Structures of disodium hydrogen citrate monohydrate, Na2HC6H5O7(H2O), and diammonium sodium citrate, (NH4)2NaC6H5O7, from powder diffraction data

The crystal structures of disodium hydrogen citrate monohydrate and diammonium sodium citrate have been solved and refined using laboratory X-ray powder diffraction data and optimized using density functional techniques.

The crystal structures of disodium hydrogen citrate monohydrate, Na 2 HC 6 H 5 O 7 (H 2 O), and diammonium sodium citrate, (NH 4 ) 2 NaC 6 H 5 O 7 , have been solved and refined using laboratory X-ray powder diffraction data, and optimized using density functional techniques. In NaHC 6 H 5 O 7 (H 2 O), the NaO 6 coordination polyhedra share edges, forming zigzag layers lying parallel to the bc plane. The hydrophobic methylene groups occupy the interlayer spaces. The carboxylic acid group makes a strong charge-assisted hydrogen bond to the central carboxylate group. The hydroxyl group makes an intramolecular hydrogen bond to an ionized terminal carboxylate oxygen atom. Each hydrogen atom of the water molecule acts as a donor, to a terminal carboxylate and the hydroxyl group. Both the Na substructure and the hydrogen bonding differ from those of the known phase Na 2 HC 6 H 5 O 7 (H 2 O) 1.5 . In (NH 4 ) 2 NaC 6 H 5 O 7 , the NaO 6 coordination octahedra share corners, making double zigzag chains propagating along the b-axis direction. Each hydrogen atom of the ammonium ions acts as a donor in a discrete N-HÁ Á ÁO hydrogen bond. The hydroxyl group forms an intramolecular O-HÁ Á ÁO hydrogen bond to a terminal carboxylate oxygen atom.
As part of our ongoing studies in this area, we now report the syntheses and structures of disodium hydrogen citrate monohydrate, Na 2 HC 6 H 5 O 7 (H 2 O), (I), and diammonium sodium citrate, (NH 4 ) 2 NaC 6 H 5 O 7 , (II).

Structural commentary
The structure of (I) was solved and refined from powder X-ray data and optimized by density functional theory (DFT) calculations (see Experimental section) and is illustrated in Fig. 1. The root-mean-square Cartesian displacement of the non-hydrogen citrate atoms in the Rietveld refined and DFToptimized structures is 0.0764 Å (Fig. 2). The excellent agreement between the two structures is strong evidence that the experimental structure is correct (van de Streek & Neumann, 2014). All of the citrate bond distances, bond angles, and torsion angles fall within the normal ranges indicated by a Mercury Mogul geometry check (Macrae et al., 2020). The citrate anion occurs in the gauche, trans-conformation (about C2-C3 and C3-C4, respectively), which is one of the two low-energy conformations of an isolated citrate ion (Rammohan & Kaduk, 2018). The central carboxylate group and the hydroxyl group exhibit a small twist (O16-C6-C3-O17 torsion angle = 10.3 ) from the normal planar arrangement. The Mulliken overlap populations indicate that the Na-O bonds are ionic. Both Na cations are six-coordinate (distorted octahedral). The bond-valence sums for Na20 and Na21 are 1.09 and 1.04 respectively.
The citrate anion triply chelates to Na20 through the terminal carboxylate oxygen atom O14, the central carboxylate oxygen atom O16, and the hydroxyl group O17. All oxygen atoms except O12 coordinate to at least one Na cation.
The structure of (II) was solved and refined from powder X-ray data and optimized by density functional theory (DFT) calculations (see Experimental section) and is illustrated in Fig. 3. The root-mean-square Cartesian displacement of the non-hydrogen citrate atoms in the Rietveld refined and DFToptimized structures is 0.067 Å (Fig. 4). The r.m.s. displacement of the sodium ions is 0.037 Å and the equivalent values for the ammonium ions N20 and N21 are 0.148 and 0.147 Å , respectively. The excellent agreement between the two structures is strong evidence that the experimental structure is correct (van de Streek & Neumann, 2014). Almost all of the citrate bond distances, bond angles, and torsion angles fall within the normal ranges indicated by a Mercury Mogul geometry check (Macrae et al., 2020). Only the O13-C5-C4 angle of 117.3 [average = 119.4 (7) , 6-score = 3.2] is flagged as unusual. Mogul finds a population of three similar angles and the standard uncertainty is exceptionally low at 0.7 , so the Z-score is not of concern. The citrate anion occurs in the trans, trans-conformation (about C2-C3 and C3-C4), which Comparison of the refined and optimized structures of (I): the refined structure is in red, and the DFT-optimized structure is in blue.

