Evaluation of the 5-ethynyl-1,3,3-trimethyl-3H-indole ligand for molecular materials applications

Synthesis

The 3H-indole ligand 2 was obtained in high yield by the KOH-promoted desilylation of 1 (Scheme 1). The ethene 3 was also obtained from this reaction. The mechanism for the formation of 3 is not well understood.

Scheme 1. 

Synthesis of 5-ethynyl-2,3,3-trimethyl-3H-indole 2 (rt, room temperature).

A similar ethene compound from the reaction of an alkyllithium reagent and a 3H-indole has been reported and was proposed to form by a radical mechanism (Supplementary Scheme S1).^[24] Electron Spin Resonance (ESR) spectroscopy confirmed the presence of a radical intermediate from the reaction of methylmagnesium iodide or organolithium reagents and a 3H-indole.^[24,25] Alternative oxidation mechanisms are proposed in the absence of such alkyl reagents in the current study (Supplementary Scheme S2).

Following the oxidation of 2 to the aldehyde, in a manner similar to that in the literature,^[26–28] an aldol condensation with another molecule of 2 and the elimination of water gives the ethene dimer 3. The propensity of molecules similar to 2 to undergo aldol condensation is known.^[29] Alternatively, the oxidation of 2 to the N-oxide is also plausible. As similar self-condensation of indolium salts has previously been observed,^[30,31] which involves the activation of the indole to the N-oxide for further reactions such as with another molecule of 3H-indole. Subsequent deprotonation and elimination of water yield the dimeric indole molecule. Schemes for these proposed mechanisms are shown in the supporting information.

The ¹H NMR spectrum of 2 shows a resonance assigned to the terminal alkyne proton at 3.07 ppm. Sharp absorptions at 3176 (ν(CH)) and 2099 (ν(C≡C)) cm⁻¹ in the IR spectrum further evince the terminal alkyne functional group. Compound 3 is fluorescent under UV light. The singlet at 3.07 and 128.51 ppm in the ¹H and ¹³C NMR spectra respectively was assigned to the alkene bridge. A weak absorption at 1607 cm⁻¹ in the IR provided further evidence for the alkene bridge (ν(C=C)). The atmospheric-pressure chemical ionisation (APCI) mass spectrum showed a peak at m/z 363.1851 consistent with the molecular ion. Single-crystal X-ray diffraction studies confirmed the connectivity of 2 and 3.

The half-sandwich monometallic σ-alkynyl complexes were synthesised using established or slightly modified literature methodologies. The complexes 4–7 were synthesised by the vinylidene route (Scheme 2).^[32] The appropriate half-sandwich ruthenium or iron halides and 2 were refluxed in methanol with NH₄PF₆ present to produce a vinylidene intermediate. After the reaction mixture had cooled to room temperature, the vinylidene species was then deprotonated by carefully adding 1,8-diazabicycloundec-7-ene (DBU) and leaving the mixture undisturbed overnight. The slow diffusion of DBU over a long period afforded well-formed crystals of the targeted complexes. Copper-assisted transmetallation^[33] of Ru(CO)₂(Cp)Cl (Cp, cyclopentadiene) and 2 in THF and NEt₃ afforded 8 in good yield (Scheme 2), whereas complex 9 was synthesised in a basic methanolic solution containing 2 and the gold chloride precursor, PPh₃AuCl (Scheme 2).^[34] Both 8 and 9 showed light sensitivity. The gold complex is of interest owing to its potential use as transmetallation reagent and catalyst.^[35–37]

Scheme 2. 

Preparation of the complexes 4–9 and 10a, b (on, overnight; rt, room temperature).

The ruthenium complexes 4 and 5 showed absorptions at ~2060 cm⁻¹ in their IR spectra assigned to the alkynyl functional group. This assigned ν(C≡C) absorption appeared at a slightly lower wavenumber, ~2050 cm⁻¹, for the iron complexes 6 and 7. Comparatively, the alkynyl stretch for 8 was assigned to the IR absorption at 2119 cm⁻¹. Strong absorbances for the symmetric and asymmetric carbonyl stretches for 8 were assigned to the IR bands at 2035 and 1994 cm⁻¹ respectively.

A single peak in the ³¹P NMR spectrum for 4–7 and 9 further evinced the synthesis of the single metal σ-alkynyl complexes. Resonances for the respective cyclopentadienyl ligands were assigned to the peaks at δ 4.51 (4), 1.68 (5), 1.56 (6), 4.32 (7), 5.48 (8). Resonances for the cyclopentadienyl and phosphine ligands are within the typical range of half-sandwich ruthenium and iron complexes.^{[20,35,38,39]}

