Modulation of σ ‑ Alkane Interactions in [Rh(L 2 )(alkane)] + Solid-State Molecular Organometallic (SMOM) Systems by Variation of the Chelating Phosphine and Alkane: Access to η 2 , η 2 ‑ σ -Alkane Rh(I), η 1 ‑ σ -Alkane Rh(III) Complexes, and Alkane Encapsulation

: Solid/gas single-crystal to single-crystal (SC − SC) hydrogenation of appropriate diene precursors forms the corresponding σ -alkane complexes [Rh(Cy 2 P(CH 2 ) n PCy 2 )(L)]-[BAr F4 ] ( n = 3, 4) and [RhH(Cy 2 P(CH 2 ) 2 (CH)-(CH 2 ) 2 PCy 2 )(L)][BAr F4 ] ( n = 5, L = norbornane, NBA; cyclooctane, COA). Their structures, as determined by single-crystal X-ray di ﬀ raction, have cations exhibiting Rh ··· H − C σ -interactions which are modulated by both the chelating ligand and the identity of the alkane, while all sit in an octahedral anion microenvironment. These range from chelating η 2 , η 2 Rh ··· interactions.


INTRODUCTION
The ability to tune the local environment around a metal center by variation of supporting ligands is an important concept widely used in homogeneous organometallic synthesis and catalysis. 1,2 A well-documented example of this comes from bidentate phosphine ML 2 -type complexes, 3,4 as by altering the L−M−L bite-angle and substitution at phosphine the resulting steric (e.g., the solid-cone angle, Θ 5 ), or electronic (e.g., oxidation state 6 or degree of bond activation 7 ), changes ultimately can provide the ability to control structure, speciation, and, through the energetics of elementary reaction steps in catalysis, activity and selectivity. Figure 1A.
Extending such concepts to heterogeneous systems is difficult given the resulting challenges associated with precisely defining single-site active centers and their extended coordination environments. 8 While chelating phosphine complexes supported by metal−organic frameworks, 9 porous coordination polymers, 10 nanoparticles, 11 mesoporous hosts, 12 and silica surfaces 13 have been reported, the role of the chelating ligand in determining structure and reactivity is less well-developed. Nevertheless, precise control of well-defined, and reactive, metal centers in heterogeneous systems could lead to enhanced activity and selectivity in catalysis, as frequently demonstrated in homogeneous processes. 1,4 We have recently shown that single-crystal to single-crystal (SC−SC) solid/gas reactions between H 2 and the appropriate [RhL 2 (diene)] + precursors form well-defined but reactive σ-alkane 14−16 complexes directly in the solid-state, e.g., [Rh(Cy 2 P(CH 2 ) 2 PCy 2 )(η 2 η 2 -NBA)][BAr F 4 ], [BAr F 4 ] (NBA = norbornane, Ar F =3 , 5 -( C F 3 ) 2 C 6 H 3 ), Figure 1B). 17−20 Such σ-complexes contain 3-center 2-electron (3c−2e) Rh···H−C bonds 21,22 and are of general interest from the fundamental challenges presented by their synthesis and characterization, 14,15 as well as their central role as intermediates in C−H activation processes. 23−26 When prepared in this way, these σ-alkane complexes show remarkable relative stability compared with species prepared by solution routes; the latter are generally characterized in situ using NMR spectroscopy, on a small scale (2−20 mg) at very low temperature, and have limited lifetimes even under these relatively constrained conditions. 27−31 This stability in the solid-state originates from the [BAr F 4 ] − anions providing a robust, octahedral, crystalline microenvironment 32 that allows for isolation, characterization, and onward reactivity of the encapsulated organometallic cation to be studied in detail ( Figure 1C). 26,33 These so-called 33 solid-state molecular organometallic (SMOM) systems are related to supported organometallic catalysts (SOMC), 13 single-site heterogeneous catalysts (SSHC), 34 and MOF-functionalized organometallics 9,35−37 but, in contrast to these, are not supported by a platform material. Moreover, SMOM systems have the desirable properties of being readily studied at the molecular-level by singlecrystal X-ray diffraction, solid-state NMR (SSNMR) spectroscopy, and computational techniques such as periodic DFT.