Figure 3
The asymmetric unit of (II) with the atom numbering and 50% probability spheres.

Figure 1
The asymmetric unit of (I) with the atom numbering and 50% probability spheres.
is one of the two low-energy conformations of an isolated citrate ion (Rammohan & Kaduk, 2018). The central carboxylate group and the hydroxyl group exhibit a very small twist [O17-C3-C6-O15 = 0.34 ] from the normal planar arrangement. The Mulliken overlap populations indicate that the Na-O bonds are ionic.
The Bravais-Friedel-Donnay-Harker method suggests that we might expect platy morphology for diammonium sodium citrate, with {100} as the major faces. A 2nd order spherical harmonic model was included in the refinement. The texture index was only 1.006, indicating that preferred orientation was not significant in this rotated capillary specimen.

Supramolecular features
In the extended structure of (I), the NaO 6 coordination polyhedra share edges to form zigzag layers lying parallel to the bc plane (Fig. 5). The layers are conveniently viewed along [110] (Figs. 6 and 7). The hydrophobic methylene groups occupy the interlayer spaces. The carboxylic acid O13-H24 group makes a strong (16.8 kcal mol À1 ) charge-assisted hydrogen bond to the central carboxylate oxygen atom O15. The energies of the O-HÁ Á ÁO hydrogen bonds were calculated using the correlation of Rammohan & Kaduk (2018  Comparison of the refined and optimized structures of (II): the refined structure is in red, and the DFT-optimized structure is in blue.

Figure 7
View of the crystal structure of (I), viewed down [110]. intramolecular hydrogen bond. In the sesquihydrate, all three independent water molecules bridge Na cations; in this monohydrate the water molecule also bridges two Na. In the sesquihydrate, there are eight-membered rings of Na cations, while in this monohydrate structure the Na coordination spheres form layers. The triclinic unit cell of Na 2 HC 6 H 5 O 7 (H 2 O) 1.5 corresponds roughly to a 1/2 subcell of the current Pbca cell. The transformation matrix from the current cell to the standard orthorhombic cell is [0 1 0 / 0 0 1 / 1 0 0], and the transformation matrix from the standard cell to the subcell is [1 0 0 / À1/2 À 1/2 1/2 / À1/2 1/2 1/2]. Given the differences in the Na substructures and the hydrogen bonding, the similarities of the cells are a coincidence.
The CRYSTAL14 (Dovesi et al., 2014) energy per formula unit of Na 2 HC 6 H 5 O 7 (H 2 O) is À1160.0 eV. The energy per formula unit of Na 2 HC 6 H 5 O 7 (H 2 O) 1.5 is À1197.9 eV. Calculated in the same way, the energy of an isolated water molecule is À76.4 eV. Thus, the energy of the sesquihydrate is thus 0.23 eV higher than that of the sum of the monohydrate and half a water molecule. The difference is only 5.4 kcal mol À1 , so the structures must be considered comparable in energy.
In the extended structure of (II), the NaO 6 coordination octahedra share corners to form double zigzag chains propagating along the b-axis direction (Figs. 8 and 9). Each hydrogen atom of the ammonium ions acts as a donor in a discrete N-HÁ Á ÁO hydrogen bond ( Table 2). The hydroxyl group O17-H18 forms an intramolecular hydrogen bond to the terminal carboxylate oxygen atom O11. The N-HÁ Á ÁO hydrogen bond energies were calculated by the correlation of Wheatley & Kaduk (2019), and the O-HÁ Á ÁO hydrogen bond energy was calculated by the correlation of Rammohan & Kaduk (2018). Despite the similarities in the formulae, the crystal structures of diammonium sodium citrate and diammonium potassium citrate (Patel et al., 2020) differ. In the current compound, the NaO 6 coordination polyhedra share corners to form zigzag chains, while in diammonium potassium citrate the KO 7 polyhedra are isolated. The powder patterns ( Fig. 10) are not particularly similar, and except for layers containing the ammonium ions (Fig. 11), the structures exhibit many differences. The crystal structure of (II), viewed down [010].

