Comment

The physical, biological and nutritional properties of fatty acids are largely determined by the number, position and configuration of their double bonds. These determine the shape of the molecules, the way molecules can pack together in solid phases, monolayers, bilayers etc., and how individual molecules can interact with enzymes and receptors. Most natural unsaturated fatty acids have cis (Z) double bonds. Trans (E) fatty acids are present in dairy fats and are produced during the catalytic partial hydrogenation used in the production of hardened fats and during deodorization of commodity oils. The labelling of foods with trans content is increasingly required due to their undesirable nutritional properties. Alternative ways of producing hardened fats, such as interesterification or blending with fully saturated fats, and milder deodorization procedures, are being developed to reduce trans content. Trans fatty acids more closely resemble saturated acids in melting point and nutritional properties, sharing an essentially linear structure which allows closely aligned packing in condensed phases. In contrast, cis double bonds introduce a bend in the alkyl chain, making packing less stable and lowering the melting point. We have determined the structure of elaidic acid (trans-9-octadecenoic acid), (I), to enable a detailed comparison of a trans fatty acid with saturated and cis-unsaturated compounds.

Relatively few crystal structures of fatty acids are available, as good crystals are difficult to obtain, often being thin plates and often crystallizing in several polymorphs. Most monoenes have low melting points and polyenes are liquids at room temperature. The crystal structures of the following saturated and monoene C₁₈ fatty acids have been reported to date: stearic acid (octadecanoic acid) (Malta et al., 1971; Kaneko et al., 1990, 1994a,b), oleic acid (cis-9-octadecenoic acid) (Abrahamsson & Ryderstedt-Nahringbauer, 1962; Kaneko et al., 1997) and petroselinic acid (cis-6-octadecenoic acid) (Kaneko et al., 1992a,b). No trans-octadecenoic acid structure has been reported to date.

Elaidic acid (I) has an essentially linear alkyl chain, with the torsion angle between saturated C atoms close to 180° (Table 1). The C7-C8-C9-C10, C8-C9-C10-C11 and C9-C10-C11-C12 torsion angles are -118.8 (4), -179.9 (4) and 118.6 (4)°, respectively, resulting in the double bond being twisted across the mean direction of the alkyl chain in a skew', trans, skew conformation. The C1-C18 distance is 21.393 (6) Å, comparable with that in fully extended stearic acid structures (21.6 Å; Malta et al., 1971; Kaneko et al., 1990, 1994b). This contrasts with the cis-octadecenoic acids, where the molecules are bent and the C1-C18 distance is reduced to between 17.8 and 19.7 Å (Abrahamsson & Ryderstedt-Nahringbauer, 1962; Kaneko et al., 1997, 1992a,b).

In the crystal structure of (I), molecules related by inversion centres are linked by O-HO hydrogen bonds to form R₂²(8) dimers (Bernstein et al., 1995) typical of carboxylic acids (Table 2).

Figure 1 A view of (I), with displacement ellipsoids drawn at the 30% probability level.

Experimental

A commercial sample of (I) (Sigma, Poole, Dorset, UK) was re-crystallized from ethanol at room temperature. The crystals were composed of very thin stacked sheets which tended to be twisted. After many attempts, a crystal was found from which it was possible to obtain a data set.

Crystal data

C18H34O2 Mr = 282.45 Monoclinic, C 2/c a = 98.48 (2) Å b = 4.9381 (3) Å c = 7.1826 (8) Å = 92.570 (12)° V = 3489.4 (8) Å3 Z = 8 Dx = 1.075 Mg m-3 Mo K radiation Cell parameters from 3830 reflections = 3.3-27.6° = 0.07 mm-1 T = 120 (2) K Plate, colourless 0.20 × 0.18 × 0.01 mm

Data collection

Bruker Nonius KappaCCD area-detector diffractometer and scans Absorption correction: multi-scan(SADABS; Sheldrick, 2003)Tmin = 0.987, Tmax = 0.999 17532 measured reflections 3830 independent reflections 1679 reflections with I > 2(I) Rint = 0.150 max = 27.6° h = -126 126 k = -6 6 l = -9 9

Refinement

Refinement on F2 R[F2 > 2(F2)] = 0.102 wR(F2) = 0.256 S = 1.08 3830 reflections 182 parameters H-atom parameters constrained w = 1/[2(Fo2) + (0.0744P)2 + 7.8445P] where P = (Fo2 + 2Fc2)/3 (/)max = 0.001 max = 0.41 e Å-3 min = -0.35 e Å-3

Table 1 Selected torsion angles (°) C1-C2-C3-C4 -170.7 (3) C2-C3-C4-C5 177.0 (3) C3-C4-C5-C6 -176.6 (3) C4-C5-C6-C7 178.9 (3) C5-C6-C7-C8 -179.1 (3) C6-C7-C8-C9 -178.4 (3) C7-C8-C9-C10 -118.8 (4) C8-C9-C10-C11 -179.9 (4) C9-C10-C11-C12 118.6 (4) C10-C11-C12-C13 178.4 (3) C11-C12-C13-C14 -179.8 (3) C12-C13-C14-C15 179.9 (3) C13-C14-C15-C16 179.8 (3) C14-C15-C16-C17 -179.6 (3) C15-C16-C17-C18 179.5 (3)

Table 2 Hydrogen-bond geometry (Å, °) D-HA D-H HA DA D-HA O1-H1O2i 0.87 1.86 2.684 (3) 158 Symmetry code: (i) .

H atoms were treated as riding, with C-H(aromatic) = 0.95 and C-H(CH₂) = 0.99 Å, with U_iso(H) = 1.2U_eq(C), C-H(methyl) = 0.98 Å, with U_iso(H) = 1.5U_eq(C), and O-H = 0.87 Å, with U_iso(H) = 1.5U_eq(O). The O-bound H atom was allowed to ride at its position as determined from a difference map. Although the best crystal was selected from many crystallization attempts, the higher than usual values for R, wR and R_int may be a result of the crystal quality. The possibilty that the crystal was twinned was investigated but this did not give any significant results.

Data collection: COLLECT (Nonius, 1998); cell refinement: DENZO (Otwinowski & Minor, 1997) and COLLECT; data reduction: DENZO and COLLECT; program(s) used to solve structure: OSCAIL (McArdle, 2003) and SHELXS97 (Sheldrick, 1997); program(s) used to refine structure: OSCAIL and SHELXL97 (Sheldrick, 1997); molecular graphics: ORTEPII (Johnson, 1976) and PLATON (Spek, 2003); software used to prepare material for publication: SHELXL97 and PRPKAPPA (Ferguson, 1999).