Homolytic N-H activation of ammonia: hydrogen transfer of parent iridium ammine, amide, imide, and nitride species.

The redox series [Ir(n)(NHx)(PNP)] (n = II-IV, x = 3-0; PNP = N(CHCHPtBu2)2) was examined with respect to electron, proton, and hydrogen atom transfer steps. The experimental and computational results suggest that the Ir(III) imido species [Ir(NH)(PNP)] is not stable but undergoes disproportionation to the respective Ir(II) amido and Ir(IV) nitrido species. N-H bond strengths are estimated upon reaction with hydrogen atom transfer reagents to rationalize this observation and are used to discuss the reactivity of these compounds toward E-H bond activation.

Introduction

The chemistry of transition metal (TM) complexes with covalently bound nitrogen ligands, i.e. amido, imido, and nitrido species LM–NR, has undergone a renaissance in recent years due to their relevance to catalytic transformations.[1] For example, efficient nitrene (N–R) transfer protocols, such as C–H amination (Scheme 1) or olefin aziridination, rely on TM catalysts,[2] particularly with late, electron rich transition metals, like rhodium and copper. At the current point of mechanistic understanding, the nitrenoid character of reactive LM–NR intermediates determines their reactivity relevant to catalysis. The electronic structures of these transient species and the precise mechanistic pathways of nitrene transfer are subjects of current experimental and computational studies. However, some general features of late TM bonding to strong π-donors like imido and nitride ligands provide qualitative rationalization. Decreasing valence d-orbital energies along the transition series[3] suggests more and more covalent metal nitrogen bonding. Furthermore, rising d-orbital occupancy can lead to destabilizing repulsive M(d)–N(p) filled–filled interactions for late transition metal ions. These trends parallel several experimental findings:

Scheme 1

Generation and Transfer of Nitrene Species via Two- (red) and One-Electron (blue) Oxidation Pathways

–The bond energies of the diatomic gas phase species [M≡N]+ decrease from Ti to Cu.[4] This observation is also reflected in the scarcity of terminal imido and particularly nitrido complexes of TMs beyond group 8 (see below).

–While early TM complexes typically exhibit distinct nucleophilic imide or nitride ligand reactivity, late TM examples often show increased electrophilic (or ambiphilic) nitrogen reactivity. Within a simple molecular orbital (MO) picture, the higher degree of covalency translates into an increasing N(p) orbital contribution to the M–N π* MOs. These often define the LUMO of closed-shell imido or nitrido complexes, providing a rationale for increasing N electrophilicity (or ambiphilicity) for such compounds.

–In turn, species that result from their (formal) reduction, i.e. with populated M–N π* antibonding levels, exhibit reduced M–N bond order and increased nucleophilic (or ambiphilic) N reactivity. Furthermore, ligand centered radical reactivity might be attributed to delocalization of spin density on the nitrogen ligand.

Similar considerations as for nitrene reactivity apply to nitrene generation. Such species are generally obtained from reaction with preoxidized nitrogen sources, such as organic azides or iminoiodinane reagents from in situ oxidation of amines with hypervalent iodine oxididant. Hence, C–H amination usually follows a two-electron nitrene generation and transfer cycle (Scheme 1).[5] Warren and co-workers reported copper catalyzed amination of benzylic substrates, directly using primary and secondary amine nitrogen sources and di-tert-butylperoxide as an oxidant.[6] The authors showed that a copper(II) intermediate with amido ligand radical character undergoes HAA from the hydrocarbon substrate. In analogy, a two-step, one-electron redox regime was proposed by Berry and Kornecki for dirhodium catalyzed nitrene transfer with amine and cerium(IV) as oxidants, as sketched in Scheme 1 (blue cycle).[7] This approach might be promising with respect to the direct employment of amines, or even ammonia, as substrates in C–H amination. It could also afford the use of more sustainable chemical oxidants or electrocatalytic amination. Furthermore, besides partial amine oxidation, full oxidation of ammonia to dinitrogen represents the anodic half-reaction in direct ammonia fuel cells (DAFC), which are under investigation in the context of nitrogen based energy storage schemes.[8] In turn, the reverse reaction, i.e. electrochemical nitrogen fixation, is a highly desirable goal for electrocatalysis.[9]

Despite this general interest in amine, especially ammonia oxidation, it is far from having been broadly examined. As for parent ammonia, both the high bond dissociation enthalpy (BDEN–H: 104 kcal mol–1)[10,11] and the low acidity (pKadmso = ∼41)[12] render activation at mild conditions a challenging task. Redox-neutral N–H heterolysis of ammonia is observed upon protonolysis of basic ligands, e.g. hydrides or alkyls (Scheme 2a), upon 1,2-addition across metal–ligand multiple bonds (e.g., M=CR2), or upon proton transfer to other basic groups in the first or second coordination sphere (“metal–ligand cooperative activation,” Scheme 2b).[13] In contrast, ammonia oxidative addition to a single (Scheme 2c) or two metal centers (Scheme 2d) is rare;[14,15] homolytic activation upon HAA from TM coordinated NH3 (Scheme 2e) was not reported.

Scheme 2

Transition Metal Mediated Ammonia Activation

In this forum article, we first want to discuss the previously reported reactivity of TM imido and nitride complexes beyond group 8, which is relevant to catalytic transformations like nitrene transfer. Some comprehensive, excellent reviews on imide/nitride chemistry were published in recent years.[1] Therefore, we do not intend to give an exhaustive summary. Instead, we want to emphasize the role of open-shell species with partial nitrogen ligand radical character with the help of experimentally well-defined examples. In the main body of this paper, we then describe an experimental and computational study of a novel iridium(II–IV) parent amido, imido, and nitrido redox series and discuss the thermochemistry of hydrogen transfer in the context of ammonia activation.

