Detection of Erionite and Other Zeolite Fibers in Soil by the Fluidized Bed Preparation Methodology

Abstract

Erionite is a zeolite mineral that can occur as fibrous particles in soil. Inhalation exposure to erionite fibers may result in increased risk of diseases, such as mesothelioma. Low level detection of mineral fibers in soils has traditionally been accomplished using polarized light microscopy (PLM) methods to analyze bulk samples providing detection limits of around 0.25% by weight. This detection level may not be sufficiently low enough for protection of human health and is subject to large variability between laboratories. The fluidized bed asbestos segregator (FBAS) soil preparation method uses air elutriation to separate mineral fibers, such as erionite, from soil particles with higher aerodynamic diameter and deposits those mineral fibers onto filters that can be quantitatively analyzed by microscopic techniques, such as transmission electron microscopy (TEM). In this study, performance evaluation (PE) standards of erionite in soil with nominal concentrations ranging from 0.1% to 0.0001% by weight were prepared using the FBAS soil preparation method and the resulting filters were analyzed by TEM. The analytical results of this study illustrate a linear relationship between the nominal concentration of erionite (as % by weight) in the PE standard and the concentration estimated by TEM analysis expressed as erionite structures per gram of test material (s/g). A method detection limit of 0.003% by weight was achieved, which is approximately 100 times lower than typical detection limits for soils by PLM. The FBAS soil preparation method was also used to evaluate authentic field soil samples to better estimate the concentrations of erionite in soils on a weight percent basis. This study demonstrates the FBAS preparation method, which has already been shown to reliably detect low levels of asbestos in soil, can also be used to quantify low levels of erionite in soil.

INTRODUCTION

Erionite and offretite are natural zeolites having different topologies and both are found in sedimentary environments, as products of hydrothermal alteration, and in vugs of altered basalts (1). Erionite is the more common of the two mineral zeolites, whereas offretite occurrences are scarce. Both erionite and offretite are aluminosilicates that can form in an acicular habit similar to amphibole asbestos, which are generally Fe-Mg silicates. Erionite occurs naturally when volcanic ash reacts with groundwater and forms fibrous masses in rock fault sand fissures. The precise mineral species of erionite is classified according to the most abundant extra framework cation present, i.e., Ca, K, or Na (2, 3). Erionite and offretite are chemically and structurally very similar species (4, 5). Erionite-offretite intergrowths may occur (6, 7) but are probably rare and restricted to samples at the high Mg limit of the erionite field (8). Chemical data alone is insufficient to differentiate between zeolite species. Structural data, such as determined by selected area electron diffraction (SAED) or powder X-ray diffraction (XRD), must also be considered before a zeolite can be identified to a specific mineral species.

Erionite is carcinogenic in laboratory animals and induces autoantibodies in mice (9–12). Both asbestos and erionite exposures can lead to pleural fibrosis and malignant mesothelioma, and there have been high rates of mesothelioma in villages located in central Turkey, where the residents have been exposed to erionite (2, 3, 13–16) and in Central Mexico (17). Erionite is listed as a known human carcinogen by the World Health Organization (18). Offretite is only suspected to be carcinogenic based on its fibrous structure and strong mineralogical similarities to erionite (19). Exposures to offretite have not been well documented in the literature.

Many geologic formations contaminated with erionite have been located throughout the western U.S. (20, 21). In North Dakota, erionite deposits are located in or near the Arikaree, Brule, and Chadron geologic formations including the Chalky Buttes, Little Badlands, and the Killdeer mountain areas in Slope, Stark, and Dunn counties. Several gravel pits in the Killdeer mountain area contaminated with erionite have been excavated in western North Dakota (Dunn County) and eastern Montana (Custer National Forest) (22, 23). The gravel was used for road base and cover. Since the 1980s, these gravels have been used to construct more than 300 miles of local roads, parking lots, baseball fields, playgrounds, and other areas (15, 24).

