AmericanMineralogist, Volume 63, pages 91 3-917, 1978

Compositionaland refractive index variations of the -hvdroxvl-herderiteseries

PnrnnB. LrevnNs Departmentof Geology,Uniuersity of Delaware Newark.Delaware l97l I

Pnrr J. DunN D epartmentof GeologicalS ciences, S miths onian Institution lhashington,D. C. 20560

,arNpRIcgAno V. Genrs R.D.] Potts t own,P ennsy luania I 9464

Abstract

The compositionalrange of the herderite-hydroxyl-herderiteseries, CaBePOo(F,OH), has not beenwell-established. Microprobe analysis of 41 herderiteseries samples from 19 local- itiesgave compositions between 98 and 53 mole percenthydroxyl-herderite. The only speci- menof trueherderite known to us is a cut gemat the SmithsonianInstitution with 60 mole percentherderite. Nineteenth-century analyses of specimensfrom Stoneham,Maine, report- ing highfluorine contents and widely cited, must be considered doubtful. Indices ofrefraction of 12 samplesshow a linear declinewith increasingF content, and least-squareslinear regressionanalysis of the data gave a = 1.615- 0.00059(7oH) r: -0.991 p: t.634-0.00056(7oH) r: -0.994

I = 1.644-0.00055(7oH) r: -0.994 where7oH is the mole percentherderite in the specimen,and r is the correlationcoefficient of the indiceswith compositionfor the fitted lines.The X-ray diffractionpattern of hydroxyl- herderiteshows no perceptibleshifts with composition.

Introduction rine, but Genth (1884) felt that Winkler's methods were totally inadequate for the detection of fluorine. Herderite, CaBePOnF,and hydroxyl-herderite, Ca- The division of the seriesinto herderite and hydroxyl- BePO.(OH), are end-membersof an isomorphous herderite was made by Palacheet al. (1951,p. 820). series. Herderite was first discovered in the tin veins Most analyses of herderite datb from the nine- of Ehrenfriedersdorf, Saxony (Haidinger, 1828); hy- teenth century and suffer from the difficulties of droxyl-herderite was first describedfrom a pegmatite fluorine analysis by the wet-chemical methods in use at Paris,Maine, ashydro-herderitebyPenfield(1894). at the time. Optical data on analysedherderites are The actual composition of the Ehrenfriedersdorf sparse. This paucity of recent and reliable data herderite has never been established.Winkler, in prompted us to analysea large number of herderites Weisbach(1884), gave two analyses,with 6.59 weight and to determine indices of refraction for a number percent and 6.54 weight percent HrO, greater than of these. A summary of our results, with a detailed the theoretical amount of water in pure hydroxyl- discussion of hydroxyl-herderite from two Brazilian herderite, and only a "doubtful reaction" for fluo- localities, is presentedelsewhere (Dunn et al.,1979). 0003404x/'r8/09r0-0913$02.00 913 914 LEAVENS ET AL,,' HERDERITE.SERIES

