Document kD34NNN9gvd6NpoD83R8GZB60

1 APPENDIX C. CHARACTERIZATION OF AMPHIBOLE FIBERS FROM ORE 2 ORIGINATING FROM LIBBY, MT; LOUISA COUNTY, VA; AND PALABORA, 3 REPUBLIC OF SOUTH AFRICA By David L. Berry, PhD U.S. EPA Region 8 Ecosystems Protection and Remediation 1595 Wynkoop Street [8EPR-PS] Denver, CO 80202-1129 . This document is a draftfor reviewpurposes only and does not constitute Agencypolicy. C'l DRAFT--DO NOT CITE OR QUOTE CT\ m CO 1 The O.M. Scott plant in Marysville, OH manufactured a number of products including 2 fertilizers, dyes, and pesticides that were bound to a vermiculite carrier as a delivery vehicle. 3 The plant received ore from Enoree, SC; Louisa County, VA; Libby, MT; and Palabora, 4 Republic of South Africa, which was processed in an exfoliation furnace to produce vermiculite 5 used in the manufacture of their commercial products. Only ore from South Carolina was used 6 in 1957 and ] 958. From 1959 to 1971, ores from South Carolina and Libby, MT were used. 7 From 1972 to 1980, ores from Libby, MT, South Africa, and Virginia were used. No ore from 8 Libby, MT was used after 1980. Only ore from South Africa and Virginia was used after 1980 9 (see Appendix F). 10 The U.S. Environmental Protection Agency (EPA) Region 8 obtained samples of ore 11 from Libby, MT, South Africa, and Virginia from Dr. James Lockey, University of Cincinnati, 12 and analyzed the samples to determine mineralogy and particle size distribution (length, width, 13 and aspect ratio) using transmission electron microscopy (TEM) and energy dispersive 14 spectroscopy (EDS) to identify the nature of the amphibole fibers. Dr. Lockey obtained the 15 South African and Virginia ore samples from the Marysville, OH facility in 1980 and the Libby, 16 MT ore (Libby #3 ore) from an expansion plant in Salt Lake City, UT, in 1981. Region 8 was 17 unable to obtain vermiculite or ore from the Enoree, SC mine complex. 18 The ore from the Rainey Creek complex (Vermiculite Mountain Mine, Libby, MT) 19 resides in large ultramafic intrusive bodies that are rich in biotite, pyroxenite, and biotitite, a rock 20 comprised of almost pure biotite. The ultramafic intrusions are cut by deposits of syenite and 21 carbonatite, and much of the biotite has been hydrothermally altered to hydrobiotite and 22 venniculite (Meeker et al., 2003; Frank and Edmund, 2001). The pyroxenite has been altered to 23 fibrous soda-rich amphiboles, and contacts with pyroxenite surrounding the biotitite contain the 24 vermiculite ore zone containing diopside, hydrobiotite, and apatite. Fibrous and nonfibrous 25 amphiboles are located in both veins and disseminated throughout the intrusive rock along 26 cleavage planes of pyroxene. Amphiboles from Vermiculite Mountain had been referred to as 27 soda tremolite, richterite, soda-rich tremolite, tremolite asbestos, and richterite asbestos by a 28 number of investigators. In 2000, Wylie and Verkouteren (20001 identified winchite as the 29 principal amphibole in the Venniculite Mountain deposit based on chemical investigation 30 referencing the classification system of Leake et al. (1997) and optical properties. Meeker et al. 31 (2003) investigated amphibole types from the mine complex using electron probe microanalysis This document is a draftfor review purposes only and does not constitute Agency policy. C-2 DRAFT--DO NOT CITE OR QUOTE 1 and X-ray diffraction analysis and reported the presence of winchite, richterite, tremolite, and 2 magnesioriebeckite. Magnesio-arfvedsonite and edenite were detected in low abundance. The 3 amphibole composition of the Libby Amphiboles is roughly winchite, richterite, tremolite, 4 magnesio-riebeckite, magnesio-arfvedsonite, and edenite (84:11:6:<1 :<1 :<1). The O.M. Scott 5 facility received ore from the Vermiculite Mountain mine complex, Libby, MT from 1959 6 through 1980. 