THE AMERICAN MINERALOGIS'I, VOL. 49, SEPTEMBER_OCTOBER, 1964

PHASE RELATIONS BETWEEN CYMRITE, BaAlSiaOr(OH), AND CELSIAN, BaAlzSizOal

Y6r.q.n6 Srn2 aNo Gnoncl C. KBNNBnv, Institute oJ Geophysics anil Planetary Physics, Uni:tersity oJ CaliJornia, Los Angeles,CatiJ. Aesrnacr

The chemical reaction: celsianfsanbornite+2 quartzfwater32 cymrite, has been experimentally studied in the temperature range of 200" to 850'c. and at pressuresup to 35 kb by the simple squeezer. Hexagonal celsian forms a complete solid solution series with cymrite by substitution of Al3+ for [Si(OH)]3+ with increasing water pressure (and/or de- creasing temperature). Some interpretations of the natural occurrence of monoclinic cel- sian, cymrite and sanbornite are presented.

IrttnooucttoN The potassium and barium feldspars have been studied at length in this laboratory at high temperaturesand pressures.Data on the potas- sium feldspars are appearing in a separate paper. The general relation- ships between the potassium and barium feldspars are shown in the following diagram. r:-.-----.1 IK.\rsiro"|+H,O Sanirline I I larer pressur.e I Neu.l,hase I

(KSi-BaAl) Solidsolution (Hyalophaneseries)

-l IlaAhSLOs :FtAt"sl,c\ l{onoclinic cclsian pressure Hexagonal celsian I I

Sanidineplus water goesto a new phase,KHAISLOs.OH at high pres- sure. This new phase is an analog of cymrite and an end member of a solid solution seriesinvolving cymrite. Cymrite, in turn is part of a solid solution series of which hexagonal celsian is the other end member.

r Publication 1304 Institute of Geophysics and Planetary Physics, University of Cali- fornia, Los Angeles, California. 2 Department of Earth Sciences,Saitama University, Urawa, Japan. 1407 1408 Y. SE:,KIAND G. C. KENNEDY

Hexagonal celsianis the high pressure-lowtemperature poiymorph of monocliniccelsian. Monoclinic celisanis an end member of a solid solu- tion seriesof which sanidine is the other member. Cymrite was describedas forming a vein within the manganiferous ore deposit of the Benallt mine in Wales b;" Smith et al. (1949). They re- ported the resultsof their microchemicaltests and r-ray study and con- cluded that this mineral with the chemicalcomposition BaAlSirOs(OH) probably had a very simple hexagonalframework tvpe structure. Hexagonal celsian was artificially prepared by Dittler and Lasch (1930), Yoshiki and Matsumoto (1951) and Davis and Tuttle (1952). Yoshiki and Matsumoto (1951) found two modifications (a and 0) of hexagonalcelsian. Tak6uchi (1958) reexaminedthe cr1'stalstructure of thesetwo types of "hexagonalcelsian" in detail and showedthat B-"hexa- gonal" celsian,which is stablebelow 300" C., is truly orthorhombic;and a-hexagonalcelsian, which is stable above 300oC., is hexagonal(D46h). No significant differencecan be observed between the r-ray powder patterns of the a- and the B-form except peak shifts owing to lattice expansion.The similarity of r-ray patterns demonstratesthe similarity of the basicframeworks of thesetwo kinds of celsian.Neither hexagonal nor orthorhombic celsianhas beenfound in nature. X-ray powder data of natural cymrite are almost identical to those of synthesized hexagonal celsian as shown in Columns D and E in Table 1. This similarity is believedto show that these two materialshave al- most the samefundamental structure. Another kind of barium-aluminum silicate, monoclinic celsian, has been described,particularlv in manganiferousore deposits(Bauer and Palache, 1926;Larsen et al., l94l; Meier, 1939; Schaller,1929; Segnit, 1946; Spencer,1943; Vermaas, 1953; Villiers, 1951; Yoshimura, 1936, 1939).The stability relation betweenhexagonal celsian and monociinic celsianhas never been clarifi.ed. It is well known that monoclinic celsian and monoclinic potassium feldspar form a complete or partial solid solution through the mineral hyalophane,which is intermediateboth in chemicalcomposition and in phvsicalproperties (Strandmark, 1903, 1904; Taylor etol.,1934;Vermaas, 1953). The present writers have recently succeededin synthesizinga new phase,KAISiaOE'HzO, from potassiumfeldspar under high water pres- sure conditions (Seki and Kennedy, 1964).The r-ray powder patterns and unit-ceil dimensions calculated for this new phase are quite similar to those of cymrite presented by Smith et al. (1949a). In order to study the relationbetween two possibiesolid solution series: c1'mrite (BaAlSiaOs(OH))-KAlSi3Os.HzOand monoclinic celsian CYM RITE-CELS I A N RELAT ION S

d.9 _o N q =- E 934e O ,h> 5

N

ts Z 'a U) ^o"b.J 9 6"d.F I ix?; Itro^ @ q q q q q q @q ..! €lq I I { I EIts O N g I N d N A

iF o@oN O+NH 'lE a>

:$6nbE60 r<€+ooohooo. q91? e.: I ": : : SNNNd Ztl aZ EA P.Z i+ AnOO@{^tO >:i E. J4 d14 (t -l Z

