Part 3. Triethylborane-air: a suitable initiator for intermolecular radical additions of S-2-oxoalkyl-thionocarbonates (S-xanthates) to olefins

Under carefully controlled conditions, the triethylborane-air combination proves to be an efficient radical initiator that allows intermolecular radical additions of S-2-oxoalkyl-thionocarbonates (S-xanthates) to olefins. Depending on both the structures of the xanthate and the olefin, the addition process can be achieved at room temperature or slightly higher.


Results and discussion
The pivotal experiments at the origin of this paper are depicted in Scheme 1. In the first experiment, 2.5 equiv. of Et 3 B were added to a mixture of xanthate 1a and 1-decene in dichloromethane under argon at 20°C. The stopper was then removed and air was allowed to enter the flask. After 1 h, purification afforded reduced starting material 1b as the only isolated compound (63%). Such a reactivity was not surprising in view of previous observations [3] and from the literature data. [9][10][11][12][13][14] Trapping with benzaldehyde gave aldol 1d (48%, Scheme 1) and thus confirmed that a boron enolate is a plausible intermediate in the reduction of compound 1a into 1b. One cannot put aside the possibility that the reduction of the transient α-acyl radical may also occur, to a minor extent, via a direct transfer from a hydrogen donor. [4,15] Scheme 1: Reactivity of 2-oxoalkylxanthates toward 1-decene in the presence of Et 3 B/O 2 : competition between addition and reduction.
Scheme 1 shows that xanthate 1a undergoes two main types of reactions. The group transfer reaction operates through a radical chain mechanism and affords the adduct 1c. [16] The reduction of compound 1a into 1b results from a bimolecular process in which Et 3 B is implicated not only in the generation of the α-acyl radical but also in the reaction with the latter, in a stoichiometric manner, to afford an intermediate boron enolate. Lowering the amount of Et 3 B would therefore minimise the premature unwanted reduction. This hypothesis was then tested. When Et 3 B (0.1 equiv.) was used in catalytic amounts, adduct 1c was isolated in a modest but remarkable 40% yield (Scheme 1), together with traces of compound 1b (<5%). We reasoned that a slow addition of Et 3 B would diminish more efficiently the unwanted reduction into 1b. On the other hand, slow addition of air would maintain a low concentration of the radical species and hence minimise the usual unwanted side reactions (dimerisations, abstractions...) that could hamper a clean addition process. Accordingly, as Et 3 B was added slowly with a syringe pump to a 0.4 M solution of xanthate 1a (0.6 mmol) and 1-decene (2 equiv.) in dichloromethane at 20°C, air was injected (10 mL/h) at the same time in the reaction medium with a second syringe pump [see Additional file 1]. The data reported in Table 1 show that, with this technique, diminishing the rate of addition of Et 3 B from 0.15 mmol/h to 0.03 mmol/h and increasing simultaneously the total amounts of Et 3 B (from 0.2 to 0.4 equiv.) resulted in a marked improvement. The yield of adduct 1c increased from 35 to 64%. At the same time, the amount of recovered 1a dropped from 30 to 11% and the reduction into compound 1b was totally suppressed.
We then turned our attention to the highly delocalised radicals derived from aromatic ketones 2a, 3a, and 4a. As anticipated, this represented one of the worst situations, as premature reduction to methyl ketone should be relatively fast when compared to intermolecular addition to an olefin. Experiments 7 and 8 validated this hypothesis. When xanthate 2a was reacted at 20°C with phenylvinyl dioxolane 13 in the presence of Et3B, no trace of the adduct 23a could be isolated, and only starting material 2a was recovered ( Table 2, entry 7). Under the same conditions, xanthate 3a also failed to add to acrolein diethylacetal 12 (entry 8). The only compounds that could be isolated were the starting material 3a (54%) and the reduced product 3b (12%). Obviously, the reaction conditions needed to be adjusted in order to favour the addition process with regard to both premature reduction of the starting material and useless degenerate reaction of the α-acyl carbon radical with its precursor. [8] We were delighted to observe that gently warming the reaction in refluxing dichloromethane (40°C) totally turned the course of the reaction. Thus, xanthate 2a added to olefin 13 in a fair 51% yield (entry 9). All the starting material was consumed and only 12% of acetophenone 2b were formed. Similarly, xanthate 3a, in the presence of  olefin 12, succeeded in giving adduct 24a (47%, entry 10) accompanied by some p-methoxyacetophenone 3b (26%). For these two substrates, comparison between experiments 7-10 showed a striking effect of the temperature on the outcome of the reaction: at 20°C no addition was observed but simply warming the reaction medium to 40°C ensured a clean intermolecular addition process. Under the same conditions, α-phenacyl xanthates 3a and 4a were also reacted with allylacetate (entries 11 and 12).
In both cases the corresponding adducts were the major products (30a and 31a, 49% and 45% yield, respectively), accompanied by some reduced starting materials (39% and 34% respectively). The astonishing effect of the temperature increase from 20 to 40°C noticed with the aromatic ketones also held for "normal" ketones, albeit to a less dramatic extent. Thus, xanthate 1a condensed with olefin 12 with a much higher yield (74%, entry 13) than at 20°C (25%, entry 6). Xanthate 1a also reacted with olefins 14 (entry 14) and 13 (entry 15) to afford adducts 25a and 26a in satisfactory 52% and 66% yields, respectively. Similarly, 1a condensed with olefins 9, 10, 11 and 15 to furnish adducts 19a, 20a, 21a, and 27a in 77%, 42%, 54% and 59% yield respectively (entries [16][17][18][19]. In the case of addition to vinyltrimethylsilane 10, the yield obtained at 40°C, lower than the one observed at 20°C, is clearly due to the volatility of vinyltrimethylsilane, as demonstrated by experiment 20 where the addition of more olefin (2.5 equiv.) during the reaction resulted in a marked increase of the yield (73%). It is important to note that, at 40°C, no trace of prematurely reduced starting material 1b could be isolated.
Nevertheless, some substrates were still refractory. Thus, at 40°C in dichloromethane, xanthate 5a failed to add to allylacetate and was recovered unchanged. However, when the reaction was performed in refluxing 1,2-dichloroethane (83°C), adduct 28a was isolated in an excellent 77% yield (  Adducts between xanthates and olefins  OMe and a' 2 , see below) constitute a body of fast but useless processes (degenerate reactions) [16] that preserve the radical character but do not let the system evolve. Route c is a relatively slow reaction when compared to the degenerate reactions or to reaction of Et• with A. Routes b and b' depend on the concentration of Et 3 B that can be controlled by maintaining a low concentration of Et 3 B. The addition to olefins (route c) is practically irreversible because of the formation of a strong C-C bond. However, the efficiency of route c, compared to routes b, b' and a 1 , is strongly linked to the structures of both xanthate A and olefin R 3 -CH=CH 2 , and can be dramatically modified by varying the reaction temperature. Fortunately, such an increase of reaction temperature enhances route c much more than route b.

