American Mineralogist, Volume 74, pages 1097-1 105, 1989
Cation distribution in the octahedral sites of hornblendes
KuNuxr M.Lxruo, Klrsurosrrr Tovrru. Department Geology and Mineralogy, Kyoto University, Kyoto 606, Japan
Ansrn-Lcr The structure refinements of five calcic amphiboles with t4rAl > 0.5 have been carried out in order to characterizethe cation distributions in the octahedral sites in amphiboles formed at diferent temperatures.The location and atomic fractions of Al and Fe3+in the octahedral M(l), M(2), and M(3) sites were determined together with other constituent cations on the basis of the site refinement and the relations between mean bond length and mean ionic radius. The studied specimensare two metamorphic pargasites,a volcanic pargasite,a volcanic magnesio-hornblende,and a hastingsitefrom a skarn. In the metamorphic pargasitesand the hastingsite,Al, Fe3+,and Ti occupy only the M(2) site, and Mg prefers the M(2) site to the M(l) and M(3) sites (Mg-Fe'z+partitioning; K${t>t"rtzr: 0.30 between the M(l) and M(2) sites). On the other hand, the volcanic pargasite and hornblende indicate a more disordered cation distribution among the octahedral sites than that from metamorphic rocks and skarn.
INrnooucrrou tetrahedral sites in calcic amphiboles was demonstrated by Robinson et al. (1973) and Hawthorne and Grundy Calcic amphibole is an important rock-forming min- (r977). eral and occurs in extremely wide varieties of metamor- Finding the ordering of trivalent cations at the smaller phic, plutonic, and volcanic rocks. A chemicaldisconti- M(2) site in Bolivian crocidolite (magnesio-riebeckite) nuity in the calcic amphiboles exists betweenAl-rich and (Whittaker, 1949) and in glaucophane(Papike and Clark, Al-poor phases(e.g., Shido and Miyashiro, 1959; Ya- 1968) helped in understanding the crystal chemistry of maguchiet al., 1983).The Al-rich phasewith I4rAl> 0.5 the octahedralsites in hornblende;the smallerAl, Fe3*, is termed hornblende, subsumingthe appropriate names and Ti cations are confined to the M(2) site in horn- for amphiboledefined by Leake(1978), such as pargasite, blende,because it is usuallymuch smallerthan the M(l) tschermakite, and actinolitic hornblende. and M(3) sites.Robinson et al. (1973)demonstrated that The basic structural unit in calcic amphibole is a dou- the mean bond lengths of the octahedral siteswere a lin- ble chain of TOo tetrahedra extending parallel to the c ear function of the mean ionic radius of the constituent axis (Warren, 1929, 1930).The octahedralcation strips cations. This relationship was extended to the grand oc- are sandwichedbetween the double chains. There are two tahedralsite, which combinesthe M(l) sitewith the M(2) distinct tetrahedralsites, T(l) and T(2), which contain Si and M(3) sites,and to individual M(l), M(2), and M(3) and Al. The octahedral sites are subdivided into three sitesof the C2/mamphlbolesbyHawthorne (1978, 1981, crystallographicallynonequivalent sites, M( I ), M(2), and 1983). M(3), which accommodatevarious cations such as Mg, In the metamorphic cummingtonites, the Ca-poor am- Fe2+,Mn, Fe3*,Al, and Ti. The M(4) site accommodates phiboles with spacegroup C2/m or P2r/m, the Mg-Fe'z+ Ca, Na, Mg, and Fe2+.A large A site may be vacant or distributionbetween the M(4) site,and the M(l, 2, 3) site may containNa and K calions. was discussedby Hafner and Ghose(197l). The temper- Using data obtained from modern techniques,the ge- ature dependencyof the Mg-Fez+ distribution between ometry and chemistry of the cation sites in hornblendes the M(4) site and the M(1, 2, 3) site was suggestedand have been discussedin detail (Papikeet al., 1969;Kita- discussed by Mueller (1962) and Ghose and Weidner mura et al., 1975; Robinson et al., 1973; Hawthorne and (1972). The kinetics of the Mg-Fe2+exchange reaction Grundy, 1973;Hawthorne et al., 1980;Ungaretti et al., betweenthe M(4) site and the M(1, 2, 3) site in an an- l98l). The stereochemistryof the tetrahedral double thophyllite were discussedby Seifert and Virgo (1975). chains and the octahedral strips inthe C2/m amphibole On the contrary, the chemical complexity of horn- wassummarizedby Hawthorne (198 I, 1983). blende has obscured the temperature dependencyof its Papikeet al. (1969) observedthe preferenceofAl for cation distribution. In particular, the site preferenceofAl the T(l) siterelative to the T(2) sitein hornblendesbased (or other trivalent cations) in the octahedraat high tem- on the (T(IFO) and (T(2)-O) lengths.The correlation peraturehas not been studied. This study was undertaken betweenthe A1 content and the mean bond lensth of the to characterizethe distribution of cations among the oc- 0003-004x/89/09l 0-l 097$02.00 r097 1098 MAKINO AND TOMITA: OCTAHEDRAL CATION DISTRIBUTION IN HORNBLENDES
