The Hydrous Components in Garnets: Grossular-Hydrogrossular Gnoncr R

American Mineralogist, Volume 76, pages I 153-I 164, l99l

The hydrous components in garnets: Grossular-hydrogrossular

Gnoncr R. RossrrlN, Rocnn D. Arxts* Division of Geological and Planetary Science,California Institute of Technology, Pasadena,California 9 I 125, U.S.A.

Ansrucr Grossular garnets from 33 localities were examined for indications of OH or HrO in their infrared spectrum. All contained OH-. Classical hydrogrossular with more than 5 wto/oHrO displays systematicspectroscopic behavior consistentwith the hydrogarnet substitution consisting oftwo absorption bands at 3598 and 3662 cm '. These spectroscopic characteristicswere generallynot observedin other glossular samples.Instead, 20 distinct absorption bands have been identified in the spectra of common grossular,occurring in groups of four to ten bands. Both the number and intensities of thesebands show a large variation, which does not correspond with the garnet's composition. Seven classesof spectra were identified in the OH region based on the position of the most intense absorption band. The spectroscopicdata suggestthat in addition to the tetrahedral site, OH groups exist in multiple other environments. The OH content of grossulargarnets can be obtained from infrared spectra using the equation HrO wto/o: 0.0000786 x integrated absorbanceper cm in the OH region near 3600 cm '. The OH content of macroscopic grossularcrystals (expressedas weight percent HrO) rangedfrom 12.80/odown to lessthan 0.0050/0.Macroscopic grossulartypically contains less than 0.3 wto/oHrO. Grossular from rodingites and low-temperature alteration vugs contained much more than that from skarn or contact metamorphic environments.

INrnooucrroN position. Becausethe hydrogarnet substitution should be prevalent grossular pyrope-almandine The hydrogamet substitution, (OH)o : SiO", has been more in than in establishedin both synthetic and natural garnets.Gros- garnets(Lager et al., 1989),we have examined a number sular analyseshave been reported with over l0o/oHrO, of grossularsamples from a variety of localities to deter- and the structure of hydrogrossularhas been determined mine if the hydrogrossularsubstitution was,indeed, com- quantify with both X-ray and neutron diffraction (Cohen-Addad monplace and to attempt to the extent of this etal.,19671,Foreman, 1968; Basso et al., 1983;Sacerdoti substitution. previously and Passaglia,1985; I-ager etal., 1987a,1989). Most hy- Infrared spectra have often been used to drogrossular with high OH contents consists of crystals characteize synthetic hydrogrossular samples. These present typically no more than a few tens of micrometers in size, spectraconfirm that OH is in hydrogrossularand often with cores of grossularor intimately admixed with should be a standard of comparison for the identification vesuvianite(Zabinski, 1966; Rinaldi and Passaglia,1989). of OH in natural samples.However, when Passagliaand (1984) Although such high HrO content garnetsare rare, previ- Rinaldi describedkatoite, a mineral in the hydro- ous infrared spectraofmacroscopic garnetshave indicat- grossularseries with lessthan one-third ofthe tetrahedral presented ed that a minor OH component is presentin a variety of sites occupiedby Si, they an infrared spectrum garnets(Wilkins and Sabine, 1973). We have previously that bearslittle resemblanceto previously published gros- reported the results of a survey of the infrared spectros- sular spectraand differs in detail from the spectraofsyn- copy of a number of pyrope-almandine garnets (Aines thetic hydrogrossular.Consequently, it seemedappropri- properties and Rossman, 1985a, 1985b) in which nearly all con- ate first to examine carefully the spectroscopic tained OH. The position of the bands in the spectrum of of garnetswith known hydrogrossularsubstitution in or- these garnets varied systematicallywith the garnet com- der to compare the spectroscopicproperties of natural position, but the amount of OH varied for garnetsfrom hydrogrossular samplesto their synthetic counterparts and different localities. to the common low OH-content grossularsamples. Such previously per- This paper is primarily concernedwith the OH content a comparison has not been critically of common, macroscopic garnets of the grossular com- formed. Expnnrprnx.rAl, DETAILS

* Presentaddress: Department ofEarth Science,Lawrence Liv- Samples were obtained primarily from museums and ermore National l,aboratories, Livermore, California 94550, private collections.Details of samples,localities, and ma- U.S.A. jor element compositions are presentedin Tables I and 0003-004x/91/0708-1 I 53$02.00 I 153 1154 ROSSMAN AND AINES: HYDROUS COMPONENTS IN GARNETS

TABLE1, Grossulargarnet samples

Variety Locality Class

42 orange-brown [McFallmine?], calcite skarn, Ramona, CA, U.S.A. 2 4.6 52 orange-brown lvesper Peak?], WA, U.S.A. D 2.7 53 pale orange Asbestos, Quebec, Canada 3,4 4.8 227 green tsavorite Campbell Bridges mine, Tsavo National Park, Kenya 2b 1.8 229 green tsavorite stream gravel, Kenya 2b 3.7 771 colorless Meralini Hills,Tanzania 2b 3.4 936 orange Bric Camula, Cogoleto, Liguria, ltaly 6 22.1 937 dark red-orange Passo del Faiallo,Genova Province,Liguria, ltaly b 13.7 938 dark red-brown Ossola Valley, ltaly 3 1.5 941 pink jade Butfelsfontein,Rustenberg, S. Africa 4 89.7 946 brown-red Auerbach, Germany 7 4.8 1037 orange Dos Cabezas, lmperial County, CA, U.S.A. 2b 1.4 1038 green Jeffrey Mine, Asbestos, Quebec, Canada 5 1.5 1042 orange-brown VesperPeak, WA, U.S.A. 5 2.8 1051 pare orange lBelvidereMtn?], EdenMills, W, U.S.A. 3'| 1.0 1058 synthetic Lager et al. (1989) powoer 1059 syntheticLager et al. (1987a) 1 powder 1113 brown-orange North Hill,Riverside County, CA, U.S.A. 7 1.7 1122 orange-brown lOommercialQuarry?], Crestmore,CA, U.S.A. 7 2.7 1124 colorless Chihuahua,Mexico 3 2.1 1125 colorless Lake Jaco, Mexico 3,7 0.7 1129 pale orange BelvidereMountain, VT, U.S.A. 3 3.2 1131 black zone Lake Jaco, Mexico 3 o.2 1198 green jade Transvaal,S. Africa 4 68.3 1326 orange Ala Valley, Piedmont, ltaly c 7.2 1327 orange Sciarborosca,Ligura, ltaly c 6.9 't329 massrvevern CommercialQuarry, Crestmore,CA, U.S.A. 1 160.9 1357 orange-reo Bric Canizzi, Liguria, ltaly 5 20.5 1358 colorless CommercialQuarry, Crestmore,CA, U.S.A. 1 138.5 1359 red-orange lron gabbro metarodingite,Gruppo di Voltri, ltaly 3.4 1360 orange Basaltic metarodingite,Voltri Massif, ltaly 15.9 1364 massrvegray Maungatapu Survey District, Nelson, New Zealand 4 90.0 1409 red-brown calcite,skarn, Saline Valley, Darwin, CA, U.S.A. 2,2b 3.C 1411 birefringent skarn. Munam. North Korea o 1.2 1412 brown Mul-Kummine, South Korea 1.0 1413 pale green VilyiRiver, Siberia, U.S.S.R. 0.0 1419 pale orange Minot Ledge,Minot, ME, U.S.A. 2b 3.2 1420 brown Rauris, Salzburg,Austria 2 8.3 't422 brown Wakefield,Ontario, Canada 2b 0.3 1423 red-brown MountainBeauty mine, Oak Grove,CA, U.S.A. 0.6 1424 orange GarnetQueen mine?, Santa Rosa Mtns, CA, U.S.A. 7 3.1 1427 massive yellow Mavora Lakes, Otago, South lsland, New Zealand 4 1429 red-brown EssexCounty, NY, U.S.A. 0.6 1430 oragne EdenMills, VT, U.S.A. 3 1,1 't444 katoite Pietramassa,Viterbo, ltaly 1 powder

