Understanding and tuning the electronic structure of pentalenides

Here we report the first example of systematic tuning of the electronic properties of dianionic pentalenides through a straightforward synthetic protocol which allows the controlled variation of substituents in the 1,3,4,6-positions to produce nine new compounds, representing the largest pentalenide study to date. Both electron-withdrawing as well as electron-donating aromatics have been incorporated to achieve different polarisations of the bicyclic 10π aromatic core as indicated by characteristic 1H and 13C NMR shifts and evaluated by DFT calculations including nucleus-independent chemical shift (NICS) scans, anisotropy of the induced current density (ACID) calculations, and natural bond orbital (NBO) charge distribution analysis. The introduction of methyl substituents to the pentalenide core required positional control in the dihydropentalene precursor to avoid exocyclic deprotonation during the metalation. Frontier orbital analyses showed arylated pentalenides to be slightly weaker donors but much better acceptor ligands than unsubstituted pentalenide. The coordination chemistry potential of our new ligands has been exemplified by the straightforward synthesis of a polarised anti-dirhodium(i) complex.


Introduction
Pentalenides have been intriguing organic and organometallic chemists ever since Katz and Rosenberger reported the rst example of the planar 10p aromatic bicyclic Pn 2− (1; C 8 H 6 2− ) in 1962. 1,2Unlike its monocyclic congener COT 2− (C 8 H 8 2− ) pentalenide ligands can access a variety of bonding modes from h 1 to h 8 that allow them to fold around a single metal centre or link two metals together. 2,30][11][12] No general synthetic protocol which would allow the controlled variation of substituents is known, and to date only eight derivatives of this ligand architecture have been reported (Scheme 1): Knox, Stone et al. described four Pn 2− variants obtained via thermal rearrangement of substituted cyclooctatetraenes and cyclooctatrienes; 13,14 Cloke et al. subjected unsubstituted Pn 2− to two successive nucleophilic substitution reactions at the 1and 4-position, creating two bis-silylated variants; 15 Ashley et al. developed a multi-step synthesis of a permethylated Pn 2− ; 16 and we recently reported the synthesis of 1,3,4,6-Ph 4 Pn 2− (2 2− ) via a two-step solution-phase synthesis from 1,4-Ph 2 CpH and chalcone. 17Contrary to the ubiquitous cyclopentadienides [18][19][20][21] no systematic investigation of substituent effects exists, meaning that currently there is no understanding of the extent and nature of electronic variation possible within pentalenides.Here we report the deprotonative metalation chemistry of several substituted dihydropentalenes (PnH 2 ) to yield nine new alkali metal pentalenides with symmetrical and unsymmetrical substitution patterns in 1,3-and 1,3,4,6-positions, and for the rst time investigate their electronic effects on the dianionic Pn 2− core by NMR spectroscopy and DFT calculations, including charge distribution analyses, nucleus-independent chemical shi (NICS) scans, and anisotropy of the induced current density (ACID) calculations.