The ¹H and ¹³C NMR spectra contained resonances consistent with the indole ligand for all complexes. Supplementary Fig. S1 shows the general atom numbering and labelling for the indole and phosphine ligands of the complexes. In general, resonances for Cα, Cβ and C5 are deshielded on metal–carbon bond formation of the free ligand. The extent of this deshielding is dependent on the electron density of the metal centre.^[40] Peaks for the Cα carbon in 4, (δ 113.8) 5 (δ 127.1), 7 (δ 122.3 ppm) and 8 (δ 81.3 ppm) and the Cβ carbon in 4 (δ 116.1 ppm), 5 (δ 111.4 ppm), 6 (δ 121.1), 7 (δ 122.7) and 8 (δ 111.0) show close agreement with expected resonances.^[20,41] The signals for Cα for complexes 4–7 appear as triplets owing to coupling to the phosphorus ligand on the metal centre. The peak for Cα in complex 9 appears as a doublet at δ 135.8 with a J_CP coupling of 146 Hz, whereas the peak for Cβ appears at δ 104.1 with a J_CP coupling of 26 Hz. The chemical shifts and coupling constants are similar to previously synthesised gold(I) alkynyl complexes.^[34] The observation of these alkynyl carbons is in contrast to a previous report that suggests these analogous resonances were not observed as a consequence of quadrupolar broadening and quenching in those gold type complexes.^[42]

The electrospray ionisation (ESI) mass spectra of complexes 4–9 showed ions consistent with an [M + H]⁺ species, except for 6, which showed a peak assigned to the [M]⁺ species. In addition, the mass spectra of complexes 4, 5 and 7 contained an ion consistent with an [M + 2H]²⁺ species, whereas the mass spectrum of complex 5 also showed an ion at m/z 663, consistent with the oxidative cleavage of the alkyne ligand to give the carbonyl cation complex [Ru(dppe)(Cp*)CO]⁺ (dppe, 1,2,-bis(diphenylphosphino)ethane).^[43] Most complexes also showed fragmentation consistent with the loss of the alkyne ligand and incomplete desolvation. The spectrum of gold complex 9 exhibited an ion at m/z 721 for the [Au(PPh₃)₂]⁺ species also seen in previous studies.^[42]

The synthesis of the alkynyl complex of 2 containing the [Ru(dppe)₂Cl]⁺ moiety by a number of alternative, established literature procedures was unsuccessful.^[44] However, the reaction of [Ru(dppe)₂Cl]OTf and 2 afforded a mixture of complexes 10a and 10b, as evinced by spectroscopic data. We suggest that the basicity of the nitrogen on the indole ligand can deprotonate the intermediate vinylidene species; however, singlet peaks in the ³¹P NMR spectrum at ~48 and 37 ppm suggest two trans configurations are in equilibrium (Scheme 2).^[45] This basic proligand effect has been seen in the syntheses of complexes from 5-ethynyl-2,2′-bipyridine using both the Ru(dppe)₂Cl and RuCp(PPh₃)₂ moieties.^[19,45] Similar effects have been noted in the formation of alkynyl complexes from Ru(dppe)₂Cl₂ and 4-ethynylaniline,^[44] but not with the half-sandwich analogues using the Ru(dppe)(Cp*) moiety.^[46] Contrarily, the syntheses of ruthenium pyridyl alkynyl complexes also report this deprotonation when the alkynyl is formed from RuCp(PPh₃)₂ halides,^[47] but not Ru(dppm)₂Cl₂ (dppm, 1,2-bis(diphenylphosphino)methane).^[48] The ¹⁹F NMR spectrum was consistent with the presence of the triflate anion. The ES mass spectrum showed a peak at m/z 1121 consistent with [M–Cl + MeCN]⁺ species. A presumed impurity, unassigned, was seen at 41 ppm in the ³¹P NMR spectrum and could not be removed by further purification steps. Attempts to deprotonate 10a/10b resulted in mixtures of decomposition and oxidation products evinced by the numerous peaks observed in the ³¹P NMR spectra obtained.

The cross-coupling of 2 with 1,4-di-iodobenzene in NEt₃ gave 11 (Scheme 3). The ¹H NMR spectrum shows resonances consistent with the geminal methyl groups and imine methyl group at δ 1.32 and 2.30 respectively. The multiplet at δ 7.46–7.50 in the ¹H NMR spectrum was assigned to the 1,4-substituted benzene and indole. The ¹³C NMR shows peaks assigned to the alkynyl carbons at 91.9 and 89.1 ppm, shifted downfield from the parent alkyne. The alkynyl bond is also evinced by a band at 2203 cm⁻¹ in the IR spectrum. Mass spectroscopy (APCI) shows a peak consistent with the protonated molecular ion at m/z 441.2338.

Scheme 3. 

Preparation of the compounds 11–18. Conditions: (i) NEt₃, 70°C, 19 h; (ii) HNⁱPr₂, reflux, 3 h, (iii) 60°C, 11 days; (iv) 55°C, 21 days (on, overnight; rt, room temperature).

Access to the platinum metal complexes trans-[Pt(L)₂(PR₃)₂]}, where L = CC-2,3,3-trimethyl-3H-indole, and R=Et (12), Ph (13) or Cy (16) (Scheme 3) was achieved by CuI-catalysed dehydrohalogenation of Pt(PR₃)₂Cl₂ and 2 in the presence of an amine base.^[49] The use of cis-Pt(PR₃)₂Cl₂ conserves the cis-orientation of the complex^[50,51]; however, heating under reflux allows access to the more thermodynamically stable trans isomer.

