American Mineralogist, Volume 7l, pages 547-556, 1986

High-temperature structure and crystal chemistry of hydrous alkali-rich berylfrom the Harding pegmatite, Taos County, New Mexico

GonooN E. BnowN, Jn., Bu,onono A. Mrr,r-stDepartment of Geology, Stanford University, Stanford, California 94305

AssrRAcr

The structure of hydrous Li-, Cs-, and Na-rich beryl from the Harding, New Mexico,pegmatite has been refined to an R factor of 0.050 using single-crystal X-ray methods. Alkaliions plus water molecules were located by difference Fourier methods in two sites withinthe channels paralleling c. Orientation of water molecules was determined from polarizedIR spectra. The resulting structural formula is

[W(I)o orW(Ds a6Css 6eIG.rQ r,]xlr[Nao 'Cao orMg orFefli'Eo ro]u'

[Al, o]vt[Ber rnLL 38410 0r]I"[Si5.e2AL.o8]ItOr 8,

where W(I) and W(I! represent Type I (H-H vector parallel to c) and Type II (H-H vectorperpendicular to c) water, respectively, and E represents vacancies. Li occupies the nonringtetrahedral site; water molecules, Cs, and K occupy the l2-coordinated channel site betweensix-membered silicate rings; and Na and Ca occupy the six-coordinated channel site withinthe rings.

The structure was also refined at 500, 800, and 24C, after heating, and cell parameterswere determined at 200, 300, 400, 500, 600, 700, and 800",C. The a and c cell parameterswere found to expand over this temperature range with the following coefficients: e r : 1.2 xl0-6 qC-' and e, : 3.1-1.5 x l0-6'C-'. The thermal expansion is consistent with changesin Gruneisen parameters, "y, and 7r, with increasing temperature and can be rationalizedin terms of (l) increases in volume of AlO. octahedra and (Be,Li)Oo tetrahedra, (2) changesin T-O-AI and O-T-O angles (T : Si,Be,Li,Al), (3) rotation of (Be,Li)Oo tetrahedra aboutaxes lying in (001), and (4) rotation of vertically adjacent six-membered rings in oppositesenses about [001]. Our results contrast with the thermal expansion of nonalkali hydrousand anhydrous synthetic beryl, emerald, and low-alkali hydrous cordierite. Heat treatmentat 800"C for 72h had little effect on the occupancies of channel sites; little dehydrationoccurred because the large alkali ions effectively plug the channels.

INrnooucrroN

Beryl is a beryllium aluminum silicate mineral oftenfound in granitic pegmatites in which it can occur as free-standing, euhedral crystals in "pockets" or as "frozen,"anhedral to euhedral crystals associated in either case withqtartz, K-feldspar, albite, muscovite, and a variety ofaccessory minerals. The beryl structure, illustrated sche-matically in Figure l, consists of six-membered rings ofsilica tetrahedra crossJinked by Be-containing tetrahedraand Al-containing octahedra to form a tetrahedral frame-work with open channels that parallel the c axis (Gibbset al., 1968). Pegmatitic beryls commonly contain varyingamounts of alkali-metal cations as well as alkaline earths,Fe, and other cations. Studies of compositional variations(see, e.g., Filho et al., 1973) and structural details (Evansand Mrose, 1968; Gibbs et al., 1968; Bakakin et al., 1969:

1 Present address: CR Exploration, 755 Gregg Street, Sparks,Nevada 89431.

0003404V86/030,14547$02.00 547

Hawthorne and Cernj', 197 7 ; Pice et al., | 97 6; Goldmanet al., 19781' Aines and Rossman, 1984) have shown thatthe open channels in the beryl structure can accommodatethese cations and water molecules similar to the channelsin cordierite (Hochella etal.,1979; Goldman etal., 1977;Carson et al., 1982).

The present study is concerned with the structure andcrystal chemistry of a high-alkali beryl that occurred as aeuhedral crystal "frozerf' or included within a smokyquartz vein in the microcline-spodumene-lepidolite-al-bite-muscorrite-quartz core zone of the Harding pegmatitein Taos County, New Mexico (Jahns and Ewing, 1976,1977). Considering Hawthorne and Cernj's (1977) struc-tural study of a Li- and Cs-rich beryl from the Tanco(Manitoba) pegmatite, our independent results on Li-Cssubstitution in beryl are discussed only briefly, and themajor emphasis is placed on (l) response of the berylstructure to temperature, (2) the role of channel constit-uents in the thermal expansion behavior of beryl, and (3)

548 BROWN AND MILLS: HYDROUS ALKALI-RICH BERYL

Table 2. Cell parameters of beryl HB-3 at eight temperatures

Temperature ("C) 4(E) c (A) v (L',)

2 4 "200"300 '400"500 '500" 2

600"700 "800 '5oo" 3

24" x

s . 236 (1 ) \9 .247 ( \ )9 . 244 ( r )9 .249 (r )9 . 2s1 (1)9 .246 ( r )9 . 2 5 3 ( 1 )9 . 256 ( r )9 .259 ( r )9 . 245 ( r )9 .233 ( t )

9 . 2 4 6 ( L ) 6 8 3 . 0 ( 2 )9 .252 ( r ) 684 .2 (2 )9 . 2s3 ( r ) 684 .7 ( 2 )9 . 2 5 7 ( 7 ) 6 8 s . 8 ( 2 )9 ,260 ( r ) 686 .3 (2 )9 . 2 5 4 ( L ) 6 8 s . 1 ( 2 )9 . 259 ( r ) 686 .5 (2 )9 .262 (7 ) 687 . 2 (2 )9 . 2 6 4 ( L ) 6 8 1 . 8 ( 2 )9 .2s2 (L ) 684 .8 (2 )9 .243 (L ) 682 .4 (2 )

Fig. l. Schematic illustration of the beryl structure showingthe hexagonal rings ofSi tetrahedra, designated T1, linked by Beand Li tetrahedra, designated T2. Possible positions for channelconstituents are also shown, although P6/mcc symmetry mustbe obeyed, on the average.

comparison of the structure and thermal expansion ofberyl and cordierite.

ExpnnrurNTAL DETATLS

A specimen of beryl from the Harding pegmatite was kindlyprovided for this study by the late R. H. Jahns of Stanford Uni-versity. This particular specimen was selected for structural studybecause an earlier wet chemical analysis (Table l) showed it tobe enriched in alkalis relative to other available, chemically ho-mogeneous, pegmatitic beryls. A polished section examined byelectron microprobe was found to be chemically homogeneous,

Table l. Chemical analysis of Harding beryl HB-3'

! t r t . Z N o r m a l i z e d - A t o n i c

N o . C a t i o n s / ] 8 O s s e n s"/ f rODOrEaOn

rllmtbers in puentheses ?ep?esent(l&) in the Last decinaL place.

zAfte! 96 hours.-Al'ter nearLng.

the estimted stmdntd ereox

with the possible exception of Li and Be, which were not de-tectable. Our microprobe analysis is consistent with the com-position in Table l. A clear fragment designated HB-3 and mea-suring 0.12 x 0.12 x 0.20 mm was examined by precessionphotography; precession photographs displayed Laue symmetryDuo and systematic absences of the type hhl, and h0l, I odd,which are consistent with space-group symmetry P6/mcc. Thecrystal was remounted in high-temperature mullite cement andsealed in an evacuated silica-glass capillary. The c axis was ori-ented parallel to the rotation axis of the goniometer head, andthe crystal was transferred to a Picker r,lcs-r X-ray diffractometerequipped with a furnace (Brown et al., 1973). The furnace wascalibrated to 1000"C by observing the beginning of melting of200-pm diameter crystals of five standard compounds placed inevacuated silica-glass capillary tubes. Cell parameters (Table 2)were obtained by least-squares refinement ofthe angular settingsof 28 automatically centered reflections in ttre 20 range 30o-50o.

