Single-crystal structure analysis of non-deuterated triglycine sulfate by neutron diffraction at 20 and 298 K: a new disorder model for the 298 K structure

A precise crystal-structure analysis using a neutron diffractometer with high-power neutron sources at the J-PARC facility has been performed on non-deuterated triglycine sulfate at 20 K and 298 K and a new double-potential-well disorder model for the O—H⋯O hydrogen bond in the 298 K structure is proposed.

Precise single-crystal structure analyses of the title compound, bis(glycinium) sulfate-glycine (1/1), 2C 2 H 6 NO 2 + ÁSO 4 2À ÁC 2 H 5 NO 2 (or C 6 H 17 N 3 O 10 S), nondeuterated triglycine sulfate (HTGS) at 20 K and 298 K were undertaken using time-of-flight neutron diffraction data. At 20 K for the O-HÁ Á ÁO hydrogen bond between the glycinium cation and the zwitterionic, unprotonated glycine molecule that is associated with the ferroelectric behaviour of HTGS, O-H = 1.070 (3), HÁ Á ÁO = 1.408 (3) [ = 0.338 (4)], OÁ Á ÁO = 2.4777 (15) Å and O-HÁ Á ÁO = 179.0 (4) , which is in good agreement with previous studies. Two reasonable structures for the same three atoms were refined for the 298 K dataset. One is a single-minimum potential-energy model, with O-H = 1.090 (12) (2) . These models did not show any significant differences in R factors. In addition, the unit-cell parameters and other structural parameters of HTGS did not show any major differences compared to those of partially deuterated TGS and fully deuterated TGS for both 20 K and 298 K.

Chemical context
Triglycine sulfate, 2(C 2 H 6 NO 2 ) + Á(C 2 H 5 NO 2 )Á(SO 4 ) 2-(TGS), is a hydrogen-bond ferroelectric material (Matthias et al. 1956) exhibiting a second-order and order-disorder-type ferroelectric phase transition at a Curie temperature (T C ) of 322 K (Triebwasser, 1958). The TGS structure belongs to the point group C 2h and the space group P2 1 /m in the paraelectric phase and C 2 and P2 1 in the ferroelectric phase, respectively (Wood & Holden, 1957). Because of its high pyroelectricity, TGS has long been used as a material for pyroelectric sensors. Therefore, determining the crystal structure of TGS is essential for understanding such physical properties.
The atomic coordinates, except for those of the hydrogen atoms, of TGS at room temperature were first determined using single-crystal X-ray diffraction (Hoshino et al., 1959). ISSN 2056-9890 The study assumed the presence of one neutral glycine molecule (C 2 H 5 NO 2 ) exhibiting a zwitterionic configuration, and two monoprotonated glycinium ions (C 2 H 6 NO 2 + ), from the detailed analysis of the bond lengths and angles of the glycine molecules. The authors also proposed a hydrogen-bonding scheme and pointed out that the hydrogen atom that lies between the oxygen atom of the carboxyl group in the glycine III cation (GIII) and the O atom in the glycine II molecule (GII) plays a crucial role in the dipole reversal. Many structural studies on TGS have subsequently been conducted (see Database survey): most of them were X-ray diffraction studies, but some of them were neutron diffraction studies. The atomic coordinates of non-deuterated TGS (hereinafter, designated as HTGS in place of TGS), including those of the hydrogen atoms at room temperature, were first revealed using singlecrystal neutron diffraction (Padmanabhan & Yadav, 1971) and the atomic arrangements including hydrogen atoms of the zwitterion and glycinium ions were directly observed. The neutron diffraction experiment revealed that the hydrogen atom forming the O-HÁ Á ÁO hydrogen bond between the GIII and GII species was closer to the GIII O atom compared to that in GII. This result agreed with that obtained by Hoshino et al. (1959). The structure refinement of HTGS with an applied external electric field at 298 K revealed the placement of all the hydrogen atoms and the unambiguous definition of the hydrogen-bonding scheme in an ordered domain structure (Kay & Kleinberg, 1973).
Crystal-structure refinements of partially deuterated TGS (DTGS), where deuterium replaced the H atoms except for the hydrogen atoms of the methylene (CH 2 ) group in each glycine molecule and those in sulfuric acid molecules at 40 K and 180 K (Protas et al., 1997) showed that the refined structures were consistent with those of the HTGS reported by Kay & Kleinberg (1973). Protas et al. (1997) also observed that HTGS and DTGS in the ferroelectric phase had a consistent structure from 40 K to 298 K. The deuterium atom lying between GIII and GII was $0.40 Å closer to the O atom of the carboxyl group of GIII than that of GII at both temperatures. In contrast, the crystal-structure refinement of HTGS at room temperature showed positional disorder over two adjacent sites of the amino group in glycinium cation I (GI) (Choudhury & Chitra, 2008). However, this is not in agreement with the refined structure of HTGS reported by Padmanabhan & Yadav (1971) where the GI species was analysed as an ordered structure.
In the crystal structure of fully deuterated TGS (FDTGS), all the hydrogen atoms in the glycine molecules and sulfuric acid molecules are substituted by deuterium atoms: the crystal structures did not show major changes between 20 K and 295 K (Hudspeth et al., 2013). The unit-cell parameters of these FDTGS structures were consistent with those of HTGS (Kay & Kleinberg, 1973;Choudhury & Chitra, 2008) and DTGS (Protas et al., 1997).
Structural analysis of DTGS at 40 K and FDTGS at 20 K have been undertaken by Protas et al. (1997) and Hudspeth et al. (2013), respectively, as mentioned above. However, a precise structural analysis of HTGS including hydrogen atoms at low temperatures has not been reported. Furthermore, two different structures of HTGS at $298 K were reported: one is an ordered structure by Padmanabhan & Yadav (1971) and the other is a disordered structure by Choudhury & Chitra (2008).

