Configuring Bonds between First-Row Transition Metals - Accounts of

Article pubs.acs.org/accounts

Configuring Bonds between First-Row Transition Metals Reed J. Eisenhart, Laura J. Clouston, and Connie C. Lu* Department of Chemistry and Center for Metals in Biocatalysis, University Of Minnesota, 207 Pleasant Street SE, Minneapolis, Minnesota 55455, United States S Supporting Information *

CONSPECTUS: Alfred Werner, who pioneered the field of coordination chemistry, envisioned coordination complexes as a single, transition metal atom at the epicenter of a vast ligand space. The idea that the locus of a coordination complex could be shared by multiple metals held together with covalent bonds would eventually lead to the discovery of the quadruple and quintuple bond, which have no analogues outside of the transition metal block. Metal−metal bonding can be classified into homometallic and heterometallic groups. Although the former is dominant, the latter is arguably more intriguing because of the inherently larger chemical space in which metal−metal bonding can be explored. In 2013, Lu and Thomas independently reported the isolation of heterometallic multiple bonds with exclusively first-row transition metals. Structural and theoretical data supported triply bonded Fe−Cr and Fe−V cores. This Account describes our continued efforts to configure bonds between first-row transition metals from titanium to copper. Double-decker ligands, or binucleating platforms that brace two transition metals in proximity, have enabled the modular synthesis of diverse metal−metal complexes. The resulting complexes are also ideal for investigating the effects of an “ancillary” metal on the properties and reactivities of an “active” metal center. A total of 38 bimetallic complexes have been compiled comprising 18 unique metal−metal pairings. Twenty-one of these bimetallics are strictly isostructural, allowing for a systematic comparison of metal−metal bonding. The nature of the chemical bond between first-row metals is remarkably variable and depends on two primary factors: the total d-electron count, and the metals’ relative d-orbital energies. Showcasing the range of covalent bonding are a quintuply bonded (d-d)10 Mn−Cr heterobimetallic and the singly bonded late−late pairings, e.g., Fe−Co, which adopt unusually high spin states. A long-term goal is to rationally tailor the properties and reactivities of the bimetallic complexes. In some cases, synergistic redox and magnetic properties were found that are different from the expected sum of the individual metals. Intermetal charge transfer was shown in a Co−M series, for M = Mn to Cu, where the transition energy decreases as M is varied across the first-row period. The potential of using metal−metal complexes for multielectron reduction of small-molecules is addressed by N2 binding studies and a mechanistic study of a dicobalt catalyst in reductive silylation of N2 to N(SiMe3)3. Finally, metal-ion exchange reactions with metal−metal complexes can be selective under appropriate reaction conditions, providing an alternative synthetic route to metal−metal species.

1. INTRODUCTION

First-row metals are particularly attractive because they adopt a wide range of spin states and because their natural abundance is relevant to sustainability. Prior to 2013, there was no example of a coordination complex with multiple bonds between different first-row transition metals. Such species had only been observed spectroscopically as naked binuclear clusters in gasphase or matrix isolation studies.8 The poor spatial overlap of 3d orbitals was believed to impede strong covalent bonding between different first-row metals. Challenging this notion, two independent reports demonstrated that Fe−Cr and Fe−V triple bonds could be stabilized with auxiliary ligands.9,10 This Account describes our recent progress in making metal−metal bonds with first-row transition metals from Ti to Cu. A signature of the work is the use of double-decker ligands. These unique platforms have provided access to diverse

Fifty years have passed since the delta bond was recognized in the structure of [Re2Cl8]2−.1 Today, the field of metal−metal bonding continues to challenge our notions of chemical bonding, electronic structure, and reactivity paradigms.2,3 The discovery of a quintuply bonded dichromium complex revived efforts to make and understand metal−metal bonds across the periodic table from the transition metal block to the lanthanides and actinides.4,5 This is an area ripe with possibilities in part because vast numbers of different metal−metal pairings remain largely unexplored. Such explorations are motivated to uncover emergent properties that are remarkably different from the properties of the individual metals or to develop new modes of cooperative reactivity.6,7 To rationally wield these properties or tailor multimetallic active sites, we seek to understand the underlying principles that govern metal−metal bonding and dictate its electronic and chemical behavior. © XXXX American Chemical Society

Received: July 20, 2015

A

DOI: 10.1021/acs.accounts.5b00336 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

Figure 1. Modular synthesis of metal−metal complexes using double-decker ligands (L1−L3) in a two-step metalation.

bimetallics of exclusively first-row metals. To understand the underlying principles of metal−metal bonding and its attendant electronic structure, properties, and reactivities, we take the approach of comparing isostructural bimetallics, where each metal is systematically varied.

The double-decker ligands have distinct binding sites for selective installation of different metals. Many heterobimetallics were prepared by a two-step metalation, where initial metalation of the tri(amido)amine pocket provides a monometallic complex that then acts as a metalloligand in a subsequent metalation. Using this two-step procedure, we were able to garner diverse metal−metal pairings (Figure 1).9,12,13,18−20 Despite the general success of this approach, there are specific limitations, which are rationalized according to the ligand’s ability to delocalize charge. The mixed amide-phosphine ligand L1 is effective for stabilizing monometallic Ti, V, and Cr complexes.9,18,21 Presumably, the trianionic amide pocket is ideal for stable trivalent metals. The upper phosphine pocket is suited to late metals, and low-valent Fe, Co, or Ni often occupies this site. As a partner, chromium is versatile and also adept at multiple bonding.2,4,22,23A dozen (CrM)+n complexes were isolated, where M = Cr to Ni and n = 2−4. However, divalent metals in the L1 amide pocket tend to be unstable. Attempts to singly metalate with FeX2 gave instead the corresponding homobimetallic (Fe2)+4 halide (Figure 2).24 The lack of late-metal metalloligands has thwarted efforts to isolate late−late heterobimetallics within this scaffold. An exception, L1FeCoCl, was obtained from a metal-swapping reaction of L1Fe2Cl and CoCl2(THF)1.5.24,25 The L2 and L3 platforms have successfully been used for assembling bimetallics of the mid-to-late first-row metals Mn to Cu. In contrast to L1, monometallic Fe(II) and Co(II) complexes were isolated with L2 and L3. Bimetallics containing an early metal are notably absent for L2 and L3, even though V(III) and Cr(III) complexes were synthesized. The V and Cr metalloligands were unreactive to further metal uptake, presumably because, in the quasi-octahedral coordination

