Poly(imidazolyliden-yl)borato Complexes of Tungsten: Mapping Steric vs. Electronic Features of Facially Coordinating Ligands

A convenient synthesis of [HB(HImMe)3](PF6)2 (ImMe = N-methylimidazolyl) is decribed. This salt serves in situ as a precursor to the tris(imidazolylidenyl)borate Li[HB(ImMe)3] pro-ligand upon deprotonation with nBuLi. Reaction with [W(≡CC6H4Me-4)(CO)2(pic)2(Br)] (pic = 4-picoline) affords the carbyne complex [W(≡CC6H4Me-4)(CO)2{HB(ImMe)3}]. Interrogation of experimental and computational data for this compound allow a ranking of familiar tripodal and facially coordinating ligands according to steric (percentage buried volume) and electronic (νCO) properties. The reaction of [W(≡CC6H4Me-4)(CO)2{HB(ImMe)3}] with [AuCl(SMe2)] affords the heterobimetallic semi-bridging carbyne complex [WAu(μ-CC6H4Me-4)(CO)2(Cl){HB(ImMe)3}].


Introduction
The poly(pyrazolyl)borate class of chelates developed by Trofimenko, colloquially known as 'scorpionates' [1][2][3], have found broad application in diverse of areas of coordination and bioinorganic and organometallic chemistry.Key features that have contributed to their widespread deployment include (i) ease of synthesis; (ii) functionalization at both the bridgehead boron and pyrazolyl rings to provide a range of steric and electronic properties; (iii) kinetic stability of the chelated cage once coordinated to a metal centre; (iv) their so-called 'octahedral enforcer' nature, whereby the topology of the cage especially favours octahedral coordination geometries; and (v) the extension of the principle to the replacement of the pyrazol-1-yl arms with a range of other heterocycles that bridge boron and the metal to which they coordinate.Amongst these, the hydrotris(3,5-dimethylpyrazol-1yl)borate ligand (HB(pzMe 2 ) 3 , Scheme 1) has proven to be especially useful in presenting a moderate degree of steric protection to the remaining three ligands in an octahedral metal complex.

Pro-Ligand Synthesis
Fehlhammer's original synthetic approach (Scheme 2) [7] involved threefold alkylation of potassium hydrotris(imidazol-1-yl)borate with Meerwein's salt [Me 3 O]BF 4 , this latter reagent being the most expensive component.Apart from blazing the original trail, Fehlhammer's approach allows for the installation of various carbene alkyl N-substituents at a late stage on a common late synthetic intermediate.
We have developed an alternative synthesis that borrows from protocols developed for more sterically encumbered examples described by Smith [11][12][13][14].Whilst demonstrating no new principles here, our approach does offer both convenience and economy, employing cheap commercially available reagents (Scheme 3).

Pro-Ligand Synthesis
Fehlhammer's original synthetic approach (Scheme 2) [7] involved threefold alkylation of potassium hydrotris(imidazol-1-yl)borate with Meerwein's salt [Me3O]BF4, this latter reagent being the most expensive component.Apart from blazing the original trail, Fehlhammer's approach allows for the installation of various carbene alkyl N-substituents at a late stage on a common late synthetic intermediate.
We have developed an alternative synthesis that borrows from protocols developed for more sterically encumbered examples described by Smith [11][12][13][14].Whilst demonstrating no new principles here, our approach does offer both convenience and economy, employing cheap commercially available reagents (Scheme 3).
The reaction of [Me3N .BH3] with bromine affords [Me3N .BHBr2] [59], which may be generated in situ without isolation.Subsequent treatment with N-methylimidazole affords the salt [HB(ImMeH)3]Br2 ( [1]Br2).This salt, whilst forming in high yields, is difficult to manipulate as it is exceedingly deliquescent and upon filtration under ambient air rapidly forms a sticky syrup.This behaviour is potentially problematic since the subsequent step calls for deprotonation via strong, moisture-sensitive bases, e.g., n BuLi or KN(SiMe3)2.Metathesis with aqueous Na[PF6], however, results in ready recovery of the hexafluorophosphate salt [HB(ImMeH)3](PF6)2 ([1](PF6)2), which is not hygroscopic and crystallizes free of water as confirmed via a crystallographic analysis (Figure 1).   .This salt, whilst forming in high yields, is difficult to manipulate as it is exceedingly deliquescent and upon filtration under ambient air rapidly forms a sticky syrup.This behaviour is potentially problematic since the subsequent step calls for deprotonation via strong, moisture-sensitive bases, e.g., n BuLi or KN(SiMe 3 ) 2 .Metathesis with aqueous Na[PF 6 ], however, results in ready recovery of the hexafluorophosphate salt [HB(ImMeH) 3 ](PF 6 ) 2 ([1](PF 6 ) 2 ), which is not hygroscopic and crystallizes free of water as confirmed via a crystallographic analysis (Figure 1).

Pro-Ligand Synthesis
Fehlhammer's original synthetic approach (Scheme 2) [7] involved threefold alkylation of potassium hydrotris(imidazol-1-yl)borate with Meerwein's salt [Me3O]BF4, this latter reagent being the most expensive component.Apart from blazing the original trail, Fehlhammer's approach allows for the installation of various carbene alkyl N-substituents at a late stage on a common late synthetic intermediate.
We have developed an alternative synthesis that borrows from protocols developed for more sterically encumbered examples described by Smith [11][12][13][14].Whilst demonstrating no new principles here, our approach does offer both convenience and economy, employing cheap commercially available reagents (Scheme 3).
The reaction of [Me3N .BH3] with bromine affords [Me3N .BHBr2] [59], which may be generated in situ without isolation.Subsequent treatment with N-methylimidazole affords the salt [HB(ImMeH)3]Br2 ([1]Br2).This salt, whilst forming in high yields, is difficult to manipulate as it is exceedingly deliquescent and upon filtration under ambient air rapidly forms a sticky syrup.This behaviour is potentially problematic since the subsequent step calls for deprotonation via strong, moisture-sensitive bases, e.g., n BuLi or KN(SiMe3)2.Metathesis with aqueous Na[PF6], however, results in ready recovery of the hexafluorophosphate salt [HB(ImMeH)3](PF6)2 ([1](PF6)2), which is not hygroscopic and crystallizes free of water as confirmed via a crystallographic analysis (Figure 1).

