The crystal chemistry of ca10-y(sio4) 3(so4)3cl2-x-2yfx ellestadite

Article abstract of DOI:10.1021/ic201673r

Fluor-chlorellestadite solid solutions Ca₁₀(SiO ₄)₃(SO₄)₃Cl_{2 -x}F_x, serving as prototype crystalline matrices for the fixation of hazardous fly ash, were synthesized and characterized by powder X-ray and neutron diffraction (PXRD and PND), transmission electron microscopy (TEM), and Fourier transform infrared spectroscopy (FTIR). The lattice parameters of the ellestadites vary linearly with composition and show the expected shrinkage of unit cell volume as fluorine (IR = 1.33 A) displaces chlorine (IR = 1.81 A). FTIR spectra indicate little or no OH^- in the solid solutions. All compositions conform to P6₃/ m symmetry where F- is located at the 2a (0, 0, 1/4) position, while Cl^- is displaced out of the 6h Ca(2) triangle plane and occupies 4e (0, 0, z) split positions with z ranging from 0.336(3) to 0.4315(3). Si/S randomly occupy the 6h tetrahedral site. Ellestadites rich in Cl (x ≤ 1.2) show an overall deficiency in halogens (<2 atom per formula unit), particularly Cl as a result of CaCl2 volatilization, with charge balance achieved by the creation of Ca vacancies (Ca²⁺ + 2Cl^- →□_Ca+ 2□_lC) leading to the formula Ca_10-y(SiO ₄)₃(SO₄)₃Cl₂- _x-2yF_x. For F-rich compositions the vacancies are found at Ca(2), while for Cl-rich ellestadites, vacancies are at Ca(1). It is likely the loss of CaCl₂ which leads tunnel anion vacancies promotes intertunnel positional disorder, preventing the formation of a P21/b monoclinic dimorph, analogous to that reported for Ca₁₀(PO ₄)₆Cl₂. Trends in structure with composition were analyzed using crystal-chemical parameters, whose systematic variations served to validate the quality of the Rietveld refinements.

Full text of DOI:10.1021/ic201673r

Article
pubs.acs.org/IC
The Crystal Chemistry of Ca10−y(SiO4)3(SO4)3Cl2−x−2yFx Ellestadite
Yanan Fang,*,† Clemens Ritter,‡ and Tim White†,§
†School of Materials Science & Engineering, Nanyang Technological University, 50 Nanyang Avenue, 639798, Singapore
‡Institute Laue-Langevin, 6 rue Jules Horowitz, 38042 Grenoble Cedex 9, France
§Centre for Advanced Microscopy, Sullivan’s Creek Road, Australian National University, Canberra ACT 0200 Australia
S
* Supporting Information
ABSTRACT: Fluor-chlorellestadite solid solutions Ca10(SiO4)3(SO4)3Cl2−xFx, serving as prototype crystalline matrices for the
fixation of hazardous fly ash, were synthesized and characterized by powder X-ray and neutron diffraction (PXRD and PND),
transmission electron microscopy (TEM), and Fourier transform infrared spectroscopy (FTIR). The lattice parameters of the
ellestadites vary linearly with composition and show the expected shrinkage of unit cell volume as fluorine (IR = 1.33 Å)
displaces chlorine (IR = 1.81 Å). FTIR spectra indicate little or no OH− in the solid solutions. All compositions conform to P63/
1
m symmetry where F− is located at the 2a (0, 0, /4) position, while Cl− is displaced out of the 6h Ca(2) triangle plane and
occupies 4e (0, 0, z) split positions with z ranging from 0.336(3) to 0.4315(3). Si/S randomly occupy the 6h tetrahedral site.
Ellestadites rich in Cl (x ≤ 1.2) show an overall deficiency in halogens (<2 atom per formula unit), particularly Cl as a result of
CaCl2 volatilization, with charge balance achieved by the creation of Ca vacancies (Ca2+ + 2Cl− →□Ca + 2□Cl) leading to the
formula Ca10−y(SiO4)3(SO4)3Cl2−x−2yFx. For F-rich compositions the vacancies are found at Ca(2), while for Cl-rich ellestadites,
vacancies are at Ca(1). It is likely the loss of CaCl2 which leads tunnel anion vacancies promotes intertunnel positional disorder,
preventing the formation of a P21/b monoclinic dimorph, analogous to that reported for Ca10(PO4)6Cl2. Trends in structure with
composition were analyzed using crystal-chemical parameters, whose systematic variations served to validate the quality of the
Rietveld refinements.
(Figure 1b). Saint-Jean and Sweden9 synthesized chlorellestadite
under different conditions (annealing time, temperature, CaCl2
1. INTRODUCTION
Structural adaptation in apatites makes them suitable for
environmental remediation,1 catalysis and energy production,2
content), but only the hexagonal phase was observed. Thus, the
precise arrangement of the metalloids in ellestadite and its
as well as the traditional biomedical uses of bone and teeth
replacement.3,4 Optimization of chemically substituted apatites
symmetry remain in question, and may be dependent on
petrogenesis or fabrication conditions.
for these applications demands a comprehensive understanding
of their chemical and crystallographic properties. Apatites also
The scope of this work was to synthesize and characterize
fluor-chlorellestadites, Ca10(SiO4)3(SO4)3Cl2−xFx, as simplified
show promise as waste form ceramics for the immobilization of
incinerator fly ash,5 however our crystallochemical knowledge is
surrogates for apatite derived from the stabilization of certain
solid waste ashes, and establish the continuity of the solid
presently insufficient to tailor formulations and predict
solution series. The role of substituted fluorine in promoting
durability reliably.