2. MODULAR SYNTHESIS OF METAL−METAL COMPOUNDS Double-decker ligands, or binucleating ligands with two decks of donor atoms shown in Figure 1, brace two metal atoms in proximity.11−13 The ligands favor metal−metal interactions by forming 5-membered chelate rings comprising two metal centers and a 3-atom buttress. The ligands are tripodal and trianionic, where an apical amine connects the three ligand arms. The ligands can be made on mutigram scale by simple elaboration of the tren backbone, N(C2H4NH2)3, or the aryl variant, N(o-NH2C6H4)3. Donor types differ across the ligands, but perhaps more importantly, so does the extent of charge delocalization. In L1, charge is localized to the amide pocket, whereas charge is delocalized in the amidinate groups of L3. The pyridylamide ligand L2 is somewhere in the middle. The value of using double-decker ligands is debatable, as many bimetallics are generated with much simpler ligands, e.g., amido-phosphines, [RN−PR′2]−.14−16 The latter are known to slip from bridging two metals (κ1N,κ1P) to one metal (κ2NP), creating open coordination site(s) at either metal center for reactivity.17 In contrast, the multidenticity of L1−L3 constrains the coordination environment, locking the number and type of donors at each metal-binding site. The only obvious open coordination site is the apical pocket at the upper metal. One advantage is that the bimetallics obtained with L1−L3 are largely isostructural, enabling systematic comparisons to be made. B


Article


structural validation. The ability of these species to transfer multiple electrons has been probed by isolating different redox states of the bimetallic core, particularly (MM′)+2,+3,+4.

3. METAL−METAL BONDING Molecular metal−metal bonding continues to be primarily characterized by X-ray crystallography, where the intermetal distance defines the metal−metal interaction. Although an inverse relationship between intermetal distances and metal− metal bond order is qualitatively understood, formulating a quantitative relationship has been impossible because of complex electronic configurations, subtlety of delta bonds, and sensitivity of intermetal distances to external factors, such as molecular charge, ligand differences, and crystal packing forces.2 Some of these complications are obviated by comparing isostructural complexes with the same ligand. Hence, the diverse bimetallics obtained with L1 make an ideal collection for understanding how metal−metal bonding varies across the firstrow period. A counterargument is that bridging ligands intrinsically enforce short metal−metal contacts. In the case of L1, the scaffold is flexible and accommodates a wide range of metal− metal distances (Figure 4). Considering just the L1CrM

Figure 2. Reactions that exemplify limitations in the two-step metalation approach.

geometry at V and Cr, all six N-atoms are tightly bound to the metal (Figures S1−2).26 A subtle limitation is the lack of heterobimetallic isomers, where only the metal sites are swapped. As an example, L2CoFeCl was formed cleanly from [L2Co]K and FeCl2. Attempts to prepare the L2FeCoCl isomer from [L2Fe]K and CoCl2 instead produced L2CoFeCl and L2Co2Cl. For labile first-row metal ions, trapping a “kinetic” heterobimetallic species appears untenable. Figure 3 sums our progress in configuring bonds across the first-row transition metal period. Half of the possible metal− metal combinations from Ti to Cu have been isolated with Figure 4. Overlay of solid-state structures of isostructural L1CrM bimetallics shown from side (a) and top (b) views. Methyl and H atoms were omitted for clarity. The bimetallics are [L1Cr2]K (red), L1CrMn (orange), L1CrFe (green), L1CrCo (cyan), and L1CrNi (blue).

complexes, the intermetal contact can differ by nearly 0.7 Å from 1.74 Å in [L1Cr2]K to 2.41 Å in L1CrNi. The Cr−Namine distance changes by 0.4 Å despite the constancy of the L1Cr metalloligand. On the basis of these stark structural changes, any inherent bias of the ligand pales to the metal−metal interaction in determining the metal−metal bond distance. Interpreting heterometallic bonding is vexing because absolute intermetal distances cannot be neatly compared. To normalize metal−metal bonding across different metals, Cotton introduced the formal shortness ratio (FSR), a quotient of the intermetal distance to the expected single-bond length.2 An underlying issue is how one determines the “expected” singlebond length between any two metals. Typically, one sums the metals’ single-bond radii, a tabulated value that depends on the metallic radii and coordination number in the lattice.27 Short intermetal distances may indicate multiple bonding and correspond to FSR values below unity. At the other extreme, long intermetal distances with FSR values above unity would be considered weak or noninteracting. The bonding interpretations were guided by experimental FSR values and

Figure 3. Array of first-row transition metal−metal complexes prepared with the double-decker ligands L1−L3. Homometallic (M2)+n and heterometallic (MM′)+n pairings are indicated by color (blue and yellow, respectively) with the redox state(s) of the bimetallic core, +n. For a complete and detailed list, see Table S1. C


Article

Accounts of Chemical Research theoretical calculations. Although FSR values appear distinctive for the number of σ- and π-bonds, they poorly reflect the presence of delta bonds. The poor spatial overlap of δsymmetric d-orbitals render delta bonds weak and of minor impact to the intermetal distance. Hence, for quadruply or higher bonded bimetallics, theory is critical to gain a precise bonding picture. The M−M stretching frequencies are arguably a better experimental readout of metal−metal bond multiplicity, but such measurements can be challenging to obtain. Bonding Trends

To extract trends in metal−metal bonding, we focus on the strictly isostructural bimetallics, [L1MM′]+,0,−, which comprise twenty-one unique compounds. The shortness of the metal− metal bond (as indicated by low FSR values, Figure 5) depends

Figure 6. Corresponding spin state map of 20 isostructural bimetallics.

