Article Cite This: Inorg. Chem. XXXX, XXX, XXX−XXX
pubs.acs.org/IC
Computational Study of Methane C−H Activation by Diiminopyridine Nitride/Nitridyl Complexes of 3d Transition Metals and Main-Group Elements Zhicheng Sun, Olivia A. Hull, and Thomas R. Cundari* Department of Chemistry and Center of Advanced Scientific Computing and Modeling, University of North Texas, 115 Union Circle, #305070, Denton, Texas 76203-5017, United States S Supporting Information *
ABSTRACT: The C−H bond activation of methane using Ph,MePDI−M≡N [Ph,MePDI = 2,6-(PhNCMe)2C5H3N] (M = V, Mn, Fe, Co, Ni, Al, or P) has been studied via three reaction pathways: [2σ + 2π] addition, hydrogen atom abstraction (HAA), and direct insertion. The activating ligand is a nitride/nitridyl (N), with diiminopyridine (PDI) as the supporting ligand. Calculations show reasonable C−H activation barriers for Co, Ni, Al, and P Ph,MePDI nitrides, complexes that favor an HAA pathway. Electrophilic Ph,Me PDI nitride complexes of the earlier metals with a nucleophilic actor ligandV, Mn, Fefollow a [2σ + 2π] addition pathway for methane activation. Free energy barriers for methyl migration, Ph,Me PDI−M(CH3)NH → Ph,MePDI−M−N(H)CH3, are also interesting in the context of alkane functionalization; discriminating factors in this mechanistic step include the strengths of the σ-bond and metal-actor ligand π-bond that are broken and the electrophilicity of the actor ligand to which methyl migrates.
1. INTRODUCTION
The great versatility of 2,6-diiminopyridine (PDI) with transition metals and main group elements have been demonstrated through the considerable catalysis investigation,1,13−15 and X-ray crystallography characterized for a broad variety of metal ions, Figure 1.1 This demonstrates the great versatility of PDI in coordination chemistry. Reported
Carbon−hydrogen bond functionalization via organometallic catalysis is an area of intense study. Conversion of cheap, abundant hydrocarbon feedstocks to more chemically useful products is a main goal; however, due to the high strengths of the C−H bonds in low molecular weight alkanes, activation of these C−H has proven difficult to accomplish.1,2 Current methods to achieve alkane conversion, such as the Fischer− Tropsch process, are extremely energy consuming.3 The challenges associated with C−H functionalization lend themselves to the field of organometallic catalysis. The development of an efficient catalytic method would greatly benefit the industrial sector and have a tremendous economic impact. Alkane functionalization under mild conditions has garnered vast attention. Aliphatic Csp3−H bonds of light alkanes have very low acidity (pKa ≈ 50−60). The tightly held highest occupied molecular orbital (HOMO) electrons are unlikely to participate in electron donation to all but the strongest Lewis super acids, with deprotonation similarly limited to super bases.4−6 Precious metal complexes are well-known for performing C−H activation, for example, Rh,7 Os,8 Ir,9 and Pt.10 Calculations reported by Cundari and co-workers suggest that later 3d transition metal (Fe, Co, Ni, Cu) imide and oxide complexes are feasible for the catalytic functionalization of methane C−H bonds.11,12 © XXXX American Chemical Society
Figure 1. Conformation of isolated PDI molecule and elements to which PDI has been reported to coordinate.1 Adapted from ref 1 and https://upload.wikimedia.org/wikipedia/commons/6/61/Periodictable.jpg (last accessed 9/12/17). Received: January 2, 2018
A
DOI: 10.1021/acs.inorgchem.7b03212 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry applications include olefin polymerization,16−18 nitrogen− hydrogen bond activation,19 hydrosilylation,20 and hydrogenation.21 Biernesser et al. reported the polymerization of lactide catalyzed by bis(imino)pyridineiron(II)-bis(alkoxide).22 PDI is renowned for its redox noninnocence.23−25 An intriguing feature of PDI is its ability to serve as an electron reservoir to source and drain electrons to/from a metal active site.1,26 Only deep, multielectron reductions may affect the oxidation state of the metal center in PDI complexes.27,28 Noninnocent or redox-active ligands like PDI are those with oxidation state(s) determined by experiments or calculations that differ from predictions made from their Lewis structures.24 The Lewis structure of PDI implies a neutral ligand. When a transition metal is coordinated to a redox noninnocent ligand, the electronic properties of both moieties are altered.1,23 In this study, PDI is expected to serve as a noninnocent ligand where it receives or donates