INTRODUCTION
Scapolite forms solid solutions between the end members marialite, Na4[Al3Si9O24]Cl=M e 0 and meionite, Ca4[Al6Si6O24]CO3=M e 100. Scapolite forms two series that meet at M e 75, Na2Ca6[Al10Si14O48](CO3)2, and the composition varies by replacement of [Na4⋅Cl]Si2 for [NaCa3⋅CO3]Al2 between M e 0–75, and by the replacement of [NaCa3⋅CO3]Si for [Ca4⋅CO3]Al between M e 75–100 (Evans et al., Reference Evans, Shaw and Haughton1969; Hassan and Buseck, Reference Hassan and Buseck1988; Deer et al., Reference Deer, Howie and Zussman1992). In addition, chemical analyses of scapolite are represented by two straight lines that meet at M e 75, again indicating two series (see Figure 12 in Hassan and Buseck, Reference Hassan and Buseck1988 or Figure 188 in Deer et al., Reference Deer, Howie and Zussman1992). The Ca–Na cations disordered on heating, but the Cl–CO3 order remains to 900°C (Antao and Hassan, Reference Antao and Hassan2002, Reference Antao and Hassan2008a, Reference Antao and Hassan2008b).
A number of studies on scapolite divided the series into three subseries that meet at M e 20–25 and M e 60–67 (e.g., Sokolova et al., Reference Sokolova, Kabalov, Sherriff, Teertstra, Jenkins, Kunath-Fandrei, Goetz and Jäger1996, Reference Sokolova, Gobechiya, Zolotarev and Kabalov2000; Teertstra and Sherriff, Reference Teertstra and Sherriff1996; Zolotarev Reference Zolotarev1996; Zolotarev et al., Reference Zolotarev, Petrov and Moshkin2003; Sherriff et al., Reference Sherriff, Sokolova, Kabalov, Teertstra, Kunath-Fandrei, Goetz and Jäger1998, Reference Sherriff, Sokolova, Kabalov, Jenkins, Kunath-Fandrei, Goetz, Jäger and Schneider2000; Teertstra et al., Reference Teertstra, Schindler, Sherriff and Hawthorne1999; Hawthorne and Sokolova, Reference Hawthorne and Sokolova2008; Sokolova and Hawthorne, Reference Sokolova and Hawthorne2008). Ulbrich (Reference Ulbrich1973a) suggested a break in cell parameters at M e 65–66 (his Figure 1).
Using transmission electron microscopy (TEM), and the type-b reflections (h+k+l=odd), three subseries were identified: series M e 0–18 and M e 90–100 have space group I4/m, whereas series M e 18–90 has space group P42/n (Seto et al., Reference Seto, Shimobayashi, Miyake and Kitamura2004). Some scapolite samples were refined in both space groups I4/m and P42/n, for example, ON8 (M e 21) and ON45 (M e 93) samples do not show type-b reflections, but that does not mean that they are body centered (Lin and Burley, Reference Lin and Burley1973c). These authors indicated that the type-b reflections are less intense than the detection limit of the X-ray experiments and suggested that all scapolite structures belong to space group P42/n, except the end members, M e 0 and M e 100, which do not occur naturally.
In space group P42/n, the T2 and T3 tetrahedra are symmetrically distinct and are not required to have equal ⟨T2-O⟩ and ⟨T3-O⟩ distances (Figure 1). Each of the T2 and T3 sites with a multiplicity of 8 in the P42/n structure is symmetrically distinct, and together they form the T2′ site with a multiplicity of 16 in space group I4/m. Sokolova and Hawthorne (Reference Sokolova and Hawthorne2008) refined the structures of their S15 sample (M e 76.9) in both space groups, I4/m and P42/n. They test for the best space group for S15 by comparing the ⟨T2-O⟩ and ⟨T3-O⟩ distances in the P42/n refinement. If these distances are equal, the space group is I4/m; if these distances are not equal, the space group is P42/n. Their relevant distances are 1.691(2) and 1.666(2) Å, so they assign the space group
Figure 1. (Color online) (a) Scapolite structure showing framework T and interstitial M and A sites, four-membered rings, and oval shaped channels. (b) A cage containing A and M ions and uncommon five-membered rings; A is coordinated by four M in a square-planar configuration. Space group P42/n.
