CUESTIONARIO DIRIGIDO A TALENTO HUMANO COAC

Calculations were carried out on a range of derivatives of N,N’-diphenylthiourea dioxide

19 in order to identify trends which could lead to the identification of isolable derivatives. A series of compounds containing electron-donating and -withdrawing groups was investigated and key data are summarised in Table 8.

The computational studies were carried out with 6-31G(d,p) basis set in the gas phase (instead of 6-311+G(3d,p) in water) so that the series could accommodate potentially

63 larger molecules in which the optimisations would converge in reasonable time (ca. 5 days). Thiourea dioxide 3 and N,N’-diisopropylthiourea dioxide 25 were also calculated at the same level in order to establish a reliable benchmark. It was decided that compounds with C-S bonds which were predicted longer, i.e. weaker, than N,N’- diphenylthiourea dioxide would be unsuitable targets. Compounds with C-S bond lengths similar to thiourea dioxide and N,N’-diisopropylthiourea dioxide were considered isolable. R C-S /Ǻ O-HN /Ǻ 1 2.28038 1.75243 2 2.20197 1.77668 3 2.17512 1.75301 4 2.15321 1.77285 5 2.11013 1.76465 6 2.08625 1.78247 7 2.02106 1.80324 8 2.04329 1.80486 9 H 2.01849 1.93878

Table 8. Calculated geometries of thiourea dioxide derivatives

RHN NHR SO2- O2N N N MeO Me2N Me Me Me Me

64 All of the analogues listed in Table 8 possess intramolecular hydrogen bonding, as described previously. The computational data show a marked trend with regard to the electronic factors affecting the C-S bond length. Steric factors are also thought to influence the C-S bond length.

The electron-poor derivatives, such as N,N’-di-p-nitrophenylthiourea dioxide (entry 1, Table 8), were predicted to have longer C-S bond lengths than the diphenyl derivative (entry 3). In contrast, electron-rich derivatives, such as the 1,3-bis(p-N,N’-

dimethylaminophenyl)thiourea dioxide (entry 5), were predicted to have shorter C-S bond lengths. Both nitro- and dimethylamino-derivatives have similar conformation i.e. similar overlap of the aromatic and amidine π-system, as shown in Figures 16 and 17. In addition, both analogues were predicted to have significantly different C-S bond lengths. Therefore, it appears that electronic factors influence the C-S bond in diaryl derivatives.

65

Figure 17. DFT predicted structure of 1,3-bis(p-N,N’-dimethylaminophenyl)thiourea dioxide

However, close inspection of the conformation of the predicted structures listed in Table 8 reveals that not all aromatic rings are coplanar with the amidine moiety. Only the pyrimidinyl analogue (entry 2) resulted in a conformer where both aromatic rings were co-planar with the amidine moiety (Figure 18).

66 The degree of overlap of the π-systems was greatly diminished when methyl groups were incorporated at the ortho-positions of the aniline ring. For example, N,N’-di-o-

tolylthiourea dioxide (entry 6) was predicted to have a shorter C-S bond length 2.08625 Å compared with the diphenyl derivative 2.17512 Å. N,N’-Dimesitylthiourea dioxide (entry 7) is predicted to have the shortest C-S bond 2.02106 Å, second only to thiourea dioxide (entry 9) 2.01849 Å.

Figure 19. DFT predicted structure of N,N’-dimesitylthiourea dioxide 31

The computational structure of N,N’-dimesitylthiourea dioxide 31 (Figure 19) demonstrated that the aromatic rings are not coplanar with the amidine moiety. Therefore, the aromatic π-system would appear to have little influence on the C-S bond. From our calculations, the diminished overlap of the aromatic ring π-system with the amidine π-system, probably due to steric repulsion, leads to dioxide analogues with shorter C-S bonds.

Overall, the computational data show that electronic and steric factors are involved in affecting the C-S bond length. From the data in Table 8, N,N’-dimesitylthiourea dioxide

67 was predicted to have the shortest C-S bond of the diaryl analogues and was therefore chosen as our next target.

The synthesis of N,N’-dimesitylthiourea dioxide 31 was successfully carried out as outlined in Scheme 39. The thiourea 30, synthesised from the isothiocyanate 29, was obtained following literature procedures using commercially available starting materials.128 _{The dioxide 31 was then isolated via the catalytic oxidation of the thiourea}

with hydrogen peroxide.

N N S N N MesNH2 Mes N C S MeCN, RT, 4 h 99% MesNH₂, toluene, ∆, 4 d Mes H N NH S Mes 26% H2O2, MoO2(acac)2 1,4-Dioxane, Et2O -5 °C, 90 min 60% Mes H N NH SO2 Mes 29 30 31

Scheme 39. Synthesis of N,N’-dimesitylthiourea dioxide 31

N,N’-Dimesitylthiourea dioxide is insoluble in water and chlorinated solvents, and gives the characteristic sulfinate asymmetric and symmetric stretching modes at 1104 and 1008 cm-1_{, respectively. The compound was incompatible with the dithionite test because it}

was insoluble in water and therefore tested negative for dithionite ions. The dioxide was found to be air stable for at least a few days but can be stored at room temperature, in the dark and away from moisture for ca. 3 months. NMR spectroscopic analysis revealed the

68 presence of apparent intramolecular hydrogen bonding as evident from the two sets of

ortho-methyl and para-methyl signals. The dioxide did not give a satisfactory elemental analysis (C: 65.32, H: 7.01, N: 7.99, S: 8.20%; theoretical composition C: 66.25, H: 7.02, N: 8.13, S: 9.31%) but was detected with high resolution LSIMS.

To conclude, it appears that N,N’-disubstituted thiourea dioxides are difficult to isolate. The structure of N,N’-diisopropylthiourea dioxide in the solid state and solution appears to demonstrate the contribution of intramolecular hydrogen bonding to the structure and probable stability of N,N’-disubstituted thiourea dioxides. The synthesis of cyclic thiourea dioxides with aqueous hydrogen peroxide proved more difficult. The three novel thiourea dioxides isolated in this work are air stable and the procedure for their syntheses was successfully repeated. An improved synthesis of the thiourea dioxides under non- aqueous conditions is outlined in chapter 5.

Characterisation has focused heavily on IR and elemental analyses, both techniques of which require sufficiently pure samples. Attempts to obtain 33_{S NMR spectroscopic data}

for all the dioxides gave very broad signals (due to the asymmetry of the sulfur centre in the dioxides) and could not be interpreted. The dithionite test has proved useful in identifying most of the dioxides generated in situ but is dependent on the water solubility of the dioxide.

Problems with characterisation are further amplified by the fact that the corresponding trioxide has been detected, especially for the cyclic derivatives, even when two equivalents of oxidising agent are used. The formation of the trioxide is probably due to oxidation of the dioxide or by other mechanisms e.g. disproportionation129, 130 which are observed in other S,S-dioxides (sulfinic acids).

There does, however, appear to be good agreement with computational predictions and synthetic work. The stability of the dioxides is largely influenced by the C-S bond length.

69 It is noted however that the calculations have not been developed to account for other factors which affect the decomposition of the dioxides.