American Mineralogist, Volume 72, pages l195-1203, 1987

X-ray diffraction and 2esi magic-angle-spinning NMR of opals:Incoherent long- and short-range order in opal-CT

B.H.W.S. nB Jor.qc

Applied Physics, SP-DV-025, Corning Glass Works, Corning, New York 14831, U.S.A.

J. vaN Hoex. W. S. Ynnvr,lNDepartment of Physical Chemistry, University of Nijmegen, Toernooiveld, 6525 ED, Nijmegen, The Netherlands

D. V. M.cssoNGemmological Institute of America, 1660 Stewart Street, Santa Monica, California 90404, U.S.A.

Dedicated to Dr. J.G.Y. de Jong on his 78th birthday

AssrRAct

X-ray diffraction and 2eSi vras Nvrn experiments have been carried out to assess thestructure of amorphous and quasi-crystalline natural and synthetic opals. We collected'?eSiMAS NMR spectra for nine opals, which, according to their X-ray signature, belong to eithercategory A (amorphous) or CT (cristobalite-tridymite). A,ll Nrvrn spectra of natural opalshave chemical shifts of about - I l2 ppm, characteristic of amorphous silica, and resemblein peak shape amorphous silica most and, for different reasons, silica gel and quartz least.The full width at half maximum (FWHM) of the Nr'an spectra increases in the order opal-CT, opal-A, and amorphous silica. None of the opal NMR spectra can be mimicked bycombining those of the crystalline-silica polymorphs. Comparisons between X-ray and'zeSiMAS NMR spectra indicate that long-range order suggestive of cristobalite-tridymite inter-growth is not strongly coupled to short-range ordering. There is a distinct diference in'eSichemical shift between the eight natural opals and the synthetic one. Synthetic opal has amaximum chemical-shift peak intensity at - 107.0 ppm, close to that for quartz (-107.4ppm), in contrast to natural opals, whose maximum chemical-shift peak intensity is around-ll2 ppm, close to the average chemical-shift value of tridymite (-lll.0). Because ofthis quartzJike character, synthetic opal has a higher density and refractive index in com-parison to natural samples. It is an important characteristic of Nvn spectroscopy that itcan distinguish between amorphous substances with similar X-ray patterns.

lNrnooucrroru Nakata, 1974; Kastner, 1979:- Kano, 1983; Graetsch etNucleation is the principal process variable in the tec- al., 1985, 1987).

togenesis-i.e., the creation of structure-of crystalline Because X-ray diffraction of opal can only distinguishcompounds from amorphous starting materials. A facet three stages, it seemed likely to us that additional stagesin this process is the interaction between short- and long- in the solid-state transition between amorphous and crys-range ordering. We shall address in the following this talline opal could show up as variations in local Si en-facet using opals as example. vironments. Such variations can be characterized by'zeSi

Opal consists of 1500- to 3000-A spheres of silica, MAS NMR, and thus a road seemed to open up toward amono- or polydisperse depending on the number of nu- refined stratigraphic marker in sedimentary basins. In ad-cleation events, and occurs in a wide variety of natural dition, study of opals in their various transition stagessettings (Frondel, 1962;Iler,l979; Gauthier, 1985; Sand- might shed some light on the nature of amorphous solidsers, 1985; Deelman, 1986). X-ray diffraction experiments and the relation between amorphous-crystalline pairs, aon natural opal from sedimentary basins reveal the oc- subject that currently is receiving increasing attentioncurrence of solid-state transitions from amorphous to (Phillips, 1980; Matsonetal., 1983; Grimmer etal, 1984;crystalline, the various stages being designated opal-A de Jong et a1., 198 l, 1984a, 1984b; Gerstein and Nicol,(amorphous), opal-C (cristobalite), and opal-CT (cristo- 1985; Engelhardt et al., 1985; Gladden et al., 1986).balite-tridymite). The final stage of burial metamorphism In this paper we show that in the opal-A to opal-CTis indicated by the presence of quarlz in opals. Thus opal transition, long-range order as determined by X-ray dif-and its various transition stages function as stratigraphic fraction occurs first followed by short-range order, as de-markers in sedimentary basins (Fl6rke, 1956; Murata and termined by 'esi MAS NMR. In addition, we show that all

0003-o04x/87/1 I 12-1 195$02.00 I 195

I 1 9 6 DE JONG ET AL.: MORPHOGENESIS OF OPAL-CT

opal-A and opal-CT samples used in this study show lo-cal Si environment distributions closely akin to thosefound in amorphous silica. Next, we show that the ,eSi

MAS NMR spectra of opals cannot be constructed by simplesuperposition of the spectra of cristobalite and tridymite.Hereafter, we show that silica in opal is completely po-lymerized and that there is a fairly narrow range of Si-G-Si angles between those of amorphous silica and trid-ymite. Finally, we suggest reasons why long-range order-ing and short-range ordering are not developed synchro-nously with one another.

ExpnnrlrnNTAl DETATLS

Eight opals from different localities and one synthetic one werecharacterized by X-ray diffraction, chemical analysis, nsn, andproton and 'zeSi magic-angle-spinning Nrran. HrO measurementson the samples were carried out in three temperature steps usinga modified Karl-Fischer assaying method.

'?eSi rrr.ns Nun spectra were collected on a home-built spectrom-eter with a 4.2-T wide-bore Oxford magnet at 35 7 56 kHz andon a cxp 300 Bruker spectrometer at 59 600 kHz. Normal rr-rwrntechniques in conjunction with u,q,s were used to obtain all spec-tra. Most spectra were collected without proton decoupling onceit was observed that such decoupling did not afect the':eSi linewidth nor line position in opals. Cylindrical rotors (8-mm inter-nal diameter) were used for collection of the "Si rrtes wun spectrawith typical spinning speeds of 2.5 kHz. A 90' pulse, varyingrecycle delay between I s and I h (30 s for the spectra discussedhere), 1000 data points, and 20-kHz spectral line width wereused in all cases. Line broadening due to exponential multipli-cation of the free-induction decay is l0 Hz. The "Si chemicalshifts reported in this paper are referenced by sample exchangeto TMS. Two spectra were run with benitoite (-94 ppm) as aninternal standard (de Jong et al., 1984a).