P42/n to sample S15. In spite of refining their S15 sample (M e 76.9) in space group P42/n, Sokolova and Hawthorne (Reference Sokolova and Hawthorne2008) claimed a phase transition at M e 60–67 based on a break in the c-cell parameters. Furthermore, they refined other samples close to M e 0 and M e 100 in space group I4/m, thereby implying two phase transitions with an intermediate series that is refined in space group P42/n.
Hassan and Antao (Reference Hassan and Antao2010) observed a discontinuity in unit-cell parameters at M e 75 and the absence of any phase transition across the scapolite series; a maximum in the c parameter occurs at M e 37.5. In this study, the structures of a marialite (M e 6) and a meionite (M e 93) sample were refined in both space groups I4/m and P42/n, and similar results were obtained. There is no reason for phase transitions at M e 20–25, M e 60–67, or M e 75 and no reason to refine the structure near M e 0 and M e 100 in the higher symmetry space group I4/m, thereby forcing the T2 and T3 sites in P42/n to be the same and constitute the T2′ site in I4/m.
The structure of scapolite was determined by Pauling (Reference Pauling1930) and Schiebold and Seumel (Reference Schiebold and Seumel1932) and refined by several researchers in either space group I4/m or P42/n (Papike and Zoltai, Reference Papike and Zoltai1965; Papike and Stephenson, Reference Papike and Stephenson1966; Lin and Burley, Reference Lin and Burley1973a, Reference Lin and Burley1973b, Reference Lin and Burley1973c, Reference Lin and Burley1974; Ulbrich, Reference Ulbrich1973b; Levien and Papike, Reference Levien and Papike1976; Peterson et al., Reference Peterson, Donnay and LePage1979; Aitken et al., Reference Aitken, Evans and Konnert1984; Comodi et al., Reference Comodi, Mellini and Zanazzi1990; Belokoneva et al., Reference Belokoneva, Sokolova and Dorokhova1991, Reference Belokoneva, Sokolova and Urusov1993; Teertstra et al., Reference Teertstra, Schindler, Sherriff and Hawthorne1999; Sokolova and Hawthorne, Reference Sokolova and Hawthorne2008). More data were obtained using Rietveld refinement and MAS NMR spectroscopy (Sokolova et al., Reference Sokolova, Kabalov, Sherriff, Teertstra, Jenkins, Kunath-Fandrei, Goetz and Jäger1996, Reference Sokolova, Gobechiya, Zolotarev and Kabalov2000; Sherriff et al., Reference Sherriff, Sokolova, Kabalov, Teertstra, Kunath-Fandrei, Goetz and Jäger1998, Reference Sherriff, Sokolova, Kabalov, Jenkins, Kunath-Fandrei, Goetz, Jäger and Schneider2000), valence-matching principle (Hawthorne and Sokolova Reference Hawthorne and Sokolova2008), and electrostatic energy calculations (Chamberlain et al., Reference Chamberlain, Docka, Post and Burnham1985). A review of scapolite crystal chemistry was given by Lin (Reference Lin1975).
EXPERIMENTAL
A marialite sample from Badakhshan, Afghanistan and a meionite sample from Mt. Vesuvius, Italy were used in this study. The marialite crystals are large (0.5×0.5×0.5 cm3), purple in color, and are of gem quality. The meionite crystal is colorless and was obtained from the Smithsonian National Museum of Natural History (B20018-1). Their chemical analyses were obtained by electron microprobe. The crystals are homogeneous based on optical observation and microprobe data. The chemical composition for the marialite sample is K0.24Na3.37Ca0.24[Al3.16Si8.84O24]Cl0.84(SO4)0.02(CO3)0.15 and corresponds to M e 6.2; the meionite sample is Na0.29Ca3.76[Al5.54Si6.46O24]Cl0.05(SO4)0.02(CO3)0.93 and corresponds to M e 92.9.
A small (about 0.2×0.2×0.2 mm3) crystal was handpicked under a binocular microscope and finely crushed in an agate mortar and pestle for synchrotron high-resolution powder X-ray diffraction (HRPXRD) experiments that were performed at beamline 11-BM, Advanced Photon Source, Argonne National Laboratory. The sample was loaded into a Kapton capillary and rotated during the experiment at a rate of 90 rotations/s. The data were collected to a maximum 2θ of about 40° with a step time of 0.1 s per step. The step size was 0.001° and 0.0005° for M e 6 and M e 93, respectively. Additional details of the experimental set-up are given elsewhere (Antao et al., Reference Antao, Hassan, Wang, Lee and Toby2008; Lee et al., Reference Lee, Shu, Ramanathan, Preissner, Wang, Beno, Von Dreele, Ribaud, Kurtz, Antao, Jiao and Toby2008; Wang et al., Reference Wang, Toby, Lee, Ribaud, Antao, Kurtz, Ramanathan, Von Dreele and Beno2008).