Ligand Installation
For installation of the pro-ligand on a suitable alkylidyne precursor, the 4-toluidyne complex trans,cis,cis-[W(≡CC 6 H 4 Me-4)(CO) 2 (pic) 2 Br] (pic = 4-picoline) (2a) was chosen to exploit the lability of the bromide and 4-picoline ligands.Whilst this complex has not been previously reported on, its synthesis (Scheme 4) is unremarkable and mirrors that of the known xylyl or mesityl analogues [60][61][62].Synthetic procedures are presented alongside those for the molybdenum analogue (2b) in the Experimental section in addition to a crystallographic analysis.

Ligand Installation
For installation of the pro-ligand on a suitable alkylidyne precursor, the 4-toluidyne complex trans,cis,cis-[W(≡CC6H4Me-4)(CO)2(pic)2Br] (pic = 4-picoline) (2a) was chosen to exploit the lability of the bromide and 4-picoline ligands.Whilst this complex has not been previously reported on, its synthesis (Scheme 4) is unremarkable and mirrors that of the known xylyl or mesityl analogues [60][61][62].Synthetic procedures are presented alongside those for the molybdenum analogue (2b) in the Experimental section in addition to a crystallographic analysis.The pro-ligand salt [1](PF6)2 was dissolved in tetrahydrofuran and cooled (dry ice/propanone) before addition of 3 equivalents of n BuLi, followed by slow warming to room temperature to provide a yellow solution of Li[HB(ImMe)3] (Li [3]) which was recooled and treated with 2a.Re-warming to room temperature resulted in a colour change to dark brown as the infrared absorptions for the starting material (2a: CO = 1986, 1898) were replaced with those of the new product (4: CO = 1958, 1873 cm −1 ).After stirring for 3 h, the product was isolated via column chromatography to yield a bright orange microcrystalline powder.
As 4 is the first tungsten complex of such a ligand, it provides an opportunity to demonstrate the special feature of HB(ImR 1 )3 chelates cf.poly(pyrazolyl)borates; scalar couplings observed in the 13 C{ 1 H} NMR spectra may serve as reporters to interrogate metal-carbon bonding.Thus, whilst the chemical shift and associated coupling for the carbon nuclei trans to the carbonyl ligands are unremarkable (e.g., cf. the conventional NHC complex [W{=C(ND i PP)2C2H2}(CO)5]: δC = 187.9, 1 JWC = 105.7 Hz, DiPP = C6H3 i Pr2-2,6) [63], the resonance for the carbon trans to the carbyne is shifted some 11 ppm to higher The pro-ligand salt [1] (PF 6 ) 2 was dissolved in tetrahydrofuran and cooled (dry ice/propanone) before addition of 3 equivalents of n BuLi, followed by slow warming to room temperature to provide a yellow solution of Li[HB(ImMe) 3 ] (Li [3]) which was re-cooled and treated with 2a.Re-warming to room temperature resulted in a colour change to dark brown as the infrared absorptions for the starting material (2a: ν CO = 1986, 1898) were replaced with those of the new product (4: ν CO = 1958, 1873 cm −1 ).After stirring for 3 h, the product was isolated via column chromatography to yield a bright orange microcrystalline powder.
Spectroscopic data were consistent with the formulation of the desired product [W(≡CC 6 H 4 Me-4)(CO) 2 {HB(ImMe) 3 }] (4).Amongst these, the most conspicuous datum is that for the carbyne resonance in the 13 ] (R 1 = Me, Et) for which 1 J PtC values were not reported [8], and [Rh(CO)(L){X 2 B(ImR) 2 ] (L = CO, PPh 3 , PCy 3 ; X = H, F; R = Ph, Cy) [31], poly(imidazolylidenyl)borates have not previously been coordinated to metal nuclei with usefully spin-active (I = 1  2 ) isotopes.As 4 is the first tungsten complex of such a ligand, it provides an opportunity to demonstrate the special feature of HB(ImR 1 ) 3 chelates cf.poly(pyrazolyl)borates; scalar couplings observed in the 13 C{ 1 H} NMR spectra may serve as reporters to interrogate metal-carbon bonding.Thus, whilst the chemical shift and associated coupling for the carbon nuclei trans to the carbonyl ligands are unremarkable (e.g., cf. the conventional NHC complex [W{=C(ND i PP) 2 C 2 H 2 }(CO) 5 ]: δ C = 187.9, 1 J WC = 105.7 Hz, DiPP = C 6 H 3 i Pr 2 -2,6) [63], the resonance for the carbon trans to the carbyne is shifted some 11 ppm to higher field and displays a dramatically reduced coupling to tungsten-183 (44.7 Hz).These may be taken as indicating a weaker W-C interaction which in turn reflects the pronounced trans influence of the alkylidyne ligand, a feature well-documented in the structural chemistry of alkylidyne complexes ligated via poly(pyrazolyl)borate ligands [58].As to the impact of the HB(ImMe) 3 ligand on the remaining co-ligands, comparison with the known complex [W(≡CC 6 H 4 Me-4)(CO) 2 (Tp*)](5) [64] (Tp* = hydrotris(dimethylpyrazoyl)borate, prepared here from K[Tp*] and 2a, see Experimental) is useful.The carbyne and carbonyl resonances for the Tp* derivative appeared at almost identical frequencies to those of the HB(ImMe) 3 complex [δ C ( 1 J WC /Hz) = 279.2(186.6), 224.0 (166.2)];however, in both cases, the magnitudes of 1 J WC values were significantly larger for 5 than for 4. Insofar as these may be taken as being indicative of the strength of the metal-carbon interaction, it would appear that the NHC donors weaken both the carbyne and carbonyl binding.This is, however, difficult to reconcile with the ν CO -associated infrared data which comprise A 1 and B 1 modes observed at 1958 and 1873 cm −1 in dichloromethane (ATR: 1949, 1867 cm −1 ).These are amongst the lowest observed for neutral complexes of the form [W(≡CC 6 H 4 Me-4)(CO) 2 (L)] where L is one of a range of nominally tripodal facially capping ligands [58,[64][65][66][67].These values are even lower than for the π-donor ligand HB(mt) 3 (1967, 1875 cm −1 ; mt = 2-mercapto-N-methyl-imidazol-1-yl) [67] and Kläui's (η 5 -C 5 H 5 )Co(PO 3 Me 2 ) 3 ligand [68].It would therefore appear that the HB(ImMe) 3 ligand makes the tungsten centre especially electron rich and this may be verified using cyclic voltammetry (Figure 2).For both 4 and 5, sweeping the voltage to ca +2 V reveals two oxidation processes, neither of which appear reversible.Limiting the sweep to ca 1.0 V indicates that the reversibility of first oxidation event increases with increasing sweep rate.For 5, ∆E p increases slightly with increased scan rate from 0.180 (0.1 Vs −1 ) to 0.250 V (0.3 Vs −1 ) suggesting the oxidation is essentially reversible with E 1 2 = 0.34 V (E p,c = 0.43 at 0.1 Vs −1 ).For 4 the dependence of ∆E p on sweep rate is more significant, increasing from 0.170 V at 0.1 Vs −1 (E p,c = 0.33 V) to 0.630 V at 5 Vs −1 (E p,c = 0.64 V) is observed.Thus, fast sweep rates are required to observe a reasonable degree of reversibility, with, however, an almost identical half-wave potential (E 1 2 0.345 V) to that of 5. Chemical oxidation of tris(pyrazolyl)borate carbyne complexes of tungsten is typically accompanied by decarbonylation [65,[69][70][71], which most likely accounts for the poor reversibility at slow sweep rates or higher voltages.