In both molecules the aryl substituents were bent away from the plane of the pentalenide core by 14-19°, with a pair at one ring pointing upwards and the other pair downwards.As no signicant aryl-metal interactions were found in the XRD data of either, and the 1 H and 13 C NMR chemical shis of Li$K [2]  and Li 2 [3] were similar in solution (Tables 1 and S1 ‡), the observation that their substituents had opposite conguration in the solid state (bent towards the metals in [Na(THF) 3 ] 2 [2] but bent away from the metals in [Na(THF) 3 ] 2 [3]) can be ascribed to packing effects that suggest a degree of exibility in these molecules.This bending of the aryl substituents also seemed to have inuenced the hapticity of the sodium ions: analysing the crystallographic data with the metal slippage vector model, 24 [Na(THF) 3 ] 2 [3] had 78% h 5 and 22% h 2 character while [Na(THF) 3 ] 2 [2] showed a more ambiguous pattern with a larger coordination preference towards the C-C bridge (58% h 5 and 42% h 2 character).In addition to this substituent bend, the aryl substituents were twisted by 15-26°relative to the planar pentalenide core in both structures, but only one set of magnetically equivalent aryl protons were observed in their room temperature NMR spectra, suggesting rapid ipping in solution.This exible, pairwise substituent bend and twist found in these tetra-substituted pentalenides is likely to relieve steric repulsion between the aryl groups in 1,3 and 4,6 positions, and has implications for electronic substituent effects by affecting the degree of p overlap between the aryl groups and the pentalenide core (see below).As previously found for [2] 2− , the use of a heterobimetallic combination of two different alkali metals improved the solubility further: subjecting 4H 2 to sequential deprotonation with rst KHMDS and then LiNEt 2 cleanly furnished Li$K [4] which did not precipitate from THF for more than 12 months and allowed solution phase analysis at variable temperature.Cooling a sample of Li$K [4] in THF to −95 °C resulted in decoalescence of the o-aryl and methyl protons of the m-xylyl substituents in the 1 H NMR spectrum at 500 MHz below −65 °C, indicating a slowed oscillation of the aryl substituents (Fig. S54-S57 ‡).The rate constant for the NMR exchange was determined to be 1500 s −1 at −80 °C, where the peaks at 7.24 and 5.85 ppm coalesced, representing a DG ‡ value of 8.55 kcal mol −1 .No signicant changes in the 7 Li chemical shi were observed during the VT NMR experiment, ruling out a reduction in symmetry of [4] 2− through increased ion pairing at lower temperatures.

b. Synthesis and properties of unsymmetrical substitution patterns
Asymmetric pentalenides with different substituents on each ring are rare.In the 1970s Knox and Stone et al. reported two examples of mono-substituted cyclooctatetraenes that isomerised in the presence of [Ru 2 CO 4 (SiMe 3 ) 2 ] to give silylsubstituted di-ruthenium pentalenide complexes which were isolated in <1% yield and characterised by 1 H NMR and IR spectroscopy. 13,14To the best of our knowledge, no method for accessing pentalenides with different substituents on each ring in synthetically useful yields has ever been reported. 80][11][12] We thus investigated the deprotonation of Ph 2p Tol 2 PnH 2 (5H 2 ) and Ph 2 m Xyl 2 PnH 2 (6H 2 ) as their respective double bond isomer mixtures 22 with LiNEt 2 in THF (Scheme 2).Pleasingly, Li 2 [Ph 2 p Tol 2 Pn] (Li 2 [5]) and Li 2 [Ph 2 m Xyl 2 Pn] (Li 2 [6]) were cleanly obtained in spectroscopically quantitative yields, as in the case of their symmetrical congeners Li 2 [3] and Li 2 [4].Both showed the same SSIP character in THF solution at room temperature and were fully characterised by NMR spectroscopy and high-resolution mass spectrometry (Fig. S19-S26 ‡).Likely as a result of their reduced symmetry (time-averaged D 2d ), the homobimetallic di-lithium salts remained fully soluble at 10 −2 M concentration in THF for several months at room temperature.Interestingly, both [5] 2− and [6] 2− displayed a distinct chemical shi difference of 0.06 ppm for their respective wingtip protons H 2 and H 5 , indicative of a different polarisation of the two ve-membered rings of the pentalenide core.
