Cooperative Al–H Bond Activation in DIBAL–H - American Chemical

Article pubs.acs.org/JACS

Cite This: J. Am. Chem. Soc. 2017, 139, 16334-16342

Cooperative Al−H Bond Activation in DIBAL-H: Catalytic Generation of an Alumenium-Ion-Like Lewis Acid for Hydrodefluorinative Friedel−Crafts Alkylation Francis Forster,† Toni T. Metsan̈ en,†,‡ Elisabeth Irran,† Peter Hrobárik,*,†,§ and Martin Oestreich*,† †

Institut für Chemie, Technische Universität Berlin, Strasse des 17. Juni 115, 10623 Berlin, Germany Department of Inorganic Chemistry, Faculty of Natural Sciences, Comenius University, 84215 Bratislava, Slovakia

§

S Supporting Information *

ABSTRACT: The Ru−S bond in Ohki−Tatsumi complexes breaks oligomeric DIBAL-H structures into their more reactive monomer. That deaggregation is coupled to heterolytic Al−H bond activation at the Ru−S bond, formally splitting the Al−H linkage into hydride and an alumenium ion. The molecular structure of these Lewis pairs was established crystallographically, revealing an additional Ru−Al interaction next to the Ru−H and Al−S bonds. That bonding situation was further analyzed by quantum-chemical calculations and is best described as a three-center−two-electron (3c2e) donor−acceptor σ(Ru−H) → Al interaction. Despite the extra stabilization of the aluminum center by the interaction with both the sulfur atom and the Ru−H bond, the hydroalane adducts are found to be stronger Lewis acids and electrophiles than the free ruthenium catalyst and DIBAL-H in its different aggregation states. Hence, the DIBAL-H molecule and its Al−H bond are activated by the Ru−S bond, but these hydroalane adducts are not to be mistaken as sulfur-stabilized alumenium ions in a strict sense. The Ohki−Tatsumi complexes catalyze C(sp3)−F bond cleavage with DIBAL-H, and the catalytic setup is applied to hydrodefluorinative Friedel−Crafts alkylations. A broad range of CF3-substituted arenes is efficiently converted into unsymmetrical diarylmethanes with various arenes as nucleophiles. Computed fluoride-ion affinities (FIAs) of the hydroalane adducts as well as DIBAL-H in its aggregation states support this experimental finding.

■

convert these monofluorides into alkanes.8 By replacing the aforementioned counteranions [X]− in [Ph3C]+[X]− with Reed’s [HCB11H5Br6]−, Ozerov and co-workers then achieved the hydrodefluorination of CF3 groups; competing transfer of hydride or an alkyl substituent from DIBAL-H was observed.9 We together with Ohki and Tatsumi introduced the coordinatively unsaturated ruthenium(II) complexes [1]+[X]−10 for cooperative Si−H11,12 and B−H13 bond activation (Scheme 1, top). The polar Ru−S bond in [1]+[X]− was shown to split H−H,10 Si−H,11,12 and B−H13 bonds into hydride and the corresponding proton or maingroup cation (Scheme 1, middle).14 We anticipated that Al−H bonds could also undergo this bond activation, thereby forming sulfur-stabilized alumenium ions for catalysis (Scheme 1, bottom).

INTRODUCTION Examples of [R2Al−LB] (R = alkyl/aryl; LB = Lewis base) are scarce while heteroatom-substituted alumenium ions as well as cyclopentadienyl complexes are well established.1 The earliest example of an [R2Al]+-type cation is the ion-like complex [Et2Al]+[HCB11H5X6]− (X = Cl or Br) disclosed by Kim and Reed.2 Weak bidentate Al···X interactions stabilize its highly electron-deficient aluminum center. Wehmschulte and coworkers later accomplished the structural characterization of the m-terphenyl-substituted complex [(2,6-Mes2C6H3)2Al]+[B(C6F5)4]− with the mesityl groups sterically shielding and stabilizing the electron gap at the aluminum center.3 Those examples were followed by similar complexes such as [Dipp*AlEt]+[HCB11H5Cl6]−,4a ([Et2Al]+)2[B12Cl12]2−,4b or {[Me2Al]+[H3CCB11F11]−}2.4c As to that, Krossing’s recent review of reactive main-group cations provides an excellent summary of alumenium ions.5 A prominent application of strongly electrophilic aluminum compounds is the hydrodefluorination of C(sp3)−F bonds.6 Rosenthal and co-workers reported the hydrodefluorination of an alkyl fluoride by the combination of DIBAL-H as the hydride source and various trityl salts [Ph3C]+[X]− {X = B(C6F5)4, Al(C6F5)4, or Al[OC(CF3)3]4} as initiator or catalyst.7 However, Terao, Kambe, and co-workers showed in the same year that stoichiometric DIBAL-H alone is able to © 2017 American Chemical Society

■

RESULTS AND DISCUSSION NMR Spectroscopic Analysis of the Adduct Formation. Initial experiments showed rapid degradation of the routinely used [BArF4]− counteranion in the presence of alumenium ions. We therefore turned toward more robust counteranions such as [B(C6F5)4]− and [B12Cl12]2− devoid of Received: September 4, 2017 Published: October 6, 2017 16334

DOI: 10.1021/jacs.7b09444 J. Am. Chem. Soc. 2017, 139, 16334−16342

Article

Journal of the American Chemical Society Scheme 1. Tethered Ru−S Complexes [1]+[X]− and Cooperative Bond Activation of E−H Bonds [E = H, Si, B, and Al]

Figure 1. Molecular structure of [1a·iBu2AlH]+[B(C6F5)4]−. Hydrogen atoms (except for Ru−H) and counteranion are omitted for the sake of clarity. Selected experimental bond lengths (Å): Ru−S, 2.41; Ru−Al, 2.81; Ru−H, 1.65; Al−S, 2.31; Al−H, 1.98. Selected experimental bond angle: C−Al−C, 119° (average of 110° and 127°). For comparison, the DFT optimized distances (Å) and angles are as follows: Ru−S, 2.42; Ru−Al, 2.78; Ru−H, 1.65; Al−S, 2.33; Al− H, 1.85; C−Al−C, 119.5°.