Imido Complexes

Within group 9, pseudotetrahedral MIII (M = Co, Rh, Ir) imidos are best characterized and were reported from the groups of Bergman, Peters, Meyer, Theopold, Smith, and Tejel (Figure 1).[16] Notably, complex E was not prepared with an azide precursor but from the respective cobalt(II) amido complex upon HAA with the 2,4,6-tri-tert-butylphenoxy radical. This reaction is consistent with a computed N–H bond dissociation enthalpy in the amido complex of 75 kcal mol–1.[16g] Hence, HAA from an unactivated hydrocarbon will be thermochemically unfavorable. Complexes A–F exhibit two-electron, electrophilic imido group reactivity, such as NR transfer to phosphines, CO, or isocyanate, as expected from their low-spin (S = 0) ground states. Interestingly, Theopold et al. (Figure 1, F) and Betley et al. (Scheme 3) reported Co imido complexes with thermally accessible open-shell states.[16e,16f,17] In fact, these compounds also undergo radical-type nitrene group transfer upon intramolecular insertion into benzylic C–H bonds (Scheme 3).

Figure 1

Pseudotetrahedral group 9 imido complexes with low-spin ground states.[16]

Scheme 3

Intramolecular C–H Amination Reported by Betley and Co-workers[17]

The groups of De Bruin and Zhang exploited imido radical reactivity for C–H bond amination and olefin aziridination catalyzed by cobalt porphyrin complexes (Scheme 4).[18] The nitrene-radical character for the key intermediates formed with organic azides (mononitrene H) and iminoiodinanes (bisnitrene G), respectively, was most recently demonstrated spectroscopically.[18g]

Scheme 4

(Top) Proposed Mechanisms for C–H Amination and Aziridination Catalyzed by Cobalt(II) Porphyrin Complexes and (Below) Formation of Spectroscopically Characterited Key Intermediates G and H (Ligand L Currently Remains Unclear, but Is Likely an Amine or Amido Donor)[18]

The three-coordinate, closed-shell nickel complexes with chelating bisphosphine and bis-NHC-carbene of type I (Scheme 5) that were established by Hillhouse et al. represent a key system in the exploration of late transition metal imido chemistry.[19] Notably, the Ni–N–R angles were found to vary from linear (180°) to strongly bent (127°). Detailed bond analysis revealed a reduced Ni–N bond order, i.e. considerably smaller than three also for virtually linear examples, which is expressed in distinctly nucleophilic imide reactivity, e.g. upon [2 + 2] cycloaddition with aldehydes or heterocumulenes.[20] Hence, oxidation to [L2Ni(NR)]+ should result in ligand centered radicals, which is in fact supported by DFT and reactivity:[21] Reaction of [(diphosphine)Ni(NAd)] with ferrocenium results in HAA even from diethyl ether to give nickel(II) amido complexes (Scheme 5). Alternatively, aryl C–C coupling was found for a dicarbene arylimido analogue (Scheme 5), attributable to spin-delocalization into the aromatic moiety.[22] This reactivity was previously reported, e.g. for dirhodium and nickel aryl imido complexes by Sharp et al. and Stephan and Bai, respectively.[23]

Scheme 5

HAA Reactivity and Aryl Coupling of Nickel(III) Imido Radical Complexes (Ad = 1-Adamantyl)

The HAA reaction of transient [(diphosphine)NiIII(NAd)]+ with weak hydrogen donors like diethyl ether contrasts with the stability of [(diphosphine)NiII(NAd)], e.g. in THF. In fact, the nickel(II) arylimides can be prepared by HAA of the respective NiI amide with the 2,4,6-tri-tert-butylphenoxyl radical.[19c,20b] This allows for an estimate of the upper limit of BDEN–H in the nickel(I) anilides of around 82 kcal mol–1. That value is around 10 kcal mol–1 smaller than BDEC–H data of typical ethereal solvents,[24] which is in agreement with the absence of solvent radical activation by the nickel(II) imides. Albeit only qualitative in nature and omitting kinetic influences, this comparison between the nickel(II) and nickel(III) imides showcases how radical reactivity is promoted by one-electron redox steps leading to ligand radical character.

Warren and co-workers proposed redox noninnocent imido ligand character for S = 1/2 nickel complex J (Scheme 6),[25] supported by large nitrogen hyperfine interaction detected by EPR and computational analysis. The ligand radical character is reflected in the HAA reactivity:[25,26] Amido complex K and benzene are obtained from the reaction with 1,4-cyclohexadiene. With benzylic hydrocarbon substrates, mixtures of amides K and L and the amine complex M are obtained, which was rationalized by HAA and subsequent competitive trapping of the hydrocarbyl radical.

Scheme 6

Benzylic C–H Activation with Diketiminato Nickel(III) Imido Complexes[26]

Efficient, catalytic adamantyl nitrene C–H insertion was reported for an analogous diketiminato copper system by Warren and co-workers.[27] The respective terminal copper nitrene complex was proposed as a reactive intermediate but remains elusive. In an elegant study, Ray and co-workers demonstrated the stabilization of a transient copper nitrene complex by various Lewis acids (Scheme 7).[28] Characterization of the scandium adduct by XAS, UV–vis, and Raman spectroscopies afforded a detailed model of the (electronic) structure. Importantly, the Lewis acid adducts still undergo HAA reactivity and nitrene C–H insertion with hydrocarbons.

Scheme 7

Lewis Acid Stabilization and HAA Reactivity of a Copper Nitrene Complex[28]

Nitrido Complexes

The chemistry of terminal transition metal nitrido complexes beyond group 8 is much less developed. Well-defined nitrides are restricted to group 9. The first example, iridium pyridinediimino (PDI) nitride N, was reported by Burger and co-workers in 2009 (Figure 2).[29] The electronic structure of diamagnetic N was described as [(IrIII)(N–)(PDI2–)]. The resulting ambiphilic nitride character is expressed, e.g. by high basicity and the facile electrophilic nitride insertion into H2 and silane Si–H bonds through a concerted, two-electron mechanism.[29a,30]

Figure 2

Characterized group 9 terminal nitrido complexes.