Occurrence Maine, Penfield(1894) noted that a prism of the The herderiteseries minerals occur typically in peg- mineral was cut to determinethe index of refraction matites,as crystalsin cavitiesformed during the late, by the methodof minimumdeviation. This prism and replacementstage of crystallization,Crystals are usu- a specimenlabelled "hydro herderite,Paris, Maine" ally only a few mm across,but in oneBrazilian local- (Brush#1973) are preserved in theBrush collection of ity reachl5cm (Dunn et al., 1979).The most com- Yale University and were made available for this monly associatedminerals are crystalline qvartz, study,Penfield (1894) found a "Irace"offluorine, but albite,muscovite, and clays.Microcline, if present,is microprobeanalysis shows that this hydroxyl-herd- etched,as at Topsham,Maine (Yatsevitch,1935), erite actuallycontains about I percentF by weight, and Virgem da Lapa, Minas Gerais,Brazil (Dunn el and is 9l mole percenthydroxyl-herderite, rather al., 1979).Hydroxyl-herderite may form by the alter- thanthe pureend-member. Penfield was a meticulous ationof beryl(Yatsevitch, 1935; Perham, 1964) or of analyst;the presenceof I percentF undetectedin the beryllonite(Palache and Shannon,1928). It rarely sampledespite his careis evidenceof the difficultyof showsalteration, but occursas bright, euhedralcrys- F analysisby the methodshe used. talsimplanted on the otherminerals. Ford (1911)described a herderitemineral from Auburn, Maine.He includeda chemicalanalysis for water and fluorine,which yieldedF : 6.04percent, Chemistry HrQ = 3.62 percent,corresponding to 48 mole per- Forty-onesamples of herderiteseries minerals were cent and 65 mole percenthydroxyl-herderite respec- analysedwith an Anl-Snlrq electronmicroprobe us- tively. A twinned crystal of the describedmaterial ing an operatingvoltage of l5kV, a beamcurrent of (Yale rt3772)was examined in this study.It hasone 0.l5pA, and a lOp beam.The standardswere fluo- end sawnand is presumablythe analysedspecimen. rapatitefor Ca, P, and F; hornblendefor Si, Al, Fe, Microprobe analysisgave 4.20 percent F, corre- and Mg; and manganitefor Mn. The datawere cor- sponding to 64 mole percent hydroxyl-herderite, rectedfor background,backscatter, absorption, and compatiblewith the HrO valuereported by Ford. fluorescenceusing a computerprogram employing Hydroxyl-herderite from the Golconda Mine, the,Bence-Albee factors. The analysesare presented MinasGerais, Brazil (NMNH #121024)was used by in Table l, alongwith the fluorineanalysis of a fac- Lagerand Gibbs (1974)for a detailedcrystal-struc- eted green gem herderitein the SmithsonianGem ture analysis.They treatedit aspure hydroxyl-herd- Collection(Dunn and Wight, 1976). erite, but our analysisof fragmentsfrom the same The analysesare averages;the fluorinecontent of specimen,a singlecrystal 7 X 5 X 4cm,showed 3.39 singlecrystals varied by asmuch as 20 percentrela- percent F, correspondingto 7l mole percenthy- tive at differentsample points. Indices of refractionof droxyl-herderite. different fragments of single crystals show corre- Completeanalyses of herderiteminerals for which spondingvariation. In addition, differentsamples F > OH werepublished by Hiddenand Mackintosh from the samelocality are in somecases quite differ- (1884)and by Genth (1884),both on materialfrom ent in averagecomposition. No attemptwas made to Stoneham,Maine. Since all six samplesfrom Stone- mappossible zonal distribution of fluorinein anyof hamwhich we analysedare hydroxyl-herderite, these the specimens. early analysesmust be reconsidered.Hidden and The analyticaldata indicate very little replacement Mackintosh(1884) gave 11.32percent F and stated of Ca by Fe, Mg, or Mn in the herderite-hydroxyl-only that "the fluorinewas calculated from theexcess herderiteseries. Most samplesconform quite well to of lime." They did not statethat they madedirect the theoreticalcontent for Ca and P. The analytical testsfor fluorine or for water. Although they stated data indicatethat herderiteis a very rarespecies. All that they had no doubt that the formula calculated the samplesstudied herein are hydroxyl-herderite; from their analysisrepresented the true composition nonewere found with a fluorinecontent which would of themineral, it is in factuncertain with respectto F. indicateF > OH. The only herderiteknown to us is Genth (1884)in summarizinghis analyticaldata, the gemstonedescribed by Dunn and Wight (1976). stated,"somewhat doubtful is the exactamount of A number of the analyzedspecimens have been fluorinewhich it contains."Although Genth thought studiedin the past,and somecomments about these hisF content(8.93 percent) too low,it isclear that he arein order. wasdoubtful of the analyticalprocedures for F deter- In his descriptionof hydro-herderitefrom paris, mination in use at that time. Penfieldand Harper LEAVENS ET AL.: HERDERITE SER/ES 915

