Experimental and Computational Studies of the Molybdenum-Flanking Arene Interaction in Quadruply Bonded Dimolybdenum Complexes with Terphenyl Ligands

1 DOI: 10.1002/chem.201xxxxxx █ MoMo Quadruple bond Experimental and Computational Studies of the Molybdenum-Flanking Arene Interaction in Quadruply Bonded Dimolybdenum Complexes with Terphenyl Ligands Mario Carrasco,[a] Irene Mendoza,[a] Eleuterio Álvarez,[a] Abdessamad Grirrane,[a] Celia Maya,[a] Riccardo Peloso,[a] Amor Rodríguez,[a] Andrés Falceto,[b] Santiago Álvarez,*[b] and Ernesto Carmona*[a] Introduction Because of its fundamental importance in chemistry, the nature of the electronic interactions involved in chemical bonds of the Lewis type is a fascinating theme of research.[1] A particularly appealing problem arises when the distance between the interacting atoms in a dative bond exceeds values expected for conventional covalent bonds. In these instances, the question arises as to whether there exists a true bond, that is, if there is sufficient a sharing of electrons to glue the atoms together.[2] Among the many contingencies of this kind that have been disclosed in the literature,[3] the long CrCarene interactions present in the molecules of the quintuply metal-metal bonded Ar’CrCrAr’ complexes reported by Power and co-workers[4] (Ar’ = terphenyl ligand) constitute a most relevant example in the context of the work reported in this contribution. These dimers exhibit a transbent CarylCrCrCaryl planar core (Chart 1, structure I) with a very short CrCr bond (ca. 1.811.83 Å) and with CrCaryl distances of ca. 2.13 Å. However, each chromium atom features also a weaker CrCarene secondary interaction (at ca. 2.29 Å, consistent with the average distance found in the CSD for chromium-arene complexes, 2.24(6) Å) with the ipso carbon of one of the aryl substituents of the terphenyl group bound to the other metal atom. Detailed structural and theoretical studies of this interaction[4,5] revealed that it is much weaker than the sigma CrCaryl bond and causes only a small weakening of the bond. Most probably, this interaction partly compensates the electron deficiency of the twelve-electron chromium centres. We have recently characterized some quadruply bonded terphenyl complexes of molybdenum and tungsten of composition M2(Ar’)(O2CR)3 and M2(Ar’)2(O2CR)2 for different terphenyl (Ar’) and carboxylate groups.[6] Referring as representative examples to the molybdenum complexes, the mono-terphenyls exhibit structure of type II (Chart 1) in which one of the Mo atoms features a MoCarene interaction at a distance of ca. 2.58 Å, slightly longer than the average molybdenum-arene distances found in the CSD, 2.38(8) Å. In the bis-terphenyl derivatives (Chart 1, structure III) the MoCarene separation increases significantly (ca. 2.78 Å although it is still much shorter than the van der Waals radii sum of 4.22 Å).[7] In the two types of compounds there is a quadruple MoMo bond of length ca. 2.10 Å. In the absence of an extensive computational investigation we [a] Dr. M. Carrasco, I. Mendoza, Dr. E. Álvarez, Dr. A. Grirrane, Dr. C. Maya, Dr. R. Peloso, Dr. A. Rodríguez, Prof. Dr. E. Carmona Instituto de Investigaciones Químicas Departamento de Química Inorgánica Universidad de Sevilla-Consejo Superior de Investigaciones Científicas Avenida Américo Vespucio 49, 41092 Sevilla (Spain) Fax: (+34) 954460565 E-mail: [email protected] [b] A. Falceto, Prof. Dr. S. Álvarez Departament de Química Inorgànica and Institut de Química Teòrica i Computacional, Universitat de Barcelona, Martí i Franquès 1-11, 08028 Barcelona (Spain) E-mail: santia[email protected] Supporting information for this article is available on the WWW under http://www.chemeurj.org/ or from the author. Abstract: To clarify the nature of the MoCarene interaction in terphenyl complexes with quadruple MoMo bonds, ether adducts of composition Mo2(Ar’)(I)(O2CR)2(OEt2) have been prepared and characterized (Ar’ = ArXyl2, R = Me, 4·OEt2; Ar’ = ArMes2, R = Me, 5·OEt2; Ar’ = ArXyl2, R = CF3, 6·OEt2) and their reactivity toward different neutral Lewis bases investigated. PMe3, P(OMe)3 and PPri3 were chosen as P- donors and the reactivity studies complemented with the use of the C-donors CNXyl and CN2C2Me4 (1,3,4,5- tetramethylimidazol-2-ylidene). New compounds of general formula, Mo2(Ar’)(I)(O2CR)2(L), 4·L6·L were obtained except for the imidazol-2-ylidene ligand that yielded a salt-like compound of composition [Mo2(ArXyl2)(O2CMe)2(CN2C2Me4)2]I, 7. The MoCarene interaction in these complexes has been analysed with the aid of X-ray data and computational studies. This interaction compensates the coordinative and electronic unsaturation of one of the Mo atoms in the above complexes but it seems to be weak in terms of sharing of electron density between the Mo and Carene atoms and appears to have no appreciable effect in the length of the MoMo, MoX and MoL bonds present in these molecules. 