CpNiBr(NHC) complexes as pre-catalysts in the chemoselective anaerobic oxidation of secondary aryl alcohols: Experimental and DFT studies

Molecular Catalysis 432 (2017) 47–56

Contents lists available at ScienceDirect

Molecular Catalysis journal homepage: www.elsevier.com/locate/mcat

CpNiBr(NHC) complexes as pre-catalysts in the chemoselective anaerobic oxidation of secondary aryl alcohols: Experimental and DFT studies Frederick P. Malan, Eric Singleton, Bryan W. Bulling, Ignacy Cukrowski, Petrus H. van Rooyen, Marilé Landman ∗ Department of Chemistry, University of Pretoria,02 Lynnwood Road, Hatfield, Pretoria, 0002, South Africa

a r t i c l e

i n f o

Article history: Received 25 October 2016 Received in revised form 26 December 2016 Accepted 28 December 2016 Keywords: Nickel N-heterocyclic carbene Alcohol oxidation Anaerobic

a b s t r a c t Nine Ni(II)-NHC complexes of the type [CpNiBr(NHC)], each containing either a symmetric or asymmetric alkyl/benzyl/phenylethyl imidazolium ligand, effectively catalysed the anaerobic oxidation of a series of secondary alcohols, including 1-phenylethanol to yield acetophenone. Catalytic reactions performed using KOt Bu as a mild base and 1,2-dibromobenzene as both oxidant and solvent cleanly converted secondary alcohols to their corresponding ketones in high yield using conventional heating, and almost quantitatively using microwave heating. Catalytic activity of the nine [CpNiBr(NHC)] complexes were evaluated separately under optimized conditions, where the more electron-donating NHC-bearing Ni complexes (1, 2, and 5) proved to be more efficient than the electron-witdrawing NHC ligand analogues (6, 8, and 9). Substrate screening studies revealed that secondary alcohols bearing electron-withdrawing functional groups were less efficiently oxidized. Catalyst concentration optimization studies were aimed at finding a catalyst concentration to limit the effects of catalyst deactivation, while also maintaining reasonable TONs. Additional characterization of selected Ni(II) complexes have been included, comprising three single crystal X-ray structure determinations, and a DFT study on the mechanistic reaction pathways of the alcohol oxidation reaction as well as ␣-ketone arylation, a potential secondary competitive catalytic reaction. © 2017 Elsevier B.V. All rights reserved.

1. Introduction The oxidation of secondary alcohols remains a topic of interest for both organic and inorganic chemists in the synthesis of sought after carbonyl-containing molecules [1–4]. While the majority of alcohol oxidation reactions are focused on employing molecular oxygen as oxidant (aerobic oxidation), where quantitative conversions may be achieved using low catalyst concentrations with fast reaction times, the potentially dangerous conditions under which these reactions occur remain to be a concern [1,2,4,5]. The anaerobic oxidation of primary and secondary alcohols only received considerable attention during the last decade, despite it being a much milder and safer route to yield functionalised ketones and aldehydes in high yields [4,6]. The elimination of the need to perform oxidations at high O2 pressures and high temperatures, as well as the lack of formation of reactive species that give rise to

∗ Corresponding author. E-mail address: [email protected] (M. Landman). http://dx.doi.org/10.1016/j.mcat.2016.12.023 1381-1169/© 2017 Elsevier B.V. All rights reserved.

side-reactions, allows for an attractive route for which large-scale application is possible [1,6]. Many catalytically active zero-valent metal complexes require ligands to help stabilise these sensitive species and prevent catalyst decomposition [2,3,7–9]. N-heterocyclic carbenes (NHCs) remain to be a favourite class of ancillary ligands that provide additional steric- and electronic ligand tuning options [8–20]. This is believed to be due to the stronger binding of NHCs to metal centres, which prevents rapid degradation under oxidizing conditions, as is the case with most common ligands [21,22]. The oxidation of secondary alcohols has successfully been performed using mainly Pd [3,7,21,23–27] and Ru [5,28,29] NHC-bearing organometallic systems. The employment of the significantly cheaper Pd analogue, Ni, was only realized in 2009 for Ni(II)-NHC complexes [1,4], and in 2013 using low-coordinated Ni(0) nanoparticles supported on alumina [6]. Furthermore, only one research group has actively investigated the anaerobic oxidation of alcohols using the catalytically versatile [CpNiX(NHC)] (X = halide) complexes [1,14,15,22]. During this process a secondary alcohol is dehydrogenated and used as hydrogen donor to an aryl halide to form the corre-

48

F.P. Malan et al. / Molecular Catalysis 432 (2017) 47–56

sponding secondary ketone and unreactive dehalogenated arene as by-products [22,27]. We have recently expanded the existing limited range [30–34] of [CpNiBr(NHC)] complexes to include bulky, flexible Ni(II)-NHC ligands that improve the efficiency of these complexes to partake in catalytic addition and elimination processes [35]. Here we evaluate the catalytic activity of nine [CpNiBr(NHC)] complexes in the oxidation of a range of secondary alcohols using KOt Bu as base and 1,2-dibromobenzene as oxidant. This serves as the first catalytic report on the use of [CpNiBr(NHC)] complexes in the oxidation of secondary alcohols to form their corresponding ketones. Furthermore, we evaluate the effect of catalyst concentration, and employ a DFT study to help elucidate the favoured secondary alcohol oxidation mechanism over the subsequent ␣-ketone arylation mechanism. 2. Material and methods 2.1. General All experiments were carried out under an argon atmosphere using standard Schlenk techniques. Solvents were dried prior to use using standard techniques [36,37]. Column chromatography was carried out under inert argon atmospheres using silica gel (particle size 0.063–0.200 mm) as the stationary phase. The synthesis, purification, and characterization of complexes [CpNiBr(NHC)] 1–9 (Fig. 1) were recently reported by our group [35]. All other chemicals were purchased from Sigma-Aldrich and used without further purification. GC analyses were carried out on a Shimadzu GC-2010 fitted with a flame ionization detector (FID) using a TRB780143 capillary column (30 m, 0.25 mm ID, thickness 0.25 ␮m) and an AOC-20i auto injector. Mass spectrometry measurements were performed using a Hewlett Packard (HP) GC 1530A coupled to an Agilent 5975C mass selective detector (MSD). 2.2. General synthesis of [CpNiBr(NHC)] complexes (1-9) The syntheses of complexes 1-9 were accomplished as described in literature [31]. A suspension of the alkyl/aryl imidazolium bromide (3 mmol) in THF (10 mL) with [Ni(C5 H5 )2 ] (0.57 g, 3 mmol) was heated under reflux between 3–16 h (depending on the NHC ligand). The reaction mixture was concentrated in vacuo, and purified by silica gel column chromatography, using gradient elution with hexane and dichloromethane. The red bands were collected and concentrated in vacuo to give red to red-brown powders in relatively high yields.