Quantification of Steric and Electronic Features
A popular and time-honoured method for assessing the donor properties of ligands involves their impact on infrared frequencies of carbonyl co-ligands.This is traditionally assayed, in the case of phosphines, using the Tolman electronic parameter ν T , viz. the frequency of the A1 mode of CO vibrations in a host of complexes of the form [Ni(L)(CO)3] [72].Although similar scales may be developed for NHC ligands coordinated to the 'Ni(CO)3' fragment [73][74][75], the toxicity of nickel carbonyl has led to the advent of alternative scales based on the RhCl(CO)2 fragment (average of A1 and B1 modes) as the preferred platform, alongside metrics derived from NMR data for the NHC bound to selenium (=Se, δSe), phenylphosphinidine (=PPh, δP) or PdBr2{C(N i Pr)2C6H4} (δC) fragments [76].These methods are not directly applicable to HnB(ImR 1 )4-n complexes due to their negative charge and chelation.While it would be reasonable to presume that, as with conventional neutral NHC ligands, these will be potent net donors, it would be useful to be able

Quantification of Steric and Electronic Features
A popular and time-honoured method for assessing the donor properties of ligands involves their impact on infrared frequencies of carbonyl co-ligands.This is traditionally assayed, in the case of phosphines, using the Tolman electronic parameter ν T , viz. the fre-quency of the A 1 mode of CO vibrations in a host of complexes of the form [Ni(L)(CO) 3 ] [72].Although similar scales may be developed for NHC ligands coordinated to the 'Ni(CO) 3 ' fragment [73][74][75], the toxicity of nickel carbonyl has led to the advent of alternative scales based on the RhCl(CO) 2 fragment (average of A 1 and B 1 modes) as the preferred platform, alongside metrics derived from NMR data for the NHC bound to selenium (=Se, δ Se ), phenylphosphinidine (=PPh, δ P ) or PdBr 2 {C(N i Pr) 2 C 6 H 4 } (δ C ) fragments [76].These methods are not directly applicable to H n B(ImR 1 ) 4-n complexes due to their negative charge and chelation.While it would be reasonable to presume that, as with conventional neutral NHC ligands, these will be potent net donors, it would be useful to be able to benchmark both the electronic and steric features of poly(imidazolylidenyl)borate ligands against those of more familiar facially capping nominally tridentate (κ 3 , η 5 or η 6 ) ligands, of which there are many.Smith has already suggested such a ranking for a small number of facial/tripodal ligands based on the ν NO stretching frequencies of complexes of the form [Ni(NO)(L)] [37].Such ligands may be grouped according to their charge (neutral, mono-or di-anionic) which in turn impacts the charge of the derived complexes (cationic, neutral or anionic, respectively).In the case of complexes of the form [W(≡CC 6 H 4 Me-4)(CO) 2 (L)] x+ , a number of these have been compared in terms of the experimentally determined infrared data for the cis-dicarbonyl oscillator [67, [77][78][79][80][81][82][83][84][85][86][87][88][89].In addition to the frequencies of the observed symmetric and antisymmetric modes (A 1 ν s(CO) , B 1 ν as(CO) ), the two numbers may be condensed into a singular Cotton-Kraihanzel force constant [90].This is reasonable in the case of [W(≡CR)(CO) 2 (L)] x+ because the two carbonyl ligands are chemically equivalent, i.e., both individual CO oscillators are identical.This is perhaps less appropriate in the 'RhCl(CO) 2 ' system, where in any event the simple arithmetic mean is usually employed.
Our previous collation was based on experimentally determined ν CO values with the caveat that some were acquired from solid-state mesurements (Nujol mulls, KBr discs, ATR, etc.) while others were obtained from a variety of solvents.Infrared data for metal carbonyls are prone to significant perturbation in the solid state due to different crystal modifications or crystallographically independent molecules within the same crystal which in each case place the CO ligand(s) in different environments.The solvent-dependent nature of IR data for metal carbonyls, due to which both the frequency and broadening are significantly impacted by the choice of solvent, has long been recognized [91].Thus, gas phase data, when measurement is viable, typically produce higher frequencies than are found in aliphatic hydrocarbons, and while such solvents provide the sharpest and therefore best-resolved peaks, comparatively few carbonyl complexes are sufficiently soluble.Dichloromethane has therefore become the solvent of choice offering the most accommodating solubility characteristics and reasonably narrow peaks.
To obviate these imponderables, we have collated infrared data for a range of complexes [W(≡CC 6 H 4 Me-4)(CO) 2 (L)] x+ derived from computational interrogation (Table 1).Our intention is not to provide the most precise current state-of-the-art investigation of the intimate bonding and thermodynamic properties of such complexes but rather to derive a readily accessible and computationally economic comparative scale.A useful corollary of this approach is that the optimised geometries used for frequency calculations may be employed to directly calculate the percentage buried volume (%V bur ) [92,93] of each ligand L. The %V bur approach to quantifying the steric impact of a ligand is especially suitable for ligands with irregular topologies, and for phosphines, such analysis reassuringly returns a correlation approximately linear with Tolman's cone angle (θ T = 3.95x%V bur + 31.5)[94].Accordingly, a scatter plot of the Cotton-Kraihanzel force constant k CO vs. %V bur (Figure 3) may be presented for ligands L that is reminiscent of the familiar ν T vs. θ T plot used to map phosphine electronic and steric space [72].For this purpose, with this combination of density functional, basis set and anharmonic scaling factor the value of the Cotton-Kraihanzel force constant reduces to the following equation:   The ωB97X-D [95,96] functional was employed with the 6-31G* basis set [97] in combination with the LANL2Dζ effective core potential for tungsten [98][99][100], and while much more sophisticated levels of theory are certainly available, this selection represents a balance between utility and computational economy for these medium-sized molecules.For larger ligands 'L', where steric bulk has or might be an intentional design feature, %V bur values obtained at the simpler semi-empirical PM3tm level of theory are used, as we are here only concerned with molecular topologies (Figure 4).Taking complexes of the ligands HB(pzMe 2 ) 3 , HB(ImMe) 3 and MeC(CH 2 PMe 2 ) 3 as test cases, the variation in %V bur calculated between ωB97X-D/6-31G*/LANL2Dζ and PM3tm methods was <3%, i.e., within the magnitude of molecular libration.Vibrational frequencies, whilst calculated to ensure local minima had been located, were imprecise at the PM3tm level and considered of little use.Accordingly, the ordinate location of such ligands in Figure 3 (shown in green) should be viewed with considerable caution.These were derived with