Development and characterization of reference materials for trace element analysis of keratinized matrices

Abstract

Biomonitoring for human exposure to lead, arsenic, mercury, and other toxic metal(loid)s often relies on analyzing traditional biospecimens such as blood and urine. While biomonitoring based on blood and urine is well-established, non-traditional biospecimens such as hair and nails can offer the potential to explore past exposures as well as the advantages of non-invasive collection and ease of storage. The present study describes the production of four reference materials (NYS RMs 18–01 through 18–04) based on caprine horn, a keratinized tissue similar to human hair and nails, intended to serve as a resource for calibration, quality control, and method validation purposes. The elemental content and homogeneity of these candidate reference materials were characterized for 17 elements using inductively coupled plasma mass spectrometry (ICP-MS). Commutability between two or more of the NYS caprine horn RMs and human nails was established for 8 elements (Ba, Ca, Cr, Cu, Mn, Pb, Sr, and Zn) based on analysis by ICP-MS/MS and ICP-optical emission spectrometry. The development and optimization of an ICP-MS/MS instrumental method for the determination of 17 elements in keratinized tissues is described. The method was validated against three certified reference materials based on human hair showing good accuracy and method repeatability better than 25% for all analytes. This study also describes sample preparation issues and addresses common challenges including surface contamination, microwave digestion, matrix effects, and spectral interferences in inorganic mass spectrometry.

Graphical Abstract

New York State Department of Health Keratin Matrix Reference Materials

1. Introduction

Biomonitoring provides information about absorbed dose from all exposure routes and is critical in assessing exposure to toxic agents. For lead (Pb), a ubiquitous environmental contaminant, exposure levels in the United States and elsewhere have declined due to regulations in recent decades. However, the element’s low-level toxicity, especially for young children, has necessitated increasingly sensitive biomonitoring techniques [1]. For assessing human Pb exposure, the most widely accepted biomarkers of recent exposure are Pb in blood (as well as serum and plasma) and urine [2]. While collecting a urine specimen is non-invasive, the practicality of using “spot” urine is complicated by the effects of dilution in the bladder, and corrections based on creatinine excretion are not always appropriate [3]. Because of the long residence time of Pb in the body, Pb exposure represented by bone, hair, nails and other non-traditional biomarkers, reflecting a range of exposure windows and time periods, may provide complementary information to blood and urine data and be valuable for many applications [4].

Hair and nails belong to a family of hard keratinized tissues that includes horns, hooves, claws, and others. Hair and nails are convenient to collect and non-invasive, such that there is the potential for larger sample sizes and higher statistical power compared to invasive matrices such as blood or bone [5]. Keratinized tissues have been shown to accumulate elements such as Hg, As, Se, and Pb [6–11]. Hard keratinized tissues also grow in increments that are metabolically inert once formed, potentially reflecting longer term exposures [2].

Nonetheless, trace element analysis of hair and nails is less analytically rigorous than traditional biomonitoring, and is not as well validated for human exposure assessment [4]. For several toxic elements, analysis of hard keratinized tissues for clinical [12] and/or public health screening [13] purposes has been questioned. High interlaboratory variability has been reported for specific elements [7,12,14–16]. The poor harmonization and reliability of trace element data in keratinized tissues may be due in part to insufficient availability and variety of well characterized keratin-based RMs [17]. As a major component of measurement accuracy and traceability [18], RMs are often used for calibration, quality control, method validation, and to assign values to other materials.

Previous studies have investigated trace element exposures using hair and nails, and numerous reports and review papers on this subject have been published over the years [19–24]. Trace element measurements in other types of hard keratinized tissues in animals, including horns [11,25], claws [26], hooves [27], animal hair [28], and tortoiseshell scutes [29] have also been reported. Several studies have addressed specific analytical challenges in trace element analysis of hair and nails, including sample washing [30–33], spectral interferences in ICP-MS [17], and matrix effects [17,34,35]. Three certified reference materials (CRMs) based on human hair have been produced and were available at the time this study was conducted: GBW 07601 [36], ERM DB001 [37], and NIES No. 13 [38]. Multielement content and homogeneity in all three CRMs have been characterized. However, the suitability of these hair CRMs for use in analyses of nails or other keratinized tissues, i.e., commutability, has not been established. To our knowledge, no other RMs are commercially available based on human nails or any keratinized material aside from hair. In addition, the range of Pb values certified in currently available hair RMs (2.14 to 5.7 μg/g) is relatively narrow compared to actual human samples. Published values for Pb vary widely from 0.04 to 240 μg/g in nails, and 0.004 to 95 μg/g in hair [39].

The primary aim of this study was to produce and characterize four homogenized keratinized RMs, based on horns collected from Pb-dosed goats and containing physiologically-bound Pb at four different levels. These materials were designed to serve as a new resource in future keratinized tissue analyses. A secondary aim of the study was to investigate sample preparation techniques and methods for analyzing keratinized tissues, paying special attention to common challenges for Pb and other key elements.

Four RMs, designated as New York State (NYS) RMs 18–01 through 18–04, were produced from 13 caprine horns. The horns were derived from a herd of goats that were periodically dosed with Pb to produce Pb-enriched blood reference pools for distribution in the NYS Department of Health (DOH) blood Pb proficiency testing (PT) program. A method was developed for the analysis of keratinized tissues for 17 elements (Ca, V, Cr, Mn, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Ba, Hg, Tl, and Pb) based on acid digestion and determination by tandem inductively coupled plasma mass spectrometry (ICP-MS/MS). The method was validated against existing hair CRMs and used to characterize the trace element content and homogeneity of RMs 18–01 through 18–04, as well as their commutability with human nail samples. Spectral interferences and matrix effects were investigated as part of the ICP-MS/MS method development. The RMs characterized in this study were then distributed to 21 laboratories around the world as part of an interlaboratory study coordinated by the Trace Elements Laboratory at the NYS DOH Wadsworth Center, as described elsewhere [16]. Additional details on the work described here can be found elsewhere [40].