Proton uls wr'rR spectra were collected in similar fashion. Dueto the much shorter proton I, relaxation time, recycle delays ofI s were used. The proton spectra were referenced by sampleexchange to benzene. The negative sign indicates that the pro-tons in opals are more shielded than those in benzene. Protonand 'zeSi MAs NMR spectra of one sample were measured afterheat treatment for 24 h at 300 qC under vacuum (0.005-mmHg). esn spectra on opals were collected on a Varian E-123 (x-band) spectrometer at room temperature.

Density and refractive-index measurements were carried outusing a picnometer and refractometer, respectively.

Rrsur,rsResults of our X-ray diffraction and MAS Nun experi-

ments are collected in Table I and illustrated in Figures1,2,3, and 4. The opals in Table I are designated eitheropal-A or opal-CT depending on their amorphous orquasi-crystalline character as determined by X-ray dif-fraction (Fig. 3). The do,, spacing of cristobalite andtridymite are 4.05 and 4.10 A, respectively. Larger dspacings are thought to be a measure of the tridymite-cristobalite transition in opal-CT (Kano, 1983). The do,,spacingin samples 2,3,and7 is4.07,4. 13, and4. l l A,respectively, suggesting that sample 2 is most and sample3 least like cristobalite. One sample, number 6, has anopal-A pattern with some quartz. The amorphous haloes

of natural and synthetic opal-A have an intensity maxi-mum at a20 of 22" (d^" :4.04 A) and 21.5" (d^,8:4.13A), respectively, slightly diferent and narrower than thatof vitreous sil ica at a20 of 2l '(d^"":4.23 A).

Procedures for manufacturing synthetic opals have beendeveloped by K. Inamori and P. Gilson. X-ray patternsof these synthetic opals, designated Inamori or Gilsonopal, show the presence of ZrOr, particularly in the Gil-son opal as noted by other workers (Simonton et al., 1986;Gauthier, 1986). The calculated crystallite size of opal-CT, based on line broadening of the (01l) diffraction line,varies between 60 and 90 A, bracketing the c dimensionof tridymite PO (81.864 A;(Ktug and Alexander,1954;Fagherazzi et al., l98l; Konnert and Appleman, 1978;Vieillard, 1986; Nukui and Fl6rke, 1987).

The '?eSi chemical shifts of all natural opals fall around- I 12 ppm (Fig. l), between the chemical shifts for cris-tobalite and tridymite. The full width at half maximum(FWHM) of 'zeSi Nun peaks varies between 5.8 and 9.8ppm for natural opals, but is substantially broader, I Ippm, for the one synthetic sample. The'zeSi chemical shiftof this sample is less negative, - 107 ppm, and closer tothe chemical shift of qtJarrz, than those of the naturalsamples.

There are large variations in fr relaxation times forsilica species. 7", for quartz is ofthe order of5 h, whereas7, for opals varies between 0.25 and l0 s, presumablybecause of the presence of either paramagnetic impuritiesor water in the latter (Gladden et al., 1986). The widthof the components building up a 2eSi resonance line inopal is 50 Hz (l ppm) as revealed by 7", determinationsobtained using the Hahn spin-echo technique (Hahn,1950). The meaning of this result is that 50 Hz is thelower resolution limit between a collection of silica speciesand hence that the width of the '?eSi MAS NMR lines inopals is due to a distribution ofsilica species rather thanto variations in relaxation efects.

Proton chemical shifts do not show large variationsbetween samples. Proton decoupling does not afect the'zeSi spectralJine width nor its position. The water con-tent, characteristic for physisorbed (20-105 'C) andchemisorbed (105-500 "C) water, varies between sampleswithout discernible trend.

The esn spectra were collected to determine if para-magnetic centers affect the opal NMR line width. Somesamples (notably 3, 6, and 8) show strong ESR signals,indicative of substantial concentrations of paramagneticcenters, whereas others (samples l, 4, and 5) do not. Thereis no significant difference in Nrurn line width betweensamples with high or low concentrations of paramagneticcenters (Table 1).

DrscussroN

Since the pioneering study of Lippmaa et al. (1980),2eSi uas NMR spectroscopy has become a well-establishedtechnique to probe the structure of crystalline alumino-silicates, particularly zeolites (Klinowski, 1984; Oldfieldand Kirkpatrick, 1984; Kirkpatrick et al., 1985). On the

other hand, 2eSi uas Nvrn experiments on silicate glassesare still fairly rare, though the number of such studies israpidly increasing (Grimmer et al., 1984; de Jong et al.,1983, 1984a, 1984b; Murdoch et al., 1985; Dupree et al.,1984, 1986; Weeding et al., 1985; Fujiu and Ogino, 1984;Gerstein and Nicol, 1985; Engelhardt et al., 1985; Glad-den et al., 1986; Kirkpatrick et al., 1986; Aujla et al.,1986; Yang et al., 1986; de Jong and Veeman, 1986; De-vine et al., 1987; Turner et al., 1987; Schneider et al.,1987). Still unique is the Nvrn study on molten silicatesofStebbins er al. (1985).

We start our discussion with the 'zeSi MAS NMR resultsfor opals and compare these results with those for silicagel and glass. Next we discuss differences between theX-ray diffraction and MAS NMR results.

