Structural Control of Metal-Centered Excited States in Cobalt(III) Complexes via Bite Angle and π–π Interactions

Abstract

Co^III complexes have recently become an important focus in photophysics and photoredox catalysis due to metal-centered excited states with strong oxidizing properties. Optimizing chelate ligand bite angles is a widely used strategy to strengthen metal–ligand interactions in coordination complexes, with the resulting enhanced ligand fields often contributing to extended excited-state lifetimes that are advantageous for photochemical applications. We demonstrate that bite-angle optimization exerts the opposite effect on Co^III polypyridines compared to previously studied transition metal complexes, as polypyridine ligands function as π-donors to Co^III rather than π-acceptors. Our findings reveal two counterintuitive paradigms: while bite-angle optimization weakens the ligand field in Co^III complexes, the resulting lower-energy metal-centered excited states can be accompanied by extended excited-state lifetimes, driven by increased rigidification through intramolecular π–π interactions. These insights, along with additional experiments investigating the possibility of photoreactions from higher excited states, advance the current understanding of the photophysics and photochemistry of first-row transition metal complexes and highlight key distinctions from the more extensively studied photoactive complexes of second- and third-row transition metals.

Introduction

Transition metal complexes with a d⁶ valence electron configuration, such as Ru^II (4d⁶), Ir^III and Os^II (5d⁶), have become essential in applications including solar energy conversion, − photocatalysis, − light-emitting devices, − bioimaging, − photodynamic therapy, − molecular sensors, and probes. − This prominence is attributed to the metal-to-ligand charge transfer (MLCT) excited states of the respective noble metal complexes, which exhibit favorable properties such as high energy storage capacity, redox activity, prolonged excited-state lifetimes, and often excellent luminescence properties. In the pursuit of more cost-effective and earth-abundant alternatives to precious compounds, first-row 3d⁶ transition metals such as Co^III, − Fe^II, − Mn^I, , and Cr⁰, , have received substantial attention over the past decade. However, fine-tuning the photophysical properties of 3d metal complexes remains challenging due to the primogenic effect, which leads to a more contracted nature of the 3d metal orbitals compared to the 4d and 5d orbitals, reducing metal–ligand orbital overlap and thus decreasing ligand field strength (10 Dq). , This limited orbital overlap leads in consequence to the stabilization of highly distorted metal-centered (MC) states, which further complicates the utilization of charge-transfer (CT) states due to their rapid nonradiative deactivation into the ground state through those MC states. ,− Recent studies show that in strong ligand fields, the lowest MC excited states of Co^III complexes, although strongly distorted, can become relatively long-lived and photochemically useful. ,,

A common design strategy for photoactive metal complexes involves using π-acceptor ligands and carefully tuning coordination bite angles to enhance metal–ligand orbital overlap and thereby maximize the ligand field strength. ,,,, This approach has proven particularly effective in Cr^III polypyridine complexes with a d³ electron configuration (Figure ), − as it increases the energy of the highly distorted ⁴T₂ excited state and thereby enables “spin-flip” emission with prolonged lifetimes and enhanced luminescence quantum yields from largely undistorted and energetically lower-lying ²E/²T₁ states.

1.

Effect of bite-angle optimization on the lifetimes (τ) of the doublet metal-centered (²E) excited state and emission quantum yields (Φ) of pseudo-octahedral Cr^III polypyridine complexes in solution at room temperature. − Selected cis- and trans-N–Cr–N angles are marked in the molecular drawings to highlight the important geometrical differences between the individual molecular structures. The geometry distortion parameter Σ was calculated as ¹²Σ_i=1|90 – φ_i|, where φ_i represents cis-N–Cr–N angles. The Σ parameter quantifies the deviation from the ideal octahedral geometry. ,,

Thus, in [Cr­(dqp)₂]³⁺ (dqp = 2,6-di­(quinolin-8-yl)­pyridine) the increased σ-donation effect associated with its comparatively highly octahedral coordination environment is further combined with the π-acceptor character of the pyridine and quinoline moieties, leading to a large ligand field splitting and consequently an excited-state lifetime (τ­(²E)) of 1.2 ms and a photoluminescence quantum yield of 5.2% in solution at room temperature (Figure c). The performance of [Cr­(tpy)₂]³⁺ (tpy = 2,2′:6′,2″-terpyridine) and [Cr­(bpy)₃]³⁺ (bpy = 2,2′-bipyridine) is exceeded by orders of magnitude (Figure a,b). − This is primarily due to the stronger Cr^III–N interactions in [Cr­(dqp)₂]³⁺, which arise from improved metal–ligand orbital overlap enabled by a coordination geometry that more closely approximates an ideal octahedron compared to [Cr­(tpy)₂]³⁺ and [Cr­(bpy)₃]³⁺. The deviation from the ideal geometry can be quantified by the octahedral distortion parameter Σ, which is defined as the sum of deviations from 90° for all 12 cis-N–Cr–N angles (¹²Σ_i=1|90 – φ_i|). The Σ parameter decreases from 104° to 67° and then to 29° across the series of complexes shown in Figure , resulting in significant enhancements in τ­(²E) and Φ as described above.