2. Materials and Methods

2.1. Reagents, standards, and certified reference materials

All HNO₃ used in this work was double distilled in-house from reagent grade acid using a DuoPUR Sub-boiling point Distillation System (Milestone Inc., Shelton, CT). Double de-ionized (DDI) (>18M Ωcm) water was produced using a NANOpure DIamond UV/UF water system (Barnstead International, Dubuque, IA). Six intermediate multielement solutions were prepared in Nalgene polypropylene volumetric flasks (Thermo Fisher Scientific, Rochester, NY) from a multielement calibration standard stock solution with various elemental concentrations (SM-108–005, High Purity Standards, North Charleston, SC), supplemented with 1000 μg/mL Ca (High Purity Standards), Rb (GFS Chemicals, Inc., Columbus, OH, USA), and Sr (High Purity Standards) single-element standards. Intermediate multielement solutions contained 1% HNO₃. Six separate intermediate calibration standards stocks for Hg containing 1.5% (v/v) HCl were prepared from a 1000-μg/mL single-element Hg standard solution (High Purity Standards).

Aqueous calibration standards and run samples were prepared for each analysis from a 20-fold dilution of intermediate standard solutions and digests, respectively, using a reagent solution containing 1% (v/v) HNO₃, 1% (v/v) HCl, 1 μg/L Au (Inorganic Ventures, Christiansburg, VA), 0.005% (v/v) Triton X-100 (Baker Analyzed, Phillipsburg, NJ), and 0.5 μg/L Rh and 1 μg/L Ir as internal standards (High Purity Standards). The reagent solution was stored in acid-washed 2-L fluorinated ethylene propylene containers (GFS Chemicals, Inc., Columbus, OH, USA).

Three human hair CRMs were used in the ICP-MS/MS analysis of horn samples for method validation and quality assurance purposes: GBW 07601a (Institute of Geophysical and Geochemical Exploration, Langfang, China), ERM DB001 (Institute for Reference Materials and Measurements, Geel, Belgium), and NIES No. 13 (National Institute for Environmental Studies, Onogawa, Japan).

2.2. Source of caprine horns

Horns were derived post-mortem from 11 Pb-dosed goats that were from a herd maintained by the NYS DOH Wadsworth Center, as described elsewhere [41,42]. Over the last three decades, more than 60 goats have been dosed periodically with Pb to produce Pb-enriched blood reference pools for the NYS DOH’s blood Pb PT program. Animal use protocols were approved by the Wadsworth Center Institutional Animal Care and Use Committee (IACUC) (number 16–096 associated with this program). The Center is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care. The herd is maintained by full time animal care staff and a licensed facility veterinarian oversees animal care. Adult goats were dosed with oral gelatin capsules containing Pb, as Pb acetate, over several multi-day events, before whole blood was collected for PT pools. Necropsies were performed in which the major organs and tissues including the liver, kidneys, heart, lungs, horns, and long bones were harvested for research purposes. Lead-enriched blood pools were previously used to validate analytical blood Pb measurements [42,43]. Previous studies from the Wadsworth Center have reported on the uptake and accumulation of Pb in caprine long bones [44], brain [45], and other soft tissues [46]. Intact horns were collected from approximately 20 animals and archived at −80° C for future research studies. Each goat is identified by year of birth – number, e.g., 2001–5, where the identifier used for each horn is the same as that of the goat from which it was derived.

2.3. Production of four candidate horn RMs and estimation of surface contamination

Horns were removed proximal to the skull using a Stryker autopsy saw (Kalamazoo, MI) during the post-mortem necropsy. In the laboratory, whole horns, ranging in length from 20 to 50 cm, were prepared for elemental analysis by cleaning, sectioning, and finally milling in three steps. Horns were first cleaned using “Ghost Wipe” lead dust wipes (Environmental Express, Charleston, SC) based on an EPA/HUD protocol for assessing Pb dust loading on surfaces in homes (ASTM Method E1728). Wipes were applied to separate, contiguous areas to capture contamination from the entire horn surface. Two wipes each were analyzed for horns from goats 2003–3 (un-dosed) and 1982–8 (22-g lifetime Pb dose), and four for the larger horn from goat 2004–7 (un-dosed). Wipes were acid-digested and analyzed in triplicate by ICP-MS. Elemental concentrations in the acid digests of each wipe (μg/L), corrected for the blank wipe, were multiplied by digest volume (L) to give total μg of elemental content. Estimated endogenous elemental content was determined by ICP-MS for acid-digested powders of the same three horns (2003–3, 1982–8, and 2004–7). To compare estimated endogenous elemental content of the three horns with the corresponding surface contamination found in wipes, the elemental mass fraction (μg/g) in the horn was converted into total elemental content (μg) by multiplying by the original horn mass (g).

Following the initial cleaning using Ghost Wipes, horns were hemisected longitudinally using a diamond disk blade attached to a Dremel tool (Racine, WI) and the bony core was removed intact. Next, each of the two horn sheaths was further separated into rectangular sections of ~50 cm² in area using the diamond disk blade. Subsections were then soaked in 20% (v/v) trace element grade H₂O₂ for around 1 hour to remove exogenous contamination, residual blood, and portions of the epidermis remaining between the bony core and keratinized sheath, as well as to clean areas affected by sawing. The soaking time used was selected based on experiments in which aliquots of the H₂O₂ solution were analyzed at approximately 1-hour intervals over 70 hours to assess leaching of elements of interest (data not shown). After soaking in H₂O₂, horn subsections were rinsed with double-deionized (DDI) water, air-dried in a class 100 clean room, and then milled as described below.

Cleaned subsections of 13 horns (derived from 11 animals) were homogenized in two mills using the brittle fracture technique: subsections were frozen in liquid nitrogen or overnight in a −80°C freezer, then introduced into the Retsch SM 2000 Cutting (knife) Mill (Haan, Germany) for a pre-grind without the use of a grate. The resulting horn fragments were re-frozen, then milled a second time in the knife mill using a 2.0-mm grate to create a coarse-milled material. This coarse material was introduced into the Retsch ZM200 Ultra-Centrifugal Mill (Haan, Germany) for further milling using Ti sieves with a 0.25-mm grate (Retsch, Haan, Germany). The resulting milled horn powder was oven-dried at 55°C overnight.

Three to five aliquots of each of the 13 homogenized horn powders were acid digested, as described below, and the Pb content of each powder was determined by ICP-MS. Based on the Pb data, homogenized horn powders were combined and blended to create four separate pools to minimize within-level variability in Pb content, maximize the range of concentrations, and maintain similar masses between levels to the extent possible. Each of the four pooled powders was sifted through a 180-μm pore plastic sieve to separate larger particles and clumps. Clumps were broken up using an agate mortar and pestle and re-sieved. Powders were thoroughly mixed by manual agitation and sample scooped into 4-mL HDPE vials for storage, with each vial containing ~1.3 g of material on average.