little rigour by simply scaling the PM3tm k CO values by 0.9089, this being the ratio of k CO values calculated at the PM3tm and DFT levels of theory for 4 and 5.That said, the peripheral inclusion of sterically obtrusive substituents in ligands often results in rather limited transmission of inductive electronic effects to the metal centre itself, as seen, for example, in experimental data for L = η 5 -C 5 H 5 (k CO = 15.24Nm −1 ) and η 5 -C 5 Me 5 (k CO = 15.29 Nm −1 ).Similarly, experimental data are not available for toluidyne complexes of all ligands L, in which cases experimental data for the corresponding phenyl or xylyl carbynes are instead provided alongside those calculated for the toluidyne.Table 1 presents νCO frequencies corrected by an anharmonic scaling factor (λ1) o 0.9740 as implemented in the SPARTAN20 ® software for the ωB97X-D/6-31G* combination [101,102], which, however, still overestimates these frequencies relative to those observed experimentally.Calculated vibrational frequencies generally exceed experimentally de termined values due to incomplete incorporation of electron correlation, neglect of me chanical anharmonicity and the use of finite basis sets [103][104][105].
This overestimation is assumed to be relatively uniform, allowing for the develop ment of generic scaling factors (λ) derived via least-squares analysis of calculated vs. ex perimental frequencies for various test sets of molecules.Such test sets typically involv small molecules comprising first and second row elements but rarely metals.Moreover single scaling factors are not universally appropriate for the entire vibrational spectros copy range (400-4000 cm −1 ) [106], and the fundamental modes from which they are de rived generally fall below the range of interest to organometallic chemists (1800-2200 cm −1 ).For the present discussion, it therefore seems appropriate to consider an alternativ scaling factor (λ2 = 0.9297), which we have derived from consideration of 18 experimenta and fundamental modes from Table 2, with the caveat that only data measured in di chloromethane solutions were used, discarding those from solid-state or alkane solution measurements.Gas phase data were calculated, since there seemed little benefit in intro ducing further artificial approximations such as conductor-like polarizable continuum molecular electron density (SMD) or conductor-like screening models (COSMO) [107-111 when the aim was to construct an approximate but internally consistent steric-electroni map rather than to seek out absolute values.
The data points may be loosely grouped according to the charge on the complex, with the general observation that as this increased from anionic through neutral to cationic, so too did the kCO value.It should, however, be noted that these groupings are not well sep arated.Rather, some cationic complexes are coordinated by strong net σ-donors, e.g.N,N′,N″-trimethyltriazacyclononane (Me3 [9]aneN3, Entry 7) and tris(dimethylpyra A bonus of the requisite frequency calculations is that the vibrational mode for the W≡C bond may be readily identified, though in contrast to similar essentially 'pure' vibrations for terminal oxo (M≡O) and toluidyne (M≡N) ligands, this is by necessity coupled to the vibration of the C-C bond connecting it to the aryl substituent.This mode appears within a remarkably narrow frequency range (1345-1356 cm −1 ), with the exception of 4 (1334 cm −1 ), perhaps also reflecting the electron-releasing nature of the HB(ImMe) 3 ligand.The intensity of this mode, however, varies substantially, such that in some cases it is unlikely to be unambiguously identified in experimental IR spectra.This invariance in the value of ν WC is also reflected in the derived Löwden bond orders (Table 2) for this bond, which fall within the very narrow range of 2.32-2.41.This is despite considerable variation in the calculated natural charge on tungsten (+0.405 to +1.177), while that for carbon is comparatively invariant (-0.105 to -0.299); i.e., electroneutrality would appear to balance charge distribution within the 'LW' unit so as to not significantly transmit this influence to the carbyne ligand.
Table 1 presents ν CO frequencies corrected by an anharmonic scaling factor (λ 1 ) of 0.9740 as implemented in the SPARTAN20 ® software for the ωB97X-D/6-31G* combination [101,102], which, however, still overestimates these frequencies relative to those observed experimentally.Calculated vibrational frequencies generally exceed experimentally determined values due to incomplete incorporation of electron correlation, neglect of mechanical anharmonicity and the use of finite basis sets [103][104][105].
This overestimation is assumed to be relatively uniform, allowing for the development of generic scaling factors (λ) derived via least-squares analysis of calculated vs. experimental frequencies for various test sets of molecules.Such test sets typically involve small molecules comprising first and second row elements but rarely metals.Moreover, single scaling factors are not universally appropriate for the entire vibrational spectroscopy range (400-4000 cm −1 ) [106], and the fundamental modes from which they are derived generally fall below the range of interest to organometallic chemists (1800-2200 cm −1 ).For the present discussion, it therefore seems appropriate to consider an alternative scaling factor (λ 2 = 0.9297), which we have derived from consideration of 18 experimental and fundamental modes from Table 2, with the caveat that only data measured in dichloromethane solutions were used, discarding those from solid-state or alkane solution measurements.Gas phase data were calculated, since there seemed little benefit in introducing further artificial approximations such as conductor-like polarizable continuum, molecular electron density (SMD) or conductor-like screening models (COSMO) [107][108][109][110][111] when the aim was to construct an approximate but internally consistent steric-electronic map rather than to seek out absolute values.
The data points may be loosely grouped according to the charge on the complex, with the general observation that as this increased from anionic through neutral to cationic, so too did the k CO value.It should, however, be noted that these groupings are not well separated.Rather, some cationic complexes are coordinated by strong net σ-donors, e.g., N,N ,Ntrimethyltriazacyclononane (Me 3 [9]aneN 3 , Entry 7) and tris(dimethylpyrazolyl)methane (HC(pzMe 2 ) 3 , Entry 17), such that comparatively low values are observed for ν CO and k CO .Likewise, the icosohedral dicarbollide complexes [W(≡CC 6 H 4 Me-4)(CO) 2 (η 5 -C 2 B 9 H 9 R 2 )] - (R = H, Me), whilst anionic, have frequencies not dissimilar to those of neutral 4 (Entry 1) and 5 (Entry 2), while the anionic docosohedral example [W(≡CC 6 H 4 Me-4)(CO) 2 (η 6 -C 2 B 10 H 10 Me 2 )] -has a considerably higher k CO value 15.04 Ncm −1 .There is no correlation obvious to us between the net charge on the complex and derived WC bond orders or W≡C bond lengths for the carbyne ligand.