Nofe.'Question n''ark denotes additional locality information that was not part of the documentationprovided with the sample but that we added based on our inspectionof the sample.Absorbance per mm refers to the strongest band in the OH region. Integratedintensity in the OH regiongenerally closely follows this parameter. Class refers to the class of spectra in the OH region discussed in text. Multiple numbers mean that different crystals from the same locality have different behavior or that different zones of the crystal show different behavior.

2. The infrared spectra of the natural crystals were ob- both a MAC5 (Bence-Albeecorrections) and JEOL 733 tained from doubly polished single crystals. Exceptions instruments (CITZAF corrections). were the fine-grained,massive hydrogarnet samples(nos. Batches of both CarAlr(OoHo)3and Ca3Al, (Sioo)rr8 1329, 1364, ar;'d 1427), which were studied as doubly (OoHo)., were provided by G. A. Lager. CarAlr(OuDo), polished polycrystalline slabs,and katoite (no. 1444) and that contained some residual H and was used in the neu- the fine-grained powders of synthetic hydrogarnet sam- tron structural refinement ofLager et al. (1987a)was also ples,which were studied as KBr pellets. Details of sample examined. The position and full width at half height of preparation are the sameas describedin Aines and Ross- the band in the OH region of its spectrum are essentially man (1985a). Details of spectrophotometermeasure- identical to that of CarAlr(OoHo)r. ments are generally the same as described in Aines and Rossman with the exception that most spectra were ac- Rnsur,rs quired with a Nicolet 60SX FTIR with an InSb detector at 2 cm-t resolution. This instrument allowed regions in Synthetic hydrogrossular the sample as small as 25 p.m in diameter to be studied Hydrogarnet ofthe composition Ca.Alr(OH)', has been selectively. Garnet analyseswere performed by electron synthesizedby a number of authors who have also re- microprobe as described in Aines and Rossman using ported its infrared spectrum (Zabinski, 1966; Cohen-Ad- ROSSMAN AND AINES: HYDROUS COMPONENTS IN GARNETS I 155

TrsLe 2, Grossular analyses: formula proportions

lnfrared vvt"/" Number Ca Mg Fe,* Mn FeP- Cr Ti si HrO 42 2.89 0.00 0.08 0.01 1.89 0.11 0.00 na 0.00 3.02 na 0.10 52 3.03 0.00 0.00 0.05 1.67 0.30 0.00 na 0.02 3.00 na 0.14 53 2.93 0.00 0.03 0.04 1.96 0.06 0.00 na 0.00 2.98 na 0.21-0.38 227 2.99 0.05 0.00 0.07 1.78 0.01 0.01 0.07 0.02 2.96 na 0.09 229 2.99 0.05 0.00 0.05 1.84 0.00 0.02 na 0.02 3.01 na 0.15 771 2.95 0.04 0.00 0.01 1.97 0.01 0.00 0.01 0.02 2.99 na 0.18 936 2.92 0.00 0.08 0.00 1.05 0.84 0.00 0.00 0.11 2.99 na 1.26 937 1.01 938 2.78 0.02 0.03 0.18 1.60 0.37 0.00 0.00 0.03 2.96 na 0.08 941 5.00 946 2.85 0.01 0.13 0.01 1.66 0.25 0.00 0.01 0.01 2.97 na 0.26 1037 2.90 0.01 0.08 0.01 1.78 0.20 0.00 0.01 0.01 3.01 na 0.06 1038 2.96 0.00 0.05 0.07 1.93 0.04 0.o2 na 0.01 3.01 na 0.08-0.19 1042 0.11-0.13 1051 2.58 0.00 0.01 0.01 1.33 0.68 0.00 na 0.00 2.98 na 0.02-0.04 1058a 3.00 na na na 2.00 na na na na 2.28 2.88 1059A 3.00 na na na 2.00 na na na na 0.00 12.00 1113 2.86 0.01 0.11 0.01 1.52 0.43 0.00 na 0.05 2.99 na 0.09 1122 2.91 0.00 0.06 0.02 1.55 0.44 0.00 na 0.02 2.99 na 0.13 1124 2.95 0.06 0.00 0.01 1.93 0.05 0.00 0.00 0.00 3.02 na 0.11 11258 2.98 0.05 0.00 0.01 1.70 0.24 0.00 na 0.02 2.99 na 0.03 1125c 3.01 0.12 0.00 0.01 1.29 0.48 0.00 na 0.28 2.80 na 0.u 1't29 2.93 0.00 0.06 0.01 1.85 0.'t7 0.00 0.00 0.00 2.97 0.17 1131 2.95 0.09 0.00 0.01 1.76 0.14 0.00 0.00 0.05 3.00 na 0.01 1198 3.59 1326 2.86 0.04 0.09 0.01 1.74 o.25 0.00 0.00 0.02 3.00 na 0.37 1327 2.74 0.03 0.20 0.03 1.66 0.29 0.00 0.00 0.05 2.99 na 0.47 1329D 3.00 na na na 2.OO na na na na 1.64 5.43 13570 2.67 0.16 0.14 0.o2 0.99 0.80 0.00 na 0.21 2.84 0.63 1358a o 3.00 na na na 2.00 na na na na 1.53 5.88 1359 2.05 0.03 0.90 0.02 1.80 0.16 0.00 0.01 0.03 2.96 na 0.28 1360 2.87 0.02 0.09 0.02 1.47 0.48 0.00 0.00 0.05 2.88 na 0.85 1409E 2.49 0.01 0.09 0.41 1.41 0.57 0.00 0.01 0.01 2.97 na 0.20 1409' 2.88 0.00 0.04 0.08 1.54 0.44 0.00 0.00 0.02 3.00 na 0.17 1411 2.92 0.01 0.03 0.04 1.27 0.68 0.00 0.00 0.0s 2.97 na 0.08 1412^ 2.85 0.15 2.00 3.00 0.10 1413 2.91 0.07 0.o2 0.01 1.71 0.27 0.00 0.00 0.02 2.98 na 0.0003 1419 2.85 0.01 o.12 0.01 1.86 011 0.00 0.00 0.03 2.96 na 0.16 1420 2.85 0.01 0.12 0.03 1,60 0.38 0.00 0.00 0.o2 2.95 na 0.26 1422 2.91 0.08 0.00 0.01 1.87 0.08 0.00 0.00 0.04 2.96 na 0.02 1423 2:t2 0.01 0.59 0.28 1.88 0.10 0.00 0.01 0.01 2.99 na 0.06 1424 2.83 0.00 0.13 0.03 1.72 0.24 0.00 0.00 0.04 2.96 na 0.17 1429 279 0.02 0.18 0.01 1.70 0.28 0.00 0.00 o.o2 2.91 na 0.05 1430 2.86 0.00 0.12 0.02 1.89 0.10 0.00 0.00 0.00 3.00 na 0.06 1444^ 2.93 0.00 1.97 0.64 9.44 Nofe: The abbreviationna : not analyzed.All other analysesare electron microprobe analysesnormalized to five cations in the combined X and Y sites, where Ti4* is assignedto the Y site, Fep*is used to fill the Y site, and the remainingFe is assignedto Fe,* in the X site. Infraredwty" HrO values are from the calibrationdiscussed in the text. ALiterature values: no. 1058 Lageret al. (1989),no. 1059 Lageret al. (1987a),no. 1357 Bassoet al. (1981),no. 1358 Bassoet al. (198i|),no.1412 Hiraiand Nakazawa(1986), no. 1444 Sacerdotiand Passagtia(1985). s Colorless rim. c Brown inner region. o From X-ray cell parameters. ERim. FInterior.