Spectral characteristics of 12 and 13 are comparable with other bis-alkynyl Pt^II complexes.^[49] The ³¹P NMR spectrum of 12 and 13 respectively shows a single resonance (with ¹⁹⁵Pt satellites) at δ 11.62 and 19.18, affirming a trans orientation of the phosphine ligands. Resonances in the ¹H NMR spectra for the alkyl groups in the triethylphosphines in 12 and phenyl groups in the triphenylphosphines in 13 are assigned to signals at δ 2.17–2.20 (CH₂) and 1.20–1.27 (CH₃), and δ 7.40–7.84 (30H) respectively. The ¹H NMR spectrum also shows singlets consistent with the geminal methyl groups and imine methyl group of the alkynyl indoles in both Pt complexes. The ¹³C NMR spectrum shows resonances assigned to the respective Cα and Cβ alkynyl carbons at δ 107.4 and 109.8 in 12. These alkynyl signals for 13 appear at δ 118.8 and 113.7. The IR spectrum of 12 shows a band at 2099 cm⁻¹ also confirming an alkynyl ligand. An absorption at 2105 cm⁻¹ in the IR spectrum of 13 is assigned to the ν(C≡C) . The HRMS contains peaks consistent with the protonated molecular ion at m/z 796.3510 and 1084.3472 for 12 (ES) and 13 (APCI) respectively.

In the synthesis of 12, extraction of the reaction residue with benzene and subsequent recrystallisation gave the coordination complex 12b (Supplementary Fig. S8). The complex is a polymeric compound consisting of 12 and CuI, with coordination at the nitrogen of the indole ligand. This compound’s connectivity was confirmed by single-crystal X-ray diffraction studies (Supplementary Fig. S9). Both ¹H and ³¹P NMR spectra were unremarkable compared with the parent 12. Examples of coordination complexes of 3H-indoles are rare, with only a few iridium, platinum and gold complexes reported.^[52–54]

Complex trans-[Pt(PCy₃)₂Cl₂]} 14 has previously been synthesised by the reaction of [Pt(COD)Cl₂] (COD, 1,5-cyclooctadiene) with PCy₃ in THF.^[55] Adapting the synthetic procedure for cis- and trans-PtCl₂(PEt₃)₂,^[56] the reaction of commercially available (NH₄)₂PtCl₄ and PCy₃ in toluene and water gives 14 in 83% yield (Scheme 3). Formation of the trans orientation is evinced by a single peak at δ 17.18 (J_PPt  2398 Hz) in the ³¹P{¹H} NMR spectrum. The increased steric bulk of PCy₃ hinders formation of the cis orientation.

Owing to the scarcity of bis-alkynyl complexes of the type trans-[Pt(PCy₃)₂(CC–Ar)₂], a brief optimisation synthesis of trans-[Pt(PCy₃)₂(CC–C₆H₅)₂] 15 was undertaken. The reaction of 14 and phenylacetylene in the presence of CuI, triethylamine and co-solvent CHCl₃ gave 15 in 59% yield (Scheme 3). The co-solvent was required owing to solubility issues of 14 in neat amine solutions. Other co-solvents tested were THF and CH₂Cl₂. The steric bulk of the PCy₃ ancillary ligands renders this reaction sluggish even at elevated temperatures. This observation is mirrored in a study that found only 50% conversion (30% isolated yield) to the bis-alkynyl Pt complexes trans-[Pt(PCy₃)₂(CC–C₆H₄SAc)₂] after 4 days at 60°C by similar methodology.^[57] The ³¹P NMR spectrum of 15 shows a single peak at δ 24.06 (J_PPt  2408 Hz). These conditions were employed to synthesise the Pt complex 16; however, this reaction took 21 days to complete and was low-yielding (Scheme 3).

A peak at δ 23.22 (J_PPt  2424 Hz) in the ³¹P NMR spectrum for 16 evinced a single trans-orientated complex. The ¹H NMR spectrum showed three resonances assigned to the aryl protons on the 3H-indole ligand at δ 7.34, 7.21 and 7.14. Broad unresolved resonances in the range δ 1.20–2.89 were assigned to signals from the cyclohexyl groups and the methyl groups on the 3H-indole. The ES mass spectrum shows a peak consistent with the protonated molecular ion at m/z 1120.6332.

The ruthenium complex 17 was synthesised by the reaction of [RuCl(dppe)₂]OTf with excess 2, in the presence of KO^tBu (Scheme 3). This reaction is analogous to established methodology using NaPF₆ and NEt₃^[44,58–60]; however, the base and halide abstracting reagent are combined as a single compound, KO^tBu. Another ruthenium complex, 18, was synthesised by reaction of trans-[RuCl₂(P(OEt)₃)₄] with 2, KPF₆ and HNⁱPr₂ in ethanol for 8 weeks (Scheme 3). The low yield is comparable with that of previous bis(alkynyl) complexes synthesised by this method.^[61] Although lithium alkynyl species can be used to obtain this type of alkynyl complexes in low to moderate yields,^[62] a simple extraction and crystallisation workup makes this method favourable.

