Crystal structure of (S)-1-O-tert-butyldiphenylsilylglycerol: eight chiral molecules in a triclinic cell

In the triclinic crystal form, the molecules are linked by hydrogen bonds into an infinite assembly propagating along the a axis; hydrophobic tert-butyl and phenyl groups form an external coating of the assembly.


Chemical context
Glycerol nucleic acids (or glycol nucleic acids, GNA) and flexible nucleic acids (FNA) are two groups of unnatural polymers that have received attention as possible precursors of the present DNA/RNA-based life (Zhang et al., 2010). A common characteristic feature of both GNA and FNA is the presence of an acyclic three-carbon unit containing a stereogenic center, which gives rise to both possible configurations (R) and (S). Some nucleoside derivatives containing an acyclic appendix (instead of a ribose or 2-deoxyribose moiety) are very active antiviral agents: Acyclovir, Adefovir, Ganciclovir, Penciclovir, Tenofovir or Cidofovir. For the latter two compounds, this appendix has three carbon atoms and can be built up from chiral glycerol. Guaifenesin [3-(2-methoxyphenoxy)propane-1,2-diol], a common expectorant medication, is another example of a substituted chiral glycerol. A simple and stereospecific method to obtain 1-O-substituted glycerols with a predetermined configuration based on 5-Osubstituted d-and l-arabinose was realized (Doboszewski & Herdewijn, 2011) and further expanded by application of 6-Osubstituted d-glucopyranose and d-galactopyranose as presented in this paper.

Structural commentary
The initial crystal structure determination had been performed at 220 K and revealed a complex chiral triclinic structure with Z = 8 and multiple disordered fragments. Numbering scheme for molecule 1 of the title compound (50% probability displacement ellipsoids). The disordered H atom (occupancy 0.22 at 123 K) is shown in yellow.

Figure 2
Numbering scheme for molecule 2 of the title compound (50% probability displacement ellipsoids).

Figure 3
Numbering scheme for molecule 3 of the title compound (50% probability displacement ellipsoids).

Figure 4
Numbering scheme for molecule 4 of the title compound (50% probability displacement ellipsoids).

Figure 5
Numbering scheme for molecule 5 of the title compound (50% probability displacement ellipsoids). designation (Marsh, 1999(Marsh, , 2005. There is also the possibility of multiple polymorphs existing at different temperatures (Desiraju, 2007). To reduce the possibility of an erroneous determination of unit-cell parameters or overlooked higher symmetry and to look at this structure at lower temperatures, the experiment was repeated at 123 K and at 173 K using different diffractometer types. Preliminary, low-quality data were also obtained at 220 K (full data not included here). The results were consistent for all three measurements.
Data obtained for 123 K (data set 1) show the lowest degree of disorder and a lower uncertainty of all parameters of the crystal structure; further discussion thus deals mainly with this dataset.
There are eight independent molecules of the title compound in a unit cell in space group . Both the bond distances and angles of all moieties are unexceptional and consistent with standard values. As a result of the relatively long Si-C bond, there is little hindrance for rotation of the phenyl and tert-butyl groups in the tert-butyldiphenylsilyl fragments. This results in higher vibrational ellipsoids for the methyl groups and for some of the phenyl groups; in molecule 7, there are two visibly disordered phenyl rings (Fig. 7). Flexible glycerol fragments can occupy different positions. In two cases, this disorder was substantial and the fragments were refined as disordered (Figs. 6 and 8). Disorder of all groups visibly increases with temperature: for example, the occupancy of the minor component of the glycerol moiety in molecule 8 is 0.12 at 123 K, 0.29 at 173 K and 0.37 at 220 K; for molecule 6 it is 0.22 at 123 K, 0.36 at 173 K and 0.46 at 220 K. Several tertbutyl and phenyl groups are also becoming disordered at 220 K. Numbering scheme for molecule 6 of the title compound (50% probability displacement ellipsoids). The second glycerol chain (occupancy 0.22 at 123 K) is shown in green.

Figure 7
Numbering scheme for molecule 7 of the title compound (50% probability displacement ellipsoids).

Figure 8
Numbering scheme for molecule 8 of the title compound (50% probability displacement ellipsoids). The second glycerol chain (occupancy 0.13 at 123 K) is shown in green.