A sampling technique known as activity-based sampling (ABS) demonstrated that average airborne concentrations of erionite (reported in terms of phase contrast microscopy-equivalent [PCME] structures) in Dunn County (detected range from 0.0004 to 0.0575 structures per cubic centimeter [s/cc]) were within the range of those found in Turkish villages (detected range from 0.0017 to 1.737 s/cc) experiencing mesothelioma rates as high as 1-in-1,000 (15). This is compared to rates of 1- to 15-per-1,000,000 in most areas of the U.S. even where asbestos exposure is high (15). No detectable increase in mesothelioma rates in North Dakota have been observed, but the latency period for expression of this tumor type may exceed 40 to 50 years depending on the level of exposure (25). Given the long latency between the development of mesothelioma and the relatively recent use of erionite-contaminated gravel in North Dakota, it still may be too early to observe any adverse health outcome. A recent study of Dunn County, ND road workers, without any history of asbestos exposure, did report evidence of fibrotic pleural disease consistent with erionite exposures (25). In an investigation near Battle Mountain, NV, erionite was discovered in road dust and a population of local residents were screened at a local community hospital for pulmonary outcomes. Results indicated the prevalence of pleural plaques without pleural calcification in 1.8% of the participants (n = 275) and blunting of the costophrenic angle in 6.2% of the participants (26).

Given the demonstrated toxicity and health outcomes associated with erionite exposures in laboratory animal and humans, it is important to develop low-level detection capabilities of these mineral particles in soil to better identify localities where potential human exposures might be significant. Detection of erionite in soils is currently limited to scanning electron microscopy (SEM) and transmission electron microscopy (TEM) via the wet preparation method (27). Screening level analysis, such as polarized light microscopy (PLM), of erionite is probably limited to only a few laboratories that have access to high-dispersion refractive index liquids with the appropriate refractive index range (n = 1.460 to 1.480) and have developed a modification to NIOSH Method 9002 to semi-quantitate zeolite fibers. Conventional TEM analysis of erionite may be hampered by the fact that the high-energy beam (up to 100 kilovolts [kV]) has been shown to deform erionite structures and may partially evaporate certain cations, such as Na, during analysis, making accurate identification problematic (28).

Previous research and method development demonstrated that low level (0.002% to 0.005% by weight) detection of asbestos fibers in soils can be readily attained using the fluidized bed asbestos segregator (FBAS) preparation method and analysis by TEM (29). Farcas, et al. (30) recently reported using the FBAS soil preparation method followed by phase contrast microscopy (PCM) analysis for detecting erionite from soils sampled in eastern Montana and western South Dakota. They compared the fluidized bed-PCM analysis to PLM analysis of erionite in soils and found the FBAS-PCM method to be more sensitive than PLM. This paper describes a performance evaluation (PE) study of the FBAS soil preparation method for soils containing erionite and presents results based on TEM analysis of these mineral fibers. This paper also describes the use of the FBAS preparation method and analysis by TEM for several authentic field samples collected from areas in North Dakota and Montana where erionite is known to be present.

METHODS

Preparation of Erionite Source

The erionite used for this study was acquired by the U.S. Environmental Protection Agency (USEPA) in May 2008 from federal land outside of Rome, OR, which is managed by the U.S. Bureau of Land Management (BLM). The USEPA Region 10 laboratory characterized the specimen by XRD (Figure 1A) and SEM coupled with energy-dispersive X-ray spectroscopy (EDS) (Figure 1B) and found it to contain erionite, volcanic glass, and minor concentrations of calcite and quartz. The XRD analysis was performed with a Scintag X-1 X-ray diffractometer equipped with a cobalt X-ray tube. The SEM-EDS analysis was performed with a JEOL JSM6510LV electron microscope equipped with an EDAX Genesis XM2 EDS Analyzer System with a lithium-drifted silicon detector.

Figure 1.

Composition and morphology of the Rome, OR source erionite as determined by 1A, powder XRD and 1B, SEM with EDS. Inset: SEM image of erionite fibers.

The Rome erionite specimen was processed at the USEPA Region 10 laboratory by lightly crushing it to a powder with the aid of a ceramic mortar and pestle. This powder was suspended in laboratory-grade deionized water and allowed to settle, resulting in separation of erionite from the larger heavier particles, and the suspension was filtered with the aid of a vacuum filtration apparatus using a 0.4 μm pore size polycarbonate filter. The resulting specimen was characterized by the Research Triangle Institute (RTI) and found to contain an average of 87% to 90% erionite. This specimen was used by RTI to prepare PE standards used for this study.