Table 1. 1H NMR Shifts and Coupling Constants of the Ruthenium(II) Hydrides in the iBu2AlH Adductsa

entry

adduct

1 2 3 4 5 613 713 813 913 1011f

+

1

H NMR [ppm]

−

[1a·iBu2AlH] [B(C6F5)4] [1b·iBu2AlH]+[B(C6F5)4]− [1c·iBu2AlH]+[B(C6F5)4]− [1d·iBu2AlH]+[B(C6F5)4]− ([1b·iBu2AlH]+)2[B12Cl12]2− [1a·Cy2BH]+[BArF4]− [1a·9-BBN]+[BArF4]− [1a·catBH]+[BArF4]− [1a·pinBH]+[BArF4]− [1a·EtMe2SiH]+[BArF4]−

−12.4 −13.0 −12.7 −11.7 −13.0 −12.1 −11.9 −10.1 −9.0 −8.1

2

JH,P [Hz] 26.1 26.3 26.4 25.3 25.9 18.9 17.7 28.4 45.1 49.9

Figure 2. Molecular structure of [1c·iBu2AlH]+[B(C6F5)4]−. Hydrogen atoms (except for Ru−H) and counteranion are omitted for the sake of clarity. Selected experimental bond lengths (Å): Ru−S, 2.40; Ru−Al, 2.83; Ru−H, 1.53; Al−S, 2.32, Al−H, 2.16. Selected experimental bond angle: C−Al−C, 115°. For comparison, the DFT optimized distances (Å) and angles are as follows: Ru−S, 2.42; Ru−Al, 2.80; Ru−H, 1.65; Al−S, 2.33; Al−H, 1.85; C−Al−C, 115.8°.

a

Experiments were performed in a J. Young NMR tube using complexes [1]+[X]− (1.0 equiv) and DIBAL-H (1.2 equiv).

C(sp3)−F bonds.15 Mixing complexes [1]+[X]− and DIBAL-H in o-C6D4Cl2 at room temperature instantly delivered the hydroalane adducts [1·iBu2AlH]+[X]− (Table 1, entries 1−5). The adduct formation is accompanied by a color change from green to yellow. The observed hydride shifts are comparable to the earlier example of the B−H bond activation in Cy2BH (δ(1H) −12.1 ppm with [1a]+[BArF4]− and δ(1H) −11.4 ppm with [1d]+[BArF4]−) (Table 1, entries 6−9).13 The 2JH,P coupling constants are in the range of 26 Hz and as such larger compared to those of hydroborane adducts (2JH,P ≈ 18 Hz) but smaller than in the hydrosilane adducts (2JH,P ≈ 50 Hz) (Table 1, entry 10).16 Observation of the 27Al nucleus (I = 5/2) by NMR spectroscopy failed because of the high line width.17 However, our relativistic DFT calculations for [1a·iBu2AlH]+ predict the 27Al resonance signal to appear at δ(27Al) 178 ppm,

thus only slightly downfield from that in parent DIBAL-H with δ(27Al) 158 ppm in toluene-d8.18,19 Structural Characterization of the Hydroalane Adduct. Single crystals suitable for X-ray diffraction were obtained at −35 °C from a solution of [1a]+[B(C6F5)4]− or [1c]+[B(C6F5)4]− and DIBAL-H in fluorobenzene layered with npentane (Figures 1 and 2). The molecular structures of the corresponding hydroalane adducts of [1a]+[B(C6F5)4]− and [1c]+[B(C6F5)4]− prove that the aluminum atom is bound to the sulfur atom with Al−S bond lengths of 2.31 and 2.32 Å, respectively (Table 2, entries 1 and 2). Compared to [1a·9BBN]+[BArF4]− (B−S bond length = 1.94 Å13) and [1a· 16335

DOI: 10.1021/jacs.7b09444 J. Am. Chem. Soc. 2017, 139, 16334−16342

Article

Journal of the American Chemical Society

The reliability of DFT optimized structures, including hydride positions, is further demonstrated by the excellent agreement between computed 1H NMR hydride shifts δ(Ru− H) and coupling constants 2JH,P with experimental data (Table S5 and Figures S3 and S4 in Supporting Information). To quantify the degree of E−H bond dissociation and bond covalencies for pertinent atom pairs, we evaluated Mayer bond orders (MBOs) for a wide series of [1]+ adducts with various group 13 and group 14 hydrides RnEH (Tables S3 and S4 in Supporting Information). Hence, MBOs of Ru−H bonds and E−H bonds show a superb correlation: the stronger and thereby shorter the Ru−H bond, the weaker the E···H bonding interaction (Figure 3). The electron-density depletion in the

Table 2. Selected Experimental Bond Lengths and Angles of E−H Adducts [E = Si, B, and Al] entry

adduct

Ru−H [Å]

E−S [Å]

C−E−C [deg]