Figure 9
View of the unit-cell contents: the hydrophilic fragments are pointing inwards and the hydrophobic substituents form an outer 'coat'.
The eight molecules form an assembly with the hydrophobic phenyl and butyl groups making an external 'coat' and the hydrophilic chiral glycerol moieties forming the inner core of the unit cell ( Fig. 9). There is visible pseudosymmetry [see Zorky (1996) for the definition and discussion of this phenomenon] in this assembly. The triclinic system allows only an inversion centre as a symmetry element; however, it is prohibited by chirality in this case. Nevertheless, when the chiral glycerol groups are excluded from consideration, the whole structure becomes close to centrosymmetric: an overlay of two paired molecules is shown in Fig. 10. An attempt to overlay all tert-butyldiphenylsilyl moieties (Fig. 11) reveals a relatively good fit for the tert-butyl groups and one of the phenyl groups; interestingly, almost free rotation is observed for the second phenyl group.

Supramolecular features
Each of the eight molecules has two hydroxyl groups; each of them can be both a donor and an acceptor of a strong O-HÁ Á ÁO hydrogen bond (Gilli & Gilli, 2013). Indeed, fifteen such bonds are observed (Tables 1 and 2, Fig. 12). One weaker hydrogen bond (O14-H14GÁ Á ÁO13) connects a hydroxyl group to a neighbouring ether oxygen atom. The whole system is additionally stabilized by weaker C-HÁ Á ÁO interactions (two of them are shown in Tables 1 and 2 Overlay of two molecules 2 and 3 after inversion of fragment 3.

Figure 11
Overlay of the tert-butyldiphenylsilanyl fragments. Colour key: 1 -black, 2 -light green, 3 -red, 4 -grey, 5 -purple, 6 -green, 7 -blue, 8 -orange. Table 1 Hydrogen-bond geometry (Å , ) at 123 K. are bound by this complex system of hydrogen bonds, forming an infinite assembly (a 'beam') along the [100] axis (analogous to a reinforced concrete beam in construction, with hydrogen bonding serving as the reinforcement, see Fig. 13). These beams have an almost rectangular cross-section (Figs. 9 and 14). They are packed in layers parallel to the (001) plane by weak intermolecular interactions with no hydrogen bonds or stacking. These layers are assembled in a peculiar fashion, resembling a 'header bond' brick wall in masonry (Fig. 14). An ideal 'brick wall' tiling belongs to the rectangular plane symmetry group c2mm (No. 9; see Hahn, 2006). In our case, it is distorted to an oblique p1. There are no strong contacts between the layers; more careful examination even shows some small voids between them. However, again following the masonry analogy, such packing should be relatively stable simply for mechanical reasons (similar to a 'dry wall' with no mortar). The resulting three-dimensional crystal is stable despite multiple disorder inside the crystal cell. The self-assembly of the relatively simple title molecule into a complex infinite entity can serve as an illustration of the feasibility of glycerol-based assemblies in biochemical systems.

Database survey
There are 55 structures of substituted glycerol compounds deposited in the Cambridge Structural Database (CSD Version 5.39;Groom et al., 2016). Of these structures, two are glycerolphosphates; all others are organic compounds with a carbon atom connected to the terminal oxygen of the glycerol. Therefore, the current structure is the first silyl derivative of glycerol and the first non-carbon substituent neutral organic compound of that type. Two of the substituted glycerol structures [refcodes OKOXIW (Bredikhin et al. (2010) and WASHIJ (Bredikhin et al. (2008)] report the space group P1 and high Z 0 (8 and 4, respectively). However, analysis of the corresponding CIF files using the ADDSYM procedures of PLATON (Spek, 2009) suggests much higher symmetry and a smaller Z 0 : for OKOXIW the space group is I2 and Z 0 = 4; for WASHIJ it is Iba2 and Z 0 = 1. These examples demonstrate importance of additional caution while working with high Z 0 numbers (Marsh, 1999 Table 2 Hydrogen-bond geometry (Å , ) at 173 K. Symmetry codes: (i) x À 1; y; z; (ii) x þ 1; y; z.