Beyond the d³ complexes of Cr^III and Mn^IV, − strong ligand fields generated by π-acceptor ligands and chelates with optimized bite angles have been most effective in complexes of d⁶ metals. In Cr⁰ and Mn^I complexes, ,,− the combined σ-donating and π-accepting electronic properties of isocyanide chelate ligands, paired with nearly ideal octahedral geometries, make it possible to achieve emissive ³MLCT excited states with lifetimes in the nanosecond range at room temperature in solution. ,,−

Among photoactive 3d⁶ systems, Fe^II complexes have received by far the most attention, and efforts to modify their photophysical properties have also focused on creating strong ligand fields and idealized coordination geometries. ,,− In pursuit to extend the ³MLCT excited state lifetimes of Fe^II complexes, several research groups have developed ligands based on N-heterocyclic carbene (NHC) and mesoionic carbene (MIC) moieties, which can act as strong σ-donors and π-acceptors, increasing the ligand field strength. ,,,,,,, Ultimately, this molecular design extended the ³MLCT lifetime up to ∼500 ps in the 6-fold MIC coordination environment of [Fe­(btz)₃]²⁺ (btz = 3,3′-dimethyl-1,1′-bis­(p-tolyl)-4,4′-bis­(1,2,3-triazol-5-ylidene)), about 5000 times longer than in the benchmark [Fe­(bpy)₃]²⁺ complex.

The higher oxidation state of Co^III relative to Fe^II inherently generates a stronger ligand field, providing a priori an advantageous electronic environment. , Recent work demonstrated how to control ³MC excited state deactivation in Co^III polypyridine complexes, leading to promising application potential of these states in photoredox catalysis. ,,, In [Co­(4,4′-R₂bpy)₃]³⁺ complexes the relationship between ligand field strength and the lifetimes of ³MC excited states was rationalized in terms of Marcus inverted region behavior, analogous to the common behavior of ³CT excited states in Ru^II or Os^II complexes. In this regime, an increase in the excited state energies leads to a decrease in the (nonradiative) deactivation rate of the excited state. This finding is essential for designing Co^III systems possessing long-lived ³MC excited states with high energy capacity, which are crucial for photocatalytic applications. Furthermore, this behavior contrasts with that commonly seen for isoelectronic Fe^II complexes. Previous observations on [Co­(PhB­(MeIm)₃)₂]⁺ (PhB­(MeIm)₃ ^– = tris­(3-methylimidazolin-2-ylidene)­phenyl borate) and [Co­(CN)₆]^3– complexes align well with the introduced concept of Marcus inverted behavior. In these complexes, the very strong ligand field splitting due to the 6-fold NHC or cyanido environments caused the ³MC states to become emissive, exhibiting lifetimes of 1 μs and 2.6 ns, respectively, owing to the sufficiently large energy gap between these states and the ground state (∼1.7–1.8 eV). − However, the impact of coordination bite angles on the photophysical properties of Co^III polypyridine complexes remains unexplored.

In the present work, we focus on investigating how the properties of ³MC excited states in Co^III complexes are influenced by modification of the ligand bite angles, expansion of the aromatic π-system of the ligand framework, and cooperative rigidity due to π–π and steric interactions between individual ligands coordinated to the same metal center. While it is well established for Cr^III that σ-donation and π-backbonding play key roles in the ligand design based on polypyridine scaffolds (Figure a,b), comparatively little is known about those effects in Co^III complexes. ,

2.

Anticipated main electronic effects upon bite-angle optimization shown in a simplified orbital scheme: a. σ-donation in Cr^III and Co^III polypyridine complexes, b. π-backbonding in Cr^III polypyridine complexes, c. π-donation in Co^III polypyridine complexes. In a–c molecular orbital diagrams show relevant bonding interactions of metal 3d orbitals and (symmetry-adapted) ligand group orbitals (LGOs). The magnitude of the illustrated electronic effects increases as the coordination geometry approaches a more ideal octahedral arrangement.

We hypothesize that improving the metal–ligand orbital overlap via enhancing the ligand bite angles for an octahedral geometry can enhance σ-donation, similar to the case for Cr^III complexes (Figures and a). However, literature reports indicate that polypyridine ligands behave as π-donors rather than π-acceptors toward Co^III, which would be expected to stabilize MC states (Figure c) and could entail shorter excited-state lifetimes. , To experimentally test this hypothesis, we compare the photophysical properties of compounds of known structure[Co­(phtpy)₂]³⁺ (phtpy = 4′-phenyl-2,2′:6′,2″-terpyridine) and [Co­(dqp)₂]³⁺ (Figure a), − where the differences between coordination geometries are similarly pronounced to those observed in the Cr^III complexes shown in Figure a,c. Our hypothesis is confirmed to the extent that π-donation, rather than π-acceptor behavior, predominates in the Co^III complexes; however, the resulting stabilization of the ³MC excited state does not lead to the anticipated shortening of its lifetime but rather to an unexpected extension. This finding suggests enhanced cooperative rigidity arising from the extended π-conjugated framework of the dqp ligand, which enables intramolecular π–π interactions that rigidify the overall coordination complex and reduce the efficiency of nonradiative ³MC relaxation. − The observed simultaneous lowering of excited-state energy and prolongation of its lifetime contradicts the typical photophysical behavior seen in most metal complexes and organic compounds but can be attributed to a rationalizable interplay between ligand field strength, excited-state distortions, and force constants.