1 2 313 411f

[1a·iBu2AlH]+[B(C6F5)4]− [1c iBu2AlH]+[B(C6F5)4]− [1a·9-BBN]+[BArF4]− [1a·EtMe2SiH]+[BArF4]−

1.65 1.53 1.80 1.61

2.31 2.32 1.94 2.24

119 115 109 110

EtMe2SiH]+[BArF4]− (Si−S bond length = 2.24 Å11f), the Al−S bond is notably longer (Table 2, entries 3 and 4). That can be attributed to a larger covalent/ionic radius of aluminum as compared to boron and silicon as well as to a still existing bonding interaction between aluminum and hydrogen (see bonding analysis below). The observed bond length is in accordance with those of aluminum sulfides (Al−S, bond lengths = 2.2720a−2.34 Å20b) while complexes with coordinating thioethers display longer distances (Al−S, bond length = 2.51 Å21a). The hydride is located at the ruthenium center with Ru−H bond lengths of 1.65 and 1.53 Å in [1a·iBu2AlH]+[B(C6F5)4]− and [1c·iBu2AlH]+[B(C6F5)4]−, respectively. This bond is comparably shorter than in the aforementioned hydroborane and hydrosilane adducts ([1a·9-BBN]+[BArF4]−: Ru−H bond length = 1.80 Å13 and [1a·EtMe2SiH]+[BArF4]−: Ru−H bond length = 1.61 Å11f). The Al−H bond in [1a· iBu2AlH]+[B(C6F5)4]− (bond length = 1.98 Å) is markedly elongated compared to known hydroalanes [(2,6Mes2C6H3)2Al−H with coordination number 3 at aluminum, Al−H bond length = 1.4322a and di/trimeric (tBuCH2)2Al−H with coordination number 4 at aluminum, Al−H bond length = 1.62−1.76 Å22b]. Interestingly, the molecular structures of [1a·iBu2AlH]+[B(C6F5)4]− and [1c·iBu2AlH]+[B(C6F5)4]− reveal a close contact between the aluminum atom and the ruthenium center with Ru−Al interatomic distances of 2.81 and 2.83 Å, respectively. The experimentally observed bond angle for the C−Al−C fragments are 119° and 115°; other known sulfur-coordinated aluminum compounds possess slightly wider angles (C−Al−C bond angles = 119−125°).20,21 Quantum-Chemical Analysis of the Hydroalane Adducts and Comparison with Their Group 13 and Group 14 Analogues. To ascertain that the close contacts of the aluminum atom with sulfur, hydrogen, and the ruthenium center in [1·iBu2AlH]+[B(C6F5)4]− complexes are not an artifact of crystal packing forces and to resolve uncertainties with hydride atom positions determined by X-ray diffraction, we carried out DFT calculations for a series of complexes [1]+, their hydroborane adducts, their hydrosilane adducts, and their heavier group 13 and group 14 analogues (Tables S3 and S4 in Supporting Information). The molecular structures (without counteranions) were fully optimized at the B3LYP-D3(BJ)/ ECP/6-31+G** level, using a quasi-relativistic small-core pseudopotential for ruthenium along with atom-pairwise corrections for dispersion forces. In addition, bulk solvent effects were simulated using an SMD solvation model (Computational Details in Supporting Information). We note an excellent match between experimental and DFT optimized structures, with differences in bond lengths of less than 0.03 Å (Figures 1 and 2). Only the precise detection of the Ru−H bond lengths and the Al−H bond lengths suffered from weak scattering of X-rays by light hydrogen atoms and thus the inaccurate hydride positions.

Figure 3. Correlation between Mayer bond orders (MBOs) of Ru−H bonds and E−H bonds within the same adduct of [1]+ with various group 13 and group 14 element hydrides (Tables S3 and S4 in Supporting Information for numerical data).

E−H bond is thus connected with electron-density accumulation in the Ru−H bond, which can be monitored spectroscopically by measuring 2JH,P coupling constants.23 In this regard, the adducts with completely broken E−H bonds and well-formed Ru−H bonds display 2JH,P coupling constants larger than 42 Hz (this is particularly the case for group 14 hydrides)11f while systems with weakly activated E−H bonds and partially formed Ru−H bonds possess much smaller coupling constants. On this basis, the E−H bond in dialkylborane adducts of [1]+ with 9-BBN or Cy2BH (2JH,P ≈ 18 Hz) is less activated, which means less heterolytically cleaved, than that of [1·iBu2AlH]+ (2JH,P ≈ 26 Hz), that is also in accordance with the MBO values computed for the E···H atom pairs (Table S3 in Supporting Information). The B···H interaction in these adducts preserves about 40− 50% bonding character as evident from MBO(E−H) values for free hydroboranes and their adducts with [1]+, while the Al···H interaction preserves (only) 25−33% bonding character, even though the E−H bond elongation upon coordination to [1]+ is of the same magnitude (∼0.25 Å) for both hydroboranes and hydroalanes. Note that apart from [1·pinBH]+, the donor−acceptor interaction between the low-valent, electron-rich ruthenium(II) d6 center and the electrophilic, coordinatively unsaturated main-group atom is a common feature for all adducts with group 13 hydrides, as also experimentally found in this work for both [1·iBu2AlH]+ complexes (Figures 1 and 2) and previously for [1a·(9-BBN)]+ (however, without noting a short Ru···B bond interaction = 2.52 Å).13 We note further that the Ru···Al distances in [1·iBu2AlH]+ adducts are notably above the sum of single-bond covalent radii 16336

