Semiheterogeneous Dual Nickel/Photocatalytic (Thio)etherification Using Carbon Nitrides

A carbon nitride material can be combined with homogeneous nickel catalysts for light-mediated cross-couplings of aryl bromides with alcohols under mild conditions. The metal-free heterogeneous semiconductor is fully recyclable and couples a broad range of electron-poor aryl bromides with primary and secondary alcohols as well as water. The application for intramolecular reactions and the synthesis of active pharmaceutical ingredients was demonstrated. The catalytic protocol is applicable for the coupling of aryl iodides with thiols as well.

. Examples of APIs containing the alkyl aryl ether motif.
Nucleophilic substitution of alkyl halides with phenolates (Williamson ether synthesis) traditionally used to synthesize these scaffolds, suffers from functional group incompatibilities due to the harsh conditions it requires. Catalytic strategies for the synthesis of ethers were developed to overcome these drawbacks. Copper can catalyze the coupling of phenols and aryl halides (Ullmann-type reaction), 4, 5 as well as aryl boronic acids and alcohols (Chan-Evans-Lam coupling). 6 Palladium catalyzed O-arylations of alcohols via Buchwald-Hartwig type cross-coupling reactions broaden the substrate scope. 7,8 Strongly basic conditions and well-designed biaryl phosphine ligands are required for efficient cross-coupling reactions. 9 Nickel catalysis gained increasing interest due to its significantly higher abundance compared to noble metals. 10,11 The low electronegativity of Ni enables facile oxidative addition into carbon-halide bonds, whereas reductive elimination, especially in case of C-O couplings, is difficult. 12-14 Thermolysis of Ni(II) oxametallacycles, for example, results in βhydride elimination of undesired carbonyl compounds, whereas oxidation to Ni(III) complexes via single electron transfer (SET) with stoichiometric oxidants can induce reductive elimination resulting in C-O bond formation. 15 Combining nickel with photoredox catalysis, [16][17][18] enables the coupling of aryl bromides with alcohols 19,20 or water 21 without stoichiometric SET oxidants. Photoredox catalysis is dominated by expensive, homogeneous iridium and ruthenium complexes. Noble-metal-free, homogeneous photoredox catalysts, such as borondipyrromethene derivatives (BODIPY), 21 require tedious purification procedures and are prone to degradation. 22 Heterogeneous semiconductors are a recyclable alternative to common homogeneous photoredox catalysts. [23][24][25][26] CdSe quantum dots 27 and CdS 28 were used for carbon-heteroatom couplings via the dual nickel/photoredox catalytic approach. Cadmium, however, is among the most toxic elements 29 and strictly regulated. 30 Its application in the synthesis of APIs is therefore not desirable.
Carbon nitride (CN) materials, a class of stable and metal-free semiconductors with low toxicity 31 that can be easily made from commodity chemicals, are able to activate Ni complexes via photosensitization. 32 Here, we show that these materials are also able to catalyze alkyl aryl ether synthesis, likely by triggering reductive elimination via SET modification of the oxidation state of Ni complexes. 19 The etherification of methyl 4-bromobenzoate with 1-hexanol served as a model reaction for initial studies (Table 1). A careful optimization of all reaction parameters showed that a carbon nitride material prepared by polymerization of urea and oxamide (CN-OA-m) 33 in combination with catalytic amounts of NiBr 2 ·3H 2 O and di-tert-butylbipyridyl (dtbbpy) results the selective synthesis of the desired ether (1) after 48 h irradiation with white LEDs in acetonitrile under mildly basic conditions (entry 1-2). The only side-products were small amounts of the corresponding phenol (2) from the cross-coupling with water, and dehalogenated methyl benzoate (3). In homogeneous dual catalysis, the addition of catalytic amounts of quinuclidine was reported to accelerate the reaction. 19,20 For similar reasons, an amine-modified organonickel complex was used in combination with photoredox catalysis for the synthesis of phenols from aryl halides and water. 21 The semiheterogeneous protocol did not result in significant rate enhancement when 10 mol% quinuclidine were added (entries [3][4]. The utilization of 6,6'-diamino-2,2'-bipyridyl instead of dtbbpy drastically reduced the efficacy of the C-O coupling in our model system (see Table S4 in the Supporting Information). A reaction with methyl 4-chlorobenzoate as substrate resulted in very low amounts of the desired ether product under optimized conditions (Table 1, entry 5). No reaction was detected in case of the mesyltate, triflate or tosylate derivatives (see Table S10 in the Supporting Information). Control experiments proved that light, CN-OA-m, NiBr 2 ·3H 2 O, dtbbpy, Ntert-butylisopropylamine (BIPA) and  oxygen-free conditions are essential for successful C-O crosscouplings (entries [6][7][8][9][10][11]. With the optimized conditions in hand, the versatility of the semi-heterogeneous cross-coupling was