Structural
The lattice constants and the key O15-H15Á Á ÁO3 i [symmetry code: (i) 3 À x, À 1 2 + y, 2 À z for the present study] bond lengths for HTGS, DTGS and FDTGS at low temperature are listed in Table 1. The parameters do not show any major differences, and H15 is 0.338 (4) Å closer to atom O15 in GIII than O3 in GII. This result shows good agreement with the data previously reported for DTGS (Protas et al. 1997), thus it may be concluded that the intermolecular distances and angles do not change significantly upon deuteration.
unit, and the features of the molecular structures are consistent with those for the 20 K structure apart the disordered N11/N11B amino group [refined site occupancies = 0.874 (8):0.126 (8)] in the GI cation and the O-HÁ Á ÁO association for GIII and GII. Two models were refined considering the H atom between O15 in GIII and O3 in GII. For one model (298 K model 1), the H15 atom was refined with a large ellipticity along the bond path between O15 and O3 as a single minimum potential energy structure [ Fig. 5(a)]. A doubleminimum potential-energy structure could be deduced because the distance between O15 and O3 i [symmetry code: (i) 1 À x, À 1 2 + y, Àz for the present study] did not increase with an increase in the temperature; thus for the other model (298 K model 2), a pair of hydrogen atoms were refined along the bond path between O15 and O3 i , the double-minimum potential structure [ Fig. 5 The key parameters for the O15-H15Á Á ÁO3 i hydrogen bond at 298 K are summarized in Table 2. The residuals for models 1 and 2 (Table 3) are almost identical: model 2 has one more variable parameter than model 1 (358 compared to 357). For model 1, H15 is 0.271 (17) Å closer to O15 in GIII than O3 i in GII. On the other hand, the distance between O15 and H15 [1.090 (12) Å ] is almost the same as that at 20 K despite there being no distance restraint for the H15Á Á ÁO3 i separation. Therefore, the mixed structure (model 2) of the major ferro-   (2) 177.3 (9) -Symmetry code for HGTS in this study: (i) 1 À x, À 1 2 + y, Àz.

Figure 5
[001] projection of the unit cell and detail of the hydrogen bond between GII and GIII of HTGS at 298 K with hydrogen bonds shown as dashed lines. The short O15-H15Á Á ÁO3 bonds are shown as pink dashed lines. Atoms N11 and N11B are disordered with occupancies of 87.5%/12.5%. Model 1 (a); a single-minimum potential energy model for H5; model 2 (b); a double-minimum potential-energy model for H15 and H3. Symmetry codes:

