Crystal structure of 2α-(1,1-diphenylethyl)-4-methyl-4α,5α-diphenyl-1,3-dioxolane: the result of a non-acid pinacol rearrangement

The title compound was produced unexpectedly from (±)-1,2-diphenyl-1,2-propanediol by a sequential non-acid Pinacol rearrangement followed by acetal formation during recrystallization in 1-butanol. The tri-substituted dioxolane ring has a twist conformation and in the crystal, molecules are linked by C—H⋯O hydrogen bonds, forming chains along [001].


Chemical context
The pinacol rearrangement is a well-documented reaction (Collins, 1960) that converts substituted 1,2-diols into aldehydes or ketones (pinacolone derivatives), usually with the aid of mineral or Lewis acid catalysis. In the present work, a pinacol rearrangement has occurred during recrystallization in the absence of a catalyst, thus transforming the intended object of our study (1), into the pinacol rearrangement aldehyde (3), which then reacts (by acetal formation) with another molecule of (1) to form the unexpected product (2) presented in this paper, as shown in the scheme below.
The pseudo-equatorial orientation of the diphenylethane group at C2 likely follows from thermodynamic control during the acetalization step. For the reaction conditions of our recrystallization, the acetalization step must proceed faster than pinacol rearrangement. A similar reaction has been described by Ciminale et al. (2005). There are very few other reports of the pinacol rearrangement occurring in the absence of catalysts: for example, the thermal rearrangement of pinacol to pinacolone in supercritical H 2 O (Ikushima et al., 2000), the conversion of 1,1,2-triphenylethane-1,2-diol to 1,2,2-triphenylethan-1-one when heated above its melting ISSN 2056-9890 point (Collins, 1960), and a vinylogous pinacol rearrangement thermally induced in the solid state (Sekiya et al., 2000).

Structural commentary
The molecular structure of the title compound, (2), is illustrated in Fig. 1. The dioxolane five-membered ring has a twist configuration on bond O1-C2, with atoms O1 and C2 at distances of 0.314 (4) and À0.330 (3) Å above and below the plane through atoms O3/C4/C5. The dioxolane ring has bond angles and distances that are within ca 3 (using the larger s.u. values from the reported structures) of the values found in published X-ray structures (see for example: Rao & Hong Chan, 2014;Jones et al., 1998). The planes of the two phenyl substituents on the dioxolane ring are inclined to one another by 44.67 (13) . They and the diphenylethyl substituent are all cis to one another, in equatorial positions. The phenyl rings of the diphenylethyl substituent are inclined to one another by 68.16 (12) . There is an intramolecular C-HÁ Á Á interaction present involving one of the diphenylethyl rings (C91-C96) and an H atom of the phenyl ring in position 4 of the dioxolane ring (Table 1).

Supramolecular features
In the crystal, molecules are linked by weak C-HÁ Á ÁO hydrogen bonds, forming chains along [001]. The chains are linked by C-HÁ Á ÁO bonds and by type I C-HÁ Á Á interactions (Malone et al., 1997), forming sheets parallel to the bc plane (Table 1 and Fig. 2).

Database survey
The Cambridge Structural Database (Version 5.36, last update February 2015; Groom & Allen 2014) was searched for structures containing the dioxolane ring. As there are several thousand dioxolane entries in the database, we selected only entries with the ring atoms plus one H atom each on C2 and C5, which includes the present structure. This search generated 594 hits, with 2227 sets of ring conformational angles that were reduced to 770 after removal of duplicates. There were 28 structures, 4% of the total, that contained near planar dioxolane rings, defined as rings where all torsional angles were less than 16 . Five-membered dioxolane rings have been described as 'puckered envelopes', 'half-chair' or 'twisted'. Arbitrarily broad criteria were used for envelope or twist conformations. Structures were identified as envelope when one torsional angle was less than 10 and at least 10 less than the remaining angles, or twist when two torsional angles were below 20 , with less than 10 difference between them. In this way, all of the remaining structures could be classified as envelope (447 structures, or 58%) or twist (295 structures, or 38%). The envelope flap was most often one of the ring oxygen atoms (309 structures). In the twist structures, the twisted bond was usually either of the O1-C2 type (145 examples), as in the structure described in this paper, or of the O1-C5 type (123 examples). Of the twist structures, there were 39 like the present structure (2), close to an ideal symmetric twist configuration, where the two smallest torsional angles are within 3 of each other.
The wide variety of dioxolane ring conformations found in the structural literature reflects well the conclusion from our own calculations (see: Sections 5 and 6 below) as well as in Willy et al. (1970), that the dioxolane ring is highly flexible. Table 1 Hydrogen-bond geometry (Å , ).

Figure 1
The molecular structure of the title compound, (2), with atom labeling. Displacement ellipsoids are drawn at the 50% probability level. One of the H atoms on the methyl group C10 was omitted for clarity.

Figure 2
A view in projection along the a axis of the crystal packing of the title compound, (2). The C-HÁ Á ÁO hydrogen bonds are shown as double dashed lines.

Density functional analysis
A B3LYP/6-311+G(d,p) density functional calculation (Spartan, 2006) of the present molecule in the gas phase shows minimum energy for a twist configuration with similar torsional angles to those in the structure presented here. A calculation where H atoms replace phenyl and diphenylethyl substituents on the dioxolane ring suggests that the large phenyl rings have little effect upon the ring conformation (Table 2).