higher chemical stability (e.g., reducing dental caries10) and
Ellestadite is a silicon- and sulfur-rich member of the apatite
family of the general composition Ca10(SiO4)3−x/2
thermal resistance (e.g., higher melting points11) is well-known,
-
(PO4)x(SO4)3−x/2(F/Cl/OH)2,6 when described in P63/m
symmetry the unit cell contains two formula units and the
crystallochemical representation is [A(1)]4[A(2)]6(BO4)6X2
(A = Ca, B = Si/P/S, and X = F/Cl/OH) (Figure 1a). The
topology is zeolite-like, in that a framework composed of
Ca(1)O6 metaprism columns, corner-connected to (Si/P/S)O4
tetrahedra, creates tunnels perpendicular to the basal plane that
enclose Ca(2)O5(F/Cl/OH) polyhedra. While phosphate
apatites are generally hexagonal, the symmetry of ellestadite
remains controversial. In a single crystal X-ray diffraction study
of a specimen from Chichibu Mine, Sudarsanan7 assigned the
space group as P21/m that differentiated the substitutional
positions of SO4 and SiO4 for PO4 groups. This ordering was
incomplete and the presence of trace elements obscured an
unambiguous solution. In another mineral study, Organova et
al.8 reported P21 symmetry which yields three independent (Si/
P/S)O4 tetrahedral sites and two discrete CaO6 metaprisms
with the interaction between fluorine and chlorine capable of
attenuating these properties. Commonly, the fluor- and chlo-
rellestadites are regarded as isostructural, although the halides
reside at different locations along the c-axis anion tunnels. In
the synthetic fluorellestadite (P63/m) endmember, F− at (0, 0,
3
1/4 and 0, 0, /4) lies in the center of a Ca(2) triangle on the
12
1
3
mirror planes at z = /4 and z = /4. In the corresponding
chlorellestadite endmember, Cl− cannot fit into the Ca(2)
triangle (ionic radii for VIF− = 1.33 Å and VICl− = 1.81 Å13),
and statistically occupies (0, 0, z) sites, ∼1.1 Å above or below
the mirror plane.9 In the case of natural hydroxyellestadite,
Sudarsanan7 refined separate sites for OH− (0, 0, 0.2033), F−
(0, 0, 1/4), and Cl− (0, 0, 0.3644) in P21/m with site
occupancies of 0.806(9), 0.15(6), and 0.092(5) respectively.
Received: August 3, 2011
Published: November 23, 2011
© 2011 American Chemical Society
12641
dx.doi.org/10.1021/ic201673r | Inorg. Chem. 2011, 50, 12641−12650

Inorganic Chemistry
Article
Figure 2. Expanded views of FT-IR spectra for the fluor-
chlorellestadite solid solution. (x in Ca10(SiO4)3(SO4)3Cl2−xFx).
Figure 1. (a) Hexagonal P63/m apatite structure of [A4][A6](BO4)6X2.
The AO6 metraprisms contain calcium, while the BO4 tetrahedra
accommodate phosphorus, sulfur and silicon. The tunnel contains
fluorine, chlorine or oxygen without intertunnel correlation. (b)
Monoclinic P21 apatite structure of [A4][A6](BO4)6X2. The AO6
metraprisms contain calcium while the BO4 tetrahedra accommodate
phosphorus, sulfur and silicon. The tunnels contain fluorine, chlorine,
or oxygen where intertunnel correlation leads a doubling of the b-axis
metric.
electron microscopy were used here to examine the structural
adaptations and nanometric properties of synthetic fluor-
chlorellestadite.
2. EXPERIMENTAL METHODS
2.1. Synthesis. Ca10(SiO4)3(SO4)3Cl2−xFx (x = 0, 0.4, 0.8, 1.2, 1.6,
and 2) were synthesized by solid-state reaction. The starting materials
were prepared by preheating CaCO3 (Merck, p.a.) at 1000 °C for
12 h; SiO2 (Sigma-Aldrich, 99.99%) and CaSO4·2H2O (Merck, p.a.) at
Onac et al.14 suggested minor substitutions of F− and Cl− at the
same site as OH− in a hydroxylellestadite (P63/m) from
Cioclovina Cave, Romania, although at the unit cell scale this is
unlikely to satisfy bond valence sum criteria. In general, the
interaction between F− and Cl−, and the influence on
Ca10(SiO4)3(SO4)3X2 (where X = F, Cl) ellestadite require
clarification. Two reports concerning the structural adjustments
and anion tunnel accommodation in binary solid solutions of
(Cl, OH) or (F, OH) in calcium phosphate apatite Ca10-
(PO4)6X2, and ternary solutions of (F, Cl, OH)15,16 have
appeared, but less is known about the (F, Cl) system. Mackie
and Young17 studied two fluor-chlorapatite compositions with
F/(F+Cl) ratios of 0.17 and 0.41, and included a second
fluorine site in addition to that on the mirror plane. Hughes
et al.18 suggested that OH− may be an essential component in
stabilizing natural fluor-chlorapatite solid solutions. The
presence of a miscibility gap in natural gem-grade apatite
from Brazil has been described by Ferraris et al.,19 where atomic
scale studies by high-resolution and analytical TEM revealed
the phase separation of fluorine-enriched and chlorine-enriched
phosphate apatite domains from ellestadite in the nanoscale.
An understanding of the structural atomic arrangement and
physical-chemical properties of fluor-chlorellestadite is an
essential prerequisite for formulating synthetic ellestadites for
hazardous waste immobilization. To this end, infrared spectros-
copy, powder X-ray and neutron diffraction, and transmission
600 °C for 4 h; and CaCl ·2H O (Riedel-de Haen, puriss.) at 400 °C
̈
2
2
for 4 h. Stoichiometric mixtures of CaO, CaSO4, SiO2, CaCl2, and
CaF2 were manually ground with acetone and calcined at 900 °C for
5 h in an alumina crucible, then reground and sintered at 950 °C for a
further 9 h.
2.2. Crystallographic Characterization. Powder X-ray diffrac-
tion (PXRD) and neutron diffraction (PND) were combined for
structural characterization. PXRD measurements were performed for
phase identification and mass balance using a Bruker D8 high-
resolution X-ray powder diffractometer (Cu−Kα radiation) operated
at 40 kV and 40 mA. Scans were collected from 8° to 140° with a step
size of 0.01° and a dwell time per step of 1 s. All specimens were
spiked with standard silicon (10 wt %, NIST660a, a = 5.40825 Å) to
determine the unit cell constants absolutely. Powder neutron
diffraction patterns were acquired at the Institute Laue-Langevin
(ILL) Grenoble (France) using the high resolution D1A diffrac-
tometer with a wavelength of 1.3894 Å, a 2-theta range of 0° to 150°
and scan step of 0.05° at room temperature. The diffractometer is
equipped with a focusing Ge monochromator set at a takeoff angle of
3
123°, and 25 standard He counter detectors. Approximately 2 g of
ellestadite was loaded into a 9 mm diameter cylindrical sample
container that was rotated during acquisition.