Taere 1, Chemicalcomoosition of the hornblendes mass, emplaced in Sanbagawaschists, in Ehime Prefec- ture, Japan.The associatedminerals are hypersthene,au- t-P gite, and plagioclase. The pargasite crystallized in the sio, 41.50 41.03 38.82 41.07 47.87 granulitefacies 'C and 5-10 kbar, as estimatedby Alros 14.40 17.34 9.07 17.94 7.52 [750 Tio, 1.66 0.42 0.58 0.18 1.35 Yokoyama (1980)l and sufferedSanbagawa metamor- FerO3 1.96 5.57 3.59 phism [epidote amphibolite facies; about 600 "C and 8- FeO 12.60 5.04 25.87 7.79 12.74 13 kbar, estimatedby Yokoyama (1980) and Takasu MnO o.12 0.0 0.25 0.14 0.43 as Mgo 11.88 14.39 1.91 13.37 15.30 (1984)1.The pargasiteis slightly poorer in Mg and Al CaO 10.65 12.71 11.79 12.05 10.95 than the Einstddingen sample describedbelow. Naro 3.22 1.21 2.79 1.83 1.29 K.o 0.32 3.19 0.92 0.31 0.44 HrO* 1.15 2.31 1.46 Einstiidingen pargasite (E-P) H.o- o.27 0.57 The pargasiteoccurs in a skarn intercalatedwith garnet cl 0.65 0.20 -Cl=O 0.15 0.05 biotite gneissesin an islet of Einstiidingen, Liitzow Holm Total 96.35 98.94 100.30 10030 97.89 Bay, EastAntarctica (Matsubaraand Motoyoshi, 1985). Cations on ihe basis of 23 oxygens The studied specimen was supplied by Dr. Matsubara. 5l 6.19 5.99 6.31 5.92 6.90 The skarn consistsof aluminous diopside, potassianpar- AI 1.81 2.01 1.69 2.08 1.10 gasite, and phlogopite. The pargasite is the product of > 8.00 8.00 8.00 8.00 8.00 t4l granulite-faciesmetamorphism. The metamorphic tem- AI 0.72 0.97 0.05 0.96 0.17 perature pressurewere estimated to be about 800- Ti 0.19 0.05 0.07 0.02 0.14 and Fe3* 0.05- 0.22 0.68 0.39 0.43' 850'C and 8-10 kbar on the basisofthe clinopyroxene- Fe2* 1.52 0.62 3.52 0.94 1.11 orthopyroxenegeothermometer (Wood and Banno, 1973; Mn o.02 0.0 0.03 0.03 0.05 geobarometer Mg 2.64 3.13 0.47 2.87 3.29 Wells, 1977)and the garnet-orthopyroxene > t6l 5.14 4.99 4.82 5.21 5.19 (Harley and Green, 1982)by Matsubaraand Motoyoshi 1.70 1.99 2.05 1.86 1.69 (1985). The pargasiteshows a high K content and Mg/ Na 093 0.35 0.88 0.51 0.36 (Mg + Fe'?*).It containshigh t6tAl(0.97 pfu). K 0.06 0.59 0.19 0.06 0.08 Note; l-P, lratsu paragasiteanalyzed by electron microprobe.E-P, Ein- Obira hastingsite (O-H) stodingen pargasite; combinationby electron microprobe and wet-chem- The hastingsitewas from a skarn in the Obira mine, ical analyses(Matsubara and Motoyoshi,1985). O-H, Obira hastingsite analyzedby wet-chemicalmethods (Matsumoto and Miyashita, 1 960). P-P, Oita Prefecture,Japan. Chemistry and occurrenceof the Parau pargasite analyzedby wet-chemicalmethods (Tomita, 1965). K-H, hastingsite were described in detail by Matsumoto and Kawanabehornblende analyzed by electron microprobe. (1960).The hastingsiteis fibrous. Associated 'The amountof Fe3*oer formulaunit in amohibolewas derivedfrom Miyashita the total occupancy of Na and K in the A site determined by refinement minerals are datolite and stilpnomelane, which crystal- (afterHawthorne et al., 1980). lized later than other skarn minerals (garnet,wollastonite, hedenbergite,actinolite). Judging from the assemblageof skarn and ore-forming minerals, the hastingsite crystal- lized during the hydrothermal stage(Matsumoto and Mi- tahedral sites in hornblendesformed at different temper- yashita, 1960).The hastingsitehas a high Fe2+/(Mg * atures. Fe'*) and contains little Al in the octahedral sites.
SpncrtunNs EXAMINED Parau pargasite(P-P) The following five specimenswere examined. Chemi- This pargasitewas found in andesitic agglomeratebe- cal compositions are given in Table 1, along with the longing to the Babeldaob agglomeratein Gapson, Parau methods of analysesfor each specimen. Island (Tayama, 1939).The Babeldaobagglomerate in- tercalateslimestone, shale,sandstone, and tufl suggesting Iratsu pargasite(I-P) submarine deposition. The andesitic agglomeratecon- Iratsu pargasiteis one of the constituent minerals of tains pargasite,clinopyroxene, and plagioclase.The crys- the basic granulite from the Iratsu epidote-amphibolite tallization temperature of the pargasitewas estimated to
TABLE2. Crystaldata for the hornblendes,space group C2lm E.P o-H a (A) 9.805(3) 9.900(2) 9.967(4) 9.834(4) 9.829(8) b (A) 17.96(1) 17.95(2) 18.269(2) 18.01(3) 18.06(1) c (A) 5.302(1) s.311(2) 5.347(1) 5.2e7(2) 5.304(1) pf) 104.93(2) 105.42(2) 104.97(2) 105.04(2) 104.70(2) v(A') 902.1(8) 910.5(9) 940.05(9) 906.0(4) 910.2(8) Size(mm) 0.15x0.10x0.25 0.15x0.10x0,30 0.15x0.10x0.50 0.30x 0.20x 0.50 0.20x 0.10x 0.30 F (weighted) 0.037 0.049 0.057 0.058 0.076 No.of F 2234 2967 1344 2701 2234 MAKINO AND TOMITA: OCTAHEDRAL CATION DISTRIBUTION IN HORNBLENDES 1099