Table l. Partialchemical analyses for the herderitesertes

# NMNII unless Legs Locality noted sio2 CaO Pzos o=F Total Mol8oH

Theoretical hydroxyl-herderite 34.82 44.06 0,00 0. 00 78. 88 100t1

Dunton Gen Mine, Newry, Maine * R4109 r.40 30.60 38.52 0.22 0.09 72.60 98+ Dunton cem Mine, Newry, Maine * L27895 0.83 30.s4 40.00 0.07 0.03 73.5L 999 Foote Mine, Kings Mountain, N.C. * 132503 3.35 30.29 37.59 0.08 o. 03 72.L6 99/ Keyes +l Mine, Orange, New Hampshire Private 0.67 34.04 42.80 0.24 0.10 77.6s 98 (" ) R16743 0.87 34.04 42.07 0.22 0.09 77.!0 98

Palemo *1 Mine, North croton, N.H. 106076-r 0. 00 34.L7 43.L4 0.19 0.08 7'1.32 98 (" ) 106076-t 0.03 33.26 43. 55 0.47 0.20 77.2L 96 Paris, Maine * Brush 1973 0.00 34.45 44.08 1.01 0.43 79.LL 9I Golconda Mine, Minas cerais, Brazil L2LO29 0.04 34.49 44.09 r.16 0.49 79.29 90 121033 0. 16 34. 42 44.24 1.69 0.7r 79.80 86 r2L032 0. 07 34.56 44.32 r.69 0.71 79.93 UO 121030 o.23 34.25 43.94 1.73 0,73 79.42 85 L2L034 o.32 34.22 44.01 L.82 0.77 79.60 84 * L2L024 0.00 34.59 42.93 3.39 1.43 79.48 ?l Fletcher Mine, North croton, N.H. 10615 7 0.00 34.78 43.97 0. 36 0. 15 75.96 97

(') 10616 7 0. 30 34.24 43.89 2.82 t-.19 80.05 76 (") 106191 0.00 34.47 43.89 L.72 0.72 79.36 85 Poland, Maine K)ZUI 0. 15 34.09 43.73 2.33 n oa ?q ?? 80 (") B13685 0.10 33.86 43.8s 3.32 1.40 79,73 Auburn, Maine 814r60 0.48 34.50 43.25 r.65 0. 69 79.L9 86

(") RI0289 0.00 34.14 43.99 2.90 L.22 79.7L 75 (") 46845 0.00 34.79 43.9? 3.78 l. 59 80.95 68 (") * laLe 37'12 0.00 33.90 43-34 4.20 L.79 79.65 64 Poland, Maine R 5201 0.00 33.8r 43.24 3.85 L.62 79.28 67 Bennett Mine, Buckfield, Maine LJ 1J6 L 0.14 34.54 44.02 2.98 I.25 80.43 (') L32682 0.14 33.89 43.52 3.41 L.44 79.52 7L Greenwood, Maine 123395 0.33 34.'lO 43.49 3.85 L.62 80.75 55 Stoneham, Maine 120 817 0. l0 34.23 43.9L 3.29 1. 38 80.15 R5203 0.00 34.13 43.31 3.43 L.44 79.43 7L B84I? 0.00 33.53 43.15 3.48 L.46 78.70 70

45077 0.00 34.55 44.0s 3.94 1.66 80.88 56 YaIe 1955 0.00 34.L? 42.99 4.r0 L.73 79.53 of, B13669 0.23 34.19 43.51 4.52 r.90 80.55 61 Topsham, Maine 1125s9 0.00 34.00 43.03 4.3r 1.81 79.53 63 virgm da Lapa, Minas Gerais, Brazil 134657 0.30 34.54 43.74 3.L7 1,33 80.42 53 (") R1814r 0.20 34.57 44.L3 4.79 2.02 81.67 59 (") 135016 0, 00 34.25 44.L4 5.31 2.34 81.36 Epprechstein, Germany * 82L357 0.00 34. 47 43.13 4.74 2.00 80,34 59 BIue chihuahua M., San Diego Co., Cal. L2L245 0.05 34.42 43.97 4.74 2.00 81.18 59 (") r2L275 0.22 34.55 43.91 s.00 2.10 8I.58 waldstein, Gemany 821356 0. 00 33.90 43.34 5.44 2.29 80. 39 5J

