<<

883

The Canadian Mineralogist Vol. 41, pp. 883-890 (2003)

CHARACTERIZATION OF , AND LIZARDITE BY FT-RAMAN SPECTROSCOPY

CATERINA RINAUDO¤ AND DANIELA GASTALDI Dipartimento di Scienze e Tecnologie Avanzate, Università del Piemonte Orientale “Amedeo Avogadro”, Corso Borsalino 54, IÐ15100 Alessandria, Italy

ELENA BELLUSO Dipartimento di Scienze Mineralogiche e Petrologiche, Università di Torino, Via Valperga Caluso 35, I–10125 Torino, Italy

ABSTRACT

We analyzed samples of antigorite, lizardite and fibrous chrysotile, three representative of the serpentine group, to determine their chemical and structural properties, and their FT-Raman spectra. The low-frequency region of the spectrum (1200Ð 200 cmÐ1), which corresponds to the lattice-vibration modes, was then analyzed, and the observed bands attributed, on the basis of our results. Raman spectroscopy thus proves useful in identifying the three species, despite their strong similarities in structure and chemical composition. The symmetric stretching modes of the SiÐObÐSi linkages vibrate at different frequencies in chrysotile and antigorite, whereas chrysotile and lizardite can easily be identified by analyzing the vibrations of the OHÐMgÐOH groups and the bands in the range 550Ð500 cmÐ1.

Keywords: Raman spectroscopy, , serpentine-group minerals, chrysotile, antigorite, lizardite.

SOMMAIRE

Nous avons analysé des échantillons d’antigorite, de lizardite et de chrysotile fibreuse, trois minéraux représentatifs du groupe de la serpentine, afin d’en déterminer les propriétés chimiques et structurales, et leurs spectres de Raman avec transformation de Fourier. La région du spectre à faible fréquence (1200–200 cmÐ1), qui correspond aux modes de vibration du réseau, a été analysée, et les bandes observées ont été attribuées. Les spectres s’avèrent ainsi utiles dans l’identification des trois espèces, malgré leurs grandes ressemblances structurales et compositionnelles. Les modes d’étirement symétriques des agencements SiÐOb–Si vibrent à des fréquences différentes dans le chrysotile et l’antigorite, tandis que chrysotile et lizardite sont facilement identifiables par analyse des vibrations des groupes OH–Mg–OH et des bandes dans l’intervalle 550–500 cmÐ1.

(Traduit par la Rédaction)

Mots-clés: spectroscopie de Raman, minéraux asbestiformes, minéraux du groupe de la serpentine, chrysotile, antigorite, lizardite.

INTRODUCTION octahedral and tetrahedral cations away from their ideal positions and to the limited Al-for-Si substitution in the The three principal minerals of the serpentine group, tetrahedral sites. In antigorite, the misfit is compensated chrysotile, antigorite and lizardite, have a very similar by an alternating-wave structure in which the sheet of chemical composition, but significantly different struc- octahedra is continuous, whereas the SiO4 tetrahedra are tures. In fact, the single ideal chemical formula tilted and periodically switch their orientation, pointing (OH)3Mg3[Si2O5(OH)] corresponds to several crystal alternatively in opposite directions. Finally, chrysotile structures that represent different solutions for the mini- has a coiled structure, which is responsible for its prop- mization of the mismatch between sheets of SiO4 tetra- erties as asbestos (Wicks & Whittaker 1975, Wicks & hedra and sheets of MgO2(OH)4 octahedra. In lizardite, O’Hanley 1988, Veblen & Wylie 1993). the resulting structure is planar, owing to shifts of the