Intensity data were gathered at 24€ using graphite-monochro-matized Mo K" radiation (36 kV, l6 mA) and a take-offangle of2. 5'. Approximately 500 symmetry-independent reflections with-in the 20 range 5o-65" were collected by using the 0-20 scantechnique, a scan rate of l' 20 per minute, and a scan range of2.5" plus the a, - a, dispersion. Background radiation was re-corded for l0 s at the high- and low-angle ends of each scan.Two standard reflections (500, 006) were measured every 27reflections to check crystal alignment and electronic fluctuations,both of which proved to be negligible over the course of theexperiment.

Cell dimensions were measure d at 24, 200, 300, 400, and 5 00qCin the same manner as for the 24C cell determination. Duringautomatic centering of reflections and intensity measurements attemperature, crystal temperature was recorded continuously us-ing a strip chart recorder and varied less than +10'C.

Intensity data were gathered at 500qC using the orientationmatrix determined from the auto-centering experiment at 500"C.Experimental parameters during data collection at high temper-ature were the same as those for the 24"C data set. Cell dimensionswere determined again at 500"C, after the 96 h required for in-tensity data to be collected at that temperature, and were sub-sequently determined at 600, 700, and 800"C. A third set ofintensity data was collected at 800qC after the crystal was equil-ibrated at this temperature for 72 h. Subsequently, cell parameterswere determined at 500 and 24C. A final set of intensity datawas collected at 24f in order to determine the effects ofpro-longed heating.

Oxide

S i O zBeOA l z O :F e z O :FeO

1 . 0 6 3. 4 6 4. 1 8 8. 0 0 1

5 . 9 22 . 5 92 . l 00 . 0 1

6 2 . 8 8 6 3 . 8 81 1 . 4 3 1 r . 6 118.92 L9 .22

o . 1 2 0 . L 2

llno tr.M C O O . l 7T i o 2 0 . 0 3L i z o 0 . 9 9N a 2 O 1 . 1 3

K z O 0 . 3 8c 6 2 o 2 , 2 IC a o 0 . 1 8I t z o 1 . 7 4

Tota l 100.18

o . i z0 . 0 31 . 0 11 . 1 5

0 . 3 9

0 . 1 80 . 0 0

1 0 0 . 0 0

. 0 0 4

. 0 0 8

. 0 0 3

. 0 9 8

0 , 0 2

0 . 3 8o . 2 L

0 . 0 50 . 0 9o . o 2

. 0 0 4

. o 3 4

. 0 1 8

rAnalgsLs ba L. E. Peck md R. E. steoens.z1Trde Deight pexeent nomlized to 100% after eacluding Hzo(See SchaL ler , e t a l . 119621) .

BROWN AND MILLS: HYDROUS ALKALI-RICH BERYL 549

Table 3. Refinement parameters for beryl HB-3 at threetemperatures

Table 4. Refined positional and thermal parameters for berylHB-3 at three temDeratures

Parameter 24"C 500"c 800 'c Alom Paramete !24" c t 24"C 500" c 800"c 24"c '

No. least-squared variables

Total No. non-equivalentref lect ions

N o . r e f l e c l i o n s 2

R 3

R (weighted) +

6 (unit weight observat ion)

31

4 2 7

352 329 4r2

0 . 0 9 3 0 . 0 7 0 0 . 0 5 3

o . o 1 7 0 . 0 5 8 0 . 0 5 0

2 . 2 3 7 . 1 4 1 . 6 5

x 0 . 5 0y 0 . 0 0z O . 2 58 r r O . o o 3 3 ( 7 ) 2B z z 0 . 0 0 2 5 ( 1 0 )B : : 0 . 0 0 3 1 ( 7 )Bv L lBzz

B ( . q ) 3 0 . 8 8 ( 1 0 )

x 0 . 6 6 6 7y 0 . 3 3 3 3z O . 2 5B r r 0 . 0 0 1 5 ( 2 )B z z F r rB : r 0 . 0 0 1 2 ( 2 )B r z \ B r r

B ( e q ) 0 . 3 8 ( 3 )

x 0 . 3 8 9 3 ( 2 )y 0 . 1189 (2 )z 0 . 0 0B r r 0 . 0 0 1 1 ( 1 )B z z 0 . 0 0 1 4 ( 2 )B : r 0 . 0 0 1 3 ( 1 )B r z 0 . 0 0 0 7 ( r )

B(eq) o .36(2)

x 0 . 3 0 5 7 ( 4 )y 0 . 2 3 s 0 ( 4 )z 0 . 0 0B r r 0 . 0 0 3 9 ( 5 )B z z 0 . 0 0 3 1 ( 4 )B : : 0 . 0 0 4 7 ( 4 )B r z O . O O 2 1 ( 4 )

B ( e q ) 1 . 0 2 ( s )

x 0 . 4 8 9 5 ( 3 )y 0 . 1 4 6 8 ( 3 )z O . I 4 4 5 ( 2 )B r r 0 . 0 0 3 4 ( 3 )B z z 0 . 0 0 4 1 ( 3 )B s s 0 . 0 0 2 5 ( 2 )B r z 0 . 0 0 2 4 ( 3 )B r : - . 0 0 1 8 ( 2 )

& z t - . 0 0 1 2 ( 2 )

B ( e q ) 0 . 8 6 ( 3 )

x 0 . 0 0y 0 . 0 0z O . 2 50 r r 0 . 0 0 6 8 ( 5 )B z z 8 r rB : e 0 . 0 0 4 3 ( 6 )B t z \ B t t

B ( e q ) 1 . 6 5 ( 9 )

x 0 . 0 0y 0 . 0 0z 0 . 0 08 r I 0 . 0 0 0 5 ( 1 2 )B z z B r rB s : 0 . 0 0 9 6 ( 2 9 )B r z \ B r r

B(eq) L I7 (32)

0 . 5 0 0 . 5 0 0 . 5 00 . 0 0 0 . 0 0 0 . 0 00 . 2 5 0 . 2 5 0 . 2 5o . o o 1 8 ( 1 0 ) 0 . 0 0 6 4 ( 1 4 ) 0 . 0 0 2 r ( 6 )0 . 0 1 9 8 ( 3 4 ) - . 0 0 2 5 ( 9 ) o . o o 2 6 ( 9 )0 . 0 0 3 5 ( 1 4 ) 0 . 0 0 6 2 ( 1 3 ) 0 . 0 0 1 8 ( 5 )\Bzz \Bzz \Bzz

2 . 2 0 ( 3 r ) 2 . 0 1 ( 1 6 ) o . s 9 ( 9 )

