Document 9325M0aL6Y6K46JeEewmO1jE5

American Mineralogist, Volume 88, pages 1955-1969, 2003 The Composition and Morphology of Amphiboles from the Rainy Creek Complex, Near Libby, Montana G.P. Meeker,1,* A.M. Bern,1 I.K. B ,rownfield 1 H.A. Lowers,1,2 S.J. Sutley,1 T.M. Hoefen,1 and J.S.Vance3 'U.S. Geological Survey, Denver Microbeam Laboratory, Denver, Colorado 80225, U.S.A. "Colorado School of Mines, Golden, Colorado, 80401, U.S.A. 3U.S. Environmental Protection Agency, Region 8, Denver, Colorado 80204, U.S.A. Abstract Thirty samples of amphibole-rich rock from the largest mined vermiculite deposit in the world in the Rainy Creek alkaline-ultramafic complex near Libby, Montana, were collected and analyzed. The amphibole-rich rock is the suspected cause of an abnormally high number of asbestos-related diseases reported in the residents of Libby, and in former mine and mill workers. The amphibole-rich samples were analyzed to determine composition and morphology of both fibrous and non-fibrous amphiboles. Sampling was carried out across the accessible portions of the deposit to obtain as complete a representation of the distribution of amphibole types as possible. The range of amphibole compositions, determined from electron probe microanalysis and X-ray diffraction analysis, indi cates the presence of winchite, richterite, tremolite, and magnesioriebeckite. The amphiboles from Vermiculite Mountain show nearly complete solid solution between these end-member composi tions. Magnesio-arfvedsonite and edenite may also be present in low abundance. An evaluation of the textural characteristics of the amphiboles shows the material to include a complete range of morphologies from prismatic crystals to asbestiform fibers. The morphology of the majority of the material is intermediate between these two varieties. All of the amphiboles, with the possible excep tion of magnesioriebeckite, can occur in fibrous or asbestiform habit. The Vermiculite Mountain amphiboles, even when originally present as massive material, can produce abundant, extremely fine fibers by gentle abrasion or crushing. Introduction The Rainy Creek alkaline-ultramafic complex (Fig. 1) con tains a world-class vermiculite deposit formed by hydrother mal alteration of a large pyroxenite intrusion. The deposit is located at Vermiculite Mountain (also called Zonolite Moun tain) approximately six miles northeast of Libby, Montana. The mine began operations circa 1920 and closed in 1990. Recent attention has been given to fibrous and asbestiform amphiboles associated with vermiculite ore produced at Vermiculite Mountain. The amphiboles are suspected to be a causative fac tor in an abnormally high number of cases of respiratory dis eases in the residents of Libby and the former mine and mill workers (Lybarger et al. 2001). The presence of fibrous and asbestiform amphiboles in the vermiculite and mine waste from Vermiculite Mountain has triggered a Superfund action that ranks among the largest and most costly in the history of the U.S. Environmental Protec tion Agency. The ultimate resolution of the problems associ ated with contamination by these materials will be years in coming, and the final costs in both human health and dollars may be enormous. These issues necessitate a very thorough understanding of the morphological and chemical properties * E-mail: gmeeker@usgs.gov of the amphiboles associated with the Vermiculite Mountain deposit. It is these properties that are of ongoing concern with respect to future regulatory policies and investigations into possible mechanisms of toxicity of fibrous and asbestiform amphiboles (Ross 1981; Langer et al. 1991; Kamp et al. 1992; van Oss et al. 1999). Previous studies of the composition and morphology of the amphiboles from Vermiculite Mountain are limited in number. Wylie and Verkouteren (2000) studied two amphibole samples from the vermiculite mine. They determined the amphibole in both samples to be winchite based in part on chemistry, using the classification system of Leake et al. (1997), and on optical properties. Gunter et al. (2003) confirmed the findings of Wylie and Verkouteren (2000) on the same two samples and analyzed three additional ones, which they also determined to be winchite based on optical microscopy, electron probe microanalysis, and Mossbauer spectroscopy. Indeed, the results of the present study demonstrate convincingly that the vast majority of the amphiboles from Vermiculite Mountain are winchite as currently de fined by the International Mineralogical Association (Leake et al. 1997). Previously, the amphibole from Vermiculite Moun tain had been called soda tremolite (Larsen 1942), richterite (Deer et al. 1963), soda-rich tremolite (Boettcher 1966b), and tremolite asbestos and richterite asbestos (Langer et al. 1991; Nolan et al. 1991). 0003-004X/03/1112-1955$05.00 1955 1956 MEEKER ET AL.: THE COMPOSITION OF AMPHIBOLES FROM THE RAINY CREEK COMPLEX phiboles in the context of existing industrial, medical, regu latory, and mineralogical definitions. FIGURE 1. Map of vermiculite mine showing amphibole sampling locations. Geology after Boettcher (1967). The geology, as depicted here, may not completely coincide with the present-day surface geology because of the mining activity between 1967 and 1992. Therefore, the sampling points may not coincide in all cases with the rock units as shown above. The chemical and physical properties of the fibrous amphiboles from Vermiculite Mountain are of significance for two reasons. The first is that most asbestos regulations specifically cite five amphibole asbestos "minerals:" tremolite, actinolite, anthophyllite, amosite, and crocidolite; and one serpentine mineral, chrysotile. These names have evolved from a combi nation of mineralogical and industrial terminology. The min eral names richterite and winchite do not appear in existing regulatory language. It is therefore important to understand fully the range of amphibole compositions present so that appropri ate terminology can be applied to this material. The second, and perhaps more important reason, is that the mechanisms for the initiation of asbestos-related diseases are not fully under stood. If the fibrous and asbestiform amphiboles from Vermiculite Mountain are truly a different type of amphibole than has been studied previously by the medical community, then it is important to understand and describe the full range of chemi cal and physical properties of this material for future toxico logical and epidemiological studies. The current study was designed to provide a systematic evaluation of the Vermiculite Mountain amphiboles and to specifically answer four important questions: (1) are the amphiboles from Vermiculite Mountain relatively uniform in composition or is there a broad range of compositions; (2) what morphologic characteristics are present within the population of Vermiculite Mountain amphiboles; (3) are there any correlations among chemistry, mineralogy, and morphol ogy; and (4) what are the chemical and physical character istics of the fibrous and asbestiform amphiboles that are of respirable size? The answers to these questions are of im portance to the members of the asbestos community who are involved with developing regulatory language, studying the health effects of asbestos, and planning responsible min ing and processing activities. The present study provides a framework with which to evaluate the range of composi tions and morphologies of the Vermiculite Mountain am- Geologic background The Rainy Creek complex (Fig. 1) has been described as the upper portion of a hydrothermally altered alkalic igneous complex composed primarily of magnetite pyroxenite, biotite pyroxenite, and biotitite (Pardee and Larsen 1928; Bassett 1959; Boettcher 1966a, 1966b, 1967). The original ultramafic body is an intrusion into the Precambrian Belt Series of northwest ern Montana (Boettcher 1966b). A syenite body lies southwest of and adjacent to the altered pyroxenite and is associated with numerous syenite dikes that cut the pyroxenites. A small fenite body has been identified to the north, suggesting the presence of a carbonatite at depth (Boettcher 1967). The amount of vermiculite within the deposit varies considerably. At different locations, the vermiculite content of the ore ranges from 30 to 84% (Pardee and Larsen 1928). Subsequent alkaline pegma tite, alkaline granite, and quartz-rich veins cut the pyroxenites, syenite, and adjacent country rock. It is in the veins and wall rock adjacent to these dikes and veins that a significant portion of the fibrous amphiboles occur as a result of hydrothermal processes (Boettcher 1966b). The dikes, veins, and associated wall-rock alteration zones range in width from a few millime ters to meters, and are found throughout the deposit. Fibrous and massive amphiboles are the most abundant alteration and vein-filling products. Estimates of the amphibole con tent in the alteration zones of the deposit range from 50 to 75% (Pardee and Larsen 1928). Accessory alteration miner als include