Preparation of Performance Evaluation Standards

The PE standards were prepared by spiking test soil from Arvada, CO (Arvada Soil series) with known amounts of erionite. Each PE standard consisted of 50 g of soil containing nominal concentrations of 0.1%, 0.01%, 0.001%, or 0.0001% erionite by weight. This was accomplished by creating a 0.1% base sample using gravimetric addition. Preparation of the base sample was performed by blending 1 g of erionite at high speed in 375 mL of water and then filtering 250 mL of the suspension onto a 0.8 μm polycarbonate filter. The lower 125 mL of suspension contained some slightly granular material and was discarded. The filter was then dried, and a 0.2 g aliquot was removed and weighed for combining with the soil. The selected aliquot was suspended in a 30 mL 50:50 mixture of water and isopropanol while being stirred and ultra-sonicated for 5 minutes with the aid of a sonic dismembrator. The suspension was then poured over 200 g of Arvada soil and then dried in an open container in a HEPA-filtered fume hood for 48 hours. The soil/erionite mixture was then shaken in a Turbula mixer for 30 minutes. Because the erionite was added by gravimetric addition to the 0.1% base sample, the percentage of erionite in the sample was known. Thus, quantification of the erionite concentration by PLM point counting or visual estimation was deemed unnecessary. The analysts preparing the PE samples were instructed to verify the 0.1% sample had a homogenous distribution of well-separated elongate erionite particles, and to look for any irregularities that would suggest that there was a problem with the spiking and/or the ho mogenization. This was achieved by examining 20 aliquots of the 0.1% base sample by PLM. The analyst determined the erionite was visible and homogenously distributed throughout the sample for all 20 aliquots.

Aliquots of the base 0.1% sample were then removed for serial dilution to make the 0.01%, 0.001%, and 0.0001% PE standards. Each of the serial dilution PE standards was re-shaken in the Turbula mixer for 15 minutes. The materials were then sealed into double-packed zipped plastic storage bags and labeled as follows: “Arvada Soil, 50 grams, xx% Erionite.”

Field Soil Samples

In addition to the PE standards, several field soil samples from two different states were also analyzed as part of this study. Soil samples from Dunn County, ND and the vicinity of Custer National Forest, MT were retrieved from archived soil samples collected by the USEPA in 2008 (24). Samples DCRD-10 and DCRD-16 (T146N/R96W) were gravel samples taken along unpaved roads in Dunn County within one mile of the Killdeer Mountains, where erionite and offretite are known to occur (31). When analyzed by PLM using visual area estimation, both samples were noted to contain visible tuff fragments; sample DCRD-10 was reported as containing “trace” amounts (<0.2%) of erionite, while sample DCRD-16 was reported as non-detect. Sample Custer-1 was taken from a gravel quarry (T3S/R59E) located near the Custer National Forest, approximately 150 miles southwest of the Killdeer Mountains. PLM analysis using visual area estimation reported the Custer-1 sample contained about 1% erionite.

Fluidized Bed Asbestos Segregator Preparation Method

The FBAS preparation method, illustrated and described previously by Januch, et al. (29), simulates soil disturbance activities in a controlled, reproducible laboratory setting. The FBAS method was used to prepare each sample to elutriate zeolite mineral fibers from the soils and deposit them onto mixed cellulose ester (MCE) filters for TEM analysis. The FBAS preparation method followed the process described in Standard Operating Procedure (SOP) OEAFIELDSOP-102 (Revision 1.0, August 2011). In brief, an aliquot (approximately 1–2 g) of erionite-containing soil was mixed with analytical grade sand (ASTM 20/30 analytical grade sand, e.g., Restek Corporation Ottawa sand) to yield a total of 20 g of sample, which was loaded into the funnel-shaped base of the glass fluidizing chamber. An air sampling pump then drew HEPA-filtered air at a rate of approximately 16 L/minute into the fluidizing chamber. The resulting action fluidized the sand/soil mixture, elutriating small particles so that they became airborne inside the chamber. A standard PCM air sampling cassette with an extended cowl collected the airborne particles at the top of the fluidizing chamber for a sampling duration of 3 minutes. To prevent overloading of the filter, an isokinetic splitter directed 1/80th of the airflow into the cassette and remainder of the airflow into a waste stream. The ratio of soil-to-sand was adjusted to achieve a filter loading less than 25% by area, as verified by initial PCM analysis of the resulting test filters. Following the fluidizing process, the MCE filter cassette was removed and prepared for shipment to the TEM laboratory. A total of five MCE filter replicates were generated for each PE standard, and three filter replicates were generated for each field soil sample. Quality control (QC) samples consisted of lot blanks, laboratory blanks, preparation blanks, and sand blanks.