Conformational analysis of 1,3-dioxolane rings
No organic five-membered ring is exactly planar because flat rings would have eclipsed C-C bonds that can have considerable torsional strain. Five-membered rings are usually identified as envelope or half-chair with more or less distortion. The planar part of the ring is described by a least-squares fit of three or four atoms in the ring, or by the torsional angle formed by four contingent atoms in the ring. When only one of the remaining atoms is a significant distance from the plane, this conformation is described as an 'envelope'. The non-planar atom defines the flap of the envelope. The torsional angles of an ideal half-chair configuration have two small angles of a given sign, two medium angles of opposite sign, and a single large angle of the same sign as the first. The Database Survey reveals that any atom in the ring can be the flap atom. When two atoms (one up, the other down) are a significant but different distance from the plane through the other three atoms, the conformation can be described as 'distorted envelope'. When two atoms have equal significant distances in opposite directions from the plane, the ring conformation can be described as 'twist', as shown in Fig. 3. A full range of conformations from ideal envelope to ideal twist is obtained with various substituents on the ring because the various ring conformations do not differ substantially in conformational energy (Willy et al. 1970).
Dioxolane is the title compound (2). The phenyl and diphenyl substituents are replaced by H atoms in column two, and CH 3 groups in column three.
shows a pattern of dihedral angles similar to the near-perfect twist found in the present crystal structure, shown in column five, where O1-C2 is the twisted bond. Fig. 5 above offers two views of the density functional theory (DFT) optimized structure. The pattern shown in column three (Ring with CH 3 groups), has a much larger (O3-C4-C5-O1) torsional angle. The calculated conformation is still that of a twist, but the twist bond in this CH 3 model is C4-C5, not O1-C2. The best plane through O1-C2-O3 has C4 + 0.28 Å above the plane and C5 À 0.28 Å below the plane, giving the CH 3 model an ideal twist conformation. The DFT-optimized CH 3 -substituted structure is depicted in Fig. 6. Column four of Table 2 shows the DFT results for the title compound. The predicted conformation is similar to the conformation found in the crystal structure. The differences between columns four and five are presumably due to packing (intermolecular) forces present in the X-ray structure. The Spartan DFT calculations do not include intermolecular forces, but are calculations in the gas phase. Comparing distance and angle values in column two (dioxolane ring with only H atom substituents) to columns four and five (dioxolane rings with phenyl and diphenylethyl substituents) suggests these larger substituents have little effect upon the ring conformation. Fig. 7 views the title compound as a distorted envelope.

Synthesis and crystallization
A sample of (AE)-1,2-diphenyl-1,2-propanediol (Ciaccio et al., 2001) was recrystallized in 1-butanol, as well as 2-butanol and 1-octanol. The solutions were mildly heated to obtain saturated solutions, cooled to room temperature and layered over water in an open test tube. In attempts to better characterize the rearrangements that occurred, we also recrystallized the starting material at the reflux temperature of 1-butanol. Thin layer chromatography revealed that the non-acid-catalyzed pinacol rearrangement was substantially complete after 8 h, and that other unknown products were also present. The experimental density of a typical recrystallization product, determined by flotation, is 1.054 g ml À1 . The melting point range was 435.9-443.2 K. Redoing the experimental density and melting point with hand-selected crystals with the same morphology as the X-ray data crystal gave values of 1.203 g ml À1 and 445.4-448.4 K, respectively. Perspective views of the dioxolane ring with hydrogen atoms as calculated with Spartan. Left: viewed as a distorted envelope. Right: viewed as twist.

Figure 6
Perspective views of dioxolane ring with methyl groups as calculated with Spartan. Left: viewed as distorted envelope. Right: viewed as twist.

Refinement details
Crystal data, data collection and structure refinement details are summarized in Table 3. The H atoms were included in calculated positions and treated as riding atoms: C-H = 0.96-0.98 Å with U iso (H) = 1.5U eq (C) for methyl H atoms, U eq (C) for methine H atoms, and 1.2U eq (C) for other H atoms.

Computing details
Data collection: CAD-4 EXPRESS (Enraf-Nonius, 1994); cell refinement: CAD-4 EXPRESS (Enraf-Nonius, 1994); data reduction: Data reduction followed procedures in Corfield et al. (1973); Data were averaged with a local version of SORTAV (Blessing, 1989); program(s) used to solve structure: Superflip (Palatinus & Chapuis, 2007); program(s) used to refine structure: SHELXL97 (Sheldrick, 2008); molecular graphics: ORTEPIII (Burnett & Johnson, 1996); software used to prepare material for publication: SHELXL97 (Sheldrick, 2008). where P = (F o 2 + 2F c 2 )/3 (Δ/σ) max < 0.001 Δρ max = 0.16 e Å −3 Δρ min = −0.18 e Å −3 Special details Geometry. All e.s.d.'s (except the e.s.d. in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell e.s.d.'s are taken into account individually in the estimation of e.s.d.'s in distances, angles and torsion angles; correlations between e.s.d.'s in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell e.s.d.'s is used for estimating e.s.d.'s involving l.s. planes. Refinement. In the monoclinic unit cell, the a and c axes are of very similar lengths, so that before data collection commenced, it was important to check that the Laue symmetry was indeed 2/m and not mmm. This was accomplished by temporarily transforming the cell to orthorhombic axes, and collecting all 8 forms of the (orthorhombic) 111 and 222 reflections. In each case, the 8 forms clearly split into two different sets of four, verifying the monoclinic symmetry.