calcite, K-feldspar, talc, vermiculite, titanite, pyrite, limonite (formed by pyrite oxidation), albite, and quartz. In addition, "primary" pyroxene, biotite, and hydrobiotite are present in varying amounts. Methods Sample collection Sampling of the amphibole from Vermiculite Mountain was done in the spring of 2000 with the purpose of collecting a representative suite of amphibole compositions contained within the mined area of the vermiculite deposit. Samples were collected based on a grid designed to provide statistically signifi cant sampling over the accessible areas of the mine. Due to the nature of both the geology of the deposit and the physical conditions in the mine resulting from past reclamation efforts, samples could only be collected from nearly ver tical "cut faces" in the mine. We therefore sampled from the closest vertical cut face to each grid node. A total of 30 locations from the mine area were sampled (Fig. 1). On aver age, samples were approximately 1-2 kilograms in weight. Samples were se lected to provide the maximum variability from location to location in an attempt to fully characterize the range of amphibole compositions and textures present in the deposit. Samples from some locations displayed a massive texture, whereas more friable materials occurred in other locations. In some locations, veins were only a few centimeters in width. At other sampling points, the veins of amphibole-rich rock were as wide as four meters. In these cases, an attempt was made to sample from the edge of the exposed vein as well as the center to look at compositional changes across the vein. In a few cases, veins and adjacent rock appeared to be nearly pure amphibole. Sample preparation All of the samples, whether fibrous and friable or massive, produced ex tremely fine fibrous dust when broken or abraded. The presence of this dust necessitated that all sample preparation steps, including preparation of polished MEEKER ET AL.: THE COMPOSITION OF AMPHIBOLES FROM THE RAINY CREEK COMPLEX 1957 thin sections, be carried out in a negative-pressure, stainless steel, HEPA-filtered hood. Each sample was examined, as collected, in the hood, and represen tative pieces were selected for X-ray diffraction (XRD), electron probe microanalysis (EPMA) using wavelength dispersive spectroscopy (WDS), and scanning electron microscopy combined with energy dispersive X-ray analy sis (SEM/EDS). For each sample location, an effort was made to find pieces that appeared to be representative of the total sample. Samples selected for EPMA were prepared as polished petrographic thin sections, and detailed optical micrographs were made for later reference. In addition, one or more SEM stubs were prepared for each sample by touching a sample stub cov ered with a disk of conductive C tape to the inside of each plastic sample bag. This method allowed us to collect and analyze the friable and fibrous components of each sample so that these portions could be distinguished from the non-friable material. The distribution of amphibole types within the friable material could thus be determined. A portion of a typical SEM mount is shown in Figure 2. Sample analysis In the present study, we used a combination of three analytical tech niques to characterize composition, mineralogy, and morphology of both the fibrous and non-fibrous components of the Vermiculite Mountain amphiboles. None of these analytical techniques alone is capable of accom plishing this task. XRD was used to determine and confirm the presence of amphibole by structural analysis. EPMA/WDS of polished thin-sections was used to derive accurate compositions of the amphiboles present, and SEM/ EDS was used to characterize the morphology and to determine the amphibole mineral distribution among individual small fibers that are of respirable size and are generally too small to mount and polish. The SEM-based EDS analysis of small, unpolished fibers does not have the accuracy to defini tively identify the amphibole types present. However, when combined and correlated with EPMA/WDS analysis for each individual sample the SEM/ EDS analyses show the distributions of the fibrous and asbestiform miner als present in the deposit. FIGURE 2. Area of the surface of a typical SEM sample stub prepared by touching the stub to the inside of the plastic sample bag. Most of the particles in the image are amphibole. Particle morphologies include acicular structures with high to low aspect ratios, bundles, and prismatic crystals. A few curved fibers can be seen in the image. Scale bar is 50 pm. X-ray diffraction analysis Splits of each sample were analyzed by XRD at the USGS analytical labo ratories in Denver. Two grams of material were prepared by hand grinding the sample in an agate mortar and pestle and then wet micronizing (to decrease lattice shear) in a micronizing mill to obtain an average grain size of 5 mi crometers. This procedure was used to minimize the orientation effects of the minerals present. The samples were air dried and packed into an aluminum holder for subsequent mineralogical analysis. The powder XRD data were collected using a Philips APD 3720 automated X-ray diffractometer with spinning sample chamber, a diffracted beam monochromator, and Ni-filtered CuKa radiation at 40 kV and 25 mA. The data were collected at room temperature in scanning mode, with a step of 0.02 o20 and counting time of 1 second at each step. The collected data were evaluated and minerals were identified using JADE+ soft ware from Materials Data Inc.1 Qualitative mineralogy was determined for each sample as major (>25% by weight), minor (5-25%), and trace (<5%). Our detection limit for these analyses was approximately 1-2 wt%. Table 1 shows samples ranging from fairly pure amphibole (samples 25, 28, and 30) to complex mixtures of many minerals (samples 7, 11, and 16). The primary amphibole minerals identi fied in each sample by matching reference X-ray data (JADE+) were winchite and richterite. Other minerals identified as major in some samples included calcite, talc, and dolomite. Minerals present at the minor level in many of the samples include calcite, K-feldspar, pyroxene, hydrobiotite, talc, quartz, vermiculite, and biotite. The arrangement of the amphiboles into subgroups and series based on crys tal-chemical considerations (Leake et al. 1997) is to a large extent a matter of convenience; considerable solid solution exists between one series and another, and even between one subgroup and another. Therefore, it is imperative that the final assignment of a specific amphibole name be based on a high-quality chemi cal analysis of the sample. 1The use of commercial product names in this manuscript is for information only and does not imply endorsement by the United States Government. Table 1. Qualitative mineralogy by XRD SAMPLE MAJOR MINOR TRACE 1 rht/wht, tlc qtz, kfs, vrm 2 rht/wht cal qtz, kfs,dol 3 rht/wht,cal kfs bt 4 rht/wht tlc, aug, hbt cal, dol 5 rht/wht cal, kfs hbt 6 rht/wht, cal qtz, kfs 7 rht/wht cal, aug bt, vrm, kfs 8 rht/wht cal tlc, vrm, kfs, bt 9 rht/wht cal vrm, dol 10 rht/wht cal, vrm 11 rht/wht cal, aug, kfs, tlc qtz, vrm 12 rht/wht kfs 13 rht/wht cal, tlc, di, kfs 14 rht/wht bt, kfs cal, dol 15 rht/wht cal, tlc kfs 16 rht/wht cal, aug, tlc, vrm qtz, kfs 17 rht/wht kfs, cal bt 18 rht/wht cal, kfs bt 19 rht/wht cal, kfs, aug bt 20 rht/wht cal vrm 21 rht/wht cal, tlc, hbt kfs 22 rht/wht cal, hbt, kfs tlc 23 rht/wht cal, kfs 24 rht/wht cal, kfs 25 rht/wht kfs, cal 26 rht/wht cal, kfs, vrm tlc, dol 27 rht/wht, cal kfs 28 rht/wht vrm, hbt 29 rht/wht, cal, dol kfs 30 rht/wht kfs Afofes.'Estimated concentration reported as major (>25 wt%), minor (>5%, <25%), and trace (<5%). Amphibole identification was determined by pat tern structure using a best fit algorithm. Positive identification of amphiboles must rely on chemistry (see text). Mineral abbreviations used: rht/ wht = richterite/winchite, tlc = talc, qtz = quartz, cal = calcite, kfs = potas sium feldspar, vrm = vermiculite, dol = dolomite, bt = biotite, aug = augite, hbt = hydrobiotite, di = diopside. 