Filter Analysis Method and Results Reporting

Filters were prepared and analyzed for mineral fibers in basic accordance with the International Organization for Standardization (ISO) Method 10312:1995(E) (32). TEM analyses were performed with a JEOL model JEM-1011 transmission electron microscope at an accelerating voltage of 100 kV and a magnification of approximately 20,000×. During TEM analysis, elemental composition of individual structures was determined with an Oxford INCA 100 EDS system.

EDS spectra were recorded for a representative sub-sample of the structures from each filter. Background-corrected peak areas were integrated for each element of interest and converted to weight percent by the Cliff-Lorimer approach. The k-factors used in this approach were calibrated with Madagascar orthoclase from the Charles M. Taylor collection for elements K and Al, and the BIR-1G basalt glass standard (33) for the elements Na, Mg, Ca, and Fe. The resulting weight percent estimates for each element were then converted to atoms per formula unit (apfu) on the basis of 72 oxygens and results were plotted on a compositional ternary plot. The ratio of Mg/(Ca+Na) can be used to help discriminate between erionite (Na₂K₂Ca₃)[Al₁₀Si₂₆O₇₂]·30H₂O) and offretite (KCaMg)[Al₅Si₁₃O₃₆]·15H₂O), with erionites showing values of less than 0.3 and offretites showing values close to 1.0 (1).

SAED patterns were recorded photographically for a representative subsample of the zeolite structures that were counted during TEM analysis. The process of indexing an SAED pattern consists of measuring the d-spacings (distances between planes of atoms in the crystal lattice) and angles between the planes and comparing them to the expected values for a particular mineral. Susceptibility of the zeolite structures to electron beam damage made photographic capture of SAED patterns difficult.

Each structure with a length of 0.5 μm or longer, an aspect ratio (length to width) of 3:1 or higher, and EDS and SAED characteristics consistent with erionite/offretite, was recorded as a countable structure. The analysis of each sample was terminated when 25 structures were counted or when the target analytical sensitivity of 25,000 per gram (g^–1) was achieved.

The concentration of erionite in soil reported from the TEM analysis may be expressed in two alternate ways: structures per gram of soil (s/g) and mass percent.

The basic formula for calculating soil concentration as structures per gram of soil (s/g) is as follows:

$${\text{C}}_{\text{soil}}\left(\text{s}/\text{g}\right)=\text{N}\times \text{S}$$

where:

• N = Number of erionite structures counted

• S = Analytical sensitivity (g^–1)

The analytical sensitivity (S) is expressed as the inverse of the amount of soil examined during the analysis (per gram; g^–1) and is calculated as follows:

$$\text{S}\left({\text{g}}^{-1}\right)=\text{EFA}/\left(\text{GO}\times {\text{A}}_{\text{GO}}\times \text{M}\times {\text{Q}}_{\text{R}}\right)$$

where:

• EFA = Effective filter area (mm²)

• GO = Number of TEM grid openings evaluated

• A_GO = Area of one TEM grid opening (mm²)

• M = Mass of soil placed in the FBAS (g)

• QR = Flow ratio; this is the fraction of the volume of air passed through the soil sample (V_total) that is filtered through the air filter (V_filter)