quickly as with 3-6H 2 but required several days to go to completion, presumably due to a higher pK a of the vemembered ring with the electron-donating p MeOPh groups.However, aer seven days at room temperature Li 2 [( p MeOPh) 2 -Ph 2 Pn] (Li 2 [7])the rst oxygenated pentalenidewas cleanly formed in spectroscopically quantitative yield (Scheme 3). 1 H NMR analysis showed that the two electron-donating p MeOPh groups led to a 0.18 ppm chemical shi difference between the wingtip protons H 2 and H 5 in [7] 2− , three times the polarisation seen in [5] 2− and [6] 2− (Table 1).Interestingly, its 7 Li NMR chemical shi of −4.78 ppm suggested a stronger interaction of the Li + with the pentalenide than in 2-6, more in the range of a solvent-shared ion pair (Fig. S27-S29 ‡). 25 To see if the introduction of aryl substituents containing electron-withdrawing groups could lead to further deshielding effects on the pentalenide core, we investigated the deprotonative metalation of an isomeric mixture of ( p FPh) 2 ( p Tol) 2 PnH 2 (8H 2 ). 22LiHMDS in THF was found to be ineffective in furnishing the corresponding pentalenide salt, implying its pK a to be insufficient for the double deprotonation of a dihydropentalene with either p-tolyl or 4-uorophenyl in the 1,3-positions (unlike in the preparation of Li 2 [2]).Using a mixture of LiHMDS and LiNEt 2 showed full consumption of the starting materials by 1 H NMR spectroscopic analysis aer 7.5 hours at room temperature, but with a multitude of new, broad signals that could not be condently assigned.Utilising a heterobimetallic base combination of LiHMDS followed by KH (which had proven successful for 2H 2 and 4H 2 ) resulted in the clean formation of the rst uorinated pentalenide Li$K [( p FPh) 2 ( p Tol) 2 Pn] (Li$K [8]) in spectroscopically quantitative yield aer 6.5 hours (Scheme 3) which was fully characterised by NMR spectroscopy and mass spectrometry (Fig. S30-S32 ‡).The weakly electron-withdrawing p FPh groups in Li$K [8] led to a marginal 1 H NMR chemical shi difference between H 2 and H 5 of 0.05 ppm, indicating a rather weak inuence on the polarisation of the pentalenide core (Table 1).
To increase the electron-withdrawing nature of the substituents on one side of the pentalenide further we also tried to Table 1 Pentalenide positional nomenclature and 1 H NMR chemical shifts of wingtip protons in variously substituted dilithium pentalenides (in THF at room temperature)  deprotonate the p CF 3 Ph-substituted dihydropentalene . 22 However, in situ NMR analysis showed that bases of higher pK a values than LiHMDS led to unidentiable heterogeneous mixtures indicative of decomposition (Scheme 4).All attempts of deprotonating a double bond isomer mixture of , where the CF 3 groups were located in para-position but the uorinated aryl group positioned differently in each dihydropentalene isomer, 22 led to the same observation (as representative example, see Fig. S58 and S59 ‡).Finally, to investigate whether CF 3 groups themselves or their location in the aryl substituents of the dihydropentalenes were incompatible with the basic conditions required for their metalation, we tested the deprotonation of (Ph) 2 (( m F 3 C) 2 Ph) 2 PnH 2 (8d-H 2 ; product of the condensation of 1,4-diphenyl-cyclopenta-1,3-diene 17 and 1,3bis(3,5-bis(triuoromethyl)phenyl)-2-propen-1-one; 26 see the ESI ‡).As observed with the other two CF 3 -functionalised dihydropentalenes, bases of similar or weaker strength than LiHMDS did not lead to pentalenide formation and stronger amide bases led to decomposition, demonstrating the incompatibility of aryl-CF 3 substituents in this deprotonative metalation protocol of dihydropentalenes with ionic bases of pK a >35.As strongly electron-withdrawing aryl substituents were found to be incompatible with deprotonative metalations using alkali metal bases, we investigated 1,3-diarylated dihydropentalenes featuring one unsubstituted ve-membered ring.An isomeric mixture of Ph 2 PnH 2 (9H 2 ) was readily prepared from CpH and chalcone following Griesbeck's protocol. 27Their deprotonation with LiNEt 2 in THF proceeded smoothly to give Li 2 [Ph 2 Pn] (Li 2 [9]) as SSIP in spectroscopically quantitative yield aer 1.5 hours at room temperature (Scheme 5).
Intriguingly, [9] 2− (representing an intermediate between [1] 2− and [2] 2− ) displayed drastically different wingtip chemical shis in the 1 H NMR spectrum: while H 5 on the unsubstituted ve-membered ring resonated at 6.06 ppm, H 2 on the diphenylated pentalenide moiety resonated at 7.05 ppm to give a chemical shi difference of 0.99 ppm (Table 1).This represents the strongest polarisation of any pentalenide reported yet, and suggests a markedly different electronic environment in each ring.Compared with [1] 2− , the introduction of the 1,3-Ph 2 substituents caused a mild remote deshielding of H 5 on the unsubstituted ring of 0.30 ppm (from 5.76 ppm (ref.15) to 6.06 ppm) but caused an even more pronounced deshielding effect on H 2 in the substituted ring than in [2] 2− (7.05 ppm in [9] 2− versus 6.79 ppm in [2] 2− ).The reaction of an isomeric mixture of ( p FPh) 2 PnH 2 (10H 2 ; see Chapter 2 of the ESI ‡) with LiNEt 2 led to the analogous Li 2 [( p FPh) 2 Pn] (Li 2 [10], Scheme 5), in which to our surprise the p FPh substituents caused a slightly poorer deshielding than the phenyl substituents in Li 2 [9]: with a 1 H NMR chemical shi of 6.92 ppm at H 2 , [10] 2− exhibits the second largest pentalenide polarisation based on a chemical shi difference of 0.86 ppm (6.06 ppm for H 5 ).