'C, be about 900 on the basis of the phaserelations in were correctedby the semiempirical method of North et tonalite (Wyllie, 1977). The pargasiteshows a high Mg/ al. (1968).Intensities less than three times the standard (Mg + Fe2*)and is enrichedin t6rAl(0.96 pfu) (Tomita, deviation of the observed intensity were rejected in this 1965).Chemically, this pargasiteand the Einstddingen refinement. pargasiteare very similar to each other (Table l). The cell parameters,crystal sizes,number of intensity - measurements,and the final R : )llF.o, l lF"")l/ Kawanabehornblende (K-H) ) lF"o.I are presentedin Table 2. The magnesio-hornblendewas from phenocrystsof an- desitic agglomeratecollected in the Kawanabe area of Kagoshima Prefecture,Japan. The agglomeratecontains Refinementprocedure hornblende,clinopyroxene, and plagioclase.This mineral The full matrix least-squaresrefinements were caffied assemblagegives about 900'C as the crystallizationtem- out by using the computer program RFINEIv (Finger and perature of the hornblende (Wyllie, 1977). The horn- Price, 1975)revised by Horiuchi (personalcommunica- blende has a similar Mg/(Mg + Fe2*) to Einstddingen tion, 1981).The initial positional parametersand iso- and Parau pargasites(Table l). However the I6rAlcontent tropic temperaturefactors for the presentrefinements were is very low. taken from ferro-tschermakite(Hawthorne and Grundy, t973). Cornparisonof the five specimens Since the X-ray scatteringpower of Mg and Al atoms Neither exsolutiontexture nor chemicalzoning was ob- is indistinguishable,Mg and Al wereregarded as one group served in the five hornblendes by optical microscopy, with one scatteringfactor. Mg and Al atoms were summed X-ray photographs, and electron-microprobe analyses. to form Mg* in atomic fraction. For the same reason, The hornblendes show various Mg/(Fe,* + Mg) ratios Fe2+,Fe3+, and Ti were assumedto form a species,Fe*. and contain more than 0.70 Al, Fe3*, and Ti pfu in the The small amount of Mn in the five hornblendes was octahedral sites. They can be divided into two groups neglected in this refinement. Because of the chemical basedon their occurrences:One group consistsofObira complexity of hornblende, the following simplifying as- hastingsite,Iratsu pargasite,and Einstddingen pargasite; sumptionswere employed. Ca was constrainedto the M(4) the other group, Parau pargasite and Kawanabe horn- site, excesscations (Fe* and Mg*) for the octahedralsites blende.The three hornblendesin the former group, which in Iratsu pargasite,Parau pargasite,and Kawanabehorn- crystallized under metamorphic or hydrothermal condi- blende were assignedto the M(4) site, and the rest of the tions, are consideredto have cooled slowly. Thus, inter- site was filled with Na. The M(4) sites of Einst6dingen site cation migration may have continued somewhat be- pargasiteand Obira hastingsite were filled with Ca and low the crystallization temperature. On the other hand, Na. Residual Na and K were assignedto the A site. the two hornblendes in the latter group, which crystal- Assuming random distribution of Al and Si in the tet- lized at much higher temperature and may have been rahedral sitesand of Mg* and Fe* in the octahedral sites cooled rapidly after volcanic eruption, are expected to and/or the M(4) site, initial site occupancieswere deter- preserve intracrystalline equilibrium frozen at a higher mined from the chemical analyses. temperature than the former three hornblendes. There- During all cycles of refinements, the tetrahedral and fore, the former are consideredto be "low-temperature" octahedral site chemistry were constrained to agreewith hornblendes, and the latter "high-temperature" horn- the chemical analyses.Positional disorder of the A site blendes. on the mirror plane and/or along the twofold axis was read from the Fourier and difference-Fourier sections. Split-atom models (Papike et al., 1969 Hawthorne and ExpnnrlrEl'qrAl DETAILS Grundy, 1973) were used, which reducedthe R factor The crystals,usually {ll0} cleavageflakes elongated more than 1.00/0.The total amount of Na and K in the A along the c axis, were hand-picked. The crystal sizesare site of Iratsu pargasite and Kawanabe hornblende was shown in Table 2. The single-crystalX-ray photographs determined, assumingrandom distribution of Na and K showed diffraction symmetry of space group Cm, C2 or cations in the A site. The anisotropic temperaturefactors C2/m. Sincethere was