Photocontrolled Cobalt Catalysis for Selective Hydroboration of α,β‐Unsaturated Ketones

Abstract Selectivity between 1,2 and 1,4 addition of a nucleophile to an α,β‐unsaturated carbonyl compound has classically been modified by the addition of stoichiometric additives to the substrate or reagent to increase their “hard” or “soft” character. Here, we demonstrate a conceptually distinct approach that instead relies on controlling the coordination sphere of a catalyst with visible light. In this way, we bias the reaction down two divergent pathways, giving contrasting products in the catalytic hydroboration of α,β‐unsaturated ketones. This includes direct access to previously elusive cyclic enolborates, via 1,4‐selective hydroboration, providing a straightforward and stereoselective route to rare syn‐aldol products in one‐pot. DFT calculations and mechanistic experiments confirm two different mechanisms are operative, underpinning this unusual photocontrolled selectivity switch.


Introduction
Thes elective reduction of a,b-unsaturated carbonyl compounds is at ransformation of widespread importance in synthetic chemistry and the underlying reactivity principles have made it into well-established text-book knowledge. [1] In this context, the decisive factor underpinning selectivity is the concept of hardness and softness,ofthe nucleophile/reducing agent and the reactant, which dictate whether attack occurs at the carbonyl (1,2-addition) or the b-carbon (1,4-addition), respectively (Scheme 1a). [2][3][4] Although significant progress has been made with the use of metal-additives to tune the hardness/softness of classical reducing agents,challenges still exist in this area.
Catalytic hydroboration is considered to be amild method of reduction, relying on commercially available boranes and displaying increased functional group tolerance and selectivity when compared with traditional reducing agents. [5][6][7] A broad range of mechanistic pathways have been reported, [8] however the vast majority of methods for a,b-unsaturated ketones are 1,2-selective (Scheme 1b). [9][10][11][12] 1,4-selective meth-ods have been restricted to linear substrates and do not work on cyclic a,b-unsaturated ketones which cannot access the required reactive s-cis conformation (Scheme 1c). [13][14][15] A method to address this significant limitation would be extremely valuable,e nabling regioselective formation of cyclic enolborates which give contrasting stereoselectivity in aldol reactions to their enolborinate and silyl enol ether counterparts. [16][17][18] With the recent attention directed towards the use of light to control reactivity [19][20][21] and selectivity [22][23][24] in catalytic reactions,w eb ecame interested in developing am ethod which, in contrast to the established hardness/softness reactivity concepts,allowed for afundamental selectivity reversal by simple use of visible light. This would enable us to exert precise control over reactions with an on-invasive,e xternal stimulus and avoid the need for any stoichiometric additives.
In the field of transition metal catalysis,l igand photodissociation has been shown as amethod to switch areaction between on and off states by revealing ligation sites at ametal centre.O ne such example is the 18 electron earth-abundant transition-metal complex, CoH[PPh(OEt) 2 ] 4 which undergoes photodissociation with visible light to generate 16 electron complex, CoH[PPh(OEt) 2 ] 3 . [25,26] It can be seen from the optimised structures [27] that any coordinative interaction of as ubstrate with the metal centre of CoH[PPh(OEt) 2 ] 4 appears improbable,w hereas in contrast, the corresponding complex generated upon light irradiation, CoH[PPh(OEt) 2 ] 3 ,has avacant coordination site (Scheme 1d). Despite this,wehave recently reported acatalytic system based on the use of CoH[PPh(OEt) 2 ] 4 in conjunction with pinacolborane for alkene isomerisation, indicating the potential for adifferent mechanistic scenario. [28] We therefore hypothesised that if two different mechanistic pathways,controlled only by the presence or absence of light, are operative,t his may lead to contrasting selectivity in the context of catalytic hydroboration of a,b-unsaturated carbonyls.I na ddition, this divergent mechanistic control may obviate the requirement of s-cis conformation to perform 1,4-hydroboration, leaving as ingle catalytic platform able to carry out both 1,2 and 1,4-hydroboration of linear and, the previously unsolved, cyclic unsaturated ketones.

Results and Discussion
Our investigations began using 5mol %o fC oH[PPh-(OEt) 2 ] 4 in conjunction with pinacolborane for the reduction of chalcone 1a.I nb enzene solvent and in the dark, product 2a was obtained upon quenching the reaction with water, selectivity in line with previous reports of hydroboration with catechol borane. [14] Interestingly,upon irradiating the reaction with blue light, ac omplete switch in selectivity was noted, yielding the allylic alcohol, 3a.Ascope of this reaction was then carried out to probe the generality of this light controlled, regioselective reduction procedure (Scheme 2).