¤ E-mail address: [email protected] 884 THE CANADIAN MINERALOGIST

Naturally occurring intergrowths of three types of As the XRD spectra are very similar for minerals serpentine minerals are very common. It is thus of great with such similar chemical and crystallographic prop- interest to find an experimental technique that permits erties, more precise characterization of each sample was identification of the mineral species without long and performed by transmission electron (Philips costly preparation of the sample. Raman spectroscopy CM12 instrument, operated at 120 kV); they were fits these criteria: it is a simple and rapid technique, and chemically analyzed with the attached EDS microprobe necessitates no sample preparation. It has been used [EDAX Si(Li) detector PV9800]. The standardless successfully by a number of investigators in the study SUPQ program was used with the default K factors to of chrysotile (Lewis et al. 1996, Bard et al. 1997, process the chemical data; the final values were obtained Kloprogge et al. 1999), but only one bibliographic ref- as normalized proportion of (wt.%). The samples erence has been found concerning its use in the identifi- were prepared by suspending the powder in isopropyl cation of antigorite and lizardite (Pasteris & Wopenka alcohol, and the suspension then underwent ultrasound 1987). In this latter work, no attribution for the bands is treatment in order to minimize aggregation; finally, made to the vibrational modes of the groups constitut- several drops of the suspension were deposited on a ing the crystal structure. carbon-supported Cu grid. In order to identify the Raman vibrations that are The percentage of each mineralogical phase detected specific to each serpentine phase studied and that would in the XRD spectrum was easily calculated using the permit rapid recognition of the mineral, several speci- EVA support program, which compares the area and mens of the three species were examined. Before intensity of two diffraction peaks, each diagnostic of one characterization by Raman spectroscopy, the crystallo- mineral component of the mixture. It is worth noting graphic and chemical properties of each sample were that the calculated values are reliable in powdered speci- established using the techniques of X-ray powder dif- mens without preferred orientation of the component fraction (XRD), scanning electron microscopy (SEM), crystallites. In our case, especially for chrysotile, some transmission electron microscopy (TEM), analytical preferred orientation of the certainly was present. electron microscopy (AEM), and energy-dispersion The percentages calculated on the basis of the XRD spectrometry (EDS). spectra were therefore compared with the data obtained using TEM and AEM. In this case, selected-area elec- EXPERIMENTAL tron diffraction (SAED) patterns and EDSÐTEM chemi- cal analyses of each point examined permitted Some of the samples of chrysotile studied were pro- conclusive attribution to the mineral phase. Many crys- vided by NIST (National Institute of Standards and tals and different points within a single bundle of fibers Technology). Others were taken from the reference col- from each specimen of chrysotile and antigorite were lections of the Dipartimento di Scienze Mineralogiche analyzed by SAED and EDSÐTEM. Thus it was pos- e Petrologiche, Università di Torino, and are found in sible to calculate the proportion of each coexisting outcrops of rocks rich in serpentine minerals in the Pied- mineral phase. mont Alps, Italy. A morphological study was performed with a Cam- All of the samples of antigorite studied were taken bridge Stereoscan 360 SEM: the samples were glued from the Dipartimento di Scienze Mineralogiche e onto aluminum stubs with colloidal graphite and then Petrologiche, Torino, and come from the same areas coated with a carbon film approximately 400 Å in thick- mentioned above. ness. The sample of lizardite in the study comes from Finally, the serpentine phases were studied in a back- Monte Fico, Elba Island, Italy; its structural and chemi- scattering geometry with a Bruker RFS100 FT-Raman cal properties were studied previously