dad et al., 19671,Harmon et al., 1982; Kobayashi and fully dried in a desiccatorand made into a KBr pellet that Shoji, I 983, I 984). The spectraofthese syntheticsprovide was itself redried after preparation. The spectrum of the a basis of comparison to the natural garnets. The pub- pellet, dominated by the major peak at 3662 cm-', shows lished spectra are dominated by a major band at 3660 essentiallyno indication of the molecular HrO contribu- cm ', w'ith a broader side absorption decreasingin inten- tionat3420 cm-' and lacksthe extensivetail toward lower sity to about 3000 cm '. The magnitude of the broad energies(Fig. l). absorption varies in the published spectra and contains Also relevant to the discussionofnatural, hydrous sil- inflections near 3420 cm-r, which are characteristic of icate garnetsare the synthetic garnets,which have partial weakly bound HrO, and frequently present as adsorbed substitutionof SiOoby (OH)". Hsu (1980),Cohen-Addad HrO. To avoid adsorbedH2O, our Ca3Alr(OH),,was care- et al. (1967),and Kobayashiand Shoji (1983, 1984)pre- I 156 ROSSMAN AND AINES: HYDROUS COMPONENTS IN GARNETS

Hydrogrossular OH band intensities

Hydnognossulan (!n

):v

0 02 04 (u Proportionof OoHoin tetrahedralsite (J t.u c Fig.2. Correlationbetween the proportionof the OH inten- IU sity in the 3662 cm-t band vs. the proportionof OoHoin the _o tetrahedralsite. The vertical OH intensityaxis representsthe L + o ratio of bandheights defined as (3662 cm-t)/(3662 cm-' 3588 a cm-r)where the band heights were estimated by manuallyfitting the spectrumto two Gaussianbands.

Natural hydrogrossular 1.0 Three samplesof natural hydrogrossularwith extensive OH substitutionwere examined.One (no. 1358)was the hydrogrossularfrom Crestmore,California, originally described by Foshag(1920) under the name "plazolite." It is one of the fragmentsgiven to Bassoby Pabst.Its structure was determined by Pabst (1937) and later refined by Basso et al. (1983). Its composition is approximately CarAlr(SiOo),,. (OH)r rr. It consisted of a clear, colorless crystal fragment approximately 600 pm in size. The second sample (no. 1329) from Crestmore, California, originally describedas "plazolite?" by Woodward et al. (1941), U 0 consistsof a colorless,somewhat turbid, fine-grainedma- 3800 3600 3400 3200 trix, 3-5 mm thick, between grains of akermanite. It is ') from the hand specimen described in the unpublished Wavenumbens(cm thesis of R. A. Crippen. The identity of the portion used for spectroscopicmeasurements was verified by X-ray Fig. l. Comparison of the infrared spectra of natural and powder diffraction (ao : 12.15 A, which indicates the synthetic hydrogrossularsamples with high OH contents. From formula CarAlr(SiOo),uo (OH), o, according to the calibra- top to bottom: synthetic hydrogarnets,CarAl.(SiOo).,, (OoHn)o ,, tion of Bassoet al., 1983).The third sampleis the katoite (no. 1058) in a KBr pellet; polycrystalline hydrogrossularno. 1329 (4.2 pm thick); singlecrystal hydrogrossularno. 1358 from of Passagliaand Rinaldi ( I 984). Its spectrumwas obtained Crestmore,California, (5.3 pm thick); katoite (no. 1444)in a KBr from a 50-pg crystal cluster that was ground and studied pellet; and synthetic CarAl?(OoHo)3(no. 1059)dispersed in a KBr as a micro-KBr pellet. pellet. The infrared spectra of the two Crestmore hydrogrossular samples are similar. They consist of a prominent band at 3598 cm-' with a secondaryband atabout3677 sented infrared spectra of these materials. They have a cm ' (Fig. l). The infrared spectra of the macroscopic major absorption band in the range 3600-3620 cm' and garnetsdiscussed in the subsequentsections do not look a minor band near 3660 cm-'. In addition, they show like this. variable amounts of broad absorption near 3420 cm ', When the natural and synthetic hydrogrossularsamples most likely from adsorbed HrO. Our spectrum of are compared in Figure l, some trends appear.The band ' Ca.Alr(SiOo)r r, (OoHo)'.r, is consistent with the published at 3662 cm in the silica-free synthetic garnet is also the spectra (Fie. l). dominant band in the katoite spectrum. It is present to ROSSMAN AND AINES: HYDROUS COMPONENTS IN GARNETS lt57