no indication of a noncentric na- were refined for Iratsu, Einst6dingen, and Parau pargas- ture from the intensity distribution, the space group ites, for which the final weightedR factorsarc 3.1,4.1, C2/m was used in the structure refinements. and 4.70/0,respectively. In Iratsu pargasite,Parau pargas- The intensity data were collectedon a RTcAKUAFc-suD ite, and Kawanabehornblende, Mg* and Fe* at the M(4) automatic 4-circle diffractometer. The unit-cell parame- site are Mg and Fe2+,respectively. ters were determined by the least-squaresmethod from From the chemical analysesand X-ray refinements,Si 20 reflectionscollected on a 4-circle single-crystaldiffrac- and Al occupanciesin the tetrahedral sites and Mg* and tometer. Intensity data were measuredwith the 20-usscan Fe* occupanciesin the octahedral sites and/or the M(4) equi-inclination techniqueusing monochqomatizedMqKa site were determined. The final positional parameters, radiation and were gatheredto 20 : 80'for Obira has- isotropic temperature factors, and site occupanciesare tingsiteand to 20: 100"for the others.Absorption effects listed in Tables'3and 4, respectively. I 100 MAKINO AND TOMITA: OCTAHEDRAL CATION DISTRIBUTION IN HORNBLENDES
TABLE3. Positionalparameters and isotropictemperature factors for the hornblendes o-H r(1) x 0.2809(1) 0.2776(1) 0.2774(3) 0.2816(1) 0.281s(2) 0.08s8(1) 0.0863(1) 0.0846(2) 0.0852(1) 0.0848(1) z 0.3014(1 ) 0.304s(2) 0.2958(7) 0.3009(2) 0.2974(4) B 0.48(3) 0.44(3) 0.45(5) 0.39(2) o.54(2) r(2) x 0 2919(1) 0.2961(1) 0.291s(3) 0.2914(1 ) 0.2909(2) 0.1731(1) 0.1741(1) 0.1717(2) o.1725(1) o.1717(11 z 0.8133(2) 0.8162(2) 0.8089(7) 0 811 3(2) 0.8063(4) B 0.48(2) 0.47(31 o.s2(4) 0.3s(2) 0.54(2) M(1) x 0.0 0.0 0.0 0.0 0.0 0.0886(1) 0.0903(1) 0.091s(1) 0.0879(1) 0.0873(1) z u.5 0.5 0.5 0.5 0.5 B 0.78(2) 0.e2(3) 0.53(4) 0.67(2) 0.95(3) M(2) x 0.0 0.0 0.0 0.0 0.0 o.1772(1) 0.1768(1 ) 0.1787(1 ) 0.1768(1 ) 0.1774(11 z 0.0 0.0 0.0 0.0 0.0 E 0.4e(2) 0.63(3) 0.37(4) o.41(2) 0.74(3) M(3) x 0.0 0.0 0.0 0.0 0.0 v 0.0 0.0 0.0 0.0 0.0 z 0.0 0.0 0.0 0.0 0.0 B 0.62(2) o.87(2) 0.38(5) 0.47(21 0.87(4) M(4) x 0.0 0.0 0.0 00 0.0 0.2789(1) 0.2809(1) 0.2806(1) 0.2785(1) 0.2768(1) z 0.5 0.5 0.5 0.5 B 0.80(3) 0.67(5) 0.91(s) 0.91(1) 0.97(2) A(m) x 0.0474(251 0.0203(40) 0.0365(24) o.0292(27) 0.0 v 0.5 05 0.5 0.5 u,c z 0.101 4(50) 0.0426(80) 0.0472(60) 0.1245(61 ) 0.0 B 2.70(31) 1.73(31) 1.74(52) 2.48(30) 3.44(36) A(2) x 0.0 0.0 0.0 v 0.4774(7) 0.4756(18) 0.4793(5) z 0.0 0.0 0.0 B 2.16(56) 1.e3(55) 2.36(45) o(1) x 0.1070(2) 0.104s(2) 0.1048(7) 0.'t074(2') 0.11 1 0(4) v 0.0896(1) 0.0902(1) 0.0896(4) 0.0874(1) 0.0873(2) z 0.2145(5) 0.2148(5) 0.2116(1 6) 0.2160(5) 0.2186(8) B 0.95(3) 0.83(4) 0.55(10) 0.79(3) 0.76(6) o(2) x o.1204(2) 0.1204(2) 0.1227(71 0.1197(2) 0.1204(4) o.1742(11 0.1750(1 ) 0.1763(4) 0.1733(1 ) 0.1727(21 z 0.7366(4) 0.7375(5) 0.7297(17) 0.7316(4) 0.7279(81 B 0.77(3) 0.82(4) 0.76(13) 0.63(3) 0.77(6) o(3) x 0.11 01(3) 0.11 23(3) 0.1235(13) 0.1087(3) 0.11 14(6) 0.0 00 0.0 0.0 0.0 z o.7140(71 0.7151(7) 0.7140(30) o.7157(7) 0.7158(1 2) B 0.86(5) 0.76(5) 1.'t4(24) 0.84(4) 0.83(8) o(4) x 0.3684(2) 0.3687(2) 0.3661(8) 0.3672(2) 0.3673(4) 0.2501(1) 0.2506(1) 0.2478(4) 0.2497(11 0.2485(2) z 0.7895(s) 0.7934(5) 0.7893(18) 0.7875(s) 0.7896(9) B 0.92(3) 0.87(4) 1.04(12) 0.90(3) 1.04(6) o(s) x 0.3500(2) 0.3491(2) 0.3442(8) 0.3495(2) o.3473(4) v 0.1404(1 ) 0.13980) 0.1363(4) 0.1403(1 ) 0.1374(2) 2 0.1084(5) 0.11 29(5) 0.0991(1 8) 0.11 00(5) 0.1012(8) B 0.88(3) 0.86(4) 1.0s(1 2) 0.90(3) 0.90(6) o(6) x o.3427(2) 0.3418(2) 0.33s7(8) 0.3440(2) 0.3436(4) 0.1181(1) 0.11 86(1) 0.1198(4) 0.11 6s(1 ) 0.11 88(2) z 0.6072(s) 0.6108(5) 0.5920(19) 0.6070(5) 0.594s(8) B 1.01(3) 0.e2(4) 0.95(12) 0.es(3) 0.e2(6) o(7) x 0.3380(3) 0.3344(3) 0.3299(10) 0.3393(3) 0.3367(6) 0.0 0.0 0.0 0.0 0.0 z 0.2765(7) 0.2880(7) 0.2928(2s) 0.2745(71 0.2839(12) B 1.23(5) 1.20(6) 1.1 0(1 7) 1.08(5) 0.87(9)
Rosur,rs from the (T(1FO) and (T(2FO) bond lengths(Table 5). T(f) and T(2) tetrahedral sites Iratsu, Einstodingen,and Paraupargasites and Obira has- In the five hornblendes of this study, Al tends to oc- tingsite,rich in rarAl(about 1.8-2.0atoms pfu), have sim- cupy the T(l) site in preferenceto the T(2) site,asjudged ilar (T(l)-O) and also (T(2)-O) lengths.The (T(IFO) MAKINO AND TOMITA: OCTAHEDRAL CATION DISTRIBUTION IN HORNBLENDES I l0l