2.4. Collection and preparation of human nails

Fingernails and toenails were collected from anonymous donors at the Wadsworth Center. The NYS DOH Institutional Review Board (IRB) determined that anonymous nail specimen collection for this study did not meet the Federal definition of Human Subjects Research and thus did not require IRB oversight. A total of 17 specimen bags containing between ~70 mg and ~2 g of nail clippings was collected. Nail specimens from each donor were analyzed independently.

While there is no universally accepted approach for washing hair and nail samples prior to trace element analysis, the samples in this study were cleaned using a nonpolar solvent, a polar solvent, and a detergent based on published procedures [17,30]. Specimens were first submerged in a 30% (v/v) acetone solution (Mallinckrodt Pharmaceuticals, Dublin, Ireland), and manually agitated for 10 minutes. Following a DDI water rinse, nail specimens were sonicated in 0.5% (w/v) Triton-X solution (Sigma-Aldrich, Darmstadt, Germany) for 1 hour and again rinsed with DDI water. Nail samples were sonicated for an additional hour in DDI water and finally oven-dried at 55°C for 1 hour. Acetone and DDI washings were analyzed for each nail specimen to characterize the elemental surface contamination removed; after washing, additional analyses of washed and unwashed sub-samples of three nail specimens were conducted, to provide a more detailed assessment of the washing procedures used.

2.5. Microwave-assisted digestion

Horn and nail samples were digested using a pressurized, high-temperature microwave method. Approximately 200 mg of keratinized tissue (caprine horn, human nail, hair CRM) sample was weighed out for all digestions, except for several nail samples where less than 200 mg was available. All keratinized tissue samples were pre-digested in 35-mL Pyrex® digestion vessels used with the SP-D (CEM Corporation, Matthews, NC) for 6–12 hours in 4 mL HNO₃ in a class 2 biohood. In the last ~1 hour of pre-digestion, 1 mL of 30% (v/v) H₂O₂ for ultra trace analysis (Sigma-Aldrich, Darmstadt, Germany) was added to promote a more complete digestion [47].

Other workers have used HCl in combination with HNO₃ to achieve more complete digestion of keratinized tissues [19]. Here, use of HCl as a digestion reagent resulted in a precipitate forming in the horn digests. This observation was also reported by a laboratory participating in the interlaboratory study, as described elsewhere [16]. In the study reported here, the precipitate was collected after filtering and found to contain detectable amounts of Ca, Zn, and Pb, likely from insoluble precipitates of Ca, Zn and Pb chlorides. Consequently, HCl was not used in the final digestion method used here.

Samples were digested using the CEM SP-D with a two-stage microwave program. In the first stage, the temperature was ramped to 100°C over 4 – 5 minutes. In the second stage, the temperature was ramped to 200°C over 4 – 5 minutes and held at that temperature for an additional 4 – 6 minutes. Exact ramp and hold times used were sample-dependent. A high maximum temperature (200°C) was used in the SP-D digestion method in order to ensure the complete decomposition of the protein-rich keratinized matrix. Acid digests were diluted with DDI water to 50 mL in 50-mL conical tubes (Sarstedt, Nümbrecht, Germany) and all appeared clear in color. Each digestion batch included 3–7 acid blanks and 1–3 hair CRMs as QC materials. Pyrex digestion vessels were cleaned by microwaving additional acid blanks in between sample batches. Averaged analyte levels in acid blanks were subtracted from those in samples. For all analytes besides Co and Cr, changes in recoveries for hair CRMs due to acid blank subtraction were typically less than the expanded uncertainty (U_CRM) given on the certificate of analysis (COA). Digests of one hair CRM (GBW 07601a) contained a white precipitate after digestion. A previous study also found an undigested residue after the digestion of this material, but reported negligible levels of the elements that are of interest in this study [17].

Digestion of Ghost Wipes that were used to assess elemental contamination of horns was analytically challenging due to the uncontrolled exothermic reaction with concentrated HNO₃. A modified digestion protocol, using dilute HNO₃ at approximately 60% (v/v), was optimized and applied to surface wipes from the three horns that were analyzed.

2.6. Instrumental methods

2.6.1. ICP-MS/MS analysis

Digested samples were analyzed using an Agilent 8800 ICP-MS/MS equipped with an octopole reaction system (ORS), situated between two scanning quadrupole mass analyzers (Agilent Technologies, Santa Clara, CA). A 400-μL min⁻¹ MicroMist nebulizer (Glass Expansion, Pocasset, MA) and a Peltier cooled (2 °C) Scott double-pass spray chamber (Agilent Technologies, Santa Clara, CA) were used. Argon (Airgas USA, LLC, Albany, NY 12205) was used as the plasma, dilution, and carrier gas. Horn and nail samples and hair CRMs were analyzed over six runs by ICP-MS/MS.

2.6.1.1. Matrix effects and selection of operating parameters

Matrix effects in ICP-MS can lead to the suppression or enhancement of an analyte signal in a digested sample relative to that in an aqueous solution [48]. Preliminary studies were carried out to optimize gas flow rates and sampling depth that minimized matrix effects affecting Pb and other elements in spiked solutions containing the digested horn matrix used in this study. Sampling depth refers to the distance between the load coil and the sampling orifice leading to the mass spectrometer [49,50]. These studies indicated that matrix effects were negligible at a range of operating parameters (data not shown). In the absence of appreciable matrix effects, gas flow rates and sampling depth were chosen based on other criteria. Lower gas flow rates of 0.6 L/min for nebulizer gas and 0.35 L/min for the dilution gas kept oxide ratios below 0.5%. Sampling depths below 8 mm, while yielding higher sensitivity in preliminary studies, may increase deposition of solids on the cones, potentially causing signal drift over time. A sampling depth of 8 mm was therefore selected.

Final operating conditions for the ICP-MS/MS instrument and acquisition parameters for the analytes are given in Tables 1 and 2. Data for this multi-element method were acquired using O₂, He, and No Gas modes. All modes utilized MS/MS technology for interference control and quadrupole resolution between adjacent isotopes. Flow rates of the two cell gases used, O₂ and He (Airgas USA, LLC, Albany, NY 12205), were selected based on ramping experiments (Table 1).

Table 1:

Agilent 8800 ICP-MS/MS instrumental parameters.

Parameter	ICP-MS/MS
RF Power	1600 W
Nebulizer gas flow rate	0.60 L/min
Dilution gas flow rate	0.35 L/min
He cell gas flow	3.0 ml/min
O2 cell gas flow	3.5 ml/min
Scan type	MS/MS
Sampling depth	8 mm
Replicates per sample	3

Table 2:

Optimized ICP-MS/MS and ICP-OES acquisition parameters for analytes in keratinized tissues.