1958 MEEKER ET AL.: THE COMPOSITION OF AMPHIBOLES FROM THE RAINY CREEK COMPLEX Scanning electron microscopy and energy dispersive Xray analysis Images were obtained of representative areas of each sample stub (Fig. 2). Thirty or more fibers were analyzed in each of the 30 samples. Isolated fibers with diameters of 3 mm and less, representing the respirable fraction, were se lected for analysis so as to minimize contributions of stray X-ray counts from nearby phases both laterally and vertically. One or more of the analyses from each sample set were discarded after later determination that the analysis con tained unacceptable cation ratios, possibly due to contributions from adhering or nearby particles. Scanning electron microscopy was performed using a JEOL 5800LV in strument, at the US Geographical Survey in Denver, operating in high-vacuum mode. Energy dispersive X-ray analysis was performed using an Oxford ISIS EDS system equipped with an ultra-thin-window detector. Analytical condi tions were: 15 kV accelerating voltage, 0.5-3 nA beam current (cup), and ap proximately 30% detector dead time. All SEM samples were C coated. Data reduction was performed using the Oxford ISIS standardless analysis package using the ZAF option. Analyses were normalized to 100%. The quality of each EDS analysis was based on cation ratios and correlation with EPMA/WDS data (see below). The matrix corrections used in these EDS analyses do not account for par ticle geometry. It is well known that such errors can be significant. However, Small and Armstrong (2000) have shown that, at 10-15 kV accelerating volt age, geometry-induced errors on particles can be relatively small. Our errors, in relative weight percent, estimated from analysis of 0.5-10 mm diameter par ticles of USGS, BIR1-G basalt glass reference material (Meeker et al. 1998) are approximately 13% (1s) for Na2O, 4% for MgO and CaO, 3% for AFO3, 2% for SiO2, and 7% for FeO. In addition to chemical EDS data on amphiboles from each sample stub, samples 4, 10, 16, 20, and 30 were selected for morphologic analysis of the amphibole particles. These samples were chosen to provide a representative range of compositions and textures. Size measurements were made using the Oxford ISIS software calibrated with a certified reference grid. For each sample, every amphibole (identified by EDS) was measured within a randomly chosen, 100 x 100 mm area of the stub. The minimum total number of particles counted was 300 per sample. One sample contained fewer than 300 amphiboles in one field of view, so a second field, not overlapping the first, approximately 25 x 25 mm in size was used to complete the data collection, using the same method as above. The maximum length and average width of each amphibole contained within or crossing into the field of view was used to calculate the aspect ratio (length/width) of each amphibole particle. Wavelength-dispersive electron probe microanalysis Electron microprobe analysis was performed on polished thin sections of 14 samples. The samples were selected based on their textural characteristics, mineralogy as determined by XRD and SEM/EDS, optical properties, and how representative the samples appeared to be of the entire suite. An attempt was made to include the full range of chemistries and textures. Quantitative EPMA of the samples was performed using a five-wavelength spectrometer (WDS), fully automated, JEOL 8900 scanning electron microprobe, at the USGS in Denver. Analyses were obtained from areas that appeared to be representative of each sample by optical microscopy. Analytical conditions were: 15 kV accelerating voltage, 20 nA beam current (cup), point beam mode, and 20 second peak and 10 second background counting time. Calibration was per formed using well-characterized silicate and oxide standards. Analytical preci sion for major and minor elements based on replicate analysis of standards was better than 2% relative concentration for major and minor elements and equal to counting statistics for trace (<1 wt%) elements. Matrix corrections were per formed with the JEOL 8900 ZAF software. The friable nature of most of the samples caused some areas of the thin sections to exhibit plucking or poor polishing. Analyses within these areas commonly re sulted in lower totals than would normally be acceptable on a polished surface. We rejected any EPMA analysis with an oxide total lower than 92 wt% (calculated H2O in the Vermiculite Mountain amphiboles ranges from 1.72-2.11 wt%). The quality of the remaining analyses were judged by cation ratios. Analyses with unacceptable cation ratios (see below) were not included in the data reduction. Data analysis The amphibole classification system of Leake et al. (1997) is based on site assignments for each cation in the structure. An accurate amphibole classifica tion based on chemical analysis requires determination of the OH, ultra-light elements (Z < 8), and halogen content, as well as the oxidation state of Fe. Our EPMA analyses of the thin sections included F and Cl. It is not possible to analyze for OH, nor is it possible to accurately determine the ultra-light ele ment content, particularly Li, by EPMA. It is unlikely, however, that Li is present in significant amounts because wet-chemical analyses of Vermiculite Mountain amphibole by previous investigators did not indicate Li (Deer et al. 1963). Also, USGS trace-element analyses of the 30 samples by ICRMS revealed Li (and other possible elemental constituents) at levels too low to be significant in cat ion calculations (P. J. Lamothe, personal communication). Finally, the stoichi ometry that was evident upon data reduction of the EPMA data indicates that no significant components are missing from the analyses. The hydroxyl ion (OH)was accounted for by the method described in Leake et al. (1997) by assuming a total anion charge of -2 for F + Cl + (OH). Analyses were judged primarily on cation ratios for data corrected to 23 O atoms. Cations were assigned to crystallographic sites based on the methods outlined in Leake et al. (1997). In particular, all Si was assigned to the tetrahe dral or T-site, followed by Al and then Ti, until the tetrahedral cation total equaled 8.00. Remaining Al and Ti, followed by Fe3+, Mg, Fe2+, and Mn, in that order, were assigned to the octahedral C-sites (M1, M2, and M3) until the C-site total equaled 5, or slightly less in some cases. Any remaining C-site cations, fol lowed by Ca and Na, were assigned to the B-site (M4) until the site total equaled 2. All K and any remaining Na were assigned to the A-site. Because the Vermiculite Mountain amphiboles only include sodic, sodic-calcic, and calcic amphiboles as defined by Leake et al. (1997), it is primarily the distribution and cation totals of Ca, Na, and K in the B- and A-sites, and Mg/(Mg + Fe2+) that determine the amphibole species. A complete and correct application of the Leake et al. (1997) classification method requires knowledge of the oxidation state of Fe. Gunter et al. (2003), have determined Fe3+/Fetotal in five samples of Vermiculite Mountain amphiboles to range from 0.56 to 0.76 using Mossbauer spectroscopy. Because of the large range of compositions of the amphiboles, we compared the results of cal culating total Fe as Fe2+ vs. total Fe as Fe3+. The difference in the handling of Fe made a small but significant difference in the distribution of the calculated amphibole species. Many analyses showed a change in mineral classification, as seen in Figure 3. The calculated stoichiometry of all EPMA analyses improved when total Fe was calculated as Fe3+. In particular, the average number of Si cations based on 23 O atoms (anion charge = 46.0) decreased from 8.08 0.07 with total Fe calculated as Fe2+ to 7.96 0.06 with total Fe calculated as Fe3+. Because the maximum Si content of the T-site in amphibole must be less than or equal to 8, within analytical error, these results suggest Fe3+> Fe2+, in agree- FIGURE 3. EPMA data from sample 14 plotted with all Fe calculated as Fe2+ and the same analyses plotted with all Fe calculated as Fe3+. The Y-axis represents the amount of Na + K in the A-site of the amphibole structure, and the X-axis the amount of Na in the B-site. The boundary between winchite and richterite, as defined by Leake et al. (1997), is shown as a horizontal line at A(Na+K) = 0.5. Note the approximate 25% decrease in the number of points plotting in the richterite field when all Fe is calculated as Fe+3. MEEKER ET AL.: THE COMPOSITION OF AMPHIBOLES FROM THE RAINY CREEK COMPLEX 1959 ment with the results of Gunter et al. (2003). To arrive at a better estimation of Fe3+/Fetotal for each amphibole mineral, we chose 169 of the best EPMA analy ses, representing a full range of compositions, and calculated Fe3+/Fetotal for each individual analysis. The value for Fe3+/Fetotal was determined by minimizing the deviation from ideal stoichiometry as described in Leake et al. (1997). The average Fe3+/Fetotal calculated from the best 169 EPMA analyses was 0.60, compatible with the values determined by Gunter et al. (2003). This aver age value was used to calculate the mineral distributions for the EDS analyses. This number will be least accurate for compositions close to tremolite and magnesioriebeckite (see below). However, the error introduced by using Fe3+/ Fetotal = 0.60 for all EDS analyses is significantly less than the analytical error for most of the major elements determined by EDS. Amphibole classification derived from EDS results was also based on Leake et al. (1997). In general, the EDS data were very similar to the quantitative WDS results from EPMA. It was found, however, that the C-site totals from the EDS data averaged 3% below the ideal 5 cations. This deficiency could be due to the less-accurate standardless quantification routine, the fact that the analy ses were performed on individual thin fibers rather than a polished surface or, more likely, a combination of both. In the Vermiculite Mountain amphibole, the primary cations in the C-site are Mg and Fe. In the cation site calculations, upon filling the C-site, any remaining