3.

(a) X-ray crystal structures of [Co­(phtpy)₂]³⁺, and [Co­(dqp)₂]³⁺ (Supporting Information); counterions and solvent molecules are omitted for clarity, and thermal ellipsoids are shown at 50% probability; π–π interactions are indicated with distance (Å) between the quinoline units. (b) ¹H NMR spectra of the free dqp ligand (top) and [Co­(dqp)₂]³⁺ in CD₃CN (bottom), signals corresponding to the protons of the quinoline moieties, involved in π–π interactions, are marked in brown, while the protons of the pyridine unit are marked in blue.

Results and Discussion

Two Sister Complexes with Different Coordination Geometries

The 4′-phenyl-2,2′:6′,2″-terpyridine ligand (phtpy) is synthetically more accessible than unsubstituted tpy, so we used phtpy instead of tpy for our study but expect the primary coordination sphere to be largely unaffected by the additional phenyl ring in the backbone. The phtpy ligand and the compound [Co­(phtpy)₂]­(PF₆)₃ were synthesized following procedures adapted from the literature (Supporting Information). , [Co­(dqp)₂]­(PF₆)₃ was synthesized in a manner similar to [Co­(phtpy)₂]­(PF₆)₃. , In the first step, the dqp ligand (2.05 equiv) reacted with CoCl₂·6H₂O (1.0 equiv) in CH₃CN, followed by Co^II/III oxidation with Br₂, and finally, the anion was exchanged with KPF₆ in aqueous solution. After purification, the complex was characterized by standard analytical techniquesNMR spectroscopy, high-resolution mass spectrometry, and combustion analysis (Supporting Information).

Orange crystals of [Co­(dqp)₂]­(PF₆)₃·2CH₃CN, suitable for X-ray diffraction (XRD) measurements, were obtained via vapor diffusion of Et₂O into a saturated CH₃CN solution. Analysis of the crystal data revealed the Co–N bond lengths to be in the range of 1.952(2)–1.972(2) Å, in line with the values for previously reported tridentate polypyridine complexes. Compared to [Co­(phtpy)₂]³⁺, upon extension of the aromatic system in [Co­(dqp)₂]³⁺, bite angles increase from 164.7(3)° to 179.42(10)°–179.21(10)°, approaching the ideal octahedral 180°, similar to what has been reported previously for [Ru­(dqp)₂]²⁺, − and for Cr^III complexes with bite-angle optimized chelate ligands. ,,,, The analysis of the octahedral distortion parameter Σ reveals a highly octahedral coordination environment with a Σ = 13° in [Co­(dqp)₂]³⁺ as opposed to a Σ = 65° in [Co­(phtpy)₂]³⁺. Interestingly, the Σ value in [Co­(dqp)₂]³⁺ is lower than in [Cr­(dqp)₂]³⁺ (Σ = 29°), presumably due to a smaller ionic radius of Co^III compared to Cr^III.

These structural properties as obtained in the solid state are in good agreement with the fully relaxed, calculated singlet ground-state structure of [Co­(dqp)₂]³⁺ in acetonitrile. Density functional theory (DFT) predicts Co–N bond lengths ranging from 1.961 to 1.978 Å, while a bite angle of 178.7° is obtained. All calculated equilibrium structures (as well as high-resolution images) are available via the free online repository Zenodo. In the crystal structure of [Co­(dqp)₂]­(PF₆)₃ π-stacking effects of the adjacent ligand “side-arms,” namely the quinoline units, are evident (Figure a), which could contribute to a favorable rigidity-enhancing effect. π–π interactions are indicated by roughly 3.4–3.5 Å distances between the relevant quinoline centroids (Figure a), matching the values reported for [Cr­(dqp)₂]³⁺. Notably, the structure exhibits helical twisting, compatible with P- and M-chirality according to Cahn–Ingold–Prelog nomenclature, and pairs of PP- and MM-enantiomers are present in the crystal unit cell, reminiscent of what has been reported for [Cr­(dqp)₂]³⁺ and related Cr^III complexes. ,,

Evidence for the intramolecular π-stacking effects between individual dqp ligands is further provided by comparison of the ¹H NMR spectra of [Co­(dqp)₂]³⁺ and the free dqp ligand (Figure b). Upon coordination, the quinoline protons (marked in brown in Figure b) of the ligand experience a drastic upfield shift, while the pyridine protons (marked in blue in Figure b) undergo a downfield shift. Typically, without any structural constraints, a downfield shift is anticipated for the ligand protons due to an electron density displacement from the ligand to the metal upon coordination. However, the observed upfield shift of the quinoline protons indicates that those units are likely located in close proximity to one another. These data are also consistent with π–π interactions observed in the ¹H NMR spectrum of [Ru­(dqp)₂]²⁺.

