I. INTRODUCTION
Three forms of the stoichiometric compound KZnPO4 have been well characterized. These forms are generally labeled α, β, and γ. The α form is stable up to a temperature of 691 °C (Wallez et al., Reference Wallez, Lucas, Souron and Quarton1999) and has a hexagonal crystal structure (space group P63) (Andratschke et al., Reference Andratschke, Range, Haase and Klement1992). The β and γ forms are high-temperature phases that have orthorhombic crystal structures (space groups Pna21 and Pnma, respectively) (Wallez et al., Reference Wallez, Lucas, Souron and Quarton1999). A second room temperature form of KZnPO4 is known to exist but has not been well characterized. This second form was first reported by Barbou des Courières and Simonot-Grange (Reference Barbou des Courières and Simonot-Grange1979) as a transformation product of KZn2H(PO4)2(H2O)2.5 in slightly alkaline aqueous KCl solutions. Barbou des Courières and Simonot-Grange (Reference Barbou des Courières and Simonot-Grange1979) used the notation KZnPO4II for this second room temperature variant while using the notation KZnPO4I for α-KZnPO4. Barbou des Courières and Simonot-Grange (Reference Barbou des Courières and Simonot-Grange1979) noted that the X-ray diffraction (XRD) powder pattern of KZnPO4II is similar to the XRD powder pattern of NH4ZnPO4I, the monoclinic form of NH4ZnPO4 (Averbuch-Pouchot and Durif, Reference Averbuch-Pouchot and Durif1968). Barbou des Courières and Simonot-Grange (Reference Barbou des Courières and Simonot-Grange1979) showed that KZnPO4II is one end member of the isostructural series Kx(NH4)1−xZnPO4 with NH4ZnPO4I being the other end member.
The aim of our study is to characterize KZnPO4II using XRD and Rietveld refinement (Rietveld, Reference Rietveld1967). Rather than use the notation KZnPO4II, the notation δ-KZnPO4 will be used in this article so as to be more consistent with the more widely used α, β, and γ notation for KZnPO4 phases.
II. EXPERIMENTAL
A. Synthesis of δ-KZnPO4
According to studies by Frazier et al. (Reference Frazier, Smith and Lehr1966) and Barbou des Courières and Simonot-Grange (Reference Barbou des Courières and Simonot-Grange1979), the form of potassium zinc phosphate that precipitates from aqueous solutions of potassium phosphates and zinc salts is dependent on the pH of the solution. In solutions with a pH of around 4 or less, KZn2H(PO4)2(H2O)2.5 forms. If left in solution, this phase slowly dehydrates in solution to the anhydrous form. At pH values between 5 and 7, α-KZnPO4 forms. At pH values of 7 and above, δ-KZnPO4 forms. Consequently, methods for synthesizing δ-KZnPO4 begin with an aqueous solution of the weak base K2HPO4. For the present study, two methods of synthesizing δ-KZnPO4 were used.
The first synthesis method is the method of Salutsky and Steiger (Reference Salutsky and Steiger1964). For this method, 92 g of K2HPO4(H2O)3 was dissolved in 200 ml of high-purity water held at a temperature of 70 °C. The beaker of K2HPO4 solution was transferred to a water bath set to a temperature of 65 °C. A second solution consisting of 3.5 g of ZnCl2 in 100 ml of high-purity water was slowly added and stirred into the K2HPO4 solution. The beaker of combined solution was left in the water bath (65 °C) for 3 days. The precipitated material in the solution was recovered using centrifugation and then washed three times using high-purity water. The washed precipitate was dried overnight in air at room temperature (approximately 20 °C). In order for this method of synthesis to produce a pure δ-KZnPO4 product, the amount of K2HPO4 used must be sufficiently high to maintain the solution at a pH of 7 or higher when the ZnCl2 solution is added and the potassium zinc phosphate phase precipitates. If insufficient K2HPO4 is used or too much ZnCl2 is used, the solution pH will drop below 7 and the precipitated material will be α-KZnPO4 or a mixture of α-KZnPO4 and δ-KZnPO4.
The second synthesis method follows the method of Barbou des Courières and Simonot-Grange (Reference Barbou des Courières and Simonot-Grange1979) and uses KOH to control the pH of the solution and allows smaller amounts of K2HPO4 and larger amounts of ZnCl2 to be used relative to the method of Salutsky and Steiger (Reference Salutsky and Steiger1964). For the second method of synthesis, 13.0 g of K2HPO4(H2O)3 and 10.0 g of KCl were dissolved in 500 ml of high-purity water. A second solution consisting of 5.0 g of ZnCl2 in high-purity water was slowly added to the K2HPO4/KCl solution while stirring. The pH of the mixture was immediately adjusted to 8.5 by the drop-wise addition of 1M KOH solution. Over the following 2 h, the pH of the mixture was monitored and the solution pH maintained close to 8.5 by the addition of drops of 1M KOH solution as required. The mixture was left stirring for 3 days at room temperature after which time the precipitate was recovered as per the first synthesis method. It was found that the concentration of the K2HPO4/KCl solution was not critical and less than 500 ml of water could be used in preparing this solution. However, if too little water was used (approximately 200 ml), the addition of the ZnCl2 solution to the K2HPO4/KCl solution caused the solution to gel.