creen Gem, Brazil Gem4948 7.01 40

Theoretical herderite 34.39 43.53 1r.65 4.91 84.66 0n

* - discussed in text i - rncludes 0.65t Alrol and 1.29t Feo. E - rncludes 0.85t A1;O; and I.258 Feo. / - Includes 0.25t At:o; and 0.639 Feo. z n - Plus 15.34* Beo. tl - Plus 5.59t H2O and 15.538 Beo.

(1886)analysed a specimenfrom Stoneham,and us- andHarper, and in the absenceof moreconfirmatory ing the ratio of F to HzOin the analysis,concluded data, the existenceof herderitefrom Stonehammust that their material was 60 mole percenthydroxyl- be considereddoubtful. We were unableto obtain herderite.They consideredthat their analysissuper- samplesof the actual material analysedby Hidden sededthose of Hiddenand Mackintoshand Genth. and Mackintoshor by Genth. Thus the compositionof Stonehammaterial was a The specimenfrom Epprechtstein,Fichtelgebirge, matter of considerabledispute among mineralogists West Germany,with 59 mole percenthydroxyl-herd- in the 1880's.Our analysessupport that of Penfield erite, is a single crystal, illustrated by D0rrfeld 916 LEAVENS ET AL.: HERDERITE SER/ES

Table 2. Indices of refraction for the herderite series distinguishablefrom those of the other hydroxyl- herderitesX-rayed in this study.

Locality* catalog uol% #o* ott Optics Keyes Mlne RL6743 98 1.5L5 t,634 L,643 Keyes Mlne Prlvate 98 r.613 L.632 L.642 Indicesof refractionfor a numberof analysedsam- Paris brusn rylj 9L 1.511 1.530 L.639 ples Auburn 814160 86 r.509 L.627 r.538 weredetermined using the spindlestage designed Aubum 46845 68 1.594 1.615 L,625 by Jones(1968). Principal vibration directions were Poland R5201 ot L.597 t.517 L.628 located by observing interferencefigures in con- Stoneham Yale 1955 65 L.592 1.613 L.524 Aubum YaLe 3772 64 1.593 L.6L4 L.525 oscopiclight, and refractiveindices were measured in Epprechtstein B2L357 59 1.591 1.610 L.620 sodium light by the Beckeline method. Refractive Virgen da l-€pa 1-35016 )4 1.588 1.608 1.619 indicesof the immersionoils werechecked in each IIalds tein 82L356 )J 1.589 1.610 1 4t O RtazLL Gem4948 40 1 .581 1.501 1.610 case by refractometer.Single index determinations