2 DOI: 10.1002/chem.201xxxxxx suggested[6a] that despite the weakness of the MoCarene contacts (2.582.80 Å) relative to the MoCaryl bond (2.162.19 Å), the MoCarene interaction compensates in some degree the electronic unsaturation and offers at the same time steric protection to the unsaturated metal centre. Chart 1. Structural formulae of some multiply MM bonded dichromium and dimolybdenum complexes with terphenyl ligands. To gain a deeper understanding on the nature of the MoCarene interaction in these molecules we planned its comprehensive examination by means of experimental and theoretical methods. The work reported in this paper comprises the synthesis and structural characterization of new complexes in which a [Mo2(Ar’)(O2CMe)2] fragment is stabilized by the additional coordination of an iodide ligand and a neutral Lewis base (Et2O, PMe3, PPri3, P(OMe)3 and CNXyl), or by two molecules of 1,3,4,5-tetramethylimidazol-2-ylidene, CN2C2Me4. As illustrated by the structure of a key synthetic precursor in this study, namely the ether adduct Mo2(ArXyl2)(I)(O2CMe)2(OEt2) 4·OEt2 (structure IV in Chart 1), the four-coordinate molybdenum atom that would have a fourteen-electron count participates in a bonding interaction with the proximal flanking aryl ring that can increase the electron count to 16. A computational study of this interaction performed with complexes 4·L, as well as with the mono- and bis-terphenyl complexes Mo2(ArXyl2)(O2CMe)3 and Mo2(ArXyl2)2(O2CH)2, respectively, is also reported, and compared with calculations on a model complex with an unsupported - coordinated benzene ring, [Mo2(H)(O2CMe)3(C6H6)]. Results and Discussion Synthesis and solution structure of new complexes 4·L7 In collaboration with the group of Power, we reported recently the synthesis and characterization of a variety of terphenyl complexes of the core that comprised monoterphenyl derivatives with composition Mo2(Ar’)(O2CR)3, for different terphenyl (Ar’) and carboxylate groups.[6] With some exceptions, these compounds did not undergo reaction with a second equivalent of LiAr’ to yield the targeted Mo2(Ar’)2(O2CR)2 complexes. At variance with this observation, Et2O solutions of the complex Mo2(Ar’)(O2CR)3, where Ar’ = ArXyl2 and R = Me (complex 1), reacted cleanly and smoothly (20 ºC, overnight) with MgI2 dissolved also in Et2O (Scheme 1) to afford Mo2(ArXyl2)(I)(O2CMe)2(OEt2), 4·OEt2, in fair isolated yields (ca. 55%). The substitution of the acetate group of 1, which is trans with respect to the -bonded Ar’ ligand, by iodide and a molecule of Et2O was accompanied by an attractive colour change from dark red to blue-violet. The new compound is soluble in common organic solvents (e.g. Et2O, C6H6, or C6H5Me) and is very reactive toward oxygen and water, both in solution and in the solid state. As represented also in Scheme 1, the related compounds Mo2(ArMes2)(O2CMe)3, 2, and Mo2(ArXyl2)(O2CF3)3, 3, reacted similarly with MgI2 to furnish analogous Et2O adducts, namely 5·OEt2 and 6·OEt2. Along with 4·OEt2, the former was isolated and characterized by spectroscopy and X-ray crystallography. The related trifluoroacetate 6·OEt2 was simply used in situ for further reactivity studies. Scheme 1. Synthesis of ether adducts of monoterphenyl iodide complexes. NMR data (1H and 13C{1H}) for new compounds are fully in accordance with the proposed formulation (see Supporting Information, SI, for details). Taking complex 4·OEt2 as an illustrative example, the room temperature 1H NMR spectrum contains somewhat broad resonances due, at least in part, to facile dissociation of the Et2O molecule (fast exchange between free and coordinated Et2O was observed in the presence of added Et2O). But in addition, the proximal and distal Xyl flanking rings ( 2.09 and 2.23) undergo exchange that becomes fast on the NMR time scale at temperatures above 45 ºC. This exchange is somewhat more facile than in the parent complex 1 and could be explained similarly[6a] assuming a 1,2-terphenyl shift from one Mo atom to the other through a Mo2(-Ar’) structure facilitated by dissociation of the molecule of Et2O. The lability of the Et2O ligand could additionally allow for a similar iodide shift, as suggested by the computational studies