Figure 14
Packing diagram: view along the [100] axis. The 'header bond' brick wall motif is highlighted.
itself spontaneously crystallizes (Kusukawa et al., 2013) in a chiral space group, P2 1 2 1 2 1 . The exact number of all known structures that crystallize in space group P1 with Z = 8 is ambiguous. Structures with large Z 0 have been reviewed in detail by Steed & Steed (2015) and Brock (2016); databases of high-Z 0 structures were created based on CSD data. A direct search of the CSD (CSD Version 5.39; Groom et al., 2016) yields 41 entries for P1, Z = 8. However, some of them are obvious typographical errors and several are unambiguously convertible to higher symmetry and consequently lower Z 0 . Most of the remaining (around 30) unambiguous structures are pseudocentrosymmetric, with an 80-95% fit for an exact centrosymmetric structure. Nevertheless, for a triclinic structure, the chirality of the molecules serves as a solid proof of space group P1, similar to our structure.
There are numerous structures of silyl-substituted molecules similar to the title compound (e.g. there are 3874 structures with a diphenylsilicon moiety and 475 tert-butyldiphenylsilyl compounds). Of these, eight are compounds with high Z 0 , five of which are chiral. The triclinic centrosymmetric structure of tert-butyldiphenylsilanol (Habtemariam et al., 2015) shows Z = 8 (Z 0 = 4); four independent molecules form a pseudotetragonal motif around four hydrogen bonds connecting the silanol groups in a fashion that remotely resembles the current structure. It was suggested by Prince et al. (2002) that weak interactions induce asymmetry in the crystal structures of triaryl derivatives of group 14 elements (Si, Ge, Sn), resulting in an abnormally large number of structures with high Z 0 .

Synthesis and crystallization
The title compound was prepared from 5-O-tert-butyldiphenylsilyl-d-glucopyranose (Tsutsui et al. 2014) or 5-O-tertbutyldiphenylsilyl-d-galactopyranose (Doboszewski & Herdewijn, 2012) using the procedure published before for 5-O-tert-butyldiphenylsilyl-d-or -l-arabinofuranose (Doboszewski & Herdewijn, 2011). To a solution of 5-O-tert-butyldiphenylsilyl-d-glucopyranose or 5-O-tert-butyldiphenylsilyld-galactopyranose (2.0 g, 5.1 mmol) in 96% ethanol, 20 mL, was added portionwise a solution of NaIO 4 (3.8 g, 17.8 mmol) in water (15 mL) over a period of 10 min in a magnetically stirred ice bath. The solution became turbid within a few seconds. After the end of addition, the mixture was left at room temperature for 1.5 h. The white solid was removed by filtration (sintered glass) and the filtrate was cooled in an ice bath. NaBH 4 (0.15 g, 4 mmol) was added with manual swirling. After 1h, the reaction mixture was transferred to a separatory funnel and extraction was performed (H 2 O-CH 2 Cl 2 ). The organic phase was washed with water, dried (MgSO 4 ) and the solids were removed by filtration. Vacuum evaporation at

Refinement
Crystal data, data collection and structure refinement details are summarized in Table 3. All C-bound hydrogen atoms were placed in calculated positions and treated as riding with U iso (H) = 1.2U eq (C) or 1.5U eq (C methyl ). The O-bound H atoms were refined as riding with O-H = 0.84 Å with U iso (H) = 1.5U eq (O). The difference-Fourier map indicated possible disorder for the H atom of hydroxyl group O3 (Fig. 1). The occupancies of this disordered group were set to be equal to those of the neighboring disordered glycerol fragment of molecule 6. It was not possible to locate and refine a hydrogen atom of hydroxyl group O34 with occupancy of 0.13 at 123 K; its position was set as identical to that of H24 (Fig. 6). At 173 K, this atom was placed at a calculated position, which appeared to be very close to the previous assumption.
In a disordered phenyl group (atoms C121-C126 and C161-C166), the bond distances were restrained to make the geometry of both rings similar; an additional set of restraints was applied to make the phenyl rings symmetrical (Fig. 7). The anisotropic parameters of atoms C121 and C161 were set to be equal. Two disordered glycerol fragments in molecules 6 and 8 were resolved; restraints were applied to all interatomic distances in these fragments to make equivalent distances approximately equal. The anisotropic parameters of two closely located pairs of atoms (C117 and C217, and O24 and O34) were set as equal (Figs. 6 and 8). An additional set of restraints was applied to make the Si6-O16 and Si-O216 as well as the Si8-O22 and S8-O32 bond distances approximately equal. An anti-bumping restraint was added to prevent a short distance between calculated hydrogen-atom positions involving the low-occupancy fragment in molecule 6.
The chirality of the title compound was known from the synthetic route. Analysis of the absolute structure using anomalous scattering (Flack, 1983;Spek, 2009) was undertaken for three different crystals and confirmed the original assignment (Table 3).

Funding information
Financial support from the State University of New York for acquisition and maintenance of the X-ray diffractometer is gratefully acknowledged.  (14) 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.

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.

Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å 2 )
x y z U iso */U eq Occ. (