7 The Palabora Igneous Complex, located near Phalaborwa, Republic of South Africa, is 8 the location of the Palabora mine. The Palabora ore deposit shares many features with the 9 Vermiculite Mountain mine complex--including zoned deposits with ultramafic rocks 10 (pyroxenite) and intrusion by alkalic rock, primarily syenite. The primary mica at Palabora is 11 phlogopite rather than biotite, and the primary alteration product that forms vermiculite ore is 12 hydrophlogopite rather than hydrobiotite (Schoeman, 1989). 13 The Palabora ore is reported to contain little or no asbestifonn fibers based on polarized 14 light microscopy by the Institute of Occupational Medicine in Edinburgh (TOM Consulting, 15 2008). Crude vermiculite from the Palabora complex was also reported to be free of asbestiform 16 fibers by polarized light microscopy (IOM Consulting. 2008). In both reports, the analysis by 17 polarized light microscopy was conducted with a detection limit of 1 ppm, and, since no 18 chrysotile or amphibole structures were detected, no further analysis by electron microscopy and 19 X-ray diffraction were conducted. 20 The ore from the Virginia Vermiculite mine in Louisa County, VA is described as mafic 21 rock intruded by a series of small pegmatites (Gooch. 19571. Meisinger (1979) classified the 22 deposits as Type 3, similar to the ores from Enoree, SC. The formations consist of potassic 23 ultramafic bodies, primarily biotite. The vermiculite ores are found primarily in hydrobiotite 24 portions of the biotite intrusions. The hydrobiotite deposits are preferentially mined because of 25 better commercial properties compared to vemriculite. 26 There is limited information on the asbestos content of the ores from the Louisa County 27 deposit. Rohl and Langer (1977) reported both chrysotile and amphibole fibers in six ore 28 samples from the Louisa County deposit. The chrysotile was reported as fibers and bundles 29 while the amphiboles fibers were classified as actinolite. Moatamed et al. (1986) analyzed a 30 Virginia ore sample collected at a processing plant in Salt Lake City, UT and reported traces of 31 fibrous amphibole asbestos identified as actionlite in the form of cleavage fragments having low This document is a draftfor reviewpurposes only and'does not constitute Agencypolicy. C-3 DRAFT--DO NOT CITE OR QUOTE 1 aspect ratios. Amphibole content for both unexfoliated and exfoliated ores ranged up to 1.3% 2 amphibole asbestos. 3 Ores from the Enoree, SC deposits are primarily hydrobiotite and biotite in origin. 4 Fluroapatite is a common mineral collocated with the hydrobiotite. Zircon is also widely 5 dispersed throughout the plutons along with minor accessory minerals including talc, chlorite, 6 chromite, rutile, titanite, corundum, anatase, and amphibole asbestos (Hunter. 19501. The 7 amphibole asbestos identified in the vermiculite deposit at Enoree, SC has been classified as 8 tremolite (Libby. 19751. 9 As previously noted, EPA Region 8 obtained samples ofore from Libby, MT, South 10 Africa, and Virginia from Dr. James Lockey, University of Cincinnati, and analyzed the samples 11 to determine the particle-size distribution (length, width, and aspect ratio), using TEM and EDS 12 to identify the mineral composition of the amphibole fibers. Region 8 was unable to acquire a 13 sample of ore from the South Carolina Enoree mine complex for analysis. Region 8 conducted 14 analysis of the ore and exfoliated materials to connect the exposures of workers to mineral fibers 15 in Marysville, OH, to the ore originating in Libby, MT. The connection is based on fiber 16 morphology, mineralogy, and fiber-size similarities. 