The structure refinements were carried out by the Rietveld method
using TOPAS20 with a pseudo-Voigt peak shape profile and a starting
model using the atomic positions of Ca10(SiO4)3(SO4)3Cl2 in
hexagonal P63/m.21 Although monoclinic forms have been docu-
mented, such occurrences were not observed in this study, as 2b
12642
dx.doi.org/10.1021/ic201673r | Inorg. Chem. 2011, 50, 12641−12650

Inorganic Chemistry
Article
Figure 3. Rietveld fitting pattern for (a) x = 0, (b) x = 0.4 and (c) x = 2. (x in Ca10(SiO4)3(SO4)3Cl2−xFx) Lines and dots denote observed and
calculated profile intensities, respectively. Short vertical bars represent Bragg reflection positions. The difference (observed − calculated) is plotted
below.
superstructure reflections were absent, and improvements in R values
with lower symmetries (P21/m; P21) were insignificant. For each data
set, a five-coefficient Chebychev function and 1/x background, a peak
shift, a zero error, unit cell parameters, and crystal size were refined
sequentially with temperature displacement parameters and occupa-
tion factors of each atomic site kept constant. After several refinement
cycles, the calcium positions (Ca(1) and Ca(2)) were released,
followed by the silicon/sulfur and oxygen positions. The mole ratio of
Si:S at the tetrahedral site was fixed at 1:1. The oxygen sites were fully
tenanted, while the occupancies of calcium, chlorine and fluorine were
allowed to vary. For nonstoichiometric ellestadites, the occupancy
factors of Ca and Cl were constrained to maintain charge balance (see
Discussion and Table 3). Isotropic temperature displacement
parameters for all atoms were refined in Ca(1)/Ca(2)/Cl/F and Si/
S/O(1)/O(2)/O(3) groups to the same value. The refinement cycles
were terminated when convergence was reached. Scattering lengths of
0.4700, 0.4149, 0.2847, 0.5803, 0.5654, and 0.9577 × 10−12 cm were
used for Ca, Si, S, O, F, and Cl, respectively.22 Difference-Fourier maps
were generated using Jana2006,23 specifically to examine the location
of anion scattering centers in the tunnels. The Ca(1)O6 metaprism
twist angle (φ) was calculated according to the method of White and
Dong24 for cystallochemical diagnosis, and verify trends bond angles as
a function of composition. To objectively assess the reliability of the
Rietveld refinements singular value decomposition as implemented in
12643
dx.doi.org/10.1021/ic201673r | Inorg. Chem. 2011, 50, 12641−12650

Inorganic Chemistry
Article
a
Table 1. Lattice Parameters and Crystallographic Data for Ca10(SiO4)3(SO4)3Cl2-xFx
x
0
0.4
P63/m
9.6239 (3)
6.87749 (3)
551.64(2)
20.44
0.8
P63/m
9.5772(2)
6.8950(1)
547.70(3)
21.89
1.2
P63/m
9.5312(2)
6.9181(2)
544.27(3)
22.01
1.6
P63/m
9.5064(2)
6.9263(1)
542.09(2)
22.63
2.0
space group
a (Å)
P63/m
P63/m
9.6702(3)
6.8543(3)
555.09(4)
19.92
9.4486(1)
6.9426(1)
536.77(2)
24.15
c (Å)
volume (Å3)
φ(deg)
Rwp
5.62%
4.30%
5.43%
6.72%
6.74%
8.55%
Rp
4.28%
3.37%
4.21%
5.24%
5.05%
6.48%
GOF
0.84
0.58
0.61
0.73
0.74
0.89
2
Ca(1), 4f (1/3, /3, z)
z
−0.0013(1)
1.88(6)
0.0009(4)
1.85(7)
0.0010(5)
1.81(6)
0.0037(9)
1.52(7)
0.0028(4)
1.52(7)
0.0026(1)
1.46(7)
B (Å2)
1
Ca(2), 6h (x, y, /4)
x
0.2603(5)
0.0012(6)
1.88(6)
0.2562(4)
−0.0017(3)
1.85(7)
0.2497(3)
−0.0068(4)
1.81(6)
0.2435(1)
−0.0101(4)
1.52(7)
0.2420(3)
−0.0094(3)
1.52(7)
0.2390(4)
−0.0088(2)
1.46(7)
y
B (Å2)
1
Si/S, 6h (x, y, /4)
x
y
0.4062(1)
0.3767(1)
1.84(3)
0.4045(2)
0.3755(4)
2.00(4)
0.4008(2)
0.3721(1)
1.79(3)
0.3996(1)
0.3705(2)
1.60(4)
0.3996(1)
0.3706(1)
1.52(4)
0.3980(1)
0.3704(3)
1.51(3)
B (Å2)
1
O(1), 6h (x, y, /4)
x
y
0.3368(5)
0.4900(4)
1.84(3)
0.3359(4)
0.4902(8)
2.00(4)
0.3331(1)
0.4885(1)
1.79(3)
0.3296(4)
0.4859(4)
1.60(4)
0.3302(4)
0.4870(3)
1.52(4)
0.3245(3)
0.4843(3)
1.51(3)
B (Å2)
1
O(2), 6h (x, y, /4)
x
y
0.5907(1)
0.4676(4)
1.84(3)
0.5898(1)
0.4667(3)
2.00(4)
0.5872(1)
0.4651(2)
1.79(3)
0.5867(1)
0.4662(4)
1.60(4)
0.58724(8)
0.4660(3)
1.52(4)
0.5868(3)
0.4676(3)
1.51(3)
B (Å2)
O(3), 12i (x, y, z)
x
y
z
0.3549(4)
0.2675(3)
0.0684(3)
1.84(3)
0.3512(1)
0.2647(2)
0.0701(1)
2.00(4)
0.3483(1)
0.2628(1)
0.0688(1)
1.79(3)
0.3443(4)
0.2599(3)
0.0700(2)
1.60(4)
0.3426(3)
0.2587(2)
0.0712(2)
1.52(4)
0.3410(3)
0.2573(2)
0.0717(2)
1.51(3)
B (Å2)
Cl, 4e (0, 0, z)
z
0.4315(3)
1.88(6)
0.4169(1)
1.85(7)
0.3818(6)
1.81(6)
0.342(6)
1.52(7)
0.336(3)
1.52(7)
B (Å2)
1
F, 2a (0, 0, /4)
Occ.
