American Mineralogist, Volume 73, pages487499, 1988 Experimental investigationof the effect of oxygen fugacity on ferric-ferrous ratios and unit-cell parametersof four natural clinoamphiboles Cpr,r,q.A. Cr-own,* Ronnnr K. Popp Department of Geology, Texas A&M University, CollegeStation, Texas 77843, U.S.A' SrBvnN J. FnIrz Department of Earth and Atmospheric Science,Purdue University, West Lafayette,Indiana 47907, U.S.A. Ansrucr To investigate the relation between oxygen fugacity, ferric-ferrous ratio (R), and unit- cell parameters,four natural clinoamphiboles (grunerite, tschermakitic hornblende, mag- nesio-hornblende,and a riebeckite-arfvedsonitesolid solution) were reacted at 650'C, I kbar at oxygen fugacitiesdefined by several solid oxygen-bufferassemblages. In order to produce more highly oxidized samples, heating in air at 700 'C was also carried out. Variation in R is accomplishedmainly by the oxidation-dehydrogenationequilibrium Fe2* + OH : Fe3* * 02 + t/z}{r, but the results suggestthat other mechanismsmay also be involved. All four amphiboles exhibited systematically higher ferric-ferrous ratios with increasing6, of equilibration. Equilibrium R values were achieved relatively rapidly and could be readily restored to onginal values by treatment at the appropriate buffer. In some cases,a metastableequi- librium of ferric-ferrous ratio was achieved before the amphibole decomposedto other Fe3*-bearingphases. Of the four amphiboles, grunerite is apparently the least able to accommodate Fe3* within its crystal structure and decomposesat relatively higher oxygen fugacities. The value of a sin B decreasessystematically uniformly as the Fe3*content of tscher- makitic hornblende, magnesio-hornblende,and riebeckite increases,reflecting increasing Fe3*in the octahedralcation sites.The variation in a sin B ofgrunerite is significantly less than for the other three amphiboles. The variation in b for the two hornblendessuggests that Fe3*produced by oxidation is not strongly ordered into the M(2) site. INrnolucrroN gen fugacity stability have been carried out on a wide variety compositions representing the major Amphiboles are the most chemically complex of the of bulk groups (e.g.,Gilbert et al., 1982). major rock-forming mineral groups. This complexity chemical of amphiboles results of the existing experimental studies have arises from the wide variation of unique structural sites The in chemical compositions and phys- available for cations. Fe is a major component in many shown that changes properties as a function ofoxygen fugacity and naturally occurring amphibole solid solutions, but the ical occur of Fe in an amphibole resultsini- factors that control the proportions of Fe3* and Fe2* in temperature.Oxidation formation of oxy-amphibole component any given amphibole are still very poorly understood tially from the quantitatively. by a dehydrogenationreaction ofthe type Two types of experimentshave been carried out in at- Fe2** OH- : Fe3*+ Ot + V2H2, (l) tempts to produce variation in the ferric-ferrous ratios in amphiboles: heating of amphiboles in air, and hydro- in which O' replacesOH- in the O(3) anion site of the thermal synthesisand/or treatment of amphibolesat oxy- amphibole. This process,first suggestedby the results of gen fugacitiesdefined by the standard solid oxygen buff- Barnes(1930), has also been proposed to be operative in ers. The earlier air-heating studies were carried out mainly other hydrous minerals such as the micas (e.9., Wones, on abestiform amphiboles, but more recently, nonasbes- 1963;Vedder and Wilkins, 1969).Most studiesin syn- tiform varieties have been so-treated (see Hawthorne, thetic systemsare complicated by the fact that 1000/oyields 198I , 1983, for reviews).Hydrothermal studiesto delim- of amphibole are not achieved. Therefore, changesin it physical properties and the pressure-temperature-oxy- either the proportions or compositions of the nonamphi- bole phasesmay causevariations in amphibole cornpo- * Present address: Department of Geological Sciences,The sitions. That is, changesobserved in physical properties University of Texasat Austin, Austin, Texas78712, U.S.A. may result from changesin amphibole bulk composition 0003-004x/88/0506-0487$02.00 487 488 CLOWE ET AL.: Fe3+/Fe'z+AND UNIT-CELL PARAMETERS OF CLINOAMPHIBOLES TABLE1. Compositionsof startingamphiboles rolite grade amphibolite (sample 73-318, Spear, 1982); a mag- nesio-hornblendefrom Strickland quarry, Portland, Vermont, Tschermakitic Magnesio- for which petrologic information has not been published (Na- Grunerite hornblende hornblende Riebeckite tional Museum of Natural History specimen A6169); and a rie- sio, 49.19 43.44 47 05 51.39 beckite-arfvedsonitesolid solution, hereafter referred to as rie- Tio, 0.03 0.60 