In order to express soil concentration as mass percent, the mass of each erionite structure observed is estimated from its dimensions. In the absence of detailed data on the true geometry of each particle, the mass is roughly approximated by assuming a simple rectangular solid shape, as follows:

$${\text{m}}_{\text{i}}\left(\text{g}\right)=1_{\text{i}}\times {\text{w}}_{\text{i}}{}^2\times \delta \times 1\text{E-}12$$

where:

• l_i = Length of structure i (μm)

• w_i = Width of structure i (μm)

• δ = Density of erionite (2.11 g/cm³)

• 1E-12 = Conversion factor (cm³/μm³)

The soil concentration as mass percent (grams of erionite per 100 g of soil) is then calculated as follows:

$${\text{C}}_{\text{soil}}\left(\text{mass percent}\right)=\mathrm{\Sigma }{\text{m}}_{\text{i}}\times \text{S}\times 100$$

where:

• Σm_i = Total mass of all observed erionite structures (g)

• S = Analytical sensitivity (g^–1)

Structure counts, concentrations, and structure dimension attributes were recorded in a standard Microsoft Excel electronic data deliverable designed specifically for FBAS TEM analyses. These electronic data were entered into a Microsoft Access database providing for ease of data analysis.

RESULTS AND DISCUSSION

Linearity of Results

Table 1 summarizes statistics of the concentration results for each type of PE standard. Mean concentrations were computed across all filter replicates for each PE standard, using a value of zero for filter replicates where no structures were observed (29). Mean concentrations are shown both as structures per gram and as mass percent. Figure 2 presents the results (as structures per gram) for each type of PE standard. Although the results are variable, there is an approximately linear relationship between the nominal concentration and the mean measured concentration. These data were fit using Poisson maximum likelihood estimation to the following linear model:

$$\text{Csoil}\left(\text{s}/\text{g}\right)=\text{k}\times {\text{C}}_{\text{soil}}\left(\text{mass percent}\right)$$

where the Poisson parameter λ is given by:

$$\mathrm{\lambda }\text{=k}\times {\text{C}}_{\text{soil}}\left(\text{mass percent}\right)/\text{S}$$

Table 1.

Erionite PE Standard Results

PE Std Nominal (%)	Replicate Filter #	Sensitivity (1/g)	N Structures Observed	Soil Conc. (s/g)			Soil Conc. (mass%)			% Recovery
				Result	Mean (across replicates)	Coeff. Var. (CV)	Result	Mean (across replicates)	Coeff. Var. (CV)
0.0001	1	3.0E+04	0	0.0E+00	0.0E+00	---	0	0	---	0%
	2	3.1E+04	0	0.0E+00			0
	3	3.0E+04	0	0.0E+00			0
	4	2.4E+04	0	0.0E+00			0
	5	2.5E+04	0	0.0E+00			0
0.001	1	3.0E+04	0	0.0E+00	2.2E+04	113%	0	0.00021	149%	21%
	2	3.0E+04	2	6.1E+04			0.00029
	3	3.0E+04	0	0.0E+00			0
	4	2.5E+04	1	2.5E+04			0.000030
	5	2.5E+04	1	2.5E+04			0.00071
0.01	1	3.0E+04	12	3.6E+05	3.3E+05	35%	0.0015	0.0017	109%	17%
	2	3.0E+04	6	1.8E+05			0.000059
	3	3.1E+04	9	2.8E+05			0.00057
	4	2.5E+04	20	5.0E+05			0.0048
	5	2.4E+04	14	3.4E+05			0.0016
0.1	1	2.2E+05	26	5.7E+06	4.4E+06	48%	0.021	0.020	79%	20%
	2	7.2E+04	25	1.8E+06			0.0036
	3	2.7E+05	27	7.2E+06			0.045
	4	1.4E+05	26	3.5E+06			0.0096
	5	1.5E+05	26	3.8E+06			0.022

Figure 2.

Results (as s/g) for erionite PE standards. Individual sample results for each filter replicate are shown as blue diamonds. The mean concentration across replicate filters is shown as a horizontal bar. Non-detects are shown as open symbols and are plotted at the achieved sensitivity. Best linear fit is shown, as determined using maximum likelihood estimation. Inset: best linear fit equation and correlation index based on the group mean and individual samples.