Retention times (min): acetophenone (5.560), ±1-phenylethanol (4.316), 4’-chloroacetophenone (6.179), 4’-chlorophenylethanol (5.565), benzophenone (16.287), benzhydrol (14.967), isonitrosoacetophenone (14.047), ␣-(hydroxyimino)acetophenone (10.869), 4’-bromophenacylbromide (7.792), 2-bromo-(4’bromophenyl)ethanol (7.345). 2.4. Computational methods All calculations were carried out using DFT with the B3LYP hybrid functional [38,39], and were implemented in the gas phase using the Gaussian 09, Revision D01 program [40]. The triple-␨ basis set 6-311G* basis set was used for all atoms (C, H, N, Br, and O), whereas the Stuttgart/Dresden (SDD) pseudopotential was used to describe the nickel metal electronic core, while the metal valence electrons were described using the def2-TZVPP basis set [41]. The gas phase geometries of all intermediates and transition states were fully optimized at this level of theory without any symmetry restrictions, ensuring that the local minima had zero imaginary vibrational frequencies [42] and the transition state had exactly one and to provide the thermal correction to free energies at 298.15 K and 1 atm. Solvent effects have also been included for the theoretical TOF calculations where all intermediates and transition state have been re-optimized at the same level of theory and calculated using defined properties (dielectric constant, molar volume, solvent radius, and density) for 1,2-dibromobenzene as the solvent. The HOMOs and LUMOs of all reaction intermediates, as well as the energy gaps were calculated from the same DFT method and level of theory. 2.5. X-ray crystallography of 5, 7, and 8 Single crystal diffraction of compounds 5, 7, and 8 were done using Quazar multi-layer optics monochromated Mo K␣ radiation (k = 0.71069 Å) on a Bruker D8 Venture kappa geometry diffractometer with duo I␮s sources, a Photon 100 CMOS detector and APEX II control software [43]. All x-ray diffraction measurements were performed at 150(2) K. Data reduction was performed using SAINT+ [43], and the intensities were corrected for absorption using SADABS [44]. All structures were solved by direct methods with SHELXS-97 [45] using the OLEX2 [46] interface. All H atoms were placed in geometrically idealized positions and constrained to ride on their parent atoms. For tables containing the data collection and refinement parameters, see Supplementary Information. 3. Results and Discussion

2.3. General procedure for the alcohol oxidation reactions To a round-bottom flask containing the secondary alcohol substrate (0.3 mmol), KOt Bu (2.6 eq., 0.7 mmol), 1,2-dibromobenzene (3 mL), was added [CpNiBr(NHC)] (3 mol%, 9 ␮mol) and the subsequent reaction mixture was purged with argon before being stirred at 110 ◦ C for 120 min before the mixture was left to cool. The resulting mixture was passed through a pad of silica, from which 2 mL of the resulting light yellow solutions were analysed by GC using heptane as internal standard. All products were confirmed by GC-MS, and by comparison of their GC retention times with those of the authentic samples. All conversions and concentrations were based on both consumption of starting material and the internal standard present, and are based on the average of three runs. The effluent was combusted using a H2 /air flame and detected using a FID. The following general GC conditions were employed: injection temperature, 250 ◦ C; initial column temperature, 80 ◦ C; initial hold time, 1 min; ion source temperature, 200 ◦ C, interface temperature 250 ◦ C.

3.1. Synthesis of cyclopentadienyl nickel(II) bromide NHC complexes In our previous report we described the synthesis and full characterization of a range of [CpNiBr(NHC)] complexes bearing flexible symmetric- and asymmetric NHC ligands. Each of these NHC ligands bears N-alkyl, −benzyl, and/or −phenylethyl moieties that exhibit steric bulk through phenyl groups, as well as flexibility through sp3 methylene carbon linkers (Fig. 1). Briefly, to a slurry containing the imidazolium bromide ligand in dry THF, one equivalent of nickelocene was added, and the resulting mixture was heated under reflux for 3–16 hours (depending on the NHC salt) [35]. After the loss of a cyclopentadiene ligand, both the neutral carbene ligand and bromide ligand coordinate to the nickel centre. All 1 H-NMR spectra of complexes 1-9 exhibited the disappearance of the acidic imidazolium (NCHN) proton (9.3-10.5 ppm) as well as the appearance of a significantly downfield shifted carbenic carbon on complexation (161.3-166.9 ppm from the 13 C-NMR spectra).

F.P. Malan et al. / Molecular Catalysis 432 (2017) 47–56

49

Fig. 1. The range of synthesised [CpNiBr(NHC)] complexes 1–9.

Fig. 3. (a) Perspective view of [CpNiBr{Im(Bn)(4-NO2 Bn)}] (6) with thermal ellipsoids drawn at 50% probability level. (b) Packing diagram of 6 as viewed along the a axis. Hydrogen-bromide and hydrogen-oxo interactions are indicated with red and teal lines. Fig. 2. (a) Perspective view of [CpNiBr{Im(EtPh)2 }] (5) with thermal ellipsoids drawn at 50% probability level. (b) Packing diagram of 5 as viewed along the c axis. Hydrogen-bromide interactions are indicated with red and teal lines.

