The formation of end-on bridging N2-complexes typically proceeds in a two step pro-cessviathe coordination of a second metal center to a end-on bound N2-moiety. Initial binding of N2 to one metal center is similar to other diatomic ligands like CO and can be understood as a combination of σ-donation of the lone pair of the N2-unit into an empty metal d-orbital with suitable symmetry andπ-backdonation from a filled metal d-orbital into an empty π∗-orbital of the N2 ligand. Differences arise from the much higher HOMO-LUMO gap of N2 (10.82 eV) compared to other diatomic ligands like CO (9.34 eV), which results in a smaller orbital overlap and weaker activation of the N2 -moiety.24 Nevertheless, binding to a metal center increases the electron density on the N2 ligand and its affinity for binding a second metal ion.34 Independent studies by the groups of Cumminsand Schneider have shown, that reduction of the end-on bound N2-complex can promote and accelerate the formation of end-on N2 bridged complexes.35,36
M N N
π-backdonation
M N N
σ−donation
Figure 2: Orbital interactions between N2 and a metal center in end-on bound N2 -complexes.
Another possibility for the formation of such N2-bridged compounds is the coupling of two terminal nitrides, the microscopic reverse to N2-splitting into terminal nitrides, which is mainly found for late transition metals.16,37–51
Since both metal ions get in close proximity the formation of N2-bridged dinuclear species can be inhibited by usage of too sterically demanding supporting ligands (Scheme 3,A). However, a certain shielding of the metal ion(s) is required to prevent the formation of strong metal-metal bonds, as it has been demonstrated byCummins
In this context, the group could also show that reversible cyclometalation can be used as a strategy to prevent metal-metal bond formation (Scheme 3,C).57
R2N Mo
Scheme 3: Influence of the supporting ligands in formation of end-on bridging N2 -complexes.35,52–57
Once formed, the degree of N2-activation and therefore the stability of the obtained N2bridged complexes can be achieved by consideration of orbital interactions, as first discussed byGrayandChatt.58,59Thereby, the molecular orbitals are generated from linear combination of the metal d-orbitals and nitrogen p-orbitals. The symmetry of the N2-bridged complex has a strong influence on the energetic order of the resulting frontier molecular orbitals (FMOs).
In N2-bridged dinuclear species with S6-symmetry, linear combination of the dxz and dyz with the two π and the two π∗-orbitals of the N2-ligand results in four sets of π-orbitals (1eu, 1eg, 2eu, 2eg) each set consisting of two degenerate orbitals (Scheme 4, black). Additionally, M-N2-σ-orbitals (au, Scheme 4, green) are generated by overlap between the dz2 orbitals of the two metal centers with the σ∗-orbital of the bridg-ing N2-ligand. Due to the lack of N2-molecular orbitals with appropriate symmetry, the two remaining d-orbitals of each metal center, dxy and dx2−y2, give two sets of NN-non-bondingδ-orbitals (3eg and 3eu) again each set consisting of two degenerate orbitals (Scheme 4, purple).
Taking these simple and qualitative MO considerations to account, the degree of N2 -activation can be correlated to population of π and π∗-orbitals. A very illustrative example stems from the group of Cummins, who investigated the degree of N2 -activation in a series of Mo-triamido-complexes, [(N2){Mo(N(R)Ar)3}2]n+, (R =tBu;
Ar = 3,5-C6H3Me2; n = 0 I, n = 1II, n = 2III). By comparing the NN-bond lengths and NN-stretching frequencies, an increase of the degree of N2-activation was observed with neutral I bearing the weakest activated bridging N2-ligand within this redox-series (Figure 3). The observed trend can be correlated to the number ofπ-electrons within the {MoNNMo}-manifold.