Comparison between 2eSi v.ls uun spectra of opals,silica gel, and silica glass

The 'zeSi MAS NMR spectra of opals cover the chemicalshift range of -99 and -123 ppm (Fig. l). This rangeindicates that within Nun detection limits, all Si atomsare coordinated to four oxygens in a three-dimensionalarray of corner-sharing tetrahedra. Thus the uncommonconfigurations postulated by various authors to exist inamorphous silica, such as edge-sharing tetrahedra orthreefold-coordinated Si (chemical shift less negative than60 ppm), and fivefold- or sixfold-coordinated Si (chem-ical shift more negative than 130 ppm), if present, do notoccur in sufficient concentrations in opal to be detected(O'Keeffe and Gibbs, 1984; Garofalini, 1984; Weyl andMarboe, 1959). The absence of silanol groups, i.e., Qo,Q,, Qr, and Q, tetrahedra with protons attached to thenonbridging oxygens, is confirmed by the lack of line nar-rowing of the Si spectrum on proton decoupling, in ac-cordance with the HrO chemical analyses, which indicate

tl97

oPAL A - i l t ,s

oPAL CT - | 24

-90 - i lo -t30

CHEMICAL SHIFT (ppm)

Fig. 1. '?eSi ves r.n"rn spectra of amorphous (opal-A) and quasi-crystalline (opal-CT) gem-quality opals.

that most water in these compounds is either physisorbedor chemisorbed, albeit that the second temperature step,500 'C, is somewhat high for chemisorbed water, aspointed out by Knauth and Epstein, 1982. A fairly strong

oE JONG ET AL.: MORPHOGENESIS OF OPAL-CT

TneLe 1. Proton and 2esi chemical shifts, full width at half maximum (FWHM), numberof Si free-induction decays (FlD), and weight percent water in opals in tem-perature steps

HrO (wt%)

Sample Description4Si shift. 'H shift.- FWHM

(ppm) (ppm) (ppm) F ID20- 105- 500-

105rc 500 'c 1000ec

1 Opal-A2 Opal-CT3 Opal-CT4 Opal-A5 Opal-A6 Opal-A + qtz7 Opal-CTI Opal-A9t Opal-A + qtz

10t Opal-A + qtz111 Opal-A + qtz12 Opal-A+

- 1 1 1 . 8 - 1 . 8- 1 1 1 . 9 - 1 . 7-1 ' 12 .4 - 1 .7- 1 1 1 5 - 1 I- 1 1 1 . 5 - 1 . 9-111 .4 - 1 .7- 1 1 1 . 8 - 1 . 8- 1 1 1 . 5 - 1 . 9- 1 10 .6 n .d .- 1 1 1 . 9-111.4 n.d.- 1 07.0 n.d.

9.8 5 0096.2 53405.8 8 5028.6 5 6508.0 13 5929.8 27727.2 7 9408.6 1 0 6589.8 1 1 188

10.6 627410.1 821 1 . 0 2 5 0 0

n.d. n.d. n.d.1.74 7 .0 0.010.58 4.4 3.8n.d. n.d. n.d.2.04 4.6 0.272.04 2.9 0.241.08 7.5 0.232.68 4.1 0.012.04 2.9 0.24n.d. n.d. n.d.2.04 2.9 0.24n.d. n.d. n.d.

'Relative to TMS, proton shifts relative to TMS: benzene, 7.3 ppm; H2O, 4.8-4.7 ppm; (H) inopal, 5.5 ppm

-t Relative to benzene.tThree other experiments were carried out on sample 6: in sample 9, protons were decoupled;

in sample 10, protons were removed atlet 24-h heat treatment at 300 rc; in sample 1 1 , experimentwas carried out on Bruker cxp 300 at 7.1 T.

+ Inamori synthetic opal

IItl

-JL_-l;JI

ttl

-JL-

or..lARTz

CRISTOBALITEstLtcaGLASS

OPAL A

l 198 DE JONG ET AL.: MORPHOGENESIS OF OPAL-CT

4 t l@tov

OPAL CT4 3 3(4 328)

OPAL A + OUARTZ

426

VITREOUS SILICA

I t l

-85 -rO5 -r25 -85 .rO5 -t25CHEMTCAL SHTFT (opm)

Fig.2. 2eSi u.e.s r.nan spectra of silica gel (Maciet and Sindorf,1980), silica glass (Murdoch et al., 1985), quartz, tridymite (Smithand Blackwell, 1983), and opal.

interaction between water and silica is suggested by theobservation that proton chemical shifts of opal are aboutl.l ppm more positive than for water (Table 1). Thisconclusion concerning the absence of silanol groups inopals differs from the one reached by Graetsch et al. ( I 9 8 5)based on rn data of opal-C.

Line broadening of the 2eSi ues NMR spectra of opalsmay be caused by dipolar coupling between protons and'zeSi nuclei. To test for this possibility we carried out threeadditional rurvrn experiments on one sample (no. 6), inwhich the protons were decoupled (sample 9, Table l),the spectrum was measured after removal of water (sam-ple l0), and the spectrum was measured at 7.1 T ratherthan the standard 4.2 T (sample I l). The results in TableI indicate that the FWHM of the 2rSi spectra were notaffected in these three experiments. Hence, the cause forthe broad FWHM of opal-A spectra has to be found in adistribution of chemical shifts.

Ofall the silica spectra considered here, the opal spec-tral shape resembles that of silica glass most and that ofeither silica gel or quartz least (Fig. 2). The differencesbetween the spectra of silica glass and opal are that thepeak maximum for opal is more negative [about -ll2vs. - I I1.5 ppm (Gladden et al., 1986) or - 110.9 ppm(Murdoch et al., 1985)l and that the FWHM for opal issmaller (5.6 to 9.8, vs. 11.6 ppm).