17 In order to analyze the fibers from the ore and vermiculite bulk material, the fibers must 18 be loaded onto filters and prepared for analysis by TEM. Three potential methods were 19 considered for transferring the fibers from the bulk material to filters: water elutriation, 20 glove-box transfer, and the fluidized bed asbestos segregator (EBAS). Of these three methods, 21 only the glove-box and FBAS involved physical disturbance of the bulk material to elutriate 22 fibers into the air that might be similar to handling and processing of ore in the Marysville, OH 23 plant. Due to the limited quantity of test material available for analysis, Region 8 employed the 24 FBAS as an analytical instrument to load the mineral fibers onto filters for TEM analysis. 25 Briefly, samples of ore and vermiculite were prepared following the procedure outlined 26 by Bern et al. (20021. Samples were dried, ground with a Wylie mill and mortar and pestle, and 27 sieved through a 230-pm (60 mesh) sieve. Samples (exactly 2.0 g) were mixed with 18 g of 28 analytical silica sand and placed in a FBAS vessel to load 25-mmmixed cellulose ester air 29 sampling filters (0.8-p pore size). The FBAS was run for 3 minutes to load the filter cassettes 30 with sufficient fibers for analysis by TEM. Five filters were loaded for each of the ore and This document is a draftfor reviewpurposes only and does not constituteAgency policy. C-4 DRAFT--DO NOT CITE OR QUOTE I vermiculite samples. After loading, the filters were prepared for TEM analysis by mounting on 2 copper grids, carbon coating, and subjected to TEM analysis (TEM-ISO 10312 method). 3 The laboratory followed fiber counting rules detailed in the Quality Assurance Project 4 Plan for the specific study using Libby-specific laboratory modifications. Total amphibole fibers 5 and Phase Contrast Microscopy equivalent (PCMe) fibers were counted for each ofthe 6 ore/vermiculite samples as described in Appendix B. A total of 1.0 mm2 area or a total of 7 200 asbestos structures were counted to achieve the desired analytical sensitivity (1/g; 1.5 * 10'1). 8 DS was performed on selected samples from each of the vermiculite/ore samples to provide 9 mineral characterization of individual fibers. Fiber counts were recorded on National Asbestos 10 Data Evaluation Sheet data sheets for further analysis. Only the Libby, MT vermiculite and II Libby, MT ore samples had sufficient fibers detected to construct a fiber-size distribution. 12 Fiber counts were determined by counting fiber numbers for a specific area of the filter 13 grid or a specific number of grid openings (whichever was achieved first) to determine total 14 fibers present As shown in Table C-l, the number of fibers forthe testmaterials varied greatly 15 depending on the source, and the grid area measurement was exceeded prior to the fiber count 16 metric (167 grid openings ~I.O mm2). 17 18 Table C-l. Fiber detected in ore and expanded product' Sample type Grid openings Structures counted LA OA c Concentration (s/g) LA OA c Virginia Ore Virginia Expanded 167 - - 167, .. 1 0.... ....0 W ' -0 0 00 ff|E3;oo&| 6. 0 0 South Africa Ore 167 20 2 26,403 0 26,403 South Africa Expanded 167 00 0 00 0 tabby #3 Ore 167 320 0 0 1,393,873 0 0 Libby Expanded - Sllli 100 ' O'- - - -o,' PfSPI o 19 20 LA = Libby Amphibole, OA = Other amphibole, C = Chrysotile. Note: the designation of fibers as Libby Amphibole 21 in this instance reflects only a qualitative morphological comparison to amphiboles of the Libby, MT series. 22 23 24 ThesLibby #3,ore.and the. Libby #3 expanded materialveontained the^greatesthumberajfss 25 fibers both in fiber counts on the filters and in calculated structures per gram ofbulk-material. This document is a draftfor reviewpurposes only and does not constituteAgencypolicy. C-5 DRAFT--DO NOT CITE OR QUOTE 1 Virginia expanded and Sbutli African ore contain ampHibole structures repre^ente^by low fiber 2 counts. South African ore also contained chrysotile fibers?as detemined by moiphology and;* 3 EDS analysis.; The absence of fibers detected in the Virginia ore and the South African4 expanded materials probably represents actual low fiber content of the ore and is a function of 5 the detection limit for the structure analysis. The estimation of structures per gram of material 6 indicated that there were 13,000 to 26,000 fibers per gram of bulk material, which was 7 approximately 18 times lower than the Libby, MT ore samples. The decrease in fibers found in 8 the Marysville, OH facility after 1980 when only ore from Virginia, Palabora, and South 9 Carolina was used (see Appendix F) is consistent with the findings of low fiber counts for the 10 Virginia and Palabora materials. In addition, numerous nonasbestiform minerals were also 11 detected including biotite, micas, and pyroxenes in the bulk materials from Virginia and South 12 Africa. 