The remaining group comprises Ni−M bimetallics with high bond polarity (ΔN = 4−6) and intermediate d-electron counts of 11−13. The FSR values are generally close to unity (0.98− 1.05). Because of the large Ni−M bond polarity, where M is Ti to Cr, the d-electrons are primarily localized at Ni, and the interaction is described as a Ni(0) → M(III) dative bond. The overall spin states, which are low to intermediate spin, simply reflect the localized spin of the M(III) center, e.g., S = 1/2 Ti(III), S = 1 V(III), and S = 3/2 Cr(III). Understanding Bonding through Theory

Complementary theoretical calculations by Gagliardi and coworkers have heightened our understanding of metal−metal bonding in the double-decker systems. Figure 7 shows the evolution of the molecular orbital diagrams across different metal−metal bonding regimes. The bond orders shown are based on the main electronic configuration, which is defined as the formal bond order (FBO). The corresponding effective bond order (EBO), which considers multiconfigurationality, is typically lower by 0.7 on average in the multiply bonded bimetallics (Table S3). The weaker ligand field associated with trigonal symmetry permits all five d-orbitals to engage in metal−metal bonding by forming σ, 2π, and 2δ molecular orbitals. Hence, 10 is a magical d-count for maximal d-bonding, and quintuple bonds are predicted for (d-d)10 [L1Cr2]− and L1CrMn. The corresponding EBO values of 3.99 and 3.85, respectively, are on par with those calculated for quintuply bonded dichromium complexes.30 Although quintuple bonding requires a (d-d)10 configuration, the converse is false. For instance, [L1CrFe]+, which has 10 d-electrons, exhibits a different electronic configuration of σ2π4(Fe dxy, dx2−y2)4. Presumably, as ΔN increases, the energy overlap between the two metals δsymmetric d-orbitals worsens, resulting in nonbonding, localized d-orbitals. Even larger ΔN results in localization of the π-symmetric, and eventually σ-symmetric, d-orbitals until all d-electrons are localized at individual metals. At this extreme, the metal−metal interaction, if any, would be dative in nature, which is characteristic of the Ni−Cr, Ni−V, and Ni−Ti bimetallics. Delta bonding is highly sensitive to electron count. In (d-d)9 L1Cr2, the intuitive formal bond order based on the canonical manifold (σ + 2π + 2δ) is 4.5. The predicted FBO is lower at 4 because of localization of one δ-symmetric d-orbital. By contrast, π-bonding is more robust. Triply bonded bimetallics

Figure 5. Formal shortness ratio (FSR) map of 21 isostructural bimetallics with the L1 platform. FSR values are plotted according to the total d-electron count and the difference in the two metals’ group numbers (ΔN).

on a combination of two factors: (1) the total d-electron count and (2) the polarity of the metal−metal interaction, which is approximated by the difference in the two metal’s group numbers, or ΔN. Metal−metal bond multiplicity (FSR < 0.97) is observed within a small range of d-electrons (9−12), whereas bond polarity is better tolerated (ΔN ≤ 5). On the basis of the complementary theoretical calculations, FSR values near 0.9 correspond to a double metal−metal bond (σ + π), whereas values close to 0.8 correspond to a triple bond (σ + 2π). Formal quintuple bonding (σ + 2π + 2δ) was predicted for [L1Cr2]− and L1CrMn, which have the lowest FSR values of 0.74 and 0.78, respectively, in this series.9,28 Multiply bonded bimetallics tend to be low-spin, S = 0 or S = 1/2 (Figure 6). The one exception, [L1CrFe]−, is S = 1, but this complex is also considered low spin, as the unpaired electrons populate degenerate d-orbitals. In addition to multiply bonded bimetallics, significant metal− metal interactions are observed in two other distinct groups. The first set is a trio of late−late metal pairings, L1Fe2, L1Co2, and L1FeCo, where the metal elements are identical or similar (ΔN = 0, 1) and d-electron counts are high from 13−15.24,29 The FSR values (0.97−1.00) approximate a single metal−metal covalent bond. Notably, this trio adopts high-spin configurations, S = 5/2−7/2. Also befitting this group, L3Co2 and L3CoFe, which have FSR values of 0.99 and 0.94, respectively, are similarly high spin.13 D


Article


Figure 7. Qualitative d-orbital manifolds of different metal−metal bonding regimes. The coloring of the energy levels and electrons denotes whether they are primarily delocalized (black) or localized (red and blue).

Table 1. Diamagnetic Anisotropies (Δχ) for (d-d)10 Multiply Bonded Bimetallics

are the majority of multiply bonded heterobimetallics and tolerate a range of d-counts and ΔN (Figure 7). A slight increase in the number of d-electrons and/or ΔN, however, can cost a bond order to give a formal metal−metal double bond. In the single covalent regime, the MO diagram estimates a slightly higher FBO of 1.5. Bonding between late first-row metals is uncommon as well as possessing a high-spin configuration. The concurrence of metal−metal bonding and a high-spin configuration seems counterintuitive. Electron unpairing cannot result from poor energy overlap as these complexes are homobimetallic, or a close analogue, e.g., L1FeCo. A high-spin phenomenon has been described in other metal−metal systems, and there is no link to geometry, coordination number, or ligand.16,31−33 However, many are homobimetallic, and all have high d-electron counts of 13−16. Presumably, as d-electrons increase well above 10, electron− electron repulsion becomes a significant problem. Hence, electron unpairing alleviates electronic repulsion while providing favorable exchange interactions.