DOI: 10.1021/jacs.7b09444 J. Am. Chem. Soc. 2017, 139, 16334−16342

Article

Journal of the American Chemical Society (2.51 Å)24 or those found in [Ru(AlCp*)5] (with the average Ru−Al bond length of 2.38 Å),25 but are comparable with that computed for hypothetical [RcAlMe2] complex (Rc = ruthenocene; d(Ru···Al) = 2.81 Å) with a large dip angle (38°) due to the Ru···Al bonding (Figures S7 and S8 in Supporting Information and ref 26 for analogues, synthetically available [FcSiR2]+ systems with Fe···Si bonding). The unusual bonding situation in [1a·iBu2AlH]+ is already apparent from the Mayer bond orders between aluminum and its neighbors (Table S3 in Supporting Information). While a considerably polarized single bond is clearly present between aluminum and sulfur,27 both the Ru−H bond as well as the Al− H bond possess partial covalent character, indicative of delocalized bonding. This is examined in more detail by the natural localized molecular orbital (NLMO) analysis and the electron localization function (ELF), both revealing a threecenter−two-electron (3c2e) bond between ruthenium, hydrogen, and aluminum (Figures 4 and 5), which can be best

Figure 5. Cut-plane plots from electron localization function (ELF) analysis of bonding in [1a·Me2AlH]+, showing Ru···Al, Al−S, and Ru− S bonding interactions. Gray-white regions represent ELF maxima (bonding attractors).

Apart from hydrocarbons, the overall E−H bond activation process for MenEH hydrides is exergonic with formation energies (ΔG0f) ranging from −38 kJ mol−1 to −146 kJ mol−1 for corresponding hydrosilanes and hydroalanes, respectively (Table S4 in Supporting Information). Our previous computational study for hydrosilane adducts showed that the heterolytic cleavage of the Si−H bond proceeds via a concerted fourmembered transition state with relatively small reaction energy barriers of ca. 10−40 kJ mol−1 depending on the hydrosilane.11f DFT simulations of the cooperative bond activation of DIBALH by [1a]+ show that the Al−H bond cleavage is practically barrierless on the electronic energy surface. Interestingly, while group 14 hydrides show a steady decrease in ΔG0f as the maingroup atom becomes heavier, nonmonotonous behavior is observed for the group 13 congeners, presumably due to the above-mentioned σ(Ru−H) → E interaction. Comparing computed CM5 partial atomic charges at the aluminum center reveals that in coherence with computed fluoride-ion affinities (ΔG0FIA), complex [1a·iBu2AlH]+ is a stronger electrophile and a stronger Lewis acid than both free [1a]+ and DIBAL-H in its different aggregation states [(iBu2AlH)n with n ≥ 2, Table 3; see also Table S8 in Supporting Information for more data].29 According to q(Al) charges, a slightly higher electrophilicity is expected for monomeric DIBAL-H, which is absent in solution and exists in nonpolar organic solvents mostly in the form of di/trimeric structures.31 As expected, oligomeric DIBAL-H structures are much weaker electrophiles than both monomeric iBu2AlH and its adduct [1a·iBu2AlH] (the molecular electrostatic potential (MEP) is displayed for the calculated hydroalane adduct [1a·Me2AlH]+ in Figure 6). The

Figure 4. Natural localized molecular orbital (NLMO) corresponding to a 3c2e bond between ruthenium, hydrogen, and aluminum atoms (isosurface plot ±0.06 au; hydrogen atoms except Ru−H are omitted for clarity).

characterized as a σ(Ru−H) → Al donor−acceptor interaction with an appreciable electron-density accumulation/depletion in the Ru···Al/Ru−H interatomic region. According to the NLMO analysis of [1a·iBu2AlH]+, the population of the sp3-hybridized aluminum atomic orbitals in this 3c2e bond is at 10% while the population of ruthenium is at 30% with predominant d-character. The electron density depletion in the Ru−H bond is accompanied by its elongation and more upfield hydride shifts in the 1H NMR spectra. In addition, the σ(Ru−H) → E interaction is responsible for (only) partial heterolytic splitting (activation) of the Al−H bond and is predicted to appear also in heavier group 13 analogues (the highest Ru···E bond covalency is predicted for the gallium analogue and markedly weaker interaction in indium and thallium congeners; Table S4 in Supporting Information).28 We note that the Ru−S bond in [1]+ becomes longer upon the hydride adduct formation (by about 0.13−0.19 Å) but still preserves an appreciable covalent bonding character (Figure 5). 16337

DOI: 10.1021/jacs.7b09444 J. Am. Chem. Soc. 2017, 139, 16334−16342

Article

Journal of the American Chemical Society Table 3. Computed FIAs (in kJ mol−1) for the Reaction of Lewis Acids with F3CO− as a Fluoride Donor (LAq + F3CO− → LA−Fq−1 + F2CO)30 and CM5 Partial Atomic Charges at the Ruthenium and Aluminum Centera Lewis acid (LA) +

[1a·iBu2AlH] [(Et3P)RuSDmp]+ (1a+) iBu2AlH (iBu2AlH)2 (iBu2AlH)3

ΔG0FIA [kJ mol−1]

q(Ru)

q(Al)

−151 −114 −99 −97 −82

0.202 0.281 − − −

0.309 − 0.352 0.223 0.222

even C(sp2)−F36b bond activation involving silicon cations were also described. Stephan and co-workers recently demonstrated that the electrophilic phosphonium salt [(C6F5)3PF]+[B(C6F5)4]− promotes broadly applicable Friedel−Crafts alkylations and benzylations initiated by fluoride abstraction.38 Both benzylic and aliphatic CF3 groups (with further hydrodefluorination)38a as well as benzylic fluorides38b participate well in this transformation with hydrosilanes as the hydride source. We then embarked on the systematic investigation of the hydrodefluorinative Friedel−Crafts alkylation of toluene with 4trifluoromethyl-1,1′-biphenyl (3a + 4a → 5aa, Table 4). The

a

B3LYP-D3(BJ)/ECP/6-31+G** results using an SMD solvation model (Computational Details in Supporting Information).