some extent in all the spectrain Figure l, although it is shifted to slightly higher energiesin the spectraof the more Hydno- silica-rich samples.With increasingsilica content, the rel- gnossulan ' ative intensity ofthe band at about 3598 cm increases. 2.O Our estimate of the proportion of the total OH intensity containedin th e 3662 cm-t band,defined as peak heightruur/ (height.uu,+ heightrrnr),is nearly identical to the propor- U (J tion of (OoHo)units in the tetrahedralposition, indicating c that the two components of the spectrum exhibit simple o _o mixing behavior (Fig. 2). ( A different type of spectrum is produced by massive -1 rn4r]. hydrogrossular with much lower OH contents. The best L.V known ofthese garnetsis the so-calledpinkjade ofSouth Africa, a massive,impure form of hydrogarnetwith about 5oloHrO. Both Frankel (1959) andZabinski (1966) showed an absorption band at about 3620 cm-' in the KBr pellet spectraof pink jade. Figure 3 compares the spectrum of a single-crystalre- gion in the pink jade from Transvaal (no. 941) with the syntheticCarAl,(SiOo)r(OH). (no. 1058).Also includedin 3800 3600 3400 3200 Figure 3 are the spectra of the massive, pale blue-gray - hydrogrossular from Nelson, New Zealand (no. 1364), Wavenumbens(cm which is similar to that of the pink jade. The spectrum of ) the pale yellow massive hydrogrossular from Mavora Fig. 3. Comparisonofthe infraredspectra ofnatural hydro- Lakes, New Z,ealand(no. 1427), also resemblesthe pink garnetand syntheticCarAlr(SiOo)r(OoHo). From top to bottom: jade but contains additional featuresnot identifiable with syntheticCarAl,(SiOJ,(OoHo) no. 1058,pink jade no. 941, mas- hydrogrossular, presumably arising from other hydrous sive hydrogrossularno. 1364,and massivehydrogrossular no. l presented I 2-pm mineralsin the rock. The 3662cm-' band of the synthetic 427.Samples 94 I and I 364are scaledto thick- ness.Sample 1427 Is scaledto 20 prn. The concentrationof silica-freehydrogrossular is not prominent in any of these sample1058 in the KBr pellethas been arbitrarily adjusted to spectra.Although following the discussionwill detail the have an intensitycomparable to slabsof the other materials. jade resemblanceof the pink spectrum to a classof mac- Spectrahave been offset vertically for clarity. roscopicgrossulars (class 4), the vast majority ofgrossular spectrado not resemblethe spectraof thesemassive, fine- grained hydrogrossularsamples. complicated spectrum from each class.These spectra il- Natural macroscopicgarnets lustrate the range of wavenumbers of peak absorption Absorption was observed in the OH region of the in- encountered in grossular; it extends from 3662 cm-r to frared spectrum of every grossularexamined. Only in the 3599 cm-'. Figures 5 through I I illustrate in more detail spectrumofgrossular l4l3 was the absorptionso weak the individual characteristicsobserved in theseminerals. as to be difficult to obtain. The spectra of these garnet samplesshow a great diversity, unlike the generally sys- Class I tematic spectral behavior observed in the natural HrO- This group is characterized by the most intense ab- rich hydrogrossularand in the pyrope-almandinegarnets sorption at 3662 cm-r. This is the only band in the spec- where the band position varies smoothly with composi- trum of the synthetic hydrogrossular(no. 1059)and is the tion (Ainesand Rossman,1985a). The absoluteintensity most intenseband in the katoite spectrum. Only one nat- of the OH bandsin grossularspans 3 orders of magnitude, ural samplehas been observedwith this band as the dom- significant variation in the appearanceofthe spectralfea- inant band. It is a single zone in the anisotropic grossular tures occurs within a single crystal, and the number of (no. FMAP9O.l I l) from the Jeffrey mine (Allen and Bu- bands and the positions of bands vary widely among gar- seck, 1988). In anisotropic garnet from Belvidere Moun- nets from different localities. tain, Eden Mills, Vermont, this band may be the most At first, it appearsthat there is nearly continuous vari- intense in one ofthe two extinction directions in spectra ation among the different types of grossularspectra. How- taken in polarized light. ever, when a large number of spectra are examined, a number of similarities appear in certain sets of spectra. Class 2 An empirical classificationscheme can be devised based The most prominent band is at 3647 cm ' (Fig. 5). A upon the wavelength of the most prominent absorption secondfeature occurs at 3688 cm-'. Its least complicated band in the spectrum. Ultimately, seven such classesof representativeis the Ramona, California, grossular (no. spectra were distinguished. Figure 4 compares the least 42), which occursin calciteveins (Lageret al., 1987b). l 158 ROSSMAN AND AINES: HYDROUS COMPONENTS IN GARNETS

0.8

-^^^^,,1-^ UI'UJ5UIdI Gr-ossulan Class 3

qJ 0) (J c tr o o _o n L ( 0.4 n n n r-1 !

v.(

0 0 3800 3600 3400 3200 3800 3600 3400 3200 Wavenumbens(cm-t ) Wavenumbens (cm-l )

Fig. 4. Comparison of the infrared spectraof the least com- Fig. 6. Class 3 grossular spectra with the strongestpeak at plicated spectrum from each class of spectra.From bottom to 3630 cm I. From top to bottom: no. 1430,307 pm, Eden Mills; top: class I to class7. Samplenumbers are indicated besideeach no. 53B-F,22.1 pm, Asbestos;no. 1051,380 pm, Eden Mills. spectral trace. Spectra thicknessesare scaled to produce com- parable absorption intensities.

Class 2b A more complicated variant of the class 2 garnet con- 7.? sists of garnet with the prominent band at 3645 cm-' in -^^^^,,1^ addition to a seriesoflower energybands, including those UI'UJ5U Id '. Class 2 at 3600. 3568. and 3548 cm This classis best exem- 1.0 plified by the colorlessto bright greenvanadian grossular samples from the calcite pods of the Tsavo district of Kenya (nos.229 and 77l). nl 0.8 O Class 3 tr rc These garnet samples (Fig. 6) have their most prominent componentat 3631 cm '. It is best exemplifiedby f 0.6 the samplefrom EdenMills, Vermont (no. 105l). A com- a ponent aI 3663 cm ' is always present and is sometimes _o o.4 prominent such as in the spectrum of no. 53B. Lower energybands near 3584 cm ' and 3560 cm ' are well developed in some members of the class.

v.a Class 4 Thesegarnet samples (Fig. 7) have their prominent band at 3621 cm-'. Sample53 is transitionalbetween groups 0 3 and 4. The class3 band at 3631 cm-' is nearly as intense 3800 3600 3400 3200 as the 3621 cm-r band. Sample941 is characteristicof (cm-t class4 garnet samples.It is a type of rock known as South Wavenumbens ) African Jade. The pink variety of this material consists Fig. 5. Class 2 grossular spectra with the strongestpeak at of an intimate mixture of two grossularphases that differ 3646cm-t. From top to bottom: no.771,100 pm, Tanzania;no. in their OH content (Zabinski, 1966; Bank, 1982). Its 1420, 54.7pm, Salzburg;no. 42,94.0 pm, Ramona. spectrumhas been frequently studied.Both Frankel (1959) ROSSMAN AND AINES: HYDROUS COMPONENTS IN GARNETS I 159

