Rietveld refinements and spectroscopic structural studies of a Na-free carbonate apatite made by hydrolysis of monetite

Seven nominally identical samples of Na-free carbonate apatite (CO3Ap) were prepared by reaction of CaHPO4 with ammonium carbonate solution at 70 °C over 3 days. They were studied by chemical analysis, Rietveld analysis of powder X-ray diffraction (XRD) data, Ca/P ratio determinations (quantitative phase analysis of CaO, Ca(OH)2 and hydroxyapatite formed after heating to 900 °C from Rietveld analysis of XRD data), He pycknometry, 1H, 13C and 31P MAS NMR spectrometry and Fourier transform infrared and Raman spectroscopy. Spectroscopy showed the apatite products were B-type CO3Aps (CO32− replacing PO43−) and XRD that one sample contained 1.6 wt% calcite with a trace in another. Mean results of the six essentially calcite-free samples were: a=9.405(5) Å, c=6.896(2) Å; 11.2 wt% CO3; unit cell contents, Ca8.241(PO4)4.344(CO3)1.656(OH)0.139·2.29H2O; mole Ca/P ratio from chemical analyses, 1.897(22) and from powder XRD phase analysis of samples decarbonated at 900 °C, 1.892(25). Density determinations indicated that the 2.29 mol of H2O were in the unit cell. Rietveld refinements were undertaken without and with explicit modelling of the CO32− ion. The latter used constraints to maintain the CO32− ion in its known geometry and the total of PO43− and CO32− ions per unit cell at six. Without the CO32− ion in the model, PO4 volume, P–O bond lengths and P occupancy were apparently reduced, consistent with CO32− replacing PO43− ions. With the CO32− ion modelled, the reductions were less and the CO32− ion occupied the “sloping” face of the replaced PO43− ion in two-fold disorder about the mirror plane. The angle between the normal to the plane of the ion and the c-axis was 34°, close to 35.3°, the equivalent angle for the PO43− ion. When modelled, the CO32− ion occupancy was 1.81 ions per unit cell, in reasonable agreement with unit cell contents determined chemically (1.66). The OH− ion occupancy was elevated (2.33 ions per unit cell versus 0.14 inferred from the charge balance), which we ascribe to H2O molecules occupying sites in c-axis channels. The Ca/P ratio from occupancies (2.31) was also elevated over that determined chemically (1.90). We attribute this to loss of Ca from Ca sites increasing the apparent anisotropic displacement parameters of remaining Ca atoms, leading to an apparently increased occupancy.