3.2. X-ray crystallographic studies The molecular structures of complexes [CpNiBr{Im(EtPh)2 }] (5, Fig. 2), [CpNiBr{Im(Bn)(4-NO2 Bn)}] (6, Fig. 3), and [CpNiBr{Im((CH2 )2 Ph)(4-NO2 Bn)}] (8, Fig. 4) all serve as representative examples of the solid-state molecular geometry of the cyclopentadienyl, NHC, and bromo-ligands around the nickel centre, and how flexibility of the N-substituents influence the catalytic efficiency of the Ni(II) complexes. All crystallographic data and data collection parameters are included in the supplementary information. Complexes 5, 6, and 8 may each be considered to exhibit a distorted trigonal planar geometry around the central metal atom, with Cpcent -Ni1-Br1 ≈ Cpcent -Ni-C6 ≈ 133◦ . Typical bond lengths include Ni1-C6 (1.881(3)-2.15(2) Å), Ni1-Br1 (2.334(2)-2.355(5) Å), and Ni1-Cpcent (1.759(2)-1.768(2) Å), and compare well with similar [CpNiBr(NHC)] complexes [32–35,47–49]. Other comparable descriptors include the bond

angles Br1-Ni1-C6 (94.46(8)-161.0(5)◦ ), N1-C6-Ni1 (127.2(3)129.6(2)◦ ), N2-C6-Ni1 (127.8(3)-127.9(7)◦ ). The flexibility in the range of NHC ligands employed is clearly demonstrated through the methylene- and ethylene-phenyl containing N-substituents of 5, 6, and 8, through the C␤ -C␣ -N1 (112.4(7)-113.8(2)◦ ) and C␤ -C␣ -N2 (110.3(7)-112.8(3)◦ ) (C␣ and C␤ are the primary and secondary methylene linker carbon atoms) bond angles, as well as the C␤ -C␣ -N1-C6 (from −103.6(3) up to 95.9(5)◦ ), C␤ -C␣ -N2-C6 (from −140.3(8) up to 98.8(3)◦ ), C␣ -N1-C6-Ni1 (from −13(1) up to 0.7(6)◦ ), C␣ -N2-C6-Ni1 (from 1.2(4) up to 10(1)◦ ) torsion angles. In sterically demanding environments the N-alkyl/-aryl groups tend to “fold-out”, such that the bulky moieties are directed away from the metal centre. Interestingly, in 5 it is seen that the N2phenethyl chain is directed towards the nickel centre, whereas the N1-phenethyl chain is directed away through radically different C␥ -C␤ -C␣ -N2 (-74.1(5)◦ ) and C␥ -C␤ -C␣ -N1 (-178.3(3)◦ ) torsion angles (C␥ refers to the ipso carbon of the phenyl group adjacent to the ethylene linker). A similar structural effect is also observed in 6 and 8, where the 4-NO2 -benzyl fragment is directed towards

50

F.P. Malan et al. / Molecular Catalysis 432 (2017) 47–56

Fig. 4. (a) Perspective view of [CpNiBr{Im((CH2 )2 Ph)(4-NO2 Bn)}] (8) with thermal ellipsoids drawn at 50% probability level. One molecule of CH2 Cl2 has been omitted for clarity. (b) Packing diagram of 8 as viewed along the a axis. Hydrogen-bromide interactions are indicated with red and teal lines.

the nickel centre, whereas the remaining benzyl and phenylethyl fragments are directed away from the metal centre. In Figs. 2–4, short contact interactions are observed between the bromide ligand, and the hydrogen atoms of the imidazolyl backbone as well as the methylene/ethylene linkers. This stabilisation effect is certainly only observed in the solid state. This is due to solvation effects, followed by dehalogenation as one of the first catalytic steps towards the formation of the catalytically active species. 3.3. Secondary alcohol oxidation activity The anaerobic oxidation of secondary alcohols has been studied for several years now, with various synthetic and catalytic routes that exist in efficiently producing key ketones [1,3,4]. NickelNHC complexes, despite exhibiting competitive catalytic activities as pre-catalysts, have received little attention compared to the well-studied platinum group metals. We employed our recently reported series of [CpNiBr(NHC)] complexes as pre-catalysts in the oxidation of secondary alcohols using 1,2-dibromobenzene as oxidant. This is the first account of [CpNiBr(NHC)] complexes employed in the anaerobic alcohol oxidation reactions, whereas only four [CpNiCl(NHC)] complexes of the [CpNiX(NHC)] (X = halide) series have ever been employed [14]. Complexes 1–9 catalysed the oxidation of racemic 1-phenylethanol efficiently, and gave high yields of the corresponding acetophenone. Complex 1, [CpNiBr{Im(Bn)2 }], was employed as the catalyst for the optimization of the anaerobic oxidation of ±1-phenylethanol (Table 1). We initially employed similar reaction conditions to Navarro et al.[14], that is, stirred suspensions of the secondary alcohol in 1,2-dichlorobenzene (3 mL) with 2.6 eq. KOt Bu, and 3 mol% catalyst at 110 ◦ C for 120 minutes. The use of 1,2-dibromobenzene as solvent and oxidant instead of 2,4-dichlorotoluene, which was used in the reported method [15], allowed for easier oxidative addition onto Ni(0) species. These changes were employed without sacrificing catalytic efficiency and substrate conversion, despite conventional heating methods that were employed (entry 5). Lower conversions were obtained compared to those reported by Navarro et al. [14] for the given optimized temperature (110 ◦ C) and reaction time (120 min). They employed microwave heating at the same temperatures in 30 min to give almost quantitative conversions. However, the same Ni-NHC systems of Navarro et al. were able to reach secondary alcohol conversions of up to only 5% when conventional heating was applied. The effect of catalyst concentrations