Cyclic voltammetry of [Co­(dqp)₂]³⁺ in dry and deaerated CH₃CN with 0.1 M (n-Bu₄N)­PF₆ revealed two reversible reduction events (Figure S1). The first wave at +0.33 V vs SCE (E _red, see below) can be assigned to the Co^II/III redox pair and the second wave at −0.99 V vs SCE can be attributed to a ligand-based reduction process. The assignment of the metal-centered reduction is supported by the quantum chemical simulations, which reveal the ${\sigma }_{x^2{-y}^2}^{*}$ character (or $d_{x^2{-y}^2}$ in an atomic orbital picture) of the lowest unoccupied molecular orbital (LUMO) of the [Co­(dqp)₂]³⁺ complex, while the LUMO-1associated with the second reductionis of π*_dqp nature (Figure S23). When comparing the electrochemical behavior of [Co­(dqp)₂]³⁺ to that of other Co^III polypyridine complexes, the metal reduction event was observed at very similar potentials of +0.32 V vs SCE and +0.26 V vs SCE in [Co­(4,4′-Br₂bpy)₃]³⁺ and [Co­(phtpy)₂]³⁺, respectively. ,,

Bite-Angle Optimization Weakens the Ligand Field

In the UV–vis absorption spectra of [Co­(phtpy)₂]³⁺ and [Co­(dqp)₂]³⁺, the strongest bands are observed below 400 nm (Figure a,b). In the case of [Co­(phtpy)₂]³⁺ and in agreement with the experimental observations our TD-DFT simulations associate the main absorption feature at 340 nm with a strongly dipole-allowed intraligand charge transfer (ILCT, ph → tpy) into S₄ (at 3.27 eV, 378 nm, Figure c and Table S4). Similarly, [Co­(dqp)₂]³⁺ shows an absorption band with mixed LC/LMCT character (LC = π–π* ligand-centered, LMCT = ligand-to-metal charge transfer) at slightly lower energy, appearing at 360 nm (ε = 26,000 M^–1 cm^–1), while the main electronic transition, predicted at 369 nm (into S₁₂, 3.36 eV, Figure d and Table S8), is related to a local excitation of the π-system of the coordinated dqp ligands with only very minor LMCT character. The bathochromic shift of the absorption band at 360 nm in [Co­(dqp)₂]³⁺ compared to that at 340 nm in [Co­(phtpy)₂]³⁺ is likely the result of the increased aromatic system on the dqp ligand.

4.

Top row: Experimental UV–vis absorption spectra in CH₃CN (filled); experimental UV–vis transient absorption spectra in deaerated CH₃CN at room temperature following excitation at 355 nm with picosecond pulses (∼6 mJ per pulse) time-integrated over 20 ns (solid traces); kinetic transient absorption traces at the indicated wavelengths, and their monoexponential fitting results (insets). The individual data sets correspond to a. [Co­(phtpy)₂]³⁺ and b. [Co­(dqp)₂]³⁺; TD-DFT calculated state energies (S₀ – singlet ground state geomerty, T₁ – first excited triplet state geometry, corresponding to the ³T₁ excited state) with their character assignment (see the insets for the color coding), and simulated ground-state absorption spectra for the respective complexes: c. [Co­(phtpy)₂]³⁺; d. [Co­(dqp)₂]³⁺.

Additionally, weak absorption bands observed in both spectra in the visible region beyond 400 nm, with a molar absorptivity below 500 M^–1 cm^–1, can be attributed to the spin-forbidden MC transitions. Gaussian deconvolution analysis of those weak bands in both [Co­(phtpy)₂]³⁺ and [Co­(dqp)₂]³⁺ (Figure ), allowed to determine the ¹MC (¹T₁) and ³MC (³T₁) transition energies, required for the ligand field parameter analysis. The mixed ¹MC states S₁–S₃ in [Co­(phtpy)₂]³⁺ are predicted by the calculations at 490–432 nm (2.53–2.87 eV) with slight LMCT contributions, similarly in [Co­(dqp)₂]³⁺ the ¹MC states appear at 506–480 nm (S₁–S₃, 2.45–2.58 eV), while ¹LMCT states are predicted at 442 and 410 nm (S₄ and S₅, 2.81 and 3.02 eV), see Tables S4 and S8 for details.

5.

Low-energy parts of the experimental UV–vis absorption spectra of a. [Co­(phtpy)₂]³⁺ and b. [Co­(dqp)₂]³⁺ in CH₃CN (brown) with Gaussian deconvolution and assignment of the ¹MC (¹T₁) and ³MC (³T₁ and ³T₂) electronic transitions. The cumulative fitted curve is shown with dashed black lines. MC transition energies: a. ΔE (¹T₁–¹A₁) = 22,600 cm^–1, ΔE (³T₂–¹A₁) = 16,600 cm^–1, ΔE (³T₁–¹A₁) = 14,500 cm^–1; b. ΔE (¹T₁–¹A₁) = 21,700 cm^–1, ΔE (³T₁–¹A₁) = 14,300 cm^–1.