Table 4. Catalyst Screening for the Hydrodefluorinative Friedel−Crafts Alkylationa

entry

Figure 6. Molecular electrostatic potential (MEP) of [1a·Me2AlH]+. Isovalue surface is displayed at an electron density of 0.04 au (B3LYP/ ECP/6-31+G** results). Blue region at the aluminum atom corresponds to the extreme positive potential, indicating the electrophilic center for a fluoride abstraction.

activation of DIBAL-H can be additionally viewed as a dissociation of the oligomeric DIBAL-H structures initiated by the Lewis acidic ruthenium center in [1]+ and accompanied by partial heterolytic cleavage of the Al−H bond. Application to C(sp3)−F Activation. As shown earlier by us, [1d]+[BArF4]− catalyzes the hydrodefluorination of CF3substituted anilines with hydrosilanes.12 These reactions require the substoichiometric addition of an alkoxide base, and the role of that additive is understood on the basis of an unusual dual role of the catalyst (not shown). The catalyst loading is also high (10 mol %), and the catalyst system is limited to highly activated CF3 groups (reaction temperature 20 °C for the para Me2N group but 100 °C for meta).32 Trying to enhance the performance of this setup, we decided to replace the hydrosilane by the more reactive DIBAL-H. Initial experiments with the far less activated 1,1′-biphenyl-4-yl-substituted CF3 group (as in 3a) quickly revealed that the alkoxide-free catalysis with the hydroalane works at room temperature33 while completely failing with Me2PhSiH (not shown). The dominant reaction pathway was however the hydrodefluorination coupled with Friedel−Crafts benzylation of the arene solvent, e.g., toluene (4a). Joining benzylic C−F bond activation with Friedel−Crafts alkylations using BF3 as the Lewis acid catalyst was reported by Olah and co-workers three decades ago.34,35 A few additional examples of inter-36a,37 and intramolecular36b Friedel−Crafts alkylations subsequent to C(sp3)−F36a,37 or

catalyst

para:ortho (%)b

conv (%)b

PhF PhF PhF PhF PhF

− 78:22 79:21 79:21 78:22

− >99 >99 >99 >99

PhF PhF PhF PhH PhCl nC5H12 PhF

79:21 79:21 70:30 76:24 78:22 −

99 99 >99 99:1

yield (%)c 56 85 72 43 49 −f −f −f 52 46 35 23

a

All reactions were performed according to general procedure 3. Product ratios and conversions were determined by GLC analysis using tetracosane as internal standard. cThe reactions were run until complete conversion of the trifluoromethyl compound. Percentages are displaying isolated yields after column chromatography. dReaction was performed at 80 °C. eReaction was quenched after 5 d. fOnly hydrodefluorination and alkylation were observed. Np = naphthyl, Th = thienyl, Fc = ferrocenyl. b

compound in 60% yield (3a → 5aa, entry 1). No parasubstitution (as in 3b) as well as a methyl and a bromo substituent (as in 3c and 3d) were well tolerated (3b−d → 5ab−ad, entries 2−4). Conversely, diaryl ether 3e reacted under exclusive hydrodefluorination (entry 5). Precursors with ortho-substitution also participated but required higher reaction temperature (3f → 5af, entry 6) or longer reaction times (3g → 5ag, entry 7). Regioselectivities were generally moderate (rs ≈ 75:25) and as such usually better than those seen with Stephan’s protocol (rs < 67:33).38 Intramolecular Friedel− Crafts reactions essentially failed, e.g., 3h → 5h (Scheme 2); traces of 5h were detected by GC-MS analysis along with hydrodefluorination (6) and anthracene (7; dehydrogenated

substituted (4c−e) as well as monohalogenated arenes (4f− h) revealed that only benzene and electron-rich arenes do react (3a → 5ba−ea, entries 1−4). For less nucleophiles, we always obtained hydrodefluorination and alkyl transfer (entries 5−7). However, neither aryl alkyl ethers (alkyl = methyl and isopropyl) nor dimethylamino groups were stable under the reaction conditions (dealkylation, not shown). For this reason, diaryl (thio)ethers 4c and 4d and diisopropylamino-substituted 4e were employed. These furnished the corresponding diarylmethanes 5ca−ea in moderate to good yields, requiring elevated temperature (80 °C) in the case of 4d and 4e. Higher reaction temperature was detrimental to regiocontrol (3a → 5da, entry 3 and 3a → 5ia, entry 8), but steric hindrance was able to compensate this trend (3a → 5ea, entry 4). Thiophene (4j) and ferrocene (4k) led to the corresponding products 5ja and 5ka, respectively, only in poor yields (entries 10 and 11). For dialkyl-substituted arenes such as xylenes and cymenes, moderate to good yields were generally observed (Scheme 3). m-Xylene (4m) reacted to the single regioisomer 5ma whereas o-cymene (4o) gave a mixture of three regioisomers (5oa). Trisubstituted mesitylene (4q) afforded 5qa in near-quantitative yield. Testing the limits of the system, 1,4-di-tertbutylbenzene did not give the desired product, but instead only hydrodefluorination was observed (not shown). Bis(4tolyl) ether (4r) led to the expected diarylmethane 5ra in a 95:5 mixture with xanthene 8 (3a → 5ra, Scheme 4).

