General Synthesis of Secondary Alkylamines by Reductive Alkylation of Nitriles by Aldehydes and Ketones

Abstract The development of C−N bond formation reactions is highly desirable due to their importance in biology and chemistry. Recent progress in 3d metal catalysis is indicative of unique selectivity patterns that may permit solving challenges of chemical synthesis. We report here on a catalytic C−N bond formation reaction—the reductive alkylation of nitriles. Aldehydes or ketones and nitriles, all abundantly available and low‐cost starting materials, undergo a reductive coupling to form secondary alkylamines and inexpensive hydrogen is used as the reducing agent. The reaction has a very broad scope and many functional groups, including hydrogenation‐sensitive examples, are tolerated. We developed a novel cobalt catalyst, which is nanostructured, reusable, and easy to handle. The key seems the earth‐abundant metal in combination with a porous support material, N‐doped SiC, synthesized from acrylonitrile and a commercially available polycarbosilane.

Abstract: The development of CÀNb ond formation reactions is highly desirable due to their importance in biology and chemistry.R ecent progress in 3d metal catalysis is indicative of unique selectivity patterns that mayp ermit solving challenges of chemical synthesis. We report here on ac atalytic CÀNb ond formation reaction-the reductive alkylation of nitriles. Aldehydes or ketones and nitriles, all abundantly availablea nd low-cost startingm aterials, undergo ar eductive coupling to form secondary alkylamines and inexpensive hydrogen is used as the reducing agent. The reaction has av ery broad scopea nd many functional groups, including hydrogenation-sensitive examples, are tolerated. We developed an ovel cobalt catalyst, which is nanostructured, reusable, and easy to handle. The key seems the earth-abundant metal in combination with ap orouss upport material, N-doped SiC, synthesized from acrylonitrile and acommercially available polycarbosilane.
CÀNb ond formation reactions are of fundamental interest in chemistry and biology,a nd amines are very important compounds and key functional groups in many bulk and fine chemicals, [1] drugs [2] and materials. [3] Differently substituted secondary alkylamines,e xamples of pharmaceuticals are shown in Scheme 1A,a re challenging to synthesize. Most of the existing catalytic methods, such as borrowing hydrogen or hydrogen autotransfer, [4,5] reductivea mination, [6,7] hydroaminomethylation [8] and hydroamination, [9,10] albeit intensively investigated, start already from an amine (Scheme 1B)a nd are restricted regardingt he synthesis of secondary alkylamines. [11] The hydrogenationo fa mides is an alternative (Scheme 1C), since it does not requirea na mine as starting material, [12,13] but again the synthesiso fd ifferently substituteds econdary alkylaminesi s rarely reported. [14] Theu se of catalystsb ased on earth-abundant metals such as Mn,F eo rC oi nr eactions classicallya ssociated with rare noblemetalsisalsoo ffundamental interest. [15][16][17][18][19][20][21][22][23] We report herein that the reductive alkylation of nitriles by aldehydes or ketones permitst he general synthesis of secondary alkylamines. The reactionh as ab road scope. Aromatic or aliphatic nitriles and benzylic or aliphatic aldehydes or ketones -dialkyl,d iaryl or arylalkyl substituted-can be employed.I n addition, we demonstratedt he synthesis and modification of bioactive compounds and pharmaceuticals. Many functional groups,i ncluding hydrogenation-sensitive examples, are tolerated. We had to develop an ovel catalyst whichi sn anostructured,r eusable, and easy to handle. The key is the earth-abundant metal cobalt (Co) in combination with ap orouss upport material, N-doped SiC synthesized from acrylonitrile and a commercially available polycarbosilane. The catalyst has ad istinct selectivity pattern;i tp ermits the selective hydrogenation of aliphatic or aromatic nitrileswhilebarely reducing aldehydes or ketones. In addition, the transiently formed imine intermediate is efficiently hydrogenated.