Schneider and co-workers reported the isolation of electrophilic, closed-shell nitride [Ir(N)(PNP)]PF6 (1; PNP = N(CHCHPtBu2)2; Figure 2) upon oxidation of iridium(II) azide 2a (Scheme 8).[31] One-electron reduction of closed-shell nitride 1 or irradiation of 2a or the rhodium analogue, respectively, afforded the open-shell (S = 1/2) formally MIV nitrides [M(N)(PNP)] (M = Ir (3a), Rh (3b)).[31,32] Spectroscopic characterization indicated distinct π-radical (partial “nitridyl”) character of the M–N fragment with M–N π* antibonding character of the SOMO. The essentially equal distribution of excess spin over the metal and the nitride moiety underlines the strong covalency of iridium and rhodium nitride bonding. The transient nitrides undergo rapid, three-electron nitride coupling in solution toward bridging dinitrogen complexes [(PNP)M(μ2:η1:η1-N2)M(PNP)] (M = Ir (4a), Rh (4b)).

Scheme 8

Formation of Iridium and Rhodium Nitrides and Coupling to N2[31,32]

Meyer and co-workers recently reported the EPR spectroscopic characterization of a transient, terminal cobalt(IV) nitride in frozen solution (Figure 2, O).[33] According to the computational analysis, highly covalent M–N bonding was pointed out leading to considerable nitride ligand radical character, as in case of 3a/b. In contrast to 3a/b, this compound decays upon intramolecular nitride insertion into a Co–carbene bond.

De Bruin and Schneider also disclosed the synthesis and isolation of a nitrido bridged RhII/RhIII complex using a sterically less protecting “pincer” ligand (Scheme 9, P).[34] Spectroscopic and computational analysis indicate an S = 1/2 ground state and spin delocalization over the Rh–N–Rh π-manifold with considerable spin density on the bridging nitrogen ligand. Reaction with CO yields the respective rhodium(I) carbonyl complex (Scheme 9). The proposed mechanism proceeds via dimer dissociation and nitride coupling of a transient, terminal rhodium(IV) nitride, as was directly observed for 3b.

Scheme 9

Proposed Mechanism for the Reaction of Bridging Nitride L with CO[34]

Nitride coupling to N2 was also reported for several octahedral MV group 8 nitrides,[35] and these low-spin complexes share an isolobal relationship with the square-planar MIV group 9 nitrides (c.f. Discussion). In the context of nitrogen fixation, it is appealing to compare these with systems that undergo the microscopic reverse, i.e. N2 splitting, through linear dinitrogen (or diazenido) complexes.

From simple qualitative orbital symmetry considerations, the {M–N–N–M} core in idealized rotational symmetry gives rise to a manifold of four degenerate pairs of π-MOs (Scheme 10).[36] The second lowest pair is M–N bonding and N–N antibonding in nature, while the next higher pair exhibits the opposite properties. The {M–N–N–M}[8] π-electron configuration features strong N2 activation, as in the case of Bercaw’s [Cp2*(N2)Zr(N2)Zr(N2)(Cp*)2].[37,38] However, the N–N σ*-MO in the N2-bridged complex drops upon N2 splitting into nitrides (Scheme 10), which proceeds through a zigzag transition state. Therefore, the {M–N–N–M}[8] π-electron configuration is two electrons short to fill all bonding σ- and π-orbitals in the resulting nitrides upon N2-splitting.[39] Hence, the {M–N–N–M}[10] π-configuration should be ideal to undergo N–N bond cleavage into stable nitrides.

Scheme 10

Qualitative Orbital Interaction Diagram for N2 Splitting/Coupling from/to a Linearly N2-bridged Dinuclear Complex for the {M–N-N-M}[10] (blue) and the {M–N-N-M}[12] π Configurations (red) (adapted with permission from ref (39), copyright 1996 American Chemical Society)

Only the relevant M–N σ- and π-type orbitals are shown.

In fact, these simple considerations apply to several systems in different coordination environments and molecular symmetries that were reported to split dinitrogen into terminal nitrides via linear μ2:η1:η1-N2 intermediates.[40] For example, each Mo ion in (tBuArN)3Mo(N2)Mo(NtBuAr)3, which splits into nitrides, contributes three electrons to the π system, resulting in an overall {M–N–N–M}[10] π configuration.[39] In turn, the open-shell nitrides 3a/b form dinitrogen-bridged complexes with square-planar coordinated MI (M = Rh, Ir). Hence, the two filled orbitals with Ir–N π symmetry give rise to a {M–N–N–M}[12] configuration since each metal provides four π electrons. The same applies to the octahedral MV (M = Fe–Os) nitrides that were shown to couple to N2.[35] Another interesting example is provided by Peters’ closed-shell iron(IV) nitride [(BP3)FeN] (BP3– = PhB(CH2PiPr2)3–), which couples to the N2-bridged complex [(BP3)Fe(N2)Fe(BP3)].[41] There, the metal ions were described as high-spin FeI in C3 symmetry resulting in a (d,d)4(d)1(d,d)2 d-electron configuration. Hence, each iron atom contributes only two electrons to the π-manifold giving a {M–N–N–M}[8] π configuration in the dimer, which is in line with the overall thermochemistry that favors nitride coupling.

Although these electronic structure/reactivity considerations are very simplistic in nature, they provide some rationale for the underlying driving force of N2 splitting and coupling. The relevant M–N bond strengths within these nitride vs dinitrogen complexes are certainly governed by many other contributions, which are beyond the scope of these rationalizations just based on orbital symmetry and occupation.