M N N M
Scheme 4: Qualitative Molecular Orbital scheme for S6-symmetric N2-bridged dinu-clear species. The symmetry of the resulting molecular orbitals (MOs) is indicated by the color with greenσ-, blackπ- and purpleδ-symmetry.59
NeutralIfeatures overall tenπ-electrons, assuming the bridging N2 ligand to be neu-tral, both Mo(III)-ions deliver three π-electrons, while the remaining four π-electrons stem from the N2ligand itself. Consequently, this electron count results in a 1e4u1e4g2e2u -configuration within the {MoNNMo}-manifold with a NN-bonding and M-N2-antibonding HOMO (2eu; Scheme 4). Upon oxidation to a 1e4u1e4g2e1u- (II) or 1eu41eg42e0u-configuration (III), this MO gets depleted, which explains the observed increase in the degree of N2 -activation upon oxidation.60,61 IV
Mo N
increasing N2 activation
Mo N
Figure 3: Structural and spectroscopic properties of the [(N2){Mo(N(R)Ar)3}2]n+(n = 0-2)-redox series (I, II, III) and the hetero-bimetallic Mo/Nb analogue (IV) by Cum-mins.60–62
This picture gains further support by magnetic measurements. Neutral I features a magnetic moment of μeff= 2.85 µB in agreement with a triplet ground state.60 The magnetic moment of monocationicIIwas determined toμeff= 1.96µBin line with the expected doublet ground state, while dicationicIIIis diamagnetic.61
The group ofCumminswas also able to prepare a Nb/Mo hetero-bimetallic analogue (IV), which has a π9-configuration isoelectronic to monocationic II. Accordingly, the NN-bond length (dNN= 1.235(10) Å) as well as the NN-stretching frequency of IV (ν˜NN= 1583 cm-1) are almost invariant from homobimetallic II. Nevertheless, com-bined DFT and EPR studies showed valence delocalisation of the odd electron over the whole {MoNNNb}-core, further supporting a covalent bonding picture.62
The σ- and π-donating abilities of the supporting ligands can have a strong influ-ence on the energetic order of the MO’s shown in Scheme 4. This was demonstrated by Floriani upon a redox-series of N2-bridged dinuclear vanadium compounds (V).
Neutral V has a diamagnetic ground state and features a relatively long NN-bond (dNN= 1.222(4) Å) in line with eight π-electrons fully populating the 1eu- and 1eg -orbitals of the {VNNV}-core (Scheme 4). Further reduction giving monoanionic V– or dianionic V2 – does not lead to significant changes in the NN-bond lengths as it would have been expected by populating the NN-antibonding 2eu-orbital following the MO-scheme shown in Scheme 4.63–65
V N N V
E
H H H H
H H 3e
2e 1δn.b /
2δn.b
n V N N V
Mes Mes Mes Mes
Mes
Mes n
n = 0, 1-, 2- n = 0 1-
2-V
Scheme 5: left: Floriani’sN2-bridged dinuclear vanadium dimerV.Right: Qualitative MO-scheme for the truncated [(N2)-{VH3}2]-model.63–65
These differences were rationalized computationally using a truncated [(N2){VH3}2 ]-model.63–65The calculations on the neutral compound revealed a diamagnetic ground state in which the HOMO consists of a set of two degenerate orbitals (2e, Scheme 5), which are both fully occupied. The LUMO is only about 1 eV higher in energy and features mostly metal δ-character. The relatively small energy gap leads to an ac-cessible exited triplet state and second order contributions in the magnetic sus-ceptibility, in agreement with the observed TIP (temperature independent param-agnetism) for V. Single electron reduction of V leads to population of the δ-orbital
and a doublet magnetic ground state. The NN-non-bonding character of this orbital is expressed by a nearly unchanged NN-bond length. Further reduction leads to a (1δ1/2δ1)-configuration. Again the metal-based character of these orbitals leads to in-significant changes in the NN-bond lengths. The missing contribution of the bridging N2ligand in theδ-orbitals results in very weak antiferromagnetically coupling of both S= 1/2 centers, in agreement with the observed magnetic behavior ofV2 –.