We now report that systematic variation of the P−Rh−P biteangle coupled with the identity of the diene in SMOM systems based upon precursors [Rh(Cy 2 P(CH 2 ) n PCy 2 )(diene)][BAr F 4 ] (n = 3 to 5, diene = norbornadiene, NBD, or 1,5-cyclooctadiene, COD, Scheme 1) results in significant changes in structure and reactivity on addition of H 2 in SC−SC reactions. This results in crystallographically characterized σ-alkane complexes that show markedly different degrees of Rh···H−C interaction in response to the changes in both phosphine and the precursor diene, while they are stabilized in the microenvironment provided by the octahedral arrangement of [BAr F 4 ] − anions: these range from chelating η 2 ,η 2 Rh···H−C, through to more weakly bound η 1 Rh··H−C and ultimately to systems where the alkane is not ligated with the metal center but sits encapsulated in the anion framework.  18 These precursor complexes have been characterized by solution NMR spectroscopy and single-crystal X-ray diffraction, the latter which shows an O h arrangement of [BAr F 4 ] − anions    17,20 The Supporting Information details their structures. For the NBD precursors, this homologous series allows for the bite-angle (β) of the various diphosphines in this environment to be compared. Unsurprisingly, 4 β becomes progressively larger with an increasing number of methylene units in the chelate backbone (Scheme 2B). The same trend, albeit interestingly with slightly smaller β-angles, is apparent for the COD precursors. With these complexes in hand a systematic study of solid/gas hydrogenation was undertaken.
The solid-state structure of the cation  The relatively long Rh···C distances, coupled with more open ∠RhHC angles, e.g., Rh1−H9A−C9 147.9(6)°, albeit with the hydrogens placed in calculated positions, suggests an η 1 Rh···H−C bonding motif for the agostic interactions, a view supported by calculations (see Section 2.7). This movement of the metal center keeps the phosphine ligand in essentially the same environment in the anion cavity and maximizes the Rh···H−C agostic interactions, with two cyclohexyl groups also enfolding the metal center. 51 It also retains the square planar geometry around Rh, with angles around the metal center = 360°. A minor, disordered, agostomeric 52 component is also present (∼10%) in which the Rh center swings up to interact with the chemically equivalent C15−H and C21−H bonds ( Figure 3C). The Rh−P distances in the major component [2.166 (2)  , or is a consequence of the metal's unusual disposition, is not clear. Although formally 2-coordinate group 9 complexes are known, 53,54 albeit rare, as far as we are aware the cation in [2][COA⊂BAr F 4 ] is the first example reported with a chelating phosphine.
The expelled alkane, COA, sits encapsulated 55,56 within the anion framework ( Figure 3D), essentially equally disordered between two boat−chair conformations. 57 The closest Rh···C distance is 3.74(1) Å (C28), approaching the Rh center above the square plane, suggesting no significant, and at best very weak, interaction with the metal center. This distance is similar to that measured in [U(ArO) 3 (neo-hexane)] [3.731(8) Å], albeit U versus Rh] where a bonding interaction was suggested, 58 although studies on related uranium complexes suggest that such as distance reflects a nonbonding interaction. 59 In the 158 K 31 P{ 1 H} SSNMR spectrum of [2][COA⊂BAr F 4 ] only two broad (fwhm ∼400 Hz) environments are observed at δ 51.3 and δ 48.1 in an approximate 1:1 ratio. Coupling to 103 Rh was not resolved. These do not vary significantly in chemical shift or relative intensity when measured at 298 K. The observation of only two signals suggests that the disorder observed in the cation could be static (with coincident signals) or dynamic 60 (and fast 61 ) on the NMR time scale, but the observation that no change is observed on changing the temperature suggests the former. 13 C{ 1 H} SSNMR spectrum shows a featureless region between δ 115 and 43, indicating that the COD has been hydrogenated to COA, which is observed as a sharp signal at δ 25.7. This assignment is supported by a nonquaternary suppression experiment (NQS) which, as well as detecting quaternary carbons, identifies CH n groups that experience motion in a frequency range similar to, or greater than, the 1 H− 13 C dipolar coupling. 62,63 In the 158 K NQS spectrum a single prominent signal remains at δ 25.7 in the alkyl region, compared with the 13 C{ 1 H} SSNMR spectrum ( Figure 4A). This peak remains at 298 K. 39 These data are consistent with the encapsulated COA undergoing a low-energy site-exchange within the cavity ( Figure 4B), suggested to be due to 1,2-jumps and/or exchange between the two disordered COA components. 