0 . 6 6 6 7 0 . 6 6 6 7 0 , 6 6 5 70 . 3 3 3 3 0 . 3 3 3 3 0 . 3 3 3 3

31

508

395

0 . 0 5 0

0 . 0 4 9

1 . 6 3

'AJ te! neaTLng.'Included Ln final cycle of reftnenent. Reflectims uithF < 3o(F) or dLth lFo-Fcl > IO ue?e rejected. Theye uere nomove thm 3 refLecti,ons satisfging the Latter condLtion ineach refLnenent.3 F = t / l F | - l F l \ / t l F I

| . c | , , - | ' o I

uR (u t . ) = xw( l ro l - l r " l ) t / r ' r l1%

Intensity data were corrected for Lorentz and polarization ef-fects, assuming a 500/o ideally mosaic monochromator crystal.Because of the r€latively small crystal size and low linear ab-sorption coefrcient @: 11.26 cm-'), no absorption correctionwas applied; differential absorption due to differences in trans-mission ranged from 0.80 to 0.87. Standard deviations were es-timated by using the formula suggested by Corfield et al. (1967)with a diffractometer factor of 0.04. Refinement of the 24, 500,800, and final 24"C stmctures was initiated in space Eronp P6/rucc by using the coordinates of Gibbs et al. ( 1968), neutral atomicscattering factors from Doyle and Turner (l 968), and anomalousdispersion corrections from MacGillavry and Rieck (1968). Theprogram RFrNE (Finger and Prince, 1975) was employed, a sec-ondary extinction correction was refined, and weights based oncounting statistics were assumed. Reflections with F < 36,(F) wereconsidered unobserved and were not used. Moreover, reflectionsfor which calculated and observed structure factors difered bymore than an arbitrary value of A-F: 10.0 were not included inthe refinement, though no more than three of these were foundin any given data set.

Difference Fourier maps, computed after refinement of the24Cdata set by using A. Zalkin's unpublished ronoln prog.ram,indicated electron density at two positions within the channels,C(l): (0, 0, Vr) and C(2): (0, 0, 0); the density at C(l) was ap-proximately four times greater than that atC(2). Cs, K, and HrOwere assigned to the larger C(l) site, whereas Na and the smallamounts of Ca, Mg, and Fe3+ were assigned to the smaller C(2)site, resulting in essentially featureless electron-density differencemaps at these sites. Li was assigned to the Be tetrahedral siteafter the models of Bakakin et al. (1969) and Hawthorne andCrm! Q977). kast-squares refinements of all data sets con-verged rapidly to the R factors listed in Table 3, and final posi-tional and thermal parameters are listed in Table 4. Selectedinteratomic distances and angles calculated using L. W. Finger'sunpublished program ERRoR are listed in Table 5, and observedand calculated structure factors for all four data sets are eiven inTable 6.2

2 To receive a copy of Table 6, order Document AM-86-299from the Business Office, Mineralogical Society of America, 1625I Street, N.W., Suite 414, Washington, D.C. 20006. Please remit$5.00 in advance for the microfiche.

' A f t e r h e a t i n g .2Nmb".. ln parentheses rePresent the est iMted standard ertor(1o) in the last decimal place.3Equivalent isotropic tenperature factors ( '&2)were calculated

u; ine the exDression of Hamil ton (1959).

Rpsur-rs AND DlscussroN

Structure and channel constituents

The room-temperature structure of Cs- and Li-rich ber-yl from the Harding pegmatite (HB-3) is similar to thestructure of hydrous and anhydrous synthetic beryls stud-ied by Gibbs et al. (1968) and Morosin (1972) and isvirtually identical to the structure of a Cs- and Li-rich

31 T2

473

3 1

4 4 7

c ( r )

c (2 )

o . 2 5o. 0041 (4 )B r r0 . 0031 (4 )l l B r r

r . 0 5 ( 7 )

0 . 3 8 9 2 ( 3 )0 . 1 1 9 4 ( 3 )0 . 0 00.0019 ( 3 )0 . 0 0 2 5 ( 3 )0 . 0 0 2 s ( 2 )0 .0012 ( 3 )

0 . 6 5 ( 4 )

0 . 3039 (9 )0 . 2340 (8 )0 . 0 00 . 0 0 6 8 ( 1 1 )0 .0039 (10)0 . 0 1 0 2 ( 1 0 )0 . 0 0 4 s ( 9 )

1 . 8 6 ( 1 4 )

o .4966(5)0 . 1456 (4 )0 . 1436 (4 )0 . 0056 (6 )0 . 0 0 6 8 ( 7 )0 . 0 0 4 3 ( 4 )0 . 0 0 4 8 ( 6 )- . 0 0 1 s ( 5 )- . 0 0 0 4 ( 4 )

r . 3 5 ( 8 )

0 . 0 00 . 0 0o . 2 50 . 0 1 3 3 ( 1 6 )F r ro , oo8o ( 15)L B r r

3 . 1 8 ( 2 7 )

0 . 0 00 . 0 00 . 0 0- . 0 0 8 3 ( 4 2 )B r r0 .0034 ( 39)\ B t t

1 . 80 (56)

o . 2 5 0 . 2 50 . 0 0 4 6 ( 3 ) 0 . 0 0 r 7 ( 1 )B r t B r r0 . 0 0 2 7 ( 3 ) 0 . 0 0 1 7 ( 2 )\ B t r \ B t t

1 . 0 9 ( 5 ) 0 . 4 8 ( 3 )

0 . 3 8 8 6 ( 2 ) 0 . 3 8 9 1 ( 2 )0 . 1 1 8 3 ( 2 ) 0 . 1 1 8 6 ( 2 )0 . 0 0 0 . 0 00 . 0 0 3 2 ( 2 ) 0 . 0 0 1 3 ( 1 )0 . 0 0 3 8 ( 3 ) 0 . 0 0 1 2 ( 2 )0 . 0 0 2 7 ( 2 ) 0 . 0 0 1 4 ( 1 )0 . 0 0 1 8 ( 2 ) 0 . 0 0 0 6 ( r )

0 . 9 1 ( 3 ) A . 3 7 ( 2 )

0 .3042(7) o .3o5s(4)0 . 2336 (5 ) 0 . 23s1 (4 )0 . 0 0 0 . 0 00 . 0 0 7 9 ( 9 ) 0 . 0 0 2 8 ( 4 )0 . 0 0 4 1 ( 8 ) o . o o 2 2 ( 4 )0 . o o 9 o ( 7 ) 0 . 0 0 5 1 ( 4 )o . o o 4 2 ( 7 ) 0 . 0 0 1 8 ( 4 )

1 . 9 2 ( 1 0 ) 0 . 9 s ( 5 )

o . 4 9 7 4 ( 5 ) 0 . 4 9 8 2 ( 3 )0 . 1 4 6 s ( 3 ) 0 . 1 4 6 8 ( 2 )0 . 1438 (3 ) O, 1447 (2 )0 . 0 0 8 3 ( 5 ) o . o o 3 8 ( 3 )o . oos6 (5 ) o . oo4o (3 )0 .0048(3) o .0026(2)o1oo36(4) 0 .0024(3)- . 0 0 3 9 ( s ) - . 0 0 1 8 ( 2 )- . 0 0 1 3 ( 4 ) - . 0 0 1 4 ( 2 )

r . 7 2 ( 6 ) 0 . 9 1 ( 3 )

0 . 0 0 0 . 0 00 . 0 0 0 . 0 00 . 2 5 0 . 2 5o . 0 2 1 2 ( 2 0 ) 0 . 0 0 6 7 ( s )B r r B r r0 . 0 1 0 6 ( 1 6 ) 0 . 0 0 5 2 ( 6 )23r t ' t3 r t