	ICP-MS/MS			ICP-OES
	Isotope (m/z)	Cell Gas Mode	Calibration Ranges	Wavelength (nm)	Plasma Viewing Position
Ca	43	No gas	500 to 16000 μg/L	315.887	Radial
V	51	O2 - mass shift	0.4 to 12.8 ng/L	292.402	Axial
Cr	52	O2 - mass shift	0.4 to 12.8 μg/L	267.716	Axial
Mn	55	He	0.4 to 12.8 μg/L	257.610	Axial
Co	59	No gas	0.1 to 3.2 μg/L	228.616	Axial
Cu	65	He	50 to 1600 μg/L	324.752	Axial
Zn	66	He	100 to 3200 μg/L	206.200	Axial
As	75	O2 - mass shift	1 to 32 μg/L	188.979	Axial
Se	80	O2 - mass shift	6 to 192 μg/L	196.026	Axial
Rb	85	He	1 to 32 μg/L	Not measured
Sr	88	He	1 to 32 μg/L	421.552	Radial
Mo	98	No gas	0.4 to 12.8 μg/L	202.031	Axial
Cd	111	O2 - on mass	0.4 to 12.8 μg/L	214.44	Axial
Ba	138	No gas	0.4 to 12.8 μg/L	233.527	Axial
Hg	202	O2 - on mass	1 to 32 μg/L	Not measured
Tl	205	No gas	0.2 to 6.4 μg/L	190.801	Axial
Pb	206+207+208	He	2 to 64 μg/L	220.353	Axial

2.6.1.2. Selection of isotopes and calibration in ICP-MS/MS

Specific isotopes and gas modes were selected to avoid isobaric interferences and minimize polyatomic interferences and are shown in Table 2. For Pb measurement, the signals from three isotopes (²⁰⁶Pb⁺, ²⁰⁷Pb⁺, and ²⁰⁸Pb⁺) were summed to control for geographic variations in isotopic signature among samples, calibrators, and CRMs. Due to a potential overlap with ²⁰⁴Hg⁺, the counts from ²⁰⁴Pb⁺ were not included in the summed signal. Vanadium, Cr, As, and Se were found to be best monitored as oxides, i.e., “mass shifted,” in order to avoid polyatomic interferences. Interferences on ⁵¹V⁺, ⁵²Cr⁺, and ⁷⁵As⁺ signals can be caused by compounds containing C, Cl, and/or S. Compounds containing Ar from the plasma gas can also interfere with the measurement of ⁷⁵As⁺ (⁴⁰Ar³⁵Cl⁺), and ⁸⁰Se⁺ (⁴⁰Ar₂⁺). To address interferences from two compounds that form double oxides, ⁹⁵Mo¹⁶O⁺ and ²⁰²WO⁺ respectively, ¹¹¹Cd⁺ and ²⁰²Hg⁺ are often analyzed in O₂ mode on-mass [51,52]. For ²⁰²Hg⁺ measurements in this study, high W levels in some horn samples, likely due to contamination from milling, caused a severe interference from ²⁰²WO⁺ that was not mitigated using oxygen mode, especially in RM 18–04. Interferences from WO on Hg at more than one m/z were also reported by participants in an interlaboratory study using NYS RMs 18–01 through 18–04 [16]. Elements whose major or sole isotope was measured, or that are enriched in keratinized tissues – Mn, Cu, Zn, Rb and Sr – were analyzed in He mode to artificially reduce high counts that could damage the ICP-MS detector over time. Lead in He mode was found to have superior accuracy and precision to Pb in No Gas mode, possibly due to collisional induced focusing.

Single-element calibration standards for Hg were analyzed before multi-element standards in each run to reduce Hg memory effects on sample measurements later in the run. Fresh Hg intermediate standards were prepared more frequently than multi-element standards due to the poor stability of Hg in solution. Calibration curve slopes were fitted using 1/x weighting. Calibration ranges of working standards are given in Table 2. Analytical performance parameters, including the limit of detection (LOD) for 17 elements, method repeatability (s_r) and relative method repeatability (% s_r) are given in Table 3. Validation data based on analysis of 3 hair CRMs are shown in Table 4.

Table 3:

Validation of detection limits, method repeatability (s_r) and relative method repeatability (% s_r) of ICP-MS/MS method for 17 elements, and ICP-OES method for 9 elements.

	ICP-MS/MS			ICP-OES
	LOD	sr	% sr*	LOD
Ca (μg/g)	57	11	2	24
V (ng/g)	16	7	23	73
Cr (μg/g)	0.37	0.09	12	0.07
Mn (μg/g)	0.09	0.03	5	0.02
Co (ng/g)	27	12	2	90
Cu (μg/g)	0.35	0.07	2	0.5
Zn (μg/g)	2.9	1	1	2.7
As (ng/g)	7	2	--	1600
Se (ng/g)	76	26	8	1100
Rb (ng/g)	26	9	4	Not measured
Sr (ng/g)	60	14	4	27
Mo (ng/g)	9	9	4	330
Cd (ng/g)	4.8	1.3	15	Not detected
Ba (ng/g)	13	8	3	40
Hg (ng/g)	6.5	1.9	--	Not measured
Tl (ng/g)	2.4	2.5	--	Not detected
Pb (μg/g)	0.03	0.01	20	0.8

^*Method repeatability divided by the mean value of NYS horn RM 18-01

Table 4:

Validation of accuracy of ICP-MS/MS method with three hair CRMs for 17 elements. $(U_{c,bias})$ represents the expanded uncertainty of the bias. $U_{CRM}$ is the expanded uncertainty given on the certificate of analysis of each CRM.