complex 1

−

[L Cr2] L1CrMn [L1CrFe]+ L1VFe [L1VCo]+ L1TiCo

FSR

ΔN

Δχ (10−36 m3 molecule−1)

0.74 0.78 0.84 0.79 0.84 0.89

0 1 2 3 4 5

−3500 −3900 −5800 −4700 −4700 −3400

Spectroscopic Evidence of Multiple Bonding

Nearly all the multiply bonded Fe-bimetallics exhibit large quadrupole splittings (ΔEQ = 4−6 mm/s) by 57Fe Mössbauer spectroscopy (Table S4). Large ΔEQ values are rare, but they are common for iron nitride complexes with a highly covalent FeN triple bond.36 By analogy, a triply bonded FeM core could give rise to the large ΔEQ values.9 There are a few exceptions, such as L1TiFe,37 V(iPrNPPh2)3FeBr and V(iPrNPPh2)3Fe(PMe3),10 which show normal ΔEQ. Hence, further studies are needed to understand this spectroscopic phenomenon.

Besides short intermetal distances, multiply bonded bimetallics should exhibit unique spectroscopic signatures. Delta bonding electrons should manifest a δ → δ* transition, which can indirectly probe the M−M bond strength.34 Although δ2→ δδ* transitions of quadruply bonded complexes are well-known, reports of δ4→ δ3δ* transitions in quintuply bonded species are lacking. In [L1Cr2]− and L1CrMn, the lowest energy transitions at 675 and 1025 nm, respectively, were assigned as δ4→ δ3δ* transitions.28 A complementary resonance Raman study of [L1Cr2]− showed a band at 434 cm−1 that was identified by theory as a Cr−Cr bond vibration. Metal−metal multiple bonding can significantly perturb 1H NMR chemical shifts of proximal protons.35 The methylene protons in L1 shift downfield by as much as 3.5 ppm in the (dd)10 bimetallics. The diamagnetic anisotropies (Δχ) in Table 1 are large, implicating multiple bonding. Unfortunately, Δχ does not correlate with bond order or FSR and appears insensitive to delta bonding. Presumably, Δχ relates to both FSR and bond polarity (ΔN) because low FSR values (≤0.84 to maximize πbonding) and bond polarization can enlarge diamagnetic anisotropy.

4. IN SEARCH OF SYNERGISTIC REDOX AND MAGNETIC PROPERTIES By cyclic voltammetry, a few metal−metal pairings exhibit rich redox profiles with 3 or 4 reversible one-electron transfers (Figure 8). For instance, L1CrNi has three reversible events: two oxidations, and one reduction. Comparing L1CrNi to its individual components was possible because both monometallic L1Cr and (H3L1)Ni are known. A single reversible oxidation at −1.0 V is observed for (H3L1)Ni, and L1Cr shows broad, irreversible features above −0.4 V. Thus, the emergence of three reversible redox events for L1CrNi is remarkable. The overall redox profile of L1CrNi exemplifies the principle that the whole is greater than the sum of the parts. The other redox-active bimetallics, L1CrCo, L1CrFe, and 1 L VFe, contain multiple bonds unlike datively bonded L1CrNi. Nearly all of these bimetallics have valence electrons and/or holes in localized d-orbitals. Consistent with this idea, quintuply bonded L1Cr2 and L1CrMn, which have highly delocalized valence electrons, exhibit fewer redox processes. Empirically, multielectron redox processes are more favorable in heterobimetallics containing Cr, which perhaps as a midmetal is E


Article


Metal-to-metal charge transfer (MMCT) is another fascinating property of multimetallic complexes that has potential applicability in photocatalysis. In the L2CoMCl series, an intense band in the visible region was proposed to be a Co → hν

M electronic excitation, or Co(II)M(II) → Co(III)M(I) (Figure 10). The MMCT energy red-shifts as M is varied

Figure 8. Cyclic voltammograms of selected L1 bimetallic complexes.

inherently more redox flexible than the early metals V and Ti. Consistent with this finding, substituting Cr for V results in globally harsher redox potentials in L1VFe than in L1CrFe. In summary, maximizing the number of accessible redox processes may correlate with localized valence electrons and/or holes, and an appropriate metal partner can favorably shift redox potentials.15 The concept that metal−metal interactions can govern magnetic behavior is demonstrated by the isostructural L2MM′Cl systems.12,20 The magnetic susceptibilities in Figure 9 were modeled using two individual spins that couple

Figure 10. Electronic absorption spectra of isostructural L2CoMCl bimetallics and [L2Co]K in CH2Cl2.

across the first-row period, M = Mn < Fe ∼ Co < Ni ≪ Cu. On the basis of electronegativity, M(II/I) redox couples should become more favorable across the period. Indeed, the Cu(II/I) potential in L2CoCuCl is the mildest of all the M(II/I) redox couples and corresponds to the lowest excitation energy (Figure S4). We wondered if the excitation energies could be quantified based on electrochemical data. If Co and M are noninteracting in the ground and excited states, then the excitation energy could be approximate to the driving force for the corresponding redox changes or ΔE = E°M(II/I) − E°Co(III/II). For L2CoMnCl, the excitation energy (λmax = 2.85 eV) is reasonably close to ΔE (2.55 V), where the ratio of ΔE to λmax is 0.89 (Figure S5). The ratio, however, decreases significantly across the first-row period: 0.73 (Fe), 0.70 (Co), 0.53 (Ni), and 0.27 (Cu). Presumably, the deviation is due to significant Co−M interactions in the ground state, which includes any intermetal electrostatic attraction. Hence, destabilizing the Co−M interaction costs additional energy, raising λmax well above ΔE. In support, the FSR of the L2CoMCl complexes decreases across the period, from 1.09 in L2CoMnCl to 1.03 in L2CoCuCl, where the stronger Co−M interactions correlate to larger deviations.