to be discussed in a following section. New complexes related to 4·OEt2, 5·OEt2, and 6·OEt2 formed upon reactions of these molecules with different Lewis bases (Scheme 2). Thus, treatment of 4·OEt2 with the phosphorus 3 DOI: 10.1002/chem.201xxxxxx donors PMe3, P(OMe)3, and PPri3 yielded the expected adducts 4·PMe3 (deep-blue), 4·P(OMe)3 (purple), and 4·PPri3 (violet). For complexes 5·OEt2 and 6·OEt2 only the reactions with PMe3 were carried out to generate 5·PMe3 and 6·PMe3, respectively, both with deep-blue colour. As shown in Scheme 2 the substitution of Et2O by the P-donor ligands occurred with a discernible stereochemical change that placed the iodide ligand in the coordination position trans with respect to the Ar’ group while the phosphorus ligand bonded to the lower coordinated Mo atom that participates in the secondary MoCarene interaction. A theoretical justification for this rearrangement will be provided in a later section. In the meantime, the different site of coordination of the neutral Lewis base may be discerned from the observation of a doublet 13C{1H} MoCaryl resonance at 173.2 ppm that exhibits a reduced 13C31P coupling of 9 Hz. This value is significantly smaller than expected for a two-bond trans-CarylMoP coupling[8] and points instead to a trans-bent CarylMoMoP ligand arrangement, subsequently confirmed by X-ray crystallography. Scheme 2. Substitution reactions in complexes 4·OEt2, 5·OEt2 and 6·OEt2 by neutral Lewis bases. See text for the nature of L. The 1H and 31P{1H} NMR spectra of 4·PMe3 (see SI) show no indications either for PMe3 dissociation or for exchange of the flanking aryl rings of the terphenyl ligand, ArXyl2. Nonetheless, the xylyl exchange becomes apparent with the aid of NOESY experiments. Benzene solutions of the related 4·P(OMe)3 and 4·PPri3 adducts display similar properties and give rise, for instance, to a MoCaryl 13C{1H} resonance that appears as a doublet around 172 ppm (3JC,P ca. 9-10 Hz). However, ether solvents like Et2O and THF induced ligand dissociation and substitution by the ether, to restore the starting complex 4·OEt2 (or the related 4·THF that was not investigated any further). For example, in the 31P{1H} NMR spectrum of a concentrated solution of 4·iPPr3 in THF-d8 signals corresponding to coordinated (47.5 ppm) and free (19.4 ppm) phosphine were recorded. Under these conditions the MoCaryl resonance appears in the 13C{1H} NMR spectrum as a singlet. UV-Vis spectroscopic data that will be discussed later provided additional evidence for this substitution process, that under sufficiently high THF : 4·PPri3 molar ratios (ca. 10-4 M solutions) yields 4·THF and free PPri3. Indeed the facile dissociation of the P-donor ligands of the above compounds finds precedent in the work of Andersen and co-workers on other phosphine containing organometallic complexes of the quadruple MoMo bond.[9] The lability of the coordinated molecule of Et2O in the adducts 4·OEt26·OEt2 made us wonder if CO coordination to their central unit could be achieved. However, no observable reaction took place between 4·OEt2 and CO (1 bar) either at room temperature or in boiling THF. Since CO is poor a -donor but excellent a -acceptor ligand, the lack of reactivity could be attributed to full involvement of the filled d orbitals of the molybdenum atoms into MoMo multiple bonding. In fact, the almost invariable MoMo distance of ca. 2.10 Å found in these and related[6] complexes is characteristic of typical, strong quadruple bonds.[10] To confirm this assumption, complexes 4·OEt26·OEt2 were reacted with CNXyl. Isocyanides[11] are better -donors and poorer -acceptors than carbon monoxide. Arylisocyanides are more efficient -acids than alkylisocyanides but behave essentially as -donors when bound to high-valent metals or to poor -donor metal units.[12,13] As depicted in Scheme 2 (L = CNXyl), treatment of solutions of the above complexes with 1 equiv of CNXyl dissolved also in Et2O, led to an immediate colour change to green and permitted the isolation of the corresponding adducts 4·CNXyl6·CNXyl as very air sensitive crystalline solids. Spectroscopic data for the new compounds (see SI) are in agreement with the proposed formulation. An informative hallmark is an IR absorption in the proximity of 2140 cm-1 due to ῡ(C≡N) of the coordinated isocyanide. This band is shifted by ca. 25 cm-1 to higher wavenumbers relative to free isocyanide (2114 cm-1). Since the isocyanide donates electron density from a molecular orbital localized on carbon that has some antibonding character (alike the HOMO of CO), -donation results in a strengthening of the CN bond and a concomitant increase of ῡ(C≡N).