electrons from/to the active site along the reaction coordinate. Atienza et al. proposed that a diiminopyridine Co intermediate with a terminal nitride actor ligand, Ph,MesPDI− CoN (Mes = mesityl), has the ability to effect intramolecular C−H activation of a benzylic C−H bond.29 It is plausible for an open-shell Co-nitride with spin density on the active nitrogen, i.e., a nitridyl complex, to activate C−H bonds via hydrogen atom abstraction (HAA).29 Presumably, an actor ligand (N) with strong radical character combined with a redox-active ligand (Ph,MesPDI) made this reaction achievable. Terminal nitride complexes of transition metals, LnM≡N, have various applications, including intermediates in N-transfer and nitrogen fixation.30 Terminal Fe-nitrides with different oxidation states and coordination numbers have been synthesized, and their reactivity has been studied.31−34 Terminal nitrides coordinated to precious metals have also been characterized. For example, Caulton and co-workers reported the crystallographic characterization of a Ru≡N complex with a formally d4 configuration (assuming nitride is in its typical N3− state).35 Schneider et al. and Burger et al. reported that coupling of nitride ligands may occur to form more thermally favored bridging dinitrogen complexes, which potentially compete for C−H activation.36,37 But dimerization of nitride was not a major focus of this research. Burger and co-workers reported a well-characterized PDI iridium terminal nitride complex that is hydrogenated to an amide complex, which is analogous to the stoichiometric reaction being modeled here with methane, viz nitride → amide.38a In a following study, they proposed intramolecular C−H activation by a terminal Rh≡N moiety to give a diradical imido intermediate (RhNH) via HAA and direct C−H bond addition for Ir≡N yielding amido products based on both experiments and density functional theory (DFT) calculations.37 The mechanistic study of Si−H activation by a PDI iridium nitride by Sieh et al. also offers insight computationally for C−H activation in methane using natural bond orbital (NBO) analysis, which shows greater interaction in silane than that in methane due to the low-lying Si−H σ*-acceptor orbital.38b Lau and co-workers characterized (salen)Ru(IV) terminal nitride species that could perform intermolecular C− H activation in alkanes via hydrogen atom transfer/abstraction followed by radical rebound giving amide product.39 They then further characterized terminal nitride complexes that could perform C−H bond activation, N···N coupling, and catalytic oxidation with Ru(VI), Os(VI), and Mn(V) complexes bearing Schiff base supporting ligands.40 Some first-row transition metal
nitride complexes capable of activating C−H bond have been reported. Smith and co-workers have isolated Fe(IV) nitrides that C−H activate 1,4-cyclohexadiene via an HAA mechanism.41 Similar to the work of Atienza et al., Zolnhofer et al. proposed a low-spin terminal Co(IV)-nitride complex after photolysis of Co(II) azide; due to its high reactivity, the nitride intermediate immediately underwent intramolecular N-migratory insertion (ΔG‡ = 2.2 kcal/mol) and subsequent Habstraction according to DFT calculations.42 Cui et al. very recently provided evidence of a PDI-supported Co2 bridging nitride intermediate that activated aliphatic C−H bonds.43 These precedents, along with the large number of elements to which PDI may bind, suggest that Ph,MePDI−M≡N and related complexes may be reasonable models of synthetically feasible complexes to activate and functionalize strong C−H bonds. Inspired by the aforementioned experimental precedents, DFT calculations are employed in the present research to investigate the feasibility of intermolecular C−H bond activation involving 3d metal and main group element complexes of the form LM≡N, where L = methyl, phenylsubstituted-2,6-bis(imino)pyridine ligand (Ph,MePDI; M = V, Mn, Fe, Co, Ni, Al, P), Figure 2. Three reaction pathways are
Figure 2. Methyl, phenyl-substituted bis(imino)pyridine metal nitride complex, Ph,MePDI−MN; M = V, Mn, Fe, Co, Ni, Al, P; nitride/nitridyl is the actor ligand.
modeled: [2σ + 2π] addition, hydrogen atom abstraction (HAA), and C−H insertion, Figure 3. The free energies of each pathway are calculated, thus determining the most likely route for methane functionalization for each Ph,MePDI−metal-nitride complex studied.
Figure 3. Three plausible transition states for C−H activation of methane via Ph,MePDI metal-nitride complexes: [2σ + 2π] addition, hydrogen atom abstraction (HAA) and insertion.