To investigate whether this exocyclic double bond formation during the deprotonation of 11H 2 was due to the cyclic structure of the alkyl substituents (perhaps caused by conformational effects) we investigated 3-Me-1,4,6-Ph 3 PnH 2 (12H 2 ; Fig. S62 ‡) 22 featuring a methyl group in 3-position instead of the fused cyclohexyl ring in 11H 2 .Deprotonation of 12H 2 with one equivalent of LiHMDS quickly generated the exocyclic double bond isomer Li[12H-exo] analogous to Li[11H-exo] (Scheme 7a).As seen with the latter, all attempts to convert Li[12H-exo] to Li 2 [12] by treating it with a stronger base and/or heating to 70 °C failed, showing exocyclic double bond formation to be a dead end for pentalenide formation from 1,2-dihydropentalenes featuring alkyl substituents in the 3-position.Indeed, the analogous deprotonations of 1,3-Me 2 -4,6-Ph 2 PnH 2 22 and 1,3-Me 2 -PnH 2 28 with LiNEt 2 or n BuLi led to the same observation, and similar reactivity has been reported for the deprotonative metalation of 1,2,3,4,5,6-Me 6 PnH 2 with n BuLi. 29This observation is likely due to these monoanionic intermediates being more allylic cyclopentadienides than hydropentalenides, meaning that although they still feature a methine hydrogen in the 1-position this is not acidied by being bound to the only sp 3 carbon between a double bond and a Cp − (ready to aromatise to one conjugated 10p system) as in an endocyclic hydropentalenide.This exocyclic deprotonation of alkyl groups in the 3-position of 11H 2 , 12H 2 and related 3-alkylated dihydropentalenes is likely the result of kinetic competition, where the base can attack either one of two allylic positions of similar pK a leading to either endocyclic or exocyclic double bond formation (Scheme 7c).The hydrogens on the alkyl substituent being sterically more accessible and statistically dominant over the ring-bound hydrogens (2 : 1 in 11H 2 and 3 : 2 in 12H 2 ) thus leads to predominant formation of the undesired allyl- cyclopentadienides with sterically demanding amide bases akin to the reactivity of 6,6-dialkylpentafulvenes. 30,31Unsubstituted and arylated dihydropentalenes avoid this issue by their lack of exocyclic allylic sites for competing deprotonation.
We hypothesised that if exocyclic double bond formation was indeed due to allylic deprotonation competition, then installing a methyl substituent in the 1-position and blocking exocyclic deprotonation with an aryl substituent in the 3-position should avoid this issue.We thus designed 1-Me-3,4,6-Ph 3 PnH 2 (12 0 H 2 ), a double bond isomer of 12H 2 , by cyclising 1,3-Ph 2 CpH with (E)-1-phenylbut-2-en-1-one (see Chapter 2 of the ESI ‡).To our delight, treating 12 0 H 2 with three equivalents of LiNEt 2 resulted in the clean formation of Li 2 [MePh 3 Pn] (Li 2 [12])the rst mixed aryl-alkyl pentalenidein spectroscopically quantitative yield (Scheme 7b and Fig. S47-S50 ‡).No signs of hydropentalenides (endo-or exocyclic) were observed in the reaction, and Li 2 [12] was stable in THF solution without any signs of rearrangement or decomposition for several months under inert conditions.Comparing [12] 2− with [2] 2− , the substitution of a phenyl substituent with single methyl group led to a substantial polarisation of the pentalenide core as indicated by a signicant 1 H NMR wingtip chemical shi difference of 0.48 ppm (Table 1).Crystal structures of the new dihydropentalenes 11H 2 , 12H 2 and 12 0 H 2 can be found in the ESI (Sections 8.2-8.4 ‡).

