Tricoordinate Coinage Metal Complexes with a Redox‐Active Tris‐(Ferrocenyl)triazine Backbone Feature Triazine–Metal Interactions

Abstract 2,4,6‐Tris(1‐diphenylphosphanyl‐1’‐ferrocenylene)‐1,3,5‐triazine (1) coordinates all three coinage metal(I) ions in a 1:1 tridentate coordination mode. The C3‐symmetric coordination in both solid state and solution is stabilised by an uncommon cation–π interaction between the triazine core and the metal cation. Intramolecular dynamic behaviour was observed by variable‐temperature NMR spectroscopy. The borane adduct of 1, 1BH3, displays four accessible oxidation states, suggesting complexes of 1 to be intriguing candidates for redox‐switchable catalysis. Complexes 1Cu, 1Ag, and 1Au display a more complicated electrochemical behaviour, and the electrochemical mechanism was studied by temperature‐resolved UV/Vis spectroelectrochemistry and chemical oxidation.


General Procedures Syntheses
All reactions and manipulations were carried out under an atmosphere of either nitrogen or argon using standard Schlenk line techniques unless stated otherwise. Thin-layer chromatography (TLC) with silica gel 60 F254 on glass or aluminium sheets available from Merck KGaA was used for monitoring the ligand synthesis reactions. Column chromatography was performed using a Biotage Isolera 1 automatic purification system with SNAP (silica, particle diameter: 0.040 to 0.065 mm) and SNAP Ultra (silica, sphe rical particle, diameter: 0.025 mm) cartridges using solvents purged with nitrogen prior to use. The fractions were detected by an integrated UV/Vis detector. Molecular sieves (3 and 4 Å) were activated at 300 °C in vacuo for a minimum of 3 h. Dry, oxygen-free solvents (THF, acetonitrile, CH2Cl2, Et2O, hexanes, and toluene) were obtained from an MBraun Solvent Purification System MB SPS-800 and directly stored over 4 Å molecular sieves, except for THF, which was further distilled from potassium/benzophenone and stored over 4 Å molecular sieves. 1,2-Dichloroethane was degassed using the freeze-pump-thaw method and stored over 4 Å molecular sieves. For the use in NMR measurements, THF-d8 was distilled from potassium/benzophenone, CD2Cl2, CDCl3 and CD3CN were dried by stirring over P2O5 at room temperature for several days, followed by vacuum transfer into a storage flask and degassing by the freezepump-thaw method. The solvents were stored over 4 Å (THF-d8, CD2Cl2, CDCl3) or 3 Å (CD3CN) molecular sieves.
NMR spectra were recorded with a BRUKER Avance III HD 400 MHz NMR spectrometer at 25 °C (frequencies of 1 H: 400.13 MHz; 11 B: 128.38 MHz; 13 C 100.63 MHz; 19 F: 376.53 MHz: 31 P: 161.99 MHz). Pseudo-triplets of ferrocenyl protons are abbreviated as pt and their observable H,H coupling constants are given. TMS was used as the internal standard in the 1 H and 13 C NMR spectra, and spectra of all other nuclei were referenced to TMS using the Ξ scale. [8] Electrospray ionisation mass spectrometry was performed with an ESI ESQUIRE 3000 PLUS spectrometer with an IonTrap analyser from Bruker Daltonics, or a MicroTOF spectrometer from Bruker Daltonics with a ToF analyser in positive mode. As solvents for the measurements, THF, CH2Cl2, CH3CN, MeOH, or mixtures of these solvents were used. Dry, oxygen-free solvents were used for airsensitive species. Elemental analyses were performed with a VARIO EL elemental analyser from Heraeus. Melting points were determined with a Gallenkamp MPD350·BM2.5 melting point device and are reported uncorrected. FTIR spectra were obtained with a PerkinElmer FT-IR spectrometer Spectrum 2000 as KBr pellets and with a Thermo Scientific Nicolet iS5 with an ATR unit in the range from 4000 to 400 cm −1 . UV/Vis spectra were recorded on a PerkinElmer UVV-IS-NIR Lambda 900 spectrometer in quartz cuvettes (d = 10 mm). Sample concentration was in the range of 3•10 -5 mol•L -1 .

Computational methods
All calculations were carried out with the ORCA program package. [16] All geometry optimisations were performed at the BP86-D3BJ/def2-TZVP [17] level of theory in the gas phase. For calculations involving silver and gold the respective effective core potentials, namely def2-ECP, [18] were used. Frequency calculations were carried out to confirm the nature of stationary points found by geometry optimisations. Density fitting techniques, also called resolution-of-identity approximation (RI), [19] were used for GGA calculations, whereas the RIJCOSX [20] approximation was used for time-dependent DFT (TDDFT) calculations. Wiberg bond indices (WBI) were calculated using the built-in NBO3.1. [21] module of Gaussian09 [22] on the geometries obtained with ORCA (using the BP86 functional in combination with the LANL2DZ [23] basis set). 1 H NMR shieldings were calculated at the TPSS/def2-TZVP [24,17c] level of theory in the gas phase using the calculated proton shift of tetramethylsilane as reference.