	ERM - DB001		GBW 07601a		NIES No. 13
	${\overline{x}}_{\text{CRM}}$ ± UCRM	${\overline{x}}_{\text{val}}$ ± Uc,bias	${\overline{x}}_{\text{CRM}}$ ± UCRM	${\overline{x}}_{\text{val}}$ ± Uc,bias	${\overline{x}}_{\text{CRM}}$ ± UCRM	${\overline{x}}_{\text{val}}$ ± Uc,bias
Ca (μg/g)		990 ± 30	0.145 ± 0.020 %	0.128 ± 0.005 %	820*	758 ± 43
V (μg/g)		0.06 ± 0.01	0.50 ± 0.18	0.35 ± 0.01	0.27*	0.28 ± 0.04
Cr (μg/g)		0.68 ± 0.18	0.41 ± 0.12	0.38 ± 0.19		0.44 ± 0.35
Mn (μg/g)		0.57 ± 0.01	2.0 ± 0.3	2.0 ± 0.1	3.9*	3.84 ± 0.32
Co (μg/g)		0.100 ± 0.004	0.045 ± 0.009	0.037 ± 0.003	0.07*	0.067 ± 0.003
Cu (μg/g)	33 ± 4	32 ± 4	14.3 ± 1.6	12.9 ± 0.4	15.3 ± 1.3	16.8 ± 2.3
Zn (μg/g)	209 ± 12	199 ± 12	137 ± 9	131 ± 3	172 ± 11	161 ± 1
As (μg/g)	0.044 ± 0.006	0.040 ± 0.007	0.28 ± 0.05	0.27 ± 0.01	0.10*	0.09 ± 0.01
Se (μg/g)	3.24 ± 0.24	2.94 ± 0.25	0.58 ± 0.12	0.63 ± 0.02	1.79 ± 0.17	1.62 ± 0.05
Rb (μg/g)		0.17 ± 0.02	0.06*	0.023 ± 0.002		0.14 ± 0.01
Sr (μg/g)		2.23 ± 0.03	7.7 ± 0.4	7.3 ± 0.2		2.81 ± 0.05
Mo (μg/g)		0.06 ± 0.01	0.17 ± 0.03	0.16 ± 0.01		0.05 ± 0.01
Cd (μg/g)	0.125 ± 0.007	0.121 ± 0.010	0.07 ± 0.01	0.078 ± 0.004	0.23 ± 0.03	0.22 ± 0.01
Ba (μg/g)		0.85 ± 0.02	11.4 ± 0.6	11.2 ± 0.2	2.0*	1.54 ± 0.05
Hg (ng/g)	365 ± 28	261 ± 62	670 ± 100	726 ± 28	4420 ± 200	3480 ± 80
Tl (ng/g)		<2.4	7.7 ± 1.1	6.1 ± 0.8		<2.4
Pb (μg/g)	2.14 ± 0.20	2.04 ± 0.20	5.7 ± 0.5	5.7 ± 0.15	4.6 ± 0.4	4.6 ± 0.1

^*Informational value

One of the hair CRMs used in this study, GBW 07601, contained high levels of Ga, a typical internal standard used for lighter elements. Based on internal standard optimization studies, Rh was selected to replace Ga as the internal standard for all elements lighter than Hg. Iridium was used as the internal standard for Hg, Tl, and Pb.

2.6.2. ICP-OES analysis

The NYS horn RMs, human nail samples, and hair CRMs were analyzed by ICP-MS/MS (six runs) and inductively coupled plasma optical emission spectrometry (ICP-OES) (three runs) in order to investigate commutability. Sample analysis by ICP-OES was carried out using a Perkin Elmer Optima 5300DV instrument (Perkin Elmer, Shelton, CT 06484) with a dual view plasma source and using an RF power of 1500 W. The sample introduction system included a built-in peristaltic pump and a removable torch, a concentric nebulizer, and a cyclonic spray chamber. Nebulizer gas flow rate was set at 0.55 L/min, auxiliary gas at 0.2 L/min, and the sample uptake rate 1.8 ml/min. The spectrometer contained an echelle polychromator with a solid state segmented-array, and a charge-coupled-device detector (SCD) for simultaneous detection with background measurement. Three replicates were measured per sample using area processing mode, auto-integration over 5 to 20 s, and 1–2-point background correction. Argon gas was used to purge the optical system during operation and as the plasma gas at a flow rate of 15 L/min. Rinse and read delay times were 55 s. Detection limits for ICP-OES were calculated from a single run in the same manner and using the same digestate as used for ICP-MS/MS detection limit calculations. The View Distance in the method was set at 15.0 mm, which in turn determined the observation heights of radial view. Table 2 shows the wavelengths used and plasma viewing positions, and Table 3 shows the LODs for the elements of interest in this study measured by ICP-OES.

Analytical methods used for ICP-OES were based on US EPA Method 6010C [53]. Briefly, after the instrument was operated in warm-up mode for about 30 min, the torch and optics were optimized for alignment by introducing a 1 mg L⁻¹ Mn solution into the plasma. The calibration standards used in ICP-OES analyses were identical to those used in ICP-MS/MS analysis.

2.7. Assessment of reference material homogeneity and commutability to human nails

From each of the four horn RMs, 12 vials were selected for a between-vial homogeneity study using a stratified random sampling scheme. Two sub-samples were analyzed from each vial in order to assess within-vial homogeneity, for a total of 24 samples (12 vials x 2 sub-samples) analyzed per RM level. Samples were acid-digested and analyzed by ICP-MS/MS for 17 elements following a scheme recommended by the International Organization for Standardization (ISO) which accounts for potential drift during analysis [54]. Homogeneity was assessed using One-Way ANOVA as described by ISO Guide 35 [54].

Commutability between the NYS horn RMs and human nail samples, as well as between the human hair CRMs and nail samples, was investigated by comparing results obtained for the same sample digests using ICP-MS/MS and ICP-OES. Digests of 17 human nail samples, three human hair CRMs, and homogeneity sample digests from each horn RM level were analyzed by each technique.

3. Results and Discussion

3.1. Keratinized reference materials

Losses in powder mass due to sieving using the 180-μm pore plastic sieve were estimated at 11%, 8%, 8%, and 4% for levels 1 through 4 of this powdered material, respectively, reflecting the variation in particle sizes between materials before sieving. Residual moisture content of all four materials was determined in duplicate by oven-drying: one sub-sample from two vials of each level was pre-weighed, dried first at 105°C for 2 hours, weighed, then dried again at 75°C for 15 hours and re-weighed. Changes in mass after the two drying stages differed by <1%. Residual moisture content was calculated as the difference between initial and final dried masses, and expressed as a percentage of the initial mass. Table 5 gives a description of the final four materials. Short-term stability of one of these materials was previously confirmed for the purpose of the interlaboratory study: RM 18–02 was found to be stable over 40 days for 16 key elements at the 99% confidence level based on a non-parametric t-test [16].

Table 5:

Description of the four NYS horn RMs, including the number of horns, approximate mass and number of vials, and residual moisture content in each material.