Scheme 2. Intramolecular Friedel−Crafts Alkylationa,b

a

The reaction was performed according to general procedure 3. boC6H4F2 was used instead of PhF to avoid side reactions with the solvent. cProduct ratios and conversions were determined by GLC analysis using tetracosane as internal standard. 16339

DOI: 10.1021/jacs.7b09444 J. Am. Chem. Soc. 2017, 139, 16334−16342

Journal of the American Chemical Society

Article

■

CONCLUSION The present work discloses a new example of cooperative catalysis at the Ru−S bond in Ohki−Tatsumi complexes.14 Al− H bond activation is now added to the activation of hydrosilanes, hydroboranes, and dihydrogen. The Ru−S functional group assists in deaggregation of oligomeric DIBAL-H structures and partial heterolytic splitting of the Al−H bond accompanied by the formation of Ru−H and Al−S bonds. Our initial hypothesis that the Al−H bond is completely cleaved heterolytically did not prove to be true and, hence, the assumed sulfur-stabilized alumenium ion (next to the ruthenium(II) hydride) is not an accurate description of these hydroalane adducts. Structural characterization by X-ray diffraction revealed a Ru−Al interaction in addition to the Ru− H and Al−S bonds. Our quantum-chemical calculations explain this bonding motif as an σ(Ru−H) → Al donor−acceptor interaction. DIBAL-H is still cooperatively activated in these Lewis adducts but not to the extreme of the hydrosilane adduct.11f Despite this only partial Al−H bond separation, the hydroalane adducts are found to be stronger electrophiles than the free ruthenium catalyst, DIBAL-H in its different aggregation states and hydrosilane adducts12 which is consistent with our experimental findings. The Ru−S-activated DIBAL-H molecule was shown to engage in C(sp3)−F bond cleavage, and the Ohki−Tatsumi complexes turned out to be excellent catalysts for hydrodefluorinative Friedel−Crafts alkylations of CF3-substituted arenes to afford unsymmetrical diarylmethanes with various arene nucleophiles.

Scheme 3. Scope IIb: Variation of the Arene Nucleophile in the Hydrodefluorinative Friedel−Crafts Alkylationa−c

a

■

All reactions were performed according to general procedure 3. Product ratios and conversions were determined by GLC analysis using tetracosane as internal standard. cThe reactions were run until complete conversion of the trifluoromethyl compound. Percentages are displaying isolated yields after column chromatography. b

ASSOCIATED CONTENT

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b09444. Experimental details, characterization and crystallographic data, and 1H, 13C, 11B, 19F, and 31P NMR spectra. Computational details, the results of DFT calculations, and more detailed analysis (PDF) CIF data (CIF) Cartesian coordinates of the selected DFT optimized structures (XYZ)

Scheme 4. Scope IIc: Variation of the Arene Nucleophile in the Hydrodefluorinative Friedel−Crafts Alkylationa−c

■

AUTHOR INFORMATION

Corresponding Authors

*[email protected] *[email protected] ORCID

Francis Forster: 0000-0002-0364-5491 Peter Hrobárik: 0000-0002-6444-8555 Martin Oestreich: 0000-0002-1487-9218 Present Address ‡

Department of Chemistry, University of Jyväskylä, 40014 Jyväskylä, Finland.

a

The reaction was performed according to general procedure 3. Product ratios and conversions were determined by GLC analysis using tetracosane as internal standard. cThe reaction was run until complete conversion of the trifluoromethyl compound. The percentage is displaying isolated yield after column chromatography.

Notes

b

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This research was supported by the Deutsche Forschungsgemeinschaft (Oe 249/8-1). T.T.M. and P.H. were funded by the Cluster of Excellence Unif ying Concepts in Catalysis of the 16340