investigated (Scheme 1). Aryl bromides substituted with electron withdrawing groups in para-position were generally isolated in good to excellent yields. A broad range of functional groups including esters (1), nitriles (4), aldehydes (5), ketones (6,15), phenylboronic acid pinacol esters (14), chlorides (10), and trifluoromethyl-(13) as well as methylsulfonyl-groups (16) were tolerated. Substrates with electron withdrawing meta-substituents (7,8) did also yield the desired products, although with lower efficiency. Ortho-substituted aryl bromides (11,12) resulted in a drastically decreased reactivity. Coupling of 1,4dibromobenzene with 1-hexanol gave a selective monoetherification as the resulting aryl alkyl ether (9) deactivates the second bromide functionality. Heteroaryl bromides (17,18) were successfully coupled under these conditions. Substrates lacking a strong electron withdrawing group gave very low amounts of the desired C-O coupling products (20,21) within 48 hours. While electronic effects dictate the reactivity of aryl bromides, steric effects are responsible for the scope and limitations of aliphatic alcohols (Scheme 2). Coupling of methyl 4bromobenzoate with methanol (22) was completed within 24 hours under standard conditions and within 8 hours when MeOH was used as solvent (see the Supporting Information). The semi-heterogeneous methodology provides an effective method to prepare deuterium labeled anisoles (23) in excellent yield. Primary alcohols with benzyl (24), allyl (26), nitrile (29), trifluoromethyl (25), and tertiary amine (30) groups were coupled with high selectivity. The secondary alcohols isopropanol (31), cyclohexanol (32) and 1-phenylethanol (33) reacted efficiently as well. Sterically encumbered secondary (34,35) and tertiary (36) alcohols resulted in low amounts of the desired ether products within 48 hours. Formation of diaryl ethers from the reaction of aryl bromides and phenols was not observed, presumably due to the low nucleophilicity of aromatic alcohols.
Coupling of methyl 4-bromobenzoate with water as nucleophile gave phenol 2 in moderate isolated yield by switching to DMF as a solvent (for details, see Table S12 in the Supporting Information). The ortho-substituted, electron-rich aryl bromide 2-(2-bromophenyl)ethanol did undergo an intramolecular C-O coupling, resulting in 2,3-dihydrobenzofurane (37). 34 An analogous preparation of chromane and 1,4-benzodioxane was not feasible (see Table S11 in the Supporting Information). The reason for this remains unclear, especially because chromanes were previously synthesized by reductive elimination from the corresponding nickel(II) oxametallacycles. 15 The semi-heterogeneous dual catalytic reaction of 1,4dibromobenzene and isopropylideneglycerol afforded 38, a potential intermediate for the preparation of ketoconazole ( Figure 1), itraconazole, terconazole and their derivatives. 35 The antidepressant Fluoxetine can be synthesized using this method. N-protected 3-methylamino-1-phenylpropanol reacted with 1-bromo-4-(trifluoromethyl)benzene resulting in N-acetyl fluoxetine (39) in 66%. The same catalytic system was evaluated for the coupling of thiols with aryl halides. [36][37][38][39][40] The reaction of aryl bromides and thiols usually requires strongly reducing photoredox catalysts, 39 whereas aryl iodides can be successfully coupled using weaker reductants. 38 When the optimized semiheterogeneous protocol (the conduction band minimum of CN-OA-m was reported to be at -1.6 V vs. Ag/AgCl 33 ) was applied on the reaction of methyl 4-bromobenzoate with methyl 3-mercaptopropionate, only 4% of the desired thioether were formed after 48 hours (Table S15). The analogous reaction using methyl 4-iodobenzoate went to completion within 48 hours resulting in 79% isolated yield of the desired thioether (40). When 2-mercaptoethanol was used, a selective C-S bond formation (41), with no detectable amount of the corresponding etherification product was obtained. In contrast to the C-O coupling, the semi-heterogeneous C-S bond formation is not limited to primary and secondary thiols (40)(41)(42) The reaction rate was significantly increased by using blue LEDs with higher light intensity (for details, see Supporting Information). The coupling of methyl 4-bromobenzoate and methanol, for example, was complete after 16 hours instead of 24 ( Figure 1) using a modified setup. These intensified conditions were used for studying the recyclability of the heterogeneous carbon nitride material ( Figure 2). CN-OA-m was recycled six times, without losing its catalytic activity (Figure 2), proofing its high potential for sustainable photocatalysis. In conclusion, a dual Ni/photocatalytic C-O coupling was developed using a carbon nitride semiconductor as recyclable photocatalyst with low toxicity. The semi-heterogeneous nickel/carbon nitride catalysis is an inexpensive, sustainable alternative to homogeneous protocols. The method selectively couples a broad range of electron-poor aryl bromides with primary and secondary alcohols as well as water in good to excellent isolated yields.