Reference material	No. animals (horns) used	Approx. total RM mass produced (g)	Number of vials produced	Residual moisture content (%)
NYS 18–01	1 (2)	90	69	7±1
NYS 18–02	4 (4)	700	519	6±1
NYS 18–03	4 (4)	400	308	5±1
NYS 18–04	3 (3)	200	149	5±1

3.2. Horn and nail cleaning studies

Sample preparation procedures, notably sample cleaning but also sample mass and digestion technique, have been found to vary considerably among laboratories and these may adversely affect interlaboratory harmonization [12,24]. The contribution to elemental content in keratinized tissues from exogenous sources is difficult to distinguish from endogenous content, and strong washing procedures may remove some endogenous content. Adding a further complication, optimal washing procedures vary by element [32]. For two types of keratinized samples – caprine horns and human nails –surface contamination and cleaning approaches were assessed in several experimental studies.

Surface wipes were analyzed to investigate multi-elemental contamination of horns 2003–3 (un-dosed), 1982–8 (22-g lifetime dose), and 2004–7 (un-dosed) (see Electronic Supplementary Material (ESM) Table S1). The most notable elemental contaminants of these horns were Pb, Mn, and Ba, possibly reflecting the high crustal abundance of these elements [55], relative to low endogenous content. Widely varying surface levels of Pb, Se, Rb, and several other elements were found among horns, possibly reflecting variability in environmental exposures among the animals. High Pb levels were found on the surface of the horn from goat 2004–7, possibly explaining the higher endogenous content found in this horn compared with that derived from the other un-dosed animal, goat 2003–3 (ESM Table S2).

Surface contamination of human nails, and washing procedures, have been investigated previously [17,31,56]. In this study, in brief, analytes were more efficiently removed from nail samples by DDI than by acetone, and As and Se were only weakly affected by washing (ESM Table S3). These findings are consistent with a previous study in hair [31].

Substantial Mn and Ba content, approximately 20–50% relative to endogenous content, was removed in both human nails and caprine horns by cleaning procedures; this likely reflects the ubiquity of these elements in the environment as well as low endogenous content in both tissue types.

3.3. ICP-MS/MS method validation: accuracy, limits of detection, and repeatability

The ICP-MS/MS method developed for this work was validated in six runs (independent calibrations) over a span of approximately six weeks. Table 4 includes results for method accuracy, evaluated by analyzing two to three digests each of three hair CRMs over the validation runs (six to ten replicates per CRM in total). Gross outliers due to instrumental problems or blunders were excluded from calculations for method accuracy. For n validation measurements for each CRM, the expanded uncertainty of the bias $(U_{c,bias})$ was calculated by propagating the SD of the validation measurements $(S_{val})$ and the $U_{CRM}$ given on the CRM COA, as shown in Equation 1 [56]. A coverage factor (k) of 2 was used to give a confidence interval of approximately 95%. For informational values where no $U_{CRM}$ is given on the COA, $(U_{c,bias})$ is equal to the standard error of the mean of the validation measurements multiplied by the coverage factor.

$$U_{c,bias}=k*\sqrt{\frac{s^{2}val}{n}+{(\frac{U_{CRM}}{2})}^2}$$ Equation 1

For all analytes except for Hg, Tl, and Se, the absolute difference between the mean validation measurements $({\overline{x}}_{\text{val}})$ and certified values $({\overline{x}}_{\text{CRM}})$ given on the COA for each of the three hair CRMs was less than U_CRM, indicating no significant bias from the certified value (Table 4). Low recoveries for Hg in ERM DB001 and NIES No. 13 may have been due to volatilization losses during digestion, or incomplete conversion of MeHg, the predominant Hg species in hair, to free Hg [57]. Low recoveries were found for Se in ERM DB001, but not in the other two materials. The low Tl recovery in GBW is likely explained by Tl content that is close to the LOD.

Method LODs for ICP-MS/MS were determined using a digest from the milled powder of horn 2003–3(B). This horn had been soaked overnight in H₂O₂ to assess leaching of Pb over time, resulting in lower final content of Pb and other elements of interest. The SD of within-run replicates (n=9) based on a single digest was multiplied by 3 to calculate the LOD for all 17 elements as shown in Table 3. Outliers (one for V, three for Mn, and two for Mo) identified by GESD were excluded from calculations. These outliers were potentially due to contamination by milling shavings, or to uncorrected spectral interferences. Repeatability represents the SD of within-run replicates (n=7) based on a single digest of NYS horn RM 18–01 (Tables 3, 6). This parameter is also used in the homogeneity study calculations discussed in section 3.4.

Table 6a:

Trace element content and homogeneity of NYS RM 18–01: means, uncertainties due to inhomogeneity, and ANOVA heterogeneity test results.

NYS RM 18-01
		Uncertainty			ANOVA Test
	sr	$\overline{x}$	sbb	swb	uc,bb	F	P	Fcrit
Ca	11 μg/g	641	36	59	37	0.82	0.63	2.72
V	7 ng/g	31	*2	*6	7	0.83	0.62	2.72
Cr	0.09 μg/g	0.73	0.22	0.26	0.23	1.77	0.17	2.72
Mn	0.03 μg/g	0.62	0.05	0.08	0.05	0.68	0.73	2.85
Co	12 ng/g	749	30	56	33	0.64	0.76	2.72
Cu	0.07 μg/g	4.52	0.18	0.38	0.20	0.52	0.85	2.72
Zn	1 μg/g	95	3	7	3	0.44	0.91	2.72
As	2 ng/g	<7	--	--	--	0.74	0.69	2.72
Se	26 ng/g	317	*20	49	33	0.46	0.90	2.72
Rb	9 ng/g	246	*6	21	10	0.21	0.99	2.72
Sr	14 ng/g	354	15	31	21	0.62	0.78	2.72
Mo	9 ng/g	231	14	27	16	0.60	0.80	2.72
Cd	13 ng/g	8.4	*1.1	5.4	1.7	0.18	1.00	2.72
Ba	8 ng/g	269	29	45	30	0.94	0.54	2.72
Hg	19 ng/g	<6.5	--	--	--	2.25	0.09	2.72
Tl	2.5 ng/g	<2.4	--	--	--	0.96	0.52	2.72
Pb	0.01 μg/g	0.50	0.03	0.07	0.03	0.56	0.83	2.72

^*Less than the method repeatability

3.4. Homogeneity of NYS horn RMs

For Pb, the main element of interest in these RMs, average and range for two within-vial Pb measurements for 12 vials are shown in Figure 1. The between-vial average and plus or minus twice the between-vial SD are plotted. The within-vial average for vial 1 of material 18–04 was identified as an outlier by GESD and excluded, as were outliers for other elements, as recommended by Fearn and Thompson [58].