2. COMPUTATIONAL METHODS The Gaussian 0944 software package is employed along with DFT for all reported calculations. Research for similar complexes has shown that the Minnesota functional M0645 is suitable for modeling of complexes akin to those being studied in this project.11 We choose the 6-31+G(d) basis set to model the target complexes. All calculations are done in the gas phase and assume 1 atm and 298.15 K. Nitride, imide, and amide complexes are overall neutral; high, intermediate, and low spin states are evaluated for each. Three methane activation transition states are sought for each metal: [insertion]‡, [HAA]‡, and [2σ + 2π]‡, Figure 2. The radical rebound step following hydrogen atom abstraction (HAA) is assumed to be barrierless. Methane and methyl radical are calculated as a singlet and doublet, respectively. Free energies are reported in kcal/mol. Single-point MCSCF and CASPT2 calculations at DFT-optimized geometries employed the complete active space (CAS) formalism, the 6-31G(d) basis set, and the GAMESS46 code. B
DOI: 10.1021/acs.inorgchem.7b03212 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry
d7-Co2+)↓(N−)↑↑] and 2[(PDI•−)↓(LS-d8-Ni2+)(N−)↑↑] descriptions, albeit with extensive mixing of other resonance structures. The manganese reactant and others have more complex electronic structures, for example, 3[(PDI•‑)↓(HS-d4Mn 3+) ↑↑↑↑(N 2−) ↓] ⇔ 3[(PDI•−) ↓(HS-d 3-Mn 4+) ↑↑↑ (N 3−)]. These MCSCF calculations implicate significant multireference character to the ground states of these nitrides, echoing the work by Ortuño and Cramer on Fe-PDI complexes.13 CASPT2 were performed for all PDI nitride/nitridyl complexes. CASPT2, CASSCF, and DFT results corroborate each other except for PDI-CoN (state ordering is unchanged, but the magnitude changes by 77 kcal/mol, likely due to the intruder state). Moreover, CASSCF and CASPT2 results differ by an average of 2−3 kcal/mol. The one exception is PDI-VN, which DFT says is T < S by ∼8 kcal/mol; CASPT2 says S < T by a mere 0.3 kcal/mol. The geometry of Ph,MePDI−MN with M = Mn, Fe, Co, Ni is closest to square planar with the sum of the four N−M−N angles ∼360°, while the geometries of Ph,MePDI−VN and Ph,Me PDI−AlN are closer to disphenoidal. Interestingly, the geometry of Ph,MePDI−PN is approximately tetrahedral. Additional geometric details are given in the Supporting Information. 3.2. Pathways of Methane Activation and Functionalization. The Ph,MePDI−MN was computed to activate the C−H bond in methane through one of three different pathways: [2σ + 2π] addition, insertion, and hydrogen atom abstraction (HAA), Figure 4. Calculations of Ph,MePDI−MN reacting with methane revealed an interesting trend with respect to metals and the preferred pathway of C−H activation.
3. RESULTS AND DISCUSSION 3.1. PDI Nitride Complexes. The primary goals of this research are to understand how metal identity affects C−H activation/functionalization barriers and the preferred reaction pathway, both of which are critical design considerations for alkane conversion catalysts. The overall charge on all reactant complexes is zero. PDI is a neutral ancillary ligand, when not behaving in a redox noninnocent fashion, and the formal charge of the actor nitride ligand is −3, which yields a +3 metal center for the reactant complex. DFT and MCSCF calculations (vide infra) suggest that the electronic structure of these open-shell complexes is more complex than this simple description. All DFT-calculated metal-nitride/nitridyl bond lengths are shown in Table 1. Also included are experimentally Table 1. Experimental and DFT-Optimized (Ph,MePDI− M≡N) M≡N Distances (Å), Calculated Ground State Multiplicity (Mult.), Metal Geometry, and Mspin Density (ρspin) on Metal and Nitride/Nitridyl Nitrogen Are Givena
V≡N Mn≡N Fe≡N Co≡N Ni≡N Al≡N P≡N
expt M≡N (Å)
calcd M≡N (Å)
1.59847 1.56148 1.51249 1.66750 1.61951 1.70552 1.47453
1.554 1.624 1.589 1.729 1.741 1.843 1.504
mult.
geometry at metal
ρspin (M) (e−)
ρspin (N) (e−)
3 3 2 3 2 1 1
disphenoidal square planar square planar square planar square planar disphenoidal tetrahedral
1.25 3.37 2.28 −0.69 0.17 0.00 0.00
−0.47 −0.77 −0.47 1.79 1.74 0.00 0.00
a
Full complex names for the experimental reference complexes are given in the citations. V≡N: (nacnac)V≡N(N[Mes]tol), (nacnac) = [ArNC(CH3)]2CH−, Ar = 2,6-(CHMe2)2C6H3); Mn≡N: (salen)Mn≡N·hexane; Fe≡N: Ph-tBuTpFe≡N·NCMe; Co≡N: Ph-tBuBP3Co≡N−pTol·C6H6; Ni≡N: Tp*Ni≡N+−O−; Al≡N: No Al≡N model was deposited in the CCDC, so an Al-imide (AlN) is used for a rough comparison. (iPr,MeNHC)Al(NDipp)(κ2-C,N-tBuC = CH−CC(tBu)N(Dipp)); P≡N: [P = N-2,4,6-C6H2tBu3]AlCl4· C7H8.