5 . 8 ' 1 ( 2 9 ) 1 . 7 3 ( 9 )

0 . 0 0 0 . 0 00 . 0 0 0 . 0 00 . 0 0 0 . 0 00.ooo2(20) 010059(18)B r t 8 r r0 . 0 2 9 3 ( 8 0 ) 0 . 0 0 2 9 ( 1 9 )\ B r r \ B t r

3 . 3 8 ( 5 0 ) 1 . 3 4 ( 3 9 )

550 BROWN AND MILLS: HYDROUS ALKALI-RICH BERYL

Table 5. Selected bond distances and angles' for beryl HB-3 at three temperatures

(24" c) (s00 'c ) ( 800 ' c )/ a L o r ^ l r ^ -\ z + u o r L r !

heat ing)

In te ra ton ic d is tances in S i0a te t rahedron

si-o1s i-01Si-O2 (2 bonds)ol-0101-02 (2 bonds)o2-o2Mean S i -O d is tance

o-S i -O ang les in S iOq te t rahedron

ol-s i- 0202-s i-ol01-si-0102-si-o2

Interatomic distances in Be0a tetrahedron

Be-O2 (4 bonds)O2-O2 (2 bonds)02-0202-02

o-Be-O angles in BeOa tetrahedron

O2-Be-02O2-Be-OZ02-Be-Q2

Interatomic distances in M1O5 octahedron

M-O2 (6 bonds)02-02 (2 bonds)02-02 (2 bonds)02-O2 (2 bonds)

Angles in Ml06octahedron

o2-M-02Ow-M-O2o2-M-02

Interatomic distances ln C(1)0rz s i te

C (1 ) -o1 (12 bonds )

Ang les i n C (1 )0 r z s i t e

01-c ( 1) -o101-c ( 1) -0r01-c ( 1) -01

Interatomic distances in C(2)Os s i te

c (2 )01 (6 bonds )C(2)-C(1) (2 bonds when Na in C(2) and

H z o i n c [ 1 ] )

Ang les i n C (2 )Oe s l t e

ol-c (2) -01ol-c (2) -01ol-c (2)-o1c (1 ) - c ( 2 ) - c ( 1 )

r . 6 , rQ )21 . 6 0 3 ( 3 )r . 6 1 4 ( 2 )2 . 559 (4 )2 .675 (3 )2 .617 (4 )1 . 610 (3 )

1 0 8 . 3 3 ( 1 0 )1 1 1 . 3 2 ( 1 1 )1 0 5 . 5 8 ( 2 s )r 1 1 . 6 9 ( 1 8 )

r . 6 7 5 ( 2 )2 . 7 2 4 ( 4 )2 . 393 (4)3 . 0 5 0 ( 4 )

9 r . 20 ( r 4 )1 0 8 . 7 5 ( 1 4 )1 3 1 . 1 1 ( r s )

L . 9 1 3 ( 2 )2 .393 (4 )2 . 7 r t ( 4 )2 .85 r (4 )

7 7 . 47 ( r 3 )9 5 . 3 0 ( 9 )9 0 . 5 2 ( 1 3 )

3 .448 (3 )

43 . 58 (3 )8 0 . 0 3 ( 6 )9 5 . 8 8 ( 8 )

2 . s59 (4)

2 . 3 1 1 ( 0 )

60 . 00120 . 00180 . 00180 .00

7 .677 (6 )r . 6 0 7 ( 6 )1 . 6 0 s ( 4 )2 . s50 (7 )2 . 6 0 e ( 7 )2 . 6 5 9 ( 8 )1 . 6 0 8 ( 5 )

7O8.04 (22)111 . 78 (20 )L04 .7 4 ( s3 )rLz .04 (36 )

1 . 681 (4 )2 . 7 2 4 ( 8 )2 . 4 2 2 ( 8 )3 . 0 5 3 ( 9 )

9 1 . 9 0 ( 3 1 )1 0 8 . 4 9 ( 2 8 )r 3 0 . s l ( 3 3 )

1 . 9 3 1 ( s )2 . 4 2 2 ( 8 )2 . 7 38 (8 )2 . 8 7 7 ( 8 )

7 7 . s 0 ( 2 4 )96 . 4 r ( r 7 )90.24(26)

3 . 444 (5 )

4 3 . 4 7 ( 6 )79.80(r2)9s . 55 (16 )

2 . 5 5 0 ( 7 )

2 . 315 (0 )

6 0 . 0 01 2 0 . 0 0180 .00180 .00

1 . 6 1 5 ( s )1 . 6 0 8 ( 4 )1 . 6 1 1 ( 3 )2 . 5s3 ( s )2 . 6 2 2 ( 5 )2 . 6 6 5 ( 6 )1 . 6 1 1 ( 4 )

108 . 74 (16 )1 1 1 . 2 8 ( 1 6 )104 . 9 3 ( 40 )111 . s9 ( 28 )

1 . 6 8 6 ( 3 )2 .738 (6 )2 .4L7 (6 )3 .064 (6 )

9L .6L (2 r )1 0 8 " s s ( 2 1 )130 . 83 ( 25 )

r . 9 2 5 ( 3 )2. 4r7 (6)2 . 7 2 9 ( 6 )2 . 8 6 6 ( 6 )

77 .93 (L8 )9 6. 12 (r3)90 . 37 (27 )

3 . 4 4 7 ( 4 )

43 . 48 (4 )7 9 . 8 2 ( 9 )95 . 60 (12 )

2 . 5s3 (5)

2 . 316 (0 )

6 0 . 0 0120 . 00180 . 00180 .00

1 . 608 ( 3 )1 . 608 ( 3 )r . 615 (2 )2 . 5 s 9 ( 3 )2 . 6 L 5 ( 3 )2 . 6 7 5 ( 4 )1 . 6 1 1 ( 3 )

1 0 8 . 4 7 ( 1 0 )1 1 1 . 1 8 ( 1 0 )LOs . 44 (23)1 1 1 . 8 3 ( 1 8 )

L . 6 7 5 ( 2 )2 . 7 2 8 ( 4 )2 .3e7 (4 )3 . o4e (4 )

9 1 . 0 9 ( 1 4 )1 0 8 . 9 5 ( 1 4 )1 3 1 . 0 1 ( 1 5 )

r . 9 r 2 ( 2 )2 . 391 (4 )2 . 7 L 2 ( 4 )2 . 8s0 (4 )

7 7 . 4 6 ( 4 6 )9 6 . 3 7 ( 9 )90 . 38 (13 )

3 . 4 4 8 ( 2 )

4 3 . 5 7 ( 3 )80 .00 (5 )e5 .84 (7 )

2 . 5 5 9 ( 3 )

2 . 3 1 r ( 0 )

o0 . 001 2 0 . 0 0180 .00180 .00

'Di.stances az,e in mtgstroms and angLes are in degz:ees.'Numbez's in parentheses v'epz'esent the estimated. standaz'd erz,on 116l in the Last deeimal pLace.

beryl from the Tanco pegmatite (Manitoba) reported byHawthorne and Cern! (1977) (Tabte 7). Therefore, thediscussion below will focus on channel constituents.

Cs+, K*, Na*, and Ca2* ions and neutral water mole-cules occupy channel positions in Harding beryl. Waterwas placed in the C(l) site following the work of Gibbs

et al. (l 968). Size constraints were used to assign the largerCs* and K* ions to the larger C(l) site and the smallerNa* and Ca2+ to the smaller C(2) site. Featureless differ-ence Fourier maps resulted from these site assignments,which are similar to those made by Evans and Mrose(1968) and Hawthorne and eern! (1977).