We could not confidently determine the energies of all of the MC transitions required for a precise estimation of the 10 Dq and Racah parameters in [Co­(dqp)₂]³⁺ and [Co­(phtpy)₂]³⁺ due to the tailing of intense CT transitions in the same spectral range. However, the observed electronic transitions from the singlet ground state (¹A₁) to the ¹T₁ and ³T₁ excited states allowed an estimation of 10 Dq and the Racah parameter C for both Co^III complexes (see Figure and the footnote of Table for details). The relevant parameters for the [Co­(bpy)₃]³⁺ complex have been reported previously, enabling a comparison of 10 Dq values across the series [Co­(phtpy)₂]³⁺, [Co­(bpy)₃]³⁺, and [Co­(dqp)₂]³⁺. Within this series, the symmetry around the Co^III center becomes progressively more octahedral while the Σ parameter decreases (Table ), mirroring the trend observed for the Cr^III complexes shown in Figure . For the 10 Dq values, however, opposite trends are observed for Co^III and Cr^III. Whereas 10 Dq increases for Cr^III across the ligand series tpy < bpy < dqp (as discussed in the Introduction), it decreases for Co^III along the series phtpy > bpy > dqp (Table ). This fundamentally different behavior of Cr^III and Co^III complexes can be rationalized by taking into account the effective nuclear charges. Due to its smaller ionic radius, Co^III has a higher effective nuclear charge, resulting in lower energies of its t_2g orbitals and a closer energy alignment with the ligand π-orbitals than in the case of Cr^III. Consequently, while the tpy/phtpy, bpy, and dqp ligands act as π-acceptors to Cr^III, they behave as π-donors toward Co^III. Whereas π-acceptors enhance the ligand field strength, π-donors lead to weaker ligand fields (Figure b,c), consistent with the trends observed in Table .

1. Ligand Field Parameters, Racah Parameters, and Octahedral Distortion Parameters in Co^III and Cr^III Complexes along with the Lifetimes of Their Lowest Electronically Excited States ,,,,,

Complex	10 Dq†/cm–1	C†/cm–1	τ(3T1)/ns	Σ/°	Complex	10 Dq/cm–1	τ(2E)/μs	Σ/°
[Co(phtpy)2]3+	26,650	4050	2.6	65	[Cr(tpy)2]3+	20,390	30	104
[Co(bpy)3]3+	25,900	3730	5.0	47	[Cr(bpy)3]3+	23,300	63	67
[Co(dqp)2]3+	25,400	3700	8.3	13	[Cr(dqp)2]3+	24,900	1200	29

^a10 Dq and Racah parameter C values for [Co­(phtpy)₂]³⁺ and [Co­(dqp)₂]³⁺ were approximated using the Tanabe–Sugano formalism: ΔE(³T₁–¹A₁) = 10 Dq – 3C, ΔE (¹T₁–¹A₁) = 10 Dq – C (experimental values were extracted from the data in Figure ).

^bAcetonitrile solution, room temperature.

^cAqueous solution, room temperature.

This is in line with recent work demonstrating that bidentate and tridentate polypyridine ligands in Co^III complexes primarily act as π-donors. This behavior contrasts the π-acceptor character of polypyridines in second- and third-row transition metal complexes, and supports the previously observed trend that [Co­(tpy)₂]³⁺, with its more distorted bite angles, exhibits a stronger ligand field splitting than [Co­(bpy)₃]³⁺ (Table ). McCusker and colleagues proposed that this difference between tpy and bpy complexes arises due to a weaker π-donation by the tpy ligand owing to less favorable metal–ligand orbital overlap (Figure b), leading to a weaker destabilization of the t_2g orbitals and a (slightly) stronger ligand field in the tpy complex compared to the bpy complex (Figure c). Following that logic in our case, the π-donation is strongest with the bite-angle optimized dqp ligand, leading to the weakest ligand field. The calculated triplet energy of [Co­(dqp)₂]³⁺ lies at 1.26 eV (see T₁ in Figure d), reflecting an approximately 0.1 eV stabilization of the lowest ³T₁ state compared to 1.35 eV in [Co­(phtpy)₂]³⁺ (see T₁ Figure c). The calculated decrease in the ³T₁ energy between these two bis­(tridentate) coordination environments thus aligns with the experimentally observed decrease of 10 Dq, at least in terms of the overall trend.

Prolonged Excited-State Lifetimes Despite Lower Energies

The excited-state dynamics of the Co^III complexes were investigated via picosecond UV–vis transient absorption spectroscopy in deaerated acetonitrile at room temperature and in combination with computational modeling. Following the excitation of [Co­(phtpy)₂]³⁺ at 355 nm, multiple excited-state absorption (ESA) bands could be observed throughout the visible region (Figure a), decaying with a lifetime of 2.6 ns. In this case, TD-DFT simulations enabled the assignment of these ESA spectral signatures to electronic transitions from the lowest ³MC state to higher-lying excited states. In particular, the ESA in [Co­(phtpy)₂]³⁺ was associated with a low-lying ³LMCT transition (into T₉ at 647 nm) as well as with ³ILCT excitation into T₃₆ at 379 nm (Figure S20 and Table S6).