The IR spectra of complexes of 17 and 18 showed absorptions assigned to ν(C≡C) bands at 2048 and 2088 cm⁻¹ respectively. Singlets in the ³¹P NMR spectrum of both complexes support trans arrangement of the ligands and are comparable with similar complexes.^[61,63] The ¹H NMR spectrum of 17 shows a slight upfield shift of the assigned signals for the geminal methyl groups at 1.12 ppm and the imine methyl group at 2.03 ppm compared with free ligand 2. This upfield shiftof the methyl groups was also seen in complex 18, but not to the same extent. Resonances assigned to the ancillary phosphine ligands in 17 were observed between δ 6.94–7.76 and 2.56 (m, 8H, PPh₂CH₂CH₂PPh₂) in the ¹H NMR spectrum. Resonances for the ethyl fragments on the phosphite groups for 18 were assigned to the quartet at δ 4.33 (CH₂) and the triplet at δ 1.21 (CH₃) (J  7.2 Hz) in the ¹H NMR spectrum. The mass spectrum (APCI for 17, ES for 18) of both complexes showed fragmentation consistent with the loss of one alkynyl ligand. In addition, 17 also showed a peak consistent with the protonated molecular ion at m/z 1263.3789 in the mass spectrum.

Numerous attempts to synthesise the buta-1,3-diyne bridged putative molecular wire 19 were unsuccessful (Scheme 3). These compounds are candidates for applications in molecular electronics, supramolecular chemistry, optoelectronics and conjugated polymers.^[64–66] Hay,^[67] Eglinton and Galbraith,^[68] Glaser^[69] and Pd-type^[70] coupling reactions failed to produce the homo-coupled product, returning starting material or unidentified decomposed species.

Electronic spectroscopy

The UV-Vis spectra obtained for complexes 4–7 all displayed similar features (Fig. 1). All complexes show a metal-to-ligand charge transfer (MLCT) just below 400 nm^[40] as well as a higher-energy transition below 300 nm. This higher-energy band is assigned to a π–π* (intraligand, IL) transition.

Fig. 1. 

UV-Vis spectra of compound 2, and complexes 4–7 in CH₂Cl₂.

The UV-Vis spectra of 11 and complexes 12, 13, 17 and 18 are shown in Fig. 2. Compound 11 shows absorption maxima at 343 and 365 nm. The platinum complexes 12 and 13 show a high-energy transition at ~300 nm and a lower-energy transition between 350 and 370 nm. The lower-energy transition was assigned as an LMCT from the π(C≡CAr) to the mixed π(C≡CAr)* and metal p orbitals.^[57,71,72] The ruthenium alkynyl complexes also exhibited a band of similar intensity between 350 and 370 nm, with a very intense band appearing below 300 nm for complex 17. The lower energy band for these and similar complexes was assigned as an MLCT and the high-energy transition as an IL charge transfer band.^[73–76]

Fig. 2. 

UV-Vis spectra of compound 11, and complexes 12, 13, 17 and 18 in CH₂Cl₂.

Molecular structures

The molecular structures of several compounds in this study were determined using single-crystal X-ray diffraction studies. Plots are shown in the figures below, and Tables 1 and 2 show selected bond lengths and angles.

The organic ligand 2 crystallised with two crystallographically distinct molecules in the unit cell (Fig. 3). The difference between the molecules lies in their clockwise or anticlockwise rotation along the C7–C7a–N1–C2 dihedral angle. However, given the near-planar arrangement of both molecules in the unit cell, this is unremarkable. This phenomenon was reported in a recent communication.^[77] The ethene compound 3 shows elongation of the N1–C2 bond and shortening of the C2–C21 bond consistent with conjugation along the N1–C2–C21–C22–N2 backbone (Supplementary Fig. S6). Both indole rings are essentially coplanar and arranged in an antiperiplanar arrangement.

Fig. 3. 

ORTEP representation of 2, and atom numbering scheme. Ellipsoids at the 50% probability level. Hydrogens removed for clarity.

Complex 4 crystallised with two CH₂Cl₂ solvate molecules (Fig. 4). One of the solvent molecules was modelled as disordered over one site. The additional solvate was hydrogen-bonded between C21–H21–Cl and CHCl₂–H–N1 with respective distances of 3.68(1) and 3.54(2) Å between the donor and acceptor atoms.

Fig. 4. 

ORTEP representations of (a) 4; (b) 5; (c) 6; (d) 8; and (e) 9, and atom numbering scheme. Ellipsoids at the 50% probability level. Hydrogens and solvates removed for clarity where appropriate.