Figure 9
The crystal structure of (II), viewed nearly down the b-axis direction, to better illustrate the chains. Table 2 Hydrogen-bond geometry (Å , ) for (II) (DFT).

Figure 11
Comparison of the crystal structures of (II) and (NH 4 ) 2 KC 6 H 5 O 7 .

Database survey
Details of the comprehensive literature search for citrate structures are presented in Rammohan & Kaduk (2018). Another pattern of the same sample of '(NH 4 )Na 2 C 6 H 5 O 7 ' measured using Cu K radiation, was indexed on a primitive monoclinic unit cell having a = 16.9845, b = 8.6712, c = 12.2995 Å , = 90.03 , V = 1800.2 Å 3 , and Z = 8 using JADE Pro (MDI, 2019). Analysis of the systematic absences using FOX (Favre-Nicolin & Č erný, 2002) suggested that the space group was Pbca. A reduced cell search in the Cambridge Structural Database (Groom et al., 2016) yielded 83 hits but no citrate crystal structures. The pattern of (NH 4 ) 2 NaC 6 H 5 O 7 was indexed with DICVOL14 (Louë r & Boultif, 2014), using the PreDICT interface (Blanton et al., 2019). Analysis of the systematic absences using EXPO2014 (Altomare et al., 2013) suggested the space group P2 1 /n, which was confirmed by successful solution and refinement of the structure. A reduced cell search of the cell in the Cambridge Structural Database (Groom et al., 2016) resulted in eleven hits, but no citrate structures.

Synthesis and crystallization
0.2415 g of (NH 4 ) 2 CO 3 (Aldrich) and 0.5376 g of Na 2 CO 3 (Alfa Aesar) were added to a solution of 1.0162 g citric acid (Sigma-Aldrich) monohydrate in 10 ml of water. After the fizzing subsided, the clear solution was dried at ambient conditions to yield a clear glass. Successive heating at 361, 394, and 410 K did not induce crystallization. The glass was redissolved in 10 ml of water and layered with 40 ml of ethanol. The beaker was covered and left to stand at ambient conditions. After three days, the solvents were blended, but the solution was clear. The beaker was uncovered and after another three days, a white solid was observed at the bottom of the beaker. The solution was decanted and the solid was dried at ambient conditions. After one day, the solid was still wet, so it was dried in a 361 K oven for a few minutes to yield a white powder of (I). The powder pattern was measured from a 0.7 mm diameter capillary specimen on a PANalytical Empyrean diffractometer equipped with an incident beam focusing mirror and an X'Celerator detector, using Mo K radiation. The pattern was measured from 1-50 2 in 0.010067 steps, counting for four seconds per step.
Diammonium sodium citrate was synthesized by dissolving 1.1231 g diammonium hydrogen citrate (Fisher Lot #995047) and 0.2713 g sodium carbonate (Alfa Aesar) in $6 ml of deionized water. When the fizzing stopped, the clear solution was layered with about 20 ml of acetone and left to stand at ambient conditions. After two days, the solvents had blended and the product was a clear syrup. The syrup was dried at 363 K for three hours to yield a white solid, (II). The powder pattern was measured from a 0.7 mm diameter capillary specimen on a PANalytical Empyrean diffractometer equipped with an incident beam focusing mirror and an X'Celerator detector, using Mo K radiation. The pattern was measured from 1-50 2 in 0.010067 steps, counting for four seconds per step.

Refinement
Crystal data, data collection and structure refinement details for (I) and (II) are summarized in Table 3. The final Rietveld plots for (I) and (II) are shown in Figs. 12 and 13, respectively. The structure of (I) was solved using Monte-Carlo simulated annealing techniques as implemented in FOX (Favre-Nicolin & Č erný 2002). The citrate anion, two sodium atoms and a nitrogen atom were used as fragments. One of the fifteen runs yielded a cost factor much lower than the others and was used as the basis for refinement.
The structure was refined by the Rietveld method using GSAS-II (Toby & Von Dreele, 2013). In the initial refinement, the U iso value of the nitrogen atom refined to a negative value and the nitrogen atom was 2.4 Å away from the two sodium atoms. Both of these facts suggested that this atom was not the nitrogen of an ammonium ion, but the oxygen of a water molecule. Thus, the compound was not the intended compound. Rietveld plot for (II). The blue crosses represent the observed data points, and the green line is the calculated pattern. The cyan curve is the normalized error plot. The row of blue tick marks indicates the calculated reflection positions. The red and cyan tick marks indicate the reflection positions for the diammonium sodium citrate and diammonium hydrogen citrate impurities. The red line is the background curve.