c. Computational analysis of electronic structure
To gain deeper insight into the electronic structures of these new pentalenides and understand the polarisation effects introduced by the various substituents, we carried out DFT calculations on [2] 2− , [3] 2− , [9] 2− , [10] 2− , and [12] 2− in comparison with the parent [1] 2− .To better align the calculated geometries with the solid-state molecular structures from X-ray crystallography and obtain meaningful frontier orbital energies we included two Li + counterions in trans h 5 position as found in the solid state.For aromaticity and charge distribution calculations we used the bare dianions as a more realistic models for the speciation in solution where solvent-separated ion pairs exist (see above).Additionally, this avoids possible distortions of charge localisation due to coulomb attraction.The computed geometries for Li 2 [1-3] agreed well with the experimentally determined solid-state XRD structures respectively, 17,32 including the ∼30°twist of the aryl substituents relative to the pentalenide core.To elucidate the degree of aromaticity in the conjugated p systems of arylated pentalenides, we calculated the anisotropy of the induced current density (ACID) as well as nucleus-independent chemical shi (NICS) scans along different axes across the molecules (as indicated in the insets in Fig. 2-4).
This analysis revealed that the pentalenide core was clearly aromatic in all compounds but to different extents.The ACID plot of [1] 2− showed a strong diatropic (aromatic) ring current around the C 8 perimeter of the pentalenide excluding the transannular C-C bond (Fig. 2, top).This nding is consistent with previous MO analyses showing haptotropic mobility of Lewis-acidic metals bound to the dianionic 10p system to be conned to the perimeter, with a forbidden path across the central C-C bridge due to unfavourable orbital overlap. 33The ACID plots of [2] 2− (Fig. 2) and [3] 2− (Fig. S67 ‡) also showed global diatropic ring currents within the pentalenide system but with larger contributions of the transannular C-C bond, in addition the local diatropic ring currents within the four aromatic substituents.
The NICS scans demonstrated a pronounced difference in the degree of the aromatic character of the different pentalenides investigated.Comparison of [1-3] 2− showed that the unsubstituted [1] 2− was the most aromatic pentalenide within this series.In each case, the NICS scan along the Z axis starting from the centre of one ve-membered ring perpendicular to the plane of the pentalenide core was indicative of an aromatic ring current.The isotropic shi was negative throughout the scan, and the shape of the curve was mainly governed by the out-ofplane contributions to the isotropic shi with a clear minimum.For [1] 2− , the calculated NICS at that minimum (Z = 1.0 Å) was −28.6 ppm, which is in the same range as that of benzene (−29.1 ppm at 1.0 Å) and slightly higher than that of Cp − (−33.8 ppm at 0.9 Å). 34 For [2] 2− and [3] 2− the minima were signicantly higher than in [1] 2− , with −12.9 ppm for [2] 2− and −12.5 ppm for [3] 2− at 1.4 Å each (see also Fig. S67 ‡).