64 Figure 5, shows that the latter presents a larger steric profile for the metal center. We suggest that this, alongside possible conformational prefere n c e sf o rm e t a lb i n d i n go ft h ea l k a n ea n dn o n c o v a l e n tF ···H−C interactions in the microenvironment (see Figures S73 and S76), are drivers for the different structural motifs observed.   Increasing the bite-angle of the phosphine promoted different reactivity in the resulting σ-alkane complex. Surprisingly given that the structural metrics have not changed significantly from the smaller bite-angle congeners, [BAr F 4 ] is not stable when exposed to a moderate vacuum for 3 days. Crystallinity is lost, and SSNMR spectroscopy shows the formation of multiple species, as yet unidentified. Thus, although the binding of the NBA ligand appears to not be influenced significantly by the increase in bite-angle of the chelating phosphine, in the measured ground-state structure steric pressure and/or enhanced stability of any decomposition products as driven by the change in phosphine appear to promote reactivity toward loss of NBA. Hydrogenation of single-crystals of [3-COD][BAr F 4 ] in a solid/gas reaction resulted in loss of crystallinity. 39 We have not characterized the product of this further. 2.6. {Rh(Cy 2 P(CH 2 ) 5 PCy 2 )} + /COA and NBA: Phosphine Ligand Backbone C−H Activation, Structural Reorganization with Retention of Crystallinity, and Rh(III) η 1 -σ-Alkane Complexes. Addition of H 2 to crystalline [BAr F 4 ] resulted in a rapid (∼5 min as measured by 31 P{ 1 H} SSNMR spectroscopy) SC−SC reaction. Analysis of the product formed using single-crystal X-ray diffraction was hampered by long-range disorder, which is also present in the starting material. The structure was modeled using a supercell (Z′ = 2) which gave a satisfactory solution (R = 15.7%) that allowed for the gross structure of the cation to be determined, Figure 7, but does not allow for detailed metrics to be discussed.
Hydrogen atoms were not located, and the NBA fragments formed by hydrogenation were necessarily modeled as rigid bodies. There are two chemically very similar, but crystallographically independent, cations in this supercell in which each has a disordered NBA over two conformations ( Figure 7C). There is no crystallographically imposed local symmetry. The O h arrangement of anions in relation to each cation is retained (Figures S83−S86).
Despite the challenges associated with structural identification, it is immediately apparent that the phosphine pentamethylene backbone has undergone a C−H activation in the solidstate to form a wide bite-angle trans-spanning phosphine PCP pincer complex. Such intramolecular C−H activation with an R 2 P(CH 2 ) 5 PR 2 ligand has precedent in solution studies, either to form a hydrido−alkyl complex 66−68 or through a further α-elimination to give a diphosphino-carbene complex.

Journal of the American Chemical Society
Article These distances reflect weak (at best) σ-interactions compared to, e.g., [BAr F 4 ], 49 while this spread suggests that the NBA fragment finds a better spatial fit with the {RhPCP} + fragment for some conformations over others. The data, clearly, do not allow the precise binding mode of the alkane (η 1 -HC, η 3 -H 2 C 70 ) to be determined. A related SC−SC N−C oxidative addition at a Rh(I) center has been reported in Rh−PNP pincer complexes. 71 (Figures S49  and S50). The 13 C NQS spectrum at 158 K shows at least four signals grouped between δ 38−35 and δ 29−27, in the region associated with aliphatic C−H groups, that indicate a lowenergy molecular motion of the NBA ligand within the cavity of the cage ( Figure S54). Such a low-energy process is consistent with the disorder of the NBA fragment modeled in the solidstate. In the 158 K FSLG HETCOR spectrum, cross peaks between these aliphatic signals in the 13 C SSNMR spectrum and low-field peaks in the 1 Hprojection(δ −1.8 to −2.6) are observed. We assign these to the Rh···H−C interactions, although we cannot discount ring-current effects from the proximal Ar F groups causing such a high-field shift in other C−H bonds. 18 Figure 8B, does not suffer from disorder and clearly shows the trans-spanning PCP pincer motif suggested for [4-NBA][BAr F 4 ]. The Rh−hydride (H1) was located, sitting trans to a vacant site, and anti to the remaining hydrogen associated with the C−Ha c t i v a t e d methylene group (H3). This stereochemistry is as expected for an intramolecular C−H activation. 66,68 ACH 2 Cl 2 molecule has displaced the labile NBA fragment. Displacement of a weakly bound σ-alkane ligand by halogenated solvent is wellestablished, 18,28,30 and the structure is consistent with other crystallographically characterized Rh−ClCH 2 Cl complexes. 49 Figure S105, Supporting Information). 