OOtsN€ 3a o€€Ne iz 27

Ei :@€r3N66 f, fr 3 39fr3 R K e = Flil r$€NNNN t4z

n< o€ohND€o6 zn ^ ,)p io+N: /4 'ia 6"' @ z ! E dv c q';o qr ,e r€oo€b4oN-:e DER: 3 S 3 5 F F r+i.id.i-.iq-.;

F .$e* d! - € tsr i I o:- ^l r : + ^ * lq N € "a:19 € o o o 4 4 X +ONdNN l

Fl

l- 98:E1393ioo-'F- F o c-o-.--=-cR59RS 1410 Y. SEKI AND G. C. KENNEDY

ll

'6n E -* V"e.i 9 ;";+ +l+l -i il EbE; qo fE x l. €6-Z @a "s | .q ! rqE | |

=JO. I OF e 6 : -il ^l x+ x os r | 3 fr 3 "al=l I

o9 di "

zz I z3 ,

NO F diN n : Y^. PeE.i 9:q"l ; i ov RREIi .hiqK vr u.E r

Yv =;6ll

R oo 'd; N r9:? :s= t E. i g€ ae9 h-b iltl zz

=.2 NN + O ++ iO __ N1 + do -o o^ 1E ix NO O i C- N- hv rc CYMRITE-CELSIANRELA:IIONS I4II

(BaAlzSizOs)-sanidine(KAlSisO8) and particularly to determine the physical conditions under which cymrite and Ba-bearing monoclinic feldspar are stable, we performed experimentalwork on the chemical reactionshown below:

celsian f sanbornitel2quafizf water - 2cymrite BaAhSizOs BaSizOa SiOz HrO BaAISiaOe(OH) under wide rangesof temperatureand water pressure. In the presentpaper, the solid solution relation betweencymrite and hexagonalcelsian and the stabiiity field of monoclinic celsian will be describedand their geologicalapplications will be discussed.

ExprnrltBlrrAr, TECHNTeuESAND Sr,tnrnrc Marnnrrlr,s Most of the experimentalwork in the present study was carried out in the piston-anvil,"simple squeezer,"apparatus which has already been describedin detail by Griggsand Kennedy (1956)and by Pistorius el ol. (1e60). Three kinds of starting materials were used. The first was a glass having the approximate chemical composition of cymrite. The glasswas made by heating a finely ground mixture of the following materials to 1300" C. followed by rapid quenchingin an air blast: Ba(oH)z.SHeo 51.6% by weight AlzOr.xHeO (x = 1) 9.8/6 by weight SiOr.xHrO (x :. 1) 38.5/6 by weight

The second starting material was an unfired mixture of the above hydrated materials. The mixture was ground under water and dried. This procedurewas repeated8 to 10 times in order to completelyhomog- enize the mixture. The third was mixtures of (1) hexagonalcelsian, sanbornite and di- barium-trisilicate (Ba2Si3O) and (2) monoclinic celsian, sanbornite and BazSiaOa.These mixtures were prepared in the simple squeezerfrom oxide mixes having the chemical composition of cymrite. Water was added to all mixesbefore subjecting them to pressure.The pressurewas raised to the desired value before heating in order to retain the water in the sample. We failed to synthesize any cystalline phase from the glassstarting mixture even when the glasswas kept at 700oand 30 kb pressurefor 21 hours. The oxide mixture of cymrite composition, on the other hand, readily transformed into a crystalline phase assemblage.At above 500o C. crystallization of the oxide mixtures was complete within fifteen minutes. 1412 TI. SEKI AND G. C. KENNEDY

Mixtures of hexagonalor monoclinic celsian,sanbornite and dibarrum- trisilicate formed from oxide mixture were used to confirm the reversi- bility of the two modifications of celsian along the boundary between the hexagonaicelsian solid solution* sanborniteI qtartz and monoclinic celsian* sanbornite* quartz fields. The products were identified chiefly bv means of r-ra1' powder pat- terns taken on a Philips diffractometer,using Cu radiation; as well as by means of the polarizing microscope.The r-rav patterns of starting ma- terials and synthesized minerai assemblagesare verl' distinctive and little difficultl' was encountered in identifying even relatively small

3 5l

EIpl tl

Frc. 1. Stability fieid of cymrite-celsian and their associated minerals formed by the chemical reaction of 2 cymriteicelsianf sanbornite+2 qtartz+Hzo' Numbers represent the duration (in hours) of the runs. amounts of eachphase. Quartz powder or siliconpowder was used as an internal standard,in order to measureaccuratelv the 20 CuKa valuesof diffraction peaks. RBsur-rs Figure 1 showsthe stability fields of some mineral assemblagesasso- ciated with the chemical reaction, celsian*sanbornite*2 quartzf H2O 32 cymrite.