The NICS X-scans for [1-3] 2− showed a plateau of maximum diatropicity extending over the two ve-membered rings, with the shallow minima above the bonds originating from s-effect contaminations. 35This feature is again consistent with an induced ring current over the ellipsoidal pentalenide perimeter, as seen in the ACID maps.The X-scans should allow to differentiate whether this was due to a global current over the entire pentalenide system, or the result of superposition of local ring currents in the two fused cyclopentadienyl subunits. 35If the latter was the case, the absence of a signicant current density at the transannular C-C bond, which is most pronounced in the ACID plot of [1] 2− , would be the result of a net cancelation of two counter-currents across this linkage.To answer this question, the NICS X values of [1] 2− and the parent cyclopentadienide were compared.In the case of a superposition of two local Cp − ring currents, the NICS values at 1.7 Å above the central C-C bridge of [1] 2− (−21.1 ppm) should be approximately twice the value for [Cp] − at the same Z height at 1 Å from its ring centre (−12.8 ppm), which is equivalent to the distance between the middle of one pentalenide ring and the transannular C 3 0 -C 6 0 bond in [1] 2− .As this was not the case, it leads to the conclusion that no local cyclopentadienide ring currents exist in [1] 2− , and the ring current is due to an overall pentalenide circuit instead.This situation prevailed in [2] 2− , where the reference 1,3-diphenylcyclopentadienide (1,3-Ph 2 Cp − ) had a NICS X value of −12.0 ppm at 1 Å and [2] 2− had a NICS value of −14.1 ppm at the C 3 0 -C 6 0 bridge (Fig. S72 ‡).The ACID plots of [9] 2− (Fig. 3) and [10] 2− (Fig. S69 ‡) also revealed global aromatic ring currents but showed that the current in the unsubstituted ve-membered ring was signicantly more pronounced than in the disubstituted ring.In fact, a closed loop of an additional local cyclopentadienide-like circuit in the unsubstituted ve-membered ring was seen in each compound.The NICS X shi above the centre of the latter was 21% higher in [9] 2− and 13% higher in  S3 ‡), which can be ascribed to the coplanar arrangement of the phenyl substituents with respect to the pentalenide core (0°dihedral twist compared to 30.2°in [2] 2− ) resulting in better conjugation with the core p

ppm (Table S1 ‡).
The ACID plot of [12] 2− (Fig. 4) showed a relatively uniform aromatic ring current around the pentalenide perimeter similar to that of [2] 2− (cf.Fig. 1).The plot of the NICS Z-scans of [12] 2− also showed little difference in aromaticity between the two subunits, while the X scan indicated a slightly stronger aromatic circuit in the Me,Ph-substituted ring (−12.8 ppm) as shown by lower NICS shis than in the Ph 2 -substituted ring (−10.9 ppm; Fig. S70 and Table S3 ‡).All three phenyl twist angles in [12]    were carried out.][38][39][40][41] While in [1] 2− the dianionic charge was delocalised across but conned within the pentalenide core, in [2] 2− and [3] 2− more than half of the overall negative charge was delocalised into the four aryl substituents.This correlated with a 1 ppm downeld shi of the two equivalent wingtip protons in [2] 2− (6.79 ppm) and [3] 2− (6.66 ppm) compared to those of [1] 2− (5.76 ppm).The slightly higher shielding of the wingtip protons of [3] 2− with respect to those of [2] 2− can be ascribed to the +I effect of the p-tolyl groups leading to slightly less effective charge delocalisation in this case.This was also reected in the comparison of the NBO charges on the two different aryl moieties of [2] 2− and [3] 2− , where the phenyl groups accepted slightly more charge density than the tolyl groups (−1.10 e for [2] 2− vs. −1.05e for [3] 2− ).The same analysis allowed understanding the effect of unsymmetrical substitution patterns.In the pentalenides [9] 2− and [10] 2− the subunits without aryl groups in the 4,6 positions showed a higher sum of charges than their phenylated or p FPh-substituted counterparts where the negative charge can be effectively delocalised into these substituents.This correlated with a higheld shi in the H 5 proton (both 6.06 ppm for [9] 2− and [10] 2− ) while the H 2 protons on the substituted ve-membered rings were found at 7.05 ppm for [9] 2− and at 6.92 ppm for [10] 2− , respectively.The slightly more shielded H 2 in the latter compared to the former can be ascribed to the +M effect of the uorine atoms in para-position of the p FPh substituents; hence, the p-systems of the p FPhgroups in [9] 2− accommodate less charge than the phenyl groups in [10] 2− , resulting in a higher charge density in its pentalenide subunit.Comparing [9] 2− with [2] 2− , the 1 H NMR shis of H 2 in [9] 2− were even more deshielded than in [2] 2− which can be ascribed to the above-mentioned coplanar arrangement of the phenyl substituents with respect to the pentalenide core (0°dihedral twist compared to 30.2°for [2] 2− ) that results in better p conjugation and thus more effective charge distribution.This was also reected in the absolute charge values of the aryl substituents, where one phenyl group in [2] 2− accepted a charge of −0.28 e while in [9] 2− it accepted −0.39 e. Changes in the polarisation of the same subunit within different pentalenides can also be explained by the NBO analyses.For example, the observed NMR shis of the wingtip protons in the unsubstituted parts of [9] 2− and [10] 2− were shied slightly downeld from those in [1] 2− .This is due to the fused nature of the two (hypothetical) subunits, since some charge density was transferred from the unsubstituted Cp 1 into the arylated Cp 2 where it may be delocalised into the electronwithdrawing substituents.The NBO charges also reected the electronic inuence of the alkyl substituent in [12] 2− well, as the C 1 carbon was markedly less negatively charged than the C 3 and C 4 carbon atoms due the +I effect of the methyl group (Fig. 5).As expected, the methyl substituent itself did not accept a signicant amount of charge from the pentalenide core, but its presence in 1-position led to a slightly higher charge density in the 3-phenyl group on the same ring compared to the two phenyls on the other subunit.Due to the electron-donating inuence of the methyl group Cp 1 showed a higher charge density than Cp 2 , consistent with its wingtip proton resonating more higheld (6.19 ppm) from that of the diphenyl substituted subunit (6.67 ppm).Due to the interaction of the fused vemembered rings this polarisation caused the H 5 chemical shi to move upeld compared to that of [2] 2− .