Sub-Series of Ligands
Tables 1 and 2 along with Figures 3 and 4 contain a number of as yet hypothetical derivatives that have yet to be prepared but which would appear to be entirely plausible based on the demonstrated viability of the ligands L in other systems.Some comments on sub-classes now follow.

Hydrotris(N-R 1 -imidazolylidenyl)borates
Central to this communication are the tris(imidazolylidene)borates HB(ImR 1 ) 3 .From Figure 3, it is clear that the ligand HB(ImMe) 3 occupies a position in a somewhat sparsely populated area of the electronic-steric map, being both strongly basic and also imparting considerable steric prophylaxis upon the carbonyl and carbyne co-ligands akin to that provided by the popular HB(pzMe 2 ) 3 ligand.The experimental and calculated values for k CO are comparable to those for Stone's dicarbollide complexes (L = η 5 -C 2 B 9 H 9 R 2 R = H, Me) [79,88] which, however, carry a net negative charge, and so it must be assumed much of the negative charge resides within the carbaborane cage.
As expected, the %V bur value for 4 is close to that of 5. Smith has developed synthetic routes to the pro-ligand salts that carry N-substituents of varying bulk ( t Bu, Cy, C 6 H 2 Me 3 -2,4,6) [4] and accordingly entries 1 (R 1 = Me, 4), 35 (R 1 = Et), 36 (R 1 = i Pr), 37 (R 1 = t Bu) and 38 (R 1 = Ph) survey the sequential inclusion of increasing steric bulk at the position β to the metal.All attempts to geometrically minimize, or indeed even reasonably construct, the derivative with R 1 = mesityl met with spectacular failure, perhaps indicating a step too far, though this ligand has been successfully installed on four-coordinate nickel [37].The phenyl derivative 38, however, is able to accommodate unsubstituted aryl groups by allowing them to interdigitate between the carbonyl and carbyne ligands such that the aryl planes are near colinear with the W . . .B vector.A very approximate value for the %V bur of 56.6% is provided by the hypothetical and implausible (distorted) octahedral complex [WMe 3 {HB(ImMes) 3 ] (PM3tm level of theory).While it is not dissimilar to the value (59.8%) estimated for L = neutral MeC(CH 2 PPh 2 ) 3 (16) and anionic MeB(CH 2 PPh 2 ) 3 (31), inclusion of this excessive steric bulk would seem problematic.It should, however, be noted that a rich organometallic chemistry has emerged for the dihydrobis(N-mesitylimidazolylidenyl)borate ligand coordinated to tantalum [33,34], for which the bidentate variant presents a considerably reduced steric impact, e.g., V bur = 39.8% in pseudooctahedral [TaMe 4 {H 2 B(ImMes) 2 }].The trifluoromethylimidazolylidenyl derivate (Entry 39) was also considered and found to be a rather modest net donor (ν CO = 15.2Ncm −1 ) while presenting a comparatively occlusive encapsulating pocket (V bur = 57.1%).The only currently available synthesis of N-trifluoromethylimidazole [111] is, however, not particularly amenable to the scales needed for an exploration of the HB(ImCF 3 ) 3 ligand.Figure 5 depicts the steric maps that arise from %V bur calculations and shows the progression in steric encumbrance as the N-substituents are replaced along the alkyl series R 1 = Me, Et, i Pr, t Bu alongside those for R = Ph and CF 3 .What is immediately apparent from Figure 4 is that replacement of the 'parent' Nmethylimidazole, which is both commercially available and cheap, with ethyl, iso-propyl or phenyl imidazoles actually results in very modest variation in the steric impact around the coordination sphere of the metal because the groups can direct their bulk away from the carbonyl and carbyne ligands.It is only with the t Bu (and to a lesser extent the CF3) derivative that this bulk is unavoidably directed towards the metals centre.This is clear when the 3.5 Å value typically and arbitrarily employed in %Vbur calculations is replaced by 4.0, 5.0 and 6.0 Å (Figure 6), respectively.Thus, inclusion of phenyl, primary or secondary alkyl groups appears to have rather a modest steric influence directly on the metal coordination sphere but may contribute in a secondary manner to compound longevity by reducing the collisional cross section (Arrhenius pre-exponential factor) for proceeding reactions.It seems that only with tertiary alkyl (e.g., t Bu) or ortho-substituted aryl substituents (e.g., mesityl) that a significant impact on the steric profile is likely to manifest in the reactivity.What is immediately apparent from Figure 4 is that replacement of the 'parent' Nmethylimidazole, which is both commercially available and cheap, with ethyl, iso-propyl or phenyl imidazoles actually results in very modest variation in the steric impact around the coordination sphere of the metal because the groups can direct their bulk away from the carbonyl and carbyne ligands.It is only with the t Bu (and to a lesser extent the CF 3 ) derivative that this bulk is unavoidably directed towards the metals centre.This is clear when the 3.5 Å value typically and arbitrarily employed in %V bur calculations is replaced by 4.0, 5.0 and 6.0 Å (Figure 6), respectively.Thus, inclusion of phenyl, primary or secondary alkyl groups appears to have rather a modest steric influence directly on the metal coordination sphere but may contribute in a secondary manner to compound longevity by reducing the collisional cross section (Arrhenius pre-exponential factor) for proceeding reactions.It seems that only with tertiary alkyl (e.g., t Bu) or ortho-substituted aryl substituents (e.g., mesityl) that a significant impact on the steric profile is likely to manifest in the reactivity.by 4.0, 5.0 and 6.0 Å (Figure 6), respectively.Thus, inclusion of phenyl, primary or secondary alkyl groups appears to have rather a modest steric influence directly on the metal coordination sphere but may contribute in a secondary manner to compound longevity by reducing the collisional cross section (Arrhenius pre-exponential factor) for proceeding reactions.It seems that only with tertiary alkyl (e.g., t Bu) or ortho-substituted aryl substituents (e.g., mesityl) that a significant impact on the steric profile is likely to manifest in the reactivity.An intriguing question does, however, arise when the steric bulk is exaggerated, in that whilst this might be expected to increase the donor strength of the NHC:→W interaction, the inter-ligand repulsion is such that there is a notable increase in the W-C bond lengths of both the NHC donors cis (mean value) and trans to the carbyne (Table 3).Thus, the simple σ-basicity vs. π-acidity of the free NHC is only part of the story if the metal-donor bond length increases (weakens?)significantly.This does not appear to be the case in the present system, in that while the t Bu derivative has especially long NHC-W bond lengths, it is nevertheless the most potent net donor (k CO = 14.41 Ncm −1 ) of all the ligands considered.In the case of the complexes [Ni(NO)(L)] where L represents a sub-set of ligands considered in Tables 1 and 2 (η 5 -C 5 Me 5 , Tp*, Hb(mt tBu ) 3 and PhB(CH 2 PPh 2 ) 3 [112][113][114][115][116]) alongside those for selected tris(imidazolylidenyl)borates RB(ImR 1 ) 3 (R = H, Ph; R 1 = Me, t Bu, Mesilyl, CH 2 Cy [37]), Smith employed nitrosyl stretching frequency as a measure of the relative donor ability of 'L'.Similar σ-donor/π-acceptor arguments apply as they do to CO with the caveat that depending on the electronic nature of the metal centre, the nitrosyl may bend; i.e., lower values for ν NO may indicate an electron rich metal centre or bending, which becomes more prevalent for late-transition metal centres with high d-occupancies [117].In the case of four-coordinate nitrosyls of nickel, the situation is complicated by subtleties in the electronic nature of the nickel that remain moot [47,49].While Smith was consistent in reporting data from the same essentially non-coordinating solvent toluene (or sometimes THF), data from other sources were acquired from a variety of media (not always stated) including the solid state (KBr, Nujol, Ar (s) , etc.).The selenoimidazolylborate is a case in point for which the reported solid-state IR spectrum comprised two ν NO bands [114].Since the crystal structure revealed a single crystallographically independent molecule, one might assume the second vibrational mode was due to an alternative crystal modification in the bulk sample.Given the two bands differ by 11 cm −1 and the entire Tolman ν T scale only spans 45 cm −1 , the importance of using solution derived data, preferably from a common solvent, is demonstrated.