[1113] Thus, the positive shift of ca. 25 cm-1 observed for complexes 4·CNXyl6·CNXyl indicates that the MoCNAr bonding interaction is mostly (or exclusively) of the sigma MoCNAr type[13] and confirms that the core of these complexes behaves as a poor -donor toward -acid ligands. Scheme 3. Formation of the salt-like compound 7 by reaction of 4·OEt2 with CN2C2Me4. To complete these synthetic studies we explored the reactivity of complex 4·OEt2 toward the N-heterocyclic carbene CN2C2Me4. In marked contrast with the reactions already discussed, the resulting product, 7, was insoluble in THF and precipitated out of the reaction mixture (Scheme 3). Characterization studies revealed that complex 7 contains two molecules of CN2C2Me4 which are bound to the two molybdenum atoms and therefore that the imidazol-2-ylidine ligand displaced both the Et2O and I- groups of 4·OEt2, to yield the cationic complex 4 DOI: 10.1002/chem.201xxxxxx [Mo2(ArXyl2)(O2CMe)2(CN2C2Me4)2]+, isolated as the iodide salt. It is worth mentioning in this regard that compound 7 was the only detectable product of the reaction regardless of the 4·OEt2 : 2 (CN2C2Me4) molar ratio utilized. It was isolated as magenta crystals from its solutions in CH2Cl2 : C6H5Me solvent mixtures. The aliphatic region of the 1H NMR spectrum of 7 recorded at room temperature in CD2Cl2 is simple and consists of four singlets at 1.90, 2.09, 2.52 and 2.80 ppm, with relative intensities attributable to 12H, 12H, 12H and 6H, respectively. Clearly, at this temperature the two flanking xylyl substituents of the terphenyl ligand undergo fast exchange in the NMR time scale. By similarity with related complexes,[6a] this dynamic behavior could be due to a fast 1,2-terphenyl shift from one Mo atom to the other through a Mo(-Ar’)Mo intermediate or transition state. We have been able to computationally locate a Mo(-Ar’)Mo transition state for the 1,2-terphenyl shift from one Mo atom to the other (Figure 1) in [Mo2(ArXyl2)(O2CMe)2(CN2C2Me4)2]+, with a relative free energy of 12.45 kcal/mol, consistent with the dynamic behavior observed in the 1H-NMR spectrum at room temperature. It must be noted that the incipient interaction between the ipso atom of the non-bonded aryl group and a Mo atom in the axial region (at a distance of 3.37 Å in the calculated structure, 3.36 Å in the experimental structure), is shortened as the central phenyl ring shifts to a bridging position in the transition state, ending up in an 1 coordination through an ortho atom. Figure 1. Optimized ground state of [Mo2(ArXyl2)(O2CMe)2(CN2C2Me4)2]+ and transition state for the 1,2-aryl shift. The acetate groups are omitted for clarity. Some relevant distances are given in Å. As briefly noted earlier, the reactions that led to complexes of type 47 were accompanied by remarkable colour changes. We conclude this section with a succinct discussion of the UV-Vis spectra of these compounds. Figure 2a compares the spectrum of the parent complex 1 with those of the 4·L adducts, where L = Et2O, PMe3 and CNXyl. Figure 2b highlights the similar optical properties of 4·PMe3, 4·P(OMe)3, and 4·PPri3. All compounds exhibit a strong absorption in the range 530630 nm (ca. 1887018020 cm-1) with max varying between 1100 (1) and 1760 (4·CNXyl) M-1 cm-1. In addition, a less intense band (max ca. 540 M-1 cm-1) can be discerned in the proximity of 440 cm-1. By analogy with other terphenyl complexes of the (Mo2)4+ unit,[6] these bands can be attributed to * transitions. The above data compare well with previous findings by Cotton[14] and Gray[15] in the complexes Mo2X4(PMe3)4, where max changes between 550 (X = F) to 633 nm (X = I). These variations were attributed to changes in the energy of the LUMO (*), rather than in the HOMO ().[14] Figure 2. UV-Vis absorption spectra of compounds 1, 4·OEt2, 4·PMe3 and 4·CNXyl (a) and of the three complexes 4·L with P-donor ligands (b) in benzene solutions (ca. 10-4 M ). At this stage, it is worth recalling that UV-Vis spectroscopy provides unequivocal evidence for partial dissociation of the P- donor ligands of 4·P(OMe)3 and 4·PPri3 when dissolved in Et2O or THF. As represented in Figure 3a, max of 4·PPri3 shifts from 560 to 540 nm upon changing the solvent from C6H6 to THF, the latter spectrum being identical to that of 4·OEt2 dissolved in THF. Similar variations were recorded for 4·P(OMe)3 (Figure 3b). Figure 3. Absorption spectra of 4·PPri3 (a) and 4·P(OMe)3 (b) in benzene (___) and THF (----) solutions (ca. 10-4 M-1). In the two figures the spectrum of complex 4·OEt2 dissolved in THF (ca. 10-4 M) is represented with dotted lines (····). 