C-site cations would be placed into the B-site. Increased residual C-site cations in the B-site would decrease the amount of Na in the B-site and increase the amount of Na in the A-site, thereby affecting the cation distributions and possibly the amphibole species classification. How ever, in our calculations using the more accurate EPMA/WDS data, residual Csite cations in the B-site were generally low or not present. Therefore, low totals in the C-site in the EDS data for these amphiboles should not cause significant errors in amphibole classification. We attribute our low C-site totals in the EDS data to particle geometry and associated matrix correction errors primarily affecting Fe and possibly Mg, and not to actual differences between the friable and non-friable minerals in the Vermiculite Mountain amphibole. Based on our estimated analytical error for Fe and Mg, derived from the analysis of basalt glass particles (see above), and on the overall quality of each EDS analysis, we chose to incorporate EDS data points in which the C-site totals were 4.7 or higher or within 94% of the ideal 5 cations. With this error, the calculated compositions and site assignments on individual EDS analyses did not appear to change significantly or affect the mineral classification relative to the WDS data. A check on the validity of this argument can be seen in the sample-by-sample correlation of compositional distributions showing good agreement between EPMA/WDS and SEM/EDS data (Fig. 4). It is interesting to note that if the error in the C-site totals in the EDS data had been high rather than low, the distribution of amphibole species in the friable materials would likely have been skewed. An error of this type would be difficult to detect without EPMA/WDS data for comparison. FIGURE 4. Cation values for Na in the B-site and Na + K in the Asite from individual samples show typical correlation between SEM (crosses) and EPMA (circles) data. Sample numbers are in the upper left corner of each plot. Results Chemistry In general, the WDS (from EPMA) and the EDS data agree with respect to the amphibole species represented in each sample (Fig. 4). For some samples, the EPMA data show a narrower compositional range than the EDS data. This result is reasonable because EPMA analyses were performed on a single polished thin section for each sample, which may not repre sent the entire range of compositions of friable material found in a sample. The data indicate that most of the Vermiculite Mountain amphiboles can be classified as one of three types, although it is possible that as many as six different amphiboles may be present, based on the Leake et al. (1997) classification criteria. Those minerals, in order of decreasing abundance, are: winchite, richterite, tremolite, and possibly magnesioriebeckite, edenite (see below), and magnesio-arfvedsonite. Representative EPMA analyses of the amphibole minerals are given in Table 2. For the respirable fraction, as determined by SEM/EDS, approxi mately 84% of the amphiboles can be classified as winchite, FIGURE 5. Amphibole compositions from the best 169 EPMA analyses, as determined from cation ratios, based on the criteria of Leake et al. (1997). End-member points for tremolite, winchite, richterite, magnesioriebeckite, and magnesio-afrvedsonite are shown. The data suggest that complete solid-solution may exist within the region defined by the tremolite, winchite, richterite, and magnesioriebeckite. Also shown (inset) are "best-fit" curves for the same data, showing calculated Fe+3/Fetotal (see text) values for individual minerals where T=tremolite, R=richterite, W=winchite (multiplied by 0.25), and M=magnesioriebeckite. 11% as richterite, and 6% as tremolite. Figure 5 shows the distribution of amphibole compositions found at the mine site at Vermiculite Mountain. The amphiboles range from nearly pure tremolite to compositions 1960 MEEKER ET AL.: THE COMPOSITION OF AMPHIBOLES FROM THE RAINY CREEK COMPLEX Table 2. Representative wavelength dispersive of amphibole minerals Sample Mineral Wt% Oxides F Na2O MgO A^O3 SiO2 Cl K2O CaO TiO2 MnO FeO-T O = F,Cl TOTAL 12 W 0.21 3.39 22.3 0.15 58.7 BDL 0.65 7.50 0.14 0.10 5.71 0.09 98.79 12 W 0.27 4.45 19.2 0.17 57.1 BDL 0.71 5.18 0.14 0.05 8.38 0.11 95.54 14 W 0.18 4.21 20.5 0.16 57.2 BDL 1.03 6.28 0.13 0.06 6.35 0.07 96.00 16 W 0.20 3.29 21.4 0.46 57.7 BDL 1.02 9.41 0.07 0.14 4.38 0.08 97.99 17 W 0.58 3.54 19.8 0.37 56.8 BDL 0.94 7.51 0.17 0.14 6.54 0.24 96.21 24 W 0.31 3.13 21.3 0.15 57.5 BDL 0.93 8.43 0.04 0.13 4.95 0.13 96.73 24 W 0.52 4.47 20.9 0.33 57.7 BDL 1.10 6.62 0.23 0.08 5.54 0.22 97.21 25 W 0.43 2.70 21.7 0.11 57.5 BDL 0.62 9.89 0.07 0.10 4.22 0.18 97.12 10 T 0.21 2.61 22.0 0.16 57.1 BDL 0.71 10.2 0.12 0.09 3.08 0.09 96.19 16 T 0.17 2.26 23.0 0.60 56.4 BDL 0.87 10.1 0.10 0.08 2.48 0.07 95.92 16 T 0.20 2.27 22.0 0.52 56.6 BDL 0.78 10.3 0.09 0.08 2.47 0.08 95.19 Structural Formula Si 7.988 Aliv 0.012 Sum T-site 8.000 8.000 0.000 8.000 7.990 0.010 8.000 7.987 0.013 8.000 7.993 0.007 8.000 7.988 0.012 8.000 7.987 0.013 8.000 7.994 0.006 8.000 7.997 0.003 8.000 7.904 0.096 8.000 7.986 0.014 8.000 Aliv Ti Fevi Mg Fe2+ Mn Sum C-site 0.012 0.015 0.340 4.531 0.103 0.000 5.000 0.029 0.015 0.981 3.976 0.000 0.000 5.000 0.016 0.014 0.682 4.264 0.024 0.000 5.000 0.062 0.008 0.052 4.423 0.455 0.001 5.000 0.054 0.018 0.498 4.153 0.271 0.005 5.000 0.012 0.004 0.415 4.421 0.147 0.000 5.000 0.040 0.024 0.524 4.306 0.106 0.000 5.000 0.011 0.007 0.178 4.489 0.313 0.001 5.000 0.023 0.012 0.038 4.599 0.323 0.006 5.000 0.004 0.010 0.024 4.813 0.149 0.000 5.000 0.073 0.010 0.036 4.627 0.254 0.000 5.000 Mg Fe2+ Mn Ca Na Sum B-site 0.000 0.207 0.011 1.093 0.688 2.000 0.031 0.000 0.006 0.777 1.187 2.000 0.000 0.035 0.007 0.939 1.018 2.000 0.000 0.000 0.015 1.396 0.589 2.000 0.000 0.000 0.012 1.131 0.858 2.000 0.000 0.013 0.015 1.256 0.716 2.000 0.000 0.011 0.009 0.982 0.998 2.000 0.000 0.000 0.010 1.473 0.517 2.000 0.000 0.000 0.005 1.536 0.459 2.000 0.000 0.118 0.010 1.511 0.362 2.000 0.000 0.001 0.009 1.552 0.438 2.000 Na K Sum A-site 0.207 0.112 0.319 0.021 0.128 0.148 0.121 0.183 0.303 0.293 0.179 0.473 0.109 0.168 0.276 0.127 0.165 0.293 0.203 0.195 0.398 0.210 0.109 0.320 0.249 0.128 0.377 0.253 0.156 0.409 0.183 0.140 0.323 Total Cations 15.319 15.148 15.303 15.473 15.276 15.293 15.398 15.320 15.377 15.409 15.32 Notes: W = winchite, R = richterite, T = tremolite, MR = magnesioriebeckite, MA = magnesio-arfvedsonite, BDL = below detectability limit. Ferric Fe determined by stoichiometry. continued approaching end-member magnesioriebeckite. The majority of the compositions lie within the ternary field temolite-winchiterichterite, and all compositions lie within the field tremoliterichterite-magnesioriebeckite. The distribution of compositions suggests that complete solid solution exists within the compo sitional field shown. These results are compatible with the study by Melzer et al. (2000), who found evidence for complete solid solution in the experimental system K-richterite-richteritetremolite. Figure 5 also shows the distributions of Fe3+/Fetotal for each amphibole species. These distributions suggest that Fe3+ is parti tioned into each amphibole mineral according to crystal-chemical requirements. The complexities of such substitutions and the dif ficulties in identifying a specific substitution mechanism in amphiboles were discussed by Popp and Bryndzia (1992). Actinolite was not found in our analyses of the Vermiculite Mountain amphiboles. Wylie and Verkouteren (2000) specu lated on the presence of actinolite but were not able to make a determination in their samples because they did not calculate or otherwise determine the Fe3+ content. If our EPMA analyses were calculated with all Fe as Fe2+, some of the analyses would be classified as actinolite, based on Leake et al. (1997). This finding suggests that during routine semi-quantitative analy ses of Vermiculite Mountain amphibole, as might be performed by an environmental asbestos analysis laboratory, the presence of actinolite might be reported. It is also possible that different laboratories could report the presence of different asbestos minerals from the same samples depending on the data reduc tion methods used. Both SEM/EDS single-fiber and EPMA/WDS thin-section data occupy approximately the same compositional space, as shown in Figure 6. A few compositions that correspond to magnesioriebeckite and one to magnesio-arfvedsonite are in dicated from the EPMA data. These amphibole types along with edenite (not identified in the EPMA data) were also found with SEM/EDS analyses. The magnesioriebeckite and magnesioarfvedsonite EDS data points are all within 1s error of richterite and/or winchite. The lack of statistically significant EDS data for magnesioriebeckite and magnesio-arfvedsonite suggests that these minerals may not exist in fibrous form. All of the EDS edenite analyses are within 2s error of being classified as tremolite. All other minerals were identified in both thin sections and in the single fiber data. This comparison indicates that