The HRPXRD method, instead of the single-crystal method, was chosen because it is rapid (about 1 h for a data set), it provides superior cell parameters, and bond-distances and angles that are comparable to those obtained by the single-crystal method. Weak reflections, including satellite reflections, are easily observed because of the high beam intensity of synchrotron X-rays. In addition, single-crystal data are available to compare with our present HRPXRD data. Moreover, Antao et al. (Reference Antao, Hassan, Wang, Lee and Toby2008) compared data from HRPXRD and single-crystal methods for several minerals and showed that the results are comparable.
Figure 2. HRPXRD traces for (a) marialite, M e 6, and (b) meionite, M e 93, in space group P42/n together with the calculated (continuous line) and observed (crosses) profiles. The difference curves (Iobs–Icalc) are shown at the bottom. The short vertical lines indicate allowed reflection positions. The traces beyond 20° 2θ are scaled by 20x.
RIETVELD STRUCTURE REFINEMENT
The crystal structure of marialite was modeled using the Rietveld method (Rietveld, Reference Rietveld1969) that is incorporated in the GSAS program (Larson and Von Dreele, Reference Larson and Von Dreele2000) and using the EXPGUI interface (Toby, Reference Toby2001). Initial structural parameters were taken from Teertstra et al. (Reference Teertstra, Schindler, Sherriff and Hawthorne1999). The structure refinement was carried out by varying parameters in the following sequence: scale factor, background, cell, zero shift, profile, atom positions, and isotropic displacement parameters. Finally, all variables were refined simultaneously until convergence was achieved. Site occupancy factors for the M cations and A anions were fixed to that obtained by chemical analyses. We assumed that the A anions completely occupy the 4/m site, the disordered trigonal CO3 group occurs in a small quantity, so its O atoms were not modeled in the refinement for marialite, but they occur in significant quantity in meionite, where they are modeled quite well. Figures 2a and 2b display the HRPXRD trace for marialite, M e 6 and meionite, M e 93, respectively. Table I contains the refinement statistics and unit-cell parameters. Table II contains the atom positions and isotropic displacement parameters, U. Table III contains selected bond-distances and angles.
RESULTS AND DISCUSSION
Marialite (M e 6)
The unit-cell parameters for marialite are a=12.047555(7) Å, c=7.563210(6) Å, and V
TABLE I. Refinement data for marialite (M e 6) and meionite (M e 93).
1 From electron microprobe chemical analyses; M e(%)=[Ca/(Ca+Na+K)]×100.
=1097.751(1) Å3. Sample S2 (M e 7) from Sokolova and Hawthorne (Reference Sokolova and Hawthorne2008) has a=12.0541(5) Å, c=7.5682(3) Å, and V=1099.7(1) Å3. The differences between S2 and our sample are 0.007 Å for a, 0.005 Å for c, and 1.95 Å3 for V.
The refined marialite structures in both space groups I4/m and P42/n gave identical results (Table III). The average ⟨T1-O⟩ distances are 1.599(1) and 1.600(2) Å for space groups I4/m and P42/n, respectively. Therefore, the T1 site contains only Si atoms. The average ⟨T2′-O⟩=1.659(1) Å in space group I4/m and is identical to the mean of ⟨T2-O⟩=1.655(2) and ⟨T3-O⟩=1.664(2) Å, which is 1.659(2) Å (=⟨T2,3-O⟩) for the lower symmetry space group P42/n. The distinct T2 and T3 sites are not identical in space group P42/n; they do not have identical ⟨T-O⟩ distances. The T2 and T3 sites are forced to be a T2′ site in space group I4/m. The Al–Si atoms are disordered in the T2 and T3 sites in space group P42/n, and the T2′ site in space group I4/m. The ⟨M-O⟩[7] (M site is coordinated to seven framework O atoms) and M-A distances are identical in both space groups (Table III). The structure in space group I4/m has a smaller number of observed reflections, Nobs=1429, whereas in space group P42/n, Nobs=2680; nearly twice as many reflections (the type-b reflections are very weak towards M e 0 and M e 100). The structural results for marialite in space groups I4/m and P42/n are identical, but the ⟨T2-O⟩ and ⟨T3-O⟩ distances are not equal. Therefore, the P42/n space group is preferred and there is no reason for a phase transition at M e 20–25.