by Mellini & Viti spectrometer equipped with an Nd:YAG laser at an ex- (1994), and the sample was kindly provided by Profes- citation wavelength of 1064 nm, and a Ge detector, sor Mellini. which requires liquid N2 for cooling. The advantage in For the XRD analyses, the samples were powdered using a laser operating at near-infrared radiation is that in an agate mortar. No problems were encountered with the fluorescence from the sample is reduced, heating and the antigorite or lizardite, but chrysotile fibers needed phase transition are avoided, and measurements are to be cut with scissors before crushing, as they are so made possible for longer periods of time. All the spec- flexible and soft. tra of each mineral were recorded on several mg of pow- Each powdered sample was mounted on a flat holder dered sample. Random orientation of the crystallites in and examined initially by the powder method using a the powder analyzed was achieved only with the Siemens D5000 X-ray diffractometer equipped with antigorite and lizardite samples; in the case of chryso- CuK radiation. The EVA software program was used tile, some preferred orientation of the fibrils is impos- to determine the mineral phases present in each X-ray sible to avoid. The Raman spectra were obtained at a powder spectrum, by comparing the experimental peaks 2cmÐ1 resolution and by adapting the laser power and with PDF2 reference patterns. number of scans in order to optimize the signal-to-noise FT-RAMAN SPECTROSCOPY OF CHRYSOTILE, ANTIGORITE AND LIZARDITE 885 ratio: 2000 scans over 100 minutes and 120 mW were proved to be: (Mg11.70Al0.08Fe0.22)12.00(Si7.98Al0.02)8.00 sufficient in order to obtain the optimum resolution. O20(OH)16. Internal calibration of the Raman spectrometer was XRD and TEM analyses of the antigorite sample assured by a reference HeÐNe laser (632.8 nm). The chosen for Raman characterization revealed an almost accuracy of the band position was also checked by mea- pure antigorite; no more than 3% of chrysotile was suring the center wavenumber of two Stokes and anti- found to be present. AEM analyses indicated a chemi- Stokes bands at low frequencies: if everything is cal composition of: (Mg11.05Fe0.53Al0.21)11.79Si8.05O20 correctly calibrated, they are equal. (OH)16. In this work, only the Raman region related to lat- For the lizardite analyzed by Raman spectroscopy, tice-vibrational modes ranging from 1200 to 200 cmÐ1 Mellini & Viti (1994) reported the composition: was analyzed. (Mg10.96Fe0.64Al0.36)11.96(Si7.72Al0.28)8O20(OH)16. XRD patterns indicated that the lizardite is mixed with RESULTS AND DISCUSSION a very small amount of chrysotile (about 5%). The samples chosen, as studied by optical micros- The three species of serpentine considered in this copy and by SEM, revealed the morphology represented work are commonly found in a single specimen; several in Figures 1aÐc. The chrysotile, as shown in Figure 1a, samples were tested, especially for chrysotile and is made up of thin and flexible fibrils, classifiable as antigorite. The samples selected for Raman spectros- asbestos because of their length:diameter ratio, which copy satisfied the criterion of having one largely pre- is >3 (Davis 1993); antigorite exhibits the columnar dominant phase, as revealed by XRD, TEM and AEM habit shown in Figure 1b, and lizardite has the plate- studies. The chrysotile, antigorite and lizardite speci- like morphology shown in Figure 1c. mens selected showed the presence of other phases as The samples were studied last by Raman spectros- well, but in minor amounts. copy; for each sample, several spectra were registered The chrysotile sample chosen accounts for about on small quantities of ground material. Figure 2 shows 95%, and antigorite, for about 5% of the whole sample. a Raman spectrum obtained on the chrysotile sample of The chemical composition of the chrysotile fibers in the Figure 1a. The spectrum is characterized by the follow- sample studied was established by several analyses per- ing main bands, assigned according to the description formed under TEM by AEM on the selected areas, and of the vibrational bands in the IR spectra of the