Synthesis of 2,4,6-tris{1-(diphenylphosphanyl)-1'-ferrocenylene}-1,3,5-triazine (1)
A stirred solution of 1.41 g (1.62 mmol, 1.00 eq.) 4 in 50 mL THF was cooled to −80 °C (ethyl acetate/N2(l)) and 3.47 mL (1.54 mol•L −1 , 5.35 mmol, 3.30 eq.) nBuLi in n-hexane were added dropwise over the course of 30 min, resulting in a strong intensification of the red colour. The mixture was kept stirring for 2 hours at −80 °C, followed by the dropwise addition of 1.23 g (5.59 mmol, 3.45 eq.) chlorodiphenylphosphane dissolved in 20 mL THF at the same temperature. After slowly warming to room temperature overnight and stirring at 60 °C for 1 hour, TLC (CH2Cl2/hexanes, 4:1) indicated full conversion of 4. For quenching, a degassed saturated aqueous solution of NH4Cl was added to the fervently stirred mixture via cannula. The phases were separated, and the aqueous phase was extracted with diethyl ether (2x 10 mL). The combined organic phases were dried over degassed MgSO4, filtered, and then the product mixture was adsorbed on Celite® and subjected to column chromatography (CH2Cl2/hexanes, gradient 20-100%). Crystals suitable for single crystal X-ray diffraction analysis were grown from a so-obtained fraction at 7 °C. Following the unification of the fractions and the removal of solvent under reduced pressure, 1 was filtered over degassed silica using CH2Cl2, and the solvent removed in vacuo, yielding 1 (1.00 g, 52% yield).

Synthesis of 2,4,6-tris{1-(diphenylphosphanyl borane)-1'-ferrocenylene}-1,3,5-triazine (1BH3)
A stirred solution of 90 mg (76 μmol, 1.0 eq.) 1 in 10 mL CH2Cl2 was cooled to 0 °C (ice/water) and 0.13 mL (2.0 mol•L −1 , 270 μmol, 3.5 eq.) BH3•SMe2 in THF were carefully added. After the gas evolution had ceased, the reaction mixture was kept stirring at room temperature for 45 min after which TLC (CH2Cl2/hexanes 3:1) indicated full conversion of 1. Further work-up was carried out under ambient conditions. The volatiles were removed under reduced pressure and the residue was filtered over a plug of silica, yielding pure 1BH3 (74 mg, 80% yield) after solvent removal. Crystals suitable for single crystal X-ray diffraction analysis were obtained by slow evaporation of a solution of 1BH3 in CH2Cl2/hexanes.                The expansion displays the Cu-mediated 2 JP,P coupling of Cuop resulting in the multiplet of an AA' spin system in detail. Figure S23. 31

Synthesis of {2,4,6-tris(1-diphenylphosphanyl-1'-ferrocenylene)-1,3,5-triazine-κP 3 } silver(I) triflate (1Ag)
Under stirring and protected from direct light, a solution of 150 mg (127 μmol, 1.15 eq.) 1 in 12 mL CH2Cl2 was added to 28.3 mg (110 μmol, 1.00 eq.) AgOTf and kept stirring at room temperature for 1 hour, after which the protection from light was discontinued. The reaction mixture was filtered via cannula, layered with 36 mL toluene and kept at 7 °C overnight. The soobtained crystalline material was isolated by filtration and dried in vacuo overnight at 40 °C, causing the deep red crystals to brittle, yielding a light-insensitive red powder (131 mg, 82% yield) which can be handled in air, but which was stored under nitrogen. Crystals suitable for single crystal X-ray diffraction analysis were obtained from layering a 1,2-dichloroethane solution of 1Ag with toluene, stored at 7 °C for several days.