Attempts to reductively link nitriles and carbonyl compounds in the gas phase [24] indicatet hat this high-temperature approachi se xtremelyl imited in conversion, scope, and functional group tolerance. Based on the inspiring development of Co catalysts fort he selective hydrogenation of nitriles, [25][26][27][28][29][30] we expectedt hat aC ocatalystc ould also be the key to developa broadly applicable catalytic process for the reductive alkylation of nitriles. Co catalysts have also been employed successfully in reductivea mination reactions, the coupling of amines or ammonia and aldehydes or ketonesi nthe presence of hydrogen as the reducing agent. [27,[31][32][33][34][35][36][37][38][39][40][41] In addition, the Co catalyzed transfer hydrogenation andc oupling of nitriles has been described. [42] Our Co catalyst( Co/N-SiC) was synthesized as shown in Figure 1A.F irstly,w es ynthesized the N-doped SiC support (N-SiC), whichc ontains 8a tom% (at %) nitrogen, as determined by elemental analysis, using am odified literaturep rocedure. [43,44] Secondly,t he N-SiC material wasw et impregnated with as olution of Co(NO 3 ) 2 in water.A fter the evaporation of the solvent, the sample was pyrolyzed under nitrogen flow at 700 8Cf ollowed by ar eduction step (N 2 /H 2 ,9 0/10) at 550 8C ( Figure 1A). Inductively coupled plasma optical emission spectrometry indicated 4.7w eight% (wt %) Co in the catalystm aterial. Homogeneously distributed Co nanoparticles with an average particle size of 6nmw ere determined with transmission electron microscopy ( Figure 1B). The catalyst has as pecific surface area (Brunauer-Emmet-Teller) of 367 m 2 g À1 with primarily micropores and about 9% mesopores. X-ray photoelectron spectroscopy analysisc onfirms the presence of metallicC o species and oxi(hydroxide) Co speciesa tt he surfaceo ft he Co nanoparticles ( Figure 1C). High resolution transmission electron microscopy of the Co nanoparticles confirms that the core of these particles is metallicC o( cubic phase, Figure 1D)a nd indicatea ne mbedding of the Co nanoparticles into the N-SiC support. In the diffraction data, I(Q), of the catalyst (Co/N-SiC), the broad reflexes of graphite are still visible ( Figure 1E). Additionally,s ome reflexesd ue to Cobalt arise in the black curve. The first reflex (002) of the graphitic support is shifted slightly to lower Qv alues in the catalystm aterial due to the loading with Cobalt compared to the bare N-SiC support, indicating expansion of the graphite interlayer spacingu pon catalyst loading. Ab iphasic refinement of the d-PDF data (Co/N-SiC-N-SiC) reveals the coexistence of crystalline Co fcc nanoparticles with diameters of 5.4 nm and smaller CoO fcc domains, with a phase ratio of Co particles to CoO domains of about 6:1.
The reductivea lkylation of benzonitrile with 4-methylbenzaldehydew as chosen as ab enchmark reaction for the optimization of the reaction conditions (Table 1, top). The most active catalystb ased on the yield of product in ag iven time was ob- Homogeneously dispersed Co nanoparticles with an average particle size of 6nm. C) X-rayp hotoelectron spectroscopyanalysis verifiest he presence of metallicC oa nd oxi(hydroxide) Cos pecies (about70%)a tt he surface of the Co nanoparticles.D )Characterization of the Co nanoparticles by high resolution transmission electronm icroscopy indicates that they are embedded in the N-SiC support and metallic E) XRD data of catalystC o/N-SiC (black)a nd support N-SiC material( green),t ogether with their difference (catalyst-support,inr ed).F )PDF refinement of the d-PDF (Co/N-SiC-N-SiC), showingthe contributions of the Co fcc phase (green)and of the CoO fcc phase( pink) to the biphasic fit (in offset for clarity), together with the difference curve (grey) only containingh igh frequency noise. Co(acac) 2 N-SiC 700 70  entries 1a nd 2). Am inor decrease in activity was observed by using catalysts with Co(OAc) 2 (OAcH = acetic acid) and Co(acac) 2 (acacH = pentane-2,4-dione). Pyrolysis temperatures of 600 and 800 8Cl ed to as ignificant decrease in the catalytic activity of the catalysts synthesized from Co(NO 3 ) 2 and N-SiC (Table 1, entries 4a nd 6). In addition, we investigated different catalysts upports (Table1,e ntries 8t ill 10;T able S13), which are all significantly less active. The other supports led to inactive catalysts for which we could not observe any product formation under these conditions. The combination of Co(NO 3 ) 2 and the porouss upport material N-SiC seem essential for the activity in the reductive alkylation of nitrilesw ith carbonyl compounds under the optimized and mild conditions. We expected mild conditions to be crucial in order to achieve ad esirable level of tolerance of functional groups.