As shown, the linear model fits the data well, with an R² value of 0.72 based on the individual samples and an R² value of 0.93 based on the group mean (across all replicates).

Accuracy

Accuracy is assessed by analyzing a number of samples for which the nominal concentration value is known (i.e., PE standards), and comparing the measured concentration values to the nominal values. Because the nominal levels in the PE standards are based on mass percent, in order to evaluate accuracy, results must also be expressed as mass percent. Table 1 presents the reported mean concentration (as mass percent) and the percent recovery for each type of erionite PE standard. Figure 3 illustrates these results (as mass percent) for each type of PE standard. Reported concentrations are consistently low, with recoveries of about 20% for the PE standards with nominal levels of 0.001% or higher. This means that only a fraction of the total mass of erionite added to the PE standard is captured on the TEM filter.

Figure 3.

Results (as mass percent) for erionite PE standards. Individual sample results for each filter replicate are shown as blue diamonds. The mean concentration across replicate filters is shown as a horizontal bar. Non-detects are shown as open symbols and are plotted at 0.00001%. The dotted line shows the line of identity, which is the 1:1 line where x=y.

Precision

Precision is the ability of a method to yield reproducible results when the same sample is analyzed more than once. Within-laboratory precision is usually characterized in terms of the coefficient of variation (CV = standard deviation/mean). Table 1 presents the within-laboratory CV results for each type of PE standard. The CV is shown based on both mass percent and structures per gram. CVs ranged from 0.35 to 1.49 and tended to be higher when nominal levels were lower. The CVs for results reported as mass percent generally tended to be higher than CVs for results reported as structures per gram. The difference in precision for results as structures per gram (Figure 2) compared to mass percent (Figure 3) is also apparent in the figures.

Method Detection Limit

The method detection limit (MDL) is usually defined in terms of the concentration that must be present in a sample such that the method has a high probability of detecting the presence of the analyte. Thus, the MDL can be determined by plotting the detection frequency as a function of the nominal concentration in PE samples, and identifying the concentration that is detected at the desired frequency. For example, if the MDL were defined as the concentration where probability of detection is 80% (an arbitrary choice), then the MDL would be about 0.003% (Figure 4). This achieved detection limit is similar to that reported in earlier FBAS method validation studies for asbestos (29) and well below the typical detection limits achieved by PLM, which is about 0.25% using point-counting techniques and <1% using visual area estimation.

Figure 4.

Method detection limit for erionite PE standards. Reported detection frequency as a function of the nominal level in the PE standard. Dotted arrow shows the estimated method detection limit based on a detection frequency of 80%; mean achieved TEM analysis sensitivity was 28,000 g^–1.

The probability of detection, and hence the MDL, is determined primarily by the analytical sensitivity of the TEM analysis. The typical TEM analytical sensitivity achieved during this study was usually about 28,000 g^–1. If the TEM examination of these samples had examined a higher number of grid openings, this would have lowered the analytical sensitivity, increased the detection frequency at low concentrations, and decreased the MDL to an even lower concentration.

Field Soil Results

Figure 5 presents the results (as structures per gram) for the field soil samples from gravel roads in Dunn County, ND and the gravel quarry in Custer National Forest, MT. In this figure, the results of the erionite PE standards are also shown for reference. Erionite concentrations in the samples from the road (8E + 04 to 2E + 05 s/g) were lower than concentrations from the quarry (8E + 06 s/g). Based on a comparison to the PE standards, the mass percent was between the 0.001% and 0.01% PE standards for the Dunn County road samples and higher than the 0.1% PE standard for the Custer National Forest quarry sample. If the linear model equation from Figure 2 were used to estimate the mass percent, the estimated erionite concentration was about 0.005% for the Dunn County road samples and about 0.28% for the Custer National Forest quarry sample.

Figure 5.

Results for field soil samples. Individual sample results (as s/g) for each authentic field soil sample. PE standard results are presented for reference and the secondary y-axis presents approximate corresponding soil concentration levels as mass percent.

