Bis‐Cyclometalated Indazole Chiral‐at‐Rhodium Catalyst for Asymmetric Photoredox Cyanoalkylations

Abstract A new class of bis‐cyclometalated rhodium(III) catalysts containing two inert cyclometalated 6‐tert‐butyl‐2‐phenyl‐2H‐indazole ligands and two labile acetonitriles is introduced. Single enantiomers (>99 % ee) were obtained through a chiral‐auxiliary‐mediated approach using a monofluorinated salicyloxazoline. The new chiral‐at‐metal complex is capable of catalyzing the visible‐light‐induced enantioselective α‐cyanoalkylation of 2‐acyl imidazoles in which it serves a dual function as the chiral Lewis acid catalyst for the asymmetric radical chemistry and at the same time as the photoredox catalyst for the visible‐light‐induced redox chemistry (up to 80 % yield, 4:1 d.r., and 95 % ee, 12 examples).


Introduction
Chiral transition-metal complexes are ap rominent and powerful class of asymmetric catalysts, traditionally assembled from chiral organicl igandsa nd metal salts or organometallic precursor complexes. [1] The chiral organic ligandsa re typically involved in the asymmetric induction but also control the relative and absolute configurationo ft he transition metal complexes. Following ad ifferents trategy,w ea nd others have recently demonstrated that chiral transitionm etal complexes composed from entirely achiral ligandsc an be exquisite transition-metal catalysts foraw ide variety of asymmetric conversions, including asymmetric photocatalysis. [2][3][4] Such chiral-atmetal complexes rely on ac onfigurationally stable stereogenic metal center [5] for generating metal-centered chirality which at the same time must be ar eactive metal center for performing the asymmetric catalysis. [6] Our initial design was based on bis-cyclometalated iridium(III) and rhodium(III)c omplexes, in which two 5-tert-butyl-2phenylbenzoxazoles (IrO [7] and RhO [8] )o r5 -tert-butyl-2-phenylbenzothiazoles (IrS [9] and RhS [10] )i mplement as tereogenic metal centerw ith either al eft-handed (L-configuration) or right-handed (D-configuration) overall helical topology ( Figure 1). These two cyclometalated ligands are configurationally inert so that the overall stereochemical information is re-tained in these complexes once they are generated in an onracemic fashion. Twoa dditional monodentatea cetonitrile ligands are labile and provide access of substrates or reagents to the metal center. We found that the nature of the cyclometalating ligand has ap rofoundi nfluence on the reactivity and stereoselectivity of the bis-cyclometalated iridium(III) and rhodium(III)c omplexes [2] and we weret herefore seeking to investigate ligandst hat differ from our previousb enzoxazole and benzothiazole systems. Here we now introduce an ew class of bis-cyclometalated chiral-at-metal rhodium(III) catalystsw hich are based on two cyclometalated 6-tert-butyl-2-phenyl-2H-indazole ligands   (L-a nd D-RhInd). We demonstrate that this RhInd catalyst is superior for the visible-light-induced enantioselective a-cyanoalkylation of 2-acyl imidazoles in which RhInd serves ad ual functionast he chiral catalystbut is also involved in the photochemicalinduction.

