity was achieved by monitoring the next most abundant isotope, ⁶⁰ Ni, in which far less interference was observed.

Table III: Analytical instrument parameters (PerkinElmer Elan II DRC)

Three sample introduction configurations were implemented to measure the effectiveness of several variables to minimize the amount of time the instrument was exposed to the sample matrix and mitigate the issues associated with dissolved solids. The first configuration closely followed the specifications outlined in the CDC analytical method, using a Cetac ASX 520HS autosampler and a Meinhard Type A quartz nebulizer with a sample flow rate of 1.2 mL/min, and included element-specific collision–reaction cell reaction gas flows. The second configuration replaced the specified nebulizer with a microflow model, the ESI PFA-ST MicroFlow nebulizer, and utilized the previously mentioned common collision–reaction cell reaction gas flow, also with a sample flow rate of 1.2 mL/min. The third configuration maintained the microflow nebulizer and replaced the specified autosampler with one that included a high-throughput sample injection system, the ESI SC-4 FAST system, with a sample flow rate of 0.3 mL/min.

In Figure 2, the total analysis time per sample has been segmented into consecutive stages, including integration, sample flush, read delay, collision–reaction cell gas flow stabilization, and wash times. The first configuration required more than 150 s per analysis and exposed the instrument to the sample matrix for more than 120 s. The second configuration eliminated the necessity for collision–reaction cell gas flow stabilization time but still required more than 120 s per analysis and exposed the instrument to the sample matrix for more than 90 s. The third configuration dramatically reduced the sample flush time, required less than 50 s per analysis, and exposed the instrument to the sample matrix for less than 40 s.

Figure 2: Total analysis times for the different sample introduction configurations.

As suggested in the CDC method, the low-level and high-level versions of standard reference material NIST SRM 2670a (toxic elements in urine) were analyzed in conjunction with the samples to establish method accuracy. However, the certificate of analysis for this standard reference material does not provide any values for Fe and only "information values" for Cr, Ni, and V. The results of the analyses for each of the sample introduction configurations appear in Table IV. Configurations including the microflow nebulizer appear to have slightly increased the recoveries of Cr and Ni, particularly for the low-level SRM. The recoveries of V appear to be unchanged regardless of the configuration or concentration.

Table IV: NIST 2670a toxic elements in urine SRM results (µg/L)

Method precision was evaluated by analyzing 15 pairs of urine samples spiked at levels 2–5 times the method detection limits (refer to Table VI) for each sample introduction configuration. The combined mean recoveries of these analyses for the first and second sample introduction configurations (both used the Cetac ASX 520HS autosampler) and separately for the third sample introduction configuration (ESI SC-4 FAST autosampler) are shown in Table V. The mean matrix spike recoveries appear to have significantly improved for Cr, Fe, and Ni with the use of the third configuration. The mean relative percent differences (RPD) between results for matrix spike pairs also appears to have significantly improved for Cr, Fe, and V with the use of the third configuration.

Table V: Matrix spike results for concentrations 2–5×the MDL (%)

The recoveries for specific pairs of matrix spike samples analyzed for Cr are displayed in Figure 3. Again, the results of the analyses from the first and second sample configurations have been averaged. The mean recovery of the matrix spikes for the first and second configurations is only 84.7%, with a mean RPD between results for matrix spike pairs of 9.7%. Moreover, the analyses were halted after the eighth pair because of poor internal standard recoveries (see more below). The mean recovery of the matrix spikes for the third configuration was significantly improved at 98.5%, with a mean RPD between results for matrix spike pairs of 2.4%.

Figure 3: Chromium recoveries for specific pairs of spiked samples.