Chemical Composition and Mineralogy

Erionite from the Rome, OR area is generally classified as erionite-K (2, 3, 14, 34); however, EDS spectra showed significant Ca peaks for some structures in the PE standards, indicating erionite-Ca may also be present.

SAED photos from eight different structures — five from the PE standards (Rome), 2 from the DCRD-10 Dunn County sample, and 1 from the Custer National Forest sample — were of sufficient quality to be indexed. Based on the indexed photos, it was determined that four of the five Rome structures were erionite, whereas the fifth Rome structure and all three of the Dunn County and Custer National Forest structures were offretite. Erionite has unit-cell parameters a ≈ 13.15 and c ≈ 15.05 Å whereas offretite has unit-cell parameters a ≈ 13.30 and c ≈ 7.60 Å (8). D-spacings and interfacial angles were calculated from these unit-cell parameters and compared to measured values from the SAED photos (Table 2).

Table 2.

Indexed SAED Patterns from Eight Different Zeolite Structures

Location	Mineral	(000l) Plane			(hkil) Plane			Interfacial Angle		Zone
		Plane	Standard d (angstroms)	Measured d (angstroms)	Plane	Standard d (angstroms)	Measured d (angstroms)	Standard (degrees)	Measured (degrees)
Rome	erionite	(0002)	7.53	7.46	(10–10)	11.39	11.19	90.0	90	[−12–10]
Rome	offretite	(0001)	7.60	7.46	(21–31)	3.78	3.73	60.2	60	[−45–10]
Rome	erionite	(0002)	7.53	7.53	(10–10)	11.39	11.35	90.0	90	[−12–10]
Rome	erionite	(000–1)	15.05	14.77	(−101–1)	9.08	9.10	52.9	53	[−12–10]
Rome	erionite	(0001)	15.05	15.06	(20–21)	5.33	5.29	69.3	69	[−12–10]
Dunn County	offretite	(0001)	7.60	7.46	(10–12)	3.61	3.56	18.3	19	[−12–10]
Dunn County	offretite	(0001)	7.60	7.46	(21–31)	3.78	3.73	60.2	60	[−45–10]
Custer	offretite	(0001)	7.60	7.46	(21–30)	4.35	4.30	90.0	90	[−45–10]

The ISO 10312 method (32) recommends that filter sections be prepared onto copper grids. However, when analyzing for zeolites, the Cu L_α X-ray peak at 930 electron volts (eV) appears as an artifact when EDS is performed on filters prepared onto copper grids. This artifact can obscure the K_α X-ray signal from Na at 1,041 eV. The problem can be avoided by preparing filters onto nickel grids. EDS spectra were recorded for a representative subset of the structures from the three locations which were prepared using nickel grids; results were plotted on a compositional ternary plot (Figure 6). An erionite standard from an outcrop of the Arikaree Formation at the Killdeer Mountains was analyzed by TEM and included in the plot for comparison purposes. These TEM results indicate that both erionite and offretite occur in the Rome and Dunn County locations, but all of the structures from the Custer National Forest sample plotted within the offretite field. The nickel grid results compare favorably with other published erionite/offretite compositions (1, 7).

Figure 6.

Ternary plot of zeolite structure compositions as measured by EDS for nickel TEM grid preparations. Each point shows the atomic proportions of the extra-framework cations of an individual structure. The gray line indicates Mg/(Ca + Na)= 0.3 and is shown as the boundary between the erionite and offretite compositional fields.

Structure Size Distribution

During the TEM analysis, the dimensions (length, width) of each countable structure are recorded. One method for comparing the structure size distributions of different datasets is through a graph that plots the cumulative distribution function (CDF), which illustrates the fraction of all structures that have a dimension less than some specified value. Figure 7 illustrates the structure size distributions for the countable structures recorded for the PE standards (Rome), Dunn County road samples, and the Custer National Forest quarry sample. Distributions are shown based on the recorded length, width, and aspect ratio (length:width).As shown, the CDFs for Rome and Custer are generally similar, but the CDF for Dunn County tends to be somewhat shorter in length have a smaller aspect ratio compared to the other two locations.

Figure 7.