Results and Discussion
Design and synthesis of the rhodium catalyst As part of our ongoing interest in expanding the structural diversity of bis-cyclometalated rhodium-complexes, we chose 2phenyl-2H-indazole as an interesting candidate. Bis-cyclometalated iridium complexes with 2-phenyl-2H-indazoles are well established [11] but the analogous rhodium(III)c omplexes have not been reported.T he overall geometry of this ligand is comparable to our previously applied benzoxazole and benzothiazole ligands, however,t he electron-rich aromatic system of 2Hindazoles provides significantly distinct electronics that might enable new catalytic transformations. The chiral-auxiliary-mediated synthesis [12][13][14] of the enantiopure catalyst RhInd started with the reaction of rhodiumt richloride hydrate with 2.0 equivalentso f6 -tert-butyl-2-phenyl-2H-indazole( 1), followed by additiono f2 .0 equivalents of AgPF 6 in MeCN to obtain bis-cyclometalated rac-RhInd in 97 %y ield (Scheme1). Afterwards, the racemic product wasr eactedw ith the monofluorinated salicyloxazoline (S)-2 [10,15,16] to provide the two diastereomers L-(S)-3 and D-(S)-4 in 40 %a nd 48 %y ield, respectively,w hich were separated by column chromatography on deactivated silicag el. The required high diastereomeric purity of the isolated auxiliary complexesw as evaluated by 1 HNMR and 19 FNMR spectroscopy.C leavage of the auxiliary ligand was subsequently performedu nder acidic conditions using trifluoroacetic acid (TFA), followed by anion exchange with NH 4 PF 6 to provide the individuale nantiomers L-RhInd (95 %yield) and D-RhInd (87 %y ield).
The absolute configuration was assigned based on ac rystal structure of L-RhInd ( Figure 2a). The CD spectra of L-a nd D-RhInd are shown in Figure 3 and confirmt heir mirror-imaged structures.H PLC analysiso nachiral stationary phase exhibited an ee of > 99 %f or both the L-a nd the D-RhInd complex ( Figure 4). Superimposition of the crystal structures of RhInd and RhS reveals as lightly larger distance between the two quaternary carbon atoms of the tert-butyl groups for RhInd (11.3 )m aking the catalytic site slightly larger comparedt oRhS (10.5 ) (Figure 2b). [10a] Initial experiments and optimization With the new enantiopure complexes in hand, we next investigated the application of RhInd in asymmetric photoredox catalysis. [17,18] After some initial reaction screening, we were delighted to find that L-RhInd (2.0 mol %) catalyzes the a-cyanomethylation of 2-acyl imidazole 5a with bromoacetonitrile (6a)i nt he presence of Na 2 HPO 4 and under irradiation with blue LEDs to provide( R)-7a with ah igh ee value of 94 %b ut in only 22 %y ield ( Table 1,   [19] Ta ble 1s hows the stepwiseo ptimization of this enantioselective, visible-light-induced cyanoalkylation. First, different solvents were investigated (entries 1-6) and it was found that MeOH/THF 4:1p rovided the best results. Changing the base from Na 2 HPO 4 to 2,6-lutidine or Cs 2 CO 3 provided higher yields of 50 %a nd 73 %b ut the enantioselectivity dropped to 87 %a nd 4%ee. (entries 7a nd 8). N,N-Diisopropylethylamine (DIPEA) as base only provided 2% yield and 42 % ee (entry 9). Despite the low yield, Na 2 HPO 4 was selected as the most suitable base with respect to enantioselectivity.Ahigher catalystl oading afforded improved yields but as ignificantly lower enantioselectivity (entries 10 and 11). The reduced enantioselectivity can be rationalized with as low RhInd-catalyzed racemizationo ft he product upon coordination to the catalyst, followed by deprotonation and reprotonation (see Supporting Information for more details). Increasing the amount of bromoacetonitrile to 6.0 equivalents improved the yield while maintaining ah igh ee (entry 12). Increasing the amount of base from 1.1 to 1.5 equivalents (entry 13) or 2.0 equivalents (entry 14) provided further improved yields of 78 %o r8 0%,r espectively.I ncreasing the amount of base to 2.5 equivalents led to as harp drop in the yield to 27 %, probably due to the resultingt urbidity from the low solubility of Na 2 HPO 4 having a negative effect on the penetration by the light into the reaction suspension (entry 15). Finally,i ti sw orth noting that we found that small amounts of water provide ab eneficial effect, probablyb yf acilitating rapid proton transfer,a nd therefore   To summarize this part, we found reactionc onditions for the photoinduced cyanoalkylation reaction 5a+ 6a!(R)-7a in 80 %y ield with 94 % ee using 2mol %o ft he chiral-at-rhodium complex L-RhInd as the single catalyst.

Substrate scope
After having established the optimized reaction conditions, we next investigated the scope of the a-cyanoalkylation with respect to 2-acyli midazoles (5a-j)a nd a-cyano bromides (6a-f) (Scheme 2). Substrate 5a providedt he best resultsw ith unbranched bromoacetonitrile (6a)w ith respect to yield and enantioselectivity (7a). Interestingly,m ethyl substituted imidazole substrate 5b only gave 2% yield and 80 % ee (7b). Mesityl substituted imidazole substrate 5c provided 52 %y ield and an ee of 94 %( 7c). The addition of electron withdrawing and electron donating groupsa tt he phenyl moiety had as lightly disadvantageous effect on the yield as well as the enantioselectiv-ity (7d-f). The implementation of an aphthyl moiety resulted in as ignificantly lower yield and am oderate enantioselectivity of 76 % ee. (7g). 2-Thiophenyl substrate 5h showed no conversion at all, whereas 3-thiophenyl substrate 5i provided 56 % yield of 7i with 78 % ee. This can be rationalizedbyabidentate coordination of the catalyst to the sulfur atom and the acyl oxygen of substrate 5h thus impeding the conversion.A liphatic substrate 5j only provided al ow yield of 16 %w ith 50 % ee (7j). Furthermore, five branched a-cyano bromides were investigated. Diastereoselectivities were validated by 1 HNMR spectroscopyo ft he crude products.P roduct 7k was formed in 64 %y ield with ad .r.o f1 .08:1. Both of the diastereomers showedh igh ee values of 94 %a nd 95 %. Product 7l was formed with ad .r.o f3 .01:1 with the major diastereomer exhibiting an ee of 94 %. When a-bromophenylacetonitrile was used, the cyanoalkylation product 7m was obtainedi n5 6% yield with ad .r.o f4 .01:1. Interestingly,t he major diastereomer showedavery good ee of 95 %w hile the minor diastereomer was obtained with only 31 % ee. Unfortunately,b utyronitrile and isobutyronitrile did not form any cyanoalkylation products (7n and 7o).