Tlgue 4, Site occuoanciesof the hornblendes TABLE5. Interatomicdistances (A) in the hornblendes o-H E.P o-H r(1) si 0.57(8) 0.50(9) 0.58 0.60(8) 0.77(101 r(1)-o(1)x1 1649(2) 1.653(2)1.664(7) 1.654(2) 1.624(4) AI 0.43 0.50 0.42 0.40 0.23 O(5)x1 1.681(3) 1.688(3)1.674(9) 1.675(3) 1.655(s) x1 1681(3) 1.683(3)1.673(9) 1.676(3) 1.657(5) r(2) si 0.98 0.99 1.00 0.89 0(6) 0.96 x1 1.656(3) 1.660(1)1.633(4) 1.6s4(2) 1.632(3) AI 0.02 0.01 0.0 0.11 0.04 O(7) Mean 1.667 1.671 1.661 1.665 1.642 M(1) Mg. 0.615(3) 0.822(410.06(1) 0.7s9(4) 0.676(4) r(2)-O(2)x 1 1.625(2) 1.625(2)'t.627(7\1.631(2) 1.620(4) Fe. 0.385 0.178 0.94 0.241 0.324 O(4)x 1 1.594(2) 1.597(2)1 593(8) 1.598(2) 1.591(4) M(2) Mg. 0.816(3) 0.838(3) 0.1s(1) 0.763(4) 0.673(4) o(s) x 1 1.630(3) 1.647(3)1.637(s) 1.642(3) 1.642(5) Fe' 0.184 0.162 0.8s 0.237 0.327 0(6)x 1 1.643(3) 1.654(3)1.662(10) 1.6s7(3) 1.655(5) Mean 1.623 1.631 1.630 1.632 1.627 M(3) Ms' 0.481(3) 0.780 0.13 0.764(6) 0.664(3) Fe- 0.519 0.220 0.87 0.236(6) 0.336 M(1)-O(1)x2 2.0s2(3) 2.049(3) 2.074(9) 2.0s0(3) 2.060(5) x2 2.138(3) 2.127(2)2153(71 2123(21 2.123(4) M(4) Ca 0.851 0.995 1.0 O(2) 0.930 0.85 x2 2.086(3) 2.121(2) 2.210(81 2.079(2) 2.088(4) Na 0.092 0.005 o(3) 0.07 Mean 2.092 2.099 2.146 2.084 2.090 Mg 0.0 0.011 0.04 F€F- 0.057 0.059 0.04 M(2)-o(1) x2 2.062(s) 2.040(2) 2.100(7) 2.095(3) 2.130(4) x2 2.048(2) 2.061(3)2.121(s) 2.069(3) 2.087(5) A Na+K 0.80(1) 0.94 0.99 0.57 0.32(2) o(21 O(41x2 1.s6s(3) 1.959(2)2.019(7) 1.ss0(2) 2.000(4) Notej Mg' : Mg + Al, f,g- : Pgz' + Fe3 + Ti. Mean 2.026 2.020 2.080 2.051 2.072 M(3)-O(1)x4 2.090(3) 2.091(2)2.107(7) 2.068(2) 2.093(4) O(3) x 2 2076(4) 2.103(4)2.1e7(8) 2.061(4) 2.075(71 Mean 2.085 2.095 2.137 2.066 2.087 length of Kawanabehornblende containing l. I pfu of t41Al is shorter than those of the pargasitesand hastingsiteof M(4)-o(2) x2 2 396(3) 2.414(2)2.421(7) 2.3s3(3) 2.382(4) O(4) x2 2 30s(3) 2.348(3) 2.347(10)2.30s(3) 2.2ss(s) this study. O(5) x 2 2.646(3) 2.618(212.755(8) 2.648(3) 2.741(41 0(6) x 2 2.564(3) 2.sss(21 2.552(8) 2.589(3) 2.560(4) The M(1), M(2), and M(3) octahedralsites Mean 2.478 2.485 2.519 2.484 2.496 The site refinements (Table 4) of the Iratsu pargasite A(m)-O(s) x2 3.002(15)3.03s(2) 3.059(18) 3.996(13)3.018(4) O(5) x 2 3.196(13)3.1ss(2) 3 193(19)3.231(11)3.018(4) and Obira hastingsite indicate that the M(2) site is en- 0(6) x 2 2.700(14)2.91 1(4) 2.981(20)2.592(12) 3.148(4) riched in Mg* over the M(l) and M(3) sites.The M(l), 0(6) x 2 3.s31(18)3.264(4) 3.461(22) 3.633(16)3.148(4) O(7) x 1 2.499(27) 2.521(6) 2.s17(35)2.s64(25) 2.463(7) M(2), and M(3) sitesin Einst6dingenand Paraupargas- O(7) x 1 3.250(25)3.487(6) 2.697(33) 3.100(22)3.730(6) ites as well as in Kawanabe hornblende contain almost O(7) x 1 2.483(24)2.519(6) 3.454(33) 2.463(22) 2.463(7) randomly distributed Mg* and Fe*. O(7) x 1 4.277(24)3.931(6) 4.013(31) 4.435(22) 3.730(6) Mean 3.115 3.099 3.161 3.180 3.089 Similar to other hornblendesreported by previous au- (M(2)-O) A(2)-O(5) x 2 2.699(10) 2.708(2612.753(8) thors, the length of these five hornblendes is O(5) x 2 3.412(12) 3.433(29)3.386(8) lessthan the (M(lFO) and (M(3FO) lengths(Table 5). 0(6) x 2 2.813(8) 2.923(21',)2.822(6) In particular, the differencesbetween the (M(2fO) length 0(6) x 2 3.403(10) 3.52e(26)3.345(7) O(7)x2 2.461(41 2.621(14)2.437(4) and the (M(IFO) and (M(3FO) lengthsare large(more o(7) x2 3.778(4\ 3.753(13) 3.776(6) than 0.06 A; in Iratsu pargasite,Einstodingen pargasite, Mean 3 095 3.161 3.087 and Obira hastingsitebut not so large (lessthan 0.03 A) in Parau pargasiteand Kawanabehornblende. Einstiidin- gen and Parau pargasitewith comparable Al and Fe3+in abe hornblende.The A(rn) sites on the mirror plane shift the octahedralsites are considerablydifferent in the mean less than about 0.3 A from the center position (2/m po- bond lengthof the M(2) site (2.020and 2.051 A, respec- sition) in the five hornblendes.The displacement of the tively). A(2) site along the twofold axis is also lessthan about 0.3 The thermal ellipsoids of the octahedral sitesin Iratsu, A from the center position (2/m). Einstddingen,and Parau pargasitesare similar in the ori- entation and the magnitudes(0.6-1.1) of their principal C,q.rroN occupANcy rN THE oCTAHEDRALsrrEs axes. Robinsonet al. (1973)and Hawthorne(1978, 1981, 1983)developed a relationshipbetween the mean bond The M(4) polyhedral site length and mean ionic radius [which is the mean of the The M(a) sites of Einstddingen pargasite and Obira effective radius of the constituent cations (Shannon and hastingsite are filled with Ca and Na cations. In Iratsu Prewitt, 1969,1970)lfor the octahedralsites in clinoam- and Parau pargasites,the M(4) sitesprefer Fe2+over Mg. phiboles. This relationship can be used to characterize The M(4) site in Kawanabe hornblende contains equal unknown site occupanciesor to test the obtained site oc- proportionsof Mg and Fe2+. cupancies.
