Interlaboratory evaluation of a standardized inductively coupled plasma-mass spectrometry method for the determination of trace elements in air filter samples: preliminary results

Abstract

An interlaboratory evaluation of a standardized inductively coupled plasma-mass spectrometry (ICP-MS) method for determining trace elements in workplace air samples was carried out, toward fulfillment of method validation requirements for international voluntary consensus standard test methods. The interlaboratory study was performed in accordance with an applicable ASTM International standard practice, ASTM E691, which describes statistical procedures for investigating interlaboratory precision. Performance evaluation materials, prepared by a contract laboratory, consisted of mixed-cellulose ester filters that were spiked with 21 elements of interest at levels of 0.50 or 5.0 micrograms (µg) per filter. Triplicates of each spiked filter, plus media blanks spiked with blank reagent, were conveyed to each volunteer laboratory; spiking levels were unknown to the participants. The laboratories were requested to prepare the filter samples by one of the three sample preparation procedures (hotplate or microwave digestion or hotblock extraction) that are described in the standard test method, ASTM D7439. Participants were then asked to analyze aliquots of the prepared samples by ICP-MS using ASTM D7439, and to report their data in units of µg per filter sample. Preliminary interlaboratory precision and recovery estimates from 20 volunteer laboratories are reported.

---

Introduction

Preventing or reducing occupational exposures to airborne metals and metalloids by inhalation often requires the ability to measure target elements at very low levels in workplace atmospheric samples.^1,2 In occupational hygiene laboratories, atomic spectrometric techniques, especially flame and graphite furnace atomic absorption spectrometry (FAAS and GFAAS, respectively) and inductively coupled plasma-atomic emission spectrometry (ICP-AES),^3,4 have traditionally been used to determine metals and metalloids in occupational air samples. In efforts to obtain even lower detection limits for toxic elements in workplace atmospheres than those offered by these and other analytical methods, inductively coupled plasma-mass spectrometry (ICP-MS) is becoming more widely used in the occupational health field.⁵

To address the need for a standardized ICP-MS method for use by occupational hygiene laboratories, an ASTM International voluntary consensus standard ICP-MS test method was recently developed.⁶ Compared to AAS and ICP-AES techniques, ICP-MS offers improved detection limits for most elements,⁷ thereby enabling ultra-trace analysis capabilities that may be required for short-term measurements and/or reduced occupational exposure limits.

The aim of this work was to conduct an interlaboratory study of the ASTM International ICP-MS consensus standard test method, ASTM D7439,⁶ with a goal of obtaining estimates of method performance for elemental analysis based on a collaborative trial. The interlaboratory study was carried out in consideration of an applicable ASTM International standard practice, ASTM E691,⁸ which describes statistical procedures for investigating interlaboratory precision of ASTM standard test methods. To date there is a paucity of validation data for ICP-MS analysis of occupational air samples, and it was our desire to endeavor to fill this data void. Preliminary interlaboratory precision estimates from performance evaluation samples (containing certified levels of 21 elements of interest in occupational health) from 20 volunteer participating laboratories are reported herein.

Experimental

The performance evaluation materials (PEMs) used for the interlaboratory study consisted of 37 mm diameter mixed-cellulose ester (MCE) filters that were spiked with 21 elements of interest (Al, Ag, As, Ba, Be, Cd, Co, Cr, Cu, Fe, Mg, Mn, Mo, Ni, Pb, Sb, Se, Sn, U, V, Zn) at levels of 0.50 or 5.0 µg per filter and were certified by the source laboratory (High Purity Standards, Charleston, SC, USA). Filter spikes were used in lieu of particulate materials in order to obtain performance data for a maximum number of elements of occupational hygiene concern. The spike values were chosen in consideration of applicable occupational exposure limit values⁹ and expected loading levels for full-shift workplace air sampling at typical flow rates. Media blanks consisted of MCE filters that were spiked with the solution reagent containing no added metals and metalloids. The as-received spiked filters were mailed in sets of five within sealed plastic containers and were isolated from one another by non-absorbent separator pads. It is noted that spiked filters of this kind are not representative of real workplace air filter samples containing particulate materials. However, filters spiked with soluble metal compounds are often used in laboratory proficiency testing programs, and performance data from such samples should offer a baseline measure of optimal interlaboratory variability.

