Coupling Constants Stereoelectronic Effects: Perlin Effects in Thiane-Derived Compounds

: Stereoelectronic effects in thianes and thiane-de-rived sulfoxides, sulfones, sulfilimines, and sulfoximines were investigated by measuring 1 J C,H coupling constants and by identi-fication of normal and reversed Perlin effects, i.e., of differences in the coupling constants for equatorial and axial C–H bonds in the methylene groups of six-membered rings. The Perlin effects were correlated with results from natural bond orbital (NBO) The data are in full agreement with published data. for T. R. (LU 835/13-1) and the HGF BIFTM for financial support.


Introduction
The stability, conformation, and reactivity, as well as various physical and in particular spectroscopic properties are significantly influenced by stereoelectronic effects. [1] We have investigated these effects especially in sulfur-based functional groups, inter alia in sulfides, sulfoxides, sulfones, and in their respective α-anions. [2] Stereoelectronic effects have, e.g., a strong influence on 1 J C,H coupling constants and these can thus be used to quantify the underlying stereoelectronic interactions. Perlin and Casu [3] observed in tetrahydropyrans (actually in carbohydrates) that equatorial hydrogens next to the oxygen show a larger 1 J C,H coupling than axial hydrogens. This so-called normal Perlin effect was attributed to an n O → σ* C,Hax interaction weakening the axial C-H bond. [4] A reversed Perlin effect is observed in 1,3-dithianes: At position C-2, where the influence of two sulfur atoms is active, the 1 J C,Hax coupling is larger than the 1 J C,Heq coupling. This was explained by the relatively poor donor ability of the sulfur's lone pair and by the most relevant 1 analyses. NMR experiments were performed with conformationally restricted dimethyl-or tert-butyl-substituted derivatives, while the parent compounds were used for calculations. It turned out that the coupling constants are not only strongly influenced by stereoelectronic interactions with antiperiplanar C-H, C-C, C-O, and C-N bonds, but by the s character of the respective C-H bonds' carbon orbital as well.

Results and Discussion
3,5-Dimethylthiane was prepared starting with diethyl methylmalonate (28), which was deprotonated and added to methyl methacrylate (Scheme 1). The resulting triester was hydrolyzed and decarboxylated yielding dicarboxylic acid 29 as a mixture of diastereoisomers. [8] Anhydride formation and basic equilibration furnished cis-dimethyl-substituted substrate 30, [9] which was reduced with lithium aluminium hydride to yield diol 31, [10] activated, and reacted with sodium sulfide [11] to thiane derivative 1. 4-tert-Butyl-substituted thiane 9 was prepared from 4-tertbutyl-cyclohexanone (32), which was subjected to a double aldol condensation to yield 33 (Scheme 2). Ozonolysis with oxidative workup furnished a dicarboxylic acid. Esterification to 34 [12] and reduction gave the diol 35 (R = tBu) [13] which was again activated and reacted with sodium sulfide [11] to yield the tert-butyl-substituted thiane 9. Functionalization of thianes 1 and 9 was achieved with proven methods (Scheme 3). Oxidation with ozone yielded the equatorial sulfoxides 2 and 10, respectively, with good yields. [14] The axial sulfoxides (3 and 11) were obtained with a known protocol [14] by oxidation with tert-butyl hypochloride, albeit with quite poor yields. Excellent yields were observed in the preparation of sulfones 4 and 12, which was achieved with potassium permanganate. [15] The sulfilimines 5, 6 and 13, 14, respectively, were accessible by reaction of the respective thianes with chloramine-T (TsNClNa) [16] and subsequent separation of the isomers by medium pressure liquid chromatography (MPLC). Since only small fractions of the diastereomeric mixtures were separated, no reasonable yield can be given for these reactions. Reaction of the equatorial sulfoxides 2 or 10, respectively, with N-tosyliminobenzyliodinane (PhINTs) and copper(II) triflate as catalyst [17] yielded the equatorial sulfoximines 8 and 16, respectively, while the axial sulfoximines 7 and 15 were accessible from the axial sulfoxides 3 and 11. Scheme 3. Functionalization of thianes 1 and 9. Conditions: a) O 3 (2: 69 %; 10: 47 % [14] ); b) tBuOCl (3: 10 %; 11: 18 % [14] ); c) KMnO  The unsymmetrical nature of 2,4-dimethylthiane (17) and its derivatives prevented a comparably simple synthetic approach. 4-Methylthiane (36) is accessible from diol 35 (R = Me) by the proven activation and reaction with sodium sulfide (Scheme 4). [11] Oxidation with potassium permanganate [15] furnished the respective sulfone 37, which could be deprotonated with butyllithium and methylated with methyl iodide. The pure isomer 19 was obtained after crystallization. Oxidation of thiane 36 with ozone gave the equatorial sulfoxide, which could similarly be methylated to furnish sulfoxide 18. Reduction to the respective thiane 17 was achieved with phosphorus pentasulfide, [18] where this reaction was performed with an analytical sample in deuterated chloroform. three sets for the 3,5-dimethylthiane-derived compounds 1-8 in Figure 1, for 4-tert-butylthiane-derived compounds 9-16 in Figure 2, and for the 2,4-dimethylthiane-derived compounds 17-19 in Figure 3. 1 J C,H coupling constants are here given as green data points with error bars for every C-H bond of the thiane rings. Perlin effects for methylene groups are given as vertical blue bars, where an upward bar indicates a (normal) Perlin effect ( 1 J C,Heq -1 J C,Hax > 0), while a downward bar denotes a reversed Perlin effect ( 1 J C,Heq -1 J C,Hax < 0). Numeric values for   The following general trends can be deduced for the investigated compounds: 1) Larger coupling constants are generally observed for the α positions.
3) Sulfoxides with equatorial S=O bonds and sulfilimines and sulfoximines with equatorial S=N bonds show similar patterns of the Perlin effects. Similar patterns are furthermore observed for sulfoxides with axial S=O bonds and for sulfilimines and sulfoximines with axial S=N bonds. 4) Virtually no Perlin effect is observed for the α positions of sulfoxides with equatorial S=O bonds and for sulfilimines and sulfoximines with equatorial S=N bonds. 5) Sulfilimines and sulfoximines with axial S=N bonds show a reversed Perlin effect for their positions. 6) All substrates show only negligibly differing coupling constants at the 4-positions.
7) The conformationally constraining substituents (methyl and tert-butyl groups, respectively) seem to have no significant influence on the Perlin effects. Comparable carbon positions show comparable Perlin effects.
These trends are discussed in the next section together with results from natural bond orbital (NBO) analyses.