Fig. 5. Anaerobic alcohol oxidation using complexes 1-9. Reaction conditions: ±1-phenylethanol (0.3 mmol), KOt Bu (0.78 mmol), catalyst (3 mol%), 1,2-dibromobenzene (3 mL), 110 ◦ C, 120 min. TOF (h−1 ) indicated in the graph is the initial TOF (h−1 ) calculated based on the amount of substrate converted within the first 5 minutes of the reaction. All GC yields are based on internal standard, average value of three runs.

lower than 3 mol% was observed in dramatic lower alcohol conversions (entries 1 and 2). The mild base KOt Bu was more effective in this study than K3 PO4 (entry 4), which was found to be superior in similar aryl halide functionalization and C C coupling reactions [35,47–49]. Separate decreases in temperature (entries 6 and 7) and reaction time (entry 8) lead to the expected decrease in alcohol conversion, whereas slightly lower equivalents of KOt Bu present in the reaction mixture did not notably affect alcohol oxidation product formation, or lead to any side-product formation (entry 9). Complexes 1–9 were all catalytically active in the anaerobic oxidation of racemic 1-phenylethanol. Screening of each separate complex in the catalytic reactions under optimized conditions showed that electronic and steric differences present in the Nsubstituents of the coordinated NHC ligands affect the performance and more notably the efficiency of the catalysts (Fig. 5). Complexes 1, 2, and 5 were the most active in the oxidation of ±1-phenylethanol to give high conversions (>80%) within 2 h. A gradual decrease in oxidation activity is observed as the electron-donating character of the coordinating NHC ligand changes. The complexes bearing NHCs with electron-withdrawing 4-NO2 -benzyl moieties showed the poorest activity, with complex 9, [CpNiBr{Im(4-NO2 Bn)2 }], only converting 63% of the secondary alcohol. This result was found to be directly related to the N-alkyl/benzyl/phenylethyl groups of the different Ni(II)-NHC complexes involved in each separate reaction, where the more electron-withdrawing aryl groups tend to destabilise the electrondonating character of the NHC ligand. This leads to a metal complex with less electron density to help stabilise air and moisture sensitive catalytically active Ni(0) species. This effect was also observed by the groups of Ritleng [48] and Buchowicz [47], where they found that electron-donating, bulky N-substituents lead to more stable and catalytically active nickel species. Furthermore, the sp3 carbon linkers on the N-substituents help provide flexibility to the relatively bulky phenyl-containing substituents to allow for unhindered substrate coordination and enhanced reductive elimination steps in the catalytic cycle [32]. The presence of aryl halides in the reaction mixtures may also provide additional stabilisation through soft ␲-interactions of the N-benzyl and phenylethyl groups, as well as close-contact hydrogen-halide interactions. A similar trend is also observed in the initial TOFs of complexes 1–9, where complexes 1, 2, 3, 5, 6, and 7 were all very efficient during the first five minutes of the reaction. Complexes 4, 8, and 9 were all much less effective, resulting in lower atom efficiency of the

F.P. Malan et al. / Molecular Catalysis 432 (2017) 47–56

51

Table 1 Optimization of reaction conditions in the oxidation of ±1-phenylethanol. Entry 1 2 3 4 5 6 7 8 9

Cat. (mol%) 1 (0.5) 1 (1) 1 (3) 1 (3) 1 (3) 1 (3) 1 (3) 1 (3) 1 (3)

Solvent/Oxidanta 1,2-DBB 1,2-DBB 1,2-DBB 1,2-DBB 1,2-DCB 1,2-DBB 1,2-DBB 1,2-DBB 1,2-DBB

Base t

KO Bu KOt Bu KOt Bu K3 PO4 KOt Bu KOt Bu KOt Bu KOt Bu KOt Bu

Time (hrs)

Temp (◦ C)

Base (equiv)

Conv. (%)b

2 2 2 2 2 2 2 1 2

110 110 110 110 110 80 25 110 110

2.6 2.6 2.6 2.6 2.6 2.6 2.6 2.6 2.0

9 56 84 49 65 58 30 76 82

General reaction conditions: ±1-phenylethanol (0.3 mmol), KOt Bu (0.78 mmol), catalyst (1, 3 mol%), 1,2-dibromobenzene (3 mL). a 1,2-DBB = 1,2-dibromobenzene. 1,2-DCB = 1,2-dichlorobenzene. b GC yields based on internal standard, average value of three runs.

Fig. 6. Time resolved plots of acetophenone conversion (%) using complexes 1-9. Reaction conditions: ±1-phenylethanol (0.3 mmol), KOt Bu (0.78 mmol), catalyst (3 mol%), 1,2-dibromobenzene (3 mL), 110 ◦ C, 120 min.

catalyst and lower substrate conversion. It is interesting to note that although the latter complexes exhibited low TOFs, the eventual substrate conversions were not severely influenced and might indicate slow(er) catalyst decomposition. From the time resolved plots of ±1-phenylethanol conversion, more insight is obtained into catalyst decomposition and catalyst performance limitations (Fig. 6). An initial fast conversion of alcohol is observed for all reactions (first 15 min), after which a gradual relaxation in the slopes are observed to reach an apparent reaction time of about 60 min. Thereafter the effects of catalyst decomposition become noticeable, inhibiting further conversion activity. Complexes 1–9 have been divided based on activity as insets in Fig. 6 for clarity. From these it can be seen that complexes 1, 2, and 5 were the most active, followed by 3, 4, and 7, and finally 6, 8, and 9. These observations strongly correlate to the overall conversions and TOFs attained using each of these catalysts. A reaction time of 60 minutes could be considered fast for conventionally heated Nicatalysed alcohol oxidation reactions [1], especially without added additives such as PPh3 [22,48].