Volunteer laboratories possessing ICP-MS equipment and having experience in environmental elemental analysis were solicited to participate in the interlaboratory study. PEM samples (spiked filters as well as media blanks fortified with solution reagent) were mailed in triplicate to each volunteer laboratory; spiking levels were unknown to the participants. The participating laboratories were requested to carry out sample preparation (hotplate or microwave acid digestion or hot-block acid extraction) and ICP-MS analysis in accordance with the procedures described in ASTM D7439.⁶ For sample dissolution, laboratories were able to choose between different candidate acids (e.g., nitric, hydrochloric, perchloric and hydrofluoric acids) and mixtures thereof recommended in this ASTM standard. Participants were requested to report their results to the coordinator of the study in units of micrograms of each element per filter sample. A list of the twenty laboratories that participated in the interlaboratory investigation and returned results is presented in Table 1. For purposes of data presentation, laboratories were identified by an identifier code to ensure confidentiality.

Table 1 List of volunteer laboratories that participated in the ICP-MS interlaboratory study

Name of laboratory	Location
a Formerly US Army Center for Health Promotion and Preventive Medicine (USACHPPM). b Formerly DataChem Laboratories.
Savannah River Site Analytical Laboratory (SRS)	Aiken, SC, USA
Health and Safety Laboratory (HSL)	Buxton, UK
RTI International	Research Triangle Park, NC, USA
US Army Public Health Command^a	Aberdeen Proving Ground, MD, USA
Institut National de Recherche et de Sécurité (INRS)	Vandœuvre-les-Nancy, France
Environmental Resource Associates (ERA)	Arvada, CO, USA
BWXT Y-12 National Security Laboratory	Oak Ridge, TN, USA
Hungarian Institute of Occupational Health (HIOH)	Budapest, Hungary
US Geological Survey (USGS)	Denver, CO, USA
Institut Technique des Gaz de l'Air (ITGA)	Saint-Etienne, France
Institute of Naval Medicine, Occupational and Environmental Safety Laboratory (OESL)	Gosport, UK
Occupational Safety and Health Administration (OSHA), Salt Lake Technical Center	Sandy, UT, USA
Institut de Recherche Robert Sauvé et en Sécurité du Travail (IRSST)	Montréal, Canada
Laboratoire Central de la Préfecture de Police (LCPP)	Paris, France
Navy Central Industrial Hygiene Laboratory (CIHL)	San Diego, CA, USA
ALS Laboratory Group^b	Salt Lake City, UT, USA
Japan National Institute of Occupational Safety and Health (JNIOSH)	Kawasaki, Japan
Bureau Veritas North America (BVNA)	Novi, MI, USA
Eurofins Environnement	Saverne, France
University of Cincinnati, Department of Chemistry	Cincinnati, OH, USA

---

Interlaboratory precision of the results reported by the participating laboratories was investigated using the statistical analysis procedures described in ASTM E691.⁸ In accordance with this consensus standard practice, repeatability and reproducibility of the results reported were calculated for each element. Repeatability (r) is an estimate of within-laboratory variability, which was computed by averaging the squares of the standard deviations of within-laboratory results for each sample, and taking the square root of this average. Thus, the average within-laboratory standard deviation for each reported result is expressed by the repeatability standard deviation, s_r. Reproducibility (R) is an estimate of the variability of both within-laboratory and between-laboratory results. The reproducibility standard deviation s_R = {(s_x)² + [(s_r)²(n − 1)n⁻¹]}^1/2, where s_x is the standard deviation of the mean value as estimated by the average of all interlaboratory results for a given PEM and n is the number of test results at a particular spiking level.

For each element, an estimate of analytical bias was calculated by dividing the difference between the mean of the laboratory-reported triplicate results and the reference value by the reference value.¹⁰ That is, bias B_i = (µ_i − R_i)R_i⁻¹, where B_i, µ_i and R_i are the bias, mean measured value and reference value, respectively, for the i^th laboratory-reported value.

Results and discussion

Comprehensive ICP-MS interlaboratory precision statistics, based on results reported by the 20 participating laboratories and treated statistically in accordance with ASTM E691,⁸ are summarized in Table 2. Overall mean values were calculated based on the pooled means for the average of the three results reported by each laboratory for each element; however, it is noted that some laboratories did not report results for all elements. Some laboratories stated that they did not possess standard solutions for certain elements (e.g., U and V), while other laboratories did not report results for certain elements (e.g., Ag and Sn) owing to high dilution of extracted sample solutions prior to ICP-MS analysis.