B. XRD data collection and structure refinement
XRD powder patterns were recorded with a PANalytical X'Pert Pro multi-purpose diffractometer using Fe filtered CoKα radiation (40 kV, 55 mA), automatic divergence slit, fixed illuminated area, 2° anti-scatter slit, and a fast X'Celerator Si strip detector. For structure refinement, an XRD powder pattern was recorded from 4° to 80° 2θ with a recording interval of 0.017° 2θ and an overall counting time of a little over 9 h. Structure refinement was carried out on the CoKα radiation data using the TOPAS version 5 software suite from Bruker-AXS. A Rigaku SmartLab diffractometer was used to collect powder patterns using CuKα radiation (40 kV, 200 mA), fixed 2° divergence slit, 0.01° step size, and a D/tex Ultra 250 Si strip detector.
Elemental concentrations of samples were determined using a PANalytical Axios, wavelength-dispersive, X-ray fluorescence (XRF) spectrometer in conjunction with the PANalytical SuperQ software package. For XRF analysis AR-grade ZnO, KH2PO4 and a mixture of ZnO and KH2PO4 were used to check the calibration of the XRF software for the elements K, Zn, and P.
III. RESULTS AND DISCUSSION
Multiple batches of potassium zinc phosphate were made using the two synthesis methods. XRD showed that the precipitates produced were the same as the material called KZnPO4II by Barbou des Courières and Simonot-Grange (Reference Barbou des Courières and Simonot-Grange1979). Furthermore, XRF analysis showed that the composition of the precipitates to be consistent with the formula KZnPO4. Both synthesis methods were found to be equally effective in producing pure δ-KZnPO4 although in one of the batches made using the synthesis method based on the method of Barbou des Courières and Simonot-Grange (Reference Barbou des Courières and Simonot-Grange1979), a significant amount of KZnPO4(H2O)0.8 (Broach et al., Reference Broach, Bedard and Song1999) was formed along with δ-KZnPO4. Both synthesis methods have a recovery rate of close to 100% based on the amount of ZnCl2 used.
Based on the observed peak widths from the X-ray powder diffraction (XRPD) data of the synthesized materials, the synthesis method based on the method of Barbou des Courières and Simonot-Grange (Reference Barbou des Courières and Simonot-Grange1979), in general terms, gave δ-KZnPO4 material with slightly higher crystallinity than the method of Salutsky and Steiger (Reference Salutsky and Steiger1964) and so δ-KZnPO4 synthesized using the former method was used for structure refinement. An attempt was made to improve the crystallinity of the δ-KZnPO4 by heating the synthesized material at a range of temperatures. Heating did not significantly improve the crystallinity of the synthesized material, but it was found that after heating for 12 h at 350 °C, the δ-KZnPO4 had partially converted to α-KZnPO4. Heating for 12 h at 400 °C completely converted the δ-KZnPO4 to α-KZnPO4.
Barbou des Courières and Simonot-Grange (Reference Barbou des Courières and Simonot-Grange1979) demonstrated that δ-KZnPO4 is isostructural with NH4ZnPO4I. The NH4ZnPO4I is described as having a tridymite derivative structure (Averbuch-Pouchot and Durif, Reference Averbuch-Pouchot and Durif1968) or an ABW zeolite topology (Bu et al., Reference Bu, Feng, Gier and Stucky1997). The structure of NH4ZnPO4I consists of corner-sharing ZnO4 and PO4 tetrahedra (Averbuch-Pouchot and Durif, Reference Averbuch-Pouchot and Durif1968; Bu et al., Reference Bu, Feng, Gier and Stucky1997). The NH4 entities lie in a network of channels that lie parallel to the a and b crystallographic directions. As a starting point for the structure refinement of δ-KZnPO4 from the observed XRPD data (Table I) the atomic coordinates for NH4ZnPO4I determined by Bu et al. (Reference Bu, Feng, Gier and Stucky1997) with K replacing NH4 were used for the δ-KZnPO4 structure and the unit cell of δ-KZnPO4 refined (Table II). The unit cell dimensions of δ-KZnPO4 are all smaller than the unit cell dimensions of NH4ZnPO4I. The unit cell volume of δ-KZnPO4 is approximately 6% smaller than the unit cell volume of NH4ZnPO4I.
X-ray powder diffraction data of δ-KZnPO4.
The data are for a fixed divergence slit of 2° and CuKα radiation. RIR = 0.53.