Figure 9. Temperature dependence of magnetic susceptibility, plotted as χMT versus T, of isostructural L2MM′Cl complexes.

magnetically, but unexpectedly, the nature of the coupling, anti- versus ferromagnetic, was correlated to differences in the metal−metal interaction. In weakly interacting systems (FSR 1.03−1.09), the overall spin state is equivalent to ΔN/2 because each divalent, high-spin metal couples antiferromagnetically. Hence, L2Co2Cl is S = 0; L2CoFeCl and L2FeMnCl are S = 1/2; and L2CoMnCl and L2CoCuCl are S = 1. Moreover, the strength of the magnetic coupling was found to increase across the period. A notable finding was that the ground spin state of L2Fe2Cl, which has an Fe−Fe bond of 2.2867(5) Å (FSR 0.98) is close to S = 3, implicating ferromagnetic coupling between the two iron centers engaged in a single metal−metal bond.

5. DINITROGEN REACTIVITY: N2 BINDING AND CATALYTIC N2 SILYLATION A few Co bimetallic complexes coordinate dinitrogen upon one-electron reduction.21,29 The promise of N2 reactivity was hinted by the CVs of the L1MCo complexes: the first reductive process was irreversible under N2 and became more reversible under argon (Figure 11). Chemical reduction provided the dinitrogen adducts [L1MCo(N2)]− for M = V, Cr, and Co, with an end-on N2 at cobalt. We have not been able to make [L1TiCo(N2)]−, perhaps owing to its harsh redox potential. Of relevance, isostructural N2 adducts were isolated in a Co−Al system in two redox states, [L1AlCo(N2)]0,−. The [L1MCoF


Article


Figure 12. Plot of initial rates (Δ[N(SiMe3)3]/ Δt) versus [L1Co2] in catalytic N2 silylation.

Figure 11. Cyclic voltammograms of L1MCo and L1AlCo(N2) complexes collected under N2 (colored lines) or argon (---) atmosphere. Asterisk (*) marks the first reductive process.

(N2)]− species shows N−N bond lengths (1.11−1.14 Å) and stretching frequencies (νN2: 1995−1971 cm−1) that are consistent with weakly activated N2. Thus, the supporting metals (M) have a limited impact on the extent of N2 activation, presumably because N2 binding at the trans position weakens the Co−M interaction. Although subtle, N2 activation decreases across the first-row period, corresponding to the trend of increasingly positive [CoML]0/− redox potentials.21 A few recent studies demonstrate that a single cobalt site can catalytically reduce N2 to NH3 or N(SiMe3)3.29,38,39 The L1Co2 bimetallic complex is highly active in catalytic N2 silylation.29 At low catalyst loadings with Me3SiCl and KC8 as the reductant, L1Co2 generates N(SiMe3)3 in high turnover number (TON) of 195 (30% yield). Reusing the catalyst in a second cycle raises the overall TON to 320.

While the supporting cobalt center maintains a constant oxidation state of +2, the active cobalt cycles between Co(0) and Co(II) in single-electron steps. Although the supporting cobalt is redox inactive during catalysis, it may play an important role in stabilizing various Co(N2(SiMe3)0−2) intermediates, akin to the postulated mechanism of Rh2catalyzed diazo-transfer reactions.41 Notably, substituting the supporting Co with Al impedes the overall catalysis, and the TON obtained with L1AlCo(N2) is 6.5-times lower than that with L1Co2. To better understand the role of the ancillary metal in N2 silylation, catalytic testing of other Co−M bimetallics are underway.

6. METAL ION EXCHANGE Metal swapping is generally undesired because it leads to complex mixtures with loss of metal site selectivity. However, metal swapping can be selective and has been used to make heterometallic analogues from their homometallic precursors.25 Simple metal halides are often used as reagents, suggesting that metal-swapping occurs via halide-bridging multimetallic intermediates.25 The L2MM′Cl complexes comprise two weakly bonded divalent first-row metal ions that are highly labile, and hence, highly susceptible to metal-swapping. We anticipated metal exchange to be most facile for those ions with the fastest aquo exchange rates, where Co2+ < Fe2+ < Mn2+. For investigation of metal exchange, the L2MM′Cl bimetallics were mixed with different MCl2 (1.2 equiv, SI) for 1 h. Longer reaction times led to rampant metal swapping, resulting in complex mixtures. The Mn(II) bimetallics L2FeMnCl and L2CoMnCl underwent productive metal exchange. With FeCl2, L2FeMnCl forms L2Fe2Cl cleanly (Figure 14). The reaction between L2FeMnCl and CoCl2 gave L2CoFeCl, where incorporation of Co in the triamido pocket necessitates both metals to rearrange. The reverse reaction to generate L2FeMnCl from MnCl2 and L2CoFeCl (or L2Fe2Cl) did not proceed. Addition of CoCl2 to L2Fe2Cl generated L2CoFeCl. The Mn(II) ion is readily displaced from L2CoMnCl using FeCl 2 or CoCl2 to generate L2 CoFeCl and L2 Co 2Cl, respectively (Figure 15). L2CoFeCl can further exchange with CoCl2 to generate L2Co2Cl. Notably, the reverse reactions do not proceed, suggesting a clear order in thermodynamic stability, L2Co2Cl > L2CoFeCl > L2CoMnCl.

cat . L1Co2(0.05 mol %)

N2(g ) + Me3SiCl(l) + KC8(s) ⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯→ N(SiMe3)3 (1atm)