Using 1, several other secondary alcohol substrates were also employed for oxidation under optimized conditions (Table 2). The similar substrate ±1-(4’-Cl)-phenylethanol was oxidized to 4-chloroacetophenone in slightly lower yields as compared to ±1-phenylethanol (entry 2), with even less of the more bulky benzhydrol converted to benzophenone (entry 3). This is surprising since a lower conversion has been associated with substrates bearing electron-donating para-substituents [50]. This is indicative of the important influence the electronic nature of the aromatic ring has in oxidation reactions. Interestingly, the presence of an aryl halide in the alcohol substrates (entries 2, 5) did not lead to any side product formation under optimized conditions, despite the possibility of the aryl halide-bearing alcohol or their oxidized species undergoing oxidative addition onto Ni(0) species present in the reaction mixture [17,48,51]. The relationship between secondary alcohol conversion over reaction time at various catalyst concentrations is illustrated in Fig. 7. The strong dependence of catalyst concentration on the conversion of the alcohol substrate is indicative of a relatively fast catalyst deactivation process [48]. At concentrations lower

52

F.P. Malan et al. / Molecular Catalysis 432 (2017) 47–56

Table 2 Conventional and microwave-heated oxidation of secondary alcohols. Time (hrs)

Temp (◦ C)

Conv. (%)b

1,2-DBB

2

110

84 (98)

1(3)

1,2-DBB

2

110

79 (97)

3

1 (3)

1,2-DBB

2

110

71 (96)

4

1 (3)

1,2-DBB

2

110

52 (90)

5

1 (3)

1,2-DBB

2

110

60 (94)

Entry

Cat. (mol%)

Solvent/Oxidanta

1

1 (3)

2

Substrate

Product

General reaction conditions: alcohol (0.3 mmol), KOt Bu (0.78 mmol), catalyst (1, 3 mol%), 1,2-dibromobenzene (3 mL), 110 ◦ C, 120 min. a 1,2-DBB = 1,2-dibromobenzene. b GC yields based on internal standard, average value of three runs. Values in parentheses are conversions when microwave heating (500W) was used.

3.4. Mechanistic and computational considerations

Fig. 7. Effect of catalyst concentration on conversion of ±1-phenylethanol. General reaction conditions: ±1-phenylethanol (0.3 mmol), KOt Bu (0.78 mmol), catalyst (1), 1,2-dibromobenzene (3 mL), 110 ◦ C, 120 min.

than 3 mol%, a significant decrease in alcohol conversion is seen, whereas a saturation point in activity and atom efficiency is quickly reached at catalyst concentrations higher than 5 mol%. At these high catalyst concentrations the turnover number (TON) (see Supplementary Information) and turnover frequency (TOF) of the reactions decreases dramatically, leading to an inefficient and expensive catalytic reaction.

In the seminal article by Navarro et al. [27], they found that when 2 or more equivalents of the aryl chloride (oxidant) are used, a subsequent domino ␣-ketone arylation takes place once the ketone has formed. These ␣-arylated products were obtained in high yields (>76%) when 2 mol% of [PdCl(allyl)(iPr)] (iPr = 2,6-diisopropylimidazolidene) was used at 80 ◦ C with 2.2 eq of KOt Bu. They did not, however, observe formation of these species when complexes [CpNiCl(NHC)] (NHC = IMes, SIMes, IPr, SIPr) were employed [14], which is in full agreement with our results. In the anaerobic alcohol oxidation mechanism (Path X, Scheme 1), the [CpNiBr(NHC)] complex is reduced and dehalogenated by the base present, after which the aryl halide present undergoes oxidative addition to the Ni(0) species (D) to form the Ni(II) species (A). The secondary alcohol is deprotonated by the excess of base present, which substitutes the bromide anion on the nickel(II) centre to form B, through which, after ␤-hydride elimination, the ketone formed is expelled from the nickel centre, leaving the nickel(II)-hydride species C. Reductive elimination of the aryl group leads to the D species, which after oxidative addition of another aryl halide molecule regenerates A. It has been shown that the ␲ complexation of haloarenes with Ni(0) is irreversible and could initiate at any C-C bond of the aromatic ring [52]. A series of ringhopping steps take place before the final addition process on the preferred aromatic C-X bond. Furthermore, species A may then also react with the ethenoxide (formed from the ketone, Path

F.P. Malan et al. / Molecular Catalysis 432 (2017) 47–56

53

Scheme 1. Mechanistic pathways for the anaerobic alcohol oxidation (Path X), and domino ␣-ketone arylation (Path Y).

Fig. 8. Relative Gibbs free energy profile for the Ni(II) singlet state of Paths X and Y.

Y) to form species E, followed by a tautomeric shift to oxidize the ethenoxide to a ␴-bonded 2-oxo-2-phenylethyl moiety (F), through which after reductive elimination yields D and 2-(2-bromophenyl)1-phenylethanone as the organic product. Gibbs free energy calculations on 1 and its catalytic intermediates were carried out to support the two proposed paths and provide insight into the high selectivity of Path X over Path Y. In Figs. 8 (singlet states) and 9 (triplet states) the relative Gibbs free energy over the course of Paths X (red) and Y (blue) have been illustrated. Both spin multiplicities with respect to the nickel species involved in the mechanistic pathways X and Y (Scheme 1) were taken into consideration, since no unambiguous assignment could be made given the chemical and geometrical nature of the species. Furthermore, notable energy discrepancies between identical species calculated using different spin multiplicities required further investigation and revealed that the lowest energy species were calculated for 10 species using singlet multiplicities, and