Figure 12
Rietveld plot for (I). The blue crosses represent the observed data points, and the green line is the calculated pattern. The cyan curve is the normalized error plot. The vertical scale has been multiplied by a factor of 5Â for 2 > 26.5 . The row of blue tick marks indicates the calculated reflection positions. The red line is the background curve.
The hydrogen atoms were included in fixed positions, which were re-calculated during the course of the refinement using Materials Studio (Dassault Systems, 2019). Initial positions of the active hydrogen atoms H18, H22, H23, and H24 were deduced by analysis of potential hydrogen-bonding patterns. The U iso values of C2, C3, and C4 were constrained to be equal, and those of H7, H8, H9, and H10 were constrained to be 1.3Â that of these carbon atoms. The U iso value of C1, C5, C6, and the oxygen atoms were constrained to be equal, and that of H18 was constrained to be 1.3Â this value. The background was described by a four-term shifted Chebyshev polynomial with an extra peak at 12.85 to describe the scattering of the glass capillary.
A density functional geometry optimization was carried out using CRYSTAL14 (Dovesi et al., 2014). The basis sets for the H, C, N, and O atoms were those of Gatti et al. (1994), and the basis set for Na was that of Peintinger et al. (2013). The calculation was run on eight 2.1 GHz Xeon cores (each with 6 Gb RAM) of a 304-core Dell Linux cluster at IIT, using 8 k-points and the B3LYP functional, and took $44 h.
The structure of (II) was solved using DASH (David et al., 2006) using a citrate ion, two nitrogen atoms, and a sodium atom as fragments, along with Mogul Distribution Bias, and <010> preferred orientation. Two of the 100 runs yielded residuals lower than the others. The structure was refined by the Rietveld method using GSAS-II (Toby & Von Dreele, 2013). The hydrogen atoms were included in fixed positions, which were recalculated during the course of the refinement using Materials Studio (Dassault Systems, 2019). All C-C and C-O bond distances and all bond angles were restrained based on a Mercury Mogul Geometry Check (Sykes et al., 2011;Bruno et al., 2004) of the molecule. The U iso values of the atoms in the central and outer portions of the citrate 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. A four-term shifted Chebyschev function was used to model the background, along with a peak at 12.5 to describe the scattering from the capillary and any amorphous component. A single phase model did not account for all of the peaks. We compared those peaks to the patterns of known ammonium and sodium citrates and identified diammonium hydrogen citrate (Wheatley & Kaduk, 2019) and disodium hydrogen citrate monohydrate [i.e., (I)] as impurities, and included them in the refinement. Their concentrations were 8.8% and 4.1% weight percentages respectively.
A density functional geometry optimization (fixed experimental unit cell) was carried out using VASP (Kresse & Furthmü ller, 1996) through the MedeA graphical interface (Materials Design, 2016). The calculation was carried out on 16 2.4 GHz processors (each with 4 GB RAM) of a 64processor HP Proliant DL580 Generation 7 Linux cluster at North Central College. The calculation used the GGA-PBE functional, a plane wave cutoff energy of 400.0 eV, and a k-point spacing of 0.5 Å À1 leading to a 2 Â 3 Â 2 mesh, and took 18 h. A single point calculation was done using CRYSTAL14. The basis sets for the H, C, N, and O atoms were those of Gatti et al. (1994), and the basis set for Na was that of Peintinger et al. (2013). The calculation was run on eight 2.1 GHz Xeon cores (each with 6 GB RAM) of a 304-core Dell Linux cluster at IIT, using 8 k-points and the B3LYP functional, and took five days.   (2) C2-C1-O11 114.7 (4) C5-C4-H9 105.6 (7) C2-C1-O12 117.6 (4) C3-C4-H10 106.5 (5)