THF, 299 K, 12 h

Mechanistic investigations strongly suggest that the active species are homogeneous and bimetallic.29 In reductive N2 catalysis, low TONs can make it challenging to define the rate law. Figure 12 shows that the catalysis is first-order in catalyst with a turnover frequency of ∼1 N(SiMe3)3 per minute. The DFT-calculated mechanism of the N2 silylation shares similar features to that reported for Mo(depf)2(N2)2.40 The overall mechanism begins with N2-binding to [L1Co2]− (A → B) followed by three sequential additions of •SiMe3 to the N2 ligand, and then expulsion of [(Me3Si)2N−N(SiMe3)]−. The latter converts spontaneously to two N(SiMe3)3 in the presence of excess reagents. Figure 13 shows the reaction coordinate diagram of the catalyzed four-electron reduction of N2 to [N2(SiMe3)3]−, where the three •SiMe3 add to the cobaltbound N2 in a distal−distal−proximal sequence. The first •SiMe3 addition forms silyldiazenido(1−) C. The second •SiMe3 addition generates disilylhydrazido(2−) D in the ratedetermining step. An exergonic dissociation of one phosphine donor to alleviate steric repulsion gives D*, where the asterisk denotes an unbound phosphine. The third •SiMe3 preferentially attacks the proximal N atom to form trisilylhydrazido E*. Finally, the phosphine reassociates with concomitant loss of [N2(SiMe3)3]− to generate F, which reduces with KC8 to A to close the catalytic cycle. G


Article


Figure 13. DFT-calculated intermediates (A−F) and transition states (TS) of the reaction coordinate diagram for dicobalt-catalyzed N2 silylation. Intermediates are shown with oxidation states and N−N bond distances (Å). Energies (including activation barriers in parentheses) are in kcal/mol.

Figure 15. Metal exchange reactivity of the L2CoMCl bimetallics. Figure 14. Metal exchange reactions of L2FeMCl bimetallics.

been engendered by platforms that support adjacent open sites across the metals.42,43 Metal−metal bonding is increasingly spotlighted as a structural motif in catalysis. In Ni−Fe hydrogenases, a transient Ni−Fe bond is critical for H2 evolution by stabilizing intermediates and serving as a proton acceptor.44 Metal−metal species are intermediates in catalytic C−H bond functionalization.45−47 The recent exciting developments suggest a wealth of opportunities in this field.

7. CLOSING REMARKS The double decker ligand design was created to investigate the effects of an ancillary metal on a single, active metal center via metal−metal interactions. Through systematic variation of the ancillary or “active” metal, key factors dictating bond order, redox and magnetic properties, and reactivity have been elucidated. The discovery of emergent properties remains largely empirical, though some periodic trends have been found. Creative ligand designs will continue to advance our understanding of metal−metal bonded complexes and innovate their uses in catalysis. Cooperative reactivity of two metals has

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.accounts.5b00336. H


Article


■

(9) Rudd, P. A.; Liu, S. S.; Planas, N.; Bill, E.; Gagliardi, L.; Lu, C. C. Multiple Metal-Metal Bonds in Iron-Chromium Complexes. Angew. Chem., Int. Ed. 2013, 52, 4449−4452. (10) Kuppuswamy, S.; Powers, T. M.; Krogman, J. P.; Bezpalko, M. W.; Foxman, B. M.; Thomas, C. M. Vanadium-iron complexes featuring metal-metal multiple bonds. Chem. Sci. 2013, 4, 3557−3565. (11) Rudd, P. A.; Liu, S.; Gagliardi, L.; Young, V. G.; Lu, C. C. Metal−Alane Adducts with Zero-Valent Nickel, Cobalt, and Iron. J. Am. Chem. Soc. 2011, 133, 20724−20727. (12) Tereniak, S. J.; Carlson, R. K.; Clouston, L. J.; Young, V. G., Jr.; Bill, E.; Maurice, R.; Chen, Y. S.; Kim, H. J.; Gagliardi, L.; Lu, C. C. Role of the metal in the bonding and properties of bimetallic complexes involving manganese, iron, and cobalt. J. Am. Chem. Soc. 2014, 136, 1842−1855. (13) Zall, C. M.; Clouston, L. J.; Young, V. G., Jr.; Ding, K.; Kim, H. J.; Zherebetskyy, D.; Chen, Y. S.; Bill, E.; Gagliardi, L.; Lu, C. C. Mixed-valent dicobalt and iron-cobalt complexes with high-spin configurations and short metal-metal bonds. Inorg. Chem. 2013, 52, 9216−9228. (14) Nagashima, H.; Sue, T.; Oda, T.; Kanemitsu, A.; Matsumoto, T.; Motoyama, Y.; Sunada, Y. Dynamic titanium phosphinoamides as unique bidentate phosphorus ligands for platinum. Organometallics 2006, 25, 1987−1994. (15) Greenwood, B. P.; Forman, S. I.; Rowe, G. T.; Chen, C.-H.; Foxman, B. M.; Thomas, C. M. Multielectron Redox Activity Facilitated by Metal−Metal Interactions in Early/Late Heterobimetallics: Co/Zr Complexes Supported by Phosphinoamide Ligands. Inorg. Chem. 2009, 48, 6251−6260. (16) Kuppuswamy, S.; Bezpalko, M. W.; Powers, T. M.; Turnbull, M. M.; Foxman, B. M.; Thomas, C. M. Utilization of Phosphinoamide Ligands in Homobimetallic Fe and Mn Complexes: The Effect of Disparate Coordination Environments on Metal−Metal Interactions and Magnetic and Redox Properties. Inorg. Chem. 2012, 51, 8225− 8240. (17) Krogman, J. P.; Foxman, B. M.; Thomas, C. M. Activation of CO2 by a Heterobimetallic Zr/Co Complex. J. Am. Chem. Soc. 2011, 133, 14582−14585. (18) Clouston, L. J.; Bernales, V.; Cammarota, R. C.; Carlson, R. K.; Bill, E.; Gagliardi, L.; Lu, C. C. Heterobimetallic Complexes that Bond Vanadium to Iron, Cobalt, and Nickel. Inorg. Chem. 2015, under revision. (19) Clouston, L. J.; Siedschlag, R. B.; Rudd, P. A.; Planas, N.; Hu, S.; Miller, A. D.; Gagliardi, L.; Lu, C. C. Systematic variation of metalmetal bond order in metal-chromium complexes. J. Am. Chem. Soc. 2013, 135, 13142−13148. (20) Eisenhart, R. J.; Carlson, R. K.; Clouston, L. J.; Young, V. G., Jr.; Chen, Y.-S.; Bill, E.; Gagliardi, L.; Lu, C. C. Influence of Copper Oxidation State on the Bonding and Electronic Structure of CobaltCopper Complexes. Inorg. Chem. 2015, accepted. (21) Clouston, L. J.; Bernales, V.; Carlson, R. K.; Gagliardi, L.; Lu, C. C. Bimetallic Cobalt-Dinitrogen Complexes: Impact of the Supporting Metal on N2 Activation. Inorg. Chem. 2015, 54, 9263. (22) Wagner, F. R.; Noor, A.; Kempe, R. Ultrashort metal-metal distances and extreme bond orders. Nat. Chem. 2009, 1, 529−536. (23) Nair, A. K.; Harisomayajula, N. V. S.; Tsai, Y.-C. Theory, synthesis and reactivity of quintuple bonded complexes. Dalton Trans. 2014, 43, 5618−5638. (24) Miller, D. L. Design and Synthesis of New Ligand Scaffolds and Transition Metal Complexes for Small Molecule Activation. Thesis, University of Minnesota, Minneapolis, MN, 2013. (25) Eames, E. V.; Hernández Sánchez, R.; Betley, T. A. Metal Atom Lability in Polynuclear Complexes. Inorg. Chem. 2013, 52, 5006−5012. (26) Zall, C. M. Design, Synthesis, and Characterization of Transition Metal Compounds Using Binucleating and Bifunctional Ligands: Strategies for the Multi-Electron Reduction of Small Molecules. Ph.D. Thesis, University of Minnesota, Minneapolis, MN, 2013. (27) Pauling, L. Atomic Radii and Interatomic Distances in Metals. J. Am. Chem. Soc. 1947, 69, 542−553.