Chalcones containing both electron rich (1b)and electron poor arenes (1c-1f)w ere selectively reduced under light or dark reaction conditions,d isplaying precise control over selectivity.N otably,t he ester functionality of 1c was untouched despite using excess pinacolborane.U pon changing from ap henyl ketone to am ethyl ketone with b-phenyl substitution, the selectivity unexpectedly changed in the dark to yield the allylic alcohol, 2g in good yield which we initially attributed to the conformational preference of the starting material. However,s ubstrates 1h and 1i gave am ixture of products under the "dark conditions". In contrast, substrate 1g underwent non-selective hydroboration in the light whereas substrates 1h and 1i gave selective,a lthough contrasting 1,4-and 1,2-products,respectively.
Starting material 1j,with amethylene group between the alkene and ketone functionality,was reduced in a1,2 fashion in the dark to give 3j as the major product. Notably we obtained product 2j when the reaction was carried out in the light which we believe arises from the following sequence: isomerisation of the olefin to give the a,b-unsaturated ketone, 1,4-selective hydroboration and protonation. Previous studies have established that this cobalt complex is able to isomerise alkenes under light irradiation. [25] As we had shown that light was able to switch the reaction outcome on ar ange of linear substrates,w et herefore then turned to more challenging cyclic a,b-unsaturated ketone substrates.T he inability of these substrates to undergo 1,4 selective hydroboration to initially form the boron enolate, had been highlighted as ar estriction in previous methodologies. [14,29,30] With 4,4-dimethylcyclohex-2-en-1-one, 1k,a s the starting material in benzene,u sing pinacolborane in the dark, we observed 1,2-selective reduction product 3k upon quenching with water-selectivity that has been reported for abroad range of catalysts. [5,8] Upon carrying out the identical reaction in the light, however, we were delighted to obtain the product 2k from conjugate reduction. This unusual control of regioselectivity using only light as an external stimulus appeared to offer an excellent route to the desired boron enolates.F urthermore,t his showcases the concept of using light to control the catalytic pathway and thus the selectivity of ahydridic reagent, rather than stoichiometric quantities of additives which generate significant waste.Further optimisation (see the Supporting Information for details) demonstrated that in the light, the saturated ketone was obtained with even higher yield when the reaction was carried out in THF.
We carried out as cope of the light-switchable reduction under the optimised conditions.Starting material 1l,containing an ester group,s howed complete selectivity for enone reduction under both sets of conditions,y ielding either 3l or 2l selectively.C yclohexenone rings substituted at the gposition (1m & 1n)w ere also suitable substrates for these reactions (albeit showing am ixture of diastereomers in the dark). Carvone, 1o,i sachallenging substrate to selectively hydroborate due to the presence of the electron rich alkene. Previously reported methods which rely on more reactive alkylboranes would also hydroborate this functionality,h owever under our conditions,t his handle remains untouched to yield either 2o or 3o in the light or dark, respectively.Wealso sought to see if this reactivity was also applicable to a,bunsaturated aldehydes,h owever, substrate 1p gave only the 1,2-reduced product in the dark with this also being the major product in the light.
Our next step sought to build upon the new reactivity discovered in the light by reacting the boron enolates with other electrophiles.L ipshutz and Papa have reported ao nepot reductive aldol reaction using air-sensitive Strykers reagent and in situ generated diethyl borane which leads to anti-selective aldol products for cyclic substrates. [15] Similarly, there are limited reports on precious metal hydride catalysis for reductive aldol reactions which are applicable to cyclic enone substrates,t hough these again favour the anti-products. [31] Unlike for linear substrates,w here control of enolate geometry is the major factor for controlling aldol stereoselectivity,c yclic substrates require ac hange of enolate. Methods to generate the syn-aldol products have relied on tin, [32] zirconium [33] or titanium [34] enolates,h owever there have also been previous reports that enolborates favour syn-aldol products in direct contrast to enolborinates. [16][17][18] Previous difficulties in directly generating these starting materials had limited the practical utility of the method hence we were attracted to the possibility of using our chemistry in this context.