Synthesis of {2,4,6-tris(1-diphenylphosphanyl-1'-ferrocenylene)-1,3,5-triazine-κP 3 } gold(I) triflate (1Au)
Under stirring and protected from direct light, a solution of 235 mg (198 μmol, 1.13 eq.) 1 in 20 mL CH2Cl2 was added to 110 mg (175 μmol, 1.00 eq.) [Au(nbe)3]OTf and kept stirring at room temperature for 1 hour, after which the protection from light was discontinued. The reaction mixture was filtered via cannula, layered with 50 mL toluene and kept at 7 °C for 2 days. The soobtained crystalline material was isolated by filtration and dried at 40 °C in vacuo overnight, causing the deep red crystals to brittle, yielding a light-insensitive red powder (190 mg, 71% yield) which can be handled in air, but which was stored under nitrogen. Crystals suitable for single crystal X-ray diffraction analysis were obtained from layering a 1,2-dichloroethane solution of 1Au with toluene, stored at 7 °C for several days.
The volatiles were removed in vacuo, the solid violet residue was washed with hexanes (2x 5 mL) and dried in vacuo. Crystals suitable for single crystal X-ray diffraction analysis were obtained from layering a 1,2-dichloroethane solution of 5 with hexanes, stored at 7 °C for several days and subsequent tempering at 35 °C for 4 days. No yield was determined, and the product obtained by crystallisation was found to be impure by means of NMR spectroscopy, yet an attributable set of spectra could be recorded and analysed at 25 °C ( 11 B, 19     . Figure S46. 31 P{ 1 H} NMR spectrum of 5 in CD2Cl2 recorded at 25 °C. The asterisks (*) denote signals attributable to a second component in solution.  (2) 130 (2) 130 (2) 130 (2) Crystal system / Space group   Ligand 1 crystallises with no crystallographic symmetry, but key structural parameters for the three ferrocenylene moieties are in good accordance with each other and within the typical values reported for related compounds. The all-syn arrangement of the three 1diphenylphosphanyl groups with respect to the central triazine core does not seem to arise from either intra-or intermolecular contacts. For the closely related compounds 1,3,5-tris(ferrocenyl)benzene [25] and 2,4,6-tris(ferrocenyl)pyridine [26] syn,anti,anti conformations are found in the solid state with two ferrocenyl groups on one side of the arene core and the third on the other, while 2,4,6-tris(ferrocenyl)-1,3,5,2,4,6-triselenatriborinane [27] (also containing an aromatic six-membered core) crystallises in the all-syn conformation, as does the 1,3,5-tris(ferrocenyl)benzene moiety with the benzene core η 6 -bound to a RuCp fragment (V in Figure 1 in the main article). [28] The tris(borane) adduct 1BH3 crystallises with crystallographic C3 symmetry, obviously also resulting in an all-syn conformation. Its structural parameters are in line with expectations towards such phosphanyl boranes.
[b] Ct C denotes the calculated centre of gravity of the carbon-substituted cyclopentadienyl ring.
[c] Ct P denotes the calculated centre of gravity of the phosphorus-substituted cyclopentadienyl ring.
[d] Angle between the mean planes through the substituted cyclopentadienyl (Cp R ) rings.
[e] Torsion about the Cp P (centroid)⋯Fe⋯Cp C (centroid) axes. Figure S49. Schematic representation of key structural parameters for the compounds under investigation.  Figure 2 in the main article).
[a] Ct C denotes the calculated centre of gravity of the carbon-substituted cyclopentadienyl ring.
[b] Ct P denotes the calculated centre of gravity of the phosphorussubstituted cyclopentadienyl ring.
[c] Angle between the mean planes through the substituted cyclopentadienyl (Cp R ) rings.
[d] Torsion about the Cp P (centroid)⋯Fe⋯Cp C (centroid) axes, given as its modulus.
Comparing oxidation product 5 to its starting material 1Au is very instructive. The two BAr F 4 − anions compensate the positive charges located at the gold(I) ion and the protonated triazine core (N1) and not, as could have been expected from the oxidation of 1Au, positive charges located at any of the iron centres. This is backed by the Ct C/P -Fe distances (Table S6) which do not differ strongly between 1Au and 5 and among themselves as would be the case if any of the iron centres in 5 would be in oxidation state +III. The most obvious change relates to the change of denticity from 3 in 1Au to only 2 in 5; accordingly, the P-Au-P angles change from just below 120° in 1Au to 168° in 5, in line with the loss of the C3N3•••Au(I) interaction in 1Au. The deviation of the ideal linear binding mode (180°) is most likely due to a weak interactions between N(2) and Au(1) (d = 4.412(2) Å). This bidentate mode seems to put slightly less strain on the general geometry of ligand 1 than the tricoordinate mode found in 1M, as the more relaxed values of Θ (closer to 0°) and α (closer to 180°) suggest. The P-O distance of 1.500(2) Å fits well with a previously reported value (1.4866(9) Å) for 1diphenylphosphoryl-1'-cyanoferrocene, as do the other relevant structural parameters, confirming the phosphoryl moiety in 5. [29] Relating to the hydrogen bond in 5 between the phosphoryl group and the protonated triazine, all its parameters fall within the expected range, [30] allowing it to be classified as a heteronuclear resonance-assisted hydrogen bond (RAHB), given the ferrocenylene group is treated as a resonant spacer in the sense of Gilli. [31] It is worth noting that the protonation of N(1) and its involvement in hydrogen bonding lead to a distortion of the otherwise regular C3N3 geometry that can be observed in 1Au. The C(1/2)-N(1) bonds become elongated, while the other non-involved C-N bonds shorten; in the same way, the torsion angles in the C3N3 ring increase (7.3° to −5.6° in 5 vs. 3.0° to −3.9° in 1Au). This puckering is also observable in N(1) moving away from a calculated C3N2 plane by 0.07 Å and from H(1) being away from a calculated C3N3 plane by 0.253 Å.

CSD Searches and Results for Comparison
In the following tables (Table S7 -Table S9), molecular structures of trigonal-planar coinage metal complexes with three phosphane ligands are assembled and their individual and average P-M-P bond angles β and <β> as well as their individual and average P-M bond lengths are listed. For the CSD search, the number of ligands on the metal M was not restricted. No restraints on the P-M bond lengths were set. The three bond angles β were set to a range of 115 to 125 ° each, to retrieve only trigonal-planar structures. Metal clusters and ligands other than phosphanes (i.e., phosphanides and (aromatic) phosphacycles) were manually sorted out. Additional donor atoms or groups in close contact to the metal M are indicated, and the column "1:1 Complex" is checked when a tris-phosphane ligand was found to bind the metal.
The number of such complexes is obviously rather small given the vast number of (phosphane) complexes of the coinage metals. As becomes apparent, trigonal-planar 1:1 complexes are even rarer and seem to require the presence of a supporting apical ligand or group (e.g., an arene). Table S7. Comparison of solid-state molecular structures of trigonal-planar tris-phosphane Cu I complexes regarding (and sorted according to) their Cu-P bond lengths and P-Cu-P bond angles β1-3 (set to 115-125°). Only trivalent phosphane donors were considered, i.e. phosphacycles and phosphanides are not part of this overview. CSD entries with more than one CuP3 unit in the same solid-state structure are labelled with (1) and (2).  (2) Å as a weak Z-type interaction. [32] Table S8. Comparison of solid-state molecular structures of trigonal-planar tris-phosphane Ag I complexes regarding (and sorted according to) their Ag-P bond lengths and P-Ag-P bond angles β1-3 (set to 115-125°). Only trivalent phosphane donors were considered, i.e. phosphacycles and phosphanides are not part of this overview. CSD entries with more than one AgP3 unit in the same solid-state structure are labelled with (1) and (2). [a] In these cases, the silver(I) ion can be considered to form part of a metalloligand (FAYYIO, [33] TUGTUK, [34] MUCVOC, [35] FUJQIM [36] ).