DOI: 10.1021/jacs.7b09444 J. Am. Chem. Soc. 2017, 139, 16334−16342

Article

Journal of the American Chemical Society

(20) (a) Postigo, L.; Maestre, M. d. C.; Mosquera, M. E. G.; Cuenca, T.; Jiménez, G. Organometallics 2013, 32, 2618−2624. (b) Kischel, M.; Dornberg, G.; Krautscheid, H. Inorg. Chem. 2014, 53, 1614−1623. (21) (a) Lamberti, M.; D’Auria, I.; Mazzeo, M.; Milione, S.; Bertolasi, V.; Pappalardo, D. Organometallics 2012, 31, 5551−5560. (b) For further crystal structures containing Al−S bonds, see: Uhl, W.; Vester, A.; Hiller, W. J. Organomet. Chem. 1993, 443, 9−17. (22) (a) Young, J. D.; Khan, M. A.; Powell, D. R.; Wehmschulte, R. J. Eur. J. Inorg. Chem. 2007, 1671−1681. (b) Uhl, W.; Appelt, C.; Backs, J.; Klöcker, H.; Vinogradov, A.; Westenberg, H. Z. Anorg. Allg. Chem. 2014, 640, 106−109. (23) A modest correlation (R2 = 0.91) with DFT optimized Ru−H bond lengths was observed (Figure S5 in Supporting Information). A similar correlation but of less quality (R2 = 0.84) is also seen for 1H hydride shifts, which tend to be more deshielded for adducts with a completely broken E−H bond (Figure S6 in Supporting Information). For the important role of relativistic effects on 1H NMR hydride shifts, see for example: (a) Hrobárik, P.; Hrobáriková, V.; Meier, F.; Repiský, M.; Komorovský, S.; Kaupp, M. J. Phys. Chem. A 2011, 115, 5654− 5659. (b) Hrobárik, P.; Hrobáriková, V.; Greif, A. H.; Kaupp, M. Angew. Chem., Int. Ed. 2012, 51, 10884−10888. (24) Pyykkö, P. J. Phys. Chem. A 2015, 119, 2326−2337. (25) Steinke, T.; Cokoja, M.; Gemel, C.; Kempter, A.; Krapp, A.; Frenking, G.; Zenneck, U.; Fischer, R. A. Angew. Chem., Int. Ed. 2005, 44, 2943−2946. (26) Müther, K.; Hrobárik, P.; Hrobáriková, V.; Kaupp, M.; Oestreich, M. Chem. - Eur. J. 2013, 19, 16579−16594. (27) Note that the E−S bonding interactions in hydroborane (E = B) and hydrosilane (E = Si) adducts are more covalent and thereby in line with the electronegativities of the corresponding main-group atoms. (28) However, saturating the coordination sphere of aluminum in parent hydride such as in [(tBuO)3AlH]− hinders the possible σ(Ru− H) → Al donation and is computed to result in an adduct with the complete Al−H bond separation (Table S3 in Supporting Information). (29) Note that the partial atomic charges may be viewed as an approximate measure of the electrophilicity only when steric effects are absent or constant across the investigated series. (30) Christe, K. O.; Dixon, D. A.; McLemore, D.; Wilson, W. W.; Sheehy, J. A.; Boatz, J. A. J. Fluorine Chem. 2000, 101, 151−153. (31) As also confirmed by 1H NMR measurements and NMR shift calculations, the hydride shift for monomeric DIBAL-H is predicted to appear at δ(1H) 4.9 ppm while only two hydride signals at δ(1H) 3.3 and 3.1 ppm are observed in C6D6; similarly, the computed 27Al NMR resonance for monomeric DIBAL-H is about 130 ppm downfield as compared to that found experimentally for DIBAL-H in toluene-d8 (Table S9 in Supporting Information). (32) Hydrodefluorination is also observed with [1a]+[BArF4]− as catalyst in the absence of alkoxide. (33) The formation of DIBAL-F was detected by 19F NMR spectroscopy: δ(19F) −148.9, −148.1, −146.1 ppm in C6D6; cf. ref 9 in CDCl3. (34) Olah, G. A.; Olah, J. A.; Ohyama, T. J. Am. Chem. Soc. 1984, 106, 5284−5290. (35) (a) Tsuchimoto, T.; Tobita, K.; Hiyama, T.; Fukuzawa, S.-i. J. Org. Chem. 1997, 62, 6997−7005. (b) Mertins, K.; Iovel, I.; Kischel, J.; Zapf, A.; Beller, M. Angew. Chem., Int. Ed. 2005, 44, 238−242. (c) Iovel, I.; Mertins, K.; Kischel, J.; Zapf, A.; Beller, M. Angew. Chem., Int. Ed. 2005, 44, 3913−3917. (d) Rueping, M.; Nachtsheim, B. J.; Ieawsuwan, W. Adv. Synth. Catal. 2006, 348, 1033−1037. (e) Schäfer, G.; Bode, J. W. Angew. Chem., Int. Ed. 2011, 50, 10913−10916. (f) Champagne, P. A.; Benhassine, Y.; Desroches, J.; Paquin, J.-F. Angew. Chem., Int. Ed. 2014, 53, 13835−13839. (g) Mo, X.; Yakiwchuk, J.; Dansereau, J.; McCubbin, J. A.; Hall, D. G. J. Am. Chem. Soc. 2015, 137, 9694−9703. (h) Ricardo, C. L.; Mo, X.; McCubbin, J. A.; Hall, D. G. Chem. - Eur. J. 2015, 21, 4218−4223. (i) Vuković, V. D.; Richmond, E.; Wolf, E.; Moran, J. Angew. Chem., Int. Ed. 2017, 56, 3085−3089.

Deutsche Forschungsgemeinschaft (EXC 314/2). M.O. is indebted to the Einstein Foundation (Berlin) for an endowed professorship.