BROWN AND MILLS: HYDROUS ALKALI-RICH BERYL

Table 7. comparison ofaverage bond lengths in anhydrous, hydrous, and alkali beryl

55r

Reference <T1 -0> <c (1 ) -o> <c (2 ) -o>

This s tudy

Gibbs et aL. (L968)

Moros in ( 1972 )

. :Hawtnorne d uerny

(r97 7 )

(24" c )

(s00" c )

( 8oo 'c )(24"c AH)2

Ilydrous beryl

Aflhydrous beryl

Hydrous beryl

Alkal i bery l

1 . 6 1 0 ( 3 ) l

1 . 6 0 9 ( 5 )

1 . 6 0 9 ( 4 )

1 . 6 1 r ( 3 )

t - . 611 (4 )

r . 6 0 7 Q )

1 . 6 0 7 ( 1 )

r . 6 0 8 ( 4 )

r . 6 7 5 ( 2 )

1 . 6 8 1 ( 4 )

1 . 6 8 6 ( 3 )

r . 6 7 s ( 2 )

1 . 6 5 4 ( 3 )

1 . 660 ( 3 )

1 . 6 5 3 ( 1 )

L . 6 7 7 ( 3 )

1 . 9 1 3 ( 2 )

1 . 9 3 r ( 5 )

L . 9 2 5 ( 3 )

L . 9 r 2 ( 2 )

1 . 9 0 3 ( 3 )

1 . 9 0 3 ( 3 )

1 . e 0 4 ( 1 )

1 . e 0 6 ( 3 )

3 . 4 4 8 ( 3 )

3 . 4 4 4 ( 5 )

3 . 4 4 7 ( 4 )

3 . 4 4 e Q )

3 . 4 3 9 ( 3 )

2 . 5 5 9 ( 4 )

2 . 5 s 9 ( 3 )

2 . 5 4 8 ( 5 )

rBond Lengths i.n cngstroms' The nwnbez's in

f ,o 800"C,

pa:nentheses aftet the bond Lengths refet: to the standatd ez'r'or (1&).

2Aftev. heat

Water within the structure of hydrous beryl has beenclassified as two different types (Wood and Nassau, 1967)depending on the orientation of the C, symmetry axis ofthe water molecule relative to the c axis of beryl. Type Iwater has its C, axis perpendicular to the c axis of beryl,whereas Tlpe II water has its symmetry axis parallel toc. In both cases, water is weakly bound to O(l) oxygens.Figure I illustrates possible positions for hydrogen bond-ing within the channels of the beryl structure. Wood andNassau proposed that Type I water occurs when there areno alkali ions in the adjacent C(2) positions and that TypeII water exists when there are, as shown schematically inFigure l. Polarized IR spectra of a sample of HB-3 (G.Rossman, 1977, pers. comm.) showed the presence ofboth types of water molecules with Type II strongly pre-dominating. This result is consistent with the high alkalicontent and with Wood and Nassau's observation that apositive correlation exists between Type II water and al-kali content in beryls. The distance between the C(l) andC(2) sites, 23ll L,is apparently too short to permit TypeII water in C(l) sites adjacent to filled C(2) sites. However,Type II water in an adjacent C(l) site aids in balancingthe charge of (Na,Ca) occurring in 23o/o of the C(2) sitesin the structure; bond-strength sums (after method ofBrown and Shannon, 1973) for (Na,Ca)Ou and (Na,Ca)OuHrO are 0.78 and 1.01, respectively. Although Haw-thorne and Cernf (1977) recognized the charge-balanceproblem resulting from bonding a second water moleculeto Na (2 bond strengths: 1.24), they hypothesized thatNa* ions in the C(2) sites are bonded to two water mol-ecules, based on a 2:l HrO:Na* correlation that theyobtained from consideration of available hydrous alkaliberyl analyses. In light of this 2:l ratio for beryl as wellas the 2: I ratio of Type II water to "C(2) type" cations inhydrous alkali cordierites (Goldman etal.,1977), we sug-gest that the 0.23 (Na,Ca) cations per C(2) site in HB-3are bonded to 0.46 Type II water molecules; this inter-

pretation leaves 0.09 Type I water molecules in C(1) siteswith no adjacent C(2) cations. This speculation is consis-tent with the high proportion of Type II water indicatedby Rossman's polarized IR results. In spite of these sup-porting data, the problem of charge balance at the C(2)site with water molecules in both adjacent C(l) sites re-mains unexplained.

When Cs*, K*, or Rb* occur in the C(l) site, it is unlikelythat Na* occurs in either ofthe adjacent C(2) sites becauseof repulsions resulting from the short C(lK(2) bond. Inview of the 3lo/ovacancy content of C(l) sites per formulaunit of HB-3, it is likely that the 0.23 (Na,Ca) occupy C(2)sites with no adjacent C(1) cations or water molecules. Ifwe assume that the 0.23 (Na,Ca) are bonded to 0.46 TypeII water molecules, 0.09 Type I water plus 0.14 (Cs,K)occupy C(l) sites that have no adjacent C(2) cations. Inorder to maintain local charge balance within the struc-ture, it is also possible that Li* occurs in T2 sites eitherabove or below filled C(2) sites and adjacent to C(l) sitesfilled by (Ca,K) as shown in Figure 1. The coupled sub-stitutions Lit + Na* : Be2+ and Li' * (Cs*,K*) : Be2+

Table 8. Percent thermal expansion for Harding beryt (HB-3)

Temperature ("C) 7"La I

7'Lc 2 7.Lv 3

24"200 '300"400"500 '500 '600 '700 '800 '500"

2 4 "

( a f t e r 9 6 h o u r s )

(after heat ing)( a f t e r h e a t i n g )

0 . 00 .0540 . 0 8 70 , 14 ro. 162o. 1080 . 1 8 4o.2L7o ,249o ,o97

-o.o32

0 . 00 . 0 6 50 . 0 7 60 . r 1 90 . 1 5 10 , 0 8 70 . 1 4 10 . r 7 30 . 1 9 50 , 0 6 5

-o.o32

0 . 0o . 1 1 60 . 2 4 90 . 4 1 00 . 4 8 30 . 3 0 70 . 5 1 20. 6140 . 7 0 30 . 2 6 4

- 0 . 0 8 8

r loo(dr -

2 2 o o ( c t -

a z q ) / a z q

c z q ) / c z q

V z + ) / V z +3 l O 0 ( I / , -

552

O o n d C

BROWN AND MILLS: HYDROUS ALKALI-RICH BERYL

- c ( H B - 3 )I

o 30.ll

o 2 5

o 1 5

P e r c e n t o j oT h e r m o I

Expons ion o oa

_oo (emero ldJ

^ c ( s y n b e r y l )

o ( e m e r o l d ) - _ , o - - , - - 6 " ' @

^--- ; -7)"- ' '_p--" ' " - . ) : ' ;_

^ - o - o ' o ' - " - ^ - u = ' o " o ( s v n b e r y l )

o o o

-o os

- o r o

-2oo" ,.lil",",,l"oo,"c)

uoo 8oo"

Fig. 2. Variation of cell dimensions with temperature forHarding beryl HB-3, synthetic hydrous beryl, and emerald. Datafor the last two are from Morosin (1972).

are consistent with the total channel alkali of 0.37 andtetrahedral Li* of 0.38 per formula unit.