7 species using triplet multiplicities. Recognising that Ni(II) (d8 ) species usually exhibit the triplet state (tetrahedral arrangement) and the singlet state (square planar arrangement) as favourable multiplicity states, and that formally no triplet state is expected for Ni(0) (d10 ) species, both spin multiplicities for Ni(II) and Ni(0) were considered for completeness. Indeed, in selected cases we found that the singlet/triplet multiplicity for Ni(II)/Ni(0) species resulted in lower Gibbs energies. The crossover to the unusual singlet state for Ni(II) species is likely mediated by the relatively large spin-orbit coupling of the Ni-centre [3], and effectively decreases the activation barrier. Spin-crossing or two-state reactivity is a concept in organometallic chemistry proposed by Shaik et al. [53] and has been used widely to explain and predict the reactivity of several organometallic and bioinorganic systems [53–55], including a report on Ni(II)-porphyrin complexes [56]. With species A (Paths X and Y) taken as the reference, formation of B (Path X) is associated with a loss of 12.2 kJ/mol, whereas formation of E (Path Y) exhibits a barrier of +11.7 kJ/mol. This result, together with the energy costs involved in converting acetophenone into potassium 1-phenylethenolate (+24.6 kJ/mol) versus conversion of 1-phenylethanol into potassium 1-phenylethanolate (-98.7 kJ/mol) would inhibit formation of E from Path Y. Both intermediates B and E are characterized by shorter C-O bonds (1.851 Å (B), 1.817 Å (E)) in the singlet state, as compared to the (more stable) triplet states (1.876 Å (B), 1.872 Å (E)). Figures and additional bond lengths of the fully optimized intermediates and transition states are included in the Supplementary Information. When only considering the singlet multiplicity case for Paths X and Y (Fig. 8), Path X exhibits an activation barrier of +46.0 kJ/mol, associated with the rate-determining state [57] of reductive elimination. In contrast, Path Y exhibits an activation barrier of +228.1 kJ/mol, associated with the rate-determining state of oxidative addition. This leads to a favourable energetic span for

54

F.P. Malan et al. / Molecular Catalysis 432 (2017) 47–56

Path X of ␦E = +58.2 kJ/mol, an exothermic pathway, as compared to the much higher ␦E = +269.3 kJ/mol for Path Y, an endothermic pathway. The oxidative addition step for Paths X and Y evaluated as either singlet or triplet multiplicities (-20.3 and +11.7 kJ/mol respectively) are all lower than similar oxidative addition steps of Pd-NHC species (> 60.5 kJ/mol) [25] and Ni-NHC species (> 23.8 kJ/mol) [58], and in the case of Path X (singlet/triplet) is indicative of a facile step. Overall, Path X is highly exergonic with −228.5 kJ/mol, and Path Y endergonic with +82.2 kJ/mol, which leads to the conclusion that Path X is the preferred path. When considering triplet multiplicity for all the intermediates involved (Fig. 9), the more stable B is formed (Path X) with a loss of 20.3 kJ/mol, whereas the more stable E (Path Y) is formed with a loss of 3.8 kJ/mol. For Path X, the rate-determining state shifts to the hydride transfer step with an activation barrier of +69.4 kJ/mol, and an associated ␦E of +89.7 kJ/mol. The latter is in clear contrast to Path Y, having the rate-determining state(s) remaining at the oxidative addition step, with an activation barrier of +174.0 kJ/mol and an associated ␦E of +177.9 kJ/mol. Similar to the singlet multiplicity, the triplet multiplicity shows Path X to be the preferred path being highly exergonic at −228.5 kJ/mol, compared to Path Y being endergonic at +56.5 kJ/mol. Efficient catalysis relies on the minimization of ␦E, as under steady-state conditions the rate of the reaction is strongly influenced by this parameter. Therefore improvements in catalyst efficiency and atom economy would include stabilization of the highest energy transition state, as well as destabilizing the most abundant reaction intermediate [59]. Hence, with all results considered, it is clear that Path X is preferred irrespective of the spin multiplicities of the intermediates involved, and agrees with our experimental findings that anaerobic alcohol oxidation is feasible under elevated temperatures. Kozuch et al. [57] developed an elegant yet simplistic method of estimating the TOF of a given catalytic cycle. They showed that for exothermal reactions (G < 0) that the TOF could be defined as

TOF =

kB T −ıE/RT e h

(1)

Fig. 9. Relative Gibbs free energy profile for the Ni(II) triplet state of Paths X and Y.

where kB is the Boltzmann constant, T is the temperature, h is Planck’s constant, ıE is the energetic span, and R is the universal gas constant. If we apply Eq. (1) to re-optimized reaction intermediates and transition states where solvent effects have been included, the TOF calculations are in agreement with the preferred Path X in both spin multiplicities, with TOF = 0.2 h−1 (X, singlet), 3.7 × 10−9 h−1 (Y, singlet), 0.1 h−1 (X, triplet), and 6.4 × 10−7 h−1 (Y, triplet). With the theoretical TOF of Path X (singlet) = 0.2 h−1 , resulting from a solvent-corrected ␦E = 94.5 kJ/mol (25 ◦ C), it is in fair agreement with our experimental TOF finding for 1 (84 h−1 at 110 ◦ C; 8 h−1 at

Fig. 10. Molecular orbital diagrams of A, B (Path X) and E (Path Y) using singlet multiplicities.

