Crystal structure of anhydrous tripotassium citrate from laboratory X-ray powder diffraction data and DFT comparison

The crystal structure of anhydrous tripotassium citrate has been solved and refined using laboratory X-ray powder diffraction data, and optimized using density functional techniques.


Chemical context
In the course of a systematic study of the crystal structures of group 1 (alkali metal) citrate salts to understand the anion's conformational flexibility, ionization, coordination tendencies, and hydrogen bonding, we have determined several new crystal structures. Most of the new structures were solved using powder X-ray diffraction data (laboratory and/or synchrotron), but single crystals were used where available. The general trends and conclusions about the 16 new compounds and 12 previously characterized structures are being reported separately (Rammohan & Kaduk, 2016a). Five of the new structures, viz. NaKHC 6 H 5 O 7 , NaK 2 C 6 H 5 O 7 , Na 3 C 6 H 5 O 7 , NaH 2 C 6 H 5 O 7 , and Na 2 HC 6 H 5 O 7 , have been published recently (Rammohan & Kaduk, 2016b,c,d,e;Rammohan et al., 2016), and two additional structures, viz. KH 2 C 6 H 5 O 7 and KH 2 C 6 H 5 O 7 (H 2 O) 2 , have been communicated to the Cambridge Structural Database (Kaduk & Stern, 2016a,b). ISSN 2056-9890

Structural commentary
The asymmetric unit of the title compound is shown in Fig. 1. The r.m.s. deviation of the non-hydrogen atoms between the Rietveld-refined and the DFT-optimized structures is 0.117 Å (Fig. 2). The maximum deviation is 0.260 Å , at O14. The good agreement between the two structures is strong evidence that the experimental structure is correct (van de Streek & Neumann, 2014). This discussion uses the DFT-optimized structure. Most of the bond lengths, bond angles, and torsion angles fall within the normal ranges indicated by a Mercury Mogul Geometry Check (Macrae et al., 2008). Only the C4-C5 bond length [refined = 1.511 (5), optimized = 1.536, Mogul average = 1.498 (12) Å , Z-score = 3.1], and the C3-C2-C1 [refined = 115 (2), optimized = 115.0, Mogul average = 103 (2) ] and O17-C3-C2 angles [refined = 107 (2), optimized = 109.6, Mogul average = 106 (2) ] are flagged as unusual. The citrate anion occurs in the trans,trans-conformation, which is one of the two low-energy conformations of an isolated citrate. The central carboxylate group and the hydroxy group occur in the normal planar arrangement. Both terminal carboxylate groups O11/O12 and O13/O14 chelate to a single potassium cation (K20 for each). The terminal carboxylate oxygen atom O12 and the hydroxy O17 atom chelate to K21, and the terminal carboxylate oxygen atoms O13 and O17 chelate to K19. The terminal/central pairs O11/ O16, O14/O16, O11/O15, and O14/O15 chelate to K21, K19, K19, and K21, respectively. The three potassium cations K19, K20, and K21 are 6-, 8-, and 6-coordinate, respectively (all irregular, using a K-O cut-off distance of 3.24 Å ). Their bond-valence sums are 1.12, 1.03, and 1.12 valence units. The metal-oxygen bonding is ionic, based on the cation charges and the Mulliken overlap populations.
Although the lattice parameters of anhydrous tripotassium citrate are in general similar to those of the monohydrate (Carrell et al., 1987;CSD code ZZZHVI01), consistent with the difference in water content, the powder patterns differ considerably. Visual examination of the structures shows that the arrangements of the citrate anions are very different. A mechanism for the transformation of one phase into the other is not obvious.
The Bravais-Friedel-Donnay-Harker (Bravais, 1866;Friedel, 1907;Donnay & Harker, 1937) morphology suggests that we might expect blocky morphology for anhydrous tripotassium citrate, with {011} as the principal faces. A second-order spherical harmonic texture model was included in the refinement. The texture index was only 1.001, indicating that preferred orientation was not significant for this rotated flat plate specimen. The asymmetric unit of the title compound, showing the atom numbering. Atoms are represented by 50% probability spheroids.