Unit cell data for δ-KZnPO4 and NH4ZnPO4I.
NH4ZnPO4I parameters from Bu et al. (Reference Bu, Feng, Gier and Stucky1997).
The calculated XRPD pattern generated by using the atomic coordinates of NH4ZnPO4I with K substituted for NH4 gave quite a good fit to the measured XRPD pattern of δ-KZnPO4. However, the smaller unit cell size of δ-KZnPO4 relative to NH4ZnPO4I means that the bond lengths of the ZnO4 and PO4 tetrahedra become unrealistically small. Rietveld refinement of the δ-KZnPO4 structure while allowing the atomic coordinates of all, but the y coordinate of one of the atoms in the structure to vary gave an extremely good fit to the measured XRD pattern of δ-KZnPO4. The requirement to keep the y coordinate of one atom fixed is dictated by the P1211 space group symmetry that does not have an inversion center in the y-direction. Although the fit from this refinement was very good, the atoms in the refined structure had moved to positions where the ZnO4 and PO4 tetrahedra were very distorted and had unrealistic bond lengths.
In order to constrain the rearrangement of the ZnO4 and PO4 tetrahedra during Rietveld refinement, bond distances, and angles of the ZnO4 and PO4 tetrahedra that were determined from the NH4ZnPO4I structure were used as constrained parameters in the software package DLS-76 (Baerlocher et al., Reference Baerlocher, Hepp and Meier1978). In this way, the ZnO4 and PO4 tetrahedra are free to move about their shared corners, but the geometry of the tetrahedra is maintained. The atomic coordinates generated for Zn, P, and O atoms by DLS-76 are, therefore, a set of coordinates that satisfy the unit cell geometry of δ-KZnPO4 while maintaining realistic Zn–O and P–O bond lengths and angles.
The atomic coordinates for the Zn, P, and O atoms generated by DLS-76 along with the thermal parameters for the corresponding atoms in the NH4ZnPO4I structure of Bu et al. (Reference Bu, Feng, Gier and Stucky1997) were used in Rietveld refinements of the δ-KZnPO4 structure. The atomic coordinates of the K atoms in the δ-KZnPO4 structure were refined by allowing their coordinates to change in an unconstrained manner while holding the atomic coordinates and thermal parameters of the Zn, P, and O atoms fixed. The final step in the refinement was to refine the thermal parameters of the K atoms while holding all other structural parameters fixed.
The structural parameters for the refined structure of δ-KZnPO4 are given in Tables II and III, and the XRD pattern and the difference between the measured XRD pattern and the calculated XRD pattern are shown in Figure 1. The R wp and R exp for the refined structure were 1.91% and 0.48%, respectively (GOF = 3.97). Refining the thermal parameters for the K atoms had virtually no effect on the R-factor of the refinement and so the values for these parameters given in Table III must be considered to have a reasonable degree of uncertainty.
Measured XRD pattern of δ-KZnPO4 (upper pattern) and the difference between the measured and calculated XRD patterns (lower pattern). (CoKα radiation.)
Structural parameters for δ-KZnPO4.
Thermal parameters for Zn, P, and O atoms from Bu et al. (Reference Bu, Feng, Gier and Stucky1997). Fractional coordinates for Zn, P, and O atoms refined using the DLS-76 software (Baerlocher et al., Reference Baerlocher, Hepp and Meier1978). Average uncertainty values in the x, y and z coordinates are 0.0009, 0.0024 and 0.0009 respectively.
The examination of the refined structures of δ-KZnPO4 (Figure 2) and NH4ZnPO4I shows that the difference between these structures is rotations of the Zn–O and P–O tetrahedra around their shared corners. These rotations allow the constituent atoms of the δ-KZnPO4 structure to be accommodated in the smaller unit cell of δ-KZnPO4 relative to the unit cell of NH4ZnPO4I. The K atoms of δ-KZnPO4 fill the cavities between the Zn–O and P–O tetrahedra in the same way that NH4 entities fill the cavities in the NH4ZnPO4I structure.
Structure of δ-KZnPO4 projected along the b-axis and showing Zn–O and P–O tetrahedra. Graphic generated using VESTA 3 (Momma and Izumi, Reference Momma and Izumi2011).
Although the methodology used in the refinement of the atomic positions of δ-KZnPO4 should not be regarded as giving a complete characterization of the δ-KZnPO4 structure, it has been found that the derived crystallographic parameters are adequate for quantitative XRD analyses of mixtures containing δ-KZnPO4.
IV. DEPOSITED DATA
Files containing the raw diffraction data for δ-KZnPO4 using CoKα radiation and CuKα radiation along with a file containing the refined structural parameters for δ-KZnPO4 were deposited with the ICDD. The data can be requested at info@icdd.com.
ACKNOWLEDGEMENTS
The authors thank Ugesh Chand of Agrichem for supplying mixtures of the various KZnPO4 phases.