Interestingly, some terminal nitrides were directly demonstrated to undergo C–H amination or proposed as transient intermediates in HAA reactions toward amido or imido complexes.[42] Within group 9, Chirik observed intramolecular nitrogen insertion into a secondary C–H bond upon irradiation of a cobalt azide complex (Scheme 11).[43] In comparison, the generation of 3a/b in the presence of H atom donors like 1,4-cyclohexadiene also resulted in nitride coupling, exclusively.

Scheme 11

Intramolecular C–H Amination upon Irradiation of a Cobalt Azide[43]

This striking preference of the open shell nitrides 3a/b toward N2 formation over HAA will be examined in the next part of this paper. Within an experimental and computational study, we describe the reactivity of a redox series of iridium(II–IV) pincer complexes with parent ammine, amide, imide, and nitride ligands relevant to ammonia functionalization. Particular emphasis was put on estimating N–H BDEs of M–NH species as they relate to the thermochemistry of HAT reactivity, such as radical C–H activation.

Results

Syntheses

Salt metathesis of iridium(II) chloro complex [IrCl(PNP)] (5)[44] with excess sodium amide in the presence of 15-crown-5 results in the selective formation of parent amido complex [Ir(NH2)(PNP)] 6 in over 60% isolated yield (Scheme 12). The magnetic moment of 6 (μeff = 1.82 μB, Evans’ method) is in agreement with an S = 1/2 ground state. The X-band EPR spectrum (see Supporting Information) reveals a rhombic g tensor (g = 2.72, g = 2.12, g = 1.93) without resolved hyperfine interactions, characteristic for a metal dominated SOMO and predominant metal centered spin density (Ir, 68%; N, 27%; see Supporting Information). This observation contrasts with the formation of an aminyl radical complex upon oxidation of rhodium(I) amido complex [RhI(trop2N)(PR3)].[45] The large g anisotropy of 6 is common for square planar d7 complexes and is well reproduced by supporting DFT calculations (g = 2.69, g = 2.06, g = 1.96; see Supporting Information). The 1H NMR spectrum of 6 exhibits three paramagnetically broadened and shifted signals for the pincer tBu and CH groups, which is in agreement with the square-planar molecular structure obtained from X-ray diffraction (see below).

Scheme 12

Syntheses of Iridum(II) and Iridium(III) Parent Amine and Amido Complexes

Protonolysis of 6 with HCl restores 5 quantitatively (Scheme 12). In contrast, the use of an acid with a weakly coordinating anion, [H(OEt2)2][BAr4F] (BAr4F = B(C6H3(CF3)2)4), allows for the isolation of the cationic iridium(II) ammine complex [Ir(NH3)(PNP)][BAr4F] (7). Single-crystals suitable for X-ray diffraction could be obtained as SbF6-salt (see below). The X-band EPR spectrum of 7 (see Supporting Information) reveals an even more rhombic g tensor (g = 3.36, g = 1.82, g = 1.49) also without resolved hyperfine interactions. The g anisotropy is well reproduced by DFT calculations, which confirm predominant metal-radical (65%) character (see Supporting Information).

Chemical oxidation of 6 with AgPF6 affords the isolation of the iridium(III) parent amido complex [Ir(NH2)(PNP)]PF6 (8) in 70% isolated yield (Scheme 12). In contrast to chloro complex [IrCl(PNP)]PF6 (9), 8 is thermally stable at room temperature. X-ray single crystal diffraction (see below) confirms square-planar coordination of the metal ion, rendering complexes 8 and 9 the only known iridium(III) complexes in this geometry.[46] As reported for 9, the sharp 1H, 13C, and 31P NMR signals of 8 indicate a diamagnetic (S = 0) ground state and a C2 symmetric molecular structure on the NMR time scale. The 1H NMR signal assigned to the NH2 group (δ = 11.35 ppm) exhibits a significant low-field shift compared with other parent iridium amido complexes (−1.3–6.0 ppm).[13i,14b,47] The diamagnetism of 8 and 9 contrasts with related d6 complex [RuCl(PNP)], which adopts an intermediate-spin (S = 1) ground state,[48] attributed to larger ligand field splitting in case of the 5d ions and strong π-donation of the parent amido ligand (8), which destabilizes the LUMO.

In an attempt to synthesize parent iridium(III) imido complex [Ir(NH)(PNP)] (10), 8 was deprotonated with equimolar amounts of strong bases, such as K[N(SiMe3)2] and KOtBu (Scheme 13). Rapid and full conversion of the starting material at room temperature was accompanied by selective formation of iridium(II) amide 6 (1H NMR yield: 66%), the dinitrogen complex 4a (33% including traces of [Ir(N2)(PNP)] (11)), and H[N(SiMe3)2] (Supporting Information, Figure S3).[49] The reaction was monitored at −60 °C by NMR spectroscopy. At this temperature, the 1H NMR spectrum revealed slow consumption of K[N(SiMe3)2] toward H[N(SiMe3)2] (Figure 3). The signals of diamagnetic starting material 8 immediately vanish despite only partial (ca. 50%) conversion of the base at this stage. However, equimolar mixtures of 8 with the final iridium(II) product 6 coalesce into extremely broad peaks at −60 °C due to self-exchange as shown with an original sample (see Supporting Information, Figure S4). Hence, full conversion of 8 cannot be assured despite the immediate absence of its NMR signals in this reaction. Importantly, four new peaks are observed at 1.24, 4.03, 5.42, and 6.27 ppm in about a 36:2:1:2 ratio of intensities, which cannot be assigned to 8 (Figure 3). At −60 °C, this intermediate is rapidly formed in high concentrations and vanishes upon shortly warming the sample to room temperature. Despite some peak broadening, the chemical shifts suggest diamagnetism for this intermediate. Hence, the peak intensities and shapes support the tentative assignment to diamagnetic imido complex 10 as a reactive intermediate, which rapidly decays even at low temperatures.