NBO Analyses
Already Alabugin [5e] and Juaristi [5f ] concluded from their investigations that there is no obvious and simple correlation between resonance energies obtained from NBO analyses and coupling constants. Contreras et al. examined the influence of stereoelectronic effects on coupling constants, [19] where they gained a much deeper insight into the theoretical foundations. They could explain both the missing of correlations and some observed trends. They divided Fermi contact interactions (as the dominant coupling mechanism) into orbital contributions of occupied and unoccupied LMOs (localized molecular orbitals). This led to contributions to the coupling constant of a C-H bond from the respective σ orbital (J b , with b: bond), from the respective σ* orbital (J ab ; ab: anti bond), and from further bonds at the coupling atoms (J ob ; ob: other bond). Contreras elegantly took advantage of model compounds, in which the "other bonds" are equivalent due to symmetry. The influence of "other bonds" is easily understood: Altering of the s character in an "other bond's" hybrid orbital by resonance has an immediate influence on the hybridization of the respective atom's other bonds, since there is a total of only one 2s orbital for every carbon. As the s character is of relevance for the Fermi contact, this must have an influence on the coupling constants. The subtle interplay of hybridization and hyperconjugation has similarly been reported in other systems and is of relevance e.g. in the blue-shifting hydrogen bonding. [20] In the light of these findings we used multiple linear regressions to test for a random number of compounds, whether there is a correlation, when resonance energies [E(2) values], occupation numbers of the bonding and of the antibonding orbitals (for the respective bond and for "other bonds") are considered. Nevertheless, satisfying linear correlations could be observed for neither of these combinations. The occurrence of saturation, when several effects coincide, has already been presumed by Juaristi et al. [5f ] Since the general trends of the coupling constants turned out to be not significantly affected by the methyl or tert-butyl groups, respectively, we used simplified compounds for computational studies, in which methyl or tert-butyl groups were omitted. Compounds with tosylimino groups were calculated in a conformation, which had been determined as the minimum conformation in the presence of the conformationally constraining substituents. We calculated bond lengths of the C-H bonds, s characters of the hybrid orbitals at the carbons, and summarized resonance energies [E(2) values] of all interactions, in which the respective C-H bonds are acting as donors or acceptors, respectively (using a threshold of 0.1 kcal/mol). The resulting sums immediately reveal, whether the considered bond is essentially acting as a donor or as an acceptor. In addition, resonance energies (NBO deletion energies E del ; Supporting Information) were calculated for all antiperiplanar donor/ acceptor pairs. 1 J coupling constants for the parent compounds 20-27 were additionally calculated for comparison (Supporting Information). The highest observed coupling constants of C-H bonds in α positions come along with high s characters of the carbons' hybrid orbitals used to build up the respective C-H bond. The p character of the carbons' hybrid orbitals in the adjacent C-S bonds is increased, most probably to allow for a better overlap with orbitals at the sulfurs to compensate the increased bond lengths. This effect is somewhat more pronounced in the S-oxidized substrates, since the respective C-S bonds are here more polarized towards the S atoms (Table 1). [21] Table 1. Bond lengths and s characters in thiane derivatives.
Thianes show reversed Perlin effects in α and positions (albeit hardly pronounced), as it has similarly been observed for the thoroughly investigated 1,3-dithianes, [5e] where this has been explained with a significant σ C,S → σ* C,Heq interaction. [5b,d, 23] Evaluation of the NBO analyses showed that further parameters have to be considered. The bond length of C-H bonds is influenced by two effects: When a σ bond acts as donor, the hybridization is changed; the s character at the carbon is reduced. [21] If σ* is an acceptor, the bond is weakened and the bond length is increased. Both effects can have an influence on the respective coupling constants. Investigation of thiane (and of further thiane derivatives) revealed that coupling constants in the α positions are mainly influenced by σ C,H donor interactions. The axial C-H bonds in the α positions are weakened by n S → σ* C,Hax