The distribution of mean Si-O-Si angles per tetrahe-dron, (Si-O-Si), for natural opals, calculated using the

40 30 ?o ro20 (degrees)

Fig. 3. X-ray diffraction patterns of opal-CT, opal-A + qt:*rrtz,opal-A, and vitreous silica. The numbers in parentheses are thed spacings of tridymite.

method of Thomas et al. (1983) and illustrated in Figures4 and 5, indicates a narrower range for the opals than forsilica glass. For opal-A, (Si-O-Si) varies between 133'and 168" with a maximum around 151". For opal-CT,(Si-O-Si) varies between I 38' and I 70' with a maximumaround 152o, whereas for silica glass, the range variesbetween 122' and 170" with a maximum around 148".The (Si-G-Si) variation of the silica glass of Dupree andPettifer (1984) indicates their sample to be most likely agel. It should be noted that the calculation of (Si-O-Si)angles assumes that 2eSi chemical shifts of crystals andglasses correspond to one another, an assertion that, aswe have demonstrated elsewhere (de Jong et al., 1984b),is not necessarily correct. It necessitates in addition thatspectralJine width is caused by a distribution of silicaspecies rather than by relaxation effects. That silica dis-tribution turns out to be the cause for line broadening hasbeen demonstrated with our spin-echo experiments.

Compared to crystalline SiO, phases, the calculatedrange of (Si-O-Si) values in opals corresponds mostclosely to that of tridymite MC, i.e., tridymite with 12nonequivalent Si positions (Baur, 1977). For tridymiteMC, the average chemical shift is - I I I ppm (Smith andBlackwell, 1983) with a correspondine (Si-O-Si) angle of150o, whereas for tridymite OP, i.e., tridymite with 80

OPAL A

I DUPREE A PETTIFER (*9)

2 MURDOCH etol

TOT ANGLE (degrees)

Fig. 4. Comparison between (SiSi) distribution of opal-Aand amorphous silica using the formula of Thomas et al. ( I 983).The sample of Dupree and Pettifer (1984) is most likely silicagel.

nonequivalent Si positions and (Si-O-Si) angle of 148.3'(Konnert and Appleman, 1978), no NMR spectrum hasbeen measured to date. The maximum and minimum(Si-O-Si) angles observed from X-ray diffraction in trid-ymite MC are 157.5'and 146.7', respectively, obviouslya range ofvalues substantially narrower than the one ob-served in opals. The single Si site in cristobalite has achemical shift of - 108.5 ppm corresponding to a (Si-O-Si) of 146.4' (Dollase, 1965). Thus in terms of chemicalshifts and variation in (Si-O-Si) distribution, local Sienvironments in opal most resemble those in tridymite.The major difference is that for opals, the correlation be-tween (Si-O-Si) and 2eSi Nrun chemical shift indicates thepresence of tetrahedra with (Si-O-Si) angles as small as133o, whereas tridymite contains no such tetrahedra.

It needs to be pointed out that there is a diferencebetween angular distributions obtained from Nun chem-ical shifts and from X-ray difraction. The latter tech-nique in its application to amorphous solids assigns an-gular distributions based upon experimentally determinedSi-O, Si-Si, and O-O distances, resulting in a distributionof individual Si-O-Si angles (Taylor and Brown, 1979).On the other hand, NMR measures a distribution of meanSi-O-Si angles, (Si-O-Si), per tetrahedron. These twodistributions are not necessarily the same, as can be il-lustrated for tridymite MC. In this crystal the minimum

tt99

i-O-Si ANGLE (degr@s)

Fig. 5. Variation in (Si-G-Si) distribution between opal-Aand opal-CT using the formula of Thomas et al. (1983).

and maximum (Si-O-Si) angles are 146.7' and 157.5o,respectively, whereas the minimum and maximum Si-O-Si angles are 143.4 and 179.1', respectively (Baur,r977\.

In addition to the empirical formula derived by Thom-as et al. (1983), three other semi-empirical formulas havebeen constructed to relate chemical shift to (Si-O-Si)angle (Radeglia and Engelhardt, 1985; Ramdas and Kli-nowski, 1984; Smith and Blackwell, 1983). These threerelationships address different facets of variation in thediamagnetic contribution to the chemical shift, focusingattention on electron-density variation on Si (Ramdas andKlinowski, 1984), on oxygen (Radeglia and Engelhardt,1985), or in the Si-O bond (Smith and Blackwell, 1983).The linear least-squares fits to the data points, correlationcoemcients, and chemical-shift cutoff values are collectedin Table 2. We have calculated the angular variation foropal sample 9 using these four formulas and illustrate theresult in Figure 6. Calculated (Si-G-Si) angular varia-tions using the different formulas are slight except for therelationship of Thomas et al. (1983), which, because itincludes in the fitting procedure the chemical shift of zun-yite (-128.2 ppm), shifts the Si-O-Si distribution tosmaller angles. We have used this last procedure not onlybecause it has the best correlation coefficient (Table 2)but also because it enables calculation of (Si-O-Si) for

os JONG ET AL.: MORPHOGENESIS OF OPAL-CT

.{z

5oI

TneLe 2. Equations of various least-squares fitted lines used to calculate (Si-O-Si) angles, A, from 4Si chemical-shift values, D

Formula (ppm)

Cutotf value ofCorrelation chemicalshiftcoetficient (ppm) Reference* and comments

6 = 0.63514 - 241.448(cos(0)/(cos(d) - 1)6 : 149.947 - 270.532(sin(0) 12)6: -170.337 - 50.9861(sec(a))d: -19.8215 - 0.608526((d))

0.9080.9100.9040.911

120.1120.61 19 .4129.4

(1) s hybridization of oxygen atoms(2) non-bonded Si-Si interactions(3) Si-O overlap integrals(4) empirical

- 1, Radeglia and Engelhardt, 1985; 2, Ramdas and Klinowski, 1984; 3, Smith and Blackwell, 1983; 4, Thomas et al., 1983.

1200 DE JONG ET AL.: MORPHOGENESIS OF OPAL-CT

OPAL A{ r 9 )

to r50 160 t70 t80

Fig 6 (si-o-si) .",J;jT:;:ff;i" rhe rour rormurascollected in Table 2. The single line represents the distributionaccording to Thomas et al., 1983.

the tail of the opal NMR spectrum that has a chemicalshift more negative than - 121 ppm.