CSD Identifier
[b] The authors state that "the N atom is non-interactive at 2.662 Å", [37] the lone pair of electrons at nitrogen is directed towards the silver(I) ion and the distance is significantly below the sum of the van der Waals radii (3.27-4.19 Å), [38] yet also clearly above the sum of the covalent radii (2.156 Å). [39] [c] Part of a disordered tetrafluoroborate anion; the authors state a 50% probability of the Ag-F bond. [40] Table S9. Comparison of solid-state molecular structures of trigonal-planar tris-phosphane Au I complexes regarding (and sorted according to) their Au-P bond lengths and P-Au-P bond angles β1-3 (set to 115-125°). Only trivalent phosphane donors were considered, i.e. phosphacycles and phosphanides are not part of this overview. CSD entries with more than one AuP3 unit in the same solid-state structure are labelled with (1) and (2). [a] In these cases, the palladium(0) atom (IZOGAF [41] ), the silver(I) (AVACUV [42] ) or gold(I) ions (DUVYUO, [43] HOTZUL, [44] OMUGAG and OMUFUZ [45] ) can be considered to form part of a metalloligand.

CSD Identifier
[b] The authors state that there is "at most a weak interaction with the nitrogen atom of the NP3 ligand" considering the Au-N distance of 2.683(6) Å. [37] This distance is significantly below the sum of the van der Waals radii (3.21-3.98 Å), [38] yet also clearly above the sum of the covalent radii (2.07 Å). [39] [c] These compounds could be considered unsupported 1:1 trigonal-planar Au I complexes if the Cu I (SEKSOQ) and Cd II ions (SEKVOT) were to be considered as part of a metalloligand. [46] [d] Even though the CSD does not recognise the element-Au I distances as signifiers of true bonds in all cases, the group of Bourissou presents convincing arguments for treating these distances, shorter than the respective sum of the van der Waals radii in all cases (except for ROZREE [47] ), as Z-type interactions. [32,47,48] The following two tables, Table S10 (coinage metals) and Table S11 (all other metals), give an overview of published solid-state structures of metal complexes featuring C3N3•••M contacts, including the respective C3N3•••M distances and deviation angles γ. For the search in the CSD, the maximum distance between the C3N3-centroid and the metal was set to 4 Å and the maximum angle γ between the C3N3 plane normal and the C3N3•••M vector was set to 20° in accordance with the definition of delocalised C3N3•••M interactions by Tiekink and co-workers. [49]  [a] These two related structures mark the only cases in which a C3N3•••M contact closer than in our system has been described. [50] For both copper(I) and silver(I), no shorter contact has been published to the best of our knowledge. In no case has an intramolecular C3N3•••M contact been disclosed so far. [a] Platinum is, as two stacked (d(Pt-Pt) = 3.32 Å) bis(acetylacetonato)platinum(II) ions, encapsulated in a C3-symmetric coordination cage; the authors do not comment on the close C3N3•••Pt distances. [51] [b] Platinum is, as one discrete molecule of bis(acetylacetonato)platinum(II), encapsulated in a C3-symmetric coordination cage; the authors note the close C3N3•••Pt distance but attribute the encapsulation mainly to donor-acceptor π⋅⋅⋅π stacking interactions. [52] [c] Nickel is, as two stacked molecules of (N,N-ethylenebis(acetylacetoneiminato))nickel(II), encapsulated in a C3-symmetric coordination cage; the authors do not comment on the close C3N3•••Ni distances. [53] [d] Platinum is, as one discrete molecule of bis(acetylacetonato)platinum(II), encapsulated in a C3-symmetric coordination cage; the authors note the small C3N3•••Pt separation but do not link it to the successful encapsulation. [54]