5 DOI: 10.1002/chem.201xxxxxx Solid-state molecular structures of new complexes The newly prepared compounds were characterized by X-ray crystallography. Figure 4 displays the molecular structure of 4·OEt2 while perspective views of the structure of the molecules of 4·PMe3, 4·P(OMe)3 and 4·PPri3 are collected in Figure 5. Figure 6 presents the structure of the bis-NHC complexes 7. The X-ray structures of some 5·L and 6·L derivatives were also ascertained and can be found in the SI (Figures S7S12). Table 1 collects some selected bond distances and angles for representative compounds. For comparative purposes, data for Mo2(ArXyl2)(O2CMe)3, 1, and Mo2(ArXyl2)2(O2CH)2 are also incorporated.[6] Inspection of the above figures evidences the significant distortions that exist in comparison with the regular paddle-wheel geometry of the parent complex Mo2(O2CMe)4[16] and with the somewhat distorted structure of the precursor[6] Mo2(ArXyl2)(O2CMe)3, 1. The deviations may be due to the large steric requirements of the Ar’, I- and other ligands present in these complexes and were anticipated by studies on other complexes with bulky monodentate ligands.[9] In compound 1 (structure II, Chart 1) the bulkiness of the Ar’ group leads to a transoid CarylMoO angle of 165.4º, with a concomitant increase[6] of the CarylMoMo angle to 103.3º (Table 1). In complex 4·PMe3 the CarylMoI angle (C1Mo1I1 in Figure 5) is significantly smaller than expected for a trans ligand distribution (142.4º), while the MoMoI and MoMoP angles amount ca. 119.7 and 100.9º, respectively. For the related adducts 4·P(OMe)3 and 4·PPri3 the CarylMoI and MoMoI angles do not vary substantially from the above values but the MoMoP angle increases to about 104º in the P(OMe)3 complex and 108º in 4·PPri3. Dihedral angles between the MoMoI and MoMoP planes in these complexes are of ca. 3.5º. Without underestimating the steric hindrance of the P-donor ligand it appears that the voluminous iodide ligand, with a van der Waals radius[7] close to the MoMo separations (2.04 vs. 2.10 Å), has a significant contribution to these distorsions. For the objectives of this work, the most important metrics are those pertaining the CarylMoMoCarene linkages of these complexes. Hence, our discussion will be centred in this analysis. In all compounds, the MoMo bond length is practically invariant (ca. 2.092.11, Table 1) and clusters around 2.10 Å. Considering the diverse steric and electronic properties of the neutral P- and C-donor ligands present in our complexes, this constancy is remarkable and reveals that substitution of two bridging acetate groups by the monodentate Ar’, I- and L groups in complexes 4·L6·L has no appreciable effect in the length of the quadruple MoMo bond. Variations in the MoCarene separations in the range 2.572.78 Å do not affect either the MoMo distance. Figure 4. X-ray molecular structure of the ether adduct [Mo2(ArXyl2)(I)(O2CMe)2(OEt2)], 4·OEt2. The MoCaryl bond length in complexes 4·L6·L does not change significantly with the nature of L, the shortest value corresponding to 4·OEt2, at 2.164(3) Å (Caryl transoid to Et2O, with a CarylMoO bond angle of ca. 143º) and the longest (ca. 2.19 Å) to 4·PPri3 and 5·PMe3 (Caryl transoid to I-, see Table 1). However, in the cationic complex 7+ where the Ar’ group has a transoid position relative to one of the CN2C2Me4 ligands (C12Mo2C19 angle of Table 1. Selected Bond Lengths [Å] and Angles [deg] for Complexes Mo2(ArXyl2)(O2CMe)3, (1), Mo2(ArXyl2)2(O2CH)2, 4·L (L = Et2O, PMe3, P(OMe)3, PPri3, CNXyl) and [Mo2(ArXyl2)(O2Me)2(CN2C2Me4)2]I, 7. Parameter 1[a] Mo2(ArXyl2)2(O2CH)2[a] 4·Et2O 4·PMe3 4·P(OMe)3 4·PPri3 4·CNXyl 7 MoMo 2.086(2) 2.095(1) 2.097(1) 2.107(1) 2.098(1) 2.102(1) 2.117(1) 2.104(1) MoCaryl MoCarene MoL MoMoCaryl MoMo(L) transCarylMoI [d] 2.192(2) 2.572(2)[b] 103.28(4) 165.36(6) 2.187(3) 2.780(2)[c] 99.24(10) 2.164(3) 2.565(3)[b] 2.250(2) 98.49(8) 118.29(6) 143.01(9) 2.176(3) 2.636(1)[b] 2.491(2) 97.85(8) 100.88(2) 142.39(8) 2.181(3) 2.636(1)[b] 2.478(1) 99.51(9) 104.14(4) 141.54(9) 2.193(2) 2.697(1)[b] 2.569(1) 98.54(6) 108.04(3) 140.16(6) 2.186(3) 2.644(1)[b] 2.118(4) 97.88(9) 96.98(9) 144.28(9) 2.229(3) 2.658(3)[b] 2.211(3)[e], 2.318(3)[f] 98.45(7) 106.47(7)[e], 110.05(7)[f] 151.49(10) [a] Structural and characterization data have already been reported in reference 6a. [b] MCarene interaction to ortho carbon. [c] MCarene interaction to ipso carbon. [d] For compounds 1 and 4·Et2O CarylMoO angles. [e] To C34 in Figure 6. [f] To C27 also in Figure 6. 