Table 2 Interlaboratory precision statistics for ICP-MS elemental measurements from spiked filter samples

Element	Spike level/µg per filter	x̄ /µg per filter	s x	s r	s R	RSD^e	Percent recovery
a Overall mean (for n reporting laboratories). b Overall standard deviation. c Repeatability standard deviation. d Reproducibility standard deviation. e Relative standard deviation (sx × x̄^−1). f Significant blank levels reported. g Statistical outlier excluded (Dixon's Q-test, P = 0.05).
Aluminium^f	5.0	5.57 (n = 18)	1.07	0.755	1.60	0.193	111
Antimony^f	0.50	0.504 (n = 18)	0.136	0.073	0.262	0.271	99.6
Arsenic^f	5.0	4.98 (n = 20)	0.412	0.238	0.473	0.083	99.2
Barium	5.0	4.87 (n = 19)	0.364	0.129	0.385	0.075	96.8
Beryllium	0.50	0.509 (n = 19)	0.055	0.019	0.061	0.108	102
Cadmium	0.50	0.505 (n = 19)	0.037	0.015	0.040	0.073	101
Chromium^f	5.0	5.07 (n = 20)	0.653	0.213	0.685	0.129	101
Cobalt	0.50	0.500 (n = 19)	0.051	0.019	0.054	0.102	100
Copper	5.0	5.17 (n = 20)	0.614	0.492	0.779	0.119	103
Iron^f	5.0	6.01 (n = 16)	1.30	0.877	1.55	0.216	120
Lead	0.50	0.500 (n = 18)	0.059	0.078	0.096	0.118	100
Magnesium^f	5.0	5.59 (n = 17)	0.861	0.429	0.956	0.154	112
Manganese^f	0.50	0.507^g (n = 18)	0.060	0.024	0.064	0.118	101
Molybdenum	0.50	0.511 (n = 17)	0.107	0.024	0.110	0.209	102
Nickel^f	0.50	0.516 (n = 18)	0.073	0.057	0.092	0.141	103
Selenium^f	0.50	0.476 (n = 19)	0.067	0.145	0.156	0.141	95.2
Silver^f	0.50	0.515^g (n = 17)	0.112	0.100	0.148	0.217	103
Tin	0.50	0.562 (n = 17)	0.137	0.042	0.143	0.244	112
Uranium	0.50	0.515 (n = 15)	0.063	0.040	0.074	0.122	103
Vanadium	0.50	0.440 (n = 14)	0.120	0.044	0.127	0.273	88.0
Zinc^f	5.0	5.59 (n = 20)	0.575	0.361	0.674	0.103	112

---

Computed interlaboratory relative standard deviations (RSDs) ranged from 0.073 to 0.273 for the spiked filter samples (Table 2). Most of these precision estimates are acceptable in view of an overall goal of obtaining interlaboratory RSDs of less than 0.20.^10–12 With exceptions for a few elements yielding RSD > 0.20, interlaboratory RSDs obtained are commensurate with those reported for metals analysis in relevant interlaboratory analytical proficiency testing programs.^13,14 Two elements (Sb and V) yielded RSDs in excess of 0.25. Recoveries ranged from 88% to 120% (Table 2), with only one result (i.e., for Fe) beyond 100% ± 15%. It is noted that applicable proficiency testing programs typically investigate only a few metals (such as As, Cd, Cr, Cu and Pb),^13,14 thus sample preparation and analysis procedures in many laboratories will normally have been optimized for particular target elements.

Significant background media blank levels were reported for about half of the elements tested (Table 2), but overall blank data were not quantifiable owing to excessively high interlaboratory variability for media blanks. With the exception of iron, elements having significant background blank levels did not generally result in a high positive bias (as estimated by percent recovery). Polyatomic interferences that could have given rise to the observed positive bias for iron, if no interference correction was applied, are ⁴⁰Ar¹⁶O⁺ and ⁴⁰Ar¹⁶OH⁺ for ⁵⁶Fe⁺ and ⁵⁷Fe⁺ isotopes, respectively.^6,15 The high interlaboratory RSD observed for vanadium (with monitoring of the ⁵¹V⁺ isotope) could be ascribed to variability in correction interference from ³⁵Cl¹⁶O^+7,15 for those laboratories using aqua regia (or reverse aqua regia) sample dissolution methods. Noticeably the vanadium recovery was low (<90%, Table 2), possibly due to analyte loss during sample preparation; the high interlaboratory RSD observed for this element could be related. We do not have a satisfactory explanation for the high interlaboratory RSD that was observed for antimony.

Since the volunteer laboratories were able to choose between various sample preparation procedures described in ASTM D7439⁶ (i.e., hot plate or microwave digestion or hot block extraction), it is probable that the greatest contribution to overall measurement uncertainty in this interlaboratory trial was the sample preparation procedure.¹⁶ As mentioned earlier, a number of laboratories were unable to report results for certain elements due to overly diluted extract solutions, which led to higher than expected overall method detection limits. Besides the potential polyatomic interferences discussed above for a few elements, the effect of acid mixtures employed for sample preparation could have had some influence on recoveries and/or precision. However, these contributions should be minimal in most instances since the filters used for performance evaluation were spiked with solutions containing target elements; hence the compounds of target elements, as spiked on filters, would have been soluble. A detailed statistical analysis of individual contributions of sample preparation steps (e.g., heating technique employed, acid mixture used and dilution factors) to interlaboratory variability is currently underway and will be reported in the future. Analytical performance from the filter PEMs investigated here represents a best-case scenario; interlaboratory variability from particulate materials can be significantly higher.¹⁷ Appreciable contribution to interlaboratory uncertainty (as measured by RSDs) could be due to several of the participants having limited experience in sample preparation and ICP-MS analysis of filter sampling media. Proficiency testing schemes typically show improved interlaboratory uncertainty with succeeding rounds as laboratories obtain experience with new analytical techniques and sample matrices.¹³