interactions and thus elongated in comparison with the respective equatorial C-H bonds. However, the 1 J coupling constants of the axial C-H bonds are larger due to a smaller s character of the equatorial C-H bonds, which here is not a result of a distinct σ C,H → σ* S,C interaction. The resonance energy of this interaction actually is significantly smaller than that of a σ C,H → σ* C,C interaction. Nevertheless, the two possible stereoelectronic interactions, in which the equatorial C-H bond acts as donor (σ C,Heq → σ* S,C and σ C,Heq → σ* C,C ) exceed the single donor interaction of the axial C-H bond (σ C,Hax → σ* C,Hax ). The axial C-H bonds in the positions act as donors in two interactions, but the σ C,Heq → σ* C,S stereoelectronic effect of the equatorial C-H bond is dominating the outcome at this position. The polarization of the C-S bond makes it a significantly better acceptor than a C-C bond, which itself is a better acceptor than an S-C bond. [1f ] Homohyperconjugative interactions like the so-called homoanomeric effect (n S → σ C3,H ), which have been proposed by Alabugin et al. for thiane and other six-membered heterocycles, [22] might have a small influence on the Perlin effect in position of the thianes [n S (p-type) → σ C3,Heq : E del = 2.10 kcal/mol]. Nevertheless, this type of interaction is actually only mentionable for the p-type lone pair of the parent thiane and is much less (E del < 0.3 kcal/ mol) in the sulfoxides and sulfilimines, where the sulfur's lone pairs have a pronounced s character. The respective values are given in the Supporting Information.
An essentially similar pattern for the axial C-H bonds can be deduced for the sulfoxides and sulfilimines with axial S=O or S=N bonds. The σ C,Heq → σ* S,C interactions are even weaker as for the respective thianes. The σ* S,C bonds in sulfoxides are higher occupied due to interaction with n O orbitals and thus have reduced acceptor abilities towards further donors. [2m] A hardly pronounced reversed Perlin effect at the thiane's α positions thus turns into a weak normal Perlin effect in the equatorial sulfoxides. Since the thiane's position is governed by a σ C,Heq → σ* C,S interaction, the change in the acceptor ability of the C-S bond has an even stronger effect. The weak reversed Perlin effect (in the thianes) thus turns into a much more pronounced normal Perlin effect.
A strong normal Perlin effect is observed for the α positions in axial sulfoxides and sulfilimines due to the high acceptor ability of the σ* S,O and σ* S,N orbitals. A further contribution might arise from n O/N → σ* C2,Hax interactions (Figure 4a), which are analogous to the homoanomeric effect (vide supra). [22] The donor ability of the oxygens' lone pair in direction of the C2-H ax bond is here more pronounced than that of the nitrogens' lone pairs (cf. Supporting Information), since the latter can additionally interact with the respective tosyl groups. This leaves smaller shares for the C-H bonds. The σ C2,Hax → σ* C3,Hax interaction is much weaker than for the equatorial analogues, while the σ C3,Hax → σ* C2,Hax stereoelectronic effect is more pronounced. The former interaction results in an increased dipole moment, while the latter would reduce it (Figure 4b).
The reversed Perlin effects at the positions of sulfilimines and sulfoximines with axial S=N bond cannot be explained sufficiently with the available data. The equatorial C-H bond is somewhat longer than the axial bond, but this cannot be traced back to specific orbital interactions. Nevertheless, an extremely weak resonance energy for the σ C3,Heq → σ* C2,S interaction in the sulfoximine with equatorial S=N bond is notable. An insignificant deletion energy of 0.07 kcal/mol was determined for this interaction. The σ* C2,S orbital is significantly occupied by interaction with n O and n N orbitals, making it a less effective acceptor for the interaction with other donor orbitals. It should be noted that a much higher E(2) energy (4.8 kcal/mol) was obtained for the σ C3,Heq → σ* C2,S interaction, being an example that E(2) energies do not consider competing interactions and are thus less valuable in the explanation of resonance effects than deletion energies.
Comparatively weak normal Perlin effects are observed for the α and the positions in the sulfones. The relevant interactions seem to have a balanced overall effect. The coupling constants at the γ positions of the investigated compounds show insignificant deviations, making long-range orbital interactions less likely. Small differences might be due to slightly differing local orbital interactions resulting from the different molecular dipoles.