Structural formula

Based on the chemical, structural, and IR spectral dataand on the crystal chemical reasoning presented above,the following structural formula for Harding beryl HB-3is proposed:

c(r)____________r[W(I)o.orW(Do ouCso orKo orEo.,]xlr

[-C(2)-------l[Nao,Ca. orMgo orFefli,tro ro]vl

r-M--.| f--r2 -]

[Al, o]'' [Ber rnl.io rrAlo or]Iv

t-Tl -

Isi5 e2Alo 08]rr o,8,

where W(I) and W(II) refer to Type I and Type II water,respectively, and tr refers to vacancies. The small amountsof Mgz+ and Fe3+ indicated by Peck and Steven's analysis(Table l) are assigned to C(2), and the 0.02 excess Al3+ions are assigned to T2 to compensate for the slight cationdeficiency there. The significance of the Mg and Fe as-signments is uncertain because the amounts of these cat-ions approach the detection limits of the wet chemicalanalysis. The calculated density, 2.72 g/cm3, based on thisformula, agrees well with that measured, 2.71 g/cm3,byBerman balance.

High-temperature crystal chemistry

Thermal expansion. The variation of a and c with in-creasing tomperature for HB-3 is listed in Table 8; com-

-200" 200" 400" 600" Boo"

T e m p e r o l u r e ( " C )

Fig. 3. Percent thermal expansion versus temperature forHarding beryl HB-3, synthetic hydrous beryl, and emerald. Datafor the last two are from Morosin (1972).

parisons of absolute and percentage changes in a and cfor synthetic beryl and emerald (Morosin, 1972) and HB-3are shown in Figures 2 and3. In each case, a increases ina linear fashion; however, c initially contracts then ex-pands in synthetic beryl and emerald but expands overthe same temperature range for HB-3. The c cell parameteris smaller than a at all temperatures for the two syntheticberyls but is larger than a for HB-3. The variation of cellparameters with temperature for low-alkali cordierites(Hochella et al., 1979; Evans et al., 1980) is similar tothat for low-alkali or alkali-free beryls with c contractingand a expanding. All available data for beryl and cor-dierite indicate that cell volume increases uniformly withtemperature.

Thermal expansion coeftcients, e, and e., were calcu-lated for HB-3 using linear and second-order polynomalfits to a and c, respectively. These values for alkalirichberyl, synthetic beryl, and emerald over the range 24-800'C are listed in Table 9. Figure 4 shows the variationof e,, e, and B for HB-3, synthetic beryl, and indialite. Ineach, e, is constant as a function oftemperature and, forberyl, decreases in magnitude as the composition becomesmore complex. The e, and B values for indialite vary non-linearly with temperature, whereas these values for bothberyls show a linear dependence on temperature. The moststriking difference between the coefficients for HB-3 andsynthetic beryl and indialite is the uniformly negative slopein e, and B with varying temperature for HB-3 versus thepositive slopes for the latter two.

The differences in thermal expansion behavior dis-cussed above are related to composition and bondstrengths. Morosin (1972) discussed the effects of smallamounts of Cr3+ impurity on the thermal expansion be-havior of beryl and noted a 200'C shift in the minimumof the c-axis expansion. This observation was rationalizedby Schlenker et al. (1977) as being due to the strong in-

9

BROWN AND MILLS: HYDROUS ALKALI-RICH BERYL

Table 9. Thermal expansion coefrcients (qC-t) for Harding beryl, synthetic beryl,' and emerald'

553

Tempera ture I 00 ' 200 ' 300 " 500 ' 600 ' 700 ' 800 "

. , p R - ?

Synthe t ic bery l

Enerald

ItB-3

Synthe t ic bery l

Emerald

HB-3

Synthe t ic bery l

Emeta ld

1 1 1 7

2 . 6 2 . 6

r . 6 9 L , 6 9

1 1 l q

3 . 0 3 . 9

2 , 2 2 . 5

6 . 9 6 . 7

8 . 2 9 . 0

5 . 5 5 . 8

1 . 2

2 . 6

L . 6 9

3 . t

- 3 . 1

- 0 . 4

e ?

2 . r3 . 0

t . 2

z . o

1 . 6 9

- 2 . z

- 0 . I

3 . 0

3 . 4

1 . 2

2 . 6

L . 6 9

2 . 8

- 1 L

0 . 4

8 . 0

3 . 8

3 . 7

1 t

2 . 6

1 . 6 9

- 0 . 5

o . 1

7 . 7

4 . 7

4 . r

' t t

2 . 6

1 . O v

2 . 3

0 . 4

r . 1

7 . 5

4 . 4

I . 2

2 . 6

I . 6 9

2 . I

L . 2

L . 4

6 . 4

4 . 8

1 2

z . o

r . o v

1 q

2 . L

1 . 8

7 . L

7 . 3

r4z,om Schlenker et aL. (1977). ALL Va| .ues of e1,t he f oZ lou ing equa t i ons : c1 + zcaTc t = i / a T ( 2 . 9 2 1 1 c | o - b ) ; c a = ; - f u 2

g = 2 c 1 + e 3 " o ' 1 ' " L

etr e3i and B uez'e calcuLated using

" 1 = 2 , 9 5 2 4 r 1 o - 5 , a n d c , = - 7 . 0 r L o - 8

e3, Lnd B az'e X IO-G

, u h e r e c o = 9 . 2 4 6 1 t

fluence exerted by Crr+ on the strain dependence of thefrequencies of the normal modes of beryl. They furtherdemonstrated that positive expansion along a and c inberyl can occur only when the ratio of the Gruneisenparameters, ?r and 73, satisfies the following inequality:

,S,, + ,S,2 > 1,, > 3f;}|, (l)

lS , r l 7 r S l

where the Su values are elastic compliance coemcients forberyl and ?r and 73 are measures of the thermodynamic

(e a)x 106

Tempero lu re ( "C )

Fig. 4. Variation of thermal expansion coefficients e , (alonga), e, (along c), and p (volume) with temperature for Hardingberyl HB-3, synthetic beryl (Morosin, 1972), and indialite (Evanset al. . 1980).

driving force of thermal expansion. When ,yrl7, dropsbelow the lower limit, c should contract, whereas d shouldcontract when the upper limit is exceeded. Experimentalvalues for the ratios (Srr + ,Srr)/1,S,, I and 21S,, l/Sr. of2.39 and 0.54, respectively, were determined by Yoonand Newnham (1973) for beryl and are not sensitive tosmall compositional changes.