The A site Relation betweenthe mean ionic radius and mean bond The positional disorder of the A site occurson the mir- length for the grand octahedral site ror plane in the five hornblendesand also along the two- The grand (M-O) length, which is the mean of the fold axis, except for Einstddingen pargasiteand Kawan- (M(lFO), (M(2)-O), and (M(3FO) lengths,increases I 102 MAKINO AND TOMITA: OCTAHEDRAL CATION DISTRIBUTION IN HORNBLENDES
"< 2to o o I I N =
065 0.70 0.75 (r*er)A (r) A Fig. 2. Plot of the (M(2FO) lengthversus the meanromc Fig. 1. Plot of thegrand mean bond length of the octahedral radiusof the M(2) site for the hornblendes.Al, Fe3+,and Ti site versusthe meanionic radiusof the constituentcations in cationsare assumed to be locatedin the M(2) site.In this rela- the hornblendes.All dataofthe studiedhornblendes are consis- tion, Obira hastingsiteand Iratsu and Einstijdingenpargasites tent with the relationshown by a solid line derivedfrom other are in agreementwith the correlationshown by a solid line C2/m amphiboles(Hawthorne, I 983). (Hawthorne,1983), but Paraupargasite and Kawanabehorn- blendeare not. Symbolsas in Fig. l. linearly with the mean ionic radius of the constituent amphiboles.He suggestedthat the relation holds good for cations in these octahedral sites in C2/m amphlboles the M(2) site, but not for the M(l) and M(3) sites.The (Hawthorne,1981, 1983).As is obvious in Figure I, the assignmentof Al, Fe3*, and Ti to the individual octahe- data of the five hornblendesin the present investigation dral sitesis examined here in the five hornblendesof this are consistentwith the correlation line. study, relying mainly on the correlation line for the M(2) site. Relation betweenthe mean ionic radius and mean bond Shortnessofthe (M(2)-O) lengthrelative to the (M(l)- length for the individual octahedral sites O) and (M(3)-O) lengthssuggests that smallerAl, Fe3+, Whittaker (1949) suggestedon accountof the signifi- and Ti cations are segregatedinto the M(2) site in the five cantly short (M(2)-O) lengthcompared to the (M(l)-O) hornblendes.Thus, Al, Fe3+,and Ti in the octahedral and (M(3)-O) lengths that trivalent cations in magnesio- sites were assumedto occupy the M(2) site, where the riebeckite are concentrated in the M(2) site. Using the residual fractions of Mg* and Fe* are Mg and Fe2+, re- samereasoning, other workers(Papike and Clark, 1968; spectively. The refined Mg* and Fe* occupanciesin the Robinsonet al., 19731-Hawthorne, 1978)showed that Al, M(l) and M(3) siteswere convertedto the Mg and Fe2* Fe3+,and Ti may be concentratedin the M(2) site in occupancies,respectively. The (M(2FO) length was plot- clinoamphiboles.Furthermore, Robinson et al. (1973) ted against the rnean ionic radius derived from the oc- demonstrateda linear relationshipbetween the mean bond cupancy of the M(2) site to be compared with the corre- length and the mean ionic radius in the octahedral sites lation line of the mean bond length versusthe mean ionic of hornblendes. Such a correlation line was developed radius (Fig. 2). Obira hastingsiteand Iratsu and Einst6- separatelyfor individual octahedral sites by Hawthorne dingen pargasitesare consistentwith the correlation line, (1978,1981, 1983), using a largerdata setof the C2/m whereasParau pargasiteand Kawanabe hornblende de- viate from the line of the M(2) site, i.e., the latter two hornblendes have observed (M(2FO) lengths that arc TABLE6. Observedand calculated mean bond lengths (A) of the larger than the lengths calculated from the mean ionic octahedralsites radius by using the correlation line of the M(2) site (Haw- Observed Calculated- Calculated-. thorne, 1983).In the same manner, the observedmean bond lengths and those calculatedby Hawthorne's equa- Parau pargasite M(1,3) 2.078 2.097 2.070 tions (Hawthorne,1983) also disagree with eachother for M(2) 2.051 1.998 2.041 the M(l) and M(3) sitesof the two hornblendes.The dif- Kawanabehornblende ferencesbetween the observedand calculatedmean bond M(1,3) 2.089 2.098 2.088 lengthsof the M(l), M(3), and M(2) sitesare shown in M(2) 2.072 2.052 2.068 Table 6 for Parau pargasiteand Kawanabe hornblende. Note.'The calculatedvalues were obtained by Hawthorne's equations Their observedgrand (M-O) lengths, however, are con- (Hawthorne,1983). The M(l,3) valuesare the averageof the M(1) and M(3)sites. sistentwith the calculatedones. Hence the positive value . Al, Fe3*,and Ti are assignedto the M(2)site. (M(2FO).b" - (M(2FO)",," ** of tends to be compensated After re-arrangement(see Table 7). by the negativevalues of the differencesfor the M(l) and MAKINO AND TOMITA: OCTAHEDRAL CATION DISTRIBUTION IN HORNBLENDES I r03
TABLE7, Octahedralsite occupanciesof the hornblendes 1.0
E-r M(1) Mg 0.615 0.822 0.06 0.62 u.oo o Fe,* 0.385 0.178 0.94 0.19 u -zo o9o AI 0.14 0.02 Fe3* 0.05 0.06 do r I M(2) Mg 0.456 0.353 0.13 0.49 u.oz X I Fe2* 0.067 0.027 045 0.13 0.14 t AI 0.360 0.485 0.02 0.27 0.05 Fe$ 0.024 0.110 0.36 0.11 o.12 Ti 0.093 0.025 0.04 0.07 X M(3) Mg 0.481 0.780 0.13 0.62 0.64 Fe2* 0.519 0.220 0.87 019 o.28 AI 0.14 o.02 X Fe3* 0.05 0.06 o o.5 1.0 /Vofe;Re-arrangements maintained refined occupancies(see Table 4) \./ ll ^Ms/r X"n * XrJ') u(t)