Although adjustments in the sample introduction system did not appear to significantly affect the recoveries of the standard reference materials, the recoveries of the matrix spikes did, in some cases, improve substantially. This inconsistency is explained primarily because of the order in which these samples were analyzed. The standard reference materials were analyzed immediately subsequent to the method blanks and before the instrument had been exposed to any of the urine samples; whereas the pairs of matrix spikes were analyzed after approximately every set of 10 urine samples. The degradation in analytical sensitivity caused by the exposure of the instrument to the greater levels of dissolved solids using the first and second sample introduction configurations is apparent when the recoveries of the internal standard (10 µg/L Ga) are observed in Figure 4. However, the recoveries for the third sample introduction configuration remain highly consistent and only begin to degrade as the number of samples approaches 200.

Figure 4: Comparison of recoveries for internal standards (10 µg/L Ga).

The method detection limits (MDL) for each of the four analytes were determined through the repeated analyses of a composite urine sample made up of samples collected from multiple laboratory volunteers. Four replicates of the composite urine sample were analyzed for each element to establish the average base urinary metal concentrations. Eight replicates of the composite urine sample were then spiked at 2–15 times the average base urinary metal concentrations. The MDLs were determined by multiplying the student-t value for n-1 (2.998) by the standard deviation of the spiked sample results.

As shown in Table VI, the MDLs for Cr and V are in low parts-per-trillion concentrations. The MDL for Ni is an order of magnitude greater because of the monitoring of the less abundant isotope. The MDL for Fe is in the low parts-per-billion range because of the relatively higher background concentrations. These ultralow method detection limits allowed for the accurate quantification of the target analytes in 84–100% of the submitted urine samples for the study.

Table VI: Achieved method detection limits

Conclusion

Though the presence of polyatomic spectral interferences and elevated concentrations of dissolved solids can be the cause of significant issues when analyzing urine samples for metals concentrations by ICP-MS, an optimized configuration of a collision–reaction cell in conjunction with the appropriate sample introduction system can largely mitigate these problems, resulting in improved precision, better spike recoveries, and ultralow detection limits.

Acknowledgments

Urine samples for method development were collected from steelworkers in conjunction with research performed by Drs. David Christiani and Jennifer Cavallari at the Harvard School of Public Health. Funding for the collection and analysis of study urine samples was provided by ES009860 and ES00002 and research staff was supported by T32 ES 07069, T42 OH008416, and the Brooks Rand Labs Internal Research Fund. The Institutional Review Board at the Harvard School of Public Health approved the study protocol, and informed written consent was obtained from each adult prior to participation.

Michela Powell-Hernandez and Andrew Maizel are with Brooks Rand Labs in Seattle, Washington.

Robert Thomas is the principal consultant for Scientific Writing Solutions in Gaithersburg, Maryland. Please direct correspondence to: robert.james.thomas@verizon.net.

References

(1) I. Pires, L. Quintino, R.M. Miranda, and J.F.P. Gomes, Toxicol. Environ. Chem.88(3), 358–394 (2006).

(2) Occupational Safety and Health Guideline for Welding Fumes, United States Department of Labor, Occupational Safety and Health Administration, April 2010. http://www.osha.gov/SLTC/healthguidelines/weldingfumes/recognition.html.

(3) R. Thomas, Practical Guide to ICP-MS: A Tutorial for Beginners (CRC Press/Taylor and Francis, 2008) p. 81–82.

(4) T.W. May and R.H. Wiedmeyer, At. Spectrosc. 19(5), 150–155 (1998).

(5) S.D. Tanner and V.I. Baranov, At. Spectrosc. 20(2), 45–52 (1999).

(6) S.D. Tanner, V.I. Baranov, J. Am. Soc. Mass Spectrom. 10(11), 1083–1094 (1999).

(7) R. Thomas, Practical Guide to ICP-MS: A Tutorial for Beginners(CRC Press/Taylor and Francis, 2008) p. 17.

(8) Urine Multi-Element ICP-DRC-MS; Method No: ITU001B. US Department of Health and Human

Services, Centers for Disease Control and Prevention, National Center for Environmental Health, Division of Laboratory Sciences, Inorganic Toxicology and Nutrition Branch. 2006.