FIG. 1a. Secondary electron SEM image of the chrysotile studied by Raman spectroscopy. The flexibility of the constituent fibers is evident. Scale bar: 50 m. 886 THE CANADIAN MINERALOGIST

FIG. 1b. Secondary electron SEM image of the antigorite sample studied by Raman spectroscopy. The sample shows a columnar habit. Scale bar: 20 m.

FIG. 1c. Secondary electron SEM image of the lizardite sample studied by Raman spectroscopy. Scale bar: 20 m. FT-RAMAN SPECTROSCOPY OF CHRYSOTILE, ANTIGORITE AND LIZARDITE 887 phyllosilicates proposed by Lazarev (1972) and Farmer assigned, respectively, to A1(1) and B2(3) vibrations (1974), and to the assignment for Raman bands in tubu- of the OÐHÐO groups (Frost 1995, 1997, Frost & Kristof lar chrysotile, as suggested by Kloprogge et al. (1999): 1997). In trioctahedral phyllosilicates, the occupancy of 1) a band at 1105 cmÐ1, attributed to antisymmetric almost all the octahedral sites by divalent cations pre- stretching modes (as) of the SiÐOnb groups; vents the formation of an OÐHÐO isosceles triangle. 2) an intense band at 692 cmÐ1, attributed to the This fact changes the symmetry of the OÐHÐO groups; Ð1 symmetric stretching mode (s) of the SiÐObÐSi groups; only one band is observed in the 200Ð300 cm range in 3) a band at 620 cmÐ1 produced, following the spectra of the trioctahedral phyllosilicates studied. Kloprogge et al. (1999), by OHÐMgÐOH translation Nevertheless, assignment of this band to a precise vi- modes on the basis of the assignment of the band at 627 brational mode of the OÐHÐO groups requires polariza- cmÐ1 observed on the infrared spectrum of . On tion of Raman measurements, and is, therefore, the other hand, Farmer (1974, p. 348) also attributed the impossible for the moment. strong bands observed in the chrysotile and antigorite In Figure 3, we report a Raman spectrum obtained IR spectra near 618Ð646 cmÐ1 to vibrations of the inner on the antigorite sample shown in Figure 1b; on the and surface hydroxyl groups of the layers; whole, the spectrum appears more disturbed by noise; 4) two bands at 389 and 345 cmÐ1, in a region of the the main bands already observed in the previous spec- spectrum where bending vibrations of the SiO4 tetrahe- trum may be found again, but lying at different frequen- dra appear. The first one was assigned by Kloprogge et cies. We see: Ð1 al. (1999) to the 5(e) bending modes of the SiO4 tetra- 1) a band occurring at 1044 cm that may corre- hedra. However, two bands at 379 and 353 cmÐ1 have spond to the band observed in the IR reflection spec- also been detected previously in from Hohen- trum of this mineral at 1048 cmÐ1 and ascribed by Hagen, Germany, and ascribed to bending vibrations of Lazarev (1972, p. 123) to antisymmetric stretching the SiO4 units (Frost & Kloprogge 2000); modes (as) of the SiÐObÐSi groups (E1) and to the band 5) one band lying at 231 cmÐ1 in a region of the spec- detected in Raman spectrum of at 1049 cmÐ1 by trum where vibrations of the OÐHÐO groups are ex- Rosasco & Blaha (1980) and also ascribed to antisym- pected [O is the non-bridging atom of the metric stretching vibrations of SiÐObÐSi groups; Ð1 tetrahedron SiO4, and H is the hydrogen atom of the OH 2) an intense band at 683 cm produced by sym- group tilted toward an octahedral cation vacancy metric stretching modes (s) of the SiÐObÐSi linkages; (Griffith 1969, Loh 1973). In the dioctahedral phyllo- 3) a band at 635 cmÐ1 ascribed, as for chrysotile, to silicates, where normally one third of the octahedral the OHÐMgÐOH translation modes. The shift of these sites are vacant, the OÐHÐO groups form isosceles tri- vibrations in the two spectra may be correlated with the angles. The Raman-active vibrational modes in an isos- higher content of Fe or Al rather than Mg in the sheets Ð1 celes triangle are: symmetric stretching (1) at 262 cm , of octahedra, as demonstrated by AEM analyses; Ð1 symmetric bending (2) at 187 cm , and antisymmetric 4) the band due to 5(e) modes of the SiO4 tetrahe- Ð1 Ð1 stretching (3) at 240 cm (Loh 1973). In fact, in the dra lies here at 375 cm , a slightly lower frequency than FT-Raman spectrum of the dioctahedral phyllosilicates in the chrysotile. Note that in chrysotile spectrum ob- studied, , , and nontronite, two tained by Kloprogge et al. (1999), a strong band at bands near 270Ð290 and 240 cmÐ1 are observed and 388 cmÐ1 lies near another band at 374 cmÐ1. In this case,

FIG. 2. An FT-Raman spectrum obtained for the chrysotile FIG. 3. An FT-Raman spectrum obtained for the antigorite sample in Figure 1a. sample in Figure 1b. 888 THE CANADIAN MINERALOGIST

