Visible‐Light‐Promoted Metal‐Free Synthesis of (Hetero)Aromatic Nitriles from C(sp3)−H Bonds

Abstract The metal‐free activation of C(sp3)−H bonds to value‐added products is of paramount importance in organic synthesis. We report the use of the commercially available organic dye 2,4,6‐triphenylpyrylium tetrafluoroborate (TPP) for the conversion of methylarenes to the corresponding aryl nitriles via a photocatalytic process. Applying this methodology, a variety of cyanobenzenes have been synthesized in good to excellent yield under metal‐ and cyanide‐free conditions. We demonstrate the scope of the method with over 50 examples including late‐stage functionalization of drug molecules (celecoxib) and complex structures such as l‐menthol, amino acids, and cholesterol derivatives. Furthermore, the presented synthetic protocol is applicable for gram‐scale reactions. In addition to methylarenes, selected examples for the cyanation of aldehydes, alcohols and oximes are demonstrated as well. Detailed mechanistic investigations have been carried out using time‐resolved luminescence quenching studies, control experiments, and NMR spectroscopy as well as kinetic studies, all supporting the proposed catalytic cycle.


Introduction
Tr ansition-metal based catalysts have an indispensable role in several chemical reactions [1] industrial production, [1a-e,h] fine and bulk chemical synthesis. [1t] During the last two decades,s everal transition-metal based catalysts [1q, 6] have been developed for the activation of C À Hbonds in sp 3 centers via thermal and photochemical pathways. [1q,2] One of the most important transformations is the conversion of petroleum byproducts into fine chemicals, [1a,b] for example,the synthesis of benzonitrile from toluene. [1b,c] Industrially,b enzonitrile is produced by ammoxidation of toluene using at ransitionmetal catalyst (vanadium) and applying ah igh NH 3 and O 2 pressure at 300-500 8 8C. [1b,c] Since the nitrile moiety is an essential functional group in various drugs and bioactive compounds, [3] as well as an important building block for the preparation of fine and bulk chemicals, [4] novel, milder, (transition)metal catalyzed methods starting from alcohols, [5] amines, [6] aldehydes [7] or more conveniently from simple CH 3 have been developed in recent years. [7f, 8] On example comprises the palladium-catalyzed cyanation of methylarenes using tBuONO as nitrogen source reported by Wang and coworkers in 2013. [8a] While these methods carry ar emarkable impact and are highly valuable,t hey operate in most cases under elevated temperature and employ hazardous reagents or (transition)metals.However,owing to the high price, [9] low abundance [10] or toxicity, [11] of (transition)metals,t he interest in using alternative metal-free catalysts is growing within the scientific community,e specially for the late-stage functionalization of inert C À Hbonds. [12] An intriguing development in this regard is the activation of aromatic compounds by an organic photocatalyst to form nitriles from C(sp 2 )ÀHbonds reported by Nicewicz [13] and others. [14] Yet, this method employs toxic cyanide as stochiometric reagent, which can be ad rawback, espcially in ab ulk-scale synthesis.B esides the use of at oxic cyanide source,af ew photochemical methods for the synthesis of nitriles have been developed using ag reener ammonium salt. [7d,e, 14a] These include the conversion of aldehydes to nitriles using ar uthenium photocatalyst with ammonium persulfate as nitrogen source (Scheme 1A) [7d] or ah eterogeneous Co@g-C 3 N 4 photocatalyst together with NH 2 OH·HCl reported by the Rai group (Scheme 1B). [7e] Another interesting transformation in this regard is the functionalization of styrenes to N-hydroxybenzimidoyl cyanides (Scheme 1C). [14a] Most photocatalytic methods furnishing nitriles in absence of cyanide use pre-functionalized starting materials.H owever,t he conversion of methylarenes to nitriles analogous to the industrial ammoxidation illustrates as impler and more efficient retrosynthetic pathway.