Cymrite-heragonalcelsion Column A in Table 1 shows the r-ray powder data, unit-cell dimensions, unit-cell volume and optical properties of cymrite synthesizedfrom oxide mixture of cymrite composition.These data are practically the same as those of natural cymrite (Col' D in Table 1). With decreasingpressures and/or increasingtemperatures' r-rav data and optical propertiesof the mineral having the cymrite structure grad- ually approach those of hexagonalcelsian (Cols. B, C, E and G, Table CYMRITE-CDLSIANRELATIONS 1413

1). A hexagonalmineral formed from the oxide mixture of cymrite com- position at 725oC. under water pressureol 2 kilobars and associated with sanborniteand qvartz (Col. C, Table 1) showspracticaliy the same ph.vsicalproperties as thoseof pure hexagonalcelsian synthesized under room pressurefrom an oxide mixture having the chemicalcomposition of cl.mrite and celsianglass (Cois. E, F, Table 1). Column G in Table 1 containsthe r-rav data reported by Thomas (1950)for the "monoclinic celsian" synthesizedby him. A comparisonbetween this column and Table 2 clearll' shows that his "monoclinic celsian" actually was not monoclinic but was hexagonalcelsian. The most distinct r-rav diffraction peaks of cymrite are (101), (102) and (110). Among these three peaks, 20 values of (102) and (110) dif- fractions should clearlv and regulariv shift with the transition from pure cymrite to pure hexagonalcelsian. Figures 2 and 3 show the shift of 2d CuKa valuesof thesetwo r-ray diffractionsof cymrite-hexagonalcelsian mineral seriessynthesized under a wide range of physical conditions. It is clear, from inspectionof Figs. 2 and3, that the chemical poten- tial of all componentsin this difficult and complex system are not chang- ing as regulariy as might be desired.This probably stems from slight non-homogeneity of our oxide starting mix and from failure to retain water pressureexactly equal to confining pressure.More or less, the trend in the systemis clear,with cymrite forming at high water pressures and low temperaturesand hexagonalcelsian forming at low water pres- suresand high temperatures.In addition, increasingchemical potential of silica with decreasingchemical potential of alumina would causethe stable field of cymrite to expand to mr-rchlower water pressures.

Monoclinic celsian. Monoclinic celsian was formed from the oxide mix- ture of c1'mritecomposition under relatively high temperatureand low pressure conditions (Fig. 1). Monoclinic celsian is associatedwith sanbornite and quartz as in the case of hexagonal celsian. X-ray data, unit-cell dimension and unit-cell volume of monoclinic celsian are shown under Column A in Tabie 2. Column C is r-ray powder data of "3AlzOr.2SiO:(mullite)" synthesizedby Thomas (1950)in his study of BaO-AlzOrSiOusystem. He said "the r-ray data lead to the conclusion, though sucha conclusionis open to questionand deservesfurther study, that all the points are probably mullite." It is clear, however, that his mullite is actually monoclinic celsian.

Sanbornite.Sanbornite (BaSizO) wasformed in associationwith cymrite- hexagonal celsian series, monoclinic celisan, dibarium-trisilicate and qtartz in our experiments. r4I4 Y. SEKI AND G. C. KENNEDY

Tl-rll 2. X-nnv Poworn Dara, UNrr-Cnlt DrltowsroNs aNl Uxrr-Crr,r VoruMss or MoNocrrNrc CnsI.tl+s

Synthesizedmonoclinic Celsian Natural Synthesized 800', 9 kb monoclinic Celsian "3AlzO:'2SiOz" (this paper) (Vermaas, 1953) (Thomas, 1950)

Measured Calculated d(A) d(A) d(A)

o20 651 4l o..)l 6.39 40 o2l 4.59 18 460 4.57 20 4.6 130 3.79 38 3.79 3 380 221 3.5s 33 3.56 112 J. JI 80 .t.Jl 3.M 60 .t 4J