Experimental data are provided (CCDC 1428234− 1428238) (PDF) Combined CIF file for L1TiFe, L1TiNi, [L1CrCo(CH3CN)]BPh4, [L1CrNi]BF4, and L2CoNiCl (CIF) CIF file for L2CoNiCl (CIF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest. Biographies Reed J. Eisenhart obtained a B.S. in Chemistry from the University of Kansas and started his Ph.D. studies at the University of Minnesota in 2010 on metal−metal bonding interactions of first row transition metals. Laura J. Clouston received a B.S. in Chemistry from the University of Nebraska. Since 2011, she has been a Ph.D. student at the University of Minnesota. Her research interests are heterobimetallic complexes and crystallography. Connie C. Lu received a B.S. in Chemistry from MIT and a Ph.D. from Caltech with Jonas Peters. She was a Humboldt postdoctoral fellow at the Max Planck Institute for Bioinorganic Chemistry with Karl Wieghardt. In 2009, she began independent research at the University of Minnesota, where she is an Associate Professor of Chemistry. Her research interests lie at the intersection of synthesis, catalysis, and spectroscopy.

■

ACKNOWLEDGMENTS We thank past and present members for their research contributions and support. We are indebted to Laura Gagliardi, Eckhard Bill (Max Planck), and their members for their partnership on this work. We thank our collaborators Victor Young, Jr., Yu-Sheng Chen (Argonne), and William Tolman. Attila Kovacs assisted with figures. The work was funded by the National Science Foundation (CHE-1254621). C.C.L. is a Sloan fellow. L.J.C. thanks the Graduate School for a dissertation fellowship.

■

REFERENCES

(1) Cotton, F. A.; Curtis, N. F.; Harris, C. B.; Johnson, B. F. G.; Lippard, S. J.; Mague, J. T.; Robinson, W. R.; Wood, J. S. Mononuclear and Polynuclear Chemistry of Rhenium (III): Its Pronounced Homophilicity. Science 1964, 145, 1305−1307. (2) Cotton, F. A., Murillo, C. A., Walton, R. A., Eds. Multiple Bonds Between Metal Atoms, 3rd ed.; Springer: New York, NY, 2005. (3) Falvello, L. R.; Foxman, B. M.; Murillo, C. A. Fitting the Pieces of the Puzzle: The δ Bond. Inorg. Chem. 2014, 53, 9441−9456. (4) Nguyen, T.; Sutton, A. D.; Brynda, M.; Fettinger, J. C.; Long, G. J.; Power, P. P. Synthesis of a stable compound with fivefold bonding between two chromium(I) centers. Science 2005, 310, 844−847. (5) Liddle, S. T., Ed. Molecular Metal-Metal Bonds; Wiley-VCH: Weinheim, Germany, 2015. (6) Stephan, D. W. Early-late heterobimetallics. Coord. Chem. Rev. 1989, 95, 41−107. (7) Cooper, B. G.; Napoline, J. W.; Thomas, C. M. Catalytic Applications of Early/Late Heterobimetallic Complexes. Catal. Rev.: Sci. Eng. 2012, 54, 1−40. (8) Morse, M. D. Clusters of transition-metal atoms. Chem. Rev. 1986, 86, 1049−1109. I