Scheme 13

Deprotonation of Complex 8 at Room Temperature

Figure 3

1H NMR spectra (d8-THF, −60 °C) of the reaction of 8 with K[N(SiMe3)2] directly after mixing (red line) and after shortly warming to room temperature (blue line), respectively.

Molecular Structures

The molecular structures of 6–8 in the solid state were derived by single crystal X-ray diffraction (Figure 4), confirming square-planar coordination of the Ir ions in all cases. The structural features within the PNP pincer backbones of iridium(II) complexes 6 and 7, such as C–N and C–C bond lengths (Table 1), strongly resemble those of the chloro analogue 5. Similarly, oxidation to iridium(III) exhibits the same trends for 8 and chloro complex 9, i.e. shortening of Ir–N and C–C bonds and lengthening of C–N bonds. This trend indicates increased N to Ir π-donation upon oxidation of iridium(II) (S = 1/2) to iridium(III) (S = 0) as a result of the π*-antibonding character of the SOMO (see Supporting Information). The Ir–NH2 distance of 6 (1.9521(17) Å) compares well with other parent iridium amido complexes with a nitrogen donor ligand in the trans position (1.92–1.95 Å)[29] and is significantly shorter than examples with C-donor ligands trans to amide (2.00–2.19 Å).[14,47,50] The Ir–NH distance decreases within the series 7 (x = 3, n = 2), 6 (x = 2, n = 2), and 8 (x = 2, n = 3) by more than 0.2 Å due to increasingly covalent σ- and π-bonding with the metal ion, resulting in a particularly short bond for 8 (1.900(2) Å). In comparison, the Ir–N triple bond length of [Ir(N)(PNP)]+ (1) was reported to be at 1.678(4) Å.[31]

Figure 4

Molecular structures of 6 (top), 7 (center), and 8 (bottom) from X-ray diffraction with thermal ellipsoids at the 50% probability level (carbon bound hydrogen atoms and one molecule of THF in 7 are omitted for clarity).

Table 1

Selected Bond Lengths and Angles of Amido Complexes 6 and 8 (x = 2) and Ammine Complex 7 (x = 3)

	6	7	8
bond lengths [Å]
Ir–NPNP	2.0194(15)	1.9685(14)	1.926(2)
Ir–NHx	1.9521(17)	2.1100(14)	1.900(2)
N–C	1.380(2)	1.391(2)	1.409(3)
	1.385(2)	1.392(2)	1.416(3)
C–C	1.348(3)	1.344(2)	1.331(4)
	1.344(3)	1.343(2)	1.329(4)
bond angles [deg]
N–Ir–N	179.74(7)	175.68(6)	179.07(10)
P–Ir–P	164.445(17)	165.703(15)	166.42(2)

Electrochemical Examination

In analogy to 5,[51] the cyclic voltammogram (CV) of 6 in THF (Figure 5) exhibits a reversible oxidation wave at E1/2 = −0.68 V (vs Fc/Fc+), which is in agreement with the isolation of oxidation product 8 (see above). Compared with IrII/IrIII chloro complexes 5/9, the potential (E1/2) of the parent amido IrII/IrIII redox couple 6/8 is shifted to lower potential by about 0.7 V, as a result of the increased amide vs chloride donor properties. Remarkably, the IrIV/IrV nitrido [Ir(N)(PNP)]0/+ redox couple (3a/1) was found at a similar potential (E1/2 = −0.86 V in THF, Supporting Information Figure S6) to that of 6/8. This observation is in agreement with the high degree of covalency within Ir–NH (x = 0–2) bonding, as previously inferred for nitride 3a from the EPR and ENDOR spectroscopic features.[31] Unlike for iridium(II) chloride 5, electrochemical reduction of amide 6 in THF was found to be irreversible at all scan rates.

Figure 5

Top: CV of 6 (10–3 mol/L) in THF (400 mV/s, [nBu4N]PF6 0.1 mol/L, Pt working, counter, and pseudo-reference electrodes). Bottom: CV of 6 (10–4 mol L–1) in the presence of K[N(SiMe3)2] (10–3 mol L–1) in THF (800 mV s–1, 0.1 mol L–1 (nBu4N)PF6, Pt working electrode and counter electrodes, Ag/Ag+ reference electrode, potentials vs. FeCp2/FeCp2+).

In contrast to the neutral conditions, the cyclic voltammogram of 6 in the presence of K[N(SiMe3)2] (10 equiv) reveals increasingly irreversible oxidation at a high relative base concentration (Figures 5 and S5).[52] Furthermore, an additional reduction wave at E1/2 = −0.86 V is observed upon cycling back after oxidation of 6, which seems reversible at high scan rates. On the basis of the peak potential, this redox event is assigned to the [Ir(PNP)(N)]0/+ (3a/1) redox couple (see above). Unfortunately, quantification of the currents that arise from the 6/8 and 3a/1 redox couples, respectively, is obscured by strong peak overlap, hampering quantitative kinetic analysis.

Reactivity of [Ir(NH)(PNP)] toward HAA

HAA within the [Ir(NH)(PNP)] (x = 2–0) redox series was examined upon reaction of 6 (x = 2) with hydrogen atom acceptors and of 3a (x = 0) with hydrogen atom donors, respectively (Scheme 14).[53] No reaction is observed upon the addition of TEMPO (2 eq.; TEMPO = 1-hydroxy-2,2,6,6-tetramethylpiperidine-1-yloxidanyl) to a solution of 6 in d6-benzene (Scheme 14, reaction A; Supporting Information Figure S8). In reverse, the addition of 10 equiv of TEMPO-H (BDEO–H(benzene) = 70 kcal mol–1)[54] to nitridyl complex 3a, generated in situ by azide photolysis at −60 °C in d8-toluene, leads to immediate formation of amide 6 (Scheme 14, reaction B; Supporting Information Figure S9).