We were interested in exploring the substrate scope and the functional group tolerance of our Co/N-SiC catalystn ext. Isolated yields of the secondary amines wereg iven for the corresponding hydrochloride salts. For the reductivea lkylation of benzonitrile with various aldehydes (Figure 2, products 1-19), the secondary amines using 4-methylbenzaldehyde (electrondonating group) and 4-chlorobenzaldehyde (electron-withdrawing group)w ere isolated in 99 %y ield (Figure 2, product 1). In addition, the influence of the substituent position was investigated by converting 2-, 3-or 4-methylbenzaldehyde and 2-, 3-or 4-chlorobenzaldehyde. meta-Substituted benzaldehydes showed no influence on the catalytic activity ( Figure 2, products 2, 5), but ad ecrease in the product yields were observed for sterically more demanding 2-methylbenzaldehydea nd 2-chlorobenzaldehyde ( Figure 2, products 3, 7). In addition, other electron-withdrawing substituents, such as fluorides and bromides, were toleratedwell;this was indicative, especially for bromide 8,t hat dehalogenationi so fm inor relevance ( Figure 2, products 4-8). The tert-butyl substituent in the para position of benzaldehyde showedn oi nfluence on the product formation and the corresponding amine 9 was isolated in 99 %y ield. Boronic esters, such as 4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)benzaldehyde, are of special importance, since they are common compounds for cross-coupling reactions and one example could be isolated in excellenty ield (Figure 2, 10). Secondary amines with methoxy and benzyloxy substituents, which are used to protect hydroxyl groups,w ere also isolated in nearly quantitative yields (Figure 2, products 11,12). Cyclic acetals, such as piperonal, which are common protectiveg roups for carbonyl groups, were also tolerated ( Figure 2, product 13). In addition, purely aliphatic alde-hydes can be employed (Figure 2, products 14-19)a lbeit with lower isolated yields.
For the reductive alkylation of various benzonitriles and benzaldehyde, methyl group and chloro-substituted benzonitriles were used to investigate the substituent positione ffect for both electron-donating and -withdrawing substituents (Figure 2, products 20-22 and 24-26). By converting 2-, 3-or 4-methylbenzonitrile, the yield decreases gradually from paravia meta-t oortho-substituted nitriles (Figure 2, products [20][21][22]. This trend could not be observed at the chloro-substituted nitriles, as the correspondinga mines were isolated in 99 % yield (Figure 2, products 24-26). Ad ecreaseo ft he product yield was observed by converting 2,6-difluorobenzonitrile, which we believe is due to the two electron-withdrawing fluoro substituents (Figure 2, product 28). The electron-donating methoxy substituent in the para positionh ad no negative influence on the yield of the corresponding amine, which was isolated in 99 %yield (Figure 2, product 29).
However,aslightd ecrease in the yield could be observed for the methoxy substituent in the meta position (Figure 2, product 30). Then aphthalene-based nitrile can be converted in good yields to the correspondings econdary amine 31.T o our delight, the reductive alkylation of an aromatic O-heterocycle proceeded very well and the corresponding amine 32 was isolated in nearly quantitative yield.
For the reductive alkylation of aliphatic nitrilesw ith various aldehydes (Figure 3), as before, the influence of the substituent positionw as investigated with the methyl-and chloro-substituted benzaldehydes (Figure 3, products 33-35, 37-39). The same trend was observed, but the yields of the methyl-substituted secondary amines 33, 34 and 35 were slightly lower,due to the lower reactivity of the aliphatic nitrile. The secondary amines with the electron-withdrawing chloro substituents at the 2-, 3-or 4-position were obtainedi n9 6-99 %i solated yield (Figure 3, products 37-39). Fluorinated and brominated benzaldehydesw eres moothly converted in the corresponding amines 36 and 40 in good yields. Thes econdary amine 41 with the electron-donating methoxys ubstituent in the para positionw as isolatedi n9 6% yield. 4-Benzyloxybenzaldehyde was converted selectively to product 42 in nearly quantitative yield without as ignificant amount of hydrogenolytic ether cleavage. Piperonal was converted to the corresponding amine 43 in very good yield, no cleavage of the acetalw as observed and ay ield of 99 %w as observed for 4-tert-butylbenzaldehyde ( Figure 3, product 44). We then varied the aliphatic nitrile and combinedi tw ith numerous, mostly purely aliphatic aldehydes. Purely aliphatic nitriles of different chain lengths were linked smoothly with benzaldehyde and the secondary amines were isolated in yields of 99 % ( Figure3,p roducts 45, 46). The synthesis of such alkyl-benzylamines seems more efficient if an aliphatic nitrile is used instead of an aliphatic aldehyde.Anitrile with ac yclohexane substituent was converted to the product 47 in good yields too. The sterically demanding pivalonitrile reacts well and the couplingp roduct with benzaldehyde was obtained in 82 %y ield (Figure 3, product 48). The unsatu-rated secondary amine 49,w hich was synthesized from valeronitrile and citronellal, was still isolated in an acceptable isolated yield of 62 %. Various pure aliphatic nitrilesa nd aldehydes with different chain lengthsw ere smoothly converted to the corresponding amines (Figure 3, products 50-54); no influence of the chain length on the yield were observed.