Mechanistic proposal
We previously reportedaseries of visible-light-induced enantioselective a-alkylations of 2-acyl imidazoles using electron deficient benzyl bromides, phenacyl bromides, perfluoroalkyl halides, and enantioselectivet richloromethylationsw ith BrCCl 3 . [2, 9a, 20, 21] These photoreactionsw ere catalyzed moste ffectively with bis-cyclometalated iridium complexes, whereas relatedp hotoinduced enantioselective a-aminations of 2-acyl imidazoles were catalyzed by the relatedb is-cyclometalated rhodium complexes. [2,22,23] These enantioselective photoredox reactions serve as the basis for the proposed mechanism of the here introduced rhodium-catalyzed photoinduced a-cyanoalkylation of 2-acyl imidazoles. Accordingly,t he catalytic reaction begins with the coordination of the 2-acyl imidazole substrate( e.g. 5a)t ot he rhodium catalyst in ab identate fashion upon release of the two labile MeCN ligandso fRhInd, therebyf orming intermediate I.As ubsequentd eprotonation induced by the added base Na 2 HPO 4 generates the rhodium enolate complex II whichi sakey intermediate of this asymmetric photoreaction. It fulfills ad ual function as reactive intermediate in the catalytic cycle and as the in situ assembled visible light activatable photoredox catalyst. Upon absorption of visible light, the rhodium enolateacts as ap hotoexcited reducing agent and transfers as ingle electron [24] to the a-cyanoalkyl bromide (e.g. 6a), which in turn fragments into bromide and the a-cyanoalkyl radical V.T his free radical V is electron deficient due to the electron withdrawing cyano group in a-position andt herefore rapidlyr eacts with the electron rich double bond of the rhodiume nolate II to form the ketyl radicali ntermediate III upon formation of an ew CÀCb ond and as tereogenic carbon,t he absolutec onfigurationo fw hich is controlled by the chiral rhodiumc omplex. [25] The ketyl radical III is a strong reducing agent and either regenerates the oxidized photoredox mediator (II!III)o rit directly transfers an electron to an ew a-cyanoalkyl bromide substrate to initiate ac hain reaction. Either way,r hodium-coordinated product IV is formed and after product release( e.g. 7a)t he coordination of new substrate leads to anothercatalytic cycle.

Mechanistic controlexperiments
The proposed catalytic cycle is consistentw ith an umber of control experiments. First, the reactionr equires both catalyst and Brønsted base for achieving conversion( Ta ble 2, entries 1 and 2), indicating the important role of the intermediate rhodium enolate ( intermediate II in Scheme3). Under air,t he CÀC coupling product is completely suppressed which is consistent with the interference of air with the proposed radical pathway, thus resulting in the formation of a-keto-2-acyl imidazole as an undesired side product (entry 3). Without any visible light, only 20 %y ield with significantly lower enantioselectivity was observed (entry 4). We propose that this product formationint he dark is the result of an on-radical S N 2-pathway.W hen the photoreaction was performed in the presenceo ft he radical trapping reagent (2,2,6,6-tetramethylpiperidin-1-yl)oxyl (TEMPO), the yield dropped to 25 %( 1.0 equiv TEMPO) and 20 % (6.0 equiv TEMPO), strongly indicating the involvement of a photoinduced radicalm echanism (entries 5a nd 6). As ignificant drop in enantioselectivity to 80 %i so bserved when H 2 O is excluded, demonstrating its crucial effect (entry 7). The benefit of small amounts of H 2 Oc an be rationalized with an improveds olubility of the base Na 2 HPO 4 in the reaction solvent. We also determined aq uantum yield for this reactions, which is 0.046 for the reaction 5a+ 6a!(R)-7a,w hich suggests that the chain propagation plays at most am inor role and instead the rhodiume nolate complex II exerts the functiono fareal photoredox catalyst which is closely coupled to the asymmetric catalysis cycle. This is different from our previousi ridiumcatalyzed a-alkylations [2, 9a, 20, 21] and rhodium-catalyzed a-aminations [2,22] which apparently follow ac hain mechanism (quantum yields > 1). Finally,U V/Vis absorption spectras hown in Figure 5d emonstrate that the 2-acyl imidazole substrate 5a, the catalyst RhInd,a nd the rhodium ketone complex I are not capable of significantly absorbing visible light but that the rhodium enolatec omplex II after deprotonation of I features a new absorption band in the bathochromic region with a shoulder above 400 nm, which shouldb er esponsible for the visible-light-induced photochemistry.I ta lso explains why the shorterw avelength of blue LEDs provides better resultsc ompared to acompact fluorescence light (CFL) bulb (entry 8).