The S2 sample (M e 7) from Sokolova and Hawthorne (Reference Sokolova and Hawthorne2008) in space group I4/m has ⟨T1-O⟩=1.607 Å and ⟨T2′-O⟩=1.662 Å, which are similar to our values given above for the same space group I4/m. Their ⟨M-O⟩[7]=2.736 Å and M-A=2.945(1) Å, whereas we obtained 2.754(2) Å and 2.914(1) Å, respectively, for both space groups (Table III). The difference between these ⟨M-O⟩[7] distances is 0.018 Å and between the M-A distances is 0.031 Å.
In sodalite, Na8[Si6Al6O24]Cl2, each Cl atom is surrounded by four Na atoms in a tetrahedral configuration and Na–Cl=2.7337(4) Å (Hassan and Grundy, Reference Hassan and Grundy1984; Hassan et al., Reference Hassan, Antao and Parise2004; Antao et al., Reference Antao, Hassan, Wang, Lee and Toby2008). In marialite, each Cl atom is
TABLE II. Atom positions and U (Å2) for marialite (M e 6) and meionite (M e 93).
1 M=(Ca,Na,K).
2 A=(Cl,C,S).
also surrounded by four Na atoms in a square-planar configuration and Na–Cl=2.914(1) Å. Repulsion of the Na atoms in the square-planar configuration causes the Na–Cl distance to be larger in marialite compared to that in sodalite.
Meionite (M e 93)
The unit-cell parameters for meionite (M e 92.9) are a=12.19882(1) Å, c=7.576954(8) Å, and V=1127.535(2) Å3. Sample S18 (M e 92.8) from Sokolova and Hawthorne (Reference Sokolova and Hawthorne2008) for a sample from the same locality has a=12.2077(5) Å, c=7.5832(3) Å, and V=1130.1(1) Å3. The differences between S18 and our sample are 0.0089 Å
TABLE III. Bond distances (Å) and angles (deg) for marialite (M e 6) and meionite (M e 93).
In space group P42/n, ⟨T2,3-O⟩=[⟨T2-O⟩+⟨T3-O⟩]/2 and mean T=[⟨T1-O⟩+⟨T2-O⟩+⟨T3-O⟩]/3. In space group I4/m, the mean T=[⟨T1-O⟩+(2×⟨T2′-O⟩)]/3.
for a, 0.0062 Å for c, and 2.57 Å3 for V. Although these samples are from the same locality and have similar composition, our cell parameters are different from those reported by Sokolova and Hawthorne (Reference Sokolova and Hawthorne2008).
The refined meionite structure in both space groups I4/m and P42/n gave similar results (Table III). The average ⟨T1-O⟩ distances are 1.660(1) Å and 1.661(1) Å for space groups I4/m and P42/n, respectively. Therefore, the Al and Si atoms are partially disordered in the T1 site. The average ⟨T2′-O⟩=1.669(1) Å in space group I4/m, and is identical to the mean of ⟨T2-O⟩=1.675(1) and ⟨T3-O⟩=1.664(1) Å, which is ⟨T2,3-O⟩=1.670(1) Å for space group P42/n. Based on ⟨T-O⟩ distances, the Al–Si order in the distinct T2 and T3 sites are not identical in space group P42/n. The Al–Si atoms are partially disordered in the T2 and T3 sites in space group P42/n, and the T2′ site in space group I4/m. The ⟨M-O⟩[7] and M-A distances are identical in both space groups (Table III). For the disordered CO3 group, a more reasonable C–O distance is obtained in space group P42/n compared to I4/m. The structural results for meionite in space groups I4/m and P42/n are similar, so there is no reason for a phase transition at M e 60–67 and have the structure in the higher symmetry space group I4/m. The meionite structure in space group I4/m has the number of observed reflections, Nobs=1594, whereas in space group P42/n, Nobs=3006; nearly twice as many reflections (the type-b reflections are very weak toward M e 93). Refinement of the structure in the lower symmetry space group P42/n shows that the T2 and T3 sites are not equivalent. In M e 6, ⟨T2-O⟩ is slightly lesser than ⟨T3-O⟩, whereas in M e 93, ⟨T2-O⟩ is slightly greater than ⟨T3-O⟩ (Table III).