The 'zeSi MAs NMR spectrum of a natural opal-CT andthe previously published spectra of silica gel, vitreous sil-ica, quartz, tridymite, and cristobalite (Smith and Black-well, 1983; Murdoch et al., 1985; Maciel and Sindorf,1980) are shown in Figure 2. Comparison of these spectraclearly indicates that the opal-CT spectrum cannot beconstructed by superposition of those of tridymite andcristobalite.

The one synthetic opal measured in this study has achemical shift of - 107 ppm, close to the value for quartz(- 107.1 ppm). By analogy with the crystalline silicas, weexpected the natural opals, with their chemical shiftsaround - I 12 ppm, to be lower in density and refractiveindex than the synthetic ones. Triplicate measurementsof refractive indices and densities of four natural opalsand three synthetic ones (two samples manufactured byInamori and one by Gilson) (Table 3) confirm this hy-pothesis. Thus, the less-negative chemical shift, i.e., themore quartz-like constitution, of synthetic opals coin-cides with higher density and higher refractive index. Asimilar phenomenon is observed in the pressure-induced160/o densification of suprasil, which causes a chemicalshift of 2.5 ppm to less-negative values (Devine et al.,1987). The same l6olo densification is also observed inneutron irradiation of vitreous silica (Maurer, 1960; Pri-mak, 1958), suggesting that, if pressure and irradiationdensification of vitreous silica lead to the same end result,no anomalous Si coordinations occur in either processwithin Nrran detection limits. The chemical-shift varia-tions for amorphous silica species point toward an im-portant feature of Nrrrn spectroscopy, namely, that it candistinguish between amorphous substances that yieldsimilar X-ray diffraction patterns.

Distribution of local Si environments in opal andlong-range ordering in opal-CT

Why did we emphasize in the above discussion thesimilarity or dissimilarity between opal and cristobalite-

TeeLe 3. Density and refractive index of nat-ural and synthetic opals

Sample Density (g/cms) Refractive index

Iz

Jfqo

R 12715'R 11399 .R 1 3779-R 11397 .R 13154 - -R 13153 . .R 13213t

2.1292.1212 1222.1222.2152 2122 209

1.4471.4481.4431.4481.4601.4641.460

- Natural samoles.*. Inamori synthetic samples.t Gilson synthetic sample-

tridymite? The reason for this is that the X-ray diffractionpatterns of opal-CT (Fig. 3) are usually interpreted asbeing due to disordered cristobalite (Fl6rke, 1956; Kast-ner, 1979; Graetsch et al., 1987). Thus from the X-rayevidence one might have expected Nvrn spectra of opalswith chemical shifts around -108 ppm, correspondingto a (Si-O-Si) distribution with a maximum frequencyaround 146'and a narrower variation in angles than ac-tually observed.

The question arises whether the three peaks in an opal-CT X-ray diffraction pattern were indeed characteristicof disordered cristobalite or whether alternative interpre-tations are possible. Cristobalite and tridymite both con-sist of six-membered rings of silica tetrahedra, the ringconformation varying between chairs and boats in trid-ymite versus chairs only in cristobalite (Taylor and Brown,1979). The oxygen atoms in crystobalite consist of a three-layer sequence equivalent to the sequence ofcubic closestpacking, whereas those in tridymite consists of a two-layer sequence equivalent to hexagonal closest packing(Graetsch et al., 1987).

More important as a clue to interpret the opal-CT X-raydifraction pattern are the characteristic radial distribu-tion maxima for vitreous silica as a function of Si-O-Siangle (Taylor and Brown, 1979). These data indicate thatamorphous SiO, with straight Si-O-Si angles should havemaxima around 1.6, 2.3, and 3.2 A plus three additionalmaxima between 3.3 and 4.5 A Gig. 7A). Si-O-Si bend-ing causes third- and fourth-nearest-neighbor distances tocoincide with maxima around 2.6,3.5, and 4 A Fig. 7B).Comparison of these distances with those observed inopal-CT X-ray patterns indicates that the 4.3-, 4.1-, and2.5-A spacings are characteristic oxygen-{xygen dis-tances in angle-constrained silicas. It is therefore equiv-ocal to interpret the opal-CT diffraction pattern as thatofdisordered cristobalite, ordered cristobalite with a trid-ymite stacking sequence, or a well-ordered, possibly trid-ymiteJike oxygen array, with cristobalite stacking faults.Any of these stacking sequences, and possibly many more,may give rise to long-range unidirectional ordering ofclose-packed oxygen atoms, while maintaining local Sienvironments between those encountered in tridymite andamorphous silica.

The pronounced character of the (l0l) X-ray diffrac-

DE JONG ET AL.: MORPHOGENESIS OF OPAL-CT t20l

oo

Oo

o

Fig. 7. (A) Distance frequencies in SiO, at Si-O-Si angle of180". (B) Distance frequencies in SiO' at Si-O-Si angle of 140'(Taylor and Brown, 1979).

tion peak of opal-CT, sampling long-range ordering of atleast 50 A, and the broad opal-CT 2eSi ves NMR spec-trum, resembling amorphous silica much more than trid-ymite or cristobalite, suggest that in silica species, solid-state transitions are initiated by an ordering ofthe oxygenarray followed by that of the Si sites.

Sensitivity of 2eSi u.r,s Nrrn vis-i-vis X-ray diffraction indetecting crystalline phases

The preceding section has raised some questions aboutthe common interpretation of X-ray diffraction patternsof opal-CT as indicating the coexistence of microcrystal-line and amorphous silica. We address in this sectionfactors pertinent to detecting small amounts of crystallinephases in amorphous silica.