Values of 7r/1, for HB-3, synthetic beryl, and emeraldare listed in Table l0 and plotted in Figure 5. In contrastwith the 7r/7, ratios for synthetic beryl and emerald-which drop below the lower limit of the above inequalityat 388 and l30oC, respectively, and which correlate pos-itively with increasing temperature-the curve of yr/y,versus temperature for HB-3 has a negative slope andsatisfies the inequality over the temperature range 24-800'C. The experimental ca values of Harding beryl arethus consistent with the thermodynamic requirements forpositive expansion along c and a. Finally, the curves for

Table 10. Ratio of Gruneisen parameters, tc/^tr, for syntheticberyl, emerald, and Harding beryl at different

temperatures

Ienperature ("c) y3lyr ( t tB-3) y 3 l ' y 3 ( S y n . B e r y l ) ' y 3 l y I ( E m e r a l d ) '

24'

100 '

200 '

300"

4 0 0 '

500"

600"

7 00 '

800"

1 . 3 0

t . 2 5

L . 2 I

t . 1 7

1 . 1 3

1 . 0 9

1 . 0 5

t . 0 1

- 0 . 3 3

-0 , 05

0 . 2 5

o . 4 5

0 . 6 0

o . 7 5

0 . 8 5

0 . 9 5

1 . 0 5

0 , 4 0

0 . 5 5

0 . 6 5

0 . 7 5

0 . 8 5

0 . 9 0

0 . 9 5

r . 0 0

1 . 0 5

1

Fron sch l .enker eL aL . (Lgz7) . \ j l r r = 2er + 2 '39 ca

3 . 7 0 3 7 e 1 + e 3

554 BROWN AND MILLS: HYDROUS ALKALI-RICH BERYL

2 4

z o

2 l s , 3 [

o 8

o 4

o 0

Tabte I l Potyhedral volumes (A') for beryl HB-3: Tl, T2,M, and C(l) sites

Tempera ture ( 'C) M c ( r )

9 . 0 1 L 2 2 . 0 6

9 . 3 5 1 0 8 . 2 5

9 . 2 6 1 0 8 . 5 6

9 .09 L34 .94

T1

t 6G r i i neisenPorometers

t!/'t, | 2

24 "

s00'

800'

24" (after heating)

2 . t 4

2 . 1 3

2 . t 4

2 . r 4

2 . I L

z . L 6

2 . L 6

2 . l L

o- -o-o . -o . -a (HB-3)

( e merord )- -s, - - :jl=a=-u- -o-"o o-

- -o( i in o"rr t )

r<

800"femperoture ( "C)

Fig. 5. Ratio of Gruneisen parameters,,yt/^yr, versus temper-ature for Harding beryl HB-3, synthetic beryl, and emerald. Thesevalues for the last two are from Schlenker et al. (1977).

all three beryls converge to a common value of 7r/1, athigh temperatures.

Polyhedral distances and volume changes. Mean T1-O, C(IFO, and C(2FO distances are constant from 24 to800qC, whereas mean T2-O and M-O distances increaseslightly (Table 5). In spite of the constancy ofTl-O, C(lFO and C(2)4 distances, the volume ofthe C(l) hexagonalantiprism decreases with increasing temperature (TableI l); the Tl tetrahedral volume is constant, but the T2and M polyhedra increase in volume. Examination ofindividual distances and Tl-O-Tl and O(lFTl4(l) an-gles (Table 5) indicates that the hexagonal silicate ringsremain rigid with increasing temperature. However, theconstancy of C(llo(l) distances with increasing temper-ature does not constrain the C(l)O' polyhedral volumeto constancy or prevent vertically adjacent silicate ringsfrom moving relative to one another. In fact, the C(l)O,polyhedron decreases in volume from 24 to 500'C andremains essentially constant in volume from 500 to 800"C.This behavior can be rationalized by noting that the O(lFC(l)-O(l) angle decreases from 24 to 500'C and remainsessentially constant from 500 to 800'C. M-O bonds ex-hibit the largest absolute expansion as do MOu octahedralvolumes over the temperature range 24-500'C. However,it should be noted that M-O bond lengths (Table 5) andMOu bond volume (Table I l) decrease slightly from 500to 800'C.

Structural interpretation of thermal expansion. Thestructural manifestations of linear thermal expansion alongthe a and c axes of beryl-cordierite-type framework sili-cates are (l) an increase in the size of the MOu octahedronand, in beryl, the T2Oo tetrahedron; (2) changes in the T-GM and O-T-O angles; (3) a rotation of nonring T2tetrahedra about axes lying in (001); and (4) rotation ofvertically adjacent six-membered rings in opposite sensesabout [001]. In White Well cordierite, thermal expansionbehavior can be adequately explained by expansion ofMoctahedra coupled with changes in GT-O and T-O-M

bond angles (Hochella et al., 1979). The Tl and T2 te-trahedra, occupied by Si4+ and Al3+, have essentially con-stant volumes and constant T-O bond distances with in-creasing temperature. Expansion of M octahedra underthese circumstances is accompanied by changes in GT-O and T-O-M angles. Because the M octahedron con-tributes largely to expansion in the a and b directions,those GT-O angles allowing expansion of the frameworkin these directions must increase with a correspondingdecrease in O-T-O angles contributing to the c cell di-mension. The result is a net decrease in c with increasingtemperature.

Beryl has T2 tetrahedra occupied by Bsz+ and Li* whenpresent. T24 bonds expand significantly (a"-o : 8.4 xl0-6'C-' for HB-3) and at a rate predicted to increase asLi* increasingly substitutes for Be2*. Another differencebetween beryl and cordierite involves the expansion of Moctahedra. For the Mg octahedron in cordierite, dy-6 :

12.6 x 10-6 .C-' (Hochella et al., 1979), whereas for theAl octahedron in HB-3, dr,r-o : 9. I x 10-u oC-t over therange 24-800"C. This latter value is similar to the averageAl4 bond expansion observed in oxides (a : 8.8 x 1Q-6oC-r' Hazen and Prewitt, 1977). On the average, both T2and M polyhedra expand with increasing temperature inHB-3. Even though O-T-O and T-GM angles behavesimilarly to those in cordierite, expansion occurs alongboth a and c in this beryl. The differences in thermalexpansion behavior among Harding beryl, White Wellcordierite, and other beryl-cordierite structure types arethus due to differences in T2 and M polyhedral expansion.For example, the differences in expansivity along c be-tween alkali-free beryl (Morosin, 1972) and the alkali-richHarding beryl correlates with the Li content of the T2tetrahedron and the presence of channel cations in thelatter.

Concomitant with changes in the dimensions of theT2Oo and MOu polyhedra and in GT-O and T-O-Mbond angles in HB-3, the vertically adjacent six-mem-bered rings rotate slightly in opposite senses about c. Thisphenomenon was also observed in White Well cordierite;the rings move such that Tl tetrahedra in vertically ad-jacent rings, which are linked to the same T2 tetrahedron,rotate in opposite directions around c (0.2" from 24 to775C; Hochella eI al., 1979). In this case, the rotationwas attributed to expansion of the M octahedron. In HB-3, two different modes of rotation about c by the six-membered rings were observed, although the magnitudesare comparable to the calculated standard errors ofbond

- o 4

angles. The first, occurring from 24 to 500"C, was rotationin the opposite sense to that observed in cordierite (0.2'ftom 24 to 500"C), whereas from 500 to 800'C, the ringsrotated (0.4') in the same sense as observed in cordierite.The expansion along c from 24 to 500"C is dominated byT2 tetrahedral expansion, which causes the rings to rotatein the opposite sense to that observed in cordierite. From500 to 800t, M-octahedral-volume change dominates.