220\ 3.336 Quartz (101)1 3.33 v.s. 2o2J J.JZI )," 002 3 273 74 J.ZIO 70 3.25 S 040 3.259 3 -262 Isza 131 3 014 75 3.014 3.00 70 3.00 m 041 2.912 30 2.916 2.901 40 2.91 201 2.834 10 2.836 132 2 795 .ttt 2.794 2.758 50 2.77 221 2.603 30 2.602 24r 2.589 2 589 2.574 r00 2.57 203 2.401 10 2.404 2.406 30 T13 2.354 10 2.354 042 2.312 15 2 3r2 2.316 30 223 2.259 8 2 259 2.263 20 r32 2.251 t2 2.252 152 I n1A 18 1 11A 060 a 11A t.) z-\lJ 2.167 100 T52 Quartz(200)1 2 -122 2.104 10 311 2.080 10 2.082 2.056 10 71n 2.021 20 2.O22

o: 8.63(1) A o: 8.63A b:r3.10Q) A b:13.10A c: 7.29@) A c: 7.2sit B:116" p:116" unit-cell volume 185.2 A3 unif.cell volume 185.0 A3

t Hidden by peaks of associated quartz.

X-ray powder data and other physical properties of our sanbornite are representedin'fable 3. BaSizOswas first synthesizedby Bowen (1918). Eskola (1922) found that synthesized BaSizOsis orthorhombic. In 1932, Teer,r 3. X-nav Pomnn Dera, UNrr-Cnn DnmNsroNs eNo UNrr-Crr-r,Vor,ulrns ol. SA.NeonNrrBs

Synthesized Sanbornite Natural Sanbornite (This paper) (Douglass, 1958)

d(A) d(A) hkr

6.74 002 6.79 30 o02 505 ot2 5.08 30 012

Hex. Celsian(011, 101)r 3.97 100 110 3.81 16 to2 3.82 10 102 3.419 60 112 3.422 50 Itrz l'004 Quartz (101)r 3.342 70 022 3.227 36 103 3.226 20 103 3 .088 100 014 3.O92 014 2.980 20 113 2.980 .l 113 2 .883 16 12l 2.888 5 t2l 2.723 40 104 2.720 55 Itzz | 104 2.710 60 122

[Iex. Celsian (003)1 2.574 l.) tt4 2.540 l2 024 2.541 023 2.395 12 032 2.394 032 2.326 8 105 z .tzl 105 2.316 20 200 2.317 t.) 200 2.281 201 i rso Hex. Celsian (013, 103)t 2.236 10 ( l1ls 2.225 45 124 2.226 30 r24 ( tnc 2.r92 28 2.193 t.) Izoz \zrr l)11 2.160 12 016 2 -162 t< 016

Quartz (200)l 2 130 40 132 2.108 28 212 2 109 10 2t2 2.O38 16 034 2.O38 5 034 2.026 106 2.025 t.) 106 .995 t2 125 ( rt< 1 990 10 .991 l.) 2r3 ) rr a .921 25 041 |.921 10 041 902 27 222 1 903 10 222

Hex. Celsian(113)I 1.850 J2r4 \o+z 22s 1.793 28 126

a: a.$ Q) ir o: 4.$ A 6: 7.6e(o) A b: 7.69A c: 13.50 A d:IJ.5.t A Unit-cell volume 120.3 A3 Unit-cell volume 120.4A3

1 Peaks hidden by the presence of peaks of celsian and quartz crystals associated with sanbornite. t4t6 Y. SEKI AND G, C. KENNEDY

-

TElTIPERATUREPC)

Frc. 2. X-ray difiraction peak of (110) of cymrite-hexagonal celsian solid solution serres. Numbers represent the 29 CuKa value (example, 57:33'57 degrees). Rogers discovereda natural occurrenceof BaSizOsin associationwith qxartz, monocliniccelsian, diopside, gillespite and tourmaline forming a vein in metamorphic rocks, and proposed a new name sanbornite. Recentll,Douglass (1958) has studied in detail the crystal structure and related solid solution of sanbornite and has found that this mineral, previouslydescribed as triclinic b-vRogers (1932), is truly orthorhombic. T'he presenceof two polymorphic forms of BaSi:Os(high temperature and low temperatureforms) hasbeen reported by Roth and Levin (1959). They stated the transition temperatureof thesetwo forms to be 1350"C. at room pressure.Sanbornite synthesized at 300-850" C. in our run is believed to be their low temperature form.

GI oS

ol ozf o27 'lHI cl

al

oaoa

EleE!4Mgfcl

Frc. 3. X-ray diffraction peak oI (102) of cymrite-hexagonal celsian solid solution series. Numbers represent the 20 CuKq value (example 30:30.30 degrees). CYM RI TE-CELS I A N RELAT ION S L+LI

42 4t 40 39 3a 57 36 36 34 }r 32 3t n 28 - E-25-24-gd 2-l " 28(CuKo)

Frc. 4. X-ray diffraction figures of 2 cymrite3hexagonal celsianfsanbornite*2 quartz+Hro reaction at various physical conditions as follows: (1) 410., 29 kb, (2) 52O., 20 kb, (3) 630o C., 17 kb, (4) 500' C., 10 kb, (5) 560. C., 7 kb; Ba: Cymrite-hexagonal celsian solid solution, Bd: Sanbornite, Bt: BarSi:Os,e: euartz.