Shifting our focus to the complex with an optimized bite angle [Co­(dqp)₂]³⁺, its TA spectrum is also compatible with the lowest excited state of ³MC character (Figure S22), with a lifetime of 8.3 ns. ESA bands of the ³MC state are observed at 410, 465, and 650 nm (Figure b), along with a ground-state bleach (GSB) at 390 nm, corresponding to the LC/LMCT band in the ground-state absorption spectrum. The computational analysis assigns these ESA bands mostly to an LMCT absorption at 627 nm (T₁₁), several highly mixed and weakly allowed transitions (e.g., T₃₁, T₃₃ and T₃₅ between 432 and 402 nm) as well as with a strongly allowed LC transition (T₅₄, 364 nm; Figure S22 and Table S10).

Across the series of Co^III complexes reported in Table , the lifetime of the lowest ³T₁ excited state increases as the ligand field strength decreases (i.e., as 10 Dq decreases) and the energy of the ³T₁ state becomes lower. Specifically, the ³T₁ lifetime increases from 2.6 ns in [Co­(phtpy)₂]³⁺ to 5.0 ns in [Co­(bpy)₃]³⁺ and to 8.3 ns in [Co­(dqp)₂]³⁺ (Table , Figure a,b). This contrasts the Marcus inverted region behavior recently unraveled for polypyridine complexes of Co^III, where an increase of the ³MC energy led to an increase of the ³MC lifetime. , It seems plausible that the discrepant behavior observed here for [Co­(phtpy)₂]³⁺ and [Co­(dqp)₂]³⁺ could be due to different extents of structural rearrangements in the ³MC excited state relative to the electronic ground state. According to its X-ray crystal structure and the DFT simulations in a solvent environment, [Co­(phtpy)₂]³⁺ (Figure a) has a strong geometric distortion in the electronic ground state (Σ = 65°, Figure a and Table ) due to the strained bite angle (164(10)°) and axial compression, with the two axial Co–N bonds of 1.862(7) Å being considerably shorter than the four equatorial Co–N bonds (1.952(5) Å).

By contrast, the ground state of [Co­(dqp)₂]³⁺ shows minor deviations from the ideal octahedral geometry (Σ = 13°, Figure a and Table ) and reveals significant π-interactions between individual dqp ligands coordinated to the same metal (Figure S23, MOs 165–178), which could impart structural rigidity and greater force constants associated with molecular distortions both in the ground state and in the ³MC excited state. This in turn could lead to more nested ³T₁/ground state potentials (Figure b), making the ³T₁ lifetime more than a factor of 2 longer in [Co­(dqp)₂]³⁺ than in [Co­(phtpy)₂]³⁺, despite a decrease in triplet energy of about 0.1 eV (see above). The ca. 40% increase in the ³T₁ lifetime from [Co­(phtpy)₂]³⁺ to [Co­(bpy)₃]³⁺ is comparatively modest and more difficult to rationalize within a simple qualitative framework. This is especially true given the difference in coordination environments (bis­(tridentate) versus tris­(bidentate)), which complicates direct comparisons. Therefore, it seems more appropriate to focus the following discussion on the tridentate systems, specifically comparing phtpy/tpy-based and dqp-based complexes.

6.

Main electronic and cooperative rigidity effects in Cr^III and Co^III complexes upon bite-angle optimization and extension of the π-conjugation. Anticipated changes in the energies, nuclear coordinates, and force constants of the electronic ground and relevant excited states upon optimization of bite angles, π-conjugation, and cooperative rigidity in simplified illustrations with harmonic potential well diagrams. Q = nuclear coordinate, δq = nuclear coordinate displacement/distortion in the excited state relative to the ground state (GS), and E _a = activation energy for relaxation from ³T₁ to ¹A₁.

This leads us to the proposed picture in Figure b, where the ³T₁ energy decreases from [Co­(phtpy)₂]³⁺ to [Co­(dqp)₂]³⁺ due to enhanced π-donation. However, the barrier (E _a) for nonradiative relaxation from ³T₁ to the singlet ground state (¹A₁) increases, as a result of larger force constants associated with the relevant potential energy surfaces, thereby extending the lifetime of the ³T₁ state. In the case of the structurally comparable Cr^III complexes, the increased σ-donation and π-acceptance from [Cr­(tpy)₂]³⁺ to [Cr­(dqp)₂]³⁺ shifts the ligand-field dependent ⁴T₂ state to higher energies (Figure a). This reduces the likelihood of reverse intersystem crossing from the luminescent ²E state to the ⁴T₂ state, thereby prolonging the ²E lifetime. ,, It seems plausible that intramolecular π–π-interactions in [Cr­(dqp)₂]³⁺ could induce a similar rigidifying effect as proposed for [Co­(dqp)₂]³⁺, contributing to the exceptionally slow deactivation rate of the luminescent ²E state reported previously.