Conclusion

Overall, preliminary results of the interlaboratory study suggest that the standardized ICP-MS method is generally effective for the determination of elements in air filter samples that are used to estimate occupational exposures to metals and metalloids. These results provide data in support of the ASTM D7439 consensus standard. It is anticipated that interlaboratory performance will improve as more experience is gained by laboratories performing ICP-MS analysis of occupational hygiene samples. Additional work is currently underway that will fulfill requirements for performance data of fully validated international standard ICP-MS procedures for workplace air measurements.

Disclaimer

Mention of company names or products does not constitute endorsement by the Centers for Disease Control and Prevention, the US Department of Energy, the UK Heath and Safety Executive, or their contractors. The findings and conclusions in this article are those of the authors and do not necessarily represent the views of the National Institute for Occupational Safety and Health or the UK Health and Safety Laboratory.

Acknowledgements

We are extremely indebted to the volunteer laboratories for their willingness to participate in this study. Leroy Dobson, Martha Waters and Kenn White kindly reviewed the draft manuscript. Funding support for the interlaboratory study was provided in part by ASTM International. This work was carried out under the auspices of a NIOSH National Occupational Research Agenda (NORA) global collaborative project.

References

1. R. P. Beliles, in Patty's Industrial Hygiene and Toxicology, Part C, ed. G. D. Clayton and F. E. Clayton, Wiley, New York, 1994, vol. 2, ch. 27.
2. Chemical Hazards of the Workplace, ed. G. J. Hathaway, N. H. Proctor, J. P. Hughes and M. J. Fischman, Van Nostrand Reinhold, New York, 3rd edn, 1991.
3. National Institute for Occupational Safety and Health, NIOSH Manual of Analytical Methods, http://www.cdc.gov/niosh/nmam, accessed 22 February, 2010.
4. Occupational Safety and Health Administration, OSHA Sampling and Analytical Methods, http://www.osha.gov/dts/sltc/methods, accessed 1 February, 2010.
5. O. T. Butler, W. R. L. Cairns, J. M. Cook and C. M. Davidson, J. Anal. At. Spectrom., 2010, 25, 103.
6. ASTMD7439, Standard Test Method for the Determination of Elements in Airborne Particulate Matter by Inductively Coupled Plasma Mass Spectrometry, ASTM International, West Conshohocken, PA, 2008.
7. R. Thomas, Practical Guide to ICP-MS, Marcel Dekker, New York, 2004.
8. ASTME691, Standard Practice for Conducting an Interlaboratory Study to Determine the Precision of a Test Method, ASTM International, West Conshohocken, PA, 2005.
9. American Conference of Governmental Industrial Hygienists, 2010 Threshold Limit Values for Chemical Substances and Physical Agents, ACGIH, Cincinnati, OH, 2010 (updated annually).
10. P. L. Drake, A. D. Marcy and K. Ashley, J. Environ. Monit., 2006, 8, 134.
11. O. T. Butler and A. M. Howe, J. Environ. Monit., 1999, 1, 23.
12. K. Ashley, M. J. Brisson, A. M. Howe and D. L. Bartley, J. Occup. Environ. Hyg., 2009, 6, 745.
13. F. I. Grunder, Am. Ind. Hyg. Assoc. J., 2001, 62, 736.
14. P. R. Stacey, Ann. Occup. Hyg., 2006, 50, 417.
15. M. A. Vaughn and G. Horlick, Appl. Spectrosc., 1987, 41, 523.
16. E. Hund, D. Luc Massart and J. Smeyers-Verbecke, Anal. Chim. Acta, 2000, 423, 145.
17. M. Gerbolis and G. Busica, Intercomparison Exercise for Heavy Metals in PM10 (EUR 23219 EN), European Commission, Luxembourg, 2008.

---

Footnotes

† Electronic supplementary information (ESI) available: Laboratory-reported results. See DOI: 10.1039/c0ay00377h

‡ This article was prepared by US and UK government employees and contractors as part of their official duties and legally may not be copyrighted in the USA or the UK.

---