F.P. Malan et al. / Molecular Catalysis 432 (2017) 47–56

25 ◦ C) and furthermore shows the strong temperature dependence on the TOF of the reaction. The molecular orbital diagrams of A and its dehalogenated products B and E (Fig. 10) help to further illustrate the electronic differences of the intermediate products of Path X over Y. The HOMO of A is uniquely populated over the ArNiBr functionalities, and is the lowest in energy of all the species involved. Following Path X, the formation of alkoxy adduct B shifts the electron density in the HOMO away from the aryl towards the Ni(alkoxy)(NHC) functionalities, and in the case of E (Path Y), shifts toward the Ni(alkenoxy) functionalities (highest in energy). The LUMOs of A and its adducts B and E all have similar electronic character with Ni(NHC)(X) (X = Br, alkoxy, alkenoxy) involved. However, in E the LUMO is highly concentrated on the metal centre and ethenoxy moiety, bearing the highest LUMO energy (-4.78 eV) of the three intermediates. It is noteworthy that the largest HOMO/LUMO energy difference (energy gap) is exhibited by the alkoxy adduct B (5.26 eV), while the smallest energy gap is exhibited by ethenoxide adduct E (2.82 eV), which further highlights the latter species’ instability and higher reactivity [50]. In terms of the Principle of Maximum Hardness (PMH) related to the molecular orbital (MO) theory, the hardness of a molecule (␩) is defined as half the energy gap between the HOMO and the LUMO [60]. Furthermore, a chemical system tends to arrange itself to maximise hardness of its respective molecules where a hard molecule tend to have a large energy gap [60], Comparing intermediates B (␩ = 2.63 eV) and E (␩ = 1.41 eV), E is softer than B, and therefore is expected to be less stable and more reactive than intermediate B [60]. 4. Conclusion A range of nine [CpNiBr(NHC) complexes, each bearing flexible, electron withdrawing/donating NHCs were inexpensively synthesised. The molecular structures of three of these complexes were also obtained and elucidated with X-ray crystallography. These Ni(II)-NHC were all catalytically active in the anaerobic alcohol oxidation reaction whereby a range of several secondary alcohols were oxidized to their corresponding ketones, along with concomitant dehalogenation of 1,2-dibromobenzene. Complexes 1, 2, and 5 were the most active, followed by 3, 4, and 7, and finally 6, 8, and 9. Among complexes 1–9, the more electron-donating NHCbearing complexes were more efficient, owing to their enhanced ability to better support sensitive Ni(0) species. The flexibility and bulkiness of the N-alkyl/aryl substituents on the NHC ligands provided accessibility to incoming substrates, while also aiding in the reductive elimination of the organic products. A strong catalyst concentration dependency was found, which is indicative of catalyst decomposition. Utilization of microwave heating methods notably increased the alcohol conversions, by decreasing the reaction times to inhibit the effects of catalyst decomposition. Furthermore, DFT mechanistic studies gave more insight into the favoured ketone product formation, as opposed to their subsequent ␣-ketone arylated products. Acknowledgements The National Research Foundation of South Africa (NRF, grant number 93638) and the University of Pretoria (UP) are gratefully acknowledged for financial support (ML, FPM). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.mcat.2016.12. 023.