B (Å2)
0.199(1)
1.85(7)
0.398(2)
1.81(6)
0.607(9)
1.52(7)
0.804(5)
1.52(7)
1.00
1.46(7)
a
Rp = (∑i | yi − yci |)/∑i yi, Rwp = [(∑i w | yi − yci |2)/Σi w | yi2]1/2, GOF = ∑i w (yi − yci)2/(n − p).
Figure 4. Difference Fourier maps of the Ca10(SiO4)3(SO4)3Cl2−xFx ellestadite tunnel indicating the positions of F and/or Cl ions that were
excluded to emphasize their location. (a) Chlorellestadite (x = 0), (b) Fluor-chlorellestadite (x = 0.8), and (c) Fluorellstadite (x = 2). Contour
interval is 0.5 e/Å3 and the first positive contour (solid line) is at 0.5 e/Å3. Negative difference nuclear density is indicated by the broken contours.
12644
dx.doi.org/10.1021/ic201673r | Inorg. Chem. 2011, 50, 12641−12650

Inorganic Chemistry
Article
Table 2. Bond Lengths and Angles for Ca10(SiO4)3(SO4)3Cl2‑xFx
composition x
0
0.4
0.8
1.2
1.6
2.0
Bond Length (Å)
Ca(1)−O(1) × 3
Ca(1)−O(2) × 3
Ca(1)−O(3) × 3
<Ca(1)-O>
2.438(3)
2.466(3)
2.791(3)
2.565
2.420(4)
2.474(2)
2.808(4)
2.567
2.419(2)
2.475(3)
2.815(2)
2.569
2.411(5)
2.496(5)
2.835(5)
2.580
2.408(2)
2.484(3)
2.841(2)
2.578
2.405(1)
2.491(2)
2.834(4)
2.577
Ca(2)−O(1)
Ca(2)−O(2)
Ca(2)−O(3) × 2
Ca(2)−O(3) × 2
<Ca(2)−O>
2.893(4)
2.283(4)
2.582(3)
2.358(3)
2.509
2.855(1)
2.313(1)
2.569(2)
2.363(1)
2.505
2.786(2)
2.369(1)
2.585(4)
2.348(6)
2.503
2.726(4)
2.400(3)
2.574(2)
2.353(2)
2.497
2.736(3)
2.408(3)
2.550(1)
2.364(2)
2.495
2.687(3)
2.409(4)
2.522(1)
2.375(1)
2.482
Ca(2)−Cl
Ca(2)−F
Ca(2)−Ca(2)
Si/S−O(1)
Si/S−O(2)
2.803(4)
2.727(1)
2.474(1)
4.286(3)
1.5442(5)
1.545(1)
1.545(1)
1.545
2.589(1)
2.425(1)
4.200(6)
1.544(1)
1.545(1)
1.544(1)
1.544
2.454(1)
2.370(1)
4.106(2)
1.545(5)
1.545(1)
1.5438(2)
1.544
2.421(6)
2.346(6)
4.063(5)
1.546(4)
1.5451(7)
1.544(1)
1.545
2.301(4)
3.985(5)
1.5450(1)
1.5452(1)
1.5453(6)
1.545
4.351(1)
1.544(5)
1.545(1)
1.545(2)
1.545
Si/S−O(3)
average
Angle (degree)
O(1)-Si/S−O(2)
O(1)−Si/S−O(3)
O(2)−Si/S−O(3)
O(3)−Si/S−O(3)
112.6(2)
112.3(2)
105.9(2)
107.4(1)
112.2(1)
112.4(1)
106.4(1)
106.5(1)
111.4(2)
112.2(1)
106.4(2)
108.06(6)
111.2(2)
111.4(2)
107.6(2)
107.5(1)
111.1(1)
111.4(1)
108.0(1)
106.67(9)
112.027(2)
111.04(1)
108.016(1)
106.469(2)
Table 3. Comparison of Neutron Data Refinement Results Based on Cation Vacancies at Ca(1)/Ca(2), Ca(1) and Ca(2),
a
respectively
occupancy factor
Ca(2)
content per unit cell
x
vacancy site
Ca(1)
Cl
Ca
Cl
F
anion vacancy
B (Å2)
RB
0
Ca(1)&Ca(2)
Ca(1)
0.954(8)
0.954(2)
1.000(5)
0.41(2)
9.82
9.82
9.83
9.72
9.75
9.71
9.87
9.87
9.87
9.92
9.93
9.92
10.01
10.00
9.99
10
1.64
1.63
1.66
1.04
1.10
1.02
0.94
0.94
0.94
0.63
0.65
0.63
0.40
0.40
0.39
0
0
0
0
0.36
0.37
0.34
0.56
0.50
0.58
0.26
0.26
0.26
0.16
0.14
0.16
0
1.88(6)
1.88(6)
1.92(6)
1.85(7)
1.86(7)
1.85(7)
1.81(6)
1.82(6)
1.82(6)
1.52(7)
1.55(7)
1.53(7)
1.52(7)
1.53(7)
1.53(7)
1.46(7)
2.86
2.86
2.91
2.10
2.07
2.17
2.67
2.62
2.72
2.54
2.56
2.53
2.61
2.63
2.62
2.81
b
1
0.408(5)
0.416(5)
0.26(2)
b
Ca(2)
1
0.972(2)
0.967(7)
0.4
0.8
1.2
1.6
2.0
Ca(1)and Ca(2)
Ca(1)
0.979(9)
0.938(3)
0.40
0.40
0.40
0.80
0.80
0.80
1.21
1.21
1.21
1.60
1.60
1.60
2.0
b
1
0.275(7)
0.255(4)
0.23(2)
b
Ca(2)
1
0.952(7)
0.988(5)
Ca(1)and Ca(2)
Ca(1)
0.985(8)
0.967(3)
b
1
0.234(6)
0.235(5)
0.158(3)
0.163(7)
0.156(7)
0.10(3)
b
Ca(2)
1
0.978(2)
0.977(7)
Ca(1) and Ca(2)
Ca(1)
1.01(1)
b
0.983(3)
1
b
Ca(2)
1
0.987(2)
0.989(7)
Ca(1) and Ca(2)
Ca(1)
1.02(1)
b
1.001(3)
1
0.102(7)
0.098(7)
0
b
Ca(2)
1
0.999(2)
0.01
0
a
b
Rb = ∑i |Iko −Ikc|/∑i Iko. Not refined.