A fourth auxiliary ligand at the metal center can lead to two different symmetries de-pending on the coordination sphere around the metal ion. Addition of a ligand trans to the bridging N2-ligand results in a trigonal bipyramidal coordination sphere around the metal, which can have a strong impact on the stability of the N2-bridge with re-spect to N2-cleavage (see chapter 1.2.2). Nevertheless, the fourth ligandtransto the N2-bridge does not lead to significant changes in the degree of N2-activation. For ex-ample, both trigonal bipyramidal π10-electron complexes reported by Schrock(VIa) and Copéret (VII) feature almost identical NN-bond lengths, very close to those of isoelectronicI(Scheme 6).66,67
Mo
N N
N
N N TMS TMS
TMS
Mo N N
N N N
TMS TMS
TMS
Mo
N O
O
OSi(OtBu)3 (tBuO)3SiO
Mo N O
O
O OSi(OtBu)3 (tBuO)2Si
tBu
Si(OtBu)2
tBu
(OtBu)3Si
Schrock Copéret
π10 dNN = 1.209(5) Å π10
dNN = 1.20(2) Å
VIa VII
Scheme 6: Trigonal bipyramidal N2-bridged dinuclear species with π10-configuration reported bySchrockandCopéret.66,67
The group of Liddlereported several Ti-congeners to Schrock’sVIa with different π-electron counts (Scheme 7). The bridging N2 ligand of neutral congener VIII with sixπ-electrons shows only very weak activation (dNN= 1.121(6) Å;ν˜NN= 1701 cm-1), due to only two electrons occupying the NN-antibonding 1eu-orbital. In line with the described MO-picture, further reduction to the potassium supportedπ8-configurated IX results in further population of the 1eu-orbital and an increased degree of N2 -activation as indicated by the NN-bond length (dNN= 1.315(3) Å) and the NN-stretching frequency (ν˜NN= 1201 cm-1). Abstraction of the two supporting potassium ions yields in the formation of dianionic X with an increased NN-bond distance as well as an slightly increased NN-stretching frequency. Notably, usage of Mg insted of KC8 as re-ducting agent does not result in formation of N2-bridged species, which illustrates the influence of the used reducing agent and the obtained cation.68,69
Ti
Scheme 7: Trigonal bipyramidal N2-bridged Ti-complexes with different π-electron count reported byLiddle(B15C5 = benzo-15-crown-5-ether).68
The addition of a fourth auxiliary ligand can also result in square-pyramidal coordina-tion of the metal ions, typically with the N2-bridge on the apical side. As a result of the changed symmetry from threefold to fourfold, the orbital overlap with the supporting ligands changes and the b1u- and b2g-orbitals drop in energy. Since these orbitals are generated by linear combination of the two metal dxy-orbitals with δ-symmetry, they provide NN-non-bonding character, which has to be taken into account when correlating the overall electron count to the degree of N2-activation (Scheme 8).
M N N M
E
L4M ML4 N2
πNN π*NN σ*NN
dxy, dxz, dyz dz2
D4h
2eg
2eg au
2eu
au 2eu
1eg
1eu
1eg 1eu
z x y b1g b2u
dx2-y2
b1u b2g
b1u b2g b1u b2g
π−π−π δn.b./δ∗n.b.
σ−σ∗−σ
π−π∗−π π∗−π−π∗ π∗−π∗−π∗
δn.b./δ∗n.b.
Scheme 8: Qualitative Molecular Orbital scheme for D4h-symmetric N2-bridged dinu-clear complexes. The symmetry of the resulting molecular orbitals (MOs) is indicated by the color with greenσ-, blackπ- and purpleδ-symmetry.59
Such (idealized) D4h-symmetric compounds are often observed, when pincer com-plexes are used. For example, the group of Schneider reported two PNPtBu-pincer (PNPtBu = [N(CH2CH2PtBu2)2]– supported N2-bridged dinuclear complexes containing either Mo (XI) or Re (XII) and therefore different electron counts within the {MNNM}-core (Figure 4).36,70
The Mo-compound XI contains overall twelve electrons within the {MoNNMo}-core, four steming from the N2 ligand itself and two times four electrons from the (formal) Mo(II)-centers. Following the MO-picture (Scheme 8) this results in a 1e4u1e4gb1u2 b22g configuration with a π-electron count of eight. The degree of N2-activation is very close to III as judged by the NN-bond-length (dNN= 1.258(9) Å) and NN-stretching frequency (ν˜NN= 1343 cm-1).70 Moving from group 6 to group 7, two additional elec-trons are introduced, which leads to occupation of the 2eu-orbitals and an overall 1e4u1e4gb21ub2g2 2e2u-configuration with anπ-electron count of ten. Accordingly, the de-gree of N2-activation in XII is smaller compared to XI, as indicated by the shorter NN-bond-length (dNN= 1.202(10) Å), which is very similar to the one of Cummins’
neutralI.36 XIII
Mo
Figure 4: The PNP-pincer supported N2-bridged Mo (XI) and Re (XII; XIII) dinuclear complexes reported bySchneider.36,70,71