69 It is also likely that the bound CH 2 Cl 2 molecule is undergoing rapid exchange at the metal center with the solvent. 77 If [4-COD][BAr F 4 ] is subjected to H 2 in the solid-state, after 30 min crystallinity is lost. However, if the reaction is stopped after only 10 min and the resulting single crystals are quickly transferred to an X-ray diffractometer and cooled to 150 K, the resulting analysis shows that a new complex is formed in 30% yield in a SC−SC process, with the remainder being unreacted The molecular cation of [4-COA] + is shown in Figure 9. This shows a C−H activated, trans-spanning, diphosphine "PCP" pincer ligand with a cyclooctane located in close proximity to the Rh ( 79 respectively. However, as far as we are aware this is the first structurally determined example of an alkane interacting η 1 with a transition metal center. Such motifs have been probed on the picosecond time scale using fast timeresolved infrared spectroscopy (TRIR) combined with DFT calculations, and these are early intermediates in the oxidative addition of alkanes to coordinatively unsaturated metal centers. 80 is continued for a total of 30 min, decomposition to an, as yet unidentified, product(s) is observed upon dissolution in CD 2 Cl 2 , from which a featureless 31 P{ 1 H} NMR spectrum and a very broad 1 H NMR spectrum are observed. We speculate that this signals the formation of a paramagnetic Rh(II) dimeric complex on dissolution, 85 but the identity of this species remains to be resolved. Attempts to obtain meaningful SSNMR data for [BAr F 4 ] were hampered by temporal and temperature sensitivity that was amplified by the requirement to use finely crushed material for analysis by SSNMR that meant that crystallinity is lost much faster than for larger samples. In the 31  ] are shown to have very similar structures, with the former's stability allowing for a more detailed characterization by SSNMR spectroscopy and the latter providing a good structural solution by X-ray crystallography. Combined, they thus provide a convincing analysis as a weakly bound σ-alkane complex bound η 1 at a Rh(III) center. We believe they are the first Rh(III) σ-alkane complexes isolated, or observed, by any method.

Journal of the American Chemical Society
Article Given the variations between computed and experimental structures across the range of systems under consideration, the subsequent analyses were based on computed structures in which the Rh, C, B, and P positions were taken from the crystallographic studies and the H and F atoms were optimized with the PBE-D3 approach. Selected distances involving key H atom positions optimized on this basis are shown in Figure 10.
A more quantitative analysis of bonding was then provided by quantum theory of atoms in molecules (QTAIM) and natural bond orbital (NBO) second order perturbation donor− acceptor interaction analyses performed on the isolated cations shown in Figure 10 Figure 11D shows the NCI plot of the [2-NBA][BAr F 4 ] ion pair and reveals a broad curved feature running roughly parallel to the H11−C1−C2−H21 bonds. This reflects the chelating nature of the NBA ligand and is predominantly stabilizing (blue) in character, while also exhibiting a central destabilizing (orange/red) region that is consistent with the presence of the ring critical point (RCP) in the QTAIM study. Some destabilizing character is also seen between Rh and the center of the two C−H bonds (see detail in Figure 11E). This suggests two cyclic {RhCH} features in the electron density topology that are consistent with an η 2 -interaction and the significant contribution of classical Rh(dπ)t oσ* C−H π-back-donation identified in the NBO analysis. This also highlights how the NCI approach can amplify the insight gained from the local QTAIM critical points. 87 F 4 ], and a side-by-side comparison is provided in Figure S100 in the Supporting Information. F 4 ]. QTAIM and NBO data for the [2] + cation are displayed in Figure 12A−C. Rh−H3B and Rh−H9A bond paths signal the presence of intramolecular agostic interactions. The associated Rh···H contacts are longer (ca. 1.94 Å), and the corresponding C−H distances shorter (ca. 1.14 Å), than in the [X-NBA] + series, suggesting weaker interactions in [2] + . This is confirmed by reduced ρ(r) values at the Rh−H BCPs and lower σ C−H to trans-σ* Rh−P σ-donation via NBO. NBO, however, also suggests a similar degree of back-donation as [2-NBA] + , although this is not classical Rh(dπ)toσ* C−H π-back-donation but rather involves contributions from both the cis-and transσ Rh−P bonds (see also the discussion of [4-COA] + below). This suggests an η 1 -C−H → Rh interaction and is supported by the NCI plot which highlights these stabilizing agostic interactions with well-defined, localized blue disks ( Figure 12D,E).