3.0 0.8 Gnossulan Gnossulan nl ^^^ t rl aaa E u -t dJ5 z+

0.6 938 nt ?.o U O C C fr (o _C] L L o.4 n o 3J a _o 1.0 u.t

94t

0 n 0 n JUUU 3600 3400 3200 3800 3600 3400 3200 Wavenumbens (cm-t ) Wavenumbens(cm-t ) Fig. 7. Class4 grossularspectra with the strongestpeak at Fig. 8. Class 5 grossular spectra with the strongestpeak at 3621cm '. Fromtop to bottom:no. 53B, 405 pm, Asbestos; no. 3612cm '. From top to bottom: no. 938,204 pm,Ossola Valley; 941,21pm, SouthAfrica. no. 1327,41 pm, Sciara;no. 1357, 38.2 pm, Bric Canizzi;no. 1042, 116 pm, Vesper Peak. and Zabinski (1966)showed an absorptionband at about 3620 cm ' in a spectrumofthe powder from South African Jade.We obtained the spectrum from a small, thin, single grossularcrystal in the pink jade. We found a peak at 3621 0.8 cm' and a secondaryband at 3672 cm-| . Of all the natural grossular samples, the spectrum of the pink jade most GnossuI ar- Class 6 closelyresembles the syntheticSir * hydrogarnet(no. I 058), appearing as a spectroscopicallybetter resolved version of the synthetic. The spectra of other, massive hydro- 0.6 grossularsamples from New Zealandresemble the spectra o of pink jade (Fig. 3). tt c o Class 5 _o o.4 The class 5 garnet sampleswere the most commonly L encountered(Fig. 8). Their spectrahave the strongestpeak a 14I T at 36ll cm ' with side peaksat about 3664 and,3562 cm-'. The spectrum of the VesperPeak, Washington, garnet (no. 1042) is the simplestin this class.A variety of o.2 other peakscan occur in thesespectra, primarily between the two side peaks. The Italian grossular samples (e.g., nos. l32l and 1357)are typical of the more complicated vJo behavior in this class. 0 Class 6 3800 3600 3400 3200 Thesegrossular samples have their main peak at 3602 cm-' (Fig. 9). Members of this class were infrequently Wavenumbens (cm-t ) encountered.The distinction between classes6 and 7 is Fig. 9. Class 6 grossular spectra with the strongestpeak at small. Grossular l4ll is transitional between the two 3602cm '. From top to bottom: no. 141l, 358 pm, Manum; no. classes. 936,27.2 pm, Bric Camula. l 160 ROSSMAN AND AINES: HYDROUS COMPONENTS IN GARNETS

1.0 0.4 Gnossulan Class 7 Gnossulan 0.8 0.3

C NA C n _o -o L ( u.c r-1 a o.4 rn _o _o 1423 n4 v.(

1359

n 0 .U 3800 3600 3400 3200 3800 3600 3400 3200 - Wavenumbens(cm-t ) Wavenumbens(cm ) Fig. 10. Class7 grossularspectra with the strongestpeak at Fig. ll. Grossularspectra that do not fit into one of the 3614cm-'. From top to bottom:no.946, lll pm, Auerbach;spectroscopicclasses. From top to bottom:no. 1423,85.5pm, no. 1122,175 pm, Crestmore; a dark interior zone of no. I125, OakGrove; no. 1359,15.0 pm, Gruppodi Voltri. 2.812mm, LakeJaco.

first overtones of the OH modes were clearly observed Class 7 near 7 100 cm-', therenever was any indication of a HrO Many of the class2b featuresare found in thesegarnet combination mode in the spectraof unaltered, inclusion- samples. However, the most intense feature of class 7 free garnet. Our results with grossular samples are con- garnetsis the bandat about 3599cm-r (Fig. l0). The 3644 sistent with our previous results (Aines and Rossman, cm ' band ofthe class I garnet samplesis replacedby one I 985b) with pyrope-almandine-spessartinegarnets, which at 3641 cm-'. These features are found in the spectra of also indicated that H substituted in garnetsas OH rather large, orange-brown grossular samples from the contact than HrO. The only molecular HrO found in the garnet zone of the calcite body at Crestmore, California, (no. spectra is associatedwith fluid inclusions visible in the ll22) and Auerbach, Germany, (no. 946) and in the rims sample.In most casesin our study, it was possibleto find of the large grossular samples from skarns east of Lake an optical path sufficiently free of inclusions (e.g., 100 pm Jaco,Mexico. diameter) for the spectroscopicmeasurements.