An additional fifth auxiliary ligand trans to the N2-bridge does not effect the en-ergetic order of the MOs with π- and δ-symmetry within the described MO-scheme (Scheme 8). It can therefore be used to rationalize the degree of N2-activation in oc-tahedrally coordinated N2-bridged dinuclear complexes, like for the weakly activated [(N2){Ru(NH3)3}2]4+ (π12δ4; dNN= 1.12 Å; ν˜NN= 2100 cm-1) or the strong activated [{(PhMe2P)4ClRe}(N2){MoCl4(OMe)}] (π8δ3; dNN= 1.21 Å;ν˜NN= 1660 cm-1).72,73 Nevertheless, the stability of such N2-bridged dinuclear species with respect to N2 -cleavage is increased compared to the square pyramidal analogues. For example, featuresPr-substituted and octahedrally coordinatedXIII(Figure 4) the same electron count and degree of N2-activation as itstBu-substituted analogueXII, but is thermally stable with respect to N2 cleavage. A more detailed discussion about the influence of a ligandtransto the N2-bridge regarding N2-cleavage will be given in chapter 1.2.2.71
Upon usage of β-diketiminate (nacnac) ligands, several end-on N2-bridged dinuclear complexes, bearing late transition metals (Fe, Co, Ni), have been reported.74–79 De-pending on the steric encumbrance of the nacnac-substituents, both {M(nacnac)}-moieties are either coplanar or perpendicular oriented, resulting in an idealized D2h -or D2d-symmetry, respectively. Furthermore, a distortion from linearity of the {M-N-N-M} is often found, indicating low bending potentials of the core within these type of compounds. The degree of N2-activation can be rationalized using the MO-diagramm depicted in Scheme 9.
Assuming the x-axis to be oriented along the {M-N-N-M}-bond, linear combination of the dxy- and dxz-orbitals with the respective N2-π-orbitals results in four sets of π-orbitals each set consisting of two degenerate orbitals. The remaining dyz, dz2
and dx2−y2-orbitals form three sets of NN-non-bonding orbitals each consisting of two (nearly) degenerate orbitals, which are all located between the π2- and π3-level.80 These considerations are in line with detailed spectroscopic and computational anal-ysis of [(N2){Fe(nacnactBu)}2] (XIVa), whose overall S= 3 ground state was ratio-nalized by two high-spin Fe(II)-centers (S= 2), which are each antiferromagnetically coupled to the bridging N22 –-ligand (S= 1), leading to an overall {↑ ↑ ↑ ↑ ↓ ↓ ↑ ↑ ↑ ↑ }-three-spin model.75,80 Accordingly, the relative strong degree of N2-activation within
XIVaand its Me-substituted congener [(N2){Fe(nacnacMe}2] (XVa) (Table 2, entries 1
Scheme 9: left: Qualitative MO-diagramm for D2h- or D2d-symmetric end-on N2 -bridged dinuclear compounds.80right: Generalized structure ofXIVandXV.
Further reduction is metal centered and leads to full occupation of the dyz/dz2-orbitals.
However, increased Fe→N2-backbonding leads to further weakening of the N2-bond (Table 2, entries 3 and 4).74,75The same trend was observed for the group 9-congener (Table 2, entries 6 and 7). Compared to the Fe-analogue the degree of N2-activation is smaller due to the lower energy of the d-orbitals, which is even more expressed for the Ni-analogue (Table 2, entry 8).76,77
Table 2: Comparison of the structural and electronic properties of several nacnac-supported N2-bridged dinuclear compounds, M2’[(N2){M(nacnacR)}2]X, and their N2
The group of Holland was also able to abstract the potassium-cations from XVb.
Thereby the N2-bond grows short accompanied by a hypsochromic shift of the NN-stretching frequency, indicating a weaker activation in the absence of a counter-cation within the complex (Table 2, entry 5).79 However, variation of the alkali-metal inXVb showed no significant changes in the degree of N2-activation.74,78The only difference arises from the different size of the alkali-metal-ions, which leads to twisting and a larger torsion-angle between the {Fe(nacnacMe)}-moieties (Na = 0◦; Cs = 50.6◦).78 Notably, upon usage of a sterically less encumbered nacnac-ligand, [MeC(CMeNC6H3 -2,6-Me2)2]–, full cleavage of the NN-bond into tetra- or trinuclear bis-µ-nitride com-plexes was observed.13,81
Besides by the number of electrons and the coordination sphere of the metal centers the degree of N2-activation is also influenced by the metal itself. This was demon-strated by the group ofSitafor a series of group 4-6 N2-bridgedη5-cyclopentadienyl/η2 -amidinate complexesXVI(Figure 5). Moving down within a group the degree of N2 -activation increases as indicated by significantly elongated NN-bond-distances (Ta-ble 3). This trend can be explained by the cathodic shift of the oxidation potentials of the metal ions and therefore increased backbonding from the metal to the N2-ligand.