Kloprogge et al. (1999) suggested for the intense band sities of these bands in the spectra shown in Figures 2 Ð1 at 388 cm a 5(e) mode of the SiO4 tetrahedra, and for and 4 are also comparable. On the contrary, the anti- Ð1 the band at 374 cm , vibrations of the MgÐOH groups. symmetric stretching modes (as) of the SiÐOnb, which In the spectrum of chrysotile that we obtained, no band appear as a clear band at 1105 cmÐ1 in the chrysotile at 374 cmÐ1 is observed (Table 1, Fig. 3a); only the vi- spectrum, here produce a weak and dubious band at Ð1 Ð1 brational modes of SiO4 tetrahedra at 389 cm are de- 1096 cm . When we now take into consideration the tected as a strong band. In the spectrum of antigorite, vibrations of OHÐMgÐOH groups, they appear to be the intensity of the band at 375 cmÐ1 is comparable to shifted to higher frequencies on the lizardite spectrum, that of the band at 389 cmÐ1 in the chrysotile spectrum; 630 cmÐ1 , when compared to the vibrations of the same in our opinion, this band can therefore be ascribed to groups in chrysotile, 620 cmÐ1. This shift, as discussed bending modes of SiO4 tetrahedra. for antigorite, may be related to the higher content of Al 5) there is no significant difference in the band near and Fe, rather than Mg, in the layer of octahedra. 230 cmÐ1 compared to the chrysotile spectrum; as dis- A band near 510 cmÐ1 has also been detected in the cussed above, it derives from the vibrations of the OÐ Raman spectrum of lizardite. It occurs in a frequency HÐO groups. range where no band appears in the chrysotile spectrum. The band occurring in the chrysotile spectrum at 346 This band is observed in all spectra of lizardite recorded, cmÐ1 is not present in antigorite; moreover, an additional and it seems to be made up of two bands lying at 506 band at 520 cmÐ1 is detected. Samples of talc show a and 515 cmÐ1. In agreement with the assignment of the Raman band of weak intensity at 519 cmÐ1 (Blaha & band at 520 cmÐ1 in the antigorite spectrum, an assign- Rosasco 1978, Rosasco & Blaha 1980) and in the IR spectrum a band at 530 cmÐ1. The latter is assigned by Lazarev (1972, p. 122) to deformation vibrations of the siliconÐoxygen layer. Similarly, we propose that the band observed in the antigorite spectrum at 519 cmÐ1 be assigned to the deformation modes of the SiO4 tetrahe- dra. In fact, as is typical in the antigorite structure, the sheet of tetrahedra periodically switches its orientation, and deformations of the SiO4 tetrahedra at the reversal zone have been proposed. In Figure 4, the Raman spectrum from the lizardite shown in Figure 1c is represented. The bands produced by vibrations of the SiO4 tetrahedra or by siliconÐoxy- gen linkages appear here at frequencies very close to those detected in the chrysotile spectra. In fact, the modes due to symmetric stretching (s) of the SiÐObÐSi linkages lie here at 690 cmÐ1 (692 cmÐ1 in chrysotile), and the 5(e) modes of the SiO4 tetrahedra appear here at 388 cmÐ1 (389 cmÐ1 in chrysotile). The relative inten-

FIG. 4. An FT-Raman spectrum obtained for the lizardite sample in Figure 1c. FT-RAMAN SPECTROSCOPY OF CHRYSOTILE, ANTIGORITE AND LIZARDITE 889 ment to deformation vibrations of the SiO4ÐAlO4 tetra- ACKNOWLEDGEMENTS hedra may be proposed. In fact, in the structure of lizardite, the mismatch between the layers of tetrahedra The authors are grateful to Professor M. Mellini of and octahedra is compensated by deviation of the sheet the Università di Siena, who kindly provided the of tetrahedra from an ideal hexagonal symmetry to a lizardite sample, Professor P.L. Stanghellini of the ditrigonal arrangement with tilting of the tetrahedra of Università di Piemonte Orientale and Professor Sidney about Ð2.8° and by a more substantial substitution of F.A. Kettle for the helpful discussions. A particular Al-for-Si in the layer of tetrahedra (Mellini 1982, thanks also goes to the referees and to Associate Editor Mellini & Viti 1994). The splitting of the band at about Mickey Gunter for their work. 510 cmÐ1 into two bands at 509 and 516 cmÐ1 may there- fore reflect deformation vibrations of the two types of REFERENCES tetrahedra, AlO4 and SiO4. In Table 1, the bands observed in the spectra of the BARD, T., YARWOOD, J. & TYLEE, B. (1997): Asbestos fiber three minerals studied are reported and compared with identification by Raman microspectroscopy. J Raman the results obtained on chrysotile by Kloprogge et al. Spectrosc. 28, 803-809. (1999). BLAHA, J.J. & ROSASCO, G.J. (1978): Raman microprobe spec- tra of individual microcrystals and fibers of talc, , CONCLUSIONS and related silicate minerals. Anal. Chem. 50, 892-896.