SVDdiagnostic,25 was used to confirm the statistical robustness of the
analysis.
electron microscopy (TEM). Specimens were prepared by ultrasoni-
cally dispersing a small quantity of powder in ethanol then placing a
drop of suspension on a holey carbon-coated copper grid dried in air.
High-resolution transmission electron microscopy (HRTEM) was
carried out at 200 keV with a JEOL 2100F TEM. Samples were
orientated with a double tilting holder and the selected area electron
diffraction (SAD) patterns recorded. Observation times were limited
due to electron amorphization, especially when a more converged
beam was used for imaging.
2.3. Infrared Spectroscopy. FTIR spectroscopy was used to
study the effect of the fluorine substitution on silicate and sulfate
absorption bands of ellestadite, and exclude the possible incorporation
of CO3 and OH−. Infrared spectra were recorded over the range
2‑
400−4000 cm−1 with a Fourier Transform Infrared Spectrometer
(Perkin-Elmer Spectrum 2000) using the KBr pellet technique, where
ellestadite (∼1 mg) was ground with oven-dried KBr (200 mg). This
homogenized mixture was pressed into 10 mm diameter transparent
pellets. Measurements were carried out with a spectral resolution of
4 cm−1 and accumulated for 32 scans.
3. RESULTS AND DISCUSSION
3.1. Metalloid Speciation. The incorporation of fluorine
did not obviously affect the FTIR spectra of chlorellestadite
2.4. Transmission Electron Microscopy. Nanometric chemical
and structural properties of ellestadite were examined by transmission
12645
dx.doi.org/10.1021/ic201673r | Inorg. Chem. 2011, 50, 12641−12650

Inorganic Chemistry
Article
Figure 6. Partial X-ray powder diffraction patterns of ellestadites with
x = 0, 0.8, 1.6. (x in Ca10(SiO4)3(SO4)3Cl2−xFx).
Figure 5. Variation in cell parameters (a) a and c, and (b) unit cell
volume with degree of fluorine substitution x in Ca10(SiO4)3-
(SO4)3Cl2−xFx. Standard deviation of data is smaller than size of
points. The open circles in (b) are taken from the literature11,12
.
Figure 7. Perspective drawing of Ca(2) triangle arrangement on the
(Figure 2). The sulfate ν3 and ν4 vibration modes were
identified by three features (at 1138, 644, 615 cm−1) as were
screw axis (a)
x = 0.4, and (b) x = 1.6. (x in
Ca10(SiO4)3(SO4)3Cl2−xFx) F− lies in the center of Ca(2) triangle
on the mirror planes, while Cl− splits into two half-occupied sites
above or below the mirror plane. The interatomic distances are in Å.
26
the silicate ν3 and ν4 modes (at 938, 561, 508 cm−1). CO3
2‑
incorporation, if present, was below the limit of detection. For
x = 0.4 and 0.8 (x in Ca10(SiO4)3(SO4)3Cl2−xFx), the OH−
librational mode appears very weakly at 660 cm−1.
1
The distances between Cl site and the mirror plane (z = /4) are
indicated by the dotted lines.
3.2. Quantitative Phase Analysis. Powder XRD shows
that ellestadite is always dominant (>95 wt %) (Figure 3).
Additionally, the reactants quartz SiO2 (<0.5 wt %) or
anhydrite CaSO4 (<3.5 wt %) persist for low fluorine sub-
stituted samples.
3.3. Crystal Structure Analysis. The details of the crystal
structure refinements including atomic positional parameters
and temperature displacement parameters are given in Table 1,
with selected atomic distances and bond angles collected in
Table 2. The Rietveld refinement and difference Fourier maps
show that across the fluor-chlorellestadite series, the halides
occupy the 2a (F−) or 4e (Cl−) Wyckoff positions. The location
of chlorine ions in the chlorellestadite endmember (x = 0) show
nuclear scattering at 4e (0, 0, 0.431), while in fluorellestadite
(x = 2), scattering at 2a (0, 0, 1/4) was assigned to F− (Table 1).
Similar bilocation was used for intermediate F/Cl compositions
(Figure 4). The Cl− position shifts from z = 0.431 in the
endmember toward the mirror plane with increasing fluorine
content (z = 0.412, 0.382, 0.342, and 0.336 for x = 0.4, 0.8, 1.2,
and 1.6, respectively) (Figure 7a). It is likely that the weak OH−
signature observed by FTIR arises from a nonellestadite phase,
since a residual feature in the Fourier map cannot be assigned to
hydroxyl, as suggested by FTIR for x = 0.4 and 0.8, and if
present, OH− is in low abundance. Examination of the difference
Fourier maps (Supporting Information Figure A1) exhibited low
electron densities in the range of −0.85 e/Å3 to 0.85 e/Å3, which
discounts the possibility of multiple Cl− positions15 or splitting of
Ca(2) site9 as suggested by earlier studies.