* Cqoplete Localitiee are given in Table 1. are believedto be precisewithin +0.001. ,-* UnLess noted, ntmbers refer to NI'INHcollectlons. Becausethe grains analysedby microprobeand thoseexamined optically were not thesame, although ( 1909),and is now a part of the Carl Boschcollection comingfrom thesame specimen and usually from the at the SmithsonianInstitution (#821357).On the samesmall crystal,and becausedifferent grains may basisof apparentdifferences in axial ratios, Diirrfeld showmarked variation in compositionand refractive concludedthat thishydroxyl-herderite contains less F indices,refractive indices were determined on two or than hydroxyl-herderitefrom Stoneham,Maine; in moregrains of mostspecimens. If differencesof 0.003 fact, its compositionis comparableto thoseof hy- or greater were found for a given index between droxyl-herderitesfrom Stonehamwhich we have grains, the specimenwas consideredexcessively in- studied. homogeneousand the datadiscarded. As canbe seen An unfortunateomission from our datais informa- from the equations below, a differenceof 0.003 in tion on herderitefrom the type locality, Ehrenfrie- refractiveindex correspondsto a diflerenceof about dersdorf.Despite considerable effort we wereunable 5 mole percentF (or OH) in the composition.If to locatea specimen.A supposedspecimen in the differencesin index wereless than 0.003,the average Carl Boschcollection proved to be . of the determinationswas used. The resultantindices Two hydroxyl-herderitesamples from the Dunton of refractionof l2 samplesare recordedin Table2, Gem Mine, Newry, Maine (NMNH #R4109, along with the indicesof the herderitegem, which #127895),and onefrom theFoote Mineral Company were determinedby refractometerto t0.002 (Dunn spodumenemine near King's Mountain,North Car- and Wight, 1976).The equationsof the best-fitlines olina (NMNH #132503),have aberrantchemistry. determinedby least-squareslinear regression analysis The materialfrom Newry is botryoidal,brown in afei color, and occursas an alterationof beryllonite(Pa- a : 1.615- 0.00059(7oH) r: -0.991 lacheand Shannon,1928). It is low in calciumand phosphorus,but containssignificantly more iron-up p:1.634-0.00056(7oH) r: -0.994 Io 1.29percent FeO-than mosthydroxyl-herderite, ^y: 1.644- 0.00055(7oH) r: -0.994 and up to 1.4percent SiO2, perhaps in substitutionas SiOo for POn, as in ellestadite,viseite, perhamite whereVoH is the mole percentherderite in the speci- (Dunn and Appleman,1977), and other silico-phos- men.and r is the correlationcoefficient of the indices phates.The summationsof the analysesare quite with compositionfor the fitted lines. low, however,and thereappear to beother problems The herderiteseries thus shows a significantde- in the chemistryof thesespecimens. A microprobe creasein indicesof refractionwith increasingF con- scan failed to reveal other detectableelements, al- tent. This changeprovides a usefulmethod for deter- though Li, Be, and B cannot be detectedon the miningthe compositionof an unknownsample. The probe, and the lack of any effervescenceof this hy- variationis not unexpected,for other mineralswith droxyl-herderitein HCI indicateslow, if any, sub- F-OH substitutionshow a decreasein indicesof re- stitutionof COgfor POn.The King's Mountainhy- fractionwith increasingF content. droxyl-herderitealso hasa high SiO, content-3.35 Calculated2V anglesshow a slight change,from percent. The X-ray powder patterns of both the 77o for pure hydroxyl-herderiteto 70ofor the mid- Newry and the King's Mountain materialare in- point composition.Thus 2V is not a usefulproperty LEAVENS ET AL.: HERDERITE SER/ES 9t7 for determiningthe compositionof membersof the analysingfor smallamounts of F. HoraceWinchell made a num- series. ber of specimensavailable from the Yale Universitycollections, including documentedspecimens exceedingly important for.this Crystallography study.Joseph Nelen of the Smithsonianlnstitution and Jun lto of the JamesFranck Instituteof the Universityof Chicagoprovided Someminerals show changesin unit-celldimen- preliminarychemical analyses which were useful in the earlystages sionswith OH-F substitution.In