Starting xanthates and olefins
For planar radicals, Rûchardt and Beckwith established that the C-H bond dissociation energy (BDE) for H-CXYZ compounds displays a good linear correlation with the measured α and β-proton ESR hyperfine splitting constants. [17][18][19] When Y = Me, Z = H, the H-C BDE follows the order for X: CH=CH 2 < Ph < PhCO = MeCO < CN < CO 2 Et < Me (Table 3, entries 1-7) The BDE for compounds where Y = Z = H follows the same order, albeit the value is of course slightly higher when compared to their methylated counterparts (entries [8][9][10][11]. On the other hand, trialkylboranes react much faster with an oxygen centered radical [20][21][22] than with a carbon radical. [23,24] Therefore, when R 1 = aryl, the highly stabilised and delocalised radical C <--> C' (R 2 CH•-COAr <--> R 2 CH=CO•Ar) has a strong propensity to react with Et 3 B on the oxygen part where a high electron density is located, thus affording the enolboronate H (Scheme 2). Therefore it is not surprising that the rare literature reports of successful intermolecular radical additions of α-oxo carbon radicals are limited to esters [2,3] that correspond to less stabilised radicals more likely to react on the carbon centre.
Nevertheless, we have shown in this article that an efficient control of the various reaction parameters (slow additions, temperature) permitted us to elude this problem. For α-oxo carbon radicals derived from aliphatic ketone derivatives, we succeeded in reducing this impediment, and the usual intermolecular addition could take place readily, even at low temperature.
In a previous paper, [7] we showed that the Et 3 B/air combination efficiently promotes the reduction of S-alkylxanthates. It is an apparent paradox that, even when "large" amounts of Et 3 B were used (i.e. 0.3-0.5 equiv, entries 1-20 and 22), the reduced product G could be detected only in a few instances (entries 2 and 14). However, in the addition process described in this paper, contrary to the reduction method (see ref. 7), the concentration of Et 3 B is maintained very low by slow addition with a syringe pump, thus minimising route e. Moreover, we demonstrated that the reduction process is relatively slow. As a consequence, the "degenerate" route a' 1 is much more efficient than the route e.

Conclusion
We have described a new, efficient, and extremely mild method for performing radical additions of 2-oxoalkylxanthates to various olefins. The efficiency of the addition process vs the premature reduction depends on the reactivity of a particular substrate toward a specific olefin for given reaction conditions. This approach can be extended to cyclisations that should operate even at low temperature.