Pleasingly,u pon addition of 1.5 equivalents of benzaldehyde to the hydroborated intermediate,weobtained product 4a product in excellent 78 %y ield with only the syn-isomer observable,i nd irect contrast to the previous copper hydride catalysed method. Ar ange of cyclic enone substrates with different ring sizes were suitable substrates for this one-pot aldol reaction (4b-4e), with the syn-diastereomer being produced with excellent selectivity in almost all cases (Scheme 3). Notably,w ew ere able to form aq uaternary centre (4f)f rom an a-substituted starting material, however this gave almost exclusively the anti-product, in line with previous literature. [17] Linear starting material also gave synaldol product 4g.Application of this methodology to amore complex substrate derived from Metandienone (Dianabol), enabled as ite selective functionalisation to give product 4h. Significantly,only the less-hindered 1,4-site underwent hydroboration and no products were observed from 1,2-hydroboration.
We then turned to the scope of the aldehydes that could be used in the reaction. Notably,e lectron rich aldehydes gave excellent d.r. (4i & 4j)w ith only the syn-diastereomer detected. More electron deficient aldehydes were less selective though the yields were still good (4k & 4l)a nd heterocyclic aldehydes worked well (4m & 4n). Theu se of cinnamaldehyde gave acompletely syn-selective product (4o) whereas the selectivity was lower using aliphatic isobutyraldehyde, 4p.
Having established this platform for selectively accessing ab road range of hydroborated enone products and demonstrated their application in the aldol reaction, our attention turned to the mechanism and understanding the observed Scheme 2. Substrate scope of the photocontrolled cobalt-catalysed hydroboration of a,b-unsaturated ketones.

Angewandte Chemie
Research Articles reactivity.Heating the reaction, rather than irradiating it with light increased the proportion of 1,4-reduced product (Scheme 4a). However,t he selectivities and yields obtained failed to match those obtained under light irradiation, demonstrating the unique ability of light to promote 1,4-selective hydroboration under mild conditions. We next sought to rule out amechanism whereby the light promoted selectivity arose from sequential hydroboration and isomerisation. To this end, the product arising from 1,2hydroboration of substrate 1owas irradiated with blue light in the presence of catalytic CoH[PPh(OEt) 2 ] 4 and pinacolborane but no 1,4-hydroborated product was produced (Scheme 4b), providing strong evidence against such ascenario.Next, to probe the role of "freed" phosphonite ligand, we carried out the reaction in the dark with an additional 5mol %o f PPh(OEt) 2 .I nt his case,w eo bserved significantly decreased reactivity but no change in the selectivity (Scheme 4c).
In order to shed further light onto the mechanistic differences between the dark (saturated 18 electron cobalt complex) and light (unsaturated 16 electron cobalt complex), we analysed the mechanism by means of DFT calculations which have emerged as apowerful methodology to analyse 3d transition metal catalysis. [35] Computations were carried out at the CPCM(benzene)/B3LYP-D3/Def2TZVPP//B3LYP-D3/-6-31G(d)/LANL2DZ level of theory (see the Supporting Information for more information).
It is well-established that light promotes phosphonite dissociation in this complex and, in almost all previously reported cases,t he dark species has been considered inactive. [25,36] Direct comparison of the optimised structures of both saturated and unsaturated catalyst (Scheme 1d), demonstrates that the available space for substrate coordination is completely different:w hile in the unsaturated species as ubstrate can directly coordinate to the cobalt centre,there is no space available in the saturated one.T herefore,g iven the difference in this starting point, it would be expected that vastly different mechanistic pathways occur.