tremolite, winchite, and richterite (and possibly edenite) all occur MEEKER ET AL.: THE COMPOSITION OF AMPHIBOLES FROM THE RAINY CREEK COMPLEX 1961 Table 2. -- continued(2) Sample Mineral Wt% Oxides F Na2O MgO Al2O3 SiO2 Cl K2O CaO TiO2 MnO FeO O=F,Cl TOTAL 16 T 0.18 2.26 22.1 0.71 55.6 0.02 0.86 10.1 0.11 0.07 2.40 0.08 94.32 20 T 0.00 1.29 21.7 0.56 55.2 0.03 0.58 10.70 0.10 0.10 4.00 0.01 94.24 Structural Formula Si 7.918 Aliv 0.082 Sum T-site 8.000 Aliv Ti Fe3+ Mg Fe2+ Mn Sum C-site 0.037 0.012 0.101 4.686 0.163 0.000 5.000 7.911 0.089 8.000 0.006 0.011 0.036 4.629 0.318 0.000 5.000 Mg Fe2+ Mn Ca Na Sum B-site 0.000 0.021 0.008 1.542 0.429 2.000 0.000 0.126 0.012 1.643 0.219 2.000 Na K Sum A-site 0.194 0.155 0.350 0.138 0.106 0.244 Total Cations 15.350 15.244 25 T 0.08 2.62 21.9 0.30 57.3 0.01 0.78 10.2 0.01 0.10 3.73 0.04 96.96 7.977 0.023 8.000 0.026 0.001 0.070 4.547 0.355 0.000 5.000 0.000 0.008 0.012 1.520 0.460 2.000 0.247 0.138 0.385 15.385 25 T 0.47 2.49 21.9 0.32 57.3 BDL 0.75 10.3 0.04 0.10 3.82 0.20 97.21 7.972 0.028 8.000 0.025 0.004 0.073 4.535 0.363 0.000 5.000 0.000 0.009 0.012 1.537 0.442 2.000 0.231 0.134 0.364 15.364 25 T 0.21 2.28 21.9 0.25 57.3 BDL 0.68 10.5 0.06 0.10 3.52 0.09 96.63 7.992 0.008 8.000 0.034 0.006 0.073 4.550 0.337 0.000 5.000 0.000 0.001 0.012 1.562 0.425 2.000 0.192 0.121 0.313 15.313 10 R 0.65 4.13 23.0 0.01 58.1 0.01 1.56 7.79 0.20 0.09 2.35 0.28 97.59 7.999 0.001 8.000 0.001 0.021 0.226 4.727 0.026 0.000 5.000 0.000 0.019 0.010 1.150 0.821 2.000 0.281 0.274 0.554 15.554 12 R 0.08 3.73 21.1 0.12 55.3 BDL 0.77 7.45 0.21 0.05 5.61 0.03 94.33 7.980 0.020 8.000 0.001 0.022 0.004 4.534 0.438 0.000 5.000 0.000 0.234 0.006 1.151 0.608 2.000 0.437 0.141 0.578 15.578 14 R 0.17 4.38 20.3 0.20 55.2 BDL 1.06 6.11 0.12 0.09 6.56 0.07 94.13 7.983 0.017 8.000 0.017 0.013 0.240 4.374 0.355 0.000 5.000 0.000 0.198 0.010 0.946 0.846 2.000 0.383 0.196 0.579 15.579 16 R 0.34 4.96 19.8 0.29 56.2 BDL 0.97 5.76 0.04 0.08 7.49 0.14 95.73 24 R 0.74 3.90 21.2 0.36 56.8 0.01 1.20 8.03 0.07 0.13 5.08 0.31 97.25 30 R 0.52 4.48 21.4 0.27 56.9 BDL 1.22 7.43 0.16 0.12 5.08 0.22 97.34 7.976 0.024 8.000 0.024 0.004 0.485 4.179 0.307 0.000 5.000 0.000 0.097 0.009 0.875 1.018 2.000 0.348 0.176 0.524 15.524 7.967 0.033 8.000 7.973 0.027 8.000 0.026 0.007 0.150 4.433 0.383 0.000 5.000 0.016 0.017 0.094 4.465 0.407 0.000 5.000 0.000 0.062 0.015 1.206 0.716 2.000 0.000 0.094 0.014 1.115 0.777 2.000 0.343 0.215 0.558 0.441 0.218 0.659 15.558 15.659 continued nextpage in fibrous or asbestiform habit in the Vermiculite Mountain rocks, and also that the EPMA data include the majority of the suite of amphibole compositions that are present in the deposit. The EDS single-fiber data provide information on the dis tribution of compositions of the friable and fibrous amphiboles. These analyses are plotted for each sample in Figure 7. For many samples, the compositions cluster in relatively small regions of the diagram as compared to Figure 6. A few samples, such as 8, 16, and 23, show a wider range of compositions. Compositions of several of the samples (5, 7, 9, 13, 21, and 24) cluster entirely within the winchite region of the diagram. Sev eral samples (1, 3, 6, 25, 28, and 29) have a significant amount of richterite, but no samples plot entirely within the richterite field. Samples 8, 20, and 23 show the highest concentrations of tremolite. The classification of a small portion of the Vermiculite Mountain amphibole as edenite (samples 4, 8, and 19) by EDS remains uncertain. A natural occurrence of fibrous fluoroedenite from Sicily was reported by Gianfagna and Oberti (2001). It is likely, however, that in our analyses, microcrystal line calcite, intergrown with the amphibole, could be contrib uting Ca to the totals, thus increasing the amount of Na assigned to the A-site. Nevertheless, some of our SEM/EDS analyses calculate as edenite with no evidence of calcite. However, these analyses are within analytical error of tremolite and richterite. Edenite usually contains Al in the T-site to balance Na in the A-site, which was not found in the Vermiculite Mountain amphibole. The classification scheme of Leake et al. (1997) is not clear with regard to calcic amphiboles of this composition, i.e., amphiboles containing more than 0.5 (Na + K) in the Asite, less than 0.5 Na in the B-site, and more than 7.5 Si in the T-site. Leake (1978) includes the term "silicic-edenite," which would cover the compositions found in the Vermiculite Moun tain amphibole. This name was dropped in the subsequent and final classification system (Leake et al. 1997) and it appears that the intended name for amphiboles of this composition is edenite. Further investigations are underway regarding the pres ence of edenite. Morphology In general, the Vermiculite Mountain amphiboles have two types of occurrence: vein-fillings and replacement of the pri mary pyroxene of the Rainy Creek complex. The textures dis played by the amphibole and associated minerals are indicative of their hydrothermal origin. Traditionally, amphibole asbes tos is thought to occur as a vein-filling mineral formed during 1962 MEEKER ET AL.: THE COMPOSITION OF AMPHIBOLES FROM THE RAINY CREEK COMPLEX hydrothermal alteration in a tensional Table 2 --continued(3) environment (Zoltai 1981) or as a low-temperature alteration product formed in a stress-free environment (Dorling and Zussman 1987). In a substantial portion of our samples, the amphiboles appear to be forming as direct replacements of pyroxene, probably by the infiltration of fluids in microfractures. Examples of these two modes of formation are shown in Figure 8. Figure 8a shows a cross section of a vein filled with sym metrically matching layers of amphibole and other minerals includ ing calcite, K-feldspar, titanite, and pyrite. The amphibole becomes finer- Sample Mineral Wt% Oxides F Na2O MgO Al2O3 SiO2 Cl K2O CaO TiO2 MnO FeO O = F,Cl TOTAL 30 R 0.56 4.59 20.9 0.45 57.3 BDL 1.29 7.31 0.05 0.12 5.29 0.23 97.65 Structural Formula Si 7.979 Aliv 0.021 Sum T-site 8.000 30 R 0.17 4.26 21.0 0.35 56.6 BDL 1.32 7.26 0.04 0.06 5.22 0.07 96.16 7.971 0.029 8.000 10 MR 0.45 7.04 17.1 0.04 56.5 0.01 1.09 2.03 0.47 0.07 11.5 0.19 96.15 7.997 0.003 8.000 10 MR* 0.30 7.11 16.8 0.07 57.1 BDL 0.95 2.07 0.15 0.09 12.3 0.13 96.76 8.006 0.000 8.006 10 MR* 0.27 6.92 16.4 0.08 56.8 BDL 0.96 1.92 0.39 0.05 12.3 0.11 96.06 8.011 0.000 8.011 10 MR* 0.45 6.85 16.8 0.07 56.9 BDL 0.98 2.32 0.44 0.09 11.4 0.19 96.13 8.022 0.000 8.022 10 MR* 0.49 6.98 17.0 0.08 56.9 0.01 0.97 2.07 0.41 0.08 11.3 0.21 96.03 8.012 0.000 8.012 12 MR 0.09 6.51 17.5 0.25 56.4 BDL 0.81 2.20 0.58 0.04 11.0 0.04 95.39 7.980 0.020 8.000 10 MA 0.52 6.76 17.8 0.08 56.8 BDL 1.06 2.70 0.25 0.04 11.0 0.22 96.89 7.993 0.007 8.000 grained toward the vein center but the composition of the winchite amphibole remains fairly constant across the vein. Figure 8b shows a portion of a sample in which the primary pyrox ene augite crystals are being replaced Alvi Ti Fe3+ Mg Fe2+ Mn Sum C-site 0.053 0.005 0.241 4.348 0.354 0.000 5.000 0.029 0.004 0.285 4.402 0.280 0.000 5.000 0.004 0.049 1.097 3.604 0.246 0.000 5.000 0.012 0.016 1.240 3.505 0.200 0.011 4.984 0.013 0.042 1.272 3.456 0.182 0.006 4.972 0.011 0.046 1.177 3.532 0.164 0.011 4.942 0.013 0.043 1.211 3.578 0.115 0.009 4.968 0.020 0.062 1.123 3.690 0.105 0.000 5.000 0.007 0.026 0.955 3.738 0.274 0.000 5.000 by fibrous amphibole winchite and richterite. Figure 8c shows a detailed view of this replacement within a single pyroxene crystal. The long axis of the fibrous amphibole is Mg Fe2+ Mn Ca Na Sum B-site 0.000 0.021 0.014 1.091 0.875 2.000 0.000 0.050 0.007 1.096 0.847 2.000 0.000 0.023 0.009 0.307 1.662 2.000 0.000 0.000 0.000 0.310 1.690 2.000 0.000 0.000 0.000 0.290 1.710 2.000 0.000 0.000 0.000 0.350 1.650 2.000 0.000 0.000 0.000 0.313 1.687 2.000 0.000 0.074 0.005 0.333 1.588 2.000 0.000 0.067 0.005 0.407 1.521 2.000 crystallo-graphically aligned with the original pyroxene crystal. In portions of all of the samples Na K Sum A-site 0.364 0.229 0.593 0.316 0.238 0.554 0.270 0.196 0.466 0.243 0.170 0.414 0.181 0.173 0.354 0.221 0.175 0.396 0.220 0.173 0.393 0.196 0.146 0.341 0.323 0.191 0.513 studied, the amphibole is intergrown with accessory minerals such as calcite, K-feldspar, quartz, and titanite. The accessory minerals range in size Total Cations 15.593 15.554 15.466 15.403 15.336 15.360 15.373 15.341 15.513 * These analyses display T site totals slightly higher than what is recommended by Leake et al. (1997) for determination of percent Fe+3, however, the T site error is well below 1% and Fe+3 values are in agreement with other analyses of similar composition. from millimeters to sub-micrometer. Extremely fine-grained crystals of these minerals are commonly intergrown and often crystallographically oriented with the am- phibole (Fig. 9). These minerals were found in thin section as well as in the SEM samples of friable dust, often in acicular form. The Vermiculite Mountain amphiboles show a range of morphologies from prismatic to asbestiform (Fig. 10). Much of the fibrous amphibole seen in the SEM micrographs (Figs. 2 and 10) is composed of acicular and, some cases, needle-like particles. Splayed ends and curved fibers are present, but are not particularly common. Fibril diameter in the Vermiculite Mountain asbestiform amphibole ranges from approximately 0.1 to 1 mm. Individual fibrils less than 0.2 mm in diameter are rare, and fiber bundles are often composed of different-sized fibrils. Many of the characteristics generally associated with "commercial-grade" asbestos, such as curved fibers and bundles with splayed ends (Perkins and Harvey 1993) are present but are not common in the Vermiculite Mountain amphibole.2 The material, however, is very friable and even gentle handling of what appears to be a solid, coherent rock can liberate very large FIGURE 6. EPMA/WDS and SEM/EDS data showing the entire range of amphibole species found from all 30 samples. See text for details. numbers of extremely fine fibers as seen in SEM images (Figs. 2 and 10) and in size-distribution plots of material sampled from the inside of the sample bags (Fig. 11). 