Conclusion
Perlin effects observed in thiane-derived compounds can be qualitatively explained with the data obtained from NBO analyses, although it turned out that no direct correlation between coupling constants and calculated resonance energies is possible. It is noteworthy to emphasize that the donor ability of the sulfur's lone pairs is commonly underestimated. The reversed Perlin effects in thianes are in fact not (only) due to a better donor quality of the S-C bond in comparison with the n S lone pair but are mostly due to the higher acceptor property of the σ* S-C orbital. The equatorial C-H bonds in thianes are acting as donors in this stereoelectronic interaction. This is in contrast to the respective tetrahydropyran (and 1,3-dioxane) derivatives, where the axial C-H bonds are acting essentially as acceptors for the oxygen's lone pair electrons and thus give rise to a normal Perlin effect. Again, it became obvious in this investigation that experimental findings are easily misinterpreted without consideration of NBO analyses.
When one and the same orbital takes part in competing interactions, it has again to be mentioned, that meaningful data of NBO analyses are only obtained from the deletion energies (E del ). The resonance in these cases is usually overestimated, when only E(2) energies are considered.

Experimental Section
NMR spectroscopic Investigations. 1 J C,H coupling constants of the thiane-derived compounds 1-19 were measured on a Bruker Avance III HD 500 MHz spectrometer using CLIP-HSQC experiments [24] and analyzed using the Topspin software package. [25] CLIP-HSQC spectra result in clean inphase doublets in the directly detected dimension, so that accurate coupling constants can be determined without further phase correction. Spectra were acquired using broadband BEBOP excitation, [26] BIBOP inversion, [27] and BURBOP refocusing pulses. [28] When a signal overlap obscured the coupling constants, we used ω1-iINEPT experiments with BIP inversion pulses during the BIRD-element [29] for clarification. [30] Number of scans as well as acquisition times and spectral widths were optimized for each compound individually. In all cases, digital resolution in the dimension with coupling evolution was below 0.1 Hz for CLIP-HSQC experiments and below 1.0 Hz for the ω 1 -iINEPT experiments. Due to highly symmetric multiplets and sufficient chemical shift difference of coupling partners second order contributions could be neglected in most cases. The individually estimated experimental errors of the coupling constants were generally on the order of the digital resolution, sometimes even below (see Figure 1-3).

Quantum Chemical Calculations.
All structures were optimized at the B3LYP [31] /6-311++G(d,p) [32] level by using the Gaussian 09 software package. [33] Coupling constants were calculated with the GIAO (gauge-including atomic orbitals) method [34] at the same level. The NBO 3.1 program for natural bond orbital (NBO) analyses [35] was used as implemented in Gaussian 09.
Synthetic Procedures -General. Compounds 10, [14] 11, [14] 31, [10] 35 [13] and 36 [11] were prepared according to published procedures. Tetrahydrofuran (THF), Et 2 O and pentane were distilled from sodium benzophenone ketyl radical prior to use and CH 2 Cl 2 was distilled from CaH 2 . All moisture-sensitive reactions were carried out under oxygen-free argon using oven-dried glassware and a vacuum line (Schlenk technique). Ozone was generated with an ozone generator 300.5 (Erwin Sander Elektroapparatebau) from dry air. Flash column chromatography was carried out using Merck silica gel 60 (230-400 mesh) and thin-layer chromatography was carried out by using commercially available Merck F 254 pre-coated sheets. Spots were detected by fluorescence quenching and staining in an iodine chamber. Medium pressure liquid chromatography (MPLC) on silica gel (LiChroprep Si 60 columns from Merck) was performed with a Laboprep MPLC pump and a Latek UVIS 200 detector. NMR spectra were recorded on Bruker Avance AV 300, Bruker Avance 400, or Bruker Avance III HD 500 spectrometers. 13 C NMR spectra were recorded with broad band decoupling and signals were assigned by HSQC experiments. The spectra were calibrated using the residual solvent signals. IR spectra were recorded on a Bruker FT-IR spectrometer "Alpha" using ATR on diamond. EI and FAB mass spectra were recorded with a Finnigan MAT-95 and ESI spectra were recorded with a Q Exactive Orbitrap (Thermo Fisher). Melting points were measured with an Optimelt MPA100 apparatus and are not corrected.