In a two-phase system, as opal-CT is according to X-raydiffraction, detectability of a phase by means of NIrrnspectroscopy depends on the relative concentration ofthephases, whether the peaks overlap, and on the spin-latticerelaxation time, 2,. I, values for opal and quartz areabout l0 s and 5 h, respectively. Thus, considering onlyI, variations, for a sample of opal-A containing quartz,a 30-s delay between pulses enhances the amorphouscomponent by a factor of 600 relative to quartz.

To determine the conditions under which a small

-80 - too - t20CHEMICAL SHIFT (PPm)

Fig. 8. (A)'zeSi r'res NIvrR spectrum of opal-A and quartz takenwith 30-s time interval between pulses' (B) 2eSi tr'tls NMR spec-trum of opal-CT taken with l-h time interval between pulses.

amount of quartz can be detected in opal-A, two exper-iments were caried out. In the first experiment (Fig' 8A)a 2eSi ues NMR spectrum of quariz was compared withthat of sample 6, a quartz-containing opal-A (experimen-tal conditions: 59.6 MHz, 30-s delay, similar spin con-centrations). One free-induction decay (FID) of quartzyields a signal to noise (S/N) ratio of three; 120 FIDs ofquartz should give a S/N ratio of 33 (3 x y'T20). Henceif 1/33 or 30/o of sample 6 consists of quartz, a peak shouldappear at -107.4 ppm with a S/N ratio of l, providedT, of qtartz in opal is similar to Z, in quartz alone. The5o/o quartz found in the X-ray diffraction pattern of sam-ple 6 does not give rise to a discernible quartz peak inthe teSi MAS NMR spectrum of this sample.

In a second experiment the recycle delay of opal-CT(sample 3) was changed from 30 s to I h in order to testfor the possibility that cristobalite in this sample has con-siderably longer Z, than opal itself. The spectrum did notchange except for a very small peak, just above back-ground at -107.4 ppm, characteristic for quartz (Fig. 8B).We conclude from these experiments that, provided thatthe I, of cristobalite is not pathologically long, the localSi environments, sampled by wrvrn on the scale of 5-6 A,reflect an arrangement comparable to that of amorphoussilica. On the other hand and as already pointed out,

1202 DE JONG ET AL.: MORPHOGENESIS OF OPAL-CT

long-range order of at least 50 A, when sampled by X-raydiffraction, shows a transition toward a crystalline array.

In this context it is ofinterest to note that oxygen in-teractions determine geometry in silica species, as dem-onstrated by de Jong and Brown (1980) and used byThattachari and Tiller (1982) in modeling the structureof amorphous silica. The data in this study on opals sug-gest that long-range ordering ofthe oxygen array has al-ready taken place in opal-CT, whereas the Si atoms havenot yet found their equilibrium positions. It is a matterof speculation if such long-range order preceding settle-ment of local Si environments is the rule rather than theexception in solid-phase transitions from amorphous tocrystalline silicates.

Sulruany AND CoNCLUSToNS

Opal-A and opal-CT yield relatively broad 2eSi unsNMR spectra that cannot be interpreted as being due to asuperposition of the spectra of crystalline silicas. TheX-ray diffraction pattern ofopal-CT can therefore not beinterpreted as being caused by line broadening due tocristabolite-tridymite microcrystallites for which Nunspectroscopy, sampling local environments in the orderof 5 to 6 A, would show a combination of tridymite andcristobalite peaks. The difference between the X-ray andMAS NMR signatures of opal-A with quartz indicates thegreater sensitivity of the former technique to the presenceof crystalline phases in this particular system in contrastto, for instance, the detection of corundum in alumino-silicate glasses (de Jong et al., 1983). Results presentedhere also indicate that long-range ordering in the transi-tion from amorphous to quasi-crystalline opals precedesshort-range ordering.

Proton Nvn does not indicate a significant amount ofsilanol groups in opals. Our work suggests the presenceofwater physisorbed on the substrate, akin to the type ofwater present in zeolites. Synthetic opals are, accordingto their'?eSi chemical shift, intrinsically more quartzJikethan natural ones. As a consequence, such opals canreadily be distinguished from natural ones by their higherrefractive index and density. Our results indicate, thus,that Nrvrn distinguishes between X-ray-comparable amor-phous phases.

AcrNowr,rocMENTS

Water analyses, X-ray diffraction, density, and refractive index mea-surements were carried out by Vincent van der Mee and Willie van Weselat Smitweld, the Netherlands, and by Pat Gray at the GemmologicalInstitute of America, U.S.A. Critica.l reviews by John Higgins and JimKirkpatrick have helped considerably in straightening out the interpre-tation of the observed phenomena

RnrnnrNcrs crlnoAujla, R.S., Leng-Ward, G., Lewis, M.H., Seymour, E.F.W., Styles, G.A.,

and West, G W. (1986) An r*rrn study of silicon coordination in y-Si-AI-O-N glasses. Philosophical Magazine, B54, L5l-L56.

Baur, W.H. (1977) Silicon-oxygen bond lengths, bridging angles Si-O-Siand synthetic low tndymite. Acta Crystallographica, B33, 2615-2619

Deelman, J C (1986) Opal-CT in bamboo. Neues Jahrbuch fiir Miner-alogie Monatshefte, 407-41 5.

de Jong, B H.W.S., and Brown, G.E. (1980) Polymenzation of silicateand aluminate tetrahedra in glasses, melts, and aqueous solutions-I.Electronic structure of HuSirO,, H,AlSiO)-, and HrAlrO] . Geochimicaet Cosmochimica Acta.44, 491-513.

de Jong, B H W.S, and Veeman, W.S. (1986)'?eSi ues and'3Na nutationNun reveal that local sodium environment is principal difference be-tween sodium silicate gel and glass. Intemational Mineralogical Asso-ciation, Abstracts with Program, 87.

de Jong, B.H.W.S., Keefer, K.D., Brown, G.E., and Taylor, Ch.M. (1981)Polymerization of silicate and aluminate tetrahedra in glasses, melts,and aqueous solutions-Ill. Local silicon environments and internalnucleation in silicate glasses Geochimica et Cosmochimica Acta, 45,r 29 l - l 308.

de Jong, B H W S , Schramm, C.M., and Parziale, V.E. (1983) Polymer-ization ofsilicate and aluminate tetrahedra in glasses, melts, and aqueoussolutions-IV. Aluminum coordination in glasses and aqueous solu-tions and comments on the aluminum avoidance principle. Geochim-ica et Cosmochimica Acta, 47, 1223-1237.