It must be noted that the relative amount of sanbornitesynthesized in our runs, as estimated from the intensity oI fi-ray diffractions, grad- ualll'decreaseswith the increaseof water pressure andfor decreaseof temperature (Figs. 4, 5 and 6). 1418 Y. SEKI AND G. C. KENNEDY

Y

TEMPERATURETC}

Frc. 5. Diagram showing the variation of the ratio *am#Hs+fiih* ^- X100 in the chemical reaction of 2 cymrite3celsian*sanbornitel2 quartztMza at various physical conditions.

c E

IgeEr4IgsErcl

Frc. 6. Diagram shor,vingthe ratio of I (014) sanbornite x 100 I (102)cym.-cel. s. s. + I (014)sanbornite + I (d:3.711) Ba2Si3Os in the chemical reaction of 2 cymrite-celsian*sanbornitel2 qtartzlHzO at various physical conditions. CYXIRITE-CELSIANRELATIONS I4I9

Dibarium-lri,silicate (Bazsi^os). The appearance of dibarium-trisilicate (BazSiaos)is limited to higher pressureand lower temperatureregions of the field of cymrite-hexagonalcelsian series*sanbornite in Fig. 1. X- ray powder data of dibarium-trisilicateassociated with cymrite-hexag- onal celsianseries and sanbornitein our runs are shown in Table 4. The intensitv ratio oI r-ray difiraction peaksclearly show that in our runs the amounts of sanbornite and dibarium-trisilicate vary inversely with changing temperatureand pressure.rt is also noteworthy that di- barium-trisilicate has never been found in associationwith quartz or other silical minerals in our experimentalwork. Thus, with increasing pressure and decreasingtemperature the following chemical reaction goesto the right:

2 Sanbornite (BaSLOs)=dibarium-trisilicate (Ba:Si:Os)*SiOz The SiOzthus liberated is used to form cymrite from hexagonaicelsian. winchell (1931)concluded from his interpretation of the data of Eskola (1922)that BarSisOsand BaSirO5form a continuousseries of mix-crystals. Levin and lJgrinic (1953)also said that BauSiaOa and BaSizO6form a com- plete solid solution though the r-ray data did not conclusiveryshow a solid solution series.Keler and Glushkova (1956),Glushkova and Keler (1957) and Roth and Levin (1959) later denied the presenceof a solid solution relation betweenthese two components.our experimentalwork clearly showsseveral double peaks which indicate the presenceof both of thesetwo phasesin the samerun and thus tend,to substantiatethe latter hypothesis.

Quortz. Quartz is always associatedwith hexagonalcelsian, monocrinic celsianand sanborniteformed at relatively low water pressuresin our runs. X-rav powder and microscopedata show, with increasingwater pressureand/or decreasingtemperature, the amount of quartz associated with these minerals generally decreasesland finally under high water pressures quartz disappears and dibarium-trisilicate appears in associa- tion with cvmrite-hexagonalcelsian series, rich in cl.mrite molecule,and sanbornite.

SutvrMqnvol RBsulrs Minerals or phasesappearing in the chemical reaction: hexagonalor monoclinic celsian*sanbornitef 2 qtartzfwaterj2 cymrite are as follows:

hexagonal celsian monoclinic celsian sanbornite quafiz 1420 Y. SEKI AND G. C. KENNEDY

Tllrr 4 X nev Po'woot Dera ol Ba2Si3O8

BarSisOsassociated u'ith cymrite- BazSisOs Ba:SLOs hex. celsian s.s and sanbornite (Levin el al.,1953) (Austin,1947)