Considering all factors,w ew ondered if ap hotocatalytic, metal-and cyanide free one-step procedure from toluene derivatives to nitriles can be developed. As activation method, we envisioned that methylarene substrates (E 1/2 =+ 2.36 Vv s. SCE for toluene) [15] may be activated by ah ighly oxidizing photocatalysts or by ahydrogen atom abstaction at the benzylic position. Herein we describe the use of the commercially available organic dye 2,4,6-triphenylpyrylium tetrafluoroborate (TPP) [16] for the synthesis of nitriles from methylarenes.I na ddition to this,adetailed mechanistic investigation was carried out to support our mechanistic hypothesis.A ll experimental results and spectroscopic analyses support the proposed catalytic cycle. To the best of our knowledge,t his is the first example for ap hotocatalytic ammoxidation of methylarenes using abundant feedstock materials,ammonium salts and molecular oxygen without the use of metals and toxic reagents.

Results and Discussion
At the start of our investigation, we screened commercially available organic dyes for the desired transformation (Table 1a nd S1). Gratifyingly,t he use of DDQ (PC-1) and acridinium-based photocatalysts (PC-2 and PC-3) gave the desired product with NH 2 OH·HCl as ammonia source and NH 4 Br as additive,h owever only in al ow yield of 12-26 % ( Table 1, entries 1-3). Interestingly,the more sensitive photocatalyst TPP (PC-4) (E 1/2 =+2.55 Vvs. SCE), [17] gave methyl 4-cyanobenzoate (2)i nagood GC-yield of 76 % ( Table 1, entry 4). Screening different ammonium salts,merely hydroxylamine hydrochloride gave the desired product 2,w hile all other tested ammonium sources,s uch as ammonium acetate and aqueous ammonia failed to render the desired product ( Table 1, entries 4-6 and Table S1, entries 5-6). Possible reasons for this observation are that the oxidation of simple ammonia (E 1/2 =+0.63 Vvs. SCE) [18] might be faster than the required oxidation of the substrate and the potential degradation of PC-4 in the presence of ab ase or nucleophiles. Crucially,control experiments revealed the necessity of light, photocatalyst, O 2 and NH 2 OH·HCl for the formation of the desired product (Table S4, entries 2-4 and 7).
Having identified ap otential photocatalyst (PC-4, TPP), the effect of temperature (a), ammonium bromide (b) and hydroxylamine loading (c) was investigated in more detail, including the formation of by-products (Figure 1). Most decisively,alower reaction temperature seems to hamper the water elimination from the intermediate oxime 3 to form nitrile 2 (Figure 1a), while the presence of NH 4 Br is crucial for as atisfactory conversion of 1 ( Figure 1b). As expected, the absence of asuitable ammonium source leads to the overoxidation to the corresponding acid 4 ( Figure 1c). Notably, the conversion of oxime 3 to nitrile 2 only proceeds under photocatalytic conditions and not in the dark under thermal conditions (Figure 1d). Thec omplete optimization process including time,s olvent, additive,c atalyst loading and wavelength variation is given in the supporting information (Table S1-S3 and Figure S3).

Mechanistic Hypothesis
Theinitial results suggest that the oxime is one of the key intermediates,a si ts depletion is accompanied by the formation of the product ( Figure 1). Further,the control experiment yielding no product in the absence of O 2 indicates that the oxime might be formed from the corresponding aldehyde. Based on this we envisioned the following catalytic cycle (Figure 2; Cycle A): Thes ubstrate (E 1/2 (1/1C + ) =+2.45 Vv s. SCE, see SI) is oxidised by the excited photocatalyst (E 1/2 (PC-4 + */ PC-4C) =+2.55 Vv s. SCE) [17] generating radical cation IIC + and the reduced photocatalyst species (PC-4C).