13 Amphiboles are a complex group of minerals characterized by double chains of silicate 14 tetrahedrons and the generic chemical formula ofAo^iBiCsTzOnlOE^z where A, B, C, and T 15 represent the various cations. The modem classification system of amphiboles is described in 16 Leake et al. (1997). To classify the mineral species of the amphibole, it is not sufficient to 17 determine its composition; the various cations must be assigned to the specific A, B,C, and T 18 sites. The cutoffs of the compositional ranges allowed for each amphibole mineral species are 19 based on the number of the cations in the various sites. The methodology to classify an 20 amphibole is to first determine its elemental compositions (e.g., as expressed as weight percent 21 oxide for each element or as atomic percent for each element). Then a normalized routine is 22 applied to the raw elemental measurements to calculate the number of each of the cations 23 contained in one formula unit. (This is a simple arithmetic calculation since the cation percents 24 have been measured, and the stoichiometry must balance the charges of the cations and anions.) 25 Generally, one formula unit is assumed to contain 23 oxygens. Next, the sites are filled up by 26 assigning cations to them subsequently, specifically: 27 28 This document is a draftfor reviewpurposes only and does not constitute Agencypolicy. C-6 DRAFT--DO NOT CITE OR QUOTE i j i i f 1 T: Si4+, Al3+, and Ti4+ 2 C: Al3+ and Ti4+ (only after the T sites are filled first) and then Mg24, Fe2+, Fe3+, and 3 then Mn2+. 4 B: Any remaining Mg2* Fe2* and Mn2+ (after the C sites are filled), all Ca2+, then 5 Na* ifthere is any room left. 6 A: Na+ and K+ only. 7 8 9 Once the cations are assigned to their sites, it is a simple matter to classify the minerals 10 based on the cutoffs of the composition field allowed for each mineral. 11 The LibbyAmp^bole?asbestos4 group'6fiminerals i's:ra:comple3t;:group;of.amphiboles ;: 12 .consisting qf.six;minerals: 13 14 15 " Winchite, CaNa[Mg, Fe24]4[Al, Fe3+]Sig022 [OH]2 16 Richterite; NaCaNa [Mg, Fe2*, Mh, Fe3+]5Si8022[0H]2 17 Tremolite, Ca2Mg5Si8022[0H]2 18 Magnesio-riebeckite, Na2[Mg3, Fe3+2]Sig022[0H]2 19 Magnesio-arfvedsonite, NaNa2[Mg4,Fe3+]Sis022[0H]2 20 Edenite, NaCa2Mg5Si7A1022[0H]2 21 22 23 Libby Amphibole is characterized by a low amount of Al in the T site---and a 24 correspondingly high Si content--so, according to Leake's (19971 classification, if the Si 25 (expressed as atoms per formula unit, apfii) is at least 7.5, and Al content in the Tsite is <0.5, all 26 6 libby Amphibole types can be plotted on a graph ofNa content of the B site versus the 27 (Na + K) content in the A site. This approach was described by Meeker et al. 12003) for the 28 Rainy Creek complex. 29 EDS spectra (TEM/EDS) were collected from all amphibole fibers found in the South 30 Africa and Virginia samples, and six randomly selected Libby Amphibole asbestos fibers in each 'The term "Libby Amphibole asbestos" is used in this document to identify the mixture of amphibole mineral fibers of varying elemental composition (e.g., winchite, richterite, tremolite, etc.), that have been identified in the Rainy Creek complex near Libby, MT. It is further described in Section 2.2. This document is a draftfor review purposes only and does not constitute Agency policy, C-7 DRAFT--DO NOT CITE OR QUOTE 1 of the Libby, MT ore and Libby, MT expanded samples. Two bundles of asbestifonn serpentine 2 (chiysotile) were found in the South African ore sample. EDS spectra were collected for one of 3 the bundles. The chemical formula of serpentine is MgjSijOsfOKTk. The EDS software package 4 collected and summarized each spectrum to determine the atomic percent of each element of 5 interest. 