d(A) d(A) d(A) I

a 11 l.) 700 12 6.06 l1 6.05 6 5.00 o 4.40 4.33 6 393 l.l 3 7rl 3.79 100 375 100 3.651 3.7r /.) 368 80 10 3.51 o 344 10 |,0, Sanbornitel J.JJ 87 330 70 25 3.28 76 3.26 42 3.15 48 .].IJ 35 o Cym.-hexagonalcelsianl 2.99 2.756 25 2.78 2.78 75 2.621 JJ 2.581 48 2.6r 8 260 o o Cym.-hexagonalcelsian 2.55 2.41 8 240 9 2.345 9 2.36 35 2.35 l+ 2.268 nA 228 45 2.27 23 Cym.-hexagonal celsianl 2.23 JL 223 13 Cym. hexagonal celsianl 221 24 2.21 2.r43 2.r4 60 2t4 4l 2.097 25 2.O9 22 2.08 18 2.O43 2l 2.04 12 2.O4 t2 1n Cym.-hexagonalcelsianr 1 981 20 1.98 t.919 25 1.918 14 r91 l4 1 881 8 1.88 7 1 1 865 10 186 1.817 20 r.825 t4 1.83 l6 |.784 9 r.789 25 |.79 20 Cym.-hexagonalcelsianr 1.760 11 r76 16 l. ,/4J 11 Cym.-hexagonal celsianl t.7r 8 l.) 1 654 22 1.6s 12 14 | 636 10 163 6 1.617 a 1.62 6 Cym.-hexagonal celsianr 1.60 6 1.58 6 Cym.-hexagonal celsianl 1 557 1l | .)o 8 1 514 10 1.518 t52 8 I 504 8 1.513 1.51 6

I Hidden by cymrite-hex. celsian and sanbornite' CVMRITE.CEISIAN RELATIONS I42I

Pure cymrite was synthesizedunder high water pressureand/or low temperatures,but p're hexagonalcelsian was formed onl1, under low water pressures(Fig. 1). Figure 7 showshow the quantitative ratios of cl,mrite-hexagonalcel- sian'series,sanbornite and dibarium-trisilicate vary under wide rangesof temperature and water pressure.From this Fig. 7 it is clear that the amount of cymrite-hexagonalcelsian series formed in our runs generally increaseswith increasingwater pressure.

- !

IEIPCNATUiEFCI

Irrc. 7 Diagram showing the variation of the ratio of

- ____I!g) cymrite hexagonal celsian s.s. I (102)cym.-hex.cel-.- + r (014);"b;it" +ld--7rt su,S;,O, in the chemical reaction of 2 cymriteJcelsianfsanbornitel2 quartz*HzO at various physical conditions.

Figures8 and 9 representthe relationsbetween the amount of cymrite- hexagonal celsian seriesformed in our experimental work and 20 CuKa valuesof (102)and (110) diffractionsof this mineral series.Both of these figures clearly show that the composition of cymrite-hexagonal celsian solid solutionbecomes richer in the cymrite moleculewith the increaseof the amount of the solid solution in the experimentalproducts (Figs. 2, 3, Table 1). Drscussron The volume relationsin the chemicalreaction are asfollows:

hexagonalfsanbornite l2quartz +water= 2 cymrite ceisian BaAlzSizOs BaSi:Os SiOz F{rO BaAISLOs(OF{) molecularvolume 19043 120As (3gx2)As 30A3 (1g9x2) A3

total volume 416 A3 378A3 t422 Y. SEKI AND G. C, KENNEDY

The chemical reaction, in going from the left to the right, involves the addition of water. Thus high water pressure as well as high hydrostatic pressurepromote the formation of cymrite. As previouslynoted, we found clear evidenceshowing a solid solution relation betweenpure cymrite and pure hexagonalcelsian by the follow- ing substitution:

At3+= [Si(oH)]3+ With substitution of Al3+by Sia+a spacebetween the (AlzSiz)Ossheets of hexaqonalcelsian must have beenoccupied by (OH)-'

/ I

o o

Inoacvx.xex ctr ss x too Itoacvx'rer cer ss + kotr)selaonrnt+ l(d'l ?nc oolsilot)

Frc. 8. Diagram showing the relation between 20 C:uKa of (102) and the amount of cymrite-hexagonal celsian solid solution formed by the chemical reaction of 2 cymrite *celsian*sanbornite*2 quartz*HrO at various physical conditions'

Thus with the increaseof water pressurethe mineralassemblages in the chemicalreaction we have studied changeas f ollows: (1) hexagonal celsian*sanbornitef quartz (2) (cymrite-hexagonal celsian) f sanbornite+Ba2slo3 solid solution (3) (cymrite-hexagonal celsian) f dibarium-trisilicate solid solution (4) cymrite Our experimentalwork showsthat in an anhvdrous svstem or under Iow water pressures hexagonal and monoclinic modifications of BaAlzSizOaare stable at high and low temperatures respectively.At atmospheric pressure,hexagonal celsian may be stable belorv about 700o C. However, natural occurrenceof hexagonalcelsian has never been re- ported. Similar observationsof hexagonalanorthite were made by Goidsmith CYMRITE-CELSIANRELATIONS 1423

and Ehlers (1952). Goldsmith and Ehlers (t952) and Donnay (1952) stated that hexagonalanorthite would not be isostructuralwith hexag- 'Iable onal celsian.However, as shown in 5 of the present paper, the similarity of r-ray powder patterns suggeststhat these t*o hexagonal phasesare isostructural.The absenceof an *-ray difiraction peak corre- sponding to (00a) of hexagonal anorthite in our hexagonal celsian may suggestcelsian has twice the unit-cell dimensionc of the anorthite. The experimental work of Goldsmith and Ehlers (lgsz) shows hexagonal

r,* r,

Frc' 9. Diagram shor'ving the relation between 20 C,tKa of (110) and the amount of cymrite-hexagonal celsian solid solution formed by the chemical reaction of 2 cymrite Scelsianf sanbornitef2 quartz*Il:O at various physical conditions.