Toluidyne Orientation
Perusal of the structures, experimentally or computationally derived, reveals a broad range of orientations of the toluidine ring with respect to the nominal coordination axes.This is of secondary importance in that for all examples, the 1 H NMR spectra involve a simple AA'BB' pattern indicating free rotation on the 1 H NMR (and 13 C) NMR timescale(s).Arbitrarily adopting the cationic carbyne formalism ([CF] + , [NO] + and CO being isoelectronic molecules), coordinated to a d 6 -ML 5 fragment, the two carbyne acceptor orbitals vary in energy by only 0.2 eV, as do the two metal retrodative orbitals (HOMO-1, HOMO-2) of, e.g., the 'W(CO) 2 (Tp)' fragment (Figure 7).The HOMO itself is invariably associated with metal-carbonyl π-bonding and is orthogonal (δ-symmetry) to the W-Carbyne vector.Accordingly, any conformational preference should be presumed to reflect interligand steric factors and/or intermolecular packing effects.For the majority of structurally characterized carbyne complexes of the M(CO) 2 (Tp*) fragment; for example, the carbyne substituent typically nestles between two dimethylpyrazolyl groups.NB: The molecular orbitals of the actual carbyne complex are, as they must be, independent of the arbitrary electron allocation to hypothetical constituent fragments; i.e., similar interpretation based on [CC 6 H 4 Me] 3-and d 2 -ML 5 3+ or neutral CC 6 H 4 Me-4 and d 5 -ML 5 deconstructions lead to the same conclusion.
Molecules 2023, 28, x FOR PEER REVIEW 17 of 30 2.4.5.Toluidyne Orientation Perusal of the structures, experimentally or computationally derived, reveals a broad range of orientations of the toluidine ring with respect to the nominal coordination axes.This is of secondary importance in that for all examples, the 1 H NMR spectra involve a simple AA'BB' pattern indicating free rotation on the 1 H NMR (and 13 C) NMR timescale(s).Arbitrarily adopting the cationic carbyne formalism ([CF] + , [NO] + and CO being isoelectronic molecules), coordinated to a d 6 -ML5 fragment, the two carbyne acceptor orbitals vary in energy by only 0.2 eV, as do the two metal retrodative orbitals (HOMO-1, HOMO-2) of, e.g., the 'W(CO)2(Tp)' fragment (Figure 7).The HOMO itself is invariably associated with metal-carbonyl π-bonding and is orthogonal (δ-symmetry) to the W-Carbyne vector.Accordingly, any conformational preference should be presumed to reflect inter-ligand steric factors and/or intermolecular packing effects.For the majority of structurally characterized carbyne complexes of the M(CO)2(Tp*) fragment; for example, the carbyne substituent typically nestles between two dimethylpyrazolyl groups.NB: The molecular orbitals of the actual carbyne complex are, as they must be, independent of the arbitrary electron allocation to hypothetical constituent fragments; i.e., similar interpretation based on [CC6H4Me] 3-and d 2 -ML5 3+ or neutral CC6H4Me-4 and d 5 -ML5 deconstructions lead to the same conclusion.