The bond lengths and angles around the metal core of the complexes are consistent with previously reported analogous alkynyl complexes.^[20,39,41] Bond lengths along Ru–CC–C5 are not significantly different between complexes 4 and 5 (Fig. 4). A smaller Fe–C bond length in 6 (Fig. 4) can be ascribed to a smaller atomic radius.^[17] The usual linear arrangement along the ruthenium–alkynyl backbone is observed, with a maximum deviation from 180° observed for 5 by −5.7°. The angles subtended by the phosphine atoms at the ruthenium atoms exemplify the pseudo-octahedral environment around the metal centres. The P–Ru–P angles are 100.27(7)° for the bis-triphenylphosphine and 83.36(2)° for the dppe. The shorter Ru–P bonds in 5 are consistent with the increased electron density at the metal centre, favouring back-bonding with the phosphine ligands. Bond lengths and angles for complex 8 (Fig. 4) are also unremarkable compared with similar complexes.

The Au–C and CC bond lengths and the bond angles about the P–A–CC chain for 9 (Fig. 4) are within the ranges for reported Au^I alkynyl complexes.^{[34,42,78–81]} A near-linear arrangement of alkynyl and phosphine ligands at the gold centre is observed. The slightly increased deviation from linearity at the Au–CC moiety has been associated with crystal packing forces.^[79,81]

The relative orientation of the indole ring is influenced by the steric encumbrance of the ancillary ligands in the solid state. The torsion angle between the centroid of the cyclopentadienyl ligand, metal atom, and the C5 and C4 atoms is used to highlight the orientation of the indole ring (Supplementary Fig. S10). In complex 8, the indole ring lies perpendicular to the cyclopentadienyl ligand. In complex 4, introducing the bulky triphenylphosphine ligands causes the indole to rotate close to 180°. In complex 5, the indole ligand lies in the plane of the cyclopentadienyl ligand owing to the increased bulk of the permethylated Cp ligand and less sterically demanding dppe. Similar parallel conformations for functionalised ethynyl aryl rings in ruthenium σ-alkynyl complexes have been reported.^[46] This is also reflected in the orientation of the indole ligand in complex 6.

Compound 11 (Fig. 5) mirrors the solid-state structure of the recently reported 1,4-bis(5-ethynyl-3,3-dimethyl-2,3-dihydrobenzo[b]thiophenyl)benzene.^[60] The central phenyl ring possesses an inversion centre tilted by 35.0(3)° from the outer indole ring. Simpler phenylene-ethynylene structures seem to adopt a more coplanar orientation in the solid state.^[60,82] Intermolecular hydrogen bonding between the N1 and imine methyl group dominates the crystal packing.

Fig. 5. 

ORTEP representation of (a) 11; (b) 12; (c) 13; (d) 15; (e) 17; and (f) 18, and atom numbering scheme. Ellipsoids drawn at 50% probability. Hydrogens and solvate omitted for clarity where appropriate.

Solid-state structures showed the expected square planar geometry around the metal centre in the platinum complexes 12, 13 and 16 (Fig. 5) as well as 15 (Supplementary Fig. S7). All complexes except 15, which exhibits a slight distortion, sit on crystallographic inversion centres and have similar bond angles that deviate only slightly from idealised square planar geometry. Hydrogen bonding between N1 and methanol solvent dominates the crystal lattice of 16. The M–P bond length is shorter in 12 compared with the other Pt complexes in this study. In general, characteristic bonds lengths and angles along the metal–alkyne fragment were unremarkable and were within the range reported in the literature.^[83–88]

Complex 17 possesses a centre of inversion about the ruthenium centre (Fig. 5). The plane of the arylethynyl rings of the indole lies between the dppe bidentate ligands. The angle between the plane of the arylethynyl ligand and the plane bisecting the dppe ligand, θ, as described recently,^[63] is 76.8°. In this orientation, there is a better overlap of the ligand π-system with the metal centre. All other interatomic parameters are not significantly different from previously described trans-Ru bis(alkynyl) complexes.

Complex 18 shows the rod-like structure and pseudo-octahedral metal environment consistent with tetrakis(triethylphosphite) complexes (Fig. 5). Moreover, the bond distances and angles are in the range previously described for these complexes.^[61] Interestingly, the angle between planes of the arylethynyl rings shows only a slight deviation from a coplanar orientation. Most complexes of this type adopt a twisted conformation between the aromatic groups.^[60,61] The M–P bond lengths are shorter in 18 than in 17 as a result of Ru–P π back-bonding.

Bonds around the metal centres show slight variations. M–P bond lengths are longer in 17 than the Pt complexes, comparable with the bond lengths in 18. There are slight geometric differences about the respective metal centres; however, these are less remarkable moving away from the metal. Bond lengths and angles in the indole rings show no significant differences between the metal complexes. All indole rings of the arylethynyl complexes lie in the same plane, allowing delocalisation along the metal–alkyne backbone.