The charge distribution within these pentalenides can be visualised by electrostatic potential (ESP) maps (Fig. 6) which illustrate the differences between unsubstituted [1] 2− , Comparing the frontier orbitals and their energies across the series further illustrates the electronic inuence of the substituents on the pentalenide (Fig. 7, top).In the HOMO of each compound investigated the largest contribution was found in the 1,3,4,6 position, and in the arylated systems some delocalisation into the aryl substituents could be seen (Fig. 7, bottom).The corresponding LUMO showed even stronger delocalisation into the aryl moieties.For the unsubstituted [1] 2− the HOMO looked similar to the substituted pentalenides, whilst the LUMO of [1] 2− was more located on the lithium cations (Fig. S74 ‡).The LUMO+12 of [1] 2− had a shape similar to the LUMOs of the substituted pentalenides, which is due to the missing distribution of electron density on aryl substituents, therefore destabilising this orbital.Although all substituted pentalenides investigated possessed a similar HOMO-LUMO gap of 3.1-3.5 eV, comparison of the absolute energy levels of unsubstituted [1] 2− versus the tetraphenyl-substituted [2] 2− showed how the aryl moieties stabilise the frontier orbitals through conjugation with the dianionic core (Fig. 7, top; Table S2

d. Transmetalation
Since the results discussed above showed a signicant inuence of the substituent pattern on the polarisation of the pentalenide core, we sought to investigate their impact on polarising a homobimetallic transition metal complex.As a proof of concept, we decided to try transmetalation of the unsymmetrically substituted [7] 2− with the symmetrical [Rh I (NBD)(m-Cl)] 2 dimer in THF.These conditions have recently been shown to lead to formation of anti-homobimetallic complexes of [2] 2− with both group 1 and group 2 pentalenide precursors. 42owever, using Li 2 [7] with [Rh I (NBD)(m-Cl)] 2 in THF at room temperature resulted in a complex mixture of unidentied products including several pentalenide species.Using the heavier alkali metal analogue Na 2 [7]  The 1 H NMR spectra of [Rh(NBD)] 2 [7] displayed a reduced wingtip shi difference of 0.1 ppm in comparison to Li 2 [7] (0.18 ppm), likely as a result of the change from a SSIP to a p complex.[7]).Bottom: single crystal X-ray structure of [Rh(NBD)] 2 [7] with thermal ellipsoids at the 50% probability level (hydrogen atoms omitted for clarity).
The latter was clearly established by characteristic 103 Rh-1 Hcouplings ( 2 J RhH = 0.9 Hz for H 2 and 0.8 Hz for H 5 ) as well as 103 Rh-13 C-couplings ( 1 J RhC = 6.3 Hz for C 2 and 5.8 Hz for C 5 ).