Miscellaneous behavior Zonation of OH Two garnet samplesproved difficult to classifyinto one The concentration of OH in these garnetsis often het- ofthe above categories.Their spectraare depicted in Fig- erogeneous.Typical examplesare illustrated in Figure 12. ure I l. The reddish orange grossular from Gruppo di There was no consistentzonation trend from core to rim. Voltari (no. I 359) has a peak at 3614 cm-' and two prom- The rims of grossularsamples 1042 (VesperPeak), I125 inent side lobesat 3566 and 3658 cm-'. (Lake Jaco),and 1409 (Saline Valley) were most concen- trated in OH, whereasthe coresof grossularsamples 14 I 2 No evidenceof structurally bound molecular HrO (Mul-Kum) and 1038 (Asbestos)contained the most. The The near-infrared spectralregion was examined for ev- greatestconcentration difference we observedwas a factor idence of molecular HrO in natural garnet. If present, it of -4 for grossularsample I4l2 from Mul-Kum. would give rise to an absorption band at about 5200 cm ' from the combination mode (bend + stretch). Because Correlation with chemical cornposition bands in the region are about 100 times lessintense than We were unable to establish any broadly applicable those in the region of the fundamental stretch of the OH correlations of individual IR bands with minor or trace groups(-3500 cm-'), it was necessaryto usethick sam- componentsin the garnet samples.Concentration profiles ples for near-infrared studies.When the even less-intense acrossa number of sampleswere examined, but they did ROSSMAN AND AINES: HYDROUS COMPONENTS IN GARNETS I l6l not lead to any noteworthy correlations. For example, garnet 1038 containsa greencore (Cr3*)and a colorless rim. Although its infrared pattern showedextensive core- to-rim variation, no correlation could be establishedwith major or minor elements. The large grossular samples from Lake Jaco, Mexico, OJ (e.g.,I 125)usually display color zonation. Frequentlythey (J C have colorless rims with interior zones that are brown ru when thin. The primary chemical differencebetween the _o L zonesis the order ofmagnitude greater Ti content in the o brown zones. The spectra ofthe two zones are very dif- a ferent (Fig. l2). Becausethe spectrum of the dark zone is similar to grossularsample ll22 from Crestmore(Fig. 9), which has about the sameTi content as the colorlesszone ofthe Lake Jacogarnet, it is unlikely that the spectroscopic differencesare necessarilydue to Ti. Very little differenceresults from the substitution of V for Al in grossularfrom Kenya. The spectrumofthe green, vanadian grossular (variety, tsavorite, rro. 227) is essentially identical to V-free grossularfrom the same locality 3800 3600 3400 3600 3400 (no. 77 l) in spite of a large differencein the V content. Wavenumbens(cm-t The lack of identifiable correlation with the minor el- ) ement contents was also observed in an extensive study Fig. 12. Four examplesof spectrathat demonstratecore-to- of the zonation of OH in grossularfrom Belvidere Moun- rim zonationofOH in grossular.Each set ofspectra is presented tain (Allen and Buseck,1988). Furthennore, changes in in the 3800to 3400cm-'region. No. 1042,Vesper Peak, 1.2 the occupation of the octahedral site had little effect on mm thick, 4.0 absorbancefull scale,rim slightlymore intense the spectrumof synthetichigh OH-content hydrogrossular thanthe core. No. ll25,Lake Jaco,0.60 mm thick,0.4 absor- samplesin the work of Kobayashiand Shoji (1984).They bancefull scale,colorless rim muchmore intense than the brown reported that the wavenumbers of OH absorption bands core.No. 1409,Saline Valley, 0.18 mm thick,2.0 absorbance in their synthetic garnet samples changed by only 4-6 full scale,rim moreintense than the core.No. 1038,Asbestos, rim less cm ' with replacementof up to l0 molo/oof andradite. 0.225 mm thick, 2.0 absorbancefull scale,colorless intensethan the g.reen core. Temperature dependenceof the spectra The spectra of about one-third of the garnet samples were obtained at -78 K. Spectroscopicresolution im- becausethey representa highly concentratedsource ofH proved and more bands were revealed. Although reso- in the volume of the sample used for H analysis. lution of featuresimproved, the same componentscould Results that we judge to be most reliable are presented be found in the room temperature spectra with second in Table 3 together with literature analysesfor some of derivative analysis. the samplesstudied. The values believed to be most ac- curate are from the garnet sampleswith the highest OH Ansolurn H coNrnNrs contents. Theseare used to establishan absolute calibra- We have tried to determine the absolute H content of tion of the 3600 cm-' region infrared spectraldata. Figure some of thesegarnet sampleswith H manometry, evolved l3 summarizesthe results. We found that somewhat bet- HrO coulometry, and thermal analysis and nuclear reac- ter correlations can be obtained from integrated absor- tion analysis. In many instanceswe find that our results bance values than from the peak height values (absor- were not reproducible with the low OH-content garnet. bance/millimeter in Table l). In most of thesecases, the analysesare being carried out From the reported OH content of Pabst's Crestmore near the limit of their accuracybecause of either the low hydrogrossular (5.88 OH per formula unit) and the OH absolute H contents or the difficulty with surface HrO content of the Crippen hydrogrossular(5.43 OH), we can acquired during the grinding of samples.Furthermore, it establish molar absorptivity values for OH in these gar- is usually necessaryto work with very small portions of nets. The € values per mole per liter concentration of OH the sample to avoid the fluid inclusions that are present calculatedfrom the main peak at about 3598 cm I are 62 in nearly all samples.The fluid inclusions have minimal and 79 for the Pabst and Crippen samples,respectively. influence upon the spectroscopicresults becausethey oc- The integrated absorbance values for these two garnet cupy a small proportion of the total area of the sample, samplesare 367 5 and38 l6 permol/L of OH, respectively, and it is usually possible to select small aperturesto de- where the integratedabsorbance is calculatedfor a crystal lineate areasfree of the inclusions. The inclusions, how- 1.0 cm thick. ever, have a major influence upon the total H analyses For purposesofthis paper,we usethe averageintegrated rt62 ROSSMAN AND AINES: HYDROUS COMPONENTS IN GARNETS

200 500 cp a a Grossular c 400 Grossular ! o o c 150 P c h f c; 1nn -c F 100 c o a I 200 E o o qn (l) 100 c N F B @ 0 10 tz 14 0 50 100 150 I H2O,Wt % Jooo cm Inteq.Intens. Fig. 13. Thecalibration of theinfrared integrated absorbance (Thousands) in the 3700-3400cm ' regionto analyticallydetermined OH integratedabsorbance in the contentsexpressed as weight percent HrO. Fig. 14. Correlationbetween the 3600crn I regionto the integratedabsorbance in the 7200cm ' reglon. absorbance value, 3746, for calculating OH concentra- tions. Specifically,a crystal 1.5 mm thick with a measured integrated absorbanceof 3 I 0 would have a calculatedin- if the calibration factors actually increasefor the iow HrO tegrated absorbanceof 2067 for a I cm path. This cor- content garnet samples as the first two points of Figure respondsto an OH concentrationof 2067/3746 : 0.546 13 suggest.We also have not determined if the different mol OH/L. If the crystal'sdensity were 3.15, it would classesof hydrogarnet spectrahave significantly different contain(0.546 moUL x l7 g OH/mol)/(3150g/L) :0.295 calibration factors. wto/oOH:0.156 wto/oHrO. The calibration can be ex- The HrO contents of our grossularsamples range from pressedin the following two equations: 12.760/o(no. 1329)to nearly zero (no. l4l3) with values commonly less than 0.25o/o.The highest HrO contents H,O wto/o: 0.0768 (0.0053) x absorbanceper mm occur in garnetsfrom rodingites, rocks subjectedto meta- H,O wto/o: 0.000078 6 (0.000000 9) somatic replacement, or from vugs containing products x integrated absorbanceper cm of presumably low-temperature alteration. Garnets from tlpicat carbonate skarns have the lowest HrO contents. where the numbers in parenthesesare the standard error of the coefficient from the regressionanalysis of the data in Table 3. An integrated absorbanceofabout 12720 per NnAn rNrn-lRED sPEcrRA cm per wt0/oHrO is indicated for the higher HrO content In some cases,experimental considerationsmake it de- garnetsif minor differencesin garnet density are ignored. sirable to measurethe near-infrared OH overtones at about Many problems and questions remain about the ab- 7200 cm-t rather than the fundamental bands near 3600 solute calibration. Our data are inadequateto determine cm r. We explored the correlation betweenthe intensities