To conclude this section, the picture for Cr^III in Figure a has become a widely accepted model in the field and has been found applicable to several other d-block metal species, ,,,,, including Fe^II, ,, Ru^II, − ,, Mn^I, and Cr⁰. ,,, Optimization of the bite angles enhances the ligand field strength and prolongs excited-state lifetimes. The model for Co^III shown in Figure is unconventional, demonstrating that optimization of bite angles with π-donor ligands decreases the ligand field strength. Nonetheless, prolonged excited-state lifetimes can still be achieved through the simultaneous optimization of rigidity.

Energy Losses between Excitation and the Photoactive Excited State

Co^III polypyridines have emerged as exceptionally strong oxidizing agents in their photoactive ³T₁ excited states, surpassing the oxidizing power of several benchmark Ir^III photocatalysts. This is particularly remarkable given the significant energy loss that occurs between photoexcitation and the photochemical reaction in this class of compounds. Co^III polypyridines typically absorb strongly only in the blue and UV regions of the spectrum, yet their ³T₁ state retains just ∼1.35 eV of energy, indicating that more than 50% of the energy from a 400 nm excitation photon is not retained. In the following, we aim to gain a deeper insight into the initial energy dissipation processes within the [Co­(dqp)₂]³⁺ complex.

Specifically, we performed femtosecond UV–vis TA spectroscopy in acetonitrile at room temperature and used spectroelectrochemistry to aid the assignment of the observed spectral features, recognizing the limitations of this approach. , The TA spectra of [Co­(dqp)₂]³⁺ (Figure , top) were analyzed in the spectral region from 390 to 490 nm using global fitting, revealing three species-associated spectra (SAS) (Figure S3). The SAS₂ corresponds to the long-lived ³T₁ state (8.3 ns); its time component was fixed as a constant for the global fit. The initially populated state (SAS₀) is tentatively assigned to a ¹ILCT state because a distinct ESA band at 460 nm matches with a broad absorption band in the differential absorption spectrum following ligand reduction in this complex (Figures S2a and S4). Unfortunately, the spectral features of the oxidized ligand are outside the spectral region accessible by the femtosecond TA experiment (Figures S2c and S3). We propose that this ¹ILCT state undergoes vibrational cooling and an intersystem crossing to the triplet manifold (SAS₁) within <1 ps. The spectral features of SAS₁ are reminiscent of the lower-lying ³T₁ state (SAS₂), differing only in the somewhat more intense ESA band at 410 nm. A lifetime of 23 ps can be attributed to the internal conversion (IC) and vibrational cooling (VC) processes between the respective higher- and lower-lying ³MC excited states. This relatively slow process is phenomenologically reminiscent of Cu^I complexes, , where prolonged ISC and IC lifetimes can result from a geometrical rearrangement in the excited-state landscape.

7.

Top: UV–vis transient absorption spectra of [Co­(dqp)₂]³⁺ in acetonitrile at 293 K, recorded at different delay times following excitation at 360 nm with pulses of ca. 190 fs duration; bottom: experimental TA spectrum, obtained at a delay time of 1 ns (solid line) and simulated TA spectrum of [Co­(dqp)₂]³⁺ (black bars) at the TD-DFT level of theory with the spin density of the triplet state displayed.

While nanosecond excited-state lifetimes are typically required for diffusion-controlled bimolecular reactions, productive photoreactions can occur on the picosecond time scale when the photocatalyst and substrate molecules are preaggregated. ,, A lifetime of 23 ps would, in principle, be sufficient for such processes. It is therefore conceivable that in appropriately preorganized systems such as pairs with anionic reaction partners, − the 23 ps excited-state lifetime permits bimolecular (anti-Kasha) reactivity. This would represent an unconventional strategy for minimizing energy losses between excitation and the ensuing photochemical reaction.

To explore whether such a reaction can be observed, we attempted a photoinduced bimolecular triplet–triplet energy transfer (TTET) from the [Co­(dqp)₂]³⁺ complex to perylene. Perylene has a triplet energy of 1.53 eV, significantly higher than the energy accessible from the relaxed ³T₁ state of [Co­(dqp)₂]³⁺ (1.26 eV) (Figure S9b). Thus, we were speculating that TTET could perhaps occur prior to internal conversion to the ³T₁ state and its vibrational cooling (Figures S9b and d), accessed via preaggregation of the Co^III complex and perylene. The NMR titration experiment suggests the presence of the anticipated complex-substrate aggregates in the ground state (Figures S6–S8). The corresponding Benesi–Hildebrand plot yielded an association constant of K _a = 190 ± 50 M^–1 in acetonitrile at room temperature.

Polypyridine Co^III complexes are known to form highly oxidizing species in their lowest triplet excited state (³T₁), , and [Co­(dqp)₂]³⁺ is no exception, exhibiting an excited-state reduction potential of E _red* = +1.59 V vs SCE. This excited-state redox potential is sufficient to oxidize perylene (E _ox = +0.85 V vs SCE) via a highly exergonic (ΔG _ET ⁰ = −0.74 eV) photoinduced bimolecular single-electron transfer (SET) process. Therefore, upon photoexcitation of [Co­(dqp)₂]³⁺ in the presence of perylene, we anticipated to observe, using TA spectroscopy, the formation of both the perylene radical cation (perylene^+·) via the SET mechanism from the fully relaxed ³T₁ state of [Co­(dqp)₂]³⁺ and the triplet perylene (³perylene) via the TTET mechanism potentially involving anti-Kasha behavior. ,, Contrary to our initial expectations, we could only observe the formation of perylene^+· (Figure S9a) with its characteristic ESA signal at 540 nm and no evidence for an energy transfer from the higher triplet excited states of [Co­(dqp)₂]³⁺.