Furthermore,e xperiments carried out in absence of pinacolborane strongly suggest that ad ifferent intrinsic mechanism must be operative:n or educed products are detected in the dark (Scheme 4d), indicating am echanism reliant on activation of pinacolborane.However,inthe light, 5% of reduced product 2k was observed after aqueous workup,i na bsence of pinacolborane (Scheme 4e), which is consistent with the borane being required only to turn over the catalytic cycle in the final steps,a fter generation of ac obalt enolate species.T his also correlates with previous literature suggesting that cobalt hydrides are "soft" in character. [37,38] We began our computational investigation of the lightmediated catalytic reaction considering the unsaturated 16 e À species I as the starting point ( Figure 1) and substrate 1k as the model reaction. First, coordination of the substrate can take place in two different orientations,o ne through the carbonyl moiety (II)a nd the other through the double bond (III). Although coordination via the carbonyl group is largely favoured, the hydride migration into the C=Ob ond, though still thermally accessible at room temperature,issignificantly higher in energy than into the C = Cb ond (18.4 kcal mol À1 vs. 10.7 kcal mol À1 ). This difference can be explained in terms of the highly distorted 4-membered ring transition state in the case of C=Oi nsertion, while in the case of C=Ci nsertion, good conjugation of the resulting formal carbanion with the carbonyl facilitates the CoÀCb ond formation. From this point, keto-enol tautomerisation takes place from IV with an overall barrier of 13.5 kcal mol À1 ,f orming intermediate V exergonically.T his result is in agreement with the experimental observation depicted in Scheme 4e,s ince HBPin is not needed to promote the reaction up to this point and intermediate V can easily proceed to the 1,4 reduced product during work-up.I nt he presence of HBPin, intermediate V easily evolves towards the final boron-enolate and Co I -H, completing the catalytic cycle. [39] Our attention then turned to the more challenging dark mechanism. Monitoring of the reaction by 1 H, 11 Ba nd 31 PNMR revealed no new cobalt hydride species during the course of the reaction and no evidence of phosphonite ligand dissociation or modification occurring. We observed small traces of BH 3 by 11 BNMR appear after several hours, [40] but replacing the cobalt complex with catalytic BH 3 ·THF under our conditions did not prove effective in generating hydroborated products (Scheme 4f).
Therefore,w ith this information in hand, we initially explored the direct outer sphere transfer of hydride to the substrate with the assistance of HBPin (TS VII-XII)(without the assistance of HBPin the reaction is even more endergonic). [41] Interestingly,t he barrier is extremely high (45.1 kcal mol À1 )a nd that is in agreement with the fact that no product is detected when HBPin is not used in the reaction mixture (Scheme 4d). Intriguingly,w eo bserved experimentally,t hat after addition of HBPin to the catalyst, an NMR signal corresponding to H 2 is clearly observed, suggesting afirst reduction step of Co I to Co 0 promoted by the reaction mixture.I no rder to clarify that point, the Co I -H to Co 0 equilibrium was explored computationally and surprisingly, the reduction is exergonic even with the concomitant formation of ar adical species (VII to VIII). [42] Encouraged by this finding,wecalculated an alternative pathway based on aCo 0 -Co I electron transfer mediated catalytic cycle.
Once VIII is formed, Co 0 complex can interact with HBPin to form IX that is then activated with the assistance of the cyclohexanone substrate through TS IX-X BS (Figure 2, left). This is the rate determining step of the reaction, where asingle electron transfer from alone pair of the cobalt centre to the HBPin occurs,forming Co I -H through ahydrogen atom abstraction. BPin radical is simultaneously trapped by the oxygen of the substrate,f orming ac onjugated radical (X). This step is exergonic by 3.0 kcal mol À1 respect to intermediate VIII,and the overall barrier is only 24.9 kcal mol À1 ,more than 20 kcal mol À1 lower than the outer sphere Co I -H attack. Thereason behind the low energy of this pathway lies in the broken-symmetry electronic structure (see the Supporting Information for further details). Finally,OBPin allyl radical X

Angewandte Chemie
Research Articles 21180 www.angewandte.org reacts with Co I -H via hydrogen atom transfer, regenerating Co 0 and substrate XI through transition state TS X-XI (Figure 2, right). Interestingly,1 ,2 recombination is faster than 1,4, which is in agreement with the experimental observations,a lthough the predicted ratio (70:30) slightly differs to the observed one (only 1,2). This difference may arise from the use of methoxy groups instead of ethoxy in the phosphonite ligand.

Conclusion
In conclusion, we have developed acatalytic platform for the hydroboration of a,b-unsaturated ketones.T his system relies on ab ench-stable,e arth-abundant cobalt hydride catalyst, uses commercially available pinacolborane,o ccurs at room temperature,a nd showcases au nique ability to control the regioselectivity depending on the presence or absence of light. As ar esult, we are able to access ab road range of products using just one reaction system, including access to cyclic boron enolates of which we have demonstrated the utility for syn-selective aldol reactions.E xperimental and computational experiments demonstrate that two distinct mechanistic pathways are operative in the dark and light which provides an explanation for the selectivities observed. Furthermore,i th ighlights the significant role that the coordination sphere of am etal complex can have on reactivity and selectivity and the underexplored potential that exists in controlling this with light.