A Heterobimetallic Hydrotris(imidazolylidenyl)borate Complex
To date, the tris(imidazolylidenyl)borate class of ligands has only been employed in monometallic systems; however, terminal carbyne ligands have an extensively documented propensity to support metal-metal bond formation, as championed by Stone [144].In particular, the addition of gold(I) reagents to monometallic carbyne complexes [145][146][147][148][149][150][151][152][153][154][155] is of interest due to the tendency of the carbyne to adopt a semi-bridging rather than the more common symmetrical bridging geometry.This is considered to arise when the carbyne bridges electronically disparate metals, and therefore, the late high d-occupancy metal (d 10 gold(I) or platinum(0)) is considered to act as a σ-donor (Z-type metal-ligand bonding [156]) to the carbyne carbon.Accordingly, the reaction of 4 with [AuCl(SMe 2 )] was investigated and found to readily provide the bimetallic complex [WAu(µ-CC 6 H 4 Me-4)Cl(CO) 2 {HB(ImMe) 3 }] (6, Scheme 4).The complex is somewhat unstable in solution, slowly depositing elemental gold during unsuccessful attempts to slowly obtain crystallographically serviceable crystals.The formulation, however, rests reliably on spectroscopic data which may be compared with precedents for other carbyne and tungsten substituents.The reaction is accompanied by a shift in the ν CO absorptions to a higher frequency (CH 2 Cl 2 : 1971, 1879 cm −1 ) than those of the precursor in the same solvent (1958, 1873 cm −1 ).The carbyne carbon resonance in the 13 C{ 1 H} NMR spectrum appears at δ C = 277.7,and while this is only marginally shifted from that of the precursor (280.7 ppm), there is a dramatic decrease in the value of 1 J WC (85 Hz cf.171.3 Hz for 4), which is consistent with the increase coordination number (reduced s-character) of both tungsten and carbon.The resonances due to the imidazolylidene donors appear at 187.7 [ 1 J CW = 90 Hz], 173.7 [ 1 J CW = 71 Hz] in a similar region to the precursor but with more similar values for 1 J WC (90,71 Hz) once the trans influence of the carbyne is alleviated upon gold adduct formation.
While the 1 H and 13 C{ 1 H} NMR spectra each confirm a locally C s symmetric environment around the tungsten, at least on these timescales, they do not distinguish between the AuCl unit lying syn or anti to the imidazolylidene units; however, based on precedent from the sterically similar HB(pzMe 2 ) 3 ligand, it seems likely that the AuCl unit nestles between two imidazolylidene rings.This geometry was adequately modelled (Figure 8) at the ωB97X-D/6-31G/LANL2Dζ/gas-phase level of DFT, from which it would appear that the W-C bond clearly retains its considerable multiple-bond character (W-C = 1.913Å).The W-C-C (148.9While the 1 H and 13 C{ 1 H} NMR spectra each confirm a locally Cs symmetric environment around the tungsten, at least on these timescales, they do not distinguish between the AuCl unit lying syn or anti to the imidazolylidene units; however, based on precedent from the sterically similar HB(pzMe2)3 ligand, it seems likely that the AuCl unit nestles between two imidazolylidene rings.This geometry was adequately modelled (Figure 8) at the ωB97X-D/6-31G/LANL2Dζ/gas-phase level of DFT, from which it would appear that the W-C bond clearly retains its considerable multiple-bond character (W-C = 1.913Å).  [157] and CH2CH2NHMe2 + Cl -2.712 Å [158]).The optimized Au-W bond length for 6 (2.812 Å) is therefore comparable to these, though towards the longer end of the range.The infrared νCO absorptions are noted at 1955 and 1899 cm −1 (λ2), while TD-DFT analysis suggests that the colour of the complex may be attributed to absorptions calculated at 420 nm (W-C ≈ z-axis: HOMO-LUMO; dxy-W=Cπ*), 357 (HOMO-LUMO+1; dxy-WAuσ*) and 344 nm (HOMO-1-LUMO; W=Cπ-W=Cπ*), the first two of which involve considerable charge transfer.

