Several core-substituted VBX reagents 1 were synthesised according to literature methods (Stridfeldt et al., 2016; Boelke et al., 2017), Scheme 1A). The synthesis of novel VBO reagents 2 started from anilines 8, which underwent a Friedel-Crafts reaction to access the amino benzyl alcohols 9 (Scheme 1B). A subsequent Sandmeyer reaction produced the required iodoarenes 10 in good yields. (Amey and Martin, 1979). The benziodoxole core was formed through oxidative chlorination and hydrolysis, without isolation in between the steps, to form hydroxy-BO 11. The vinyl moiety was introduced from the corresponding boronic acid using TMSOTf and pyridine (Boelke et al., 2017) to form VBO 2 (yields from 10 given in Scheme 1B). It should be noted that ortho-substituted compound 2f was incredibly unstable, making isolation and analysis difficult (See the Supplementary Material for details).

Variations of the side-arm were next investigated to obtain novel compound classes for vinylation. The synthesis of VBZ 3 proceeded in good yields from 2-iodophenyl N-tosylbenzamides 12a-c, using our one-pot method developed for VBX (Stridfeldt et al., 2016) with mCPBA/triflic acid and (E)-styrylboronic acid (Scheme 1C). To our surprise, reactions with the ortho-methyl-substituted substrate 12d behaved differently, and resulted in the formation of the novel compound class VBXI 4, which has an I-O hypervalent bond instead of the expected I-N bond. It appears that the sterical congestion caused by the ortho-methyl group promotes formation of the BXI core as opposed to the BZ core. Products 5 were obtained from 2-iodophenyl sulfonic acids 12d-e in moderate yields due to incomplete conversion of the starting material.

For the sake of comparison to their cyclic counterparts, a series of vinyliodonium salts 6 with ortho functionalities were also synthesised from the iodoarenes 13 (Scheme 1D). Our one-step method (Stridfeldt et al., 2016) without the basic workup was used to obtain these compounds in good to high yields. Interestingly, when the one-pot synthesis of VBX⁶ 7 was attempted, product formation alongside an inseparable impurity was observed. (See the Supplementary Material for details) However, when isolated compound 6e was treated with an aqueous basic solution, 7 could be isolated in good yields with high purity. This strategy was also attempted for the synthesis of VBX⁷ from 6d, but was unsuccessful (See the Supplementary Material for details).

The S-vinylation protocol developed by our group was used to evaluate the vinylating reagents, as this reaction had already proved sensitive to the VBX core structure. (Castoldi et al., 2020b). 4-Bromothiophenol was thus vinylated with reagents 1-7 to provide thioether 14 with vinyl iodide 15 sometimes formed as side-product (Scheme 2). The result obtained with the novel VBX reagent 1f followed the trend in the original work, (Castoldi et al., 2020b), in which reagents with electron-donating groups (EDG) gave higher yields than those with electron-withdrawing groups (EWG). VBO reagents 2 performed markedly worse, with yields of 14 ranging from 25% to 51%. VBZ 3 behaved similarly to VBX, with 3c providing the highest yield of 90%, whereas VBXI 4 and VBT 5 gave significantly lower yields, which could be due to solubility problems. E/Z ratios were recorded in of each of these reactions, but there was no observed trend with regards to the reagent used (See the Supplementary Material for details).

Additionally, the compounds with ortho-substituents performed worse than their counterparts (1a vs 1g, 2a vs 2f). Due to the rapid decomposition of those reagents during isolation and analysis, we hypothesise that the ortho-substituents increase the reactivity of these compounds, especially since the majority of product in these reactions were the vinyl iodide 15. Indeed, ortho-substitution in hypervalent iodine compounds has earlier been reported to cause considerable reactivity changes. (Guilbault and Legault, 2012; Malmgren et al., 2013; Abazid and Nachtsheim, 2020). Additionally, VBXI 4 produced a very poor yield of 14 in comparison to its analogues VBZ 3, possibly due to its BXI core. Similar to what was reported with acyclic vinyliodonium salts in the original S-vinylation paper, (Castoldi et al., 2020b), reagents 6 provided little product, with 15 again being the major product in these reactions. VBX⁶ provided a much lower yield than VBX, indicating the importance of the 5-membered ring for the application of this reagent.