Figure 1:

Plots of Pb homogeneity in NYS RMs 18–01 through 18–04. Error bars represent two within-vial measurements (max, min). The solid black line represents the characterization study mean value derived from between-vial (bb) measurement, and dashed lines represent the between-vial SD multiplied by 2.

For all elements and RMs, Table 6 shows the grand mean of all homogeneity measurements $(\overline{x}),$ the pooled within-vial standard deviation from 12 pairs of vials (s_wb), the between-vial standard deviation (s_bb), the method repeatability (s_r), and the combined uncertainty due to between-vial inhomogeneity (u_c,bb). The combined uncertainty was calculated using Equation 2, based on recommendations from ISO in Guide 35 [54].

$$u_{c,bb}=\sqrt{s_{bb}^2+s_r^2}$$ Equation 2

The within-bottle standard deviation, s_wb, may be used as a measure of method repeatability, s_r, in some cases [54]; however, in this study, because only two within-vial measurements were available, a separate repeatability study was conducted to obtain a more reliable estimate. As expected, Table 6 shows relatively good agreement between s_wb and s_r for most analytes, with the exceptions of Cr in NYS RM 18–03 and Ba and Pb in NYS RM 18–04.

As a measure of precision, method repeatability indicates the size of the effects that can be resolved by an analytical method [60]. Therefore, in order for the estimates of within- and between-vial homogeneity to be quantitative in the current study, s_r should be smaller than s_wb and s_bb [59]. Table 6 shows that this criterion is met in most cases; s_wb and s_bb values that are lower than s_r are denoted by an asterisk. Results of the ANOVA test reveal homogeneous between-vial distribution of all 17 elements with the exceptions of Co in RM 18–03 and Mn in RM 18–04. In these two instances, contamination likely resulted from the milling apparatus used for homogenization of the horn materials.

3.5. Commutability analysis

Commutability is a property of a RM that makes it suitable for the analyses of authentic human samples by at least two measurement procedures [60]. Results for clinical samples obtained using two or more measurement procedures (i.e., instrumental techniques) that are calibrated using commutable RMs will be equivalent [61]. Therefore, commutability may be assessed by analyzing clinical samples and RMs together by two different analytical techniques, and assessing agreement in the resulting data.

Commutability between the NYS horn RMs and human nail samples, as well as between hair CRMs and nail samples, was assessed by analyzing samples by both ICP-MS/MS (six method validation runs including all 24 digests per NYS RM) and ICP-OES (three runs including 1 digest per NYS RM). Values included in the commutability analysis represent quantifiable measurements from three ICP-OES runs. Commutability analysis of keratinized RMs to human nails was carried out for nine elements (Ba, Ca, Cr, Cu, Mn, Pb, Se, Sr, and Zn) using 95% confidence intervals of the regression line for the nail samples measured by ICP-MS/MS and ICP-OES, based on a previous approach [62]. The confidence interval, shown in gray in Figure 2, indicates the range within which horn materials and hair CRMs are deemed to have acceptable commutability to human nails.

Figure 2:

Commutability of NYS horn RMs and hair CRMs to human nails based on ICP-MS/MS and ICP-OES. Error bars represent the standard deviation of 3 runs for ICP-OES data, and the combined uncertainty for ICP-MS/MS data.

Figure 2 shows a plot of Pb values obtained for all nail samples, NYS horn RMs, and hair CRMs by ICP-OES (Y-axis) and ICP-MS/MS (X-axis). Table S4 (see ESM) summarizes the results of commutability analysis for all nine elements; symbols indicate the RMs that do not fall within the 95% confidence interval for human nails, suggesting a lack of commutability.

For Pb, many of the human nail measurements reflect low (background) Pb exposures and are below the ICP-OES method LOD, which limited the robustness of the commutability study. Nonetheless, the limited data support commutability of two of the horn RMs to human nails (Figure 2). The Pb values in CRMs NIES No. 13 and GBW 07601 are out of the range of the human nail Pb values, and Pb in horn RMs 18–01 and 18–02 are below the ICP-OES LOD, such that the commutability of these materials cannot be assessed.

According to these data, several RMs are not commutable to human nails for specific elements: 1) GBW 07601 for Sr, 2) NIES No. 13 for Zn, 3) horn RM 18–03 for Ba, Cr, and Zn, and 4) horn RM 18–02 for Mn (ESM Table S4). For several of these elements, i.e., Ba, Mn, Sr, and Zn, lack of commutability is related to the excellent precision and agreement of results between ICP-MS/MS and ICP-OES, resulting in a narrow 95% confidence interval. The discrepancy in measurements reported by the two methods for Cr in RM 18–03 was due to the analysis of a single digest by ICP-OES that was likely contaminated (ESM Table S4). A total of 24 digests was analyzed by ICP-MS/MS, and Cr contamination was confirmed in the single digest analyzed by ICP-OES. Consequently, the Cr data for RM 18–03 are unreliable, and cannot be used to establish Cr commutability for that particular RM. For Se, the LOD by ICP-MS/MS is ~0.08 μg/g, while the LOD by ICP-OES is ~1.1 μg/g, an order of magnitude worse. The Se data obtained for these horn RMs by ICP-OES cannot be quantified with confidence, and therefore cannot be used to establish commutability.

One or more of the hair CRMs contains levels that are above the highest human nail measurement for the elements Ba, Ca, Cu, Mn, and Se (ESM Table S4). For all elements but Se, the NYS horn RMs contained similar elemental levels to the human nail samples, i.e., horn RM measurements were within the range of human nail measurements.

These data suggest that the caprine horn RMs produced and characterized in this study are commutable to human nails, and therefore suitable for use as quality control materials for the determination of Pb and up to eight other elements in human nails. The data also demonstrate that the elemental levels present in typical human nail samples may be more similar to the NYS horn RMs than to currently available hair CRMs.

4. Conclusions

While keratinized tissues, such as hair and nails, may offer some advantages for assessing trace element status, analytical issues in sample preparation procedures, instrumental methods, and an insufficient availability and variety of relevant CRMs present challenges to validating measurements for some key toxic elements, including Pb.