Scheme 14

Reactions of Amide 6 and Nitride 3a with HAT Reagents

An upper limit for the N–H BDEs was estimated upon reaction with 2,4,6-tri-tert-butylphenol (TBP-H; BDEO–H(benzene) = 82 kcal mol–1)[54] and the 2,4,6-tri-tert-butylphenoxy radical (TBP), respectively. Photochemically generated 3a (−60 °C, d8-toluene) does not react with TBP–H (Scheme 14, reaction C; Supporting Information Figure S10). In contrast, 6 is immediately consumed in the presence of 2 equiv of TBP (reaction D, Supporting Information Figure S11), yet not as clean. Besides 3a, considerable amounts of dinitrogen complex 4a are also observed. The mechanism for the formation of 4a remains unclear, as binuclear nitridyl coupling of 3a is slower at −60 °C than the here observed formation of 4a.[31] However, the generation of 3a suggests that the absence of HAA from TBP–H by 3a (Scheme 14, reaction C) can be attributed to the overall reaction thermochemistry rather than kinetics. Importantly, the reaction of 6 with only 1 equiv of TBP results in a mixture of approximately equimolar amounts of 3a and 6 besides 4a (Scheme 14, reaction E; Supporting Information Figure S12).

Assuming that imido complex [Ir(NH)(PNP)] (10) is an intermediate in the HAT reactions with amide 6 and nitride 3a, and that HAT is rapid under the experimental conditions, the following thermochemical estimates are made from the H-transfer experiments described above (Scheme 15):

Scheme 15

HAT Reactions from Amide 6 to Nitride 3a via Proposed Imide Intermediate 10

Computational Examination

The N–H BDEs were also computationally evaluated with density functional theory (DFT). As previously reported for nitride 3a,[31] the computed EPR features and molecular structure of amide 6 (S =1/2) are in excellent agreement with experimental results. However, assigning an electronic ground state structure and energy to postulated imido species 10 is much less straightforward. DFT geometry optimizations of 10 in triplet (T), open shell singlet (OSS), and closed shell singlet (CSS) states place all three configurations within around 1 kcal mol–1 (see Supporting Information). The OSS configuration turned out to be the most stable, which would in fact be in agreement with the tentative assignment of NMR signals to 10 (see above). In the triplet state, the Ir–N–H angle (142.0°) is considerably closer to linearity as compared with the CSS (114.2°) and OSS (117.1°) states, respectively, resulting in two energetically almost degenerate SOMOs with strong d(Ir)–p(N) π* character for the triplet (see Supporting Information). Ir–N–H bending reduces this interaction for one of the two orbitals and stabilization of the singlet states. However, in all cases Ir–NH bonding is comprised of six electrons in orbitals with σ- (2e–) and π-type (4e–) symmetry and two electrons in π* orbitals with strongly covalent bonding to the metal ion resulting in low-lying open-shell states (see Supporting Information for further discussion).

Applying doublet (6 and 3a) and OSS (10) ground state configurations, respectively, the resulting N–H BDENH2 (eq 1) and BDENH (eq 2) values (see Supporting Information for details) confirm the experimentally estimated limits from reactions A–D. As a result, disproportionation of the imido species to the amide and nitride is exothermic by ΔHDiss ≅ −24 kcal mol–1 (ΔGDiss ≅ −22 kcal mol–1). Kinetic barriers for bimolecular HAT were also calculated using the PMe2-truncated models 6, 3a, and 10. Low transition state barriers at both the triplet (ΔG‡ = +8.4 kcal mol–1) and the open-shell singlet (ΔG‡ = +12.9 kcal mol–1) potential energy surfaces (see Supporting Information) were obtained. These results are in line with reaction E and the transient nature of proposed intermediate 10.

We previously reported that N–N coupling of iridium(IV) nitrido species 3a to 4a is exclusively observed, even in the presence of excess HAT reagent 1,4-cyclohexadiene.[31] Therefore, the hypothetical HAT reaction with 1,4-cyclohexadiene was also evaluated computationally. Overall, double HAT of 1,4-cyclohexadiene with nitride 3a to amide 6 and benzene is strongly exergonic by −63.7 kcal mol–1 (Scheme 16). However, imido complex 10 and the cyclohexadienyl radical define unfavorable intermediates, and the slightly endergonic first HAT reaction (ΔG0 = +1.5 kcal mol–1) is accompanied by a large kinetic barrier (ΔG‡ = 24.4 kcal mol–1). In comparison, bimolecular nitride coupling of 3a to 4a, which is strongly downhill by ΔG0 = −108.0 kcal mol–1 (−54.0 kcal mol–1p. 3a), was computed to proceed via a much lower transition state barrier (ΔG‡ = 13.7 kcal mol–1).[31]

Scheme 16

DFT Computed Free Energies (ΔG298K0) and Kinetic Barriers (ΔG298K‡) for HAT Reactions from 1,4-Cyclohexadiene to 3a and 10 and for Nitride Coupling of 3a, Respectively (Energies in kcal mol–1)

In contrast to the first hydrogen transfer, the second HAT step from the cyclohexadienyl radical to 10 (Scheme 16) exhibits a much smaller barrier (ΔG‡ = 9.2 kcal mol–1), as expected qualitatively from the large driving force (ΔΔG0 = −65.2 kcal mol–1). Similarly, HAT of the reactive imido complex with parent cyclohexadiene is exergonic (ΔΔG0 = −21.2 kcal mol–1) with a moderate barrier (ΔG‡ = +16.7 kcal mol–1).