The S18 sample (M e 92.8) from Sokolova and Hawthorne (Reference Sokolova and Hawthorne2008) has ⟨T1-O⟩=1.658 Å and ⟨T2′-O⟩=1.683 Å, compared to our ⟨T1-O⟩=1.660(1) Å and ⟨T2′-O⟩=1.669(1) Å in space group I4/m. Their ⟨M-O⟩[7]=2.639 Å and M-A=3.196(1) Å, whereas we obtained 2.655(1) Å and 3.1804(5) Å, respectively, for both space groups (Table III). The difference between these respective ⟨M-O⟩[7] and M-A distances is 0.016 Å for both, and probably arise from the different cell parameters used.
Why are the type-b reflections unobserved near M e 0 and M e 100? The simple answer is that they are too weak to be observed. Lin and Burley (Reference Lin and Burley1973c) did not observe type-b reflections in their ON8 (M e 21) and ON45 (M e 93) samples, but they refined the structure of these samples in both space groups. With TEM, type-b reflections were observed between M e 18–90 (Seto et al., Reference Seto, Shimobayashi, Miyake and Kitamura2004). Phakey and Ghose (Reference Phakey and Ghose1972) and Hassan and Buseck (Reference Hassan and Buseck1988) showed that type-b reflections arise from Cl–CO3 order, instead of Al–Si order, and give rise to antiphase domain boundaries (APBs). Complete order of Cl–CO3 is achieved where their ratio is 1:1, which occurs at M e 37.5, [Na5Ca3[Al8Si16O48]Cl(CO3)] (Hassan and Buseck, Reference Hassan and Buseck1988). For the M e 6 sample, the amounts of Cl and CO3 are 0.84 and 0.15, respectively. This means that 0.15 CO3 can couple with 0.15 Cl to form an ordered distribution, whereas the remaining 0.69 Cl are distributed randomly throughout the structure. This 0.15 CO3–0.15 Cl order distribution occurs in small domains (about 15% of the sample). Only if such small domains are observed by TEM, then the type-b reflections may be observed in a high-voltage TEM. The chance of observing such small domains becomes challenging toward M e 0, but they are easily observed at M e 37.5, where 1:1 Cl–CO3 order occurs throughout the crystal (Hassan and Buseck, Reference Hassan and Buseck1988). Using high-resolution transmission electron microscopy (HRTEM), domains from series-1 scapolite (M e 0–75) were observed as inclusions in series-2 scapolite (M e 75–100), so type-b reflections were observed in a M e 79.6 scapolite sample, but a discontinuity in cell parameters and composition occurs at M e 75, [Na2Ca6[Al10Si14O48](CO3)2], where the A site is completely filled with CO3 groups from M e 75–100 (Hassan and Buseck, Reference Hassan and Buseck1988; Hassan and Antao, Reference Hassan and Antao2010). Scapolite analyses fall on two straight lines that meet at M e 75. A discontinuity occurs in the unit-cell parameters at M e 75 and no phase transition occurs across the scapolite series.
The structure of marialite (M e 6) and meionite (M e 93) refines equally well in both space groups I4/m and P42/n, and similar structural results were obtained. There is no reason to refine samples close to M e 0 and M e 100 in the higher symmetry space group I4/m, thereby implying transitions and forcing the T2 and T3 sites to be the same T2′ site. In space group P42/n, M e 6 and M e 93 have ⟨T2-O⟩ and ⟨T3-O⟩ distances that not identical. Therefore, space group I4/m should not be used for scapolite. We reconfirm that scapolite forms two solid solutions that meet at M e 75, where a discontinuity occurs in chemical composition and unit-cell parameters (Hassan and Antao, Reference Hassan and Antao2010), as was previously noted (Evans et al., Reference Evans, Shaw and Haughton1969; Hassan and Buseck, Reference Hassan and Buseck1988). Moreover, there is no phase transition across the scapolite series.
ACKNOWLEDGMENT
We thank J. Post for the M e 93 sample. Use of the Advanced Photon Source was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-06CH11357. This work was supported by a University of Calgary grant, a discovery grant from the National Science and Engineering Research Council of Canada, and an Alberta Ingenuity New Faculty Award.