Thep hotocatalytic cycle is closed by single electron transfer from PC-4C to O 2 ,f orming as uperoxide anion (O 2 C À ). [19] Intermediate IIC + loses ap roton to form the more stable benzylic radical (IIIC), which combines with the superoxide anion to peroxide derivative V after protonation. Alternatively,b enzyl radical IIIC can directly add to O 2 to give ap eroxide radical, which may accept an electron from PC-4C to close the photocatalytic cycle and form intermediate V. [20] Elimination of water from V leads to the corresponding Scheme 1. Photocatalytic nitrile formationsu sing ammonium salts. aldehyde intermediate (VI), which yields oxime VII upon condensation with NH 2 OH·HCl. As the oxime is not converted to the final product in the absence of light (Figure 1d), af ollowing photocatalytic cycle is proposed (Figure 2, Cycle A'). Here,the oxime (VII)isoxidised to VIII ·+ via reductive quenching [7e] of PC-4 + *f ollowed by elimination of water to give intermediate IXC + .T he reduced PC undergoes SET with IXC + to render the desired product X,c losing the photocatalytic cycle and completing ar edox-neutral process.A s shown in Figure 1, the presence of NH 4 Br is crucial to obtain as atisfactory yield. Thus,a nother catalytic cycle may be operative,yielding the same reaction product (Figure 2, Cycle B). Bromide anions are known hydrogen atom transfer (HAT) catalysts. [21] In presence of visible-light, the excited PC can oxidise Br À to generate the BrC radical, which is capable to abstract the H-atom from the methylarene to yield ab enzyl radical (IIIC)a nd HBr. Thec ombination of superoxide anion and IIIC yields intermediate IV À ,w hich deprotonates HBr to regenerate Br À and form intermediate V.After its formation, V follows the same mechanistic pathway as described above (Cycle Aand A').
To support this proposed hypothesis,w ep erformed several mechanistic experiments,s tarting with spectroscopic investigations.T ime-resolved luminescence quenching ex-    (1)can be oxidized by the excited TPP (PC-4) ( Figure S4a,b), as aslight decrease of the luminescence lifetime with increasing concentration of 1 was observed (K SV = 0.283 M À1 ). Further,n o life-time quenching was detected with product 2 or proposed aldehyde intermediate VI ( Figure S4c). On the other hand, the oxime intermediate (3)s howed as uperior quenching to methyl 4-methylbenzoate (1)( Figure S4c, K SV = 25.9 M À1 ), indicating an interaction of the excited photocatalyst with the oxime intermediate to form the desired product under the applied reaction conditions (Cycle A'). To support that an electron transfer is feasible,c yclic voltammetry measurements revealed ap otential of (E 1/2 (3C + /3) =+1.80 Vv s. SCE, see SI) for the oxime,which lies within the oxidation window of PC-4. Theq uenching efficiencies of NH 2 OH·HCl and NH 4 Br were investigated as well, with the former (K SV = 23.5 M À1 ) exhibiting ap oorer quenching than the later one (K SV = 75 M À1 )( Figure S6), which suggests that Cycle Bc an be active as well. Comparing the luminescence lifetime quenching of NH 4 Br and the methyl 4-methylbenzoate starting material (1), NH 4 Br seems to decrease the lifetime much more effectively ( Figure S7). However,a ne lectron-rich substrate (4-methyl anisole) opposed to electron-poor substrate 1 proved to be ap otent quencher as well ( Figure S8, K SV = 83.7 M À1 ), showing aslightly superior quenching ability to NH 4 Br. Thus,b ased on the time-resolved luminescence quenching experiments,both proposed catalytic cycles Aand Bc an be operative,w ith their respective importance likely being dependent on the electronic nature of the starting material.
Next, several mechanistic control experiments were performed to directly or indirectly detect reaction intermediates vital for the mechanistic process (Scheme S2). Themodel reaction was performed under standard reaction condition in presence of TEMPO,y ielding no product, which indicates ar adical pathway (Scheme S2, Experiment-1). Further,t he radical cation intermediate (IIC + )c ould be trapped in the presence of pyrazole or 4-cholorpyrazole as nucleophiles under standard reaction condition, in as imilar reaction as reported by Nicewicz, [22] supporting its formation (Scheme S2, Experiment-2). Looking at the more stable intermediates,t he methylarene is proposed to be oxidized to the corresponding aldehyde (VI). Theg eneration of an Scheme 2. Substrate scope for the synthesis of functionalizedc ycanobenzenes. Reactionsc onditions: a 0.1 mmol substrate, 20 mol %P C, 3equiv. NH 2 OH·HCl, 2.5 equiv.NH 4 Br,2 5mg4 MS, 1bar O 2 ,2mL acetonitrile (0.05 M), 455 nm, 40 8 8C, 24 h, isolated yields. b GC yields using n-decane as standard. c Same as "a" using 1equiv.o fPTSCl. d Same as "a" 48 hinstead of 24 h. aldehyde intermediate is supported by its detection when avoiding an ammonium source (Scheme S2, Experiment-3).