In an attempt to shut down ordinary diffusion-controlled SET reactivity from the relaxed ³T₁ excited state of [Co­(dqp)₂]³⁺, as observed in the case of perylene discussed above, we aimed to explore the possibility of SET to more strongly oxidizing triplet excited states (with excited-state potentials exceeding 1.85 V vs SCE), accessible only through anti-Kasha reactivity of [Co­(dqp)₂]³⁺. To this end, we investigated the possibility of photoinduced SET between [Co­(dqp)₂]³⁺ and redox-challenging aromatic compounds such as durene (E _ox = +1.75 V vs SCE), 3,3′-dimethyl biphenyl (E _ox = +1.84 V vs SCE) and biphenyl (E _ox = +1.95 V vs SCE). , However, in all cases, no evidence of radical cation formation was observed, suggesting that both thermodynamic and kinetic electron transfer remained unfavorable. Direct laser spectroscopic evidence for photoreactivity from higher excited states of [Co­(dqp)₂]³⁺ is therefore difficult to provide at present, but could perhaps be found more indirectly in long-term light irradiation experiments, such as those carried out in classical synthetic photoredox catalysis.

The inherent photostability of [Co­(dqp)₂]³⁺ in deaerated acetonitrile at room temperature is promising for potential photocatalytic applications, as the photodegradation quantum yield under these conditions is remarkably low, only 0.003% (Figure S5). To put this value into perspective, the benchmark complex [Ru­(bpy)₃]²⁺ has been reported to exhibit a significantly higher photodegradation quantum yield under comparable conditions. Furthermore, among the Co^III complexes, [Co­(ppy)₃] and [Co­(CN)₆]^3– were previously shown to undergo photodissociation from the ³T₁ excited state. , The [Co­(ppy)₃] complex undergoes significant Jahn–Teller distortion in an electronically excited state, leading to its ∼9 ps deactivation via Co–C_ph bond rupture. The photoaquation reaction of [Co­(CN)₆]^3– occurs in a similar fashion.

Conclusions

Our work reveals two counterintuitive insights into the molecular design principles and photophysical behavior of d-metal complexes. First, the widely accepted paradigm that optimizing the coordination bite angle strengthens the ligand field is not supported in this study. ,,,,,,,, Instead, we find that such optimization can actually weaken the ligand field when the ligands function primarily as π-donors. Second, the well-known “energy gap law,” which states that nonradiative excited-state relaxation accelerates as the excited-state energy decreases, does not hold in the case investigated here. This is because intramolecular rigidification exerts a counteracting influence that can significantly prolong excited-state lifetimes. Both of these key findings are, in principle, predictable based on ligand field theory (Figure ) and simple considerations of the relevant potential well diagrams (Figure ). Although π-donor ligands in photoactive metal complexes have been receiving growing attention, ,,,,,− and the concept of cooperative rigidity among ligands coordinated to the same metal center is becoming increasingly well established, ,− ,, the direct comparison between [Co­(phtpy)₂]³⁺ and [Co­(dqp)₂]³⁺ presented in this work provides rare and fundamental insight into both of these aspects.

Our study further suggests the potential for conducting bimolecular photochemistry from higher electronically excited states. This approach could help minimize energy losses between light absorption and photochemical energy storage, especially in Co^III polypyridines, which absorb in the blue spectral range but have low-lying excited states that store significantly less energy than those of traditional Ru^II polypyridines or cyclometalated Ir^III complexes.

The specific insights gained from our study can guide the future designs of photoactive complexes based on both abundant first-row and noble or non-noble second- and third-row transition metals, as metal-centered excited states and their deactivation pathways universally play a critical role in determining the photophysical and photochemical properties. The broader concept that a ligand may act as a π-acceptor with one metal but as a π-donor with another could become increasingly relevant as unconventional metal oxidation states are more widely explored, and novel ligand frameworks continue to emerge. Similar effects to those observed in this study may therefore be anticipated in yet-to-be-discovered metal complexes, where a delicate interplay between metal and ligand orbital energies, orbital overlap, and intramolecular π–π interactions shapes the excited-state energy landscape that ultimately governs luminescence properties and photochemical behavior.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.5c09616.

• Experimental procedures, materials and methods; synthesis and characterization of complexes, single crystal X-ray data, NMR and HR-ESI mass spectra, photophysical and electrochemical data as well as computational details and in-depth quantum chemical results (PDF)

Swiss National Science Foundation (grant number 200020_207329) and Deutsche Forschungsgemeinschaft (DFG) within the TRR 234 CataLight (project A4; project no. 364549901).

The authors declare no competing financial interest.