Structure size distribution. Cumulative distribution frequency for 7A, length; 7B, width; and 7C, aspect ratio of structures in PE standards (green series, n= 228 structures), Dunn County road samples (red series, n=36 structures), and Custer National Forest sample (blue series, n=111 structures). Frequency for erionite structures in Dunn County ABS air samples (dotted line series, n=778 structures) also presented for reference.

In 2006, a human exposure study was conducted in Dunn County to evaluate potential concentrations of erionite in air due to source disturbance activities, such as driving, raking, and sweeping (24). Erionite exposures were evaluated using ABS. Structure size distributions from the 2006 Dunn County ABS air samples are also shown in Figure 7. The CDF of countable structures observed in the Dunn County ABS air samples tends to be slightly shorter and thicker than the CDF of structures generated by the FBAS preparation of the Dunn County soil samples; however, the aspect ratio CDFs are nearly identical.

SUMMARY AND CONCLUSIONS

The FBAS preparation method provides a technique for low-level analysis of erionite and other min eral fibers in soils. The analysis of samples obtained with this method yield results that show an approximately linear relationship between the nominal erionite concentrations and the reported erionite concentrations. Analysis of grouped means resulted in a correlation coefficient of R² = 0.93. Percent recovery in this study, while only 20%, was improved compared to earlier PE studies of the FBAS method performed with asbestos (29), which is likely due to recent changes in the FBAS unit hardware. The low recovery suggests only a fraction of the total erionite mass present in soil is releasable to air, which is similar to what is expected in the natural environment. There is variability between filter replicates for a given sample. Precision is much better for results reported as s/g compared to results reported as mass percent. The between-replicate variability also tends to be higher at lower nominal concentrations, due primarily to analytic uncertainty arising from low structure counts.

The MDL achieved in these studies (0.003% by mass) is substantially more sensitive than other available techniques for assessing erionite concentrations in soil (e.g., 0.25% by PLM analysis). The FBAS preparation method allows detection of erionite fibers in soils that are often reported as non-detect when analyzed by PLM. For example, this study showed the Dunn County gravel sample reported as non-detect by PLM contained erionite fibers that were releasable into air. This finding was corroborated by the ABS conducted along the Dunn County roads, which resulted in fiber detections from several exposure scenarios (25).

Use of the FBAS preparation method with PCM has been shown to be a more sensitive method than PLM (30). However, use of FBAS preparation method with TEM not only provides for low level fiber detections, but also offers the benefit of fiber identification using the combination of EDS analysis to determine chemistry and SAED to determine crystal structure of the fibers, rather than solely on structure morphology. The erionite used to prepare the performance evaluation samples came from a location near Rome, OR and was shown to be predominantly erionite-K, as has been reported earlier (2). In this study, both erionite and offretite were identified in the Dunn County samples, which is consistent with previous work in the Killdeer Mountains (31). Fibers in the Custer National Forest quarry sample were identified as offretite.

The FBAS method can report soil concentrations in units of s/g, which is likely to be more directly related to airborne fiber concentrations, which are reported in terms of structures per cubic centimeters of air (s/cc), not mass. The structure size distribution resulting from air elutriation of soils with the FBAS were compared to the structure size distribution from air samples that were collected by ABS from human disturbance of soils. The structure dimension graphs showed the structure distributions were generally similar for Dunn County soil samples generated by the FBAS and air samples collected during ABS conducted on Dunn County roads. These data indicate the structure sizes generated by the FBAS are comparable to structure sizes generated by human disturbance of soils via ABS, which suggests the FBAS provides useful data to inform risk managers on locations where human exposure has the potential to be higher.

This study demonstrates the FBAS preparation method works equally well for both spiked standards and authentic field samples for detection of erionite fibers. When prepared filters are analyzed by TEM, the method allows for quantitative fiber identification using EDS and SAED, and the achieved MDL (0.003% by mass) is much lower than conventional analytical techniques for erionite in soil. The structure size distribution for FBAS-prepared filters is similar to that for activity-based air samples. Because it is possible to process up to 20 soil samples per day using the FBAS preparation method, it is a less expensive and faster alternative to ABS methods in identifying areas with higher potential for generating airborne fibers.