Synthesis of [Mo(≡CC6H4Me-4)(CO)2(pic)2(Br)] (2b)
A solution of 4-bromotoluene (6.842 g, 40.00 mmol) in diethylether (50 mL) was cooled to 0 °C before lithium (1.3 g, 190 mmol, hammered and cut wire) was added.This was stirred vigorously at 0 °C for 30 min before being allowed to slowly warm to room temperature and being stirred for a further 3.5 h.The lithium reagent was added dropwise to a suspension of [Mo(CO)6] (6.338 g, 24.01 mmol) in diethyl ether (60 mL) until negligible [Mo(CO)6] remained, as indicated by in situ IR spectroscopy.The red solution was cooled to -78 °C before trifluoroacetic anhydride (3.40 mL, 24.3 mmol) was added dropwise over a period of 10 min.After being stirred for 45 min at -78 °C, 4-picoline (6.0 mL, 62 mmol) was added.The suspension was allowed to slowly warm to room temperature and stirred overnight.The yellow precipitate was isolated via cannula filtration and extracted with dichloromethane (50 mL) and the extracts filtered through diatomaceous earth, followed by washing with further dichloromethane (6 × 5 mL).The volume was reduced to 50 mL under reduced pressure before slow dilution with hexane (120 mL) to precipitate a yellow  A solution of 4-bromotoluene (6.842 g, 40.00 mmol) in diethylether (50 mL) was cooled to 0 • C before lithium (1.3 g, 190 mmol, hammered and cut wire) was added.This was stirred vigorously at 0 • C for 30 min before being allowed to slowly warm to room temperature and being stirred for a further 3.5 h.The lithium reagent was added dropwise to a suspension of [Mo(CO) 6 ] (6.338 g, 24.01 mmol) in diethyl ether (60 mL) until negligible [Mo(CO) 6 ] remained, as indicated by in situ IR spectroscopy.The red solution was cooled to -78 • C before trifluoroacetic anhydride (3.40 mL, 24.3 mmol) was added dropwise over a period of 10 min.After being stirred for 45 min at -78 • C, 4-picoline (6.0 mL, 62 mmol) was added.The suspension was allowed to slowly warm to room temperature and stirred overnight.The yellow precipitate was isolated via cannula filtration and extracted with dichloromethane (50 mL) and the extracts filtered through diatomaceous earth, followed by washing with further dichloromethane (6 × 5 mL).The volume was reduced to 50 mL under reduced pressure before slow dilution with hexane (120 mL) to precipitate a yellow solid that was freed of supernatant via cannula filtration and dried under high vacuum (8.473A 1 L three-necked flask was fitted with a stirrer bar, water-cooled reflux condenser, pressure-equalizing dropping funnel and a gas outlet leading to a NaOH scrubber.The entire apparatus was flushed with nitrogen for 30 min before trimethylamine-borane complex (7.32 g, 100 mmol) was added, followed by 150 mL degassed chlorobenzene.To the dropping was added 50 mL chlorobenzene and bromine (7.8 mL, 85 mmol Br 2 ).The bromine solution was added to the flask at a rate of about one drop/second whilst the reaction was flushed with a gentle stream of nitrogen.This reaction is initially very exothermic and the rate of bromine addition should be adjusted accordingly; caution should also be exercised, since hydrogen gas is also liberated at this stage.After approximately half of the bromine had been added, the exothermicity was less pronounced and rate of addition of the remainder could be increased safely.The mixture was then stirred for 3 h at ambient temperature, during which time the orange colour of bromine slowly faded to a pale yellow.Hydrogen bromide was liberated during this time as nitrogen was continuously swept over the reaction.N-methylimidazole (28 mL, 330 mmol) was added to the mixture, and the was carefully transferred to a heating mantle, where it was brought to reflux for 4-6 h.Upon heating, a white crystalline solid precipitated from the reaction mixture; extended heating is to be discouraged, as this leads to the formation of tarry yellow materials and poor yields of product.The chlorobenzene layer was decanted from the solids while warm, and the flask was then rinsed with 3 × 100 mL portions of toluene; the washings were subsequently discarded.The white solid was dissolved into 100 mL methanol and added slowly to a vigorously stirred solution of NaPF 6 (35 g, 210 mmol; NaBF 4 may also be used) in 100 mL methanol, from which the product precipitated as a fluffy white solid.The white solids were collected via filtration, washed with 3 × 50 mL portions each of methanol and Et 2 O and dried under suction.Purity was sufficient for synthetic purposes, though an analytically pure sample was obtained via re-crystallisation from acetone/Et 2 O (vapour diffusion).Isolated yield 11.50 g (21 mmol, 21%) as the PF 6 salt or 12.90 g (30 mmol, 30%) as the BF 4 salt.
C{ 1 H} NMR spectrum (CD 2 Cl 2 : δ C = 280.7, 1 J WC = 171.4Hz).Consistent with the inferred C s symmetry of the molecule, the carbonyls gave rise to a single resonance (δ C = 223.3, 1 J WC = 132.0Hz) while the tungsten-bound carbon nuclei of the NHC donors gave rise to two resonances at a ratio of 2:1 with markedly different chemical shifts and 1 J WC couplings (δ C / 1 J WC = 192.0/95.2),181.3/44.7).With the exception of the complexes [Pt{H 2 B(ImR 1 ) 2 Molecules 2023, 28, x FOR PEER REVIEW 6 of 30

Figure 3 .
Figure 3. Electronic (kCO) vs. steric (%Vbur) map for a range of facially coordinating ligands derived computationally (DFT:ωB97X-D/LANL2Dζ(W)).A small number (shown in green) were calculated at the semi-empirical PM3tm level of theory due to their large atom count, for which the ordinate positions should be treated with appropriate reservation.

Figure 3 .
Figure 3. Electronic (k CO ) vs. steric (%V bur ) map for a range of facially coordinating ligands derived computationally (DFT:ωB97X-D/LANL2Dζ(W)).A small number (shown in green) were calculated at the semi-empirical PM3tm level of theory due to their large atom count, for which the ordinate positions should be treated with appropriate reservation.

Figure 6 .
Figure 6.Steric map dependence on radius of coordination sphere employed.Figure 6. Steric map dependence on radius of coordination sphere employed.

Figure 6 .
Figure 6.Steric map dependence on radius of coordination sphere employed.Figure 6. Steric map dependence on radius of coordination sphere employed.
• ) and Au-C-C (121.5 • ) angles indicate semi rather than symmetrical bridging such that the C-C and W-Au vectors form an obtuse angle of 101.4 • .Despite numerous (>80) examples of structurally authenticated W-Au bonds, only two have bonds that are not supported by bridging ligands, viz. the compounds [WAu(CO) 3 (PPh 3 )(η 5 -C 5 H 4 R)] (R = H 2.698 Å [157] and CH 2 CH 2 NHMe 2 + Cl -2.712 Å [158]).The optimized Au-W bond length for 6 (2.812 Å) is therefore comparable to these, though towards the longer end of the range.The infrared ν CO absorptions are noted at 1955 and 1899 cm −1 (λ 2 ), while TD-DFT analysis suggests that the colour of the complex may be attributed to absorptions calculated at 420 nm (W-C ≈ z-axis: HOMO-LUMO; d xy -W=Cπ*), 357 (HOMO-LUMO+1; d xy -WAuσ*) and 344 nm (HOMO-1-LUMO; W=Cπ-W=Cπ*), the first two of which involve considerable charge transfer.Molecules 2023, 28, x FOR PEER REVIEW 18 of 30 reliably on spectroscopic data which may be compared with precedents for other carbyne and tungsten substituents.The reaction is accompanied by a shift in the νCO absorptions to a higher frequency (CH2Cl2: 1971, 1879 cm −1 ) than those of the precursor in the same solvent (1958, 1873 cm −1 ).The carbyne carbon resonance in the 13 C{ 1 H} NMR spectrum appears at δC = 277.7,and while this is only marginally shifted from that of the precursor (280.7 ppm), there is a dramatic decrease in the value of 1 JWC (85 Hz cf.171.3 Hz for 4), which is consistent with the increase coordination number (reduced s-character) of both tungsten and carbon.The resonances due to the imidazolylidene donors appear at 187.7 [ 1 JCW = 90 Hz], 173.7 [ 1 JCW = 71 Hz] in a similar region to the precursor but with more similar values for 1 JWC (90, 71 Hz) once the trans influence of the carbyne is alleviated upon gold adduct formation.
The W-C-C (148.9°) and Au-C-C (121.5°)angles indicate semi rather than symmetrical bridging such that the C-C and W-Au vectors form an obtuse angle of 101.4°.Despite numerous (> 80) examples of structurally authenticated W-Au bonds, only two have bonds that are not supported by bridging ligands, viz. the compounds [WAu(CO)3(PPh3)(η 5 -C5H4R)] (R = H 2.698 Å