anorthite to be stable below 350" C., but hexagonalanorthite has never been reported to occur in nature. It is noteworthy that monocliniccelsian is transformedinto the hexag- onal form, moderatelyrich in cvmrite, under more than 4 kb water pres- suresin our experiments.Below 4 kb, monocliniccelsian does not change into the hexagonalform even at 650oC. for 33 hrs. Thus it is quite prob- able that monoclinic celsian persists into the low temperature field if water pressureis not high. Sanborniteis a rather rare mineral but is usually associatedwith mono- clinic celsianand quartz (Rogers,1932). Only one occurrenceof cymrite in nature has been reported (Smith et al.,1949a).The high water pressureor high chemicalpotential of water which is necessarvfor the formation of cymrite is believedto be a reason why this mineral is so rare in nature. The physical conditionswhich pre- r424 Y, SEKI AND G. C, KENNEDY

Tasln 5. Srurr-arrrv ol X-na.v Powonn Dere ol Hnxacorer' ANonrrrrrn ro Tsose or Hrx.qcoxer- CrrsraN

Synthesized hexagonal anorthite Synthesized hexagonal celsian (Davis el a1,.,1952) (Takduchi,1958)

hkI d(A) hkl d(A)

N2 7.37 001 77e t12

100\ 100\ 4.43 010.1 010J

102\ 101\ 3.80 3.963 or2) 011J

004 > 100

104\ 012\ 2.84 0r4) 1o2J

110 2.555 9 110 2.659 006 2.456 t4 003 2.602 22 t12 2.414 3 111 020\ .) 2ooJ 2.304

106\ 103\ 016j 0131

202\ 201\ 2.1r9 o22J o2rJ

008 > 100 004 tt6 9 113

108\ 104\ 1.701 018,l 0r4)

o:5.110+0.02A o:5.313+0.05A c:D17.369+0.02A c:7.805+0.05A

1 Davis el al., noted 2.417 but is probably a misprint.

vailed in the formation of the cymrite in wales ma)r have been below 300" C. and above5 kb water pressure.

AcrNowr.noGMENT We were given much valuable advice and criticism by Dr. R. C. New- town of The lJniversity of Chicago,Dr. R. C. Colemanand R. C. Erd of CY XI RI TE-C ELS I A N RELA TION S 1425