Figure 4
The molecular structure of HTGS at 298 K (model 2) showing 50% displacement ellipsoids for all atoms. electric phase and minor paraelectric phase is strongly suggested, because the occupancies of N11 and N11B and H15 and H3 i are related by symmetry.
The unit-cell parameters and bond lengths for HTGS, DTGS, and FDTGS at 298 K are listed in Table 2. The lattice parameters did not show any major differences and this result shows good agreement with that previously reported for DTGS (Protas et al., 1997). We may conclude that the intermolecular distances and angles do not change significantly upon deuteration.
In the previous studies using single-crystal neutron diffraction, Kay & Kleinberg (1973) proposed an ordered structure of HTGS because the domains were oriented by applying an external electric field. Hudspeth & Goossens (2012) proposed an ordered structure for FDTGS because T C for FDTGS increased by approximately 12 K compared to HTGS. Choudhury & Chitra (2008) proposed a disordered structure for the GI amino group with unequal occupancies of N11 (88%) and N11B (12%); this occupancy ratio is in excellent agreement with the results in this study. For the hydrogen atom between the oxygen atom of the carboxyl group in GIII and that in the GII, the OÁ Á ÁO distance was 2.470 (9) Å , and the H atom was approximately 0.241 Å closer to the GIII O atom than that in GII. They concluded that the structure of HTGS at room temperature has a single minimum potential energy in the O-HÁ Á ÁO hydrogen-bond path between GIII and GII. In this study, two reasonable structures were refined as a single-minimum potential-energy model and a double-minimum model without any significant differences. Therefore, we conclude that there is a significant possibility of a double-minimum potential-energy model for HTGS at 298 K.

Supramolecular features
Hydrogen bonds in the refined structures were consistent with those reported previously (see supporting information) and no additional intermolecular interactions were found. Therefore, the 20 K and 298 K structures form essentially the structural motif of a three-dimensional network of N-HÁ Á ÁO and O-HÁ Á ÁO hydrogen bonds between glycinium cations, glycine molecules and sulfate ions.

Synthesis and crystallization
The HTGS crystals were grown in an aqueous solution by the slow evaporation method at $293 K. Glycine (13.06 g; FUJI-FILM Wako Pure Chemical Corporation; purity ! 99.0%) and sulfuric acid (3.1 ml; FUJIFILM; molar ratio 3:1) was added to 50 ml of water in a 100 ml beaker. They were dissolved by heating at $313 K with a 300 r.p.m. magnetic stirrer. After completely dissolving them, plastic films were double-wrapped around the beaker, and some holes were knocked in the films to evaporate the water slowly. The beaker was left to stand at $293 K. HTGS was crystallized after approximately a month, and then the solution was filtered. The collected crystals were dried in a desiccator at $293 K.

Refinement
Crystal data, data collection, and structural refinement details are summarized in Table 3. All data were collected using the single-crystal neutron diffractometer SENJU (Ohhara et al., 2016) at beamline BL18 of the Materials and Life Science Facility, Japan Proton Accelerator Research Complex. The crystal (colourless cube, $2.8 mm edge length) mounted on an aluminum pin was cooled to 20 K in a closed-cycle helium cryostat. The crystal was surrounded by 41 two-dimensional scintillation area detectors during the data collection. The same crystal was used for the measurement at 298 K after warming to room temperature. Three-dimensional data of (x, y, ) were measured in 16 different orientations for each dataset. The measurement time was 1.5 h for one orientation; the raw data were processed using STARGazer (Ohhara et al. 2009) to generate HKLF files and visualize (x, y) slice maps and merged TOF profiles.

Figure 6
A difference scattering density map for the 298 K structure without the H atom between O15 and O3. The red dotted lines and green solid lines show negative and positive density distribution, respectively. A nuclear density distribution with a large ellipticity along the bond path between O15 and O3 is observed. For all structures, data collection: STARGazer (Ohhara et al., 2009); cell refinement: STARGazer (Ohhara et al., 2009); data reduction: STARGazer (Ohhara et al., 2009); program(s) used to solve structure: A reported structure determined by single-crystal X-ray diffraction (Hoshino et al., 1959) was used as the initial structure model.. Program(s) used to refine structure: SHELXT2018/3 (Sheldrick, 2015) for (20K); SHELXL2018/3 (Sheldrick, 2015) for 298KModel1, 298KModel2. For all structures, molecular graphics: Mercury (Macrae et al., 2020); software used to prepare material for publication: PLATON (Spek, 2020) and publCIF (Westrip, 2010).

Special details
Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes. Refinement. reflns_Friedel_fraction is defined as the number of unique Friedel pairs measured divided by the number that would be possible theoretically, ignoring centric projections and systematic absences.′

Special details
Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes. Refinement. reflns_Friedel_fraction is defined as the number of unique Friedel pairs measured divided by the number that would be possible theoretically, ignoring centric projections and systematic absences.

Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å 2 )
x y z U iso */U eq Occ. (<1) S1 0.9997 (5) 0.7500 (5) 0.7732 (8) 0.0132 (7)  Extinction coefficient: 0.106 (5) Absolute structure: Indeterminate for a neutron structure Special details Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes. Refinement. reflns_Friedel_fraction is defined as the number of unique Friedel pairs measured divided by the number that would be possible theoretically, ignoring centric projections and systematic absences.