On the basis of our findings, Raman spectroscopy DAVIS, J.M.G. (1993): In vivo assays to evaluate the pathogenic provides a reliable and simple-to-perform technique for effects of minerals in rodents. In Health Effects of Mineral distinguishing among the three common phases of ser- Dusts (G.D. Guthrie Jr. & B.T. Mossman, eds.). Rev. Min- pentine, chrysotile, antigorite and lizardite. In fact, after eral. 28, 471-487. comparing the spectra shown in Figures 2, 3 and 4, we conclude that: FARMER, V.C. (1974): The Infrared Spectra of Minerals. The Mineralogical Society, London, U.K. a) the frequencies of the stretching modes of SiÐObÐ Si and SiÐOnb linkages allow chrysotile to be distin- FROST, R.L. (1995): Fourier transform Raman spectroscopy of guished unequivocally from antigorite (s and as of kaolinite, dickite and halloysite. Clays Clay Minerals 43, antigorite less than s and as of chrysotile); 191-195. b) chrysotile and lizardite are easily distinguishable in the region where the frequencies of the bands pro- ______(1997): The structure of the kaolinite minerals Ð a duced by the vibrations of the OHÐMeÐOH groups oc- FT-Raman study. Clay Minerals 32, 65-77. cur. In fact, the spectrum of lizardite closely resembles that of chrysotile where the different siliconÐoxygen or ______& KLOPROGGE, J.T. (2000): Raman spectroscopy of nontronites. Appl. Spectrosc. 54, 402-405. SiO4 tetrahedra vibrations occur, but it resembles the antigorite spectrum where the vibrations of the octahe- ______& KRISTOF, J. (1997): Intercalation of halloysite: a drally coordinated cations linked to OH occur. More- Raman spectroscopic study. Clays Clay Minerals 45, 551- over, another clue to identifying the mineral phase can 563. be found in the presence in the lizardite spectrum of a band (actually a double band) near 510 cmÐ1 in the range GRIFFITH, W.P. (1969): Raman studies on rock-forming miner- of 550Ð500 cmÐ1, in a region where no band has been als. I. Orthosilicates and cyclosilicates. J. Chem. Soc. (A) observed in the chrysotile spectrum by us or other re- 9, 1372-1377. search groups (Lewis et al. 1996, Bard et al. 1997, KLOPROGGE, J.T., FROST, R.L. & RINTOUL, L. (1999): Single Kloprogge et al. 1999). crystal Raman microscopic study of the asbestos mineral c) antigorite and lizardite are easily distinguishable chrysotile. Physical Chemistry, Chemical Physics 1, 2559- from one another on the basis of their s and as fre- 2564. quencies of the siliconÐoxygen bonds. Both these min- erals show bands in the range of 550Ð500 cmÐ1. The fact LAZAREV, A.N. (1972): Vibrational Spectra and Structure of that these bands are detected only in the spectra of Silicates. Consultant Bureau, New York, N.Y. lizardite and antigorite, where the normal development of the sheet of tetrahedra is structurally disturbed (in the LEWIS, I.R., CHAFFIN, N.C., GUNTER, M.E. & GRIFFITHS, P.R. first case by switching, in the second by tilting and by (1996): Vibrational spectroscopic studies of asbestos and comparison of suitability for remote analysis. Spectrochim. Al-for-Si substitution), also suggests that these bands Acta A 52, 315-328. should be assigned to deformed modes of the SiO4 and AlO4 tetrahedra. LOH, E. (1973): Optical vibrations in sheet silicates. Solid State Phys. 6, 1091-1104. 890 THE CANADIAN MINERALOGIST

MELLINI, M. (1982): The crystal structure of lizardite-1T: hy- Mineral Dusts (G.D. Guthrie Jr. & B.T. Mossman, eds.). drogen bonds and polytypism. Am. Mineral. 67, 587-598. Rev. Mineral. 28, 61-137.

______& VITI, C. (1994): Crystal structure of lizardite-1T WICKS, F.J. & O’HANLEY, D.S. (1988): Serpentine minerals: from Elba, Italy. Am. Mineral. 79, 1194-1198. structures and petrology. In Hydrous Phyllosilicates (Ex- clusive of ) (S.W. Bailey, ed.). Rev. Mineral. 19, 91- PASTERIS, J.D. & WOPENKA B. (1987): Use of a laser micro- 167. probe to trace geological reactions. Microbeam Analysis, 205-209. ______& WHITTAKER, E.J.W. (1975): A reappraisal of the structures of the serpentine minerals. Can. Mineral. 13, ROSASCO, G.J. & BLAHA, J.J. (1980): Raman microprobe spec- 227-243. tra and vibrational mode assignments of talc. Appl. Spectrosc. 34, 140-144.

VEBLEN, D.R. & WYLIE, A.G. (1993): Mineralogy of Received December 8, 2002, revised manuscript accepted and 1:1 layer silicates. In Health Effects of June 23, 2003.