General remarks
Substrates, reagents, and solvents were purchased from commercial suppliers and used without further purification. N-tert-butylisopropylamine (BIPA), 1  The synthesis for CN-OA-m was carried out using a slightly adapted version of the literature procedure (Scheme S1) 10 : for each batch of the photocatalyst, urea (10 g, 166.5 mmol) and oxamide (0.5 g, 5.7 mmol) were mixed in 10 mL of DI water to generate a homogeneous mixture. After drying at 373 K, the resulting solids were grinded, transferred into a crucible with a cover and heated up in an air-oven with a heating rate of 4.3 K/min to 773 K. After keeping the mixture for 2h at 773 K, the sample was allowed to cool to room temperature. Subsequently, KCl (3.3 g, 44.3 mmol) and LiCl (2.7 g, 63.7 mmol) were added and the solids were grinded to obtain a homogeneous mixture which was heated in an inert atmosphere (N 2 flow: 5 mL/min) to 823 K with a heating rate of 4.6 K/min. After keeping the mixture for 2 h at 823 K, the sample was allowed to cool to room temperature and the resulting solids were collected on a filter paper and washed with H 2 O (3 x 100 mL). The resulting yellow material was dried at 373 K (average yield per batch: ~400 mg).
Each batch was tested under the same set of conditions and obtaining always similar catalytic activities (+/-5% based on 1 H-NMR with internal standard).
The cost of CN-OA-m was calculated to be 4.2 € g -1 based on the prices of urea, oxamide, LiCl and KCl from Sigma-Aldrich (Merck). 11 As a comparison, the price of Ir(ppy) 3 is 2124 € g -1 . 11 The UV/Vis spectrum of CN-OA-m shows a strong absorption up to ~460 nm and a comparably weaker absorption band up to ~700 nm ( Figure S1, A) which are attributed to the π-π* electron transition of the sp 2 hybridization of C and N in the heptazine framework and n-π* electron transition involving the lone pairs of the edge nitrogen atoms in the heptazine units, respectively. 10 The capability of harvesting low energy light is therefore superior compared to Ir and Ru photocatalysts (see Figure S1, B for the UV/Vis spectrum of Ir[dF(CF 3 )ppy] 2 (dtbbpy)PF 6 as a representative example) which have only a low absorption S5 band between 400 and 500 nm in the visible region, which corresponds to the metal-to-ligand charge transfer transition.

RGB photoreactor (low intensity)
A flexible, red/green/blue LED strip 12 (RGB, 5m, 24 W/strip; Tween Light, BAHAG AG, Germany) was wrapped around a 115 mm borosilicate crystallization dish ( Figure S2). White light (illumination of all three LED colors -red/green/blue) was used at full power (For emission spectra of a single diode, see Figure S3). The evaporating dish was filled with ethylene glycol and the temperature was set to 40°C to maintain a constant temperature. The sealed reaction vessels were placed at the same distance from the LED strip during all experiments ( Figure S2). All reactions were performed with a stirring speed of 1400 rpm.

440 nm photoreactor (high intensity)
Blue LED lamps 13 at 50% power (440 nm, 40W, PR160, Kessil Photoredox, for emission spectrum, see Figure S5) were used for experiments on CN-OA-m recycling ( Figure S4). Two sealed reaction vessels were placed on a stirring plate 4.5 cm away from a single lamp. To avoid heating of the reaction mixture, a fan was used for cooling. All reactions were performed with maximum stirring speed.   and subjected to 1 H-NMR analysis. For representative NMR spectra, see Figure S6.    Several carbon nitride materials were tested: mesoporous graphitic carbon nitride (mpg-CN), 14    During the study it was observed that the LED strips used in the RGB photoreactor become less efficient as they are used. As reactivity depends on the light intensity, periodic replacement of the LED strips was found to be necessary. Results reported in Table S4 were obtained right after replacing the light source, and are therefore higher than those obtained in the same conditions from the previous experiement (Table S3, entry 5).
6,6'-diamino-2,2'-bipyridyl (Entry 7) was tested because it was reported to improve the reaction when water is used as coupling partner in the nickel/photoredox catalyzed formation of phenols. 18 S15 4.6. Base screening N-tert-butylisopropylamine (BIPA) performed best during the screening of different bases (Table S5).
No conversion of the starting material was detected using common inorganic bases (Table S5, entries 9-13). The addition of quinuclidine (10 mol%, Sigma Aldrich, 13786 € mol -1 ) increased the product formation by only 15% and was therefore not used in the subsequent tests. 11   A reduction of the amount of alcohol was realized using a higher amount of BIPA (Table S7, entry 3). Reducing the amount of the Nickel catalyst resulted in significantly lower conversion within 48 hours.
The CN-OA-m (Table S8, entry 3) can be reduced resulting in slightly lower yields. Since the photocatalyst is inexpensive and recyclable, 3.33 mg ml -1 was maintained as loading for further experiments. S19