characterized metal-nitride complexesor the most reasonable facsimile thereofof each studied metal to calibrate the optimized M≡N bond lengths even though the experimental systems in Table 1 are not necessarily supported by PDI. Bona f ide Al and P nitrides have not been characterized, to our knowledge, so Al and P imido compounds are shown in order to provide a rough estimate of the computed M≡N bonds. The two PDI nitride complexes with main group elements, Al and P, have closed-shell singlet ground states; uM06 calculations with the guess = mix option in Gaussian09 were utilized in an attempt to find open-shell singlet solutions, but these yielded very small spin densities, thus collapsing to a closed-shell singlet solution. The V, Mn, and Co complexes each have a triplet ground state, and a doublet is predicted for both the Fe and Ni reactants, Table 1. Both Hirshfeld and Mulliken population analysis were performed since the latter is more basis set sensitive. Results from the two population analysis schemes are very similar. According to Mulliken analyses, both the Ph,MePDI supporting ligand and N actor ligand have substantial spin density in the ground electronic state of these “nitrides.” The vanadium reactant, for example, is best viewed as3(PDI•‑)↑(d1V4+)↑(N3−)] from both DFT and MCSCF (CAS(12,12)) calculations, while the DFT-derived Mulliken spin densities and CAS(12,12)/6-31G(d) wave functions suggest 3[(PDI•‑)↑(LS-
Figure 4. Reaction pathways computed for C−H activation of methane by PDI metal-nitride complexes forming an amide product: [2σ + 2π] addition (top), insertion (middle) and hydrogen atom abstraction (HAA) (bottom).
In methane C−H activation via a [2σ + 2π] pathway, a fourcentered transition state is followed by formation of a methyl/ imide intermediate Ph,MePDI−M(CH3)NH, Figure 4. The metal center is expected to maintain its formal oxidation state upon [2 + 2] activation, and the coordination geometry of the metal center changes to square pyramidal. The methyl group of Ph,Me PDI−M(CH3)NH then migrates from the metal center to the imide nitrogen giving amide product Ph,MePDI−M− N(H)CH3, during which the metal center is formally reduced from +3 to +1. In the insertion pathway, the nitrogen of the nitride interposes itself between hydrogen and methyl group of the substrate, yielding a three-centered transition state; the C
DOI: 10.1021/acs.inorgchem.7b03212 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry
Figure 5. Ph,MePDI−M-nitride complexes that are calculated to favor a hydrogen atom abstraction pathway for C−H activation: Al (black), P (red), Ni (blue), Co (green). Calculated ground state multiplicities are denoted to the right of the elemental symbols. Note that direct formation of the amide via radical rebound of the metal onto the imide N of the (Ph,MePDI)MNH intermediate was assumed given the thermodynamic stability of the amide product; however, methyl migration is an important step in alkane functionalization and so was calculated for completeness. The quoted values are free energies in kcal/mol and are relative to separated reactants, Ph,MePDI−MN plus methane.
Mn nitrides nor the main group nitrides despite multiple initial starting guess geometries.
amide product is formed directly with no intervening steps. In the HAA pathway, a hydrogen radical is abstracted from methane giving an imide intermediate Ph,MePDI−MNH and a methyl radical. Upon C−H activation in the HAA pathway, the metal center is formally reduced from +3 to +2. Methyl radical rebound to the imide N then leads to formation of the amide product, Figure 4. These changes in metal and actor ligand formal oxidation states assume, of course, that the spectator and actor ligands maintain their redox poise during the course of each particular chemical step. The geometries and free energy of methane-complex adducts were also calculated and compared to separate reactants. The free energy of methanecomplex adducts (ΔGadd) for vanadium and aluminum is slightly lower than that of separate reactants (ΔGadd = −2.7 kcal/mol for V and −10.2 kcal/mol for Al). The Mn, Fe, Co, Ni, P methane-complex adducts have slightly higher free energies than separate reactants (ΔGadd = 3.0 kcal/mol for Mn, 8.3 kcal/mol for Fe, 6.0 kcal/mol for Co, 5.4 kcal/mol for Ni, 5.1 kcal/mol for P). Bond distances between the actor ligand N and the closest H on methane are in the range of 2.2 to 2.8 Å, and the methane is slightly perturbed from its isolated geometry in these adducts. Both the free energy differences and the geometric changes indicate that there is a weak interaction between methane and PDI complex. Mulliken charge analysis shows a very small positive charge on methane (