0.31 0.78 from a granitic pegmatite(Scofield and Gilbert, 1982). Al,o3 0.51 13.82 10.24 0.80 beckite, FeO 44.30 12.53 12.54 21.67 Fe,O" N.D. 3.70 1.05 11.85 Separationtechniques MnO 0.66 0.25 0.74 0.71 MgO 3.21 9.80 12.24 0.25 In order to ensureclose to 100o/opurity in the starting mate- CaO O.32 11.57 11.93 0.15 rials, each amphibole sample was subjectedto a variety of sep- Na,O 0 04 1.45 1.13 8.87 aration techniques.All sampleswere disaggregated,washed in K,O N.D. 0.51 0.59 1.54 magnetic and heavy-liquid sepa- F 0.17 0.13 0.72 1.97 acid, subjectedto isodynamic cl o.12 N.D. N.D. N.D. ration techniques,and hand picked to produce 99-1000/opurity. Sum 98.55 97.80 98.54 99.98 Additional details ofthe separation techniques are given in Clowe si 7.94 6.41 6.87 800 (1987).The final grain sizesofsamples are reportedin Table 1. 'vAl 0.06 1.59 1.13 With the exceptionof a small proportion of crushedgrains, there vrAl 0.04 0.83 0.64 0.15 was no changein amphibole gmin size during runs. Ti 0.01 0.07 0.04 0.09 Fe,* 5.99 1.54 1.54 2.83 Fe3* 0.42 o.12 1.40 Phase identification and characterization Mn 009 0.04 009 0.09 Electron-microprobeanalysis of the starting materials and se- Mg 0.78 2.16 2.67 0.06 Ca 0.06 183 1.87 0.03 lected run products was carried out using the nnI- instrument in Na 0.10 0.41 0.32 2.68 the Department of GeologicalSciences at Virginia Tech, Blacks- K 0.09 0.11 0.30 burg, Virginia. The analytical proceduresand operating condi- cl 0.01 tions were essentiallyidentical to those describedby Solbergand F 0.09 0.06 0.33 0.98 o 23.00 23.00 23.00 23 00 Speer(l 982). Fe3+/Fe2+ratios were determined using a single-dissolution Grainsize (mm) Max 1.0 x 0.025 0.19 x 0.025 0.5 x 0.1 0.15 x 0.05 techniqueapplicable to small samplesizes (Fritz and Popp, 1985). Min 0.1 x 0.025 0.075 x 0.075 0.1 x 0.2 0.1 x 0.02 Sample aliquots ranging from 0.95 to 11.96 mg were digested with HF and H'SO. in the presenceof o-phenanthroline, after Nofe; N.D. : not detected. Chemicalcompositions (in wt% oxides and as atoms per 23 oxygens) and grain sizes of amphibolesused in experi- which Na-citrate and boric acid solutions were added. The so- ments. Chemicalcompositions were determinedby electron microprobe. lutions were then analyzedcolorimetrically for FeO. Total Fe as Ferric-ferrousratio determinedby wet-chemicalanalysis. Amphibole com- FerO, was determined by atornic absorption spectrophotometry positions were normalized to 23 oxygens because of unknown oxy-am- of the sampleswith the quartz blank used ohibolecontent. after further dilution in the FeO determination. Duplicate solutions were made for each sample in the colorimetric analysis for FeO, and each of those solutionswas diluted and analyzedtwice for FerO,. Ferric- as oxygenfugacity varies, rather than solely from changes ferrous ratios, reported as molar Fe3*/(Fe3*+ Fe'?*)and abbre- in the ferric-ferrous ratio. For most synthetic hydrother- viated R, are consideredto be preciseto +0.01 (Fritz and Popp, mal studies, precise chemical compositions and, more 1985). importantly, ferric-ferrous ratios of the amphiboles have A complication arisesin electron-microprobe analysesofphases not been determined routinely. that contain oxy-amphibole component becausetotal Fe is gen- reported as FeO and the formulas are normalized to 23 This study reports the results of experimentsdesigned erally oxygens.For an end-member grunerite, for example, this pro- to elucidate further ferric-ferrous equilibrium in amphi- cedureis equivalent to expressingthe chemical formula in terms boles.Four natural amphiboleswere annealed over a range of oxideswith one molecule of water presentin the stucture,i.e., ofoxygen fugacitiesat constanttemperature and pressure : to determine the variation in Fe3* content and in unit- Fe'SirO',(OH), TFeO'8SiOz'H:O. cell parameters.Natural amphiboleswere chosenin order The formulas are normalized to 23 oxygensbecause the electron to avoid the complication of less than 1000/oyields in microprobe cannot analyzefor H, and therefore, the amount of synthetic systemsand also to provide crystals large enough oxygen necessaryto form "hydrogen oxide" is not included in for single-crystalstructure refinements. Results that de- the normalized formula. In the caseof a totally dehydrogenated scribe the variation ofoptical properties as a function of grunerite,it is necessaryto normalize microprobe analysesto 24 23 oxygens,i.e., ferric-ferrous ratio and the results of crystal-structurere- rather than finements have been presentedelsewhere (Phillips et al., Fel"Fel-SirO.o: 5FeO FerOr'8SiOr. 1986;Cloweand Popp, 1987;Phillipset al., 1988). Thus, normalization of the formula of an amphibole containing a significant oxy-component to 23 oxygensresults in a total ox- ExpnnrvrnurAl METHoDS ide weight