Electrochemistry

The electrochemical properties of the indole complexes were studied to compare the redox potentials as a function of the metal–ligand fragment (Table 3). The cyclic voltammograms of all complexes show single reversible oxidations at potentials between −640 and −60 mV (v. Fc/Fc⁺) (Eox1) and additional irreversible oxidations between 580 and 780 mV (v. Fc/Fc+) (Eox2). The oxidation potentials are consistent with the indoline moiety being electron-donating.^[39] The Cp* ligand makes the oxidation potential 200–300 mV more cathodic in both systems. The iron complexes undergo the first oxidation at potentials 300–400 mV more cathodic than their isostructural ruthenium complex. The second oxidation, presumably at the nitrogen of the indoline, is more dependent on the metal in the ruthenium complexes than in the iron complexes. This behaviour is predictable given the larger degree of delocalisation of electron density over the ruthenium acetylide.

Single-molecule conductance studies

Normalised conductance histograms (on the left) and density plots (on the right) are shown in Fig. 6 and 7 for compounds 11 and 12 respectively. The conductance histograms show clear quantum of conductance G₀ ((G₀ = 2e²/h = 77.5 μS) (e, elementary charge; h, Planck constant)) peaks at log(G/G₀) = 0, which correspond to the single channel Au–Au metal atom point contact. The 2-D density plots display log(conductance) v. electrode separation during the scanning tunnelling microscopy (STM) tip retraction from the metal point contact to 2.0 nm extension. As seen from these plots, there is little evidence of extended molecular junction formation and, hence, it is not possible to attribute a conductance peak to these compounds. Possible reasons for the absence of an apparent junction formation for these compounds could be a low junction formation probability or conductance values falling below the noise floor of ~10^−5.5 G₀. A low junction probability could result from steric hindrance from the vicinal methyl group to the proposed nitrogen dative binding site of the specific indole terminal groups employed here. Low binding strength of amine contacting groups has been attributed to the steric influence of hydrogens.^[89] However, if a junction is forming, then the conductance itself could be lower than the noise floor (10^−5.5 G₀), which would mean that it would lie outside the measurement window of the conductance determination. The molecular backbone is chemically similar to previous studied compounds (and their conductance): trans-Pt(CC-CC₆H₄-CC-SiMe₃)₂(PEt₃)₂ (3.2 ± 1.3 × 10⁻⁵ G₀),^[56] trans-Pt(CC-cyclo-3-C₄H₃S)₂(PEt₃)₂ (2.70 ± 0.66 × 10⁻⁴ G₀) and 1,4-C₆H₄(CC-cyclo-3-C₄H₃S)₂ (2.83 ± 0.65 × 10⁻⁴ G₀),^[90] all of which displayed molecular conductance values greater than 10⁻⁵ G₀ (i.e. greater than the noise floor of our present measurements). Additionally, oligophenylethylenes (OPEs) with a similar backbone structure to compound 12 show conductance values of 2.6 ± 1.6 × 10⁻⁵ to 3.6 ± 1.6 × 10⁻⁵ G₀.^[91] Therefore, it seems rational to assume the primary cause for no apparent molecular junction formation is inherent to the new contact group. This lack of junction formation may be related to insufficient dative bonding between the N of the indole terminal groups and the gold electrodes or may arise from steric hindrance of this bonding as a result of the vicinal methyl group.

Fig. 6. 

Normalised 1-D conductance histogram (left) and 2-D conductance-displacement density plot (right) for 11 using STM-BJ at 100 mV bias. More than 3000 individual breaking traces were compiled with no data selection.

Fig. 7. 

Normalised 1-D conductance histogram (left) and 2-D conductance-displacement density plot (right) for 12 using scanning tunnelling microscopy break junction (STM-BJ) at 100 mV bias. More than 3000 individual breaking traces were compiled with no data selection.

Steric hindrance can impede potential anchoring groups binding to gold contacts; for example, previous studies have noted that the high conductance hollow geometric formation seen in bipyridyl ethyne is inhibited by introducing vicinal C–F bonds.^[92] It is instructive to consider how junctions might form in the STM-BJ experiment for compounds 11 and 12. Analogous pyridyl anchoring groups (e.g. in 4,4′ bipyridine molecular bridges with gold contacts^[93]) evolve from a face-on alignment of the pyridyl ring to an end-on dative binding of the terminal N atom to a Au apex atom. As an anchor group slides across the pyramidal tip through a side configuration, it can jump to an apex position to form the final molecular junction.^[94] If a similar mechanism is assumed here for our indole contact group, this last jump could be impeded by the vicinal methyl group, leading to a complete rupture. Another possible issue might be the chemical stability of the molecules under the experimental conditions needed to form Au–molecule–Au junctions. A pyrrolic-type indole-terminated diaryltetrayne has been shown not to exhibit junction formation owing to the chemical and thermal instability of the moiety under the employed experimental conditions.^[95] However, given the ambient stability of the compounds synthesised in the present study, we believe that that degradation during single-molecule conductance experiments is less likely to cause absence of junction formation features. The last reason we postulate for the absence of junctions for compounds 11 and 12 is the slight increase in basicity of the indole contact group (pKa of the conjugated acid 6.33) compared with the pyridine contact group (pKa of the conjugated acid 5.25). Hard–soft acid–base (HSAB) theory has been shown to predict the formation of metal–molecule junctions with lone pairs on the anchoring group (e.g. pyridyl, amine) and gold electrodes.^[96] Although pyridine is of an ‘intermediate’ character on the hard–soft continuum, it seems possible that the increase in basicity of the 3H-indole reduces its affinity for the gold electrodes if there is a less favourable HSAB match.