12646
dx.doi.org/10.1021/ic201673r | Inorg. Chem. 2011, 50, 12641−12650

Inorganic Chemistry
Article
chlorellestadite endmember, compared with stoichiometric
chlorellestadite (a = 9.6773(3) Å, V = 556.25(4) Å3) reported
previously.9 To examine this question, the occupancies of
calcium and chlorine were refined with the constraint Ca2+
+
2Cl− →□Ca + 2□Cl applied in three ways: (i) the occupancies
of Ca(1) and Ca(2) were jointly refined together with Cl, (ii)
the Ca(1) site was constrained with Cl site, with the Ca(2) site
fully occupied, and (iii) the Ca(2) site was released with Ca(1)
fixed to full occupancy. In this way, any preference to create Ca-
site vacancies within the framework or tunnel could be
distinguished (Table 3). While the differences in refinement
residuals are small, it does appear that for Cl-rich compositions,
Ca(1) vacancies may be present, while for F-rich ellestadites the
reverse is true. However, further analysis would be required.
It is noted that modeling with the Ca(1) and Ca(2) sites fully
occupied lead to substantially inferior residuals, and in addition,
cannot charge balance halogen substoichiometry. On the basis
of these assessments, the formulas for these compounds are
most generally written as Ca10−y(SiO4)3(SO4)3Cl2−x−2yFx, rather
than the nominal Ca10(SiO4)3(SO4)3Cl2−xFx.
Singular value decomposition (SVD) as adapted by Mercier
et al.25,28 for the evaluation apatite structures, found the
numerical stability of the Rietveld refinements was satisfactory:
in particular, the tunnel site occupancy was reliable. Typical
outputs (Supporting Information Table A1) from SVDdiag-
nostic of the preconditioned normal matrices for the models in
which the Ca(1) and Ca(2) sites were refined simultaneously
(see also Table 3), give condition numbers of around 102−103
indicative of good conditioning of the least-squares matrices,
that confirms the occupancy factors refinements are reasonable
and not over interpreted. Similar results were obtained if Ca(1)
or Ca(2) were refined alone.
As the content of fluorine increases, a contracts while c
dilates, but overall the unit cell volume is reduced (Table 1,
Figure 5), because fluorine is smaller than chlorine. In addition,
the splitting of the intense doublet reflections (121)/(211) and
(112)/(030) in the chlorellestadite endmember pattern into a
resolved triplet (121)/(211), (112), and (030) provides further
confirmation of the mixed anion structure21 (Figure 6). The
loss of CaCl2 from the structure also contributes to unit cell
shrinkage. According to Young and Elliott,29 the small increase
in c may be related to the different atomic positions of halogen
Figure 8. Dependence of (a) zCl and (b) twist angle on degree of
fluorine substitution x in Ca10(SiO4)3(SO4)3Cl2−xFx.
The Cl content was less than nominal for samples with x ≤
1.2. Halogen loss might occur during the solid state processing,
as chlorine is susceptible to volatilisation, with possible charge
compensation mechanisms being
1
ions. As fluorine is centered in the Ca(2) triangle (z = /4), it
will not affect the c axes. However, chlorine is displaced >1 Å
along c, weakly bonding to the neighboring calcium Ca(2)
triangles that leads to slight contraction of the c axis.
(1)
(2)
(3)
Differences in the anion content of the tunnel negligibly
disturbs the (Si/S)O4 tetrahedra, which are essentially rigid
bodies, and the Ca(1)O9 polyhedra (Table 2). There are only
minor changes in the (Si/S)−O bond lengths and no evidence
for a Si/S ordering which would lower the symmetry to mono-
clinic. The refined mean bond length Si/S−O = 1.545 Å, close
to the average value of 1.55 Å calculated from Si−O = 1.61 Å in
silicates30 and S−O = 1.49 Å in sulphates,31 is also consistent
with a statistical distribution of SiO4 and SO4 tetrahedra.
The major structural response to fluorine substitution occurs
in the Ca(2)O5(F/Cl)(O) polyhedron (a seventh weak bond
to O(1) exists in fluorellestadite) (Table 2). As noted the
Mechanism 1 is not supported by FTIR or structure
refinement. In the case of mechanism 2, attempts to insert
extra oxygen at additional (0, 0, z) positions on the hexagonal
axis were unsuccessful, as the temperature displacement
parameter of oxygen atom became negative. Prener27 studied
nonstoichiometry in calcium chlorapatite, and stated that
synthetic chlorapatite (often P21/b) became hexagonal P63/m
when grown from the melt due to Cl deficiency created by
loss of CaCl2. In our study, the formation of Ca and Cl
structural vacancies (mechanism 3) is possible as CaCl2
partially evaporates during sintering at 950 °C (mp(CaCl2) =
782 °C). This would be consistent with the smaller unit cell
parameter a (9.6702(3) Å) and volume (555.09(4) Å3) of the
1
3
Ca(2) atoms located on the mirror planes at z = /4 and /4
form triangles that are centered around the [001] 3 axis
̅
containing the F/Cl (Figure 7). The Cl− ion located in 4e shifts
from z = 0.4315(3) in the chlorellstadite endmember toward
the Ca(2) triangles as z decreases to 0.336(3) (Figure 8a).
12647
dx.doi.org/10.1021/ic201673r | Inorg. Chem. 2011, 50, 12641−12650

Inorganic Chemistry
Article
Figure 9. Correlations observed between polyhedral distortion parameters: (a) δCa1 versus ϕO3−Ca2−O3, (b) <τO−B−O> versus ϕO3−Ca2−O3, (c)
ΨCa1−O1Ca1 z=0 versus αCa1, and (d) αCa2 versus ϕO3−Ca2−O3. These figures should be compared with Figure.8 in Mercier et al.33 Circles: published data;
large filled squares: this study. (δCa1:counter-rotation angle of CaO6 polyhedra; ϕO3−Ca2−O3: O3−Ca2−O3 bond angle; <τO−B−O>: O−B−O bond-
bending angle; ΨCa1−O1Ca1 z=0: angle that an Ca1z=0−O1 bond makes with respect to c; αCa1: orientation of CaO6 polyhedra with respect to a; αCa2
orientation of Ca2 triangles with respect to a.).
