Formation of 3,3,4-Trimethyl-1,7-dibromonorbornane-2-one: a Spectroscopic and Computational Study

The structure and origin of the major by-product in the synthesis of 8-bromocamphor from (+)-3,3,8-tribromocamphor has been confirmed using NMR, coset and single crystal X-ray analysis and DFT-level computational techniques.

One-and two-dimensional NMR spectroscopy of a byproduct isolated during the synthesis of (+)-8-bromocamphor 5 appeared to be consistent with (1R,4S)-3,3,4-trimethyl-7,7-dibromonorbornan-2-one 9. It was assumed that the by-product is formed during the reaction of the dibromo compound 7, but the results of a coset 7 analysis of possible rearrangement pathways from 3,3-dibromocamphor 7 to compound 9 challenged this structural assignment.
In the coset analysis the maximum number of rearrangement steps in a given sequence was limited to 13 and the permissible operations to: Wagner-Meerwein rearrangements (WM), 2,3-exo-(23x), 2,3-endo-(23e) and 6,2-(62) shifts. Within these limits, the analysis generated the four potential pathways summarised in Fig. 1 for formation of compound 9 (in its protonated form 9H + ) from the protonated dibromocamphor starting material 7H + ; the rearrangement was expected to be acid-catalysed, thus warranting the use of protonated species.
An examination of the alternative routes (2)(3)(4) to the cationic species 9H + (Fig. 1) revealed the same energetically unfavourable hydride and bromide shifts in each pathway. These observations raised doubt concerning the assignment of structure 9. Money and co-workers 3-6 had also isolated a by-product to which they assigned structure 10 using 1 H NMR data; this was subsequently supported by X-ray crystallographic analysis. 11 Re-examination of our one-and two-dimensional NMR data confirmed their consistency with structure 10. Thus, the 1 H NMR spectrum clearly indicates the presence of: three methyl singlets at 1.08, 1.23 and 1.39 ppm; multiplets characteristic of the 5-and 6-methylene groups; and a singlet at 4.22 ppm corresponding to the relatively deshielded 7-methine proton. The 13 C NMR spectrum revealed the requisite number of methine, methylene, methyl, quaternary and carbonyl carbon signals, while the HMQC and HMBC data confirmed the proton-carbon connectivities -all of which, superficially at least, are also consistent with structure 9! Single crystal X-ray analysis of the byproduct isolated in our study 12 confirmed it to be the same as the compound isolated previously by Money and co-workers, 5 viz. 3,3,4-trimethyl-1,7-dibromonorbornan-2-one 10 and not the isomeric system 9.
There remains, however, some disagreement about the mechanistic pathways followed in the transformation of 3,3-dibromo-  2 In order to explore the competing mechanistic proposals, we conducted a modelling study using the Accelrys DMol 3 DFT code in Materials Studio. Stable ground state structures could not be generated for either of the intermediates 13 or 14 in Pathway I. However, stable structures were located for the intermediates 16 and 18 in Pathway II. These species appear to be linked by a single transition state with a relatively low activation energy (7.59 kcal mol -1 ), implying that the Wagner-Meerwein rearrangement (16 → 17) Table 1 and illustrated in Fig. 2.
In our synthesis of 8-bromocamphor 5, there was no spectroscopic evidence for the presence of the by-product 10 in the reaction mixture until after the final, Zn dust/AcOH-mediated reaction step. Both the computational and experimental evidence thus indicate Pathway II, as proposed by Antkowiak and Antkowiak, 13 to be a more likely route to compound 10 than RESEARCH ARTICLE I.T. Sabbagh and P.T. Kaye, 31 S. Afr. J. Chem., 2020, 73, 30-34, <https://journals.co.za/content/journal/chem/>.  Pathway I, as suggested by Money and co-workers. 5 A combination of techniques, including coset, advanced one-and twodimensional NMR and theoretical analysis, has thus permitted confirmation of the structure of a minor, terpenoid rearrangement product 10 and provided support for a mechanism involved in its formation.

Computational Methods
Density functional calculations were conducted using the Accelrys DMol 3 DFT code in Materials Studio (version 2.2) 14 on LINUX-based Pentium IV PCs. All calculations involved use of the generalized gradient approximation (GGA) functional by Perdew and Wang (PW91) 15 and the 'double numerical plus polarization' (DNP) basis set: a polarized split valence basis set of numeric atomic functions which are exact solutions to the Kohn-Sham equations for the atoms. 16 Geometry optimizations were subjected to convergence criteria of threshold values 2 × 10 -5 Ha, 0.004 Ha/Å, 0.005 Å and 1 × 10 -5 Ha for energy, force, displacement and self-consistent field (SCF) density, respectively. All calculations employed a method based on Pulay's 17 direct inversion of iterative subspace (DIIS) technique to accelerate SCF convergence using, where necessary, a small electron thermal smearing value of 0.005 Ha.
Preliminary transition state geometries were obtained using the integrated linear synchronous transit/quadratic synchronous transit (LST/QST) method, 18 and then subjected to full TS optimization using an eigenvector following algorithm. Where necessary, these geometries were confirmed using intrinsic reaction path (IRP) calculations, based on the nudged elastic band (NEB) algorithm, 19 to map the pathways connecting the relevant reactant, transition state and product geometries. All structures identified as stationary points were subjected to frequency analysis, to verify their classification as equilibrium geometries (zero imaginary frequencies) or transition states (one imaginary frequency). The reported energies reflect Gibbs free energy corrections to the total electronic energies at 298.15 K and include zero-point energy (ZPE) corrections.