6 DOI: 10.1002/chem.201xxxxxx 151.5º), the MoCaryl distance increases to ca. 2.23 Å. This observation denotes a large trans-influence of the NHC ligand that can indeed be advanced on the basis of its calculated[17] Tolman electronic parameter (TEP) of 2051.7 cm-1. On the other hand, the late position expected for the Ar’ group in a transinfluence series on account of that experimentally determined for the C6H5- ligand in square-planar Pt(II) complexes,[18] is evidently responsible for the substantial difference in the MoC bonds to the NHC ligands of 7+. Consistent with the weaker Mo2C8 bond to the flanking arene, the carbene trans to it has a much shorter Mo2C34 bond (2.211(3) Å) than the carbene trans to the - bonded aryl, Mo1C27 (2.318(3) Å). The strong trans-influence foreseen for Ar’ ligands also explains the longer MoI bonds in complexes 4·L (L = PMe3, P(OMe)3, PPri3 and CNXyl) in comparison with the ether adducts 4·OEt26·OEt2. In the former compounds the MoI distance has an average value of about 2.82 Å, whereas in the ether complexes is of 2.76 Å. When the compounds reported here and the related bisterphenyl complexes reported elsewhere[6] are jointly considered, a clear trend for a trans influence affecting the MoCarene distances is revealed. Thus, the weakening effect on the MoCarene distance increases depending on the nature of the trans donor atom, from weaker to stronger: O (2.552.57 Å) ~ I (2.57 Å) < C (2.642.66 Å in 7 and 4·CNXyl) ~ P (2.642.69 Å) < Caryl (2.762.80 Å). Regarding the MoCarene interaction, before discussing our results it is pertinent to cite some literature precedents. In Power’s Ar’CrCrAr’ molecules,[4] the CrCarene distance of 2.29 Å was found to be ca. 7.5% longer than the CrCaryl bond (2.13 Å). On the basis of quantum mechanical calculations the CrCarene interaction was viewed as a feeble one that causes only a small weakening of the quintuple bond.[5c] Besides, an also feeble CrCarene interaction was recently disclosed in the cationic bimetallic complex (OC)4Cr(-CPh2)Au(PCy3)+, where there is a CrCarene contact of 2.28 Å to the ipso carbon atom of one of the carbene phenyl substituents.[19] This contact is 7% longer than the CrCcarbene bond and whereas it results in important structural consequences, in accordance with DFT calculations it does not imply significant bonding interaction between the ipso carbon and chromium atoms.[19] It must be said, however, that these CrCarene distances are similar to those found in -bonded arene complexes, with an average value of 2.24(6) Å for structures in the CSD. In the new complexes 4·L7, as well as in the precursor compounds 13, the terphenyl-bound molybdenum atom Mo1 displays a distorted square-pyramid coordination geometry with the other metal atom, Mo2, at the apex. Disregarding for the time 4·PMe3 4·PPri3 4·P(OMe)3 Figure 5. X-ray molecular structures of compound [Mo2(ArXyl2)(I)(O2CMe)2(L)] with P-donor ligand. 7 DOI: 10.1002/chem.201xxxxxx being the secondary MoCarene interaction, Mo2 has a four coordinated structure also derived from a square-pyramid but with an empty basal site. The molecular geometry of the terphenyl ligand places one of the flanking aryl rings in the proximity of this empty coordination site so that for each of the compounds investigated either its ipso or one of the ortho carbon atoms (or both, vide infra) gets closer to Mo2 than the other Carene atoms. For the mono-terphenyl complexes Mo2(Ar’)(O2CR)3,[6] and also for compounds 4·OEt2 and 5·PMe3, the shortest MoCarene contact to one of the ortho carbon atoms has a distance of ca. 2.58 Å while in all other complexes 4·L6·L, this distance is longer and has an average value of 2.64 Å. In some of these complexes, e.g. 4·PMe3, there is another only slightly longer contact to the ipso carbon at ca. 2.71 Å, so that the interaction between the flanking arene and the Mo atom may be viewed as approaching ƞ2. It is pertinent to note that complexes 4·PMe3 and 4·P(OMe)3, that possess similar structure but P-donor ligands of different electron donor properties (TEP of 2064 and 2080 cm-1, respectively)[20] exhibit MoCarene interactions of identical length (ca. 2.64 Å). In the bis-terphenyl complex Mo2(ArXyl2)2(O2CH)2, also included in Table 1, d(MoCarene) increases to 2.78 Å.[6] Figure 6. X-ray molecular structure of [Mo2(ArXyl2)(O2CMe)2(CN2C2Me4)2]I, 7. Comparison of MoCaryl and MoCarene distances in these complexes reveals that the interaction with the flanking aryl is characterized by a MoCarene distance that is 1520% longer than the  MoCaryl bond. Furthermore, the MoCarene contacts are also significantly longer than MoCarene bonds in classical mononuclear Mo(II) arenes (e.g. 2.262.35 Å in MoMe2(6- C6H6)(PMe2Ph)2).