VT NMR Studies
Variable-temperature 1 H and 31 P{ 1 H} NMR studies for all complexes have been conducted in the temperature ranges according to solvent and/or solubility limitations. For 1Ag and 1Au, the breakdown from C3v to C3 symmetry due to the inversion of helicity becoming slow at the NMR timescale can be followed nicely. For 1Cu, the high-temperature limit spectra in CD2Cl2 and CD3CN suggest C3v symmetry, while cooling down reveals a slow equilibrium between the C3-symmetric Cucl and a bidentate, open form Cuop No significant anion influence can be discerned going from 1Cu over 1CuBF4 to 1CuBAr F 4 in CD2Cl2.   For 1Au, a 1 H, 1 H COSY experiment at −50 °C was carried out ( Figure S57) to determine the connectivities in the phenyl rings, in order to understand the reason for the significant shielding and deshielding of certain resonances ( Figure S55). The cross peaks observed in this experiment also unambiguously establish the connectivities in the substituted cyclopentadienyl rings (v, orange, and w, purple) of this complex and, given the similarities between the low-temperature 1 H NMR spectra, also of 1Ag (and, to the Cucl fraction of 1Cu, 1CuBF4, and 1CuBAr F 4). After the two phenyl rings had been separated regarding their resonances (x, green, and o, red), DFT calculations were carried out to assign the resonances to the DFT-optimised structure (depicted as simplified model in Figure S56).
Shielding tensors were calculated for all 54 protons of 1Au (cf. section 1.1, Computational methods), combined into isotropic chemical shielding (in ppm), averaged for the three protons that, due to the C3 symmetry of the 1 H NMR spectra at low temperatures, are expected to be equivalent for the solution-state structure of 1Au (standard deviations are given in brackets), and referenced to the chemical shift for TMS obtained by the same method. Results are summarised in Table S12. Since the error of the method is quite high and the chemical shifts so obtained are far from realistic values, only the general trend can be discussed. For an easier discussion of these trends, the calculated shifts have been scaled to match the experimental spectra separately for the ferrocenyl and the phenyl region. From the calculations, it is obvious that proton x 1 yields the most deshielded resonance, while proton o 5 appears as most shielded. The unusual proton shifts for x 1 (proton 1 in Figure S56) and o 5 (proton 10 in Figure S56) are in line with their distinguished positions in close vicinity to the respective metal ions. Protons x 1 have a mean Fe···H distance of 3.60 Å, the sum of the van der Waals radii being 3.61 Å; [38] however, this distance is far from that of a covalent bond for which distances between 1.34 Å and 1.62 Å have been experimentally determined. [55] It is interesting to note that for an Fe-bound hydride in [Cp2Fe(H)]PF6, a 1 H NMR chemical shift of −1.3 ppm was recorded (in HF/PF5) , showing intense shielding in contrast to the deshielding observed for x 1 . [55] Alternatively, the strong deshielding could thus also relate to the vicinity to the C3N3 nitrogen atoms (mean N···H distance 2.36 Å, sum of van der Waals radii 2.86 Å). For protons o 5 , the mean Au···H distance amounts to 3.31 Å with the sum of the van der Waals radii equalling 3.66 Å. [38] Figure S56. Optimised structure of 1Au used for the calculation of its 1 H NMR shifts. [a] Referring to the numbering scheme shown in Figure S56. [b] Mean deviations (MD) were calculated as the deviations of the three individual shielding parameters obtained for the non-identical protons at the chemically equivalent positions.
[d] Referring to the signals depicted in the corresponding 1 H, 1 H COSY NMR. spectrum ( Figure S57) and based on the DFT results and the connectivity inferred from the COSY NMR experiment.  Table S12). The insert aims to give a better view of the broad resonance at higher temperatures.  Although at 40 °C 1Cu appears C3v-symmetric in CD2Cl2, its corresponding 31 P{ 1 H} NMR spectrum at the same temperature still hints at the swiftly equilibrating bidentate form Cuop. This is most likely due to the greater difference in resonance frequencies for the two nuclei which leads to a higher temperature to attain coalescence in the 31 P{ 1 H} NMR spectrum.
In CD3CN, 1Cu appears C3v-symmetric, as judged from both its 1 H and 31 P{ 1 H} NMR spectra, already at room temperature, the signals sharpening considerably at higher temperature. Lowering the temperature, decoalescence in both the 1 H and the 31 P{ 1 H} NMR spectrum is observable, indicating that CD3CN acts as an auxiliary ligand promoting the fast equilibration between the three theoretical coordination modes for 1Cuop (for this, see also the titration experiment with CN − (followed by UV/Vis) in section 2.5). In the 1 H NMR spectra at −20 °C, 1Cucl cannot be detected from its well-distinguishable resonances x 1 at 9.7 ppm (cf. Figure S57 and Figure S59), suggesting that CD3CN acts as ligand rather than the third phosphanyl group.   A comparison of the 1 H NMR spectra of three Cu I complexes with triflate (1Cu), tetrafluoroborate (1CuBF4), and BAr F 4 (1CuBAr F 4) as their respective anion at −70 °C ( Figure S63) does not reveal significant differences in both position and multiplicity for the signals attributed to 1Cucl and 1Cuop. With respect to their relative intensities, the two signal sets centred at 3.90 ppm (blue box, 1Cuop) and 3.17 ppm (red box, 1Cucl) have been used. The results are assembled in Table S13 and illustrate that, for 1CuBAr F 4, a slight increase in the relative amount of 1Cucl in this equilibrium can be detected, potentially hinting at a weak involvement of the smaller anions in stabilising the bidentate, open form 1Cuop. In the corresponding 19 F{ 1 H} NMR spectra in CD2Cl2, no change in the respective signals can be found. In their 31 P{ 1 H} NMR spectra at −70 °C ( Figure S64), the signal sets appear very similar. Slight differences in signal width and exact position of the free phosphanyl group of 1Cuop in the case of 1CuBAr F 4 are noticeable. The relative integrals are similar, but their exact comparison is hampered by the pulse sequence (zgpg for fast-relaxing nuclei, 1 H-decoupled) and the uneven baseline.  The ratio between 1Cucl and 1Cuop is, however, solvent-dependent. In MeCN-d3, the room-temperature 1 H and 31 P{ 1 H} spectra suggest C3v symmetry (Section 1.5, Figure S61, Figure S62); cooling down again decoalesces the 31 P resonance ( Figure S62). Donor solvents like MeCN and THF might stabilise 1Cuop through weak coordination of the Cu I ion to different degrees, even though such complexes cannot be detected by HR-ESI MS from solutions of 1Cu in pure MeCN or THF. In this way, 1Cu(OTf) and 1CuBF4 exhibit different solvent behaviour in THF-d8 ( Figure S65). While 1CuBF4 behaves similarly in THF-d8 and in CD2Cl2 when lowering the temperature as monitored by 1 H ( Figure S66) and 31 P{ 1 H} ( Figure S67) NMR spectroscopysignals for both Cucl (+) and Cuop (#) are detectable, the ratio being strongly shifted towards Cuop which seems to be stabilised by THF-d8 -1Cu(OTf) gives markedly different results. Most prominently, other complex species seem to emerge at lower temperature as seen from the 31 P{ 1 H} NMR spectrum at −60 °C ( Figure  S68, top left). The 19 F resonance, sharp at 25 °C (∆ν = 7.1 Hz), broadens considerably (and much more so than the 19 F resonance of 1CuBF4, increasing from 4.8 to 16 Hz, determined for the more intense peak attributed to 11 BF4 − ) and splits up into at least three signals, one of which is very broad (∆ν = 400 Hz), speaking for an interaction with the quadrupolar 63/65 Cu (I = 3/2) nuclei of 1Cu. Even more so, the stark colour difference between solutions of 1Cu in DCM, THF and MeCN ( Figure S75) is in line with this finding. Notably, 1Cu and 1CuBF4 yield different UV/Vis spectra in THF (but not in CH2Cl2 and MeCN), futher supporting an involvement of the anion in this case, triflate being a better donor for Cu I than BF4 − , thus breaking the C3N3-Cu I bond in combination with THF, hypsochromically shifting the absorption maximum to that found for both Cu I complexes in MeCN.   (−60 °C) at slightly lower field is due to the isotopic shift between 10 B and 11 B (cf. Figure S26).