Comparison with other catalysts
The performance of the new catalyst RhInd was compared with some relateda nd previously reported bis-cyclometalated complexes for the here introduced photoinduced cyanoalkylation. The bis-cyclometalated phenylbenzothiazole complex L-IrS,w hich was very successfully applied to av ariety of enantioselectivep hotoinduced a-alkylations of 2-acyl imidazoles, [2, 9a, 20, 21] provided ah igh enantioselectivity of 94 % ee butw ith just 23 %y ield (Table 2, entry 10). The low yield can be explained by the inhibition of the catalyst by blocking the active site of the catalyst through coordination of the bromoacetonitrile substrate or the cyanoalkylated product. This is consistent with the fact that the bis-cyclometalated iridium catalyst displays am uch slowerl igand exchange kinetics comparedt oi ts rhodium congener and thus should be more sensitivet oc ompetingc oordinating functional groups. [22] Indeed, when we increasedt he reaction temperature to 50 8Ct os peed up ligand exchange, L-IrS gave as ignificantly highery ield of 62 %b ut provided ar acemic mixture of the product, which might be due to an uncatalyzed background reactiona th igher temperatures (entry 11). On the other hand, the bis-cyclometalated phenylbenzothiazolec omplex L-RhS,w hich provedh ighly suitable for av ariety of photoinduced a-aminations of 2-acyl imidazoles, [2,10,22,23] provided the cyanoalkylation product with 83 %y ield but as lightly lower enantioselectivity of 90 % ee (entry 12). At af irst glance,t his lower enantioselectivity is surprising since the more constrained active site of the benzothiazole catalyst (see Figure2b) should provide ah igher asymmetric induction. This is exactly what we observed for photoinduced a-aminations of 2-acyl imidazoles in which the benzothiazolec atalyst RhS provided significantly higher ee values comparedt ot he benzoxazole analogue RhO. [10a] We suggest that the higher enantioselectivity of RhInd over RhS for the photoinduced cyanoalkylation is due to as lower S N 2b ackground catalysis with RhInd,a nd this is crucial because we observed al ower enantioselectivity for this reaction pathway. Indeed, in the presence of air and absence of light, L-RhS provided the cyanoalkylationp roduct in 38 %y ield and with 88 % ee (entry 13), as comparedt oay ield of only 20 %w ith 86 % ee for L-RhInd under the same conditions (entry 4). Finally,s ome modified RhInd-catalysts were also tested but provided inferiorresults(entries 14-16).

Experimental Section
General procedure for enantioselective a-alkylation of 2acyl imidazoles 2-Acyl imidazole (0.10 mmol), L-RhInd (2.00 mol %) and Na 2 HPO 4 (0.20 mmol) were dissolved in MeOH/THF 4:1( 0.5 mL) under inert gas atmosphere and H 2 O( 0.56 mmol) was added. The resulting mixture was stirred for 5min before bromoacetonitrile (0.60 mmol) was added and the mixture was thoroughly degassed via freezepump-thaw for three cycles. The reaction mixture was then stirred for 24 hu nder inert gas atmosphere at r.t. in front of blue LEDs (24 W, 10 cm). Afterwards, the solvent was evaporated under vacuum and the precipitate was purified by column chromatography on silica gel (n-pentane/EtOAc 5:1!2:1) to afford pure products. For compounds 7k-m diastereomeric ratios were determined by 1 HNMR spectroscopy of the crude products.