2The definition of asbestiform found in Perkins and Harvey (1993) is for optical identification of commercial-grade asbes tos used in building materials. MEEKER ET AL.: THE COMPOSITION OF AMPHIBOLES FROM THE RAINY CREEK COMPLEX 1.0 1963 0.5 0.0 1.0 0.5 0.0 1.0 0.5 Na+K (A) 0.0 1.0 0.5 0.0 1.0 0.5 0.0 1.0 0.5 0.0 0.5 1.5 0.5 1.5 0.5 1.5 0.5 1.5 0.5 1.5 Na (B) FIGURE 7. EDS data for 30 samples showing the distribution of compositions of the fibrous and friable amphibole for each sample location at the mine (see Fig. 1). Sample number is in the top left corner of each plot (see Fig. 1). Mineral fields are the same as shown in Figure 6. The data shown in Figure 11 are plotted as diameter vs. length and diameter vs. aspect ratio, respectively. These data, which were obtained from samples 4, 10, 16, 20, and 30, repre sent the range of amphibole compositions sampled. For the most part, all of the samples produce fibers in a similar size range. It is important to remember that these samples were not ground to produce these particles. The fibers were collected on the SEM stubs by touching the stub to the inside of the original sample bag after it was received from the field and other sample mate rial was removed. Approximately 40% of the particles are greater than 5 mm in length and have aspect ratios greater than 3. This finding means that, based on size, these particles are countable as asbestos by most approved methods such as Crane (1992). Even if more conservative counting criteria are employed, such as <0.5 mm diameter with aspect ratios of >10, approximately 30% of the particles would be included. These observations dem- 1964 MEEKER ET AL.: THE COMPOSITION OF AMPHIBOLES FROM THE RAINY CREEK COMPLEX FIGURE 9. Back-scattered electron image of an area of a thin section of sample 24 showing massive and fibrous amphibole (Amp) intergrown with secondary calcite (Cal), titanite (Ttn), and quartz (Qtz). Note the fibrous amphibole enclosed by the large titanite grain at lower right, indicating order of crystallization. FIGURE 8. Transmitted-light images of entire polished thin sections showing: (a) amphibole filling a vein with symmetric dark and light (center of the vein) layers and (b) amphibole (dark areas) replacing pyroxene crystals. (c) A large single pyroxene crystal (bright areas) partly replaced by amphibole (dark areas) along crystallographically oriented planes is shown in transmitted, cross polarized light. onstrate that the Vermiculite Mountain amphiboles, with minimal disturbance, can easily degrade into highly acicular particles that are less than 3 mm in diameter and are therefore respirable (Na tional Academy of Sciences 1984). Discussion The amphibole samples analyzed in this study show a large range in chemical composition. This range is consistent with varying degrees of, and possibly different episodes of, alter ation of the original pyroxenite body by hydrothermal fluids associated with the intrusion of syenite and related rocks. The variations in composition seen in the EDS data in Figure 7 do not appear to correlate directly with sample location. Samples 5, 6, and 7 were collected in close proximity to each other. Samples 5 and 7 show a similar compositional distribution, but the compositions in Sample 6 are distinctly different. Sample pairs 4 and 28, and 27 and 29 were collected from locations that are relatively close to each other and well within the biotite pyroxenite. Both of these show distinctly different amphibole compositions within each pair. These data suggest that the compositional differences are not due to location or gross zoning within the intrusion. The variations are more likely due to the reaction of pyroxene with different compositions of hy drothermal fluids associated with the quartz-rich veins and the trachyte, phonolite, and syenite dikes described by Boettcher (1966b, 1967). The variations also could be due to differences in the duration of fluid-rock interaction. In addition to compositional variations among samples, EPMA data show compositional variations on the micrometer scale. Several samples showed changes in the amphibole min eral within single grains or fiber structures. Figure 12a shows a non-fibrous amphibole crystal with concentric zoning from magnesioriebeckite in the core to winchite at the rim. Figure 12b shows a single amphibole grain with compositions rang ing from tremolite to winchite. The variability of compositions on the micrometer scale can produce single fibrous particles that can have different amphibole names at different points of the particle. This type of varia tion has implications for the regulatory community. Morphologically, such structures might be considered fibers by most analytical protocols (Crane 1992, 1997; Baron 1994). However, by some current regulations and approved analytical methods, the variable chemistry of these particles could ex clude them from being classified as "asbestos." This complexity creates a dilemma for the analyst who is charged with determining MEEKER ET AL.: THE COMPOSITION OF AMPHIBOLES FROM THE RAINY CREEK COMPLEX 1965 FIGURE 10. Electron micrographs of typical morphological types of Vermiculite Mountain amphiboles. The morphologies range from prismatic crystals (upper left) to long fibers and bundles (lower right). 1966 MEEKER ET AL.: THE COMPOSITION OF AMPHIBOLES FROM THE RAINY CREEK COMPLEX FIGURE 11. Amphibole particle size data from samples 4, 10, 16, 20, and 30 for particle diameters 5 mm and less, plotted as length vs. diameter (a) and aspect ratio (length:diameter) versus diameter (b). whether asbestos is present in a sample and at what level. A further dilemma arises from the fact that none of the present regulatory analytical methods (with the possible ex ception of well-calibrated SEM/EDS analysis using calibra tion standards similar to EPMA/WDS) can accurately differentiate the amphiboles present in the asbestiform materi als from Vermiculite Mountain. Even with standard optical tech niques, the results can be ambiguous (Wylie and Verkouteren 2000). This ambiguity arises because the mineralogical com munity currently classifies amphiboles on the basis of crystal chemistry, and high precision and accuracy in the microanalytical technique employed are required to classify an amphibole accurately. Analytical electron microscopy (TEM/EDS) provides compositional information, but the thickness of the sample must be known to provide accurate chemistry. This in formation is normally not available during routine TEM analy sis of asbestos fibers as would be performed when following approved asbestos analysis methods such as ISO 10312 (1995). The problem of classification is complicated further when the oxidation state of Fe is considered. This complication is illustrated in Figure 3, where the amphibole-species distribu tion is seen to shift significantly when the analyses are calcu lated using pure Fe+2 and Fe+3 end-members. The degree of accuracy and precision required to determine the correct oxi dation state of Fe is not achievable during routine microanaly sis of small, unpolished, single structures by SEM/EDS or TEM/ EDS. Therefore, any regulatory distinction between minerals that requires knowledge of the oxidation state of Fe, such as the distinction between tremolite and actinolite, is technically not possible without a full quantitative chemical analysis. Our analysis of unpolished, micrometer-sized particles of a basalt glass standard by SEM/EDS resulted in the 2s errors as high as 25% relative for Na and 14% for Fe. Without the ability to correlate unpolished single-fiber SEM/EDS analyses with EPMA data from polished samples, it would be extremely difficult to confirm the presence of any of the amphibole min- FIGURE 12. (a) Back-scattered electron image showing a prismatic amphibole grain with a rim of winchite (point 1) and a core of magnesioriebeckite (point 2), partially surrounded by fibrous amphibole. Other grains of similar composition can be seen above and below. (b) Backscattered electron image