Tlele 3. Summaryof calibrationdata for grossularHrO analyses

Integ. abs. HrO content Method of analysis Grossular 53' 2877 0.18 PrOscell coulometry Grossular 53 2877 0.18 H2evolved gas volume Grossular 53F 4796 0.28 1,Fnuclear reaction Grossular 941 61 250 5.00 PrOscell coulometry Hydrogrossular 1329 153420 11.71 calculatedfrom X-ray ao Grossular 1357 15746 1.20 Bassoet al. (1981) Hydrogrossular 1358 158 750 12.75 Bassoet al. (1983) Grossular 1360 10 839 0.97 Basso et al. (1984a) * Number13 of Ainesand Rossman(1985b). ROSSMAN AND AINES: HYDROUS COMPONENTS IN GARNETS I 163 of the near infrared and the fundamental OH bands. Fig- authors.We assignthe band at 3598 cm-', which is dom- ure l4 illustratesa good correlation betweenthe integrated inant in the silica-rich hydrogrossularsamples, to (OoHo) intensitiesof the bandsat -3600 cm 'and -7200 cm I groups adjacentto SiOogroups. The spectrum ofthe high in the mid-IR. The peakheights of the -3600 cm-'bands OH-content hydrogrossular samplesthen can be viewed are about 650 times more intense than those near 7200 as a superpositionof the two types of (OoHo)components cm-', whereas the integrated intensities are about 340 dominated by the Si-free katoiteJike sites. times greater in the 3600 cm-' region than in the 7200 cm-tregion. The hydrogrossularspectrurn in common garnet The resultingcalibration equations are: The spectraofmost ofthe low OH content, macroscopic grossular samplesdo not resemblethose of garnets with (H,O) wto/o: 49.7 (0.78) x absorbanceper mm verified hydrogrossularsubstitution. Only the class4 spec- : (H,O) wto/o 0.0255 (0.0014) tra resemblethe spectraof the authentic intermediate SiOo x integrated absorbanceper cm content hydrogrossulars.The complexity and diversity of the spectraofthe low OH content grossularargue that the where the numbers in parenthesesare the standard error hydrogrossularsubstitution found in the high OH content ofthe coefficient from the regressionanalysis. garnetsis not the dominant mode of substitution at lower OH contents. BrnrrnrNcnNT GARNETS A number of authors have proposed that H occupies Anisotropic OH absorption was previously reported in sites other than the tetrahedral Si site in hydrogarnets. a birefringent grossularfrom Asbestos,Canada, (no. 53) Bassoet al. (1984a, 1984b),Basso and Cabella(1990), leading Rossman and Aines (1986) to suggestthat the and Birkett and Trzcienski (1984) proposed that H may ordering of the OH could be responsiblefor the deviation substitute in both the dodecahedraland octahedral sites. from cubic symmetry. Three other strongly birefringent Kalinichenko et al. (1987) suggestthat the protons in hy- garnet samples(nos. l4ll and l4l3), including the an- drogrossularare situated in both the tetrahedral and the dradite-grossularfrom Mul-Kum (no. l4 l2) describedby octahedralsites. The complexity of the spectraof low OH Hirai and Nakazawa(1986), were examined in this study. content grossularin the OH region is intuitively consistent The latter three samplesdiffered from the Asbestosgarnet with a multiple site occupancy.However, an attempt by by being much more strongly birefringent and by con- Lager et al. (1989) to test this hypothesis found no evi- taining much lessOH. The latter three garnetshave spec- dence for multisite occupancyin the X-ray study of hy- tra that are only weakly anisotropic in the OH region. drous titanian andradites.Unfortunately, the garnetsthat often show the greatest spectral complexity in the OH DrscussroN region contain concentrationsof OH too low to be studied The spectraofthe natural and synthetic hydrogrossular by conventional diffraction techniques. Our initial at- samplesand katoite are most readily interpreted in terms tempts to examine this problem with NMR methods have of two types of OH environments that accountfor the two also beenthwarted by the low OH concentrationsin most bandsat 3662 cm-' and 3598 cm'. One pair of candidates natural samples(Yesinowski et al., 1988). for the two sites comes from the X-ray data of Sacerdoti and Passaglia(1985) and Bassoet al. (1983)who suggest that the OH sites are different in katoite and plazolite. CoNcr,usroNs Sacerdoti and Passagliahave the O-H vector in katoite l. Essentiallyall grossularcontains OH. projecting into the volume of the d-site tetrahedron, 2. The fine-grainedgrossular samples classically called whereasBasso et al. ( I 983) have the O-H vector projecting hydrogrossularare the most OH-rich. outside the tetrahedron. This distinction is difficult to 3. Among macroscopicgrossular, that from rodingites supportin view ofthe resultsoflager et al. (1987a,1989) is the richest in the OH component. who also find the O-H vector projecting outside the vol- 4. Garnets from carbonate skarns contain the lowest ume of the tetrahedron in both end-member katoite and concentrationsof OH. hydrogrossularsamples of intermediate composition. 5. The OH is readily observedin infrared spectra. Hydrogarnet structuresdisplay positional disorder that 6. The infrared spectraof macroscopicgarnet samples has been describedin terms of two O sites(the split-atom are significantly different from the spectraof true hydro- model of Ambruster and Lager, 1989). They correspond grossular, suggestingthat the hydrogarnet substitution SiOo to the O sites in anhydrous grossularand the sites in Si- : (OoHo)is not the only means of incorporating OH. free katoite. Our interpretation ofthe hydrogarnetspectra 7. For macroscopic grossular examined in this study, is that it representsthe superposition of these two sites. sevendifferent classesofinfrared spectracan be identified The3662 cm I band in the silica-freehydrogrossular, fol- based on the energy of the most intense OH absorption lowing the assignmentof Harmon et al. (1982),represents band. O.Ho groups that are surrounded by other OuHogroups. 8. Grossular can be zoned in both the concentration of The component at 3598 cm-I correspondsto the band in OH and in details of its site in the crystal. the 3600-3620 cm-r rangepreviously reported by various 9. The details of the OH absorption in the infrared are t164 ROSSMAN AND AINES: HYDROUS COMPONENTS IN GARNETS so characteristic that in many cases it was possible to et AlrCa.