the caseof tremo- ofthis study.Frederick Keidel provideda sampleofthe hydroxyl- lite (Cameronand Gibbs, 1973),the effectis small, herderitefrom the KeyesMine. amountingto a reductionof about0.054 in the b-axis References dimensionof syntheticfluor-tremolite as compared to hydroxyltremolite, with smallerreductions in the Cameron,M. and G. V. Gibbs(1973) The crystalstructure and otheraxial dimensions.To testwhether a changein bondingof fluor-tremolite:a comparisonwith hydroxyl tremo- lite. Am. Mineral.,J8, 879-888. lattice dimensionswith compositionis detectablein Dunn, P. J. and W. Wight ( 1976)Green gem herderite from Brazil. hydroxyl-herderite,an X-ray powder photograph J. Gemmology,1 5, 27-28. wasmade of hydroxyl-herderitewith 98 mole percent - and D. E. Appleman (1977) Perhamite,a new calcium OH (PalermoQuarry, N. H.; NMNH #106076-l) aluminumsilico-, and a reexaminationof vis- mixed with an equal amount of hydroxyl-herderite 6ite.Mineral. Mag.,4l, 437442. -. C. W. Wolfe. P. B. Leavensand W. E. Wilson(1979) from Auburn, Maine, with 64 mole percentOH Hydroxyl-herderitefrom Braziland a guideto speciesnomencla- (Ford, l9l l; Yale #3772),using nickel-filtered CuKa ture for the herderite-hydroxyl-herderiteseries. Mineral, Rec- radiationand a I14.6 mm Debye-Scherrercamera. ord in press. There was no detectablebroadening or splittingof DUrrfeld,V. ( 1909)Die Drusenmineraliendes Waldsteingranits in the lines as comparedwith the pattern of Palermo Fichtelgebirge.Z. Kristallogr.,16, 583-585. Ford, W. E. (l9ll) On someherderite crystals from Auburn, Quarryhydroxyl-herderite alone; thus, any shift must Maine.Am. J. Sci.,32,283-286. beslight, and X-ray powderdiffraction does not pro- Genth,F. A. (1884)On herderite.Proc. Am. Phil.Soc.,21,694- vide a methodfor determiningF contentin members 699. of the series.Because OH- and F- havethe same Haidinger,W. (1828)On herderite,a new mineralspecies. Pftr|. numberof electrons,their mutualsubstitution is un- Mag.,4, l-3 (not seen;extracted from Palacheet al,, l95l). Hidden,W. E. and J. B. Mackintosh(1884) On herderite(?), a likely to causeintensity changes in the X-ray diffrac- gluciniumcalcium phosphate and fluoride from Oxford County, tion pattern.Powder diffraction data of the Palermo Maine.Am. J. Sci.,27,135-138. hydroxyl-herderitehave been submittedindepend- Jones,F. T. (1968)Spindle stage with easilychanged liquid and ently to the Jcpos. improvedcrystal holder. Am. Mineral.,53, 1399-1404. Penfield(1894) and Ford (19l l) investigatedthe Lager,G. A. and G. V. Gibbs (1974)A refinementof the of herderite, CaBePOTO H. Am. M ineral., 5 9, 9 19-925, variation of the morphologicalaxial ratios of hy- Palache,C. and E. V. Shannon(1928) Beryllonite and other phos- droxyl-herderitewith its composition,although each phatesfrom Newry, Maine.Am. Mineral., 13,392-396, observedthat the usualrounding and unevennessof -, H. Bermanand C. Frondel(1951) Dana's System of Min- the facesof crystalsmade precise goniometric mea- eralogy,Tth ed. Wiley, New York. surementsdifficult. Penfield reported a variation, Penfield,S. L. (1894)On the crystallizationof herderite.Am, J. Sci..32, 107-l10. whereasFord did not. Diirrfeld (1909)used Pen- - a1d D. N. Harper(1886) On thechemical composition of field's(1894) data in his studyof Epprechtsteinhy- herderiteand beryl.Am. J. Sci.,32, 107-110. droxyl-herderitediscussed under Chemistry,Our X- Perham,F. C. (1964)Waisanen mine operation-summer, 1963. ray study did not detect any changein the lattice Rocksand Minerals, 39, 341-347. spacingswith compositionand henceno changein Weisbach,A. (1884)Uber Herderite.Neues Jahrb. Mineral. Geol. Palaeontol., 2, | 34-l 36(not seen;extracted from G enth,| 884). theaxial ratios, supporting the work of Ford (l9l I ). Yatsevitch,G. M. (1935)The crystallographyof herderitefrom Topsham, Maine. A m. Mineral., 20, 426-437. Acknowledgments Clifford Frondelfirst pointedout the problemswith the dataon the herderiteseries to the seniorauthor more than l0 yearsago; Manuscriptreceiued, March 27, 1978;accepted solutionof theseproblems had to wait for adequatemethods for for publication,May 30, 1978.