:
The interatomic distances between Ca(2) and Cl decreased
from 2.803(4) Å in the chlorellestadite (x = 0) to 2.421(6) Å
when x = 1.6. The Ca(2)-F bond length also follows this trend,
decreasing from 2.474(5) Å in x = 0.4 to 2.3018(1) Å in
fluorellestadite (x = 2.0) (Table 2). As the F− ion lies in the
center of the Ca(2) triangle, the decrease of the Ca(2)−F dis-
tances with reduction of chlorine content leads to a contraction
in the unit cell a-parameter (Table 1, Figure 5b).
The metaprism twist angle (φ), defined as the (001) pro-
jected angle of the O(1)−Ca(1)−O(2), was used to validate
the Rietveld refined ellestadite structures (Table 2). The twist
angle (φ) is related to the relative sizes of the A, B, and X ions32
and also sensitively detects disequilibrium and nonstoichiom-
etry. In passing from fluorellestadite (x = 2) to chlorellestadite
(x = 0), the twist angle (φ) decreases from 24.1° to 19.9°, with
increasing chlorine content leading to a tunnel dilation and to
increased Ca(2)−F/Cl distances. The near linear change in
twist angle (φ) is expected, and small departures from linearity
may be indicative of Cl deficient nature of some compositions
(Figure 8b).
As further validation of the refined structures, the crystal-
chemical parameters proposed by Mercier et al.33 (Supporting
Information Table A2) were extracted for the refined ellestadite
structures; the geometrical derivation of these parameters can
be found in this earlier work. The principle correlations among
those crystal-chemical parameters, including data from this
study and published work are shown in Figure 9. As the
fluorine content increases, the (Si/S)O4 tetrahedra (τO3−Ca2−O3
decreases) and CaO6 polyhedra (δCa1 decreases) are distorted
to accommodate the increase of the O3−Ca2−O3 bond angle
(ϕO3−Ca2−O3). These correlations conform to the trends
observed by Mercier et al.33 (Figure 8 in ref 33.), and of
particular interest, fill a gap noted in their work for apatites with
131° < ϕO3−Ca2−O < 135°. Furthermore, the crystal chemical
3
parameters are clustered for F-rich compositions consistent
with loss of Cl, and confirm there is little change in halide
content for these materials.
12648
dx.doi.org/10.1021/ic201673r | Inorg. Chem. 2011, 50, 12641−12650

Inorganic Chemistry
Article
4. CONCLUSION
The synthesis of fluorine-substituted chlorellestadites has been
achieved by the solid state method, and the crystal structures
refined via powder X-ray and neutron diffraction using P63/m
symmetry. Ordering of the Si and S atoms within the tetra-
hedra, or halides between the tunnels can be excluded, as this
would lead to P21/m or P21/b structures respectively, with a
possible doubling of the unit cell in the a or b directions. In
natural ellestadites, described in P21/m and P21, equilibration
over very long geological periods may favor lower symmetry.
For the synthetic ellestadites studied here, increasing fluorine
content, leads to a progressive shift of Cl− toward the center of
the Ca(2) triangle; the formation of a shorter, stronger bond
between Ca(2) and Cl may enhance the chemical stability of
ellestadite. Minor loss of CaCl2 during sintering induces Ca-
vacancies within ellestadite that was confirmed by powder
neutron diffraction, and verified by the spread of Ca(1)O6
metaprism twist angles as a function of composition, although
the overall trend is linear. For charge balance, halide vacancies in
the anion tunnel were compensated by calcium vacancies, and
may be required to provide structural accommodation for larger
Cl ions, or reduce structural strain due to the mismatch of the
framework and tunnel dimensions. The reliability of Rietveld
refinements were evaluated by Singular Value Decomposition
and correlations of various crystal−chemical parameters. Future
reports will focus on compositionally complex ellestadites and
chemical tailoring that improves the stability and effectiveness
of waste form ceramics.
Figure 10. (a) Wide field [001] HRTEM image of Ca10(SiO4)3-
(SO4)3Cl0.4F1.6, (experimental, FFT and simulated SAED reflections
are provided.) (b) The as-collected images (left) and processed image
(right) constrained by p6 symmetry are shown.
ASSOCIATED CONTENT
■
S
* Supporting Information
Examples of the output of SVDdiagnostic of Rietveld refine-
ments of ellestadites, crystal-chemical parameters extracted
from the crystallographic descriptions, neutron crystallographic
files in CIF format, experimental and calculated powder X-ray
and neutron diffraction patterns, difference Fourier map of
nuclear density for Ca10(SiO4)3(SO4)3Cl0.8F1.2, and bond-
valence sums (BVS) at site Ca1 and Ca2 plotted as a function
of the degree of fluorine substitution x, x in Ca10(SiO4)3(SO4)3-
Cl2−xFx. This material is available free of charge via the Internet
at http://pubs.acs.org.
None of the ellestadites show measurable deviation from the
hexagonal metric. This behavior is distinct from the
Ca10(VxP1−xO4)6F2 apatites,34 which at the largest vanadium
contents become triclinic. For these materials, the approach of
αCa2 to 120° was seen to precede a reduction in symmetry.
However, in this study it was found that Ca10(SiO4)3-
(SO4)3Cl2 with αCa2 = 120.23° could be convincingly modeled
in P63/m. It has already been noted that mineral ellestadites
may be monoclinic, and long annealing of synthetic Ca10-
(SiO4)3(SO4)3Cl2 may allow SiO4/SO4 ordering with stabiliza-
tion as P21/m or P21 structures.