Investigations on the metal-triazine interactions
The calculated Wiberg bond indices (WBI) between the respective metal and the carbon and nitrogen atoms of the triazine ring in 1Cu, 1Ag, and 1Au, as found in their solid-state geometries, are listed in Table S14. From the comparison of the data it can be concluded that the interaction between the triazine core and the coinage metals follows the order Cu < Au < Ag, thus reflecting the covalent radii of the coinage metals. Table S14. WBI between the triazine ring atoms and the respective coinage metal ion in 1Cu, 1Ag, and 1Au. Numbering scheme according to Figure S51. Next, we investigated the interaction of the parent triazine (C3H3N3) with the coinage metal ions. Therefore, the geometry of the triazine core and the position of the coinage metal were restrained to the structural parameters obtained for the gas-phase structures of 1Cu, 1Ag, and 1Au, respectively. Then, the interaction energy between metal ion and triazine was calculated as Eint = ETriazine-M − EM − ETriazine. The results are summarised in Table S15. The same trend as derived from analysis of the WBI of the complexes 1Cu, 1Ag, and 1Au are found with silver showing the largest interaction energy. To get insights into the nature of the interaction between the triazine core and the coinage metals, energy decomposition analyses (EDA) based on natural orbitals of chemical valence (EDA-NOCV) [56] were carried out at the BP86-D3BJ/def2-TZVP level of theory. In this approach, the interaction between two fragments is decomposed into pairs of corresponding donor and acceptor orbitals. For our model systems, C3H3N3−M + , three such pairs account for 71% of the total interaction energy for M = Cu (Ag: 73%, Au: 75%). Looking at the shape of these pairs (Table S16), they clearly correspond to cation•••π interactions, namely the donation of electron density from the occupied π orbitals of the triazine core to empty s and p orbitals of the coinage metal ions. Other significant contributions to the interaction energy (Cu: 23 %, Ag: 22 %, Au: 20%) are rearrangements of electron density (polarisation) within the triazine moiety, probably induced by the coordination of the positively charged coinage metals (one pair is exemplarily shown for each metal in Table  S16). The remaining part of the interaction energy can be assigned to back-bonding from occupied d orbitals of the metals to the triazine core (M−π* back-bonding, one pair is exemplarily shown for each metal in Table S16). Based on the analyses of the NOCV, the triazine molecule can be classified as a donor ligand toward coinage metal cations, whilst back-bonding interactions play an insignificant role. Table S16. Corresponding pairs of natural orbitals of chemical valence (NOCV) for C3H3N3-Cu + , C3H3N3-Ag + , and C3H3N3-Au + . The donor orbital is rendered in blue while the acceptor orbital is rendered yellow (isosurface values set to 0.06-0.08). The frontier molecular orbitals of the complexes (calculated without considering their respective anions; for 1Cu, the closed, C3symmetric form Cucl has been considered solely) are degenerate, owing to the threefold symmetry of these molecules. Table S17 thus only representatively shows HOMO−1, HOMO and LUMO, alongside their respective energies in eV, to illustrate that the HOMO is located at and distributed over the three iron atoms of the ferrocenylene moieties, while the LUMO features contributions of both the triazine core and the ferrocenylene moieties.