showing a large single amphibole structure (center) exhibiting fibrous habit at the ends and along the margins. Point 1 is tremolite and point 2 is winchite. MEEKER ET AL.: THE COMPOSITION OF AMPHIBOLES FROM THE RAINY CREEK COMPLEX 1967 erals identified in this study by EDS alone. We therefore rec ommend that the International Mineralogical Association clas sification system (Leake et al. 1997) for amphiboles not be used for regulatory purposes in cases where high analytical precision and accuracy cannot be demonstrated. If a microanalytical technique does not have the precision and accuracy to classify amphibole asbestos correctly accord ing to current mineralogical criteria, then how should asbes tos, such as that found at Vermiculite Mountain, be classified? Within much of the existing asbestos literature, mineral names are not applied in a uniform manner and are not all consistent with presently accepted mineralogical nomenclature and defi nitions. Tremolite and actinolite (members of the solid-solu tion series tremolite-ferroactinolite), and anthophyllite are mineral names recognized by the Subcommittee on Amphiboles of the International Mineralogical Association (Leake et al. 1997). For these three amphibole species, the term "asbestiform" usually must precede the mineral name, or the term asbestos must be added after the mineral name to denote a regulated material. The name amosite, derived from an acro nym for Asbestos Mines of South Africa, is generally consid ered to refer to the asbestiform varieties of minerals in the cummingtonite-grunerite solid-solution series (Rabbitt 1948; Vermas 1952; Bowles 1959). Crocidolite is the asbestiform variety of the amphibole riebeckite. This inconsistency in the application of nomenclature can cause significant problems for asbestos analysts, medical professionals, and regulators who are unfamiliar with the principles of mineralogic classification, including solid-solution. In addition to the five amphibole as bestos "minerals" normally cited in the regulatory literature, many other amphibole minerals have been reported to occur in asbestiform and fibrous habit (Zoltai 1981; Wylie and Huggins 1980). A few methods and regulations (e.g., ISO 10312, method for TEM analysis of asbestos) recognize the possible existence of other asbestiform amphiboles, but make no attempt to iden tify or define them mineralogically. To complicate further the problems in nomenclature cited above, the nuances of mineralogical classification systems are often not specified or are not well defined in the regulatory literature for many potentially fibrous and asbestiform amphiboles (Lowers and Meeker 2002). In many cases, nominal com positions are given for a mineral but no chemical boundaries are specified. Furthermore, the techniques and methods avail able and approved for the analysis and classification of asbes tos by regulatory entities are often not capable of adequately identifying or distinguishing many of these minerals accord ing to current mineralogical guidelines (such as Leake et al. 1997). This problem is particularly true for microanalytical tech niques such as TEM and SEM employing qualitative or semiquantitative EDS. By virtue of the age of regulatory documents, the current regulatory language (i.e., Bridbord 1976; OSHA 1992) omits richterite and winchite. A better alternative for regulatory no menclature, consistent with modern mineralogical terminology and analytical capabilities, would be to replace the names of the five amphiboles, tremolite asbestos, actinolite asbestos, crocidolite, amosite, and anthophyllite asbestos by the term "asbestiform amphibole" as suggested by Wylie and Verkouteren (2000) or by "fibrous amphibole," if such a de scription is deemed necessary by the medical and health sci ence community. Barring any such changes in the current regulatory language, the Vermiculite Mountain amphibole as bestos could, for the purposes of regulation only, be consid ered equivalent to tremolite or soda-tremolite asbestos in accordance with current and past industrial terminology for the Vermiculite Mountain amphiboles. In addition to chemistry, morphology is a primary factor in evaluation of the asbestiform and fibrous amphiboles. Nomen clature is again a key issue in a discussion of morphological characteristics of amphiboles, particularly those from Vermiculite Mountain. Amphiboles can occur in fibrous and non-fibrous forms. Fibrous amphiboles can further be classified as asbestiform and non-asbestiform. The term asbestiform is usu ally applied to populations of single-crystal fibrils (the small est structural unit of a fiber), which occur in bundles and possess certain characteristics including high aspect ratio, high tensile strength, and flexibility (Zoltai 1981; Perkins and Harvey 1993; Wylie 2000). Another class of amphibole particles, cleavage fragments, can exist in blocky or acicular habit. Regardless of aspect ratio, cleavage fragments are formed by the breaking of a larger crystal. Interestingly, Ahn and Bueck (1991) have de scribed asbestiform riebeckite from Western Australia that ap pears to have formed by the separating or breaking of larger crystals on dislocation planes of weakness along (100) and (110). A similar formation mechanism was proposed by Veblen (1980) for a sample of asbestiform anthophyllite. These find ings obscure somewhat the traditional definition of asbestiform. The Vermiculite Mountain amphiboles serve to underscore the fact that traditional morphological definitions of asbestos may not adequately define amphibole mineral fibers from a toxico logical and regulatory perspective. Within the asbestiform amphibole minerals median diam eters vary. Veblen and Wylie (1993) presented data suggesting that tremolite asbestos and anthophyllite asbestos fibers have larger diameters (median about 0.45 mm), and riebeckite as bestos fibers have smaller diameters (median about 0.2 mm). Byssolite is a term that is sometimes applied to single acicular amphibole crystals with an average diameter of about 1-2 mm (Veblen and Wylie 1993) or "often wider than 1 mm" (Wylie 1979). Our data show the median diameter of the respirable fibrous component (less than 3 mm in diameter) of five Vermiculite Mountain amphibole samples to be 0.44 mm. The av erage diameter for the same set of particles is 0.56 0.45 mm (1s). From these data, the diameter of the Vermiculite Moun tain amphiboles appears to be at the upper range for asbestos and overlaps with the size range cited for byssolite. Cleavage fragments were specifically excluded from mate rial regulated by OSHA in 1992 (OSHA 1992). Therefore, re gardless of any mineralogical, physical, or toxicological differences that might exist among acicular cleavage fragments, byssolite, and asbestiform fibers, differentiation among these classes of particles has become an issue. With the amphiboles, the morphologic distinction between asbestiform fibers and cleavage fragments can be made readily in many cases. This distinction is particularly true when "high-grade" asbestos of commercial value is being compared to blocky cleavage frag 1968 MEEKER ET AL.: THE COMPOSITION OF AMPHIBOLES FROM THE RAINY CREEK COMPLEX ments generated by grinding an amphibole such as massive tremolite. The distinction is not as clear when non-commer cial-grade fibrous amphiboles, like those from Vermiculite Mountain, are being evaluated. For example, by some defini tions (e.g., Perkins and Harvey 1993; Wylie 2000), a popula tion of "true" asbestos fibers should have a minimum mean aspect ratio of 20 for individual fibers longer than 5 mm. In our size analysis of five of the Vermiculite Mountain samples plot ted in Figure 11, three samples had mean aspect ratios slightly higher than 20 whereas two had mean aspect ratios slightly lower. The mean aspect ratio for all five samples, for fibers longer than 5 mm, was only 22. The task of distinguishing be tween what traditionally has been considered asbestos from byssolite and cleavage fragments can become much more diffi cult if not impossible when only single amphibole particles are being evaluated, and a representative population of the amphibole material is not present. Such a situation can be encountered in the analysis of environmental samples of air, soil, or water. The Vermiculite Mountain amphiboles display characteris tics that include all of the above morphological classes in a continuum, from blocky crystals to acicular, non-flexible cleav age fragments, to extremely long flexible fiber bundles (Fig. 10). Most of the individual particles display features that are intermediate between cleavage fragments and long flexible fi bers. There are no distinct morphological boundaries by which to categorize the amphiboles. In addition, the mineralogy of these amphiboles is not typical of most regulated asbestos. Given the variations and ambiguities in much of the morpho logical and mineralogical terminology expressed in the mineralogical, medical, industrial, and regulatory literature (Lowers and Meeker 2002), the Vermiculite Mountain amphiboles present a