[21] Therefore, X-ray data suggest that in terms of electron density sharing, the interactions between Mo2 and Carene are weak. Although for the complexes with the longest Mo2Carene separations one could think that the proximal flanking aryl ring has mostly a protective role and acts as steric guard for the low-coordinate Mo2 atom, a detailed analysis of this interesting structural and bonding problem by means of theoretical calculations presented in the following section, as well as a parallel study on aryls coordinated in a :1-mode to platinum atoms, point to the two-electron donor role played by such rings.[22] Computational studies Arene coordination to MoII2 fragments. Modelization of [Mo2(ArXyl2)(O2CMe)3][6b] as a complex with an independent benzene ring coordinated in a monohapto mode in [Mo2(H)(O2CMe)3(1-C6H6)] should allow us to get an estimate of the bonding energy between the side phenyl group of the terphenyl ligand and the Mo2 unit. Figure 7 presents the potential energy curve for the dissociation of an independent benzene ring to the Mo2(H)(O2CMe)3 fragment. A fitting of the calculated energies to a Morse potential yields MoC = 2.645 Å and a dissociation enthalpy of 9.0 kcal/mol, while full optimization gives practically the same distance (2.644 Å), in excellent agreement with that found experimentally for the anchored phenyl side groups, and a free energy of dissociation of 13.3 kcal/mol at room temperature. Figure 7. Potential energy of [Mo2(H)(O2CMe)3(1-C6H6)] as a function of the shortest MoC(benzene) distance, relative to that of the optimized geometry. In the subsequent discussion, we will often need to find a systematic way of classifying the hapticity of a coordinated arene as ,  or . While some intermediate situations may exist, in general plotting two structural parameters in a "hapticity map", as discussed by us elsewhere, has been found useful to discriminate the low hapticities. In brief, our proposal consists in comparing the three shortest MCarene distances (d1 < d2 < d3) by means of the ratios defined in equations 1 and 2. A scatterplot of those ratios in a hapticity map has three regions that can be approximately associated with ,  or  coordination modes, although with somewhat imprecise borderline regions (Figure 8).   = d2 / d1 [1] = d3 / d1 [2] If the MoC arene distance ratios obtained for the models are plotted in such a hapticity map, the benzene ring of [Mo2(H)(O2CMe)3(C6H6)] can be clearly classifed as 2- 8 DOI: 10.1002/chem.201xxxxxx coordinated, at variance with the anchored aryl groups in the experimental structures, that place themselves in regions closer to the 1-3 border (Figure 8). These results indicate that the 2- coordination is the preferred one for an independent benzene ring, and it seems reasonable to assume that the anchoring of the phenyl groups in the terphenyl complexes can force the slightly less favorable 1 or 3 coordination modes, most clearly seen in 2, 4·OEt2, 4·P(OMe)3 and Mo2(ArXyl2)2(O2CH)2. To verify this assumption we discuss next calculations on more realistic models. Figure 8. Hapticity map that represents the ratios of the three shortest MC distances defined in equations 1 and 2, with limits of the -, - and - regions set arbitrarily, in which the flanking arene rings of calculated [Mo2(H)(O2CMe)3(-C6H6)] and [Mo2(R-Ph-R')(O2CMe)3] complexes (R = H, Ph, Xyl; R' = Ph, Xyl) (circles) and experimental (squares) structures are represented. Optimization of anchored analogues [Mo2(-C6H4-- R)(O2CMe)3] (R = Ph, Xyl), in which the hydride and benzene ligands have been replaced by a biaryl group results in coordination of the flanking arene through its ortho carbon atom, as in the experimental structures, and at the same MoC distance (2.570 and 2.571 Å for the calculated and experimental structures, respectively). Unlike the experimental structures, and in spite of the excellent agreement in the MoC bond distance, this calculated compound is unequivocally identified as a monohapto species and has also a much larger MoC···C flap angle  (see structure A) than the experimental structure (99 and 7884º, respectively). Searching for the cause of the different hapticity and flap angle in our model biphenyl compounds [Mo2(-C6H4-- R)(O2CMe)3], we have performed geometry optimizations for the series of terphenyl compounds [Mo2(RC6H3R')(O2CMe)3], where R = Ph, Xyl; R' = Ph, Xyl. Substituting the biphenyl by