For reductivealkylation of nitriles with ketones, the imine intermediate formed after the condensation is sterically more protected and, thus, more difficult to hydrogenate. Higher reaction temperatures (110 8C) and catalyst loadings of 8mol % Co were required to obtain good yields. We systematically exploredt he coupling of an aromatic nitrile,n amely,b enzonitrile or ap urely aliphatic nitrile,n amely,p entanenitrile with ad iaryl, aryl-alkyl,d ialkyl or cyclic ketone. Isolated yields between 57 and 81 %w ere obtained (Figure 4, product 55 till 62). The yield of the secondary amines in the reductive alkylation of benzonitrile decreases gradually from diaryl via aryl-alky to dialkyl ketones ( Figure 4, products 55-57). The cyclic ketone was converted smoothly in the secondary amine 58.T he yield (85 %) was, compared to the other ketones,h igher due to as maller steric demand. The same trends but lower isolated yields were observed for the reactiono ft he aliphatic nitrile pentanenitrile with the same ketones ( Figure 4, products 59-62).
For the modification and synthesisofb iologically active molecules, nabumentonea nd pentoxifylline, common drug molecules, were converted to the respective amines in up to 78 % isolated yield (Figure 4, products 63-66). To our delight, pen-  toxifylline was converted without hydrogenation of the C=O bond. Our synthesis protocol may also be applied to synthesize biologically activem olecules. Tecalcet, which is used as a calcimimetic agent,c an be synthesized from the two commercially availablee ducts2 -chlorohydrocinnamonitrile and 3-methoxyacetophenone in 66 %i solated yield (Figure 4, product 67). Nitriles can also be alkylated with varioussteroid derivatives. The reductive alkylation of benzonitrile and valeronitrile with estrone, stanolone and testosteronep roceededs moothly, and the products were obtained in yields up to 80 % ( Figure 4, products 68-73). Te stosterone, which contains the CÀCd ouble bond, was convertedi nto the corresponding unsaturated amines 72 and 73.T he yields of the products,w hich were synthesized with an aliphatic nitrile, were slightly lower than those obtained with an aromatic nitrile. Various tests were performed to demonstrate the catalyst stabilitya nd leaching of cobalt species( see SupportingI nformation). The reductive alkylation of benzonitrile with 4-methylbenzaldehyde at about 65 %y ield of product was chosen to demonstrate the reusability of the catalyst and to establish its efficiency.F ive consecutive runs showed no decrease in the catalytic activity (Figure S10). An upscaling of the reactionw as performed using variouss ubstrates. Therefore, 5mmolo ft he nitrilesw as converted into the secondary alkylamines on ag ram scale. No decrease of the yield obtained was observed as the reactionw as upscaled (Table S14).
We propose ah ydrogenation-condensation-hydrogenation pathway regarding the reductive alkylation of nitriles with carbonyl compounds as shown in Scheme 2, top. Mechanistici nvestigations were carriedo ut under the given conditions (Scheme 2, bottom), and product and intermediates are obtained in the yields listed. We used milder reactionc onditions to better distinguish between the rates of the individual reaction steps. The hydrogenation of the nitrile proceeds slowly in comparison to condensation and imine hydrogenation and is probably the rate-determining step in the product formation sequence. The benzaldehyde is slowly converted to the corresponding alcohol (23 %u nder the conditions given) if no nitrile is present.N oh ydrogenation of benzaldehyde to the corresponding alcoholw as observed if the reactioni sp erformed in the presence of nitriles.