Article

Accounts of Chemical Research (28) Eisenhart, R. J.; Rudd, P. A.; Planas, N.; Boyce, D. W.; Carlson, R. K.; Tolman, W. B.; Bill, E.; Gagliardi, L.; Lu, C. C. Pushing the limits of delta bonding in metal-chromium complexes with redox changes and metal swapping. Inorg. Chem. 2015, 54, 7579−7592. (29) Siedschlag, R. B.; Bernales, V.; Vogiatzis, K. D.; Planas, N.; Clouston, L. J.; Bill, E.; Gagliardi, L.; Lu, C. C. Catalytic Silylation of Dinitrogen with a Dicobalt Complex. J. Am. Chem. Soc. 2015, 137, 4638−4641. (30) La Macchia, G.; Li Manni, G.; Todorova, T. K.; Brynda, M.; Aquilante, F.; Roos, B. O.; Gagliardi, L. On the analysis of the Cr-Cr multiple bond in several classes of dichromium compounds. Inorg. Chem. 2010, 49, 5216−5222. (31) Cotton, F. A.; Daniels, L. M.; Falvello, L. R.; Matonic, J. H.; Murillo, C. A. Trigonal-lantern dinuclear compounds of diiron(I,II): the synthesis and characterization of two highly paramagnetic Fe2(amidinato)3 species with short metal-metal bonds. Inorg. Chim. Acta 1997, 256, 269−275. (32) Jones, C.; Schulten, C.; Rose, R. P.; Stasch, A.; Aldridge, S.; Woodul, W. D.; Murray, K. S.; Moubaraki, B.; Brynda, M.; La Macchia, G.; Gagliardi, L. Amidinato− and Guanidinato−Cobalt(I) Complexes: Characterization of Exceptionally Short Co−Co Interactions. Angew. Chem., Int. Ed. 2009, 48, 7406−7410. (33) Zall, C. M.; Zherebetskyy, D.; Dzubak, A. L.; Bill, E.; Gagliardi, L.; Lu, C. C. A Combined Spectroscopic and Computational Study of a High-Spin S = 7/2 Diiron Complex with a Short Iron−Iron Bond. Inorg. Chem. 2012, 51, 728−736. (34) Hopkins, M. D.; Gray, H. B.; Miskowski, V. M. δ→δ* Revisited: What the energies and intensities mean. Polyhedron 1987, 6, 705−714. (35) San Filippo, J. Diamagnetic anisotropy induced by metal-metal multiple bonds. Inorg. Chem. 1972, 11, 3140−3143. (36) Smith, J. M.; Subedi, D. The structure and reactivity of iron nitride complexes. Dalton Trans. 2012, 41, 1423−1429. (37) Clouston, L. J.; Bernales, V.; Gagliardi, L.; Lu, C. C. Synthesis and Characterization of Metal-Metal Complexes that Bond Titanium to Iron, Cobalt, and Nickel; in preparation. (38) Imayoshi, R.; Tanaka, H.; Matsuo, Y.; Yuki, M.; Nakajima, K.; Yoshizawa, K.; Nishibayashi, Y. Cobalt-Catalyzed Transformation of Molecular Dinitrogen into Silylamine under Ambient Reaction Conditions. Chem. Eur. J. 2015, 21, 8905−8909. (39) Del Castillo, T. J.; Thompson, N. B.; Suess, D. L. M.; Ung, G.; Peters, J. C. Evaluating Molecular Cobalt Complexes for the Conversion of N2 to NH3. Inorg. Chem. 2015, 54, 9256. (40) Tanaka, H.; Sasada, A.; Kouno, T.; Yuki, M.; Miyake, Y.; Nakanishi, H.; Nishibayashi, Y.; Yoshizawa, K. Molybdenum-Catalyzed Transformation of Molecular Dinitrogen into Silylamine: Experimental and DFT Study on the Remarkable Role of Ferrocenyldiphosphine Ligands. J. Am. Chem. Soc. 2011, 133, 3498−3506. (41) Pirrung, M. C.; Liu, H.; Morehead, A. T. Rhodium Chemzymes: Michaelis−Menten Kinetics in Dirhodium(II) Carboxylate-Catalyzed Carbenoid Reactions. J. Am. Chem. Soc. 2002, 124, 1014−1023. (42) Velian, A.; Lin, S.; Miller, A. J. M.; Day, M. W.; Agapie, T. Synthesis and C−C Coupling Reactivity of a Dinuclear NiI−NiI Complex Supported by a Terphenyl Diphosphine. J. Am. Chem. Soc. 2010, 132, 6296−6297. (43) Pal, S.; Uyeda, C. Evaluating the Effect of Catalyst Nuclearity in Ni-Catalyzed Alkyne Cyclotrimerizations. J. Am. Chem. Soc. 2015, 137, 8042−8045. (44) Kampa, M.; Pandelia, M.-E.; Lubitz, W.; van Gastel, M.; Neese, F. A Metal−Metal Bond in the Light-Induced State of [NiFe] Hydrogenases with Relevance to Hydrogen Evolution. J. Am. Chem. Soc. 2013, 135, 3915−3925. (45) Powers, D. C.; Xiao, D. Y.; Geibel, M. A. L.; Ritter, T. On the Mechanism of Palladium-Catalyzed Aromatic C−H Oxidation. J. Am. Chem. Soc. 2010, 132, 14530−14536. (46) Kornecki, K. P.; Briones, J. F.; Boyarskikh, V.; Fullilove, F.; Autschbach, J.; Schrote, K. E.; Lancaster, K. M.; Davies, H. M. L.; Berry, J. F. Direct Spectroscopic Characterization of a Transitory Dirhodium Donor-Acceptor Carbene Complex. Science 2013, 342, 351−354.

(47) Mazzacano, T. J.; Mankad, N. P. Base Metal Catalysts for Photochemical C−H Borylation That Utilize Metal−Metal Cooperativity. J. Am. Chem. Soc. 2013, 135, 17258−17261.

J


Configuring Bonds between First-Row Transition Metals - Accounts of

Recommend Documents