-(1984a) Polymerization of silicate and aluminate tetrahedra inglasses, melts, and aqueous solutions-V. The polymeric structure ofsilica in albite and anorthite composition glass and the devitrificationof amorphous anorthite. Geochimica et Cosmochimica Acta, 48,2619-2629

-(1984b) zeSi magic angle spinning Nrrrn study on local silicon en-vironments in amorphous and crystalline lithium silicates. AmericanChemical Society Journal, 106, 4396-4402

Devine, R.A.B., Dupree, R., Farnan, I., and Capponi, J.J. (1987) Pres-sure-induced bond-angle variation in amorphous SiOr. Physical Re-view B35, 2560-2562.

Dollase, W.A. (1965) Reinvestigation of the structure of low cristobalite.Zeitschrift fiir Kristallographie, I 2 l, 369-37 7 .

Dupree, R., and Pettifer, R.F. (1984) Determination of the Si-O-Si bondangle distribution in vitreous silica by magic angle spinning rcMn. Na-ture. 308. 523-525

Dupree, R, Holland, D., McMillan, P.W., and Pettifer, R.F. (1984) Thestructure of soda-silicate glasses: A MAS NMR study. Journal of Non-Crystalline Solids, 68, 399-410.

Dupree, R., Holland, D., and Williams, D.S. (1986) The structure ofbinary alkali silicate glasses. Joumal of Non-Crystalline Solids, 8 l, 185-200

Engelhardt, G, Nofz, M., Forkel, K., Wihsmann, F.G., Miigi, M., Sa-moson, A., and Lippmaa, E (1985) Structural studies ofcalcium alu-mjnosilicate glasses by high resolution solid state ,rSi and ,?Al magicangle spinning nuclear magnetic resonance Physics and Chemistry ofGlasses. 26. 157-165.

Fagferazzi, H P., Zardetto, G., and Raiteri, F. (1981) X-ray line broad-ening Fourier analysis. IL Application to a glass-ceramic system andto copper filings. Gazetta Chimica Italia, I 1 l, 137-143.

Fliirke, O.W (1956) Zv frage des "hoch"-cristobalit in opalen bentonitenund gliisem. Neues Jahrbuch fiir Mineralogie Monatshefte, 217-223.

Frondel, C. (1962) The system of mineralogy, vol. III, 334 p Wiley, NewYork.

Fujiu, T., and Ogino, M. (1984) 2eSi Nvn. study on the slructure oflead-silicate glasses. Journal ofNon-Crystalline Solids, 64, 28'1-290

Garofalini, S.H. (1984) Defect species in vitreous silica-A moleculardynamics simulation. Journal of Non-Crystalline Solids, 63, 337-345.

Gauthier, J.P (1985) Direct observation of different types of opal bytransmission electron microscopy: I. Gem opal. Journal de Microscopieet Spectroscopie Electronique, 10, ll7-128

-(1986) Synthdse et imitation de I'opal noble. Revue de Gemmo-logie A.F G., 89, 16-23.

Gerstein, B.C., and Nicol, A.T. (1985) Crystallinity and local chemicalenvironments of silicon in a silicon-aluminum-oxygen glass ceramic.Journal of Non-Crystalline Solids, 75, 423-428.

Gladden, L.F., Carpenter, T.A., and Elliott, S R ( I 986),,Si lasNun stud-ies of the spinJattice relaxation time and bond-angle distribution invitreous silica. Philosophical Magazine, B53, L8l-L87.

Graetsch, H , Fl<irke, O.W., and Miehe, G (1985) The nature of water inchalcedony and opal-C from Brazilian agate geodes. Physics and Chem-istry ofMinerals, 12, 300-306.

- (1987) Structural defects in microcrystalline silica. physics andChemistry of Minerals, 14, 249-257.

Grimmer, A R, Miigi, M, Hiihnert, M., Stade, H., Samoson, A, Weiker,W, and Lippmaa, E. (1984) High resolution solid state reSi nuclearmagnetic resonance spectroscopic studies ofbinary alkali silicate glass-es Physics and Chemistry ofGlasses, 25, 105-109.

Hahn, E.L (1950) Spin echoes Physical Review, 80, 580-594.Iler, R.K (1979) The chemistry of silica. Wiley, New York, 866 p.Kano, K. (l 983) Ordering ofopal-CT in diagenesis. Geochemical Journal,

17,87-93Kastner, M. (1979) Silica polymorphs. Mineralogical Sociery of America

Reviews in Mineralogy, 6, 99-109.Kirkpatrick, R.J., Smith, K.A., Schramm, S., Turner, G., and Yang, W.H.

(1985) Solid state magnetic resonance spectroscopy ofminerals. AnnualReviews ofEarth and Planetary Sciences, 13,29-47

Kirkpatnck, R.J , Oestrike, R, Weiss, C.A , Smith, K.A., and Oldfield,E. (1986) High-resolution rTAl and 2esi NMR spectroscopy ofglasses andcrystals along the join CaMgSirO6-CaAlrSiO6 American Mineralogist,7 1, 705-7 t l

Klinowski, J. (1984) Nuclear magnetic resonance studies ofzeolites. Pro-gress in NMR Spectroscopy, 16,237-309

Klug, H.P , and Alexander, L.E. (1954) X-ray diffraction procedures. Wi-ley, New York, 716 p.