significant challenge to the analyst, to anyone attempt ing to classify the material with respect to existing definitions, and particularly to those attempting to extrapolate those mor phological features and chemical compositions to potential toxi cological properties. Acknowledgments This paper has been greatly improved by review comments and suggestions from David Jenkins, Malcolm Ross, and Jill Pasteris. The authors thank Brad ley VanGosen, Geoffrey Plumlee, and Douglas Stoeser for extremely helpful discussions and internal USGS review. We also thank Paul Lamothe and Monique Adams who contributed trace element analyses, and Roger Clark, Gregg Swayze, Chris Weis, Paul Peronard, Robert Horton, George Desborough, Raymond Kokaly, Joseph Taggart, Ben Leonard, and Mickey Gunter for their helpful dis cussions and input. This work was funded by the U.S. Geological Survey and the U.S. Environmental Protection Agency. References cited Ahn, J.H. and Bueck, P.R. (1991) Microstructures and fiber-formation mechanisms of crocidolite asbestos. American Mineralogist, 76, 1467-1478. Baron, P.A. (1994) Asbestos by TEM, 4th ed. NIOSH Method 7402. Bassett, W.A. (1959) The origin of the vermiculite deposit at Libby, Montana. Ameri can Mineralogist, 44, 282-299. Boettcher, A.L. (1966a) Vermiculite, hydrobiotite, and biotite in the Rainy Creek Igneous complex near Libby, Montana. Clay Minerals, 6, 283-297. ----------- (1966b), The Rainy Creek igneous complex near Libby, Montana, 155 p. PhD thesis, The Pennsylvania State University, University Park. ----------- (1967) The Rainy Creek alkaline-ultramafic igneous complex near Libby, Montana, part I: Ultramafic rocks and fenite. Journal of Geology, 75, 526-553. Bowles, O. (1959) Asbestos, a materials survey. U.S. Bureau of Mines Information Circular 7880, 94 p. Bridbord, K. et al. (1976) Revised recommended asbestos standard. DHEW (NIOSH) Publication No. 77-169, 96 p. Crane, D.T. (1992) Polarized light microscopy of asbestos. OSHA, Method No. ID- 191, 20p; available on the worldwide web at http://www.osha.gov/dts/sltc/methods/inorganic/id191/id191.html ----------- (1997) Asbestos in air. OSHA, Method No. ID-160, 20p; available on the worldwide web at http://www.osha-slc.gov/dts/sltc/methods/inorganic/ id160/ id160.html Deer, W.A., Howie, R.A., and Zussman, J. (1963) Rock forming minerals, volume 2 chain silicates, 379 p. Longmans, Green and Co. Ltd, London. Dorling, M. and Zussman, J. (1987) Characteristics of asbestiform and nonasbestiform calcic amphiboles. Lithos, 20, 469-489. Gianfagna, A. and Oberti, R. (2001) Flouro-edenite from Biancavilla (Catania, Sic ily, Italy): Crystal chemistry of a new amphibole end-member. American Min eralogist, 86, 1489-1493. Gunter, M.E., DyarD.M., Twamley B., Foit, F.F. Jr., and Cornelius, S. (2003) Com position, Fe3+/Fe, and crystal structure of non-asbestiform and asbestiform amphiboles from Libby, Montana, U.S.A. American Mineralogist, This Volume. ISO (1995) Ambient air determination of asbestos fibres, direct-transfer transmis sion electron microscopy method. ISO 10312:1995(E), 51p. Kamp, D.W., Graceffa, P., Pryor, W.A., and Weitzman, S.A. (1992) The role of free radicals in asbestos-induced diseases. Free Radical Biology & Medicine, 12 293-315. Langer, A.M., Nolan, R.P., and Addison, J. (1991) Distinguishing between amphibole asbestos fibers and elongate cleavage fragments of their non-asbestos ana logues. In R.C. Brown et al. Eds., Mechanisms in Fibre Carcinogenesis, p. 253 267. Plenum Press, New York. Larsen, E.S. (1942) Alkalic rocks of Iron Hill, Gunnison County, Colorado. U.S. Geological Survey Professional Paper 197-A, 64p. Leake, B.E. (1978) Nomenclature of amphiboles. Mineralogical Magazine, 42, 533 563. Leake, B.E., Woolley, A.R., Arps, C.E.S., Birch, W.D., Gilbert, M.C., Grice, J.D., Hawthorne, F.C., Kato, A., Kisch, H.J., Krivovichev, V.G., Linthout, K., Laird, J., Mandarino, J.A., Maresch, W.V, Nickel, E.H., Rock, N.M.S., Schumacher, J.C., Smith, D.C., Stephenson, N.C.N., Ungaretti, L., Whittaker, E.J.W., and Youzhi, G. (1997) Nomenclature of the amphiboles: Report of the subcommit tee on amphiboles of the International Mineralogical Association, Commission on New Minerals and Mineral Names. American Mineralogist, 82, 1019-1037. Lowers, H.A. and Meeker, G.P. (2002) Tabulation of asbestos-related terminology. U.S. Geological Survey Open-File Report 02-458; available on the worldwide web at http://pubs.usgs.gov/of/2002/ofr-02-458/index.html Lybarger, J.A., Lewin, M., Peipins, L.A., Campolucci, S.S., Kess, S.E., Miller, A., Spence, M., Black, B., and Weis, C. (2001) Medical testing of individuals po tentially exposed to asbestiform minerals associated with vermiculite in Libby, Montana, A report to the community [report dated August, 23, 2001]. Agency for Toxic Substances and Disease Registry; available on the worldwide web at http://www.atsdr.cdc.gov/asbestos/doc_phl_testreport.html Meeker, G.P., Taggart, J.E., and Wilson, S.A. (1998) A basalt glass standard for multiple microanalytical techniques. Microscopy and Microanalysis, 4, 240 241. Melzer, S., Gottschalk, M., Andrut, M., and Heinrich, M. (2000) Crystal chemistry of K-richterite-richterite-tremolite solid solutions: a SEM, EMP, XRD, HRTEM and IR study. European Journal of Mineralogy, 12, 273-291. National Academy of Sciences (1984) Asbestiform fibers: nonoccupational health risks, p. 25-47. Committee on Nonoccupational Health Risks of Asbestiform Fibers, Board on Toxicology and Environmental Health Hazards, National Re search Council. National Academy Press, Washington, D.C. Nolan, R.P., Langer, A.M., Oechsle, G.E., Addison, J., and Colflesh, D.E. (1991) Association of tremolite habit with biological potential. In R.C. Brown, Ed., Mechanisms in Fibre Carcinogenesis, p. 231-251. Plenum Press, New York. OSHA (1992) 29 CFR Parts 1910 and 1926 [Docket No. H-033-d], Occupational exposure to asbestos, tremolite, anthophyllite and actinolite. Federal Register, 57, 24310-24331; available on the worldwide web at http://www.osha.gov/pls/ oshaweb/owasrch.search_form?p_doc_type=PREAMBLES&p_toc_level= 1&p_keyvalue=Asbestos~(1992--Original)&p_text_version=FALSE Pardee, J.T. and Larsen, E.S. (1928) Deposits of vermiculite and other minerals in the Rainy Creek District, near Libby, Mont. In Contributions to economic geol ogy, U.S. Geological Survey Bulletin 805, 17-29. Perkins, R.L. and Harvey, B.W. (1993) Test Method for the determination of asbes tos in bulk building material. U.S. Environmental ProtectionAgency, EPA/600/ r-93-116. Popp, R.K. and Bryndzia, L.T. (1992) Statistical analysis of Fe3+, Ti and OH in kaersutite from alkalic igneous rocks and mafic mantle xenoliths. American Mineralogist, 77, 1250-1257. Rabbitt, J.C. (1948) A new study of the anthophyllite series. American Mineralo gist, 33, 263-323. Ross, M. (1981) The geologic occurrences and health hazards of amphibole and serpentine asbestos. In D.R. Veblen, Ed., Amphiboles and other hydrous pyriboles, p. 279-323. Reviews in Mineralogy, Mineralogical Society of America, Washington, D.C. Small, J. and Armstrong, J.T. (2000) Improving the analytical accuracy in the analy sis ofparticles by employing low voltage analysis. Microscopy and Microanaly sis, 6, 924-925. MEEKER ET AL.: THE COMPOSITION OF AMPHIBOLES FROM THE RAINY CREEK COMPLEX 1969 van Oss, C.J., Naim, J.O., Costanzo, P.M., Giese, R.F., Jr., Wu, W., and Sorling, A.F. (1999) Impact of different asbestos species and other mineral particles on pul monary pathogenesis. Clay and Clay Minerals, 47, 697-707. Veblen, D.R. (1980) Anthophyllite asbestos microstructures, intergrown silicates, and mechanisms of fiber formation. American Mineralogist, 65, 1275. Veblen, D.R. and Wylie, A.G. (1993) Mineralogy of amphiboles and 1:1 layer sili cates. In G.D. Gutheri, Jr. and B.T. Mossman, Eds., Health effects of mineral dusts, p.61-137. Reviews in Mineralogy, 28, Mineralogy Society of America, Washington, D.C. Vermas, F.H.S. (1952) The amphibole asbestos of South Africa. Transactions and Proceedings of the Geological Society of South Africa, 55, 199-232. Wylie, A.G. (1979) Optical properties of the fibrous amphiboles. Annals of the New York Academy of Sciences, 330, 611-619. ----------- (2000) The habit of asbestiform amphiboles-implications for the analysis of bulk samples. In M.E. Beard and H.L. Rooks, Eds., Advances in environmental measurement methods for asbestos. American Society for Testing of Materials Special Technical Publication 1342, 53-69. Wylie, A.G. and Huggins, C.W. (1980) Characteristics of a potassian winchite-asbestos from theAllamoore talc-district, Texas. Canadian Mineralogist, 18, 101 107. Wylie, A.G. and Verkouteren, J.R. (2000) Amphibole asbestos from Libby, Mon tana, aspects of nomenclature. American Mineralogist, 85, 1540-1542. Zoltai, T. (1981) Amphibole asbestos mineralogy. In D.R. Veblen, Ed., Amphiboles and other hydrous pyriboles, p. 237-278. Reviews in Mineralogy, Mineralogical Society of America, Washington, D.C. Manuscript received March 7, 2003 Manuscript accepted July 13, 2003 Manuscript handled by Jill Pasteris