[2][COA⊂BAr
The closest Rh···COA contact in [2][COA⊂BAr F 4 ] is via H21 with a computed distance of 2.844 Å. This is well within the sum of the van der Waals radii of Rh and H (3.64 Å), 65 and a weak BCP is computed between these centers (ρ(r)=0.011eÅ −3 ). 91 No equivalent donor−acceptor interaction is computed with NBO, although the NCI plot does suggest a weak, stabilizing feature between Rh and H21 (see Figure 12D,E). This is part of a broad area of weakly stabilizing interactions between the COA and the [2] + cation, suggesting that any direct covalent Rh···H21 interaction is at best very weak, if it exists at all. The COA is further stabilized within the cavity by dispersive interactions with the two proximate aryl substituents of the [BAr F 4 ] − anion. 92    Figure 10). These, along with the computed Rh···H11 BCP metrics (Figure 13), suggest a somewhat weaker σ-interaction than in the [2] + cation, although NBO indicates a similar degree of σ-donation and, if anything, greater back-donation in this case. A blue/green disk in the NCI plot confirms this η 1 C1−H11 → Rh σ-interaction as well as highlights broad areas of stabilizing dispersion interactions with the cyclohexyl substituents and the [BAr F 4 ] − aryl groups. An interesting aspect of the NBO analysis of both agostic [2] + and [4-COA] + is the degree of donation into σ* C−H from the trans-σ Rh-L bonding orbitals (L = P2 or C32, respectively). In [4-COA] + , this is supported by donation from the cis-σ Rh−H2 bonding orbital (see Figure 14). These interactions reflect the η 1 -orientation of the C−H bonds in these species ([2] + , ∠RhHC calc (ave) = 138°; [4-COA] + , ∠RhHC calc =1 3 3°). Similar σ-donation has been identified with the onset of the C−H···Rh ("pregostic") interaction 93,94 and is consistent with the end-on approach of CH 4 in Burgi−Dunitz trajectories 95 and of H 2 in oxidative addition reactions. 96,97 3. CONCLUSIONS The studies reported herein provide a demonstration of the power of solid-state molecular organometallic chemistry (SMOM-Chem), and in particular single-crystal to single-crystal transformations to provide access to and, when combined with periodic DFT calculations, characterize a wide range of different σ-alkane M···H−C coordination motifs by systematic variation of the ligand set. Figure 15 highlights these, alongside selected structural and computationally determined bonding parameters for the corresponding Rh···alkane interaction. The complexes    Journal of the American Chemical Society Article described present snapshots along a continuum of M···H−C interactions that would be very difficult to probe in σ-alkane complexes using solution techniques due to the instability of such systems combined with fluxionality between different M···H−C bonds that often results in time-averaged structures in solution, even at very low tempertaures. 14 Thus, with the NBA alkane ligand and simple, chelating phosphines, relatively strong bidentate chelating η 2 η 2 motifs are observed, e.g., [2-NBA]-[BAr F 4 ]. With a trans-spanning C−H activated PCP pincer ligand, reduced access to the metal center results in unprecedented η 1 Rh(III)···H−C motifs being observed in the solid-state, e.g., [BAr F 4 ], a model for the early stages of C−H activation at metal centers. Finally, when the steric requirement of the phosphine and the alkane combine with the ability for the phosphine to engage in stabilizing intramolecular agostic interactions, the alkane is expelled from the metal center but remarkably stays encapsulated within the anion framework, providing baseline experimental structural data for a close, but essentially nonbonding, approach of an alkane with a metal center: viz. [2][COA⊂BAr F 4 ]. Underpinning these remarkable structures is the stabilizing effect of the anion microenvironment, and in particular the role that intermolecular dispersion interactions play in stabilizing these alkane ligands within the binding pocket. Reflecting this, none of the complexes reported are stable in solution even at low temperature. The SMOM technique thus complements elegant low-temperature in situ solution techniques for the synthesis and characterization of σ-alkane complexes. 29,31,81 That such wide variations of alkane binding modes and structures are observed while a well-defined molecular environment is maintained in the solid-state provides an exemplar of a potentially tunable, molecular heterogeneous system that can be precisely characterized. By using a core metal−ligand fragment, i.e., {Rh(diphosphine)} + , it also offers a wide range of potential opportunities for transformations where variation of metal− ligand interactions is likely to influence rate, stability, and selectivity in catalysis. 98 It will be interesting to see if this can be translated to productive C−H activation reactions of hydrocarbons using SMOM systems, and our efforts are currently focused in this direction.