Four RMs based on caprine horns derived post-mortem from goats that had been dosed with Pb were produced to support calibration, quality control, and method validation for keratinized tissue trace element analysis. These RMs, NYS RMs 18–01 through 18–04, were developed as a resource for use in method development studies and for validating trace element analysis of keratinized tissues. After producing the four materials from 13 caprine horns, the content of 17 trace elements in the RMs was characterized, and homogeneity and commutability to human nails was assessed. The method was validated against three CRMs based on human hair, finding good accuracy, detection limits, and method repeatability better than 25% for all analytes. All but two elements (Co and Mn) measured were found to be homogeneous in each of the four RMs using an ANOVA test. Commutability between two or more of the caprine horn RMs and human nails was established for Ba, Ca, Cr, Cu, Mn, Pb, Sr, and Zn, based on analyses by two different techniques (ICP-MS/MS and ICP-OES). The homogeneity and commutability data indicate that, for the elements listed above, NYS RMs 18–01 through 18–04 are generally suitable for use as RMs in trace element analyses of human nails. Additional work is required to confirm more thoroughly the commutability of these horn RMs to human hair and analyze the long-term stability of the horn RMs.

The RMs described here, and the discussion of their production and characterization, have the potential to provide a critical resource to the analytical community and to improve harmonization for trace element analysis of keratinized tissue in the future. As a result of this study and an associated interlaboratory study [16], a set of four practical RMs with assigned target values were produced and limited amounts are available to other research groups for use in method development and validation studies.

Table 6b:

Trace element content and homogeneity of NYS RM 18–02: means, uncertainties due to inhomogeneity, and ANOVA heterogeneity test results.

NYS RM 18–02
		Uncertainty			ANOVA Test
	sr	$\overline{x}$	sbb	swb	uc,bb	F	p	Fcrit
Ca	11 μg/g	591	15	27	19	0.63	0.77	2.72
V	7 ng/g	23	*4	*4	8	1.80	0.16	2.72
Cr	0.09 μg/g	0.35	0.11	0.21	0.14	0.52	0.85	2.72
Mn	0.03 μg/g	0.84	0.05	0.06	0.06	1.56	0.23	2.72
Co	12 ng/g	535	22	32	25	0.94	0.54	2.85
Cu	0.07 μg/g	4.61	0.07	0.07	0.10	1.57	0.23	2.72
Zn	1 μg/g	113	1	2	2	1.58	0.22	2.72
As	2 ng/g	11	3	3	4	2.68	0.05	2.72
Se	26 ng/g	302	*18	41	32	0.40	0.93	2.72
Rb	9 ng/g	352	9	14	13	0.85	0.61	2.72
Sr	14 ng/g	353	*10	15	17	0.93	0.54	2.72
Mo	9 ng/g	86	*6	10	10	0.61	0.79	2.72
Cd	13 ng/g	6.6	1.4	2.6	1.9	0.54	0.84	2.72
Ba	8 ng/g	293	15	30	17	0.52	0.85	2.85
Hg	19 ng/g	7.6	5.5	9.2	5.8	0.73	0.70	2.72
Tl	2.5 ng/g	<2.4	--		--	0.82	0.63	2.72
Pb	0.01 μg/g	0.77	0.03	0.06	0.03	0.37	0.95	2.72

^*Less than the method repeatability

Table 6c:

Trace element content and homogeneity of NYS RM 18–03: means, uncertainties due to inhomogeneity, and ANOVA heterogeneity test results.

NYS RM 18–03
		Uncertainty			ANOVA Test
	sr	$\overline{x}$	sbb	swb	uc,bb	F	p	Fcrit
Ca	11 μg/g	562	18	66	21	0.15	1.00	2.85
V	7 ng/g	29	*3	*5	8	0.11	1.00	3.02
Cr	0.09 μg/g	1.10	1.15	1.72	1.15	0.90	0.57	2.72
Mn	0.03 μg/g	0.78	0.05	0.08	0.06	0.82	0.62	2.72
Co	12 ng/g	498	240	42	240	66.57	<0.01	3.02
Cu	0.07 μg/g	3.78	*0.05	0.13	0.09	0.29	0.96	3.02
Zn	1 μg/g	115	1	4	2	0.27	0.97	2.85
As	2 ng/g	13	5	8	5	0.77	0.67	2.72
Se	26 ng/g	230	29	26	39	2.56	0.06	2.72
Rb	9 ng/g	279	*8	*8	12	2.32	0.08	2.72
Sr	14 ng/g	355	*12	20	18	0.73	0.70	2.72
Mo	9 ng/g	72	12	14	15	0.95	0.53	2.85
Cd	1.3 ng/g	6.8	1.6	2.5	2.0	0.78	0.66	2.72
Ba	8 ng/g	366	70	104	71	0.92	0.55	2.72
Hg	1.9 ng/g	6.7	5.1	6.9	5.4	1.09	0.44	2.72
Tl	2.5 ng/g	<2.4	--	--	--	0.24	0.99	2.72
Pb	0.01 μg/g	1.20	0.03	0.07	0.03	0.27	0.98	2.72

^*Less than the method repeatability

Table 6d:

Trace element content and homogeneity of NYS RM 18–04: means, uncertainties due to inhomogeneity, and ANOVA heterogeneity test results.

NYS RM 18–04
		Uncertainty			ANOVA Test
	sr	$\overline{x}$	sbb	swb	uc,bb	F	p	Fcrit
Ca	11 μg/g	637	33	43	35	1.34	0.32	2.85
V	7 ng/g	58	*6	*5	9	1.49	0.25	2.72
Cr	0.09 μg/g	0.98	0.12	0.20	0.14	0.77	0.67	2.72
Mn	0.03 μg/g	1.17	0.04	0.03	0.05	3.37	0.04	3.02
Co	12 ng/g	1145	52	97	53	0.64	0.75	3.02
Cu	0.07 μg/g	4.72	0.07	0.15	0.10	0.44	0.88	3.02
Zn	1 μg/g	124	3	4	3	0.92	0.55	2.85
As	2 ng/g	18	2	5	3	0.33	0.96	2.72
Se	26 ng/g	291	26	41	37	0.92	0.55	2.72
Rb	9 ng/g	335	19	30	21	0.86	0.60	2.72
Sr	14 ng/g	390	29	37	32	1.33	0.31	2.72
Mo	9 ng/g	134	17	24	19	1.13	0.42	2.85
Cd	1.3 ng/g	7.6	1.3	2.2	1.8	0.76	0.67	2.72
Ba	8 ng/g	570	75	126	76	0.78	0.65	2.72
Hg	1.9 ng/g	§ 53	--	--	--	0.61	0.79	2.72
Tl	2.5 ng/g	<2.4	--	--	--	1.23	0.36	2.72
Pb	0.01 μg/g	3.53	0.12	0.15	0.12	1.39	0.30	2.85

^*Less than the method repeatability

^§See text for detailed explanation