3.4. Local Structure. A high-resolution TEM image of
Ca10(SiO4)3(SO4)3Cl0.4F1.6 shows a well developed crystal
without obvious imperfection, in the sense that no extended
defects are observed (Figure 10a). A comparison of the inten-
sity distribution of experimental and simulated SAED
reflections was consistent with P63/m symmetry. The SAED
pattern did not show split reflections, precluding the possibility
of coexisting mixed-phase ellestadite,19 or localized order with
respect to anionic sequences within the anion tunnels that
would lower the symmetry to monoclinic. Fourier reconstruc-
tion of the images yield potential maps (Figure 10b) most
plausibly constrained to the plane group p6 which is the [001]
symmetry projection for P63/m. In the HR image, the darker
feature represents the positions of Ca(1)O9 polyhedral columns
which share (001) pinacoid faces to form chains parallel to
c, while the bright contrast corresponds to the anion columns at
the tunnel centers.
AUTHOR INFORMATION
Corresponding Author
*E-mail: fang0027@e.ntu.edu.sg.
■
ACKNOWLEDGMENTS
Yanan Fang thanks Nanyang Technological University, Singapore,
for the NTU Research Scholarship supporting this study.
■
REFERENCES
■
(1) Dong, Z.; White, T. J.; Wei, B.; Laursen, K. J. Am. Ceram. Soc.
2002, 85, 2515−2522.
(2) Nishikawa, H.; Omamiuda, K. J. Mol. Catal. 2002, 179, 193−200.
(3) Driessens, F. C. M.; Verbeeck, R. M. H. J. Cryst. Growth 1981,
53 (1), 55−62.
(4) Hertz, A.; Bruce, I. J. Nanomedicine 2007, 2 (6), 899−918.
(5) Neubauer, J.; Erlangen, H. P. N. Jb. Miner. Abh. 1995, 168 (3),
237−258.
(6) McConnell, D. Am. Mineral. 1937, 22, 977−986.
(7) Sudarsanan, K. Acta Cryst. B: Struct. Sci. 1980, 36, 1636−1639.
(8) Organova, N. I.; Rastsvetaeva, R. K.; Kuz’mina, O. V.; Arapova,
G. A.; Litsarev, M. A.; Fin’ko, V. I. Kristallografiya 1994, 39, 278−282.
12649
dx.doi.org/10.1021/ic201673r | Inorg. Chem. 2011, 50, 12641−12650

Inorganic Chemistry
Article
(9) Saint-Jean, S. J.; Sweden, S. H. Solid State Sci. 2005, 7, 97−102.
(10) Dean, H. T. Public Health Rep. Wash. 1938, 53, 1443−1452.
(11) Lexa, D. Metall. Mater. Trans. 1999, 30A, 147−153.
́
(12) Pajares, I.; de la Torre, A. G.; Martínez-Ramírez, S.; Puertas, F.
Powder Diffr. 2002, 17, 281−286.
(13) Shannon, R. D. Acta Crystallogr., Sect. A: Cryst. Phys., Diffr.,
Theor. Gen. Crystallogr. 1976, 32, 751−767.
(14) Onac, B. P.; Effenberger, H.; Ettinger, K.; Panzaru, S. C. Am.
Mineral. 2006, 91, 1927−1931.
(15) Hughes, J. M.; Rakovan, J. Rev. Mineral. Geochem. 2002, 48 (1),
1−12.
(16) Mccubbin, F. M.; Mason, H. E.; Park, H.; Phillips, B. L.; Parise,
J. B.; Nekvasil, H.; Lindsley, D. H. Am. Mineral. 2008, 93, 210−216.
(17) Mackie, P. E.; Young, R. A. J. Solid State Chem. 1974, 11, 319−
329.
(18) Hughes, J. M.; Cameron, M.; Crowley, K. D. Am. Mineral. 1990,
75, 295−304.
(19) Ferraris, C.; T. J. White, T. J.; Plev
́
ert, J.; Wegner, R. Phys. Chem.
Mineral 2005, 32, 485−492.
(20) Bruker. TOPAS Version 3, Bruker AXS Inc., Madison, WI, USA,
2005.
(21) Saint-Jean, S. J.; JØns, E.; Lundgaard, N.; Hansen, S. Cem. Concr.
Res. 2005, 35, 431−437.
(22) Sears, V. F. In International Tables for Crystallography; Wilson, A.
J. C., Ed.; Kluwer Academic Publishers: Dordrecht, 1993; Vol. C,
p 383.
(23) Petricek, V.; Dusek, M.; Palatinus, L. Jana2006. The Crystallo-
graphic Computing System; Institute of Physics: Praha, Czech Republic.
2006.
(24) White, T. J.; Dong, Z. L. Acta Cryst. B: Struct. Sci. 2003, 59, 1−16.
(25) Mercier, P. H. J.; Le Page, Y.; Whitfield, P. S.; Mitchell, L. D.
J. Appl. Crystallogr. 2006, 39, 458−465.
(26) Griffith, W. P. Nature 1969, 224, 264−266.
(27) Prener, J. S. J. Solid State Chem. 1971, 3, 49−55.
(28) Mercier, P. H. J.; Le Page, Y.; Whitfield, P. S.; Mitchell, L. D.
J. Appl. Crystallogr. 2006, 39, 369−375.
(29) Young, R. A.; Elliott, J. C. Archs. Oral. Biol. 1966, 11, 699−707.
(30) Smith, J. V.; Baily, S. W. Acta Crystallogr. 1963, 16, 801−811.
(31) Wuench, B. J. Sulfur, 16-A Crystal chemistry. InHandbook of
Geochemistry; Wedepohl, K. H., Ed.; Springer-Verlag: Berlin, 1970;
Vol. II/2, 16A, pp 1−19.
(32) White, T. J.; Ferraris, C.; Kim, J.; Madhavi, S. Rev. Mineral.
Geochem. 2005, 57, 307−401.
(33) Mercier, P. H. J.; Le Page, Y.; Whitfield, P. S.; Mitchell, L. D.;
Davidson, I. J.; White, T. J. Acta Cryst. B: Struct. Sci. 2005, 61, 635−655.
(34) Mercier, P. H. J.; Dong, Z.; Baikie, T.; le Page, Y.; White, T. J.;
Whitfield, P. S.; Mitchel, L. D. Acta Cryst. B: Struct. Sci. 2007, 63, 37−48.
12650
dx.doi.org/10.1021/ic201673r | Inorg. Chem. 2011, 50, 12641−12650