Krauth, L.P., and Epstein, S. (l 982) The nature ofwater in hydrous silica.American Mineralogist, 67, 510-520.

Konnert, J.H., and Appleman, D.E. (1978) The crystal structure of lowtridymite Acta Crystallographica, B34, 39 l-403.

Lippmaa, E., Miigi, M., Samoson, A, Engelhardt, G., and Grimmer, A.R.(I980) Structural studies ofsilicates by solid-state high-resolution rrsiNun. American Chemical Society Journal, 102, 4889-4893.

Maciel, G E., and Sindorf, D.W. (1980) Silicon-29 nuclear magnetic res-onance study ofthe surface ofsilica gel by cross polarization and magic-angle spinning. American Chemical Society Joumal, 102,'1606-j607.

Matson, D.W., Sharma, S.K., and Philports, J.A. (1983) The strucure ofbigh silica alkali-silicate glasses: A Raman spectroscopic investigation.Journal of Non-Crystalline Solids, 58, 323-352.

Maurer, R.D. (1960) Light scattenng by neutron irradiated silica. Inter-national Joumal of Physics and Chemistry of Solids, l7 , 4+51.

Murata, K.J., and Nakata, J.K (1974) Cristobalite stage in the diagenesisofdiatomaceous shale. Science. 184. 567-568

Murdoch, J.B., Stebbins, J.F., and Carmichael, I.S.E (1985) High reso-lution ,,Si Nun study ofsilicate and aluminosilicate glasses: The affectof network modifying cations. American Mineralogist, 70, 332-343.

Nukui, A., and Flijrke, O.W. (1987) Three tridymite structural modifi-cations and cristobalite intergrown in one crystal. American Mineral-o91st,12, 167-169

O'Keeffe, M., and Gibbs, G.V. (1984) Defects in amorphous silica: Abinitio calculations. Journal ofChemical Physics, 81, 876-8j9.

Otdfield, E, and Krrkpatrick, R.J. (198a) High resolution magneric res-onance of inorganic solids. Science, 22'7 , 1537-1544

Phillips, J.C. (1980) Chemical bonding, intemal surfaces, and the topol-ogy of noncrystalline solids. Physica Status Solidi, Bl0l, 4734j9.

t203

Primak, W. (1958) Fast-neutron-induced changes in quartz and vitreoussilica. Physical Review, I 10, 1240-1254.

Radeglia, R., and Engelhardt, G. (1985) Correlation ofSi-O-T (T: Sior Al) and 2eSi Nr*,rn chemical shifts in silicates and aluminosilicates.Interpretation by semi-empirical quantum-chemical considerations.Chemical Physics Letters, I 14, 28-30.

Ramdas, S, and Klinowski, J (1984) A simple correlation between iso-tropic "Si chemical shifts and T-O-T angles in zeolite frameworks.Nature.308.521-523.

Sanders, J V. ( I 985) Structure of opals. Joumal de Physique, 46, C3, 1-c3, 8.

Schneider, E., Stebbins, J.F., and Pines, A (1987) Speciation and localstructure in alkali and alkaline earth silicate glasses: Constraints from'?esi NMR spectroscopy. Joumal ofNon-Crystalline Solids, 89, 371-383.

Simonton, TC., Roy, R, Komarneni, S., and Breval, E. (1986) Micro-structure and mechanical properties of synthetic opal: A chemicallybonded ceramic. Joumal ofMaterials Research. 1.667-674.

Smith, J.V., and Blackwell, C.S. (1983) Nuclear magnetic resonance ofsilica polymorphs Nature, 303, 223-225.

Stebbins, J.F., Murdoch, J.B., Schneider, E, Carmichael, I.S.E., and Pines,A. (1985) A high temperature high-resolution xlrn study of ,3Na, ,7A1,

and':esi in molten silicates. Nature, 314, 250-252Taylor, M., and Brown, G.E (1979) Structure of mineral glasses-I. The

feldspar glasses NaAlSi3OE, KAlSirOE, CaAlrSirOr. Geochemica et Cos-mochimica Acta, 43, 6l-7 5.

Thattachari, YT., and Tiller, W.A. (1982) Steric factors in the spatialstructure of silicas. In R Srinivasan and R.H. Sarma, Eds., Confor-mation in biology, p. 445-452. Adenine Press, New York.

Thomas, J.M., Klinowski, J , Ramdas, S., Hunter, B.K., and Tennakoon,D T.B. (1983) The evaluation ofnonequivalent tetrahedral sites fromsilicon-29 Nun chemical shifts in zeolites and related aluminosilicatesChemical Physics Letters, 102, 158-162.

Turner, G.L., Kirkpatrick, RJ., Risbud, S.H., and Oldfield, E. (1987)Multinuclear magic-angle simple-spinning nuclear magnetic resonancespectroscopic studies of crystalline and amorphous ceramic materials.American Ceramic Society Bulletin, 66, 65 6-663.

Vieillard, P. (1986) Relation entre structure et parametres thermodyna-miques des phases de la silica. Bulletin de Mindralogre, 109,219-238.

Weeding, T.L., de Jong, B.H.W.S, Veeman, W.S., and Aitken, B.G. (1985)Silicon coordination changes from 4-fold to 6-fold on devitrification ofsilicon phosphate glass. Nature, 318, 352-353.

Weyl, W.A., and Marboe, E.Ch. (1959) Structural changes during meltingofcrystals and glasses. Joumal ofthe Sociely ofGlass Technology,43,417-437.

Yang, W.H., Kirkpatrick, R.J., and Tumer, G ( I 986) I'P and,,Si uesmvrninvestigation of the structural environments of phosphorus in alkaline-earth silicate glasses American Ceramic Society Journal, 69, C2Z2-

MeNuscnrrr REcETvED DrceMsen 18, 1986M.rrNuscnrpr AccEprED Jurv 14, 1987

DE JONG ET AL.: MORPHOGENESIS OF OPAL-CT