Spectroelectrochemistry

Spectroelectrochemical methods have become useful tools to determine the nature of redox processes and redox products for metal alkynyl complexes.^[97] The complexes 4, 7 and 17 were chosen for spectroelectrochemical investigations to act as models across the series of compounds in this study owing to the ease of comparison with other similar species studied in the literature.

The infrared spectroelectrochemical spectra of complexes 4 and 7 during the first oxidation are illustrated in Fig. 8. The IR spectra of the neutral complex 4 shows a single absorption at 2065 cm⁻¹ assigned to the ν(Ru–C≡C) band. On oxidation, the band decreases in intensity. A small IR absorption for the ν(Ru–C≡C) band was observed for the oxidised complex [4]⁺ at 1920 cm⁻¹ (Fig. 8a).^[39] Complex [4]⁺ was unstable on the timescale of spectroelectrochemical experiments, with decay of the band assigned to the ν(Ru–C≡C) at 1920 cm⁻¹ over a short period of time. However, we can confidently assign this band to ν(Ru–C≡C) as reduction exchanges this band for the ~2060 cm⁻¹ of complex 4 (Fig. 8b). We note also the possible formation of the [Ru(Cp)(PPh₃)₂(CO)]⁺ cation on the appearance of the band at ~1970 cm⁻¹ (Fig. 8b). For the iron complex 7, two active ν(Fe–CC) modes were observed on oxidation. A shift of Δ80 cm⁻¹ in the ν(Fe–CC) absorption band is in agreement with iron complexes of the type Fe(CCR)(dppe)Cp, and is consistent with a large degree of metal character in the first oxidation process (Fig. 8c).^[39] The appearance of two broad active ν(Fe–C≡C) modes can be rationalised as a population of local states on the potential energy surface, where several accessible conformers differ by the orientation of the arylethynyl ligand with respect to the metal orbitals.^[98]

Fig. 8. 

IR spectra of complexes (a, b) [4]ⁿ⁺ (n = 0, 1); (c) [7]ⁿ⁺ (n = 0, 1); and (d) [17]ⁿ⁺ (n = 0, 1) recorded in a spectroelectrochemical cell (CH₂Cl₂/0.1M Bu₄NPF₆).

The IR spectra on oxidation of 17 to [17]⁺ is illustrated in Fig. 8d. Complex 17 is characterised by an absorption at 2054 cm⁻¹ assigned to the ν(Ru–C≡C) stretch. On oxidation to the monocations, [17]⁺, the ν(Ru–C≡C) assigned absorption band shifts by 165–1889 cm⁻¹. This large shift is indicative of the large contribution of the alkynyl ligand to the oxidation process. The broad and asymmetric nature of the ν(Ru–C≡C) envelope in [17]⁺ is consistent with a range of conformers that vary as a result of the relative orientation the arylethynyl ligands with respect to the metal centre.^[63] A weak feature between 2049 and 2028 cm⁻¹ is consistent with additional conformers with electronic structures localised over one arylethynyl ligand.^[60] The additional absorptions at 1601 and 1558 cm⁻¹ are likely assignable to aryl breathing modes, further consistent with the large contribution of the ethynyl ligand in the oxidation process.

General

All reactions were carried out under high-purity argon or nitrogen using Schlenk techniques and oven-dried glassware. All solvents for reactions were dried by distillation under argon over an appropriate drying agent or by an Innovative Technologies solvent purification system (SPS) and deoxygenated (freeze/pump/thaw or sparging) before use. Hexanes and EtOAc for column chromatography were distilled before use. All other solvents were of AR grade and used without further purification. Unless otherwise stated, no special precautions were taken to exclude air or moisture during workup and purification. All other reagents were used as received from their supplier. Flash column chromatography was carried out using silica gel 60 (0.04–0.063 mm) or alumina (Brockmann activity IV, basic). Thin-layer chromatography was carried out using Merck silica gel 60 F₂₅₄ pre-coated aluminium sheets.