The NBD ligands gave rise to two doublets for the olenic carbons at 41.3 ppm and 41.2 ppm respectively (see Fig. S53 ‡), again showing the asymmetry of the two [Rh I (NBD)] + fragments bound to [7] 2− .The observation of distinct resonances and coupling constants also indicated a static bonding situation without dynamic exchange (via decoordination or ringwalking) 33 of the two metals.[Rh(NBD)] 2 [7] was further characterised by mass spectrometry and XRD analysis, the latter con-rming the expected anti-arrangement of the two [Rh(NBD)] fragments 42 (see Fig. 8 and Chapter 8.5 in the ESI ‡).In the solid state both C 5 -centroid to Rh distances were 1.93 Å, very similar to those reported for [Rh(NBD)] 2 [2] (1.94 Å). 42

Conclusion
We have described the synthesis and properties of nine new symmetrically and unsymmetrically arylated pentalenides, including a strategy for the introduction of alkyl substituents with b-hydrogens that otherwise lead to exocyclic deprotonation.The dialkali metal salts crystallise in the usual trans h 5 coordination mode, but in ethereal solution solvent-separated ion pairs exist that feature rapid (likely concerted) substituent ipping.DFT calculations have conrmed a dihedral twist angle of ∼30°(as found experimentally in the solid state) as the energetic minimum due to steric clash competing with the electronic preference for co-planar conjugation.Even at a ∼30°t wist, four aryl substituents have been shown to withdraw over 50% of the charge density from the core and reduce its aromaticity by up to 20%.The NMR chemical shis of the wingtip protons in the 2 and 5 positions of the pentalenide serve as sensitive probes for the polarisation, charge density, and degree of aromaticity of the two ve-membered subunits, and the 13 C NMR shis of the quaternary carbons connected to the aryl substituents are related to the dihedral angle and degree of conjugation.Whereas unsubstituted C 8 H 6 2− is best described as a fully conjugated sp 2 system containing 10 p electrons around its perimeter (with negligible electron density at its transannular bond), when aryls are introduced in the 1,3,4,6 positions conjugation with the substituents leads to charge localisations in these pentalenides that are more accurately depicted as two allyl units joined by a shared C]C bond.
Frontier orbital analysis has shown such arylated pentalenides to be slightly weaker donors (i.e. less reducing) but much better acceptor ligands than unsubstituted (and likely also alkylated) pentalenides, making them promising "so" ligands for electron-rich metal centres.As a proof of concept, we provide an example of transmetalation of an unsymmetrically substituted pentalenide to a d-block element and report a polarised dirhodium(I) complex where each metal atom with its auxiliary ligands are electronically as well as sterically distinct.The ease and modularity of our synthetic protocol paired with the quantum chemical insights reported will hopefully enable more widespread utilisation of this ligand framework and nally allow its full potential in organometallic chemistry to be realised.][11][12] Further variations in substitution patterns, including multiple alkyl groups introduced via complementary strategies, as well as their use in organometallic chemistry will be reported in due course.

Scheme 1
Scheme 1 Substituted pentalenides reported to date (top) and work reported here (bottom).
system.Unlike in[2] 2− , co-planarity is possible in[9] 2− due to the absence of substituents on the other half of the pentalenide, showing electronic preference for full conjugation where sterically possible.Experimentally this change in conjugation due to different dihedral angles between the pentalenide core and the aryl substituents could also be seen in the NMR shis of the quaternary C 1 and C 3 atoms: whereas in [2] 2− they resonated at 109.5 ppm (30.2°twist angle) in [9] 2− they were shied upeld to 103.5 2− fell in the range of 30-34°, showing a methyl group to cause the same level of partially interrupted conjugation as in the tetraaryl pentalenides [2] 2− and [3] 2− .In order to understand the degree of charge localisation and substituent effects in these pentalenides, NBO (natural bond orbital) calculations of [1] 2− , [2] 2− , [3] 2− , [9] 2− , [10] 2− , and [12] 2−

Fig. 5
Fig. 5 Left: separation of the pentalenide core into two Cp subunits (indicated by red circles) and calculated sums of NBO charges for each subunit compared to the experimentally obtained 1 H NMR shifts.Right: plot of sums of NBO charges for each part (charges on the shared C 30 and C 60 atoms were equally distributed between both subunits; for compounds [1-3] 2− the charges on Cp 1 and Cp 2 are added up to eliminate artifacts from minor symmetry deviations).