Discussion

We recently presented square-planar complexes in a wide range of formal oxidation states, such as [IrICl(PNP)]−, [IrIICl(PNP)] (5), [IrIIICl(PNP)]+ (9), [IrIVN(PNP)] (3a), and [IrVN(PNP)]+ (1).[32,44,51] In the present study, parent iridium(II) amide and ammine complexes 6 and 7 and iridium(III) amide complex 8 were isolated and fully characterized. These compounds represent intermediates within a synthetic cycle of ammonia oxidation to dinitrogen (Scheme 17).

Scheme 17

Synthetic Cycle of NH3 Oxidation with the Ir(PNP) Platform (Reaction with Dashed Arrow Not Demonstrated in This Work)

Once again the isolobal relationship with octahedral group 8 complexes shall be stressed (Figure 6). This analogy emphasizes the conservation of a square-planar geometry around iridium, which relies on steric shielding through the bulky pincer ligand. In several respects, the square-planar iridium compounds resemble the reactivity of octahedral polypyridyl Os–NH complexes, which undergo a fascinating redox chemistry.[55] For example, the octahedral nitrides [OsVI(N)Cl2(L)]+ (L = tris(1-pyrazolyl)methane (tpm); 2,2′:6,2″-terpyridine (tpy); isolobal with 1) also undergo nitride coupling to the N2 bridged OsIII/OsIII complexes upon one-electron reduction.[35e] Furthermore, the reversible interconversion of the OsVI nitrides and OsIII ammine complexes, [OsIII(NH3)Cl2(L)]+ (isolobal with 7), was studied electrochemically in depth by Meyer and co-workers.[56] The transient osmium(V) nitride stage (isolobal with 3a) was not directly observed electrochemically. However, it could be trapped upon use of an anionic tridentate ligand: [(Tp)OsVI(N)Cl2] (Tp = hydridotris(pyrazol-1-yl)borate) reduction in strongly acidic solution afforded the isolation of parent OsV imide [(Tp)OsV(NH)Cl2].[57] In neutral aqueous solution, the electrochemical studies of the tpy and tpm systems show that the redox couple OsV(N)/OsIV(NH2) (isolobal with 8) provides a kinetic bottleneck for ammine/nitride interconversion as it is associated with multiple proton transfer steps. The parent OsIV(NH) imide (isolobal with 10) is a likely intermediate, but the exact proton compositions of electrochemically detected OsIV(NH) species remain unclear. Notably, the spectroscopic identification of intermediates on the OsV and OsIV redox stages is hampered due to their inherent tendency to disproportionate, e.g.[56,55]

Figure 6

Schematic d-orbital splitting diagrams illustrating the isolobal relationship of octahedral d2 (OsVI) and square-planar d4 (IrV) nitrido complexes.

In the current work, we particularly focused on the putative, parent IrIII imido intermediate 10, owing to its possible relation to reactive imides that show nitrene transfer reactivity. Well characterized terminal, parent TM imides are not very frequent and for the most part limited to high-valent, early TMs (group 4–6).[42b,58] Notably, Basolo and co-workers postulated a transient, electrophilic iridium parent imido, [Ir(NH)(NH3)5]3+, as an intermediate in the decomposition of the respective azide [IrN3(NH3)5]2+, under acidic conditions.[59] All efforts to isolate 10 upon deprotonation of 8, HAA from 6, or hydrogen transfer to 3a were unsuccessful. NMR monitoring of the first reaction suggests detection of diamagnetic 10 as a reactive intermediate at low temperatures. However, at room temperature, spectroscopic and electrochemical examination indicates disproportionation of 10 into 3a and 6 (Scheme 17), which is succeeded by nitride coupling, again in analogy with the osmium system. The HAT experiments of the nitride 3a and amide 6 with TEMPO/TEMPO-H and TBP/TBP-H, respectively, confirm the disproportionation (reaction E) and allow for an estimate of boundaries for the amide and imide N–H BDEs. These are in agreement with the DFT computed values of 95 kcal mol–1 (6) and 71 kcal mol–1 (10), respectively. Also, a relatively high barrier for HAT from model substrate 1,4-cyclohexadiene to iridium(IV) nitride 3a was computed, which cannot compete with decay upon bimolecular nitride coupling. Hence, the absence of reactivity of nitride 3a with hydrocarbon HAT reagents and high selectivity toward nitride coupling is attributed to both thermodynamics and kinetics. Imide 10 seems more suitable for radical C–H activation due to the formation of a strong N–H bond. However, this reaction will compete with imide disproportionation proceeding at low barriers (10: ΔG‡ = +8.4 kcal mol–1). Hence, clean C–H activation and NH-group transfer will be a difficult task for kinetic reasons.

These results can be rationalized with qualitative frontier MO considerations within the IrNH redox series. The SOMO of the amido species 6 (Figure 7 left, π*) is mainly metal centered (d) with some Ir–Namido π-antibonding contribution. HAA and formation of imide 10 results in overlap of another nitrogen lone-pair with an Ir d orbital (d), which is a subject of the Ir–N–H bond angle. The high degree of covalency within the two Ir–N orbital interactions with π symmetry produces two high lying orbitals (Figure 7 center, π1* and π2*), which are occupied with overall two electrons in the three accessible states (CSS, OSS, T). In comparison, the two Ir–N MOs with π*-character in nitride 3a (Figure 7 right, π1* and π2*) are only occupied with overall one electron. Hence, disproportionation of the IrIII-imide by intermolecular HAT to the IrII-amide and IrIV-nitride results in stabilization upon both reduction of covalent Ir–N interactions (6) and reduction of the population of antibonding orbitals (3a), respectively.

Figure 7

Qualitative representation of the frontier orbital interactions within amide 6, imide 10 (only triplet state is shown), and nitride 3a.

Our results indicate that, for such square-planar group 9 MII–NH2/MIII=NH/MIV≡N compounds, the d6 imido species should be best suited for E–H activation via HAT. The same arguments should qualitatively also apply to isolobal platforms. However, prevention of parent imide disproportionation will be a challenging task, and application of primary amines as a nitrogen source is advised.