To evaluate how the core substituents and side-arms influenced the structure, we collected single crystal X-ray diffraction data on selected compounds (Figure 2). Much of the crystal structure data in the literature focusses on vinyliodonium salts, (Hinkle and McDonald, 2002; Ochiai et al., 2007; Yoshimura et al., 2021), and varied substitution patterns on the vinyl groups of VBO (Wu et al., 2016; Wu J. et al., 2017; Wu et al., 2019; Ding et al., 2020; Pisella et al., 2020; Chai et al., 2021; Laskar et al., 2021; Wu et al., 2023) and VBX. (Stridfeldt et al., 2016; Caramenti et al., 2019; Tessier et al., 2019). On the other hand, there appears to be no crystal structure investigations on the effect of cyclic vinylation reagents with different side-arms and core-substituents, as well as non-covalent interactions in ortho-substituted iodonium salts, on reaction outcome.

We evaluated the effect of the side-arm by comparison of core-unsubstituted VBX 1a, VBO 2a, VBZ 3a and VBT 5a, which showed very similar C₁-I bond lengths, 2.127, 2.118, 2.127 and 2.112 Å respectively. The same is also true for C₃-I bond lengths, as they measured at 2.100, 2.103, 2.140 and 2.128 Å respectively. However, X-I bond lengths were drastically different. 1a and 3a were somewhat similar with bond lengths of 2.510 and 2.581 Å respectively, whilst 2a had a shorter bond length (2.346 Å) and 5a has a longer bond length (2.649 Å). This is perhaps indicative of the increased trans effect caused by this functional group. (Ochiai et al., 2006). Additionally, measured X-I-C₃ bond angles showed that all compounds expressed a T-shaped conformation with 1a having the smallest angle, 165.9°, and VBO 2b the largest, 170.9°. The X-ray crystal structure of VBX⁶ 7 showed a strained 6-membered ring in the side-arm, with similar bond lengths and hypervalent bond angle with VBX 1a. However, 7 has a C₂-C₁-I-C₃ bond angle of 36.4°, which is far higher than that of 1a (4.3°).

To ascertain whether core-substituents made a measurable difference on any structural properties, several analogues of each class of vinylating reagent were also crystallised. Generally, the same trends followed within each class of compounds. Interestingly, the crystallographic data of ortho-substituted VBXI 4 showed a markedly lower X-I-C₃ bond angle of 155.7°, which is the lowest angle of any hypervalent iodine vinylating reagent in the literature. Additionally, 4 has a C₂-C₁-I-C₃ bond angle of 38.5°, which is far from the more idealised angle of 1°–8° for the other compounds. These two measurements show that the ortho-methyl substituent induces sufficient steric strain to disrupt the hypervalent bond and ultimately leads to the formation of an I-O bond, as opposed to the I-N bond found in VBZ 3. We hypothesise that this key difference contributes to the reagent’s poor reactivity under the S-vinylation conditions. Furthermore, this characteristic likely contributes to their unstable and over-reactive nature as seen in Scheme 2. Interestingly, iodonium salt 6a had similar C₁-I (2.123 Å) and C₃-I (2.073 Å) bond lengths to other compounds. Though, it has a much longer X-I bond length (2.673 Å), which is unsurprising considering the methyl ester ligand is not covalently bound, but it is markedly smaller than the X-I bond length of 6f (The crystal structure data of 6f is included as comparison to 6a; we did not use 6f in other parts of the study) (2.879 Å). This shows that whilst not having a covalently bound group will affect the structural properties, a much more significant effect will be observed when non-ligating substituents are used in the ortho position to the iodine.