Further,a ldehydes as well as oximes could be used as starting materials for the synthesis of the final nitrile product (Scheme S2,. In both cases,product 2 was formed with ay ield of 91 %a nd 80 %, respectively.N otably, the presence of O 2 is not required in both cases,w hich is in accordance to the proposed reaction mechanism (Figure 2, Cycle A'). Ther eaction progress could be followed by NMR as well ( Figure S15-18), clearly showing the formation of the oxime followed by the product formation under the applied conditions.A sm entioned above,t he oxime could not be converted to the nitrile under thermal conditions.T oindicate the necessity of more than one photon for the formation of one product molecule,t he product yield dependent on the irradiation intensity was investigated (see SI, Figure S19). [23] Collectively,t he described control experiments and spectroscopic investigations all support the proposed photocatalytic cycles.Inaddition, the catalyst deactivation pathway was also studied under the standard reaction conditions,s uggesting 2,4,6-triphenylpyridine as the degradation product, which could further be observed in the scale-up batch as minor byproduct supported by HRMS (Scheme-S3).
With the successful conditions in hand, we explored the C(sp 3 )-H functionalization of different methylarenes (Scheme 2). Substrates bearing electron-donating and -withdrawing groups,a sw ell as heterocycles,g ave the respective products in good to excellent yields (Scheme 2, 5-39). Functional groups such as carboxylic acid, ester,a mide,h alogens, cyanide and boronic ester were untouched during the reaction (Scheme 2, 5-6, 9-12, 15-18 and 22-23). Applying this methodology,adinitrile could also be synthesized in ao nepot procedure with am oderate yield (24). Structurally complex, bioactive-and drug-molecules could be employed for as elective cyanation as well, rendering the desired product in good to excellent yields up to 76 %( Scheme 2, 29-39). Delightfully,s ubstrates bearing multiple oxygen atoms,w hich are usually not stable under photochemical conditions in the presence of oxygen, are viable,too. [24] Using the developed protocol, sugar derivatives 29 and 38 were obtained in 62 %a nd 60 %y ield, respectively.F urther, more challenging substrates such as cholesterol, isoborneol, amino acid and peptide derivatives gave the corresponding products in 45-70 %y ield. Interestingly,a nti-inflammatory drug (Celecoxib, 37)g ave the desired product in 70 %y ield. Fort he example of ab ulk-scale preparation, compound 2 and Celecoxib were cyanatedi na1gs cale to give the desired product in 65 and 60 %yield, respectively (Scheme S1). After the screening of methylarenes,w ew ere interested to apply this methodology using alcohols,a ldehydes and oximes as starting materials for the synthesis of aromatic nitriles as well (Scheme 3). Similar to methylarenes,alcohols,aldehydes and oximes gave good to excellent yields up to 91 %. Notably, isophthalaldehyde (Scheme 3) was cyanided twice to isophthalonitrile (44)i n7 8% yield. Sterically crowded 2,6dichlorobenzaldehyde (Scheme 3) was aviable substrate,too, giving 2,6-dichlorobenzonitrile (DCBN, 45)i n8 2% yield, which is used as herbicide and regarded as ap otential intermediate for pesticides and agrochemicals.

Conclusion
In conclusion, we present the first metal-and cyanide-free visible-light-induced photocatalytic ammoxidation of C(sp 3 ) À Hb onds using an abundant ammonia source and molecular oxygen. Adetailed mechanistic investigation was carried out to support the proposed mechanistic hypothesis including various spectroscopy experiments.A pplying this methodology,m ore than 50 aromatic and heteroaromatic substrates, as well as steroids and existing drug molecules containing methyl groups could be converted to the nitrile in good to excellent yields.I na ddition to this,t he method could be executed on gram scale and alcohols,a ldehydes and oximes could be used as starting materials for their conversion to nitriles in up to 91 %yield in the same manner.