(SiO")r,u(OH),'u de type grenet. Acta Crystallographica,23, identify the locality from which an individual grossular 220-230. Foreman, D.W. (1968) Neutron and X-ray diffraction study of came once a standard spectrum was available. Ca.Al,(O"Do),, a garnetoid. Joumal of Chemical Physics, 48, 3037- 10. The analytical amount of OH in grossular can be 304r. estimated from the infrared spectmm. An integrated ab- Foshag,W.F. (1920) Ptazolite,a new mineral. American Mineralogist, 5, sorbanceof 3746 per cm thicknesscorresponds to an OH 183-1E5. (l garnet South African jade from Trans- concentration H2O I mol per liter Frankel, J.J. 959) Uvarovite and expressedas of of HrO vaal. American Mineralogist, 44, 565-59 l. of sample. Harmon, K.H., Gabriele, J.M., and Nuttall, A.S. (1982) Hydrogen bonding. Part 14. Hydrogen bonding in the tetrahedral OoHl- cluster in AcxNowr-rncMENTs hydrogrossular.Journal of Molecular Structure, 82, 213-219. Hirai, H., andNakazawa,H. (1986)Yisualizinglowsymmetryofa grandite We thank D.G. Schlom for assistingwith our synthesisof hydrogarnet garnet on precessionphotographs. American Mineralogist, 71, l2l0- and G.A. I-ager (University of Kentucky) for providing samples of his 1213. synthetic hydrogarnetsand numerous helpful discussionsabout the hy- Hsu, L.C. (1980) Hydration and phaserelations ofgtossular-spessartine drogamet substitution. Other garnetsfor this study were donated by A.L. garnetsat PH,o: 2Kb. Contributions to Mineralogy and Petrology,7 I, Albee (Caltech), F.M. Allen (Arizona State University), G Amthauer 407-415. (Salzburg),D. Atkinson (Santa Barbara), R. Basso (Genoa), C. Bridges A.M., Proshko, V.Ya., Matyash, I.C., Pavlishin, Y.I., and (Nairobi), B. Cannon (Seattle),R. Coleman (Stanford),R.H. Currier (Ar- Kalinichenko, M.Ya. (1987) NMR data on crystallochemicalfeatures of cadia), P. Flusser (l,os Angeles),C. Francis (Harvard Mineral Museum), Gamarnik, hydrogrossular.Geochemistry International, 24, 132-135 (translated M. Gray (Culver City), G. Harlow (American Museum), W.A. Henderson no.9, 1363-1366,1986). (Stanford),H S. Hill (Altadena),P. Keller (Los AngelesCounty Museum), from Geokhimiya, Kobayashi, S., and Shoji, T. (1983) Infrared analysis ofthe grossular- G.A. Novak (California StateUniversity, I-os Angeles),and Y. Takeuchi hydrogrossularseries. Mineralogical Journal, I l, 331-343. (Tokyo). This study was funded in part by NSF grants EAR-7919987, - (1984) Infrared analysisofthe grossular-hydrogrossularseries with EAR-8313098,EAR-8618200, and EAR-8916064.Contribution no. 4449. a small amount of andradite molecule.Mineralogical Journal, 12, 122- RnrnnnNces crrED I 36. Lager, G.A., Armbruster, T., and Faber, J. (1987a) Neutron and X-ray Aines, R.D., and Rossman, G.R. (1985a) The water content of mantle diffraction study of hydrogarnetCa3Al,(O4H4). American Mineralogist, garnets.Geology, 12, 720-723. 72,756-765. - (1985b) The hydrous component in garnets:Pyralspites. American Lager,G.A., Rossman,G.R., Rotella, F.J., and Schultz,A.J. (1987b)Neu- Mineralogist, 69, | | 16-1 126. tron-diffraction structure of a low-water grossular at 20 K. American Allen, F.M., and Buseck,P.R. (1988) XRD, FIIR, and TEM studies of Minerafogist, 72, 766-7 68. optically anisotropicgrossular garnets. American Mineralogist,73, 568- Lager, G.A., Armbruster, Th., Rotella, F.J., and Rossman, G.R. (1989) 584. The OH substitution in garnets: X-ray and neutron diffraction, infrared Armbruster, T., and Lager,G.A. (1989)Oxygen disorder and the hydrogen and geometric-modelingstudies. American Mineral ogist,7 4, 840-8 5 l. position in garnet-hydrogarnetsolid solutions. European Journal of Pabst,A. (1937) The crystal structureof plazolite.American Mineralogist, Minerafogy, l, 363-369. 22,86r-868. Bank, H. (1982) Uber Grossular und Hydrogrossular. Z€itschrift der Passaglia,E., and Rinaldi, R. (1984) Katoite, a new member of the DeutschenGemmologischen Gesellschaft, 3 l, 93-96. Ca.Alr(SiOo),-CarAl,(OH),,series and a new nomenclaturefor the hy- Basso,R., and Cabella,R. (1990) Crystal chemical study of garnetsfrom drogrossulargroup of minerals. Butletin de Min6ralogie, 107, 605-6I 8. metarodingites in the Voltri Group metaophilites (Ligurian Alps, Italy) Rinaldi, R., and Passaglia,E. (1989) Hibschite topotype: Crystal chemical NeuesJahrbuch fiir Mineralogie Monatshefte, 127-136- characterization.European Journal of Mineralogy, l, 639-644. Basso,R., Giusta, A.D., andZnfiro, L. (1981) A crystal structure study of Rossman,G.R., and Aines, R.D. (1986) Birefringent grossularfrom As- a Ti-containing hydrogarnet.Neues Jahrbuch fiir Mineralogie Monat- bestos,Quebec, Canada. American Mineralogist, 7 l, 77 9-7 80. shefte,230-236. Sacerdoti,M., and Passaglia,E. (1985) The crystal structureofkatoite and - (1983) Crystal structurerefinement ofplazolite: A highly hydrated implications within the hydrogrossulargroup of minerals. Bulletin de natural hydrogrossular. Neues Jahrbuch fiiLr Mineralogie Monatshefte, Min6ralogie, 108, l-8. 25t-258. Wilkins, R.W.T., and Sabine,W. (1973)Watercontent of somenominally Basso,R., Cimmino, F., and Messiga,B. (1984a) Crystal chemical and anhydrous silicates.American Mineralogist, 58, 508-5 I 6. petrological study of hydrogarnets from a Fe-gabbro metarodingite Woodward, A.O., Crippen, R.A., and Garner, K.B. (1941) Sectionacross (Gruppo di Voltri, Western Liguria, Italy). Neues Jahrbuch fiir Minera- Commercial Quarry, Crestmore,California. American Mineralogist,26, logie Abhandlung, l5O, 247-258. 35l-38 l. - (1984b) Crystal chemistry of hydrogarnetsfrom three different Yesinowski, J.P., Ecken, H., and Rossman,G.R. (1988) Characterization 'H microstructural sites of a basaltic metarodingite from the Voltri Massif of hydrous speciesin minerals by high-speed MAS-NMR. Journal (Western Liguria, Italy). Neues Jahrbuch fiir Mineralogie Abhandlung, ofthe American Chemical Society, ll0, 1367-1375. r48,246-258. Zabinski, W. (1966) Hydrogarnets.Polska Akademia Nauk, Oddzial Kra- Birkett, T.C., and Trzcienski, W E (1984) Hydrogamet: Multi-site hy- kowie, Komisja Nauk Mineralogicznych,Prace Mineralogiczne, 3, 1- drogen occupancy in the garnet structure Canadian Mineralogist, 22, 69. 675-680. Cohen-Addad, C., Ducros, P., and Bertaut, E.F. (1967) Etude de la sub- Menuscnrrr REcErwDAucusr 20,1990 stitutiondugoupementSiO4par(OH)4danslescomposdsAlrCa,(OH),, Meuuscntrr AcCEPTEDAprIr 18, 1991