Experimental Section

All experiments were carried out using Schlenk and glovebox (argon atmosphere) techniques. All solvents were dried by passing through columns packed with activated alumina. Deuterated solvents were obtained from Euriso-Top GmbH, dried over Na/K (d6-benzene, d8-THF, and d8-toluene) or CaH2 (CD2Cl2), respectively, distilled by trap-to-trap transfer in vacuo and degassed by three freeze–pump–thaw cycles. NaNH2 (Acros), AgPF6 (ABCR), AgSbF6 (ABCR), TEMPO (Acros) and TBP–H (Acros) were used as purchased. 3a,[44] [Ir(N3)(PNP)],[31] TEMPO–H,[60] [H(Et2O)2][BAr4F],[61] and TBP[62] were prepared according to published procedures. Irradiation was carried out using a LOT-Quantum design 150 W xenon short-arc lamp. Elemental analyses were obtained from the analytical laboratories at the Georg-August University on an Elementar Vario EL 3. NMR spectra were recorded on Bruker Avance III 300 or 400 MHz spectrometers and were calibrated to the residual solvent proton resonance (d6-benzene: δH = 7.16 ppm, δC = 128.39 ppm; d8-THF: δH = 3.58 ppm, δC = 67.2 ppm; d8-toluene: δH = 2.09 ppm, δC = 20.4 ppm; CD2Cl2: δH = 5.32 ppm, δC = 53.84 ppm). 31P chemical shifts are reported relative to external phosphoric acid. Signal multiplicities are abbreviated as s (singlet), d (dublet), t (triplet), q (quartet), m (multiplet), and br (broad). Magnetic moments were determined in benzene at room temperature by Evans’ method as modified by Sur and corrected for diamagnetic contribution.[63]

[Ir(NH2)(PNP)] (6)

15-Crown-5 (43 μL; 0.21 mmol; 1 equiv) is added to a mixture of 5 (126 mg; 0.21 mmol; 1 equiv) and NaNH2 (90.0 mg; 2.15 mmol; 10 equiv) in THF (10 mL, freshly dried over Na/K) and stirred for 3.5 h. All volatiles are removed in vacuo, the residue is extracted with benzene (3 × 7 mL), and the solvent is evaporated to dryness. After extraction of the crude product with pentanes (8 × 5 mL) and filtration over Celite, the solvent is removed, and the product is lyophilized overnight out of benzene (15 mL). 6 is obtained as green crude product and purified by sublimation (yield: 55%). Anal. Calcd for C20H42N2P2Ir (564.73): C, 42.54; H, 7.50; N, 4.96. Found: C, 42.87; H, 7.41; N, 4.73. NMR (d6-benzene): 1H (300 MHz, 20 °C): δ 5.96 (br, CH3), −53.88 (br), −73.11 (br). μeff297 K = 1.82 μB.

[Ir(NH3)(PNP)][B(ArF)4] (7)

[H(OEt2)2][B(ArF)4] (47.0 mg; 65.9 μmol; 1.24 equiv) and 6 (30.0 mg; 53.1 μmol; 1 equiv) are dissolved in d8-THF (0.5 mL) and shaken for 5 min. The product is precipitated with pentanes (2 mL), collected by filtration and washed with pentanes (3 × 2 mL). 7 is isolated as a blue solid (30.9 mg; 21.6 μmol; 41%). Anal. Calcd for C52H55BF24N2P2Ir (1428.96): C, 43.71; H, 3.88; N, 1.96. Found: C, 43.91; H, 3.82; N, 1.61. NMR (d8-THF, 20 °C): δ 38.3 (br, 2H), 13.5 (br, 36H, CH3), 7.77 (s, 8H, o-C6H3(CF3)2), 7.52 (s, 4H, p-C6H3(CF3)2).

[Ir(NH2)(PNP)]PF6 (8)

A mixture of 6 (50.0 mg; 88.5 μmol) and AgPF6 (22.4 mg; 88.5 μmol) is dissolved in THF (20 mL) and stirred for 5 min. After removal of the solvent, the residue is washed with benzene (4 × 5 mL), and the crude product is extracted with THF (8 × 5 mL). The solution is concentrated, layered with pentanes (40 mL), and crystallized at −32 °C for 16 h. Deep purple crystalline 8 is filtered off, washed with pentanes (4 × 5 mL), and dried in vacuo (50.3 mg; 63.0 μmol; 71%). Anal. Calcd for C20H42N2F6P2Ir (709.70): C, 33.85; H, 5.97; N, 3.95. Found: C, 33.99; H, 5.78; N, 3.48. NMR (CD2Cl2, 20 °C), 1H NMR: 11.35 (br, 2H, NH2), 6.35 (ABXX′B′A′, N = |2JHP + 4JHP| = 4.7 Hz, 3JHH = 6.3 Hz, 2H, NCHCHP), 6.11 (ABXX′B′A′, N = |3JHP + 5JHP| = 17.7 Hz, 3JHH = 6.3 Hz, 2H, NCHCHP), 1.66 (A18XX′A′18, N = |3JHP + 5JHP| = 7.5 Hz, 36H, CH3). 13C (75.47 MHz): δ 169.6 (AXX′A′, N = |2JCP + 3JCP| = 6.0 Hz, NCHCHP), 121.2 (AXX′A′, N = |1JCP + 3JCP| = 18.4 Hz, NCHCHP), 36.5 (A2XX′A′2, N = |1JCP + 3JCP| = 11.7 Hz, PCCH3), 31.4 (A6XX′A′6, N = |2JCP + 4JCP| = 2.7 Hz, PCCH3). 31P (121.49 MHz): δ 48.2 (s, PtBu23), −145 (hept, 1JPF = 710.5 Hz, PF6).