Effects of heat treatment. Only slight differences be-tween the beryl structures before and after heating to 800"tCwere observed. The conventional R factors from the two24'C refinements are identical (0.050), although standarderrors are slightly better for the 24C structure after heat-ing. In the higher-temperature refinements, electron den-sity in the channel sites becomes more diffuse, consistentwith high thermal motion (Table 4) and in a manner pro-portional to the ratio ofelectron densities ofthe C(l) andC(2) sites in the 24'C structure. However, when the tem-perature was reduced to 24C after heating at 800"C for72 h, the electron densities at the C(l) and C(2) sitesreturned to their values prior to the initial heating; sug-gesting that little net change occurred in channel constit-uents during the high-temperature experiments. Our sug-gestion is consistent with the results ofTGA measurementson a sample of Harding beryl (B. Chakuomakos, 1985,pers. comm.) that indicate only 0.15 wto/o loss at 800"C.These results differ somewhat from the findings of Ainesand Rossman (1984), who reported some dehydration oftheir beryl after heat treatment at 700"C for 2 h. Becausethey did not report alkali contents, it is not possible toreconcile these findings. Positional parameters and bondlengths and angles were also essentially unchanged beforeand after heating. In this regard, the difference in C(l)polyhedral volumes before and after heating (Table I l) isanomalous. Our result differs from the observed loss ofchannel water after several days at 775"C in White Wellcordierite. This difference is due to the presence of sig-nificant amounts of Na*, Cst, and K* in HB-3, whichapparently "plug" the channels, and to their absence inWhite Well cordierite.

AcxNowr-nocMENTS

We wish to thank the late R. H. Jahns (Stanford University)for providing the chemically analyzed sample of Harding berylused in this study, and we are grateful to George Rossman (Cal-ifornia Institute ofTechnology) for running polarized IR spectraofthis beryl and for helpful discussions concerning channel con-stituents in beryl-cordierite-type framework silicates. We alsowish to thank K. D. Keefer (Sandia National Laboratories) andM. P. Taylor (Corning Glass Works) for help with various aspectsof the computer work and B. Chakoumakos (University of NewMexico) for carrying out TGA measurements on a sample ofHarding beryl. M. F. Hochella, Jr. (Stanford), Joan R. Clark(formerly with the USGS), B. Chakoumakos, and an anonymousreviewer are thanked for critical reviews ofthe manuscript. Thisstudy was supported by the Experimental and Theoretical Geo-chemistry Program ofthe National Science Foundation throughGrant EAR-80-1691 I Grown).

555

RnrnnnNcpsAines, R.D., and Rossman, G.R. (1984) The high temperature

behavior of water and carbon dioxide in cordierite and beryl.American Mineralogist, 69, 319-327.

Bakakin, V.V., Rylov, G.M., and Belov, N.V. (1969) Crystalstructure of a lithium-bearing beryl. Akademie Nauk SSSRDoklady, 188,659462.

Brown, G.E., Jr., Sueno, S., andPrewitt, C.T. (1973)Anewsingle-crystal heater for the precession camera and four-circle dif-fractometer. American Mineralogist, 58, 698-704.

Brown, LD., and Shannon, R.D. (1973) Empirical bond-strengrh-bondJength curves for oxides. Acta Crystallographica, 429,266-282.

Carson, D.G., Rossman, G.R., and Vaughan, R.w. (1982) Ori-entation and motion of water molecules in cordierite: A protonnuclear magnetic resonance study. Physics and Chemistry ofMinerals, 8, 1,1-19.

Corfield, R., Doedens, R.J., and Ibers, J.A. (1967) The crystaland molecular structure of nitrododichlorobis (triphenylphos-phine) rhenium (V), ReNCr(P(CuHr)rL. Inorganic Chemistry,6, 197-204.

Doyle, P.A., and Turner, P.S. (1968) Relativistic Hartree-FockX-ray and electron scattering factors. Acta Crystallographica,424.390-397.

Evans, D.L., Fischer, G.R., Geigher, J.E., and Martin, F.W' (1980)Thermal expansion and chemical modifications of cordierite'American Ceramic Society Journal, 63, 629-63 4.

Evans, H.T., Jr., andMrose, M.E. (1968) Crystalchemical studiesof cesium beryl. Geological Society of America Special Paperl0l. Abstracts for 1966. 63.

Filho Sampaio, H. de Alrneido, Sighinolfi, G.P', and Galli, E.(1973) Contributions to the crystal chemistry of beryl. Con-tributions to Mineralogy and Petrology, 38,279-290.

Finger, L.W., and Prince, E. (1975) A system of ronrnaN rvcomputer programs for crystal structure computations. UnitedStates Department Commerce, National Bureau of StandardsTechnical Note 854.

Gibbs, G.V., Breck, D.W., and Meagher, E.P. (1968) Structuralrefinement of hydrous and anhydrous synthetic berylAlr(BerSiu)O,, and emerald, Al, ,Cro ,(BerSiu)O". Lithos, 1, 275-285.

Goldman, D.S., Rossman, G.R., and Dollase, W.A. (1977) Chan-nel constituents in cordierite. American Mineralogist, 62,1144-I 1 5 7 .

Goldman, D.S., Rossman, G.R., and Parkin, K.M. (1978) Chan-nel constituents in beryl. Physics and Chemistry of Minerals,3.225-235.

Hamilton, W.C. (1959) On the isotropic temperature factorequivalent to a given anisotropic temperature factor. ActaCrystallograph ica, 12, 609 4 | 0.

Hawthorne, F.C., andCern!, P. (1977) The alkali-metalpositionsin Cs-Li beryl. Canadian Mineralogist, 1 5, 414, 421.

Hazen, R.M., and Prewitt, C.T. (1977) Effects of temperatureand pressure on interatomic distances in oxygen-based min-erals. American Mineralogist, 62, 309-31 6.

Hochella, M.F., Jr., Brown, G.8., Jr., Ross, F.K., and Gibbs,G.V. (1979) High temperature crystal chemistry of hydrousMg and Fe rich cordierites. American Mineralogist, 64,337-3 5 1 .

Jahns, R.H., and Ewing, R.C. (1976) The Harding mine, TaosCounty, New Mexico. New Mexico Geological Society Guide-book, 27th Field Conference, Vermejo Park,263-276.

----41977) The Harding mine. Mineralogrcal Record, 8, 115-126.

MacGillavry, C.H., and Rieck, G.D. (1968) International tablesfor X-ray crystallography, volume lll, 202-205.

Morosin, Bruno. (1972) Structure and thermal expansion of ber-yl. Acta Crystallographica, B28, 1899-1903.

Price, D.C., Vance, E.R., Smith, G., Edgar, A., and Dickson, B.L.(l 976) Mossbauer effect studies ofberyl. Journal de Physique,

BROWN AND MILLS: HYDROUS ALKALI-RICH BERYL

556 BROWN AND MILLS: HYDROUS ALKALI-RICH BERYL

Colloque C6, supplement au no. 12, Tome 37, Decembre, C6-81r{6-817.

Schaller, W.T., Stevens, R.E., andJahns, R.H. (1962)Anunusualberyl from Arizona. American Mineralogist, 47, 672-699.

Schlenker, J.L., Gibbs, G.V., Hill, R.G., Crews, S.S., and Myers,R.H. (1977) On the relation between negative thermal expan-sion, strain dependence ofentropy and elasticity in indialite,emerald and beryl. Physics and Chemistry of Minerals, l, 243-2s5.

Wood, D.L., and Nassau, K. (1967) Infrared spectra of foreignmolecules in beryl. Journal of Chemical Physics, 47, 2220-2228.

Yoon, H.S., and Nevrnham, R.E. (1973) The elastic propertiesof beryl, Acta Crystallographica, 429,507-509.

Mnuuscnrsr REcETvED Mnv 24, 1985M,lNuscRrpr AccEprED Novprrasrn 8. 1985