Instruments

¹H, ¹³C{¹H}, ³¹P{¹H} and ¹⁹F NMR spectra were recorded at 25°C on a Bruker 600, 500  or 400 MHz spectrometer. Chemical shifts (δ) are referenced to the internal undeuterated residual solvent signal (¹H), deuterated solvent signal (¹³C) or external standard (³¹P, 85% H₃PO₄). First-order multiplets are as follows: s, singlet; d, doublet; t, triplet; q, quartet. Apparent multipets are labelled with the prefix ‘app’. Broad signals are labelled with the prefix ‘br’. Multiplets of higher order or that were undeterminable are labelled as m. Coupling constants (J) are reported in hertz (Hz). Signal assignments were made with help from HSQC (heteronuclear single quantum coherence), HMBC (heteronuclear multiple bond correlation) and ¹H COSY (correlation spectroscopy) experiments, spectrum prediction software and literature precedent. Infrared spectroscopy was carried out with neat products using an ATR module-fitted FTIR spectrometer. ESI or APCI mass spectrometry was carried out on a Waters LCT Premier time-of-flight (TOF) spectrometer. Electron-impact ionisation was carried out using a Waters GCT Premier or Shimadzu GCMS QP2010. In most cases, HPLC grade acetonitrile was used as the solvent. If needed, a couple of drops of dichloromethane were used to help solubilise the compound for analysis. Melting points were collected on a Buchi M-565. Elemental analyses were performed by Stephen Boyer, London Metropolitan University, London, United Kingdom.

Synthesis

Compounds 1, [RuCl(dppe)₂]OTf,^[44] Ru(dppe)(Cp*)Cl,^[99] CpRu(PPh₃)₂(Cp)Cl,^[100] Fe(dppe)(Cp*)Cl,^[99] Fe(dppe)(Cp)Cl,^[99] PPh₃AuCl^[101] and Ru(CO)₂(Cp)Cl^[102] were synthesised according to literature procedures.

Electrochemistry

Cyclic voltammetry was performed in 3 mL reaction vials (eDAQ) in a glovebox under an argon atmosphere using a Pt working disc electrode (1 mm diameter; eDAQ) and Pt-coated Ti rods as the counter and pseudo-reference electrodes (eDAQ). The surface of the working electrode was polished with an alumina paste (0.05 μm alumina powder) on a polishing cloth, rinsed with MilliQ water, ethanol and CH₂Cl₂ and dried under a stream of nitrogen before use. Ferrocene (Fc), decamethylferrocene (Fc*) or diacetylferrocene (Ac₂Fc) were used as internal standard and all potentials are reported v. the Fc/[Fc]⁺ couple. Measurements were conducted in CH₂Cl₂ solution containing 0.1 M tetrabutylammonium hexafluorophosphate (Bu₄NPF₆) as the supporting electrolyte. The electrochemical potential within the cell was controlled using a PalmSens EmStat3+ potentiostat and PSTrace software.

Spectroelectrochemistry

All spectroelectrochemical (SEC) measurements were conducted using an OTTLE cell of Hartl design^[104] using dry CH₂Cl₂ as the solvent and 0.1 M Bu₄NPF₆ as the supporting electrolyte. For each experiment, the SEC cell was filled with the analyte solution inside a glovebox and sealed before being removed for operation in the spectrometer. UV-Vis/NIR (NIR, near-infrared) spectra were recorded on an Agilent Cary 5000 spectrometer. IR spectra were recorded on an Agilent Cary 660 FT-IR/NIR spectrometer. The electrochemical potential within the cell was controlled using a PalmSens EmStat3+ potentiostat. For each measurement, the potential was incrementally increased, and a spectrum was recorded at each step.

Crystallography

X-ray diffraction measurements were carried out at 100 K on an Oxford Diffraction Gemini-R Ultradiffractometer fitted with a MoKα or CuKα radiation source at the University of Western Australia. Following analytical absorption corrections and solution by direct methods, the structures were refined against F2 with full-matrix least-squares using the SHELXL software suite. The .cif files have been deposited at the Cambridge Crystallographic Data Centre, numbers CCDC 2255621–2255635.

Single-molecule conductance

We performed STM-BJ^[105] experiments to determine single-molecule conductance. In these experiments, the Au STM tip is driven into the Au STM substrate to form a metallic contact, and then it is withdrawn at a constant speed (13 nm s⁻¹ in this study). As the tip is retracted, the metallic contact thins to a point contact having conductance of G₀, which is then ruptured to yield two atomically sharp nanoelectrodes. The experiment is performed in a solution of the target molecule (1 mM in 1,2,5-trichlorobenzene in this study), which can self-assemble in the freshly formed nanogap to fabricate a molecular junction. The tip is further withdrawn so that the junction is stretched to its maximum length, and eventually ruptured. The tip is then driven again into the substrate and the process is repeated. Data (as G = I/V) (G, conductance; I, current; V, voltage) are continuously recorded and compiled in statistical 1-D conductance histograms and in 2-D conductance/electrode separation plots.

A modified commercial instrument (Keysight Technology 5500 SPM) was used in this study. The original electronics were used to apply the junction bias (V) and to drive the piezoelectric actuator. A custom current amplifier based on the design originally introduced by Mészáros et al.^[106] was used as I/V converter, and the resulting four-channel data were acquired using a National Instruments DAQ (NI-9215). We used Au/Cr/glass slides (Arrandee) as substrate and 99.999+% Au wire (Goodfellow Cambridge) as tip.