Calculation of UV/Vis spectra
TDDFT calculations on 1Cu, 1Ag, and 1Au were conducted at the TPSSh/def2-TZVP [17c, 57] level of theory in the gas phase. For the sake of computational cost, only 100 transitions were calculated for 1Cu, whilst 240 transitions were calculated for both 1Ag and 1Au.
The calculated spectra are in good agreement with the experimental spectra (cf. section 2.5.1) measured in dichloromethane ( Figure  S73 - Figure S75). In the visible part, the spectra are dominated by 3d-π* transitions from the iron centres to the triazine core (see Table S18 for selected difference densities of 1Ag). At higher wavelengths, π-π* as well as M-π* transitions are observed as well.
As can be seen from Figure S72, the hypothetical structure for 1Cuop with its presumed Cu-C3N3 contact ( Figure S72) does lead to a broadening of the 3d-π* and 3d-3d transitions for the absorption band at about 500 nm, in line with the experimental UV/Vis spectrum which might be understood as the sum of both forms, 1Cucl and 1Cuop.       Figure S72.

d [Å] Angle [°]
P(1)-Cu(1) 2.247 P(1)-Cu (1)   While for 1Ag ( Figure S73) and 1Au ( Figure S74) the solvent does not result in significant shifts of both the position and the extinction coefficient of the electronic transitions in the UV/Vis region (the strongest influence can be seen for the ligand-centred π-π* and the M(Fe,Au)-π* transitions which decrease in the order CH2Cl2 > THF > CH3CN, correlating with the donor strength of the solvent), 1Cu ( Figure S75) shows a more pronounced solvent effect, especially regarding the d(Fe)-π*(C3N3)/d(Fe)-d(Fe) transitions which, in CH2Cl2 (maroon curve), are batho-and hyperchromically shifted as well as broadened with respect to the peaks in THF (yellow) and CH3CN (black). This effect also becomes apparent when comparing the three complexes with each other; the visibly more intense and darker colour of 1Cu in CH2Cl2 when compared to its heavier homologues is reflected in the UV/Vis spectra ( Figure S76). As also noted in the VT NMR experiments, the anion plays a role for 1Cu in THF ( Figure S77). In contrast to the BF4 − anion, the triflate anion in 1Cu seems to bind to Cu I , disturbing the C3N3-Cu I contact responsible for the bathochromic shift.    While both the ligand precursor 4 and the borane-protected ligand 1BH3 are reversibly oxidisable in both the BF4 − -and the BAr F 4 −based SE ( Figure S80) and even display three individual redox events when measured in the very weakly coordinating BAr F 4 − -based SE, free ligand 1 yields non-reversible oxidation events and, tied to the first of several oxidation events, a delayed reduction hinting at an electron transfer-chemical reaction (EC) oxidation mechanism ( Figure S81). The electrochemical behaviour of ligand 1 in the BAr F 4 − -based SE is shown in Figure S82. While the first oxidation at about 340 mV (top, dashed lilac) is reversible to some degree (vide infra), the later oxidations at 500 mV (top, dash-dotted lavender) and 860 mV (top, solid blue) are irreversible and cancel out the partial reversibility of the first oxidation. Similar to the properties of 1 in the BF4 − -based SE, the second and third oxidation induce broad yet weak reductive events at cathodic potentials around −1.5 V which are absent if the scan direction is reversed immediately after the first oxidation. The bottom part of Figure S82 confirms this behaviour. If the cycle starts with a cathodic scan, no reductive events can be identified, while the broad reductive features appear after the first oxidations have been carried out in the second cycle.
Recording the first oxidation with three different scanning speeds (20, 100, and 500 mV•s −1 ) and normalising the currents according to the Randles-Sevčik equation for the square root of the scanning speed ( Figure S83), [58] the increase of reversibility indicates an EC(EE) mechanism operating in that case. The first oxidation, likely iron-centred, thus generates a species undergoing a chemical transformation yielding an electroactive product which participates in further redox events.  The electrochemical characterisation of 1Au in the BF4 − -based SE ( Figure S84) reveals similar behaviour of this complex in comparison to free ligand 1. Two oxidation events at 350 mV (pink dotted, top) and 490 mV (red dash-dotted, top) can be isolated by CV, the first of which has some degree of reversibility, while the second is linked to a cathodically shifted reduction at −1.3 V. This dependence on prior oxidation is also found when the scan direction is reversed (orange dashed, bottom). A third oxidation event at 690 mV appears to be (quasi)reversible, yet leads to a further broadening at the reductive event at around −1.3 V. Analysing a 1:1 mixture of 1 (red dotted) and 1Au (blue dotted) under the same conditions ( Figure S85) however demonstrates that the electrochemical characteristics of 1Au are most likely not a result of 1Au degrading into 1 during the experiment, since the recorded cyclic voltammogram (black bold) appears as the sum of both individual traces. In the BAr F 4 − -based SE ( Figure S86), the electrochemistry of 1Au displays marked differences. The first (dotted pink) and, to some degree, reversible, and the second (dashed red), less reversible, oxidations appear more separated than in the BF4 − -based SE, in line with the lower ability of BAr F 4 − anions to compensate positive charges by coming in close vicinity to the positive charges. [59] The third oxidation at about 1.2 V is much less reversible. Most notably, the cathodically shifted reduction observed for 1Au in the BF4 − -based electrolyte seems almost absent. Only at higher scanning speeds does a reductive event at about −1.2 V become apparent ( Figure  S87), in line with describing the electrochemistry of 1Au as following an ECE mechanism. This Randles-Sevčik plot also indicates changes in the reductions at more anodic potentials.   