■

REFERENCES

(1) (a) Atwood, D. A. Coord. Chem. Rev. 1998, 176, 407−430. (b) Dagorne, S.; Atwood, D. A. Chem. Rev. 2008, 108, 4037−4071. (2) Kim, K.-C.; Reed, C. A.; Long, G. S.; Sen, A. J. Am. Chem. Soc. 2002, 124, 7662−7663. (3) Young, J. D.; Khan, M. A.; Wehmschulte, R. J. Organometallics 2004, 23, 1965−1967. (4) (a) Klis, T.; Powell, D. R.; Wojtas, L.; Wehmschulte, R. J. Organometallics 2011, 30, 2563−2570. (b) Kessler, M.; Knapp, C.; Zogaj, A. Organometallics 2011, 30, 3786−3792. (c) Ivanov, S. V.; Peryshkov, D. V.; Miller, S. M.; Anderson, O. P.; Rappé, A. K.; Strauss, S. H. J. Fluorine Chem. 2012, 143, 99−102. (5) Engesser, T. A.; Lichtenthaler, M. R.; Schleep, M.; Krossing, I. Chem. Soc. Rev. 2016, 45, 789−899. (6) Stahl, T.; Klare, H. F. T.; Oestreich, M. ACS Catal. 2013, 3, 1578−1587. (7) Klahn, M.; Fischer, C.; Spannenberg, A.; Rosenthal, U.; Krossing, I. Tetrahedron Lett. 2007, 48, 8900−8903. (8) Terao, J.; Begum, S. A.; Shinohara, Y.; Tomita, M.; Naitoh, Y.; Kambe, N. Chem. Commun. 2007, 855−857. (9) Gu, W.; Haneline, M. R.; Douvris, C.; Ozerov, O. V. J. Am. Chem. Soc. 2009, 131, 11203−11212. (10) (a) Ohki, Y.; Takikawa, Y.; Sadohara, H.; Kesenheimer, C.; Engendahl, B.; Kapatina, E.; Tatsumi, K. Chem. - Asian J. 2008, 3, 1625−1635. (b) Lefranc, A.; Qu, Z.-W.; Grimme, S.; Oestreich, M. Chem. - Eur. J. 2016, 22, 10009−10016. (11) (a) Klare, H. F. T.; Oestreich, M.; Ito, J.-i.; Nishiyama, H.; Ohki, Y.; Tatsumi, K. J. Am. Chem. Soc. 2011, 133, 3312−3315. (b) Königs, C. D. F.; Klare, H. F. T.; Ohki, Y.; Tatsumi, K.; Oestreich, M. Org. Lett. 2012, 14, 2842−2845. (c) Königs, C. D. F.; Müller, M. F.; Aiguabella, N.; Klare, H. F. T.; Oestreich, M. Chem. Commun. 2013, 49, 1506− 1508. (d) Königs, C. D. F.; Klare, H. F. T.; Oestreich, M. Angew. Chem., Int. Ed. 2013, 52, 10076−10079. (e) Hermeke, J.; Klare, H. F. T.; Oestreich, M. Chem. - Eur. J. 2014, 20, 9250−9254. (f) Stahl, T.; Hrobárik, P.; Königs, C. D. F.; Ohki, Y.; Tatsumi, K.; Kemper, S.; Kaupp, M.; Klare, H. F. T.; Oestreich, M. Chem. Sci. 2015, 6, 4324− 4334. (g) Omann, L.; Oestreich, M. Angew. Chem., Int. Ed. 2015, 54, 10276−10279. (h) Metsänen, T. T.; Oestreich, M. Organometallics 2015, 34, 543−546. (i) Bähr, S.; Simonneau, A.; Irran, E.; Oestreich, M. Organometallics 2016, 35, 925−928. (12) Stahl, T.; Klare, H. F. T.; Oestreich, M. J. Am. Chem. Soc. 2013, 135, 1248−1251. (13) Stahl, T.; Müther, K.; Ohki, Y.; Tatsumi, K.; Oestreich, M. J. Am. Chem. Soc. 2013, 135, 10978−10981. (14) Omann, L.; Königs, C. D. F.; Klare, H. F. T.; Oestreich, M. Acc. Chem. Res. 2017, 50, 1258−1269. (15) Complex ([1b]+)2[B12Cl12]2− could not be isolated in pure form (see Supporting Information). (16) Coupling constant of [1a·Cy2BH]+[BArF4]− is 2JH,P = 19 Hz. See ref 13 for further details. Coupling constant of [1a· EtMe2SiH]+[BArF4]− is 2JH,P = 50 Hz. See ref 11f for further details. (17) Akitt, J. W. Prog. Nucl. Magn. Reson. Spectrosc. 1989, 21, 1−149. (18) Performance of two-component ZORA-SO method in 27Al NMR shift calculations for a broad range of Al(III) compounds is demonstrated in Table S7 and Figure S9 in Supporting Information. The experimental 27Al NMR data are collected in: Benn, R.; Janssen, E.; Lehmkuhl, H.; Rufínska, A. J. Organomet. Chem. 1987, 333, 155− 168. (19) The surprisingly small deshielding in [1a·iBu2AlH]+ can be rationalized by an extra stabilization of the alumenium ion by electron donation from multiple ligand atoms (see the structural characterization and discussion below). 16341

DOI: 10.1021/jacs.7b09444 J. Am. Chem. Soc. 2017, 139, 16334−16342

Article

Journal of the American Chemical Society (36) (a) For an intermolecular version, see: Douvris, C.; Ozerov, O. V. Science 2008, 321, 1188−1190. (b) For an intramolecular version, see: Allemann, O.; Duttwyler, S.; Romanato, P.; Baldridge, K. K.; Siegel, J. S. Science 2011, 332, 574−577. (37) Lühmann, N.; Panisch, R.; Müller, T. Appl. Organomet. Chem. 2010, 24, 533−537. (38) (a) Zhu, J.; Pérez, M.; Caputo, B. C.; Stephan, D. W. Angew. Chem., Int. Ed. 2016, 55, 1417−1421. (b) Zhu, J.; Pérez, M.; Stephan, D. W. Angew. Chem., Int. Ed. 2016, 55, 8448−8451. (39) Yartys, V. A.; Denys, R. V.; Maehlen, J. P.; Frommen, C.; Fichtner, M.; Bulychev, B. M.; Emerich, H. Inorg. Chem. 2007, 46, 1051−1055.

16342

DOI: 10.1021/jacs.7b09444 J. Am. Chem. Soc. 2017, 139, 16334−16342