a terphenyl group (R = R' = Ph) does not significantly affect the hapticity of the side phenyl ring ( = 1.14,  = 1.19). Introduction of methyl groups at the ortho positions of the distal aryl group (R = Xyl, R' = Ph) does not change the situation either ( = 1.12,  = 1.20, and flap angle of 97º). It is the presence of the ortho methyl groups in the -coordinated arene ring (i.e., R = Ph, Xyl; R' =Xyl) that places that phenyl ring close to the  region where the experimental structures are (= 1.08,  = 1.12) and puts down the phenyl flap (MC···C angle of 86º), with a simultaneous increase of the Mo···C distance. It must also be noted that the isomeric compound (R = Xyl, R' = Ph) is slightly more stable in our calculations (1.3 kcal/mol) than the alternative isomer (R = Ph, R' = Xyl). It thus seems that the attachment of the flanking arene group to the -bonded ring prevents it from achieving an  coordination and favours the  mode, but the steric bulk introduced by the ortho methyl groups forces a somewhat longer MoC distance with the corresponding weakening of the  interaction, which is in part compensated by a shift towards an  mode via a smaller flap angle. It is worth mentioning in passing that the flanking arene rings in the quintuply bonded Ar’CrCrAr’ complexes reported by Power and co-workers[4] occupy the same region in the hapticity map (not shown in Figure 8 for simplicity) than the Mo2 compounds presented here, and have similar flap angles as well. Figure 9. An occupied  bonding MO of the flanking arene ring showing a bonding contribution from the Mo atom in [Mo2(terphenyl)(O2CMe)3]. 9 DOI: 10.1002/chem.201xxxxxx The existence of a bonding interaction between the  system of the flanking arene and the Mo atom shows up in the significant mixing with a molybdenum atomic orbital (Figure 9). In addition, a topological analysis of the electron density shows a bond critical point between the metal atom and the 1-coordinated carbon atom of the arene ring in the biaryl- and terphenyl complexes, and two MC bond critical points for the 2-coordinated benzene. An interesting finding is the existence of a bond critical point between the non-bonded arene ring R and the Mo atom -bonded to the central phenyl group in the [Mo2(RC6H3Ph)(O2CMe)3] complexes, at quite long Mo···C distances (3.12 and 3.17 Å for R = Ph and Xyl, respectively) along the MoMo axis. When the flanking arene ring is a xylyl, that distance is slightly longer (3.23 and 3.30 Å for R = Ph and Xyl, respectively), and the bond critical point vanishes, indicating a steric avoidance of such a weak non-bonding interaction. As argued below such an incipient interaction may be instrumental in facilitating the 1,2-shift fluxional process. Wiberg calculated bond indices of the order of 0.10.2 for the MCarene contact in the 1-coordinated species are consistent with the existence of bonding. Moreover, bond orders of around 0.050.10 with the two neighboring carbon atoms indicate that the electron pair involved in the donation towards the metal atom is not fully localized and that electronically the coordination has a small 3-character, regardless of the geometrical hapticity discussed above. For the 2-bonded species the Wiberg bond indices also have similar values for the two MC bonds (between 0.1 and 0.2). Substitution reactions on [Mo2(ArXyl2)(I)(O2CMe)2(L)]. An intriguing geometrical feature of this family of compounds is that the starting complex (L = Et2O) has the iodide ligand coordinated in trans to the same Mo atom than the -bonded phenyl ring, whereas substitution of the ether by ligands such as PMe3, P(OMe)3, PPri3, and CNXyl results in the alternative isomer with the iodide trans to the :1 coordinated phenyl ring. Our DFT calculations on the [Mo2(ArXyl2)(I)(O2CMe)2(L)] complexes (L = Et2O, PMe3) (Figure 10) consistently indicate the experimental isomer to be the most stable one. Moreover, the dissociation of Et2O is calculated to cost barely 0.4 kcal/mol, and to yield an intermediate with a bridging iodide. Since the entering ligand can in principle react with either Mo atom, it seems reasonable to assume that the steric protection provided by the unbonded xylyl Figure 10. Above: Relative free energies at 298 K of the two isomers of [Mo2(ArXyl2)(I)(O2CMe)2(L)] (L = Et2O, PMe3), a dissociated intermediate [Mo2(ArXyl2)(- I)(O2CMe)2], and two transition states for the ligand substitution reaction. The acetate groups are omitted for clarity. Distances are given in Å. Below: Calculated molecular structures of the four low-energy species found along the substitution pathway.