Regarding its electrochemical behaviour in the BF4 − -based SE ( Figure S88), 1Ag behaves similar to 1Au. There are two oxidation events at 255 mV (dotted light green) and 500 mV (dash-dotted forest green) which are not reversible and entail cathodically shifted reduction events at −500 mV (linked to the first oxidation, itself again inducing another oxidation event at −450 mV) and −1.3 V (linked to the second oxidation). Further oxidation events at more anodic potentials intensify the reductive current at both potentials (solid navy), but none of the reductive events is present when the measurement is begun with a cathodic scan (bottom) but only appear after the first oxidative cycle. Comparing the cyclic voltammograms at different scan rates ( Figure S89) confirms these reductions to be linked to one or more chemical transformations that 1Ag undergoes after initial and following oxidations, as both reductions disappear at slow scanning speeds. As for 1Au, one or more EC(E) mechanisms seem to operate here. In much the same way, following the electrochemistry of 1Ag in the BAr F 4 − -based SE ( Figure S90) leads to a disappearance of the reduction events at about −500 mV, while the even more cathodic and oxidation-dependent reductions are still present. As for the BF4 −based SE, slower scan speeds result in the disappearance of the reductive currents (not shown).   The electrochemical characterisation of the copper(I) complexes in the BF4 − -based SE ( Figure S91) was carried out using 1CuBF4. In contrast to 1Au and 1Ag under the same conditions, only one isolable first oxidation at 270 mV (top, dotted light blue) can be identified, tied to a cathodically shifted reduction event at about −1.4 V which is not present when the complex is initially scanned in cathodic direction (bottom, dashed berry) The first oxidation is followed by a broad oxidation event (bottom, solid dark purple) at about 700 mV (and potentially by further oxidations), which are linked to a comparatively sharp reduction peak at 380 mV and at least one more cathodically shifted reduction at about −1.3 V. While lowering the scanning speeds entails the disappearance of the reduction peaks ( Figure S92), it also results in a second prominent yet also irreversible oxidation at higher potential (1.1 V), an unprecedented behaviour in comparison to both 1Au and 1Ag and potentially linked to the generation of an (unstable) Cu 0 species (vide infra). When following the electrochemistry of 1Cu in the BAr F 4 − -based SE ( Figure S93), two oxidations at 640 mV (top, dotted light blue) and 890 mV (top, dash-dotted royal blue) can be identified again. In contrast to both 1Au and 1Ag, an initial cathodic scan (bottom, dashed berry) reveals a reduction at about −2 V original to 1Cu, invariant to scanning speed ( Figure S94) unlike the other reduction events at about −1.5 V (bottom, solid dark purple) related to the most anodic oxidation. The nature of this original reduction is only speculative but might correspond to the Cu I /Cu 0 redox couple. The absence of a corresponding oxidationno corresponding peak is found during the anodic scan even at higher scanning speeds and the oxidation events are independent of prior reductionwould be in line with the reported instability of mononuclear copper (0)    In all cases, the second of three consecutively measured cycles is shown.         As becomes apparent from the oxidation and reduction experiments, 1Au cannot be reversibly oxidised at room temperature or at −50 °C; the UV/Vis spectra obtained after one oxidation-reduction cycle are not identical; hence, they most likely correspond to different compounds. The UV/Vis spectral change on the first oxidation at −80 °C is markedly different from the changes recorded at the other temperatures, and only the second oxidation brings about the same spectral features that have been found on oxidation at −50 °C and at 25 °C. Comparing the spectra obtained from double oxidation at −80 °C ( Figure S103, orange) and from the isolated oxidation product 5 (104, solid red) reveals them to be quite similar, particularly with respect to the intense absorption at about 330 nm and the shifted d(Fe)π*(C3N3)/d(Fe)-d(Fe) transitions at about 575 nm. Though it is only speculative and a reaction with adventitious traces of water in the spectroelectrochemical setup seems unlikely, the similarity of the two spectra could point to the formation of 5 or a very similar species under these conditions, supporting the proposed ECE-type oxidation from 1Au to 5 concurrent with oxidation at a phosphorus atom.
. Figure S104. Comparison of the UV/Vis spectra of 1Au (black, dotted) and isolated oxidation product 5 (red, solid), both in CH2Cl2. The spectra have been arbitrarily scaled for better comparability

Spectroelectrochemical measurements of 1Ag
Figure S105. Time-dependent UV/Vis spectra recorded during the oxidation of 1Ag at −80 °C in (nBu4N)BAr F 4/CH2Cl2. The green line represents the starting material, the red line represents the spectrum obtained after the oxidation and the blue line represents the spectrum after the subsequent reduction. Arrows indicate the direction of intensity changes during oxidation.

Spectroelectrochemical measurements of 1Cu
Figure S106. Time-dependent UV/Vis spectra recorded during the oxidation of 1Cu at −80 °C in (nBu4N)BAr F 4/CH2Cl2. The green line represents the starting material, the red line represents the spectrum obtained after the oxidation. Arrows indicate the direction of intensity changes. The insert shows the region relevant to MMCT/IVCT transitions in detail. Figure S107. Time-dependent UV/Vis spectra recorded during the reduction of 1Cu at −80 °C in (nBu4N)BAr F 4/CH2Cl2 after previous oxidation. The green line represents the starting material before oxidation, the red line represents the spectrum obtained after the oxidation and the blue line represents the spectrum obtained after subsequent reduction. Arrows indicate the direction of intensity changes during the reduction (red to blue). The insert shows the region relevant to MMCT/IVCT transitions in detail.