Next, the possible correlation between structural parameters and reaction outcome was investigated (Figure 3). Firstly, the X-I bond lengths were plotted against the yield of S-vinylation, showing an upward slope from 2.341 Å (2e) to 2.581 Å (3a), which was proceeded by a downward slope towards 2.673 Å of 6a (ester-bound iodonium) (Figure 3A). Interestingly, 4 and 7 were outliers to this trend (circled, hollow diamond). Thus, reagents with X-I bond lengths of ∼2.55 Å represent a “sweet-spot” under these reaction conditions. Then the hypervalent bond angles (X-I-C₃) were plotted against reaction yields (Figure 3B). Within this parameter, it was observed that the higher the angle, and thus closer to the idealised 180°, the worse the reagent performed, with a peak of ∼165°. Again, reagent 4 was an outlier in this trend. These results show that there is indeed a link between these two structural parameters and the reaction yield, and that both electronic and steric factors influence the reaction outcome.

We also wanted to investigate how the reduction potentials of the reagents were affected by substituents of the aromatic ring core and the side-arm (Figure 4). Whilst there have been some reports of redox potentials of hypervalent iodine reagents in the literature, (Kokkinidis et al., 1989; Kokkinidis et al., 1991; Choi et al., 2015; Le Vaillant and Waser, 2017; De et al., 2019; Ramkumar et al., 2023), and even a computational study, (Radzhabov et al., 2020), there is currently no data on vinylation reagents and certainly no quantitative studies linking this parameter to reaction outcome.

To begin, we measured the potentials of each of the reagent used in this study. The higher the reduction potential (closer the value to 0 V (vs Fc/Fc⁺)), the more easily the reagent can be reduced. It was found that there was a tight range of reduction potentials within each class of reagent. VBX ranged from −1.33 V (1f) to −1.59 V (1c), which matches that EWG should make the reagent easier to reduce. Considering its strong likeness to 1a, VBX⁶ 7 had a very different potential of −1.61 V. Whilst VBZ 3a-c had slightly higher potentials between −1.32 and −1.48 V, VBXI 4 (−1.21 V) and VBT 5 (−1.21 and −1.17 V) had lower potentials. The lowest reduction potentials were measured for VBO 2, ranging from 2c (−1.98 V) to 2f (−1.68 V). As controls, the potentials of some vinyliodonium salts 6 were measured. Direct comparison of the uncyclized and cyclised analogues (6c vs 1e and 6e vs 6) showed that the iodonium salts were indeed much easier to reduce. This was, however, not the case for 6b (−1.16 V), which was very similar to its cyclised VBT counterpart 5b (−1.17 V).

Next, to ascertain whether there is a relationship between reduction potentials of the vinylation reagents and their yield in S-vinylation, the two were plotted against each of other (Figure 5). Firstly, we plotted the VBO, VBZ and only the EDG-substituted VBX reagents, as EWG-substituted VBX performed poorly in S-vinylation (Figure 5A). Reagents with lower reduction potentials were found to have a positive effect on the reaction yield. Secondly, we plotted the EWG-substituted VBX reagents, VBT reagents and iodonium salts (Figure 5B). In this case, lower reduction potential resulted in lowered reaction yield together with increased levels of vinyl iodide (see data in Scheme 2). This is interesting because the less reactive reagents (e.g., VBO) provided lower reaction yields, whereas the more reactive reagents (e.g., iodonium salts) gave low yields and more vinyl iodide. Furthermore, the results suggest that a reagent with a reduction potential between −1.3 and −1.5 V, represents the peak of this reaction, with potentials on either side being ultimately detrimental for the reagents. However, this could be due to the reaction itself being optimised on VBX 1a. Curiously however, if both plots are overlayed with the previously excluded reagents, it becomes clear that there are some outliers to this trend (green triangles, Figure 5C). These are VBX⁶ 7 and ortho-substituted VBO 2f and VBXI 4, and the results might reflect the lack of idealised T-shaped conformation or conjugation in those structures. Clearly the reduction potential does not account for the change in T-shaped conformation and steric factors very well. Overall, it appears that the reduction potentials can be a good signifier for the efficiency of the vinylation reagent in the S-vinylation under these reaction conditions, but further reaction optimisation could potentially alter the outcome. Finally, a fine correlation was observed between X-I bond lengths and reduction potentials (Figure 5D), signifying that X-I bond lengths could affect the reduction potentials greatly.