6 Several assumptions were made in the treatment of the TEM/EDS data: 7 8 9 1. Numbers of cations per formula unit are calculated on the basis of23 oxygens. This may 10 or may not be correct because an [OH] site in the amphibole crystal can be occupied by 11 either OH", F", Cl" or O2-. The calculated cation numbers will be affected if a significant 12 quantity of O2" is in the OH site. 13 2. A persistent problem with amphiboles is that they can contain both ferric [3+] and ferrous 14 [2+] iron in die same crystal. For the purposes of this report all Fe was assumed to be 15 Fe2+. A method for calculating the ratio of Fe2+ to Fe31- is described in Leake et al. 16 (1997), but it is very complex, applies to polished sections, and was not attempted for this 17 report. 18 3. For the purposes of this report, the T sites were assumed to be filled completely full to 19 8 apfo, and the C sites were assumed to be completely foil to 5 apfo. All Ca and any Mg, 20 Fe, and Mh remaining after the C site was foil were then assigned to the B site. Next, Na 21 was assigned to the B site until it was foil (2 apfo), then any remaining Na and all K were 22 assigned to the A site. 23 24 25 Applying these assumptions to the TEM/EDS data produces a useable graph of the Na 26 and K content of the amphibole fibers. As shown in Figure C-l, Libby #3 ore and Libby #3 27 Expanded amphiboles were characteristic of winchite and treinolite. Virginia Expanded and 28 South African ore both contained amphibole fibers characteristic of non-Libby (Na and K) in the 29 tremolite series. 30 31 This document is a draftfor reviewpurposes only and does not constitute Agencypolicy. C-8 DRAFT--DO NOT CITE OR QUOTE Sodium and Potassium Content of Individual Amphibole Fibers as Measured by TEM/EDS 3 E o 0) L <0 E (0 c <0 < c + -------------------- 0---------- gdeniio Rit-Merite .A A Q TremoliteActirsolite , ^ Winchite A I i__ ___________________ 0.0 0.5 1.0 1.5 Na in B Site (atoms per formula unit) o Virginia Expanded v South Africa Ore a Libby #3 Ore Libby #3 Expanded 1 2 Figure 0-1. Cation values for Na In theJJ site and the Na + K in the .4 site 3 from individual amphibole fibers. 4 5 6 Following all assumptions described above and the approach ofplotting Na in the B site 7 versus Na + K in the A site as described by Meeker et al. (2003). the mineral species of the 8 Marysville, OH fibers can be described as: 9 10 11 The single Virginia amphibole asbestos fiber is an actinolite 12 Both of the South African amphibole fibers are tremolite 13 8 of the Libby Amphibole asbestos fibers from Libby, MT are winchite 14 4 of the Libby Amphibole asbestos fibers from Libby, MT are tremolite This document is a draftfor reviewpurposes only and does not constitute Agency policy. C-9 DRAFT--DO NOT CITE OR QUOTE CDF CDF ' Length (uim) Width (um) 1 2 Figure C-2. Fiber-size distribution of Libby Amphibole asbestos. This document is a draftfor reviewpurposes only and does not constitute Agencypolicy. C-10 DRAFT--DO NOT CITE OR QUOTE CDF 1 Fiber-size distributions for amphibole fibers from the Libby #3 ore and Libby #3 2 expanded sources were conducted on the fibers counted during the TEM analysis of the filter 3 grids. Due to the low fiber count detected in the Virginia and South Africa sources, it was not 4 possible to develop a fiber-size distribution for these fibers. The Libby Amphibole asbestos 5 fiber-size data were plotted as a cumulative distribution frequency for fiber length, fiber width, 6 and aspect ratio. These data were compared to