Fig. 3. Mg-Fe'?*distribution betweenthe M(l) and M(3) sites M(3) sites (Table 6). It follows that the foregoing assign- in the hornblendes.Solid line: KSltrvtzr: 1.0. + symbol: ferro- ment of Al, Fe3+,and Ti to the M(2) site is not adequate tschermakite(Hawthorne and Grundy, 1973).x symbol: pargas- in Parau pargasiteand Kawanabe hornblende. The dis- itic hornblende(Hawrhorne et al., 1980).Other symbolsas in agreement in the calculated and observed mean bond Fig. 1. lengths of the two hornblendes arises from overestima- tion of the amounts of smaller Al. Fe3+.and Ti cations in the M(2) site and from underestimation of the amounts hornblendesfrom volcanic rocks and "low-temperature" of smallercations in the M(l) and M(3) sites. ones from metamorphic rocks and skarn. The noncon- vergent cation order-disorder among the octahedral sites Re-arrangementof cation occupanciesin the will be discussedhere for the five hornblendesas well as octahedralsites for a ferro-tschermakite(Hawthorne and Grundy, 1973) From the abovediscussion, Al, Fe3+,and Ti appearto and a pargasitichornblende (Hawthorne et al., 1980).The be located only at the M(2) site in Obira hastingsiteand ferro-tschermakiteoccurred in the Frood mine, Sudbury, Iratsu and Einsttidingen pargasites.Their site occupan- Canada, and the pargasitic hornblende occurred in an ciesare listed in Table 7. amphibolite sequenceat the margin of Tallan Lake sill, The optimum arrangementsof Al, Fe3+,and Ti cations PeterboroughCounty, Ontario. The two hornblendesare within the octahedral sites for Parau pargasiteand Ka- consideredto belong to the "low-temperature" horn- wanabehornblende were estimated so that the mean ion- blendesbecause they are of metamorphic origin. ic radius of each octahedral site was close to the calcu- + lated one from the observed mean bond length by Mg-Fsu order-disorderamong the octahedralsites Hawthorne'sequations (Hawthorne, 1983) under the fol- Figure 3 is a plot of the Mg/(Mg * Fe'?+)of the M(l) lowing constraints. The refined Mg* and Fe* fractions of site against that of the M(3) site. The Mg-Fe'?+dis- every site were retained. Total contents of Al and Fe3+ tribution constantbetween the M(1) and M(3) sites is over the octahedral sites were constrained to agreewith given by the chemical analyses.Every octahedralsite was assumed /.-\ //-. \ to contain the same AllFe3* as the ratio obtained by rv.r: (*l (r) chemicalanalyses. The M(1) and M(3) siteswere treated K|1 {+*) / \^Fe2-/M(l)/ \^Fe2+/M{t} as one set. Ti, which is small in amount in the octahedral site,was assigned to the M(2) site.The cationoccupancies whereX is the atomic fraction in the sites. of the octahedral sites determined in this way are pre- K$rtruo is about 0.6 for "low-temperature" horn- sentedin Table 7. blendes except for Obira hastingsite and is nearly unity Thermal ellipsoids of Iratsu, Einstddingen, and Parau for "high-temperature" hornblendes. pargasitesshow no evidenceofpositional disorderin the The Mg-Fe'z+distribution between the M(2) and the octahedral sites. Therefore, there is no problem in using M(1) sitesis shownin Figure4. The distribution constant the relations between the mean bond length and ionic betweenthe M(2) and M(l) sitesis given by radius for the examinedhornblendes. /-.\ t/-.^""\ \ C.l,rIoN ORDER-DrsoRDERAMONG THE OCTAHEDRAL Kftt'vr:,: (+\ / [ (2) SITES IN HORNBLENDE tXoe-1r,,, / \Xr.,'/',r,' On the basis oftheir occurrences,the five hornblendes where X is the atomic fraction in the sites. can be divided into two groups: "high-temperature" "Low-temperature" hornblendes plot near the curve I 104 MAKINO AND TOMITA:OCTAHEDRAL CATION DISTRIBUTION IN HORNBLENDES
but the "high-temperature" hornblendesof this study are from agglomeratedeposited on the sea floor. The rapid quenching in seawater may be responsiblefor the pres- s! ervation ofthe random distribution ofcations in "high- + temperature" hornblendes. &o L Similar disordering of divalent and trivalent cations in X o.s the octahedral sites of experimentally oxidized tsherma- + kitic hornblende and riebeckite was reported by Phillips et al. (1988).Al and Fe3+migrate from the M(2) site to X the M(l) and M(3) sitesof the riebeckitetO(3) : 0.1O(OH) f + 0.90(O)lheated in air. Paraupargasite [O(3) : 0.7I(OH) X + 0.29(0)1, in which Al and Fe3+ are distributed over the octahedralsites, is different from the riebeckite in the o o.s 1.o degreeof oxidation. Thus, cation disordering in the oc- Xrn/(Xrn+ XrJ')uo) tahedral site of amphibole is consideredto dependon the crystallization temperature as well as on the oxidation Fig. 4. Mg-Fgu+ distribution betweenthe M(2) and M(l) sites degreeof hornblende.Which is the most important factor in the hornblendes. Solid line: Kytt>utzr : 1.0. Broken line: in the cation disordering is not known. Further exami- K${tru12;: 0.3. Symbolsas in Figs. I and 3. nations will be neededto make it clear.