Figure 2
Comparison of the refined and optimized structures of anhydrous tripotassium citrate. The refined structure is in red, and the DFToptimized structure is in blue.

Figure 3
The crystal structure of K 3 C 6 H 5 O 7 , viewed down the c axis, with coordination spheres of the potassium cations in polyhedral representation.

Supramolecular features
The [KO n ] coordination polyhedra share edges and corners to form a three-dimensional framework (Fig. 3), with channels running down the c axis. The only hydrogen bond is an intramolecular one (Table 1) involving the hydroxy group and the central carboxylate group, with graph-set motif S(5). The Mulliken overlap population in the hydrogen-acceptor bond is 0.076 e. By the correlation in Rammohan & Kaduk (2016a), this hydrogen bond accounts for 15.1 kcal per mole of crystal energy.

Database survey
Details of the comprehensive literature search for citrate structures are presented in Rammohan & Kaduk (2016a). A reduced-cell search of the cell of anhydrous tripotassium citrate in the Cambridge Structural Database (Groom et al., 2016) (increasing the default tolerance from 1.5 to 2.0%) yielded 208 hits, but limiting the chemistry to C, H, K, and O only resulted in no hits. The powder pattern is now contained in the the Powder Diffraction File (ICDD, 2015) as entry 00-064-1370.

Synthesis and crystallization
Potassium citrate monohydrate was synthesized by dissolving 2.0796 g (10.0 mmole) H 3 C 6 H 5 O 7 (H 2 O) in 20 ml deionized water. 2.0731g K 2 CO 3 (15.0 mmole, Sigma-Aldrich) was added to the citric acid solution slowly with stirring. The resulting clear colourless solution was evaporated to dryness in a 333 K oven. The powder pattern matched PDF entry 02-064-1651, confirming the structure as potassium citrate monohydrate (Carrell et al., 1987). The monohydrate was heated at 15 K min À1 to 498 K, and held there for two minutes (the white solid started to discolour). The white solid was removed from the oven, and immediately placed in a sealed glass jar to cool.

Refinement details
Crystal data, data collection and structure refinement details are summarized in Table 2. The white solid was ground and blended with NIST SRM 640b Si internal standard in a mortar and pestle. The specimen was protected from the atmosphere by an 8 micron Kapton film attached to the sample holder with petroleum jelly. (The sample hydrates slowly on contact with ambient atmosphere.) The pattern (Fig. 4) was indexed on a primitive orthorhombic unit cell using ITO (Visser, 1969). Manual examination of the systematic absences suggested space group Pna2 1 . Pseudo-Voigt profile coefficients were as parameterized in Thompson et al. (1987) with profile coefficients for Simpson's rule integration of the Pseudo-Voigt function according to Howard (1982). The asymmetry correction of Finger et al. (1994) was applied and microstrain broadening by Stephens (1999  Č erný, 2002) using a citrate moiety and three potassium atoms as fragments. The structure was refined by the Rietveld method using GSAS/EXPGUI (Larson & Von Dreele, 2004;Toby, 2001). All C-C and C-O bond lengths were restrained, as were all bond angles. The hydrogen atoms were included at fixed positions, which were recalculated during the course of the refinement using Materials Studio (Dassault Systemes, 2014). The U iso values of the atoms in the central and outer portions of the citrate anion were constrained to be equal, and the U iso values of the hydrogen atoms were constrained to be 1.3Â those of the atoms to which they are attached.
The ADDSYM module of PLATON (Spek, 2009) suggested the presence of an additional centre of symmetry, and that the space group was Pnam. Refinement in this space group yielded poorer residuals, so we believe that Pna2 1 is the correct space group.

DFT calculations
After the Rietveld refinement, a density functional geometry optimization (fixed experimental unit cell) was carried out using CRYSTAL09 (Dovesi et al., 2005). The basis sets for the C, H, and O atoms were those of Gatti et al. (1994), and the basis set for K was that of Dovesi et al. (1991). The calculation used 8 k-points and the B3LYP functional, and took about 66 h on a 2.4 GHz PC. The U iso values from the Rietveld refinement were assigned to the optimized fractional coordinates.