Libby Amphibole asbestos fibers collected in 7 Libby, MT as part of EPA's ongoing ambient air monitoring program and the Libby Asbestos 8 Superfund site (see Appendix B). The Libby, MT ore and expanded material showed an 9 increased frequency of longer and wider fibers than the fibers from the Libby, MT ambient 10 air-sampling program. Aspect ratios were nearly identical. The differences between the length 11 and width frequency were not outside of the expected range for Libby Amphibole asbestos fibers 12 and were consistent with fiber-size distributions for soil activity-based-sampling data from 13 Libby, MT. 14 Based on the TEM morphological analysis of filter grids, TEM/EDS analysis for the fiber 15 mineralogy, and the fiber-size distribution data, it can be concluded that the amphibole fibers 16 detected in the Libby # 3 ore samples from the Salt Lake Expansion facility are consistent with 17 data from authentic Libby Amphibole fibers (Meeker et al- 20031 found in Libby, MT (see also 18 Appendix B). Further, ore samples from Virginia and South Africa contained amphibole and 19 chiysotile fibers but at a much lower frequency of detection than the Libby Amphibole ore as 20 reported in Appendix F. 21 22 This document is a draftfor reviewpurposes onfy and does not constitute Agencypolicy. C-11 DRAFT--DO NOT CITE OR QUOTE t i i i Cl. REFERENCES Bern. AG; Meeket^GP: Brownfield. 1. (2002). Guide to analysis of soil samples from Libby, Montana for asbestos content by scanning electron microscopy and energy dispersive spectroscopy. In USGS Open File Report. USGS. ' Frank. D: Edmund. L. (2001). Feasibility for identifying mineralogical and geochemical traces from vermiculite ore deposits. (EPA 910-R-01-002). Seattle, WA: U.S. EPA Region 10. Gooch, EO. (1957). Vermiculite. 3:1-5. . Hunter. CE. (19501. Vermiculite ofthe southeastern, states. In Symposium on mineral resources of the southeastern United States, Knoxville, TN: University ofTennessee Press. . IOM Consulting. (2008). Sampling and analysis of crude vermiculite samples for possible asbestiform fibre and quartz content. (609-02386). Surrey, England: Palabora Mining Co, Palabora Europe Ltd. Leake. BF.: Woolley. AR: Ams. CF.S: Birch. WD: Gilbert, MC: Grice. JD; Hawthorne, FC: Katn. A: Kisch. HJ; Krivovichev. VG: Linthout, K: Laird. J: Mandarine. J: Maresch. WV: Nickel. EH: Rnck. NMS: Schumacher. JC: Smith. DC: Shephenson. NCN: Ungaretti. L: Whittake. EJW: Ynnzhi. G. (1997). Nomenclature ofamphiboles: Report ofthe Subcommittee on Amphiboies ofthe International Mineralogical Association Commission on New Minerals and Mineral Names. Mineral Mag 61: 295-321. Libbv. SC. (1975) The origin of potassic ultramafic rocks in the Enoree `'Vermiculite" District, South Carolina. Pennsylvania State University, University Park, PA. Meeker. GP: Bern. AM: Brownfield. IK: Lowers. HA: Sutlev. SI: Hoefen. TM: Vance. JS. (2003). The composition and morphology of amphiboles from the Rainy Creek Complex, near Libby, Montana. American Mineralogist 88: 1955-1969. http://dx.doi.org/10.1093/annhvg/38.inhaied Particles VIL639. Meisinger, AC. (1979). Vermiculite. In Minerals Yearbook 1978-1979, Metals and Minerals. Washington, DC: U.S. Bureau of Mines. Moatamed. F: Lockev. JE: Parry, WT. (19861. Fiber contamination ofvermiculites: a potential occupational and environmental health hazard. Environ Res 41: 207-218. Rohl. AN: Langer. AM. (19771. Mineral analysis of core samples from the Green Springs area. Virginia vermiculite deposit: Unpublished letter report from Mt. Sinai School of Medicine. Schoeman. II. U989V Mica and vermiculite in South Africa. 1 South African Institute of Mining and Mineralogy 89: 1-12. Wviie. AG: Verkouteren. JR. (2000). Amphibole asbestos from Libby, Montana: Aspects ofnomenclature. American Mineralogist 85: 1540-1542. This document is a draftfor review purposes only and does not constitute Agencypolicy. C-12 DRAFT--DO NOT CITE OR QUOTE