the u' s. Geologicalsurvey at Menlo park, california to whom our sin- cerethanks are due. Dr. L. H. Adams and Mr. T. J. Thomas of the uni- versity of Californiaassisted us greatllrin the progressof this work. Partial financial support by the GeophvsicsBranch of the office of Naval Research,Mineral Svnthesis,Nonr 233(28)is gratefuily acknowl- edged. RrltnrNcns Ausrtn, (1947) A. E. X-ray diffraction data for compounds in systemsLirO-SiO, and BaO_ SiOz.Iouy. Am. Ceram. Soc.3O,27B-22O. B'turn, L. H. eNo c. P,lracuB (1926) Hyalophane from Frankrin Furnace, New Jersey. Am. M ineraL ll, 172-17 4. Bo'r'rx, (1918) N L. crystals of barium-disilicate in optical glass.Jour.wash. Acad. Sci., 8, 265-269 Devrs, G. L. nNn o. F. Turr'B (1952) Trvo nerv crystalline phases of the anorthite com- position, . CaO.ALO: 2SiOr.Am. Jour. Sci.,Bouen VoI , 107 114. Drturn, E' aNo H. Lascrr (1930) Synthetische Untersuchungen iiber die Mischkristall- bildung des Barium- und Strontiumfeldspates mit orthoklas Anz. Ako.d wiss. wi.en, M ath. notuttuis s. KL., 201,203. DoNN.rv, G. (1952) Hexagonal CaALSiTO:.Acta Cryst.S, 153. Doucrass, R. M. (1958) The crystal structure of sanbornite, BaSLOa.Am. Minual. 43, s17-536. Esxora, P. (1922) The silicatesof strontium and barium. Am. rour. sci,.4,33r-32s. Gtusurove, B. B. eNo E. K. Krlnn (1957) Reaction between barium perioxide and sili_ condioxide.Zhur. Neorg. Khi.nx 2,1001-1006. Gorosurru, J. R. eNn E. G. Enrrens (1952) The stability relations of anorthite and its hexagonalpolymorph in the system CaAlzSirOu.HrO. Iour. Geol 60. 3g6 392 Gnrccs, D. T. aNo G. C. KnNNr:ov (1956) A simple apparatus for high pressures anrl tem_ peratures.Am. J out. Sci. 254, 722-7 35. Kn'rn, E' K. 'r.Nl v. B. Grusmove (1956) condition of formation of barium silicate Zhur. Neorg. Khim. t,2283 2293 KrNNtrv, G. c. ,q.NoR. c. Nrwrow (1963) SolidJiquicl and solid-solid phase transitions in some pure metals at high temperatures and pressures, solids under pressure, paul & Warschauer ed. McGraw-Hill Book Co.. 163-17g. LAR.EN, E. S., C. S. HunnunrJn., B. F. Burr, aNn C H. Buncnss (1941) Igneousrocks of the Highwood Mountains, part VI. BulI. Geol Soc.Am.52, 1g41_1g56. LnvrN, E. M. aNo G. M. ucnnrc (1953) The system barium oxide-boric oxide-sirica Jour. Res. NatI. Bnr. Stand 51,37-56. Mnrrn, A. E' (1939) Association of harmotome and barium feldspar at GIen Ricldle, Pennsylvania. Am. M,inu aI. 24, 540-560. Prsronrus, C. W. F. T , G. C. KrNNroy n*l S. SouarnafeN (1962) Some relations between the phase anorthite, zoisite and lawsonite at high temperatures and pressures.,4r2. Jout- Sci.26O, 44-56. Rocrns, (1932) A r. sanbornite, a new barium silicate minerar from Mariposa county, California. Am. M ineral lT, 761 122. porymorphism Rorn, R' s' .ryn E. M. Lrvrx (1959) in barium disilicate. Am. Mi,neror.44. 452-45J. scn.rrr.nn, w . T ' (1929) The properties and associated minerars of ginespite. Am. M.inerar. 14,319-322. Srcxrr, (1946) E. R Barium-feldspars from Broken Hill, Ner,v south wales. Mineral. Mag 27,16G174. 1426 Y. SDKI AND G. C. KENNLDY

high pressure Snxr Y. aNn G. C. KoxNnov (in press) The breakdorvn of potash-feldspar at and high temperature. mineral Surrn. W. C.. F. A. BeNNrsrnn mo M' H. Hnv (1949a) Cymrite, a new barium from the Benalit manganese mine, Bhiw, Carnarvonshtre' Mineral" Ma* 28' 676-681' from wales.,4finerol. Mog. seoNcnn, L. J. (1943) Barium-feldspar (celsian and paracelsian) 26,231-245. om celsian och andra baryt- Srn,uouerx, J. E. (1903, 1904) Bildrag till Kannedomen -133' faltspater. Geol F i;r. F drh., 25, 289-319,26, 97 BaAl'SizOs Ter<6ucnr, Y. (1958) A detailed investigation of the structure of hexagonal with referenceto its inversion. Min. Iour.2, 311-332' Tlvr-on,w.H.,J'A.DealysrrrnnANDH.STRUNZ(1934)AnX-rayinvestigationofthe feldspars. Zeit. Krist 87, 464-498. Tnouns, k. H. (1950)phase equilibrium in a portion of the ternarl' system BaO-Al:O.SiO:. Jour. Am. Ceram. Soe' 33' 35-4t4. S' W' Africa Vnnueas, F. H. S. (1953) A new occurrence of barium-feldspar at Otjosondu, andanr.raymethodfordeterminingthecompositionofhyalophane'Am.M'i.neral', 38, 845-857. S' W' Africa' Trans' GeoI' Vrrurns, J. E. or (1951) The manganeseores of Otjosondu, Soc.South AJriea 54,89-98. wiley & sons, wrNcner,r, A. N. (1931) Microscopi.c charocters oJ Arlifeial, Mineral,s. John N. Y.403. barium feldspar' Yosnrrr, B. emn K. Mersuuoro (1951) HiSh temperature modification of Iour. Am. Cerant'.Soc' 34,283-286. Yosnnnuna, T. (1936) A barium-fetdspars from the Kaso mine, Tochigi Prefecture' Japan (in Japanese).Iour. Geol' Soc.Iapan,43,877-9lO' mine' --- (1939) Studies on the minerals from the manganese deposit of the Kaso Japan Jour. Fac Sci' Hokkaido Unitt.4,315-451'

Monuscr'i'pt receiteil, February 7, 1961; acce4ledlor publication, June 8, 196:l'