55

References [1] C. Berini, O.H. Winkelmann, J. Otten, D.A. Vicic, O. Navarro, Chem. Eur. J. 16 (2010) 6857–6860. [2] L.M. Dornan, G.M.A. Clendenning, M.B. Pitak, S.J. Coles, M.J. Muldoon, Catal. Sci. Technol. 4 (2014) 2526–2534. [3] K.M. Gligorich, M.S. Sigman, Angew. Chem. Int. Ed. 45 (2006) 6612–6615. [4] C. Berini, D.F. Brayton, C. Mocka, O. Navarro, Org. Lett. 11 (2009) 4244–4247. [5] S. Manzini, C.A. Urbina-Blanco, S.P. Nolan, Organometallics 32 (2013) 660–664. [6] M.N. Kopylovich, A.P.C. Ribeiro, E.C.B.A. Alegria, N.M.R. Martins, L.M.D.R.S. Martins, A.J.L. Pombeiro, Advances in Organometallic Chemistry, Elsevier Inc, 2015, pp. 91–174. [7] J.A. Mueller, C.P. Goller, M.S. Sigman, J. Am. Chem. Soc. 126 (2004) 9724–9734. [8] R.H. Crabtree, Chem. Rev. 115 (2015) 127–150. [9] L.P. Bheeter, M. Henrion, L. Brelot, C. Darcel, M.J. Chetcuti, J.-B. Sortais, V. Ritleng, Adv. Synth. Catal. 354 (2012) 2619–2624. [10] A. Collado, A. Gómez-Suárez, A.R. Martin, A.M.Z. Slawin, S.P. Nolan, Chem. Commun. 49 (2013) 5541–5543. [11] R.H. Crabtree, Coord. Chem. Rev. 257 (2013) 755–766. [12] G.C. Fortman, S.P. Nolan, Chem. Soc. Rev. 40 (2011) 5151–5169. [13] O. Kühl, Chem. Soc. Rev. 36 (2007) 592–607. [14] B. Landers, O. Navarro, Inorg. Chim. Acta. 380 (2012) 350–353. [15] B. Landers, O. Navarro, Eur. J. Inorg. Chem (2012) 2980–2982. [16] S.R. Patrick, A. Collado, S. Meiries, A.M.Z. Slawin, S.P. Nolan, J. Organomet. Chem. 775 (2015) 152–154. [17] A.P. Prakasham, P. Ghosh, Inorg. Chim. Acta 431 (2015) 61–100. [18] O. Santoro, A. Collado, A.M.Z. Slawin, S.P. Nolan, C.S.J. Cazin, Chem. Commun. 49 (2013) 10483–10485. [19] O. Schuster, L. Yang, H.G. Raubenheimer, M. Albrecht, Chem. Rev. 109 (2009) 3445–3478. [20] R. Visbal, A. Laguna, M.C. Gimeno, Chem. Comm. 49 (2013) 5642–5644. [21] L.-Y. Wang, J. Li, Y. Lv, H.-Y. Zhang, S. Gao, J. Organomet. Chem. 696 (2011) 3257–3263. [22] V. Ritleng, M. Henrion, M.J. Chetcuti, ACS Catal. 6 (2016) 890–906. [23] M. Gholinejad, H.R. Shahsavari, M. Razeghi, M. Niazi, F. Hamed, J. Organomet. Chem. 796 (2015) 3–10. [24] S. Meiries, K. Speck, D.B. Cordes, A.M.Z. Slawin, S.P. Nolan, Organometallics 32 (2013) 330–339. [25] Y. Li, Z. Lin, Org. Chem. Front. 1 (2014) 1188–1196. [26] S. Dayan, N.O. Kalaycioglu, Appl. Organometal. Chem. 27 (2013) 52–58. [27] B. Landers, C. Berini, C. Wang, O. Navarro, J. Org. Chem. 76 (2011) 1390–1397. [28] D. Jantke, M. Cokoja, A. Pöthig, W.A. Herrmann, F.E. Kühn, Organometallics 32 (2013) 741–744. [29] C. Pranckevicius, D.W. Stephan, Chem. Eur. J. 20 (2014) 6597–6602. [30] W. Buchowicz, W. Wojtczak, A. Pietrzykowski, A. Lupa, L.B. Jerzykiewicz, A. ´ Eur. J. Inorg. Chem. (2010) 648–656. Makal, K. Wozniak, [31] F.E. Hahn, M.C. Jahnke, Angew. Chem. Int. Ed. 47 (2008) 3122–3172. [32] A.R. Martin, Y. Makida, S. Meiries, A.M.Z. Slawin, S.P. Nolan, Organometallics 32 (2013) 6265–6270. [33] A.M. Oertel, V. Ritleng, A. Busiah, L.F. Veiros, M.J. Chetcuti, Organometallics 30 (2011) 6495–6498. [34] A.M. Oertel, J. Freudenreich, J. Gein, V. Ritleng, L.F. Veiros, M.J. Chetcuti, Organometallics 30 (2011) 3400–3411. [35] F.P. Malan, E. Singleton, P.H. van Rooyen, M. Landman, J. Organomet. Chem. 813 (2016) 7–14. [36] R.J. Errington, Advanced Practical Inorganic and Metalorganic Chemistry, Blackie Academic & Professional, London, 1997. [37] D.F. Shriver, M.A. Drezdzon, The Manipulation of Air-Sensitive Compounds, 2nd ed, Wiley, New York, 1986. [38] A.D. Becke, J. Chem. Phys. 98 (1993) 5648–5652. [39] C. Lee, W. Yang, R.G. Parr, Phys. Rev. B 37 (1988) 785–789. [40] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G.A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H.P. Hratchian, A.F. Izmaylov, J. Bloino, G. Zheng, J.L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J.A. Montgomery Jr, J.E. Peralta, F. Ogliaro, M. Bearpark, J.J. Heyd, E. Brothers, K.N. Kudin, V.N. Staroverov, T. Keith, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J.C. Burant, S.S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J.M. Millam, M. Klene, J.E. Knox, J.B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski, R.L. Martin, K. Morokuma, V.G. Zakrzewski, G.A. Voth, P. Salvador, J.J. Dannenberg, S. Dapprich, A.D. Daniels, O. Farkas, J.B. Foresman, J.V. Ortiz, J. Cioslowski, D.J. Fox, Gaussian 09, Revision D.01, Gaussian Inc, Wallingford CT, 2010. [41] F. Weigend, R. Ahlrichs, Phys. Chem. Chem. Phys. 7 (2005) 3297–3305. [42] J.W. McIver, A.K. Komornicki, J. Am. Chem. Soc. 94 (1972) 2625–2633. [43] APEX2 (including SAINT and SADABS), Bruker AXS Inc, Madison, WI, 2012. [44] G.M. Sheldrick, SHELXL96, Program for the Refinement of Crystal Structures, University of Göttingen, Germany, 1996. [45] G.M. Sheldrick, Acta Cryst. A 64 (2008) 112. [46] O.V. Dolomanov, L.J. Bourhis, R.J. Gildea, J.A.K. Howard, H. Puschmann, J. Appl. Cryst. 42 (2009) 339–341. ´ [47] W. Buchowicz, Ł. Banach, J. Conder, P.A. Gunka, D. Kubicki, P. Buchalski, Dalton Trans. 43 (2014) 5847–5857.

56

F.P. Malan et al. / Molecular Catalysis 432 (2017) 47–56

[48] A.M. Oertel, V. Ritleng, M.J. Chetcuti, Organometallics 31 (2012) 2829–2840. [49] V. Ritleng, A.M. Oertel, M.J. Chetchuti, Dalton Trans. 39 (2010) 8153–8160. [50] S. Thangavel, S. Boopathi, N. Mahadevaiah, P. Kolandaivel, P.B. Pansuriya, H.B. Friedrich, J. Mol. Catal. A. 423 (2016) 160–171. [51] C.-C. Lee, W.-C. Ke, K.-T. Chan, C.-L. Lai, C.-H. Hu, H.M. Lee, Chem. Eur. J. 13 (2007) 582–591. [52] T. Sperger, I.A. Sanhueza, I. Kalvet, F. Schoenebeck, Chem. Rev. 115 (2015) 9532–9586. [53] S. Shaik, H. Hirao, D. Kumar, Acc. Chem. Res. 40 (2007) 532–542. [54] B.K. Mai, Y. Kim, Angew. Chem. Int. Ed. 54 (2015) 3946–3951.

[55] D. Schröder, S. Shaik, H. Schwarz, Acc. Chem. Res. 33 (2000) 139–145. [56] S. Thies, C. Bornholdt, F. Köhler, F.D. Sönnichsen, C. Näther, F. Tuczek, R. Herges, Chem. Eur. J. 16 (2010) 10074–10083. [57] S. Kozuch, S. Shaik, Acc. Chem. Res. 44 (2011) 101–110. [58] T. Mesganaw, A.L. Silberstein, S.D. Ramgren, N.F.F. Nathel, X. Hong, P. Liu, N.K. Garg, Chem. Sci. 2 (2011) 1766–1771. [59] A.T. Normand, K.J. Hawkes, N.D. Clement, K.J. Cavell, B.F. Yates, Organometallics 26 (2007) 5352–5363. [60] R.G. Pearson, Acc. Chem. Res. 26 (1993) 250–255.