Similarly, a correlation between the degree of activation and the oxidation potential of the respective metal was also found within a row.82–85
Notably, within group 4 the Ti-complexXVIais the only one featuring a end-on bridg-ing µ2:η1:η1-N2-ligand, best described as N22 –. In contrast, both higher homologes, Zr (XVIb) and Hf (XVIc), feature a highly activated side-on boundµ2:η2:η2-N2-ligand, which is best described as an N24 –-ligand. The higher degree of activation was as-signed to the different oxidation potentials, which renders further oxidation of Ti(III) to Ti(IV), while the side-on coordination was attributed to the larger covalent radii of Zr and Hf (both 1.75 Å) compared to Ti (1.60 Å).83,86
M N
iPrN NiPr
N M NiPr
iPrN
M = Ti, V, Nb, Ta, Mo, W
XVI
Figure 5: Sita’s isostructural [(N2)(MCp*am)2] complexes (am = [N(iPr)C(Me)N(iPr)]–)·
Table 3: NN bond distances in the N2-bridged η5 -cyclopentadienyl/η2-amidinate complexes reported bySita.aexchange of methyl group in amidinate with NMe2. bexchange of methyl group in amidinate with phenyl.82–85
metal coord. mode dNN [Å] Ref.
TiXVIa µ2:η1:η1 1.270(2) 83 Zra XVIb µ2:η2:η2 1.518(2) 86 Hf XVIc µ2:η2:η2 1.611(4) 86 VXVId µ2:η1:η1 1.225(2) 84 NbbXVIe µ2:η1:η1 1.300(3) 84 TaXVIf µ2:η1:η1 1.313(4) 82 MoXVIg µ2:η1:η1 1.267(2) 83 WXVIh µ2:η1:η1 1.277(8) 83
The group of Chirik illustrated the influence of the supporting ligands based on a redox-series of overall five end-on N2-bridged terpyridine supported dinuclear Mo-complexes, [(N2){Mo(TpyPh)(PMe2Ph)2]n+ (XVIIn+; n = 0-4; TypPh= 4’-Ph-2,2’,6,6’2”-terpyridine) (Figure 6, left). Computional analysis describe dicationic XVII2+ as two Mo(II)-ions bridged by an N22 –-ligand, which is in good agreement with the observed NN-stretching frequency (ν˜NN= 1563 cm-1) and NN-bond length (dNN= 1.203(2) Å).
Due to mixing with the π-system of the terpyridine ligand, the degeneracy of the Mπ−∗N−πNπ−∗M orbitals is lifted, which results in a singlet ground state forXVII2+. Accord-ingly, due to removal of an electron from an NN-bonding orbital, oxidation ofXVII2+ to XVII3+ results in a significant increase of the degree of N2-activation (ν˜NN= 1482 cm-1). Interestingly, the double oxidation productXVII4+ features an al-most identical NN-stretching frequency (ν˜NN= 1477 cm-1), which was rationalized by removal of an electron from a metal centered b1u-orbital in agreement with the ob-served triplet ground state for XVII4+. Intriguingly, reduction ofXVII2+ also leads in weakening of the NN-bond, which was substantiated by the mainly TypPh-π∗-character of its LUMO (2b2g) and therefore mostly ligand-centered reduction in line with EPR measurements ofXVII+ (Figure 6,right).87
E
b3u
b2g b1g N
N N
Ph
Mo N
N MoN
Ph N N
PPh2Me PPh2Me PPh2Me
PPh2Me
n 2b2g
M N N M
L L
[Mo2N2]4+
S = 1 νNN = 1477 cm-1
[Mo2N2]3+
S = 1/2 νNN = 1482 cm-1
[Mo2N2]2+
S = 0 νNN = 1563 cm-1
[Mo2N2]+ S = 1/2 νNN = 1530 cm-1
[Mo2N2] S = 1
XVIIn
Figure 6: left: Chirik’s terpyridine supported Mo-N2-dimer redox seriesXVIIn+(n = 0-4). right: Qualitative FMO scheme of the redox series and the corresponding spin states and NN stretching frequencies.87