CoNcr-usron for K$tr>trrzr: 0.3. "High-temperature" hornblendes,es- The structure refinements of the "high- and low-tem- pecially Parau pargasite,plot near the curve for KY(IFM(2)perature" hornblendeshave been carried out. The mean : 1.0 and thus indicate little preferencefor ordering of bond lengths and the calculatedmean ionic radius of the Mg and Fe2+between the M(l) and M(2) sites.So "low- octahedralsites in the "low-temperature" hornblendesare temperature" hornblendesshow the Mg/Fe2+to be M(2) consistent with the equation proposed by Hawthorne >> M(l) > M(3). At high temperature, the octahedral (I983). However the "high-temperature" hornblendesare sites of hornblendesmay become indistinguishable from inconsistent with that relation. The inconsistencyarises each other for the Mg and Fe2* cations. mainly from the assignmentof Al and Fe3+only to the The Mg-Fe'z+partitioning coefficientbetween the M(4) M(2) site, thereby suggestingthat the conventional meth- and M(1, 2, 3) of metamorphiccummingtonites is 0.02- od of assignment is unsuitable for the "high-tempera- 0.08 (Ghoseand Weidner,1972). Geometricallythe M(2), ture" hornblendes.Therefore, cation distributions among M(l), and M(3) sitesof clinoamphiboleare more or less the octahedralsites were estimated so as to be consistent similar sites,but the M(4) site differs much from the oc- with the refined mean bond lengths, on the basis of the tahedral sites in size and coordination number. Accord- relation betweenmean ionic radius and mean bond length. ingly, it is reasonablethat the K$
mingtonite, (Mg,Fe),SirOrr(OH)r:A new geothermometer.Earth and distribution in three homblendes.American Journal of Science,273A, PlanetaryScience Letters, 16, 346-354. 522-535. Hafner,S S, and Ghose,S. (1971)Iron and magnesiumdistnbution in Seifert,F, and Virgo, D. (1975) Kinetics ofthe Fe'?*-Mg,order-disorder cummingtonites (Fe,Mg)TSLOdOH)r. Zeitschrift fiir Kristallographie, reactionin anthophyllites:Quantitative cooling rates. Science, 1 88, 1107- 133.301-326. I 109. Harley, S.L., and Green, D H (1982) Gamet-orthopyroxenebarometry Shannon,R.D, and Prewitt,C.T. (1969)Effective ionic radii in oxides for granulitesand peridotites. Nature, 300, 697-701. and fluorides. Acta Crystallographica,82 5, 925 -9 46. Hawhorne, F.C. (1978)The crystalchemistry of the amphiboles.VI. The -/1970\ Revised values of effective ionic radii Acta Crystallo$a- stereo chemistry of the octahedral strip. Canadian Mineralogrst, 16, phica,826, 1046-1048. 37-52 Shido, F., and Miyashiro, A (1959) Hornblendes of basic metamorphic -(1981) Crystal chemistry of the amphiboles.Mineralogical Society rocks.Journal ofFaculty ofScience,University ofTokyo, 12, 85-102. of America Reviews in Mineralogy, 9A, l-102. Takasu, A. (1984) Progradeand retrogradeeclogite in Sanbagawameta- -(1983) The crystal chemistry of the amphiboles. Canadian Min- morphic belt in Besshi district, Japan. Journal of Petrology, 25, 619- eralogist,21,173480. 643. Hawthome, F.C., and Grundy, H.D (1973) The crystal chemistry of the Tayama, R. (1939) Geology and mineral resourcesin Babeldaob(ofPar- amphiboles.I. Refinementof the crystal structureof ferrotschermakite au) islands.Bulletin ofthe Tropical Industry Institute ofParau, South Mineralogical Magazine, 39, 3648. SeaIslands, Japan, 3, 1-19 (in Japanese) -(1977) The crystal chemistry of the amphiboles. III. Refinement Tomita, K. (1965) Studieson oxyhomblende. Memoirs of the Collegeof of the crystal stmcture of a subsilicic hastingsite.Mineralogical Maga- Science,University ofKyoto, SeriesB, 32,47-87 zine.4l.43-51. Ungaretti,L, Smith, D.C., and Rossi,G. (1981) Crystal-chemistryby Hawthorne,FC., Griep, J.L., and Cunis, L. (1980)A three-amphibole X-ray structurerefinement and electron microprobe analysisofa senes assemblagefrom the Tallan Lake sill, PeterboroughCounty, Ontario. ofsodic-calcic to alkali-amphibolesfrom the NybO eclogitepod, Nor- Canadian Mineralogist, 18, 27 5-284. way. Bulletin de la Soci6t6Frangaise de Min6ralogie et de Cristallo$a- Kitamura, M, Tokonami,M., and Morimoto, N. (1975)Distnbution of phie, 104, 400412. titanium atoms in oxy-kaersutite.Contributions to Mineralogy and Pe- Warren, B E. (1929) The structure of tremolite H'Ca'Mgr(SiOJr. Zeit- trology,51, 167-172. schrift fiir Kristallographie, 72, 42-57. Leake,B.E. (1978) Nomenclatureof amphiboles.American Mineralogist, -(1930) The crystal structure and chemical composition of the 63, t023-r052. monoclinic amphiboles.Zeitschrift fiir Kristallo$aphte, 72, 493-5 17. Matsubara, S., and Motoyoshi, Y. (1985) Potassiumpargasite from Ein- Wells, P.R.A. (1977) Pyroxenethermometry in simple and complex sys- stddingen, Liitzow-Holm Bay, East Antarctica. Mineralogical Maga- tems Contributions to Mineralogy and Petrology,62,129-139. zlne. 49.703-707 whittaker, E.J.w. (1949)The structureof Bolivian crocidolite.Acta Crys- Matsumoto,Y., and Miyashita, M (1960)Ferrohastingsite from the Obira tallographica,2, 312-317 mine, Kyushu, Japan. Journal of the Mineralogical Society of Japan, Wood, 8.J., and Banno, S. (1973) Garnet-orthopyroxeneand orthopyrox- 4,372-382 (in Japanesewith English abstract). ene-clinopyroxenerelationships in simple and complex systems.Con- Mueller, R.F. (1962) Energeticsof certain silicate solid solutions. Geo- tributionsto Mineralogyand Petrology,42,109-124. chimicaet CosmochimicaActa, 26, 581-598. Wyllie, PJ. (1977) Crustal anatexis: An experimental review. Tectono- North, A.C.T.,Phillips, D C., and Mathews,F.S. (1968) A semi-empirical physics,43, 4 1-7I . method of absorption correction Acta Crystallographica,A24, 351- Yamaguchi,Y., Shibakusa,H., and Tomita, K. (1983)Exsolution of cum- 3s9. mingtonite, actinolite and sodic amphiboles in hornblende in high- Papike,J.J , and Clark, J.R. (1968) The crystal structure and cation dis- pressuremetamorphism. Nature, 304, 257-259 tnbution of glaucophane.American Mineralogrst, 53, I I 56-1 I 73. Yokoyama, K. (1980)Nikubuchi peridotite body in the Sanbagawameta- Papike,J.J., Ross,M., and Clark, J.R. (1969) Crystal-chemicalcharacter- morphic belt; thermal history of the "Al-pyroxene-rich suite" perido- ization of clinoamphiboles based on five new structure refinements. tite body in high pressuremetamorphic terrain. Contributions to Min- Mineralogical SocietyofAmerica SpecialPaper 2, 117-136. eralogyand Petrology,73, l-13. Phillips,M.W., Popp,R.K., and Clowe,C.A. (1988)Structure adjustmenr accompanying oxidation-dehydrogenationin amphiboles. American Mineralogist,73, 500-506 Mer.ruscnrrr RECEIVEDAucusr 18, 1988 Robinson,K., Gibbs, G.V., Ribbe, P.H, and Hall, M R. (1973)Cation Me.ruscnrpr ACcEPTEDMev 15, 1989