(8-Amino)quinoline and (4-Amino)phenanthridine Complexes of Re(CO)₃ Halides

Abstract

In this report, we present a study on the synthesis, structure, and electronics of a series of (8-amino)quinoline and (4-amino)phenanthridine complexes of Re(CO)₃X, where X = Cl and Br. In all cases, the (amino)heterocycles bind as bidentate ligands, with surprisingly symmetric modes of binding based on Re-N bond lengths. Between the complexes of (8-amino)quinolines and (4-amino)phenanthridines studied in this report, we do not observe much structural variation, and remarkably similar UV-visible absorption spectra. Expansion of the π-system in the (4-amino)phenanthridine complexes does result in an increase in the intensity of the lowest energy transitions (λ_max), which computational modeling suggests are more purely MLCT in character compared with the mixed π-π*/MLCT character of these transitions in the smaller (8-amino)quinoline-supported complexes. DFT and TDDFT modeling further showed that consideration of spin-orbit coupling (SOC) is essential; omitting SOC misses the π-π* contributions to λ_max and is unable to accurately model the observed electronic absorption spectra.

Introduction

The simple amine-functionalized heterocycle (8-amino)quinoline (Figure 1) has been used as a metal chelate for decades, with the first reports of metal carbonyl bound examples appearing in 1959 [1]. As a ligand, (8-amino)quinoline usually binds in a bidentate fashion via the two nitrogen positions, although examples of bridging binding modes are also known [2]. Typically, the ligand binds as a neutral species, but can occasionally lose a proton from the exocyclic nitrogen atom position and bind as an anion [3,4]. In addition to studies into the fundamental coordination chemistry of this ligand, complexes of (8-amino)quinoline and its derivatives have been used to catalyze organic transformations, as well as in construction of materials with novel physical properties. In the former case, (8-amino)quinoline has been used as a directing group for C-H bond functionalization [5], and in the latter these ligands have been used as components of metal binding azo dyes [6].

Figure 1:

Syntheses for compounds 1–10.

Despite its ready availability, the coordination chemistry of (8-amino)quinoline has not been explored nearly as extensively as that of other bidentate systems, such as 2,2’-bipyridine or phenanthroline. There have been some investigations into the main group [7–9] and lanthanide [10] element coordination chemistry, with more focus on transition metals, especially elements of the first row [2,11]. Nevertheless, overall, comparably little has been reported on the organometallic chemistry of this chelate, even though some of the first examples of metal complexes of (8-amino)quinoline were in fact group VI metal carbonyl compounds [1]. In addition to the under-explored regions of the periodic table, the impact of chemical modification of (8-amino)quinoline on metal binding has also not been extensively probed. Recently, Herbert and coworkers have been exploring the coordination chemistry of phenanthridine-based multidentate ligands, which can be considered π-expanded analogs of quinolines (phenanthridine = 3,4-benzoquinoline) [12]. Comparing the properties of (4-amino)phenanthridines and (8-amino)quinoline as ligands, therefore, provides an opportunity to study how π extension (‘benzannulation’) affects the asymmetric bidentate binding mode of (8-amino)quinoline.

In this report, we present the first study into the Re(CO)₃ chemistry of (8-amino)quinoline and its π-expanded phenanthridine congener. Specifically, we examined the coordination of (8-amino)quinoline, (8-amino-6-methyl)quinoline, as well as three comparable (4-amino)phenanthridines, with methyl, tert-butyl, and trifluoromethyl substitution at the 2-position. In all cases, the ligand binds to the metal with expected bidentate coordination. The resultant complexes have been fully characterized, and the structures of most of the Re(CO)₃ adducts elucidated by single-crystal X-ray diffraction methods. We have also probed their electronic structures using DFT and TDDFT methods, which are used to interpret their electronic absorption spectra.

Experimental Section

All reagents and starting materials were purchased from commercial vendors and used without further purification. The phenanthridine ligands were prepared as previously described: (4-amino-2-methyl)phenanthridine [12], (4-amino-2-tert-butyl)phenanthridine [13], (4-amino-2-trifluoromethyl) phenanthridine [14]. Deuterated solvents were purchased from Cambridge Isotope Laboratories and used as received. NMR spectra were recorded on 750 MHz, 500 MHz, and 300 MHz spectrometers and chemical shifts were given in ppm relative to residual solvent resonances (¹H NMR and ¹³C{¹H} NMR spectra). High-resolution mass spectrometry experiments were performed on a Bruker MicroTOF-III instrument. Infrared spectra were collected on Thermo Scientific Nicolet iS5 which was equipped with an iD5 ATR. UV-visible spectra were recorded on a Hitachi 3010 spectrometer.

X-ray intensity data were measured on a Bruker CCD-based diffractometer with dual Cu/Mo ImuS microfocus optics (Cu Kα radiation, λ = 1.54178 Å, Mo Kα radiation, λ =0.71073 Å). Crystals were mounted on a cryoloop using Paratone oil and placed under a steam of nitrogen at 100 K (Oxford Cryosystems). The detector was placed at a distance of 5.00 cm from the crystal. The data were corrected for absorption with the SADABS program. The structures were refined using the Bruker SHELXTL Software Package (Version 6.1), and were solved using direct methods until the final anisotropic full-matrix, least squares refinement of F² converged. Tables S9–S11 lists the X-ray data collection and structure parameters for all structures in this report.

Synthesis of 1. The synthesis of 1 is representative for the preparations of 2–4 but with an alternate rhenium starting material (Re(CO)₅Br) or (8-amino)quinoline. A sample of 50 mg of Re(CO)₅Cl (0.14 mmol) dissolved in 10 mL of hot toluene and 20 mg of (8-amino)quinoline (0.14 mmol) were mixed together and allowed to reflux for 2 h. After cooling to room temperature, the mixture was cooled to −5 °C to promote precipitation. The solid was then filtered, washed with toluene, and dried in a vacuum oven. Yield: 54 mg (86%). IR (CO stretch, cm⁻¹): 2016 (s), 1911 (m), 1866 (s). ESI MS (negative mode) [M-H]⁻ calculated C₁₂H8ClN₂O₃Re 448.98 m/z, found 448.80. ¹H NMR (300 MHz, DMSO-d₆, ppm): δ = 9.22 (dd, 1H; Ar CH), 8.70 (dd, 1H; Ar CH), 7.99 (m, 1H; Ar CH), 7.73 (m, 4H; 2 Ar CH and 2 NH₂), 6.95 (d, 1H; Ar CH). ¹³C{¹H} NMR (300 MHz, DMSO-d₆, ppm): δ = 154.62 (C Ar), 145.92 (C Ar), 141.37 (C Ar), 139.71 (C Ar), 130.03 (C Ar), 129.89 (C Ar), 128.66 (C Ar), 127.03 (C Ar), 123.90 (C Ar), carbonyl quartenary carbon positions not observed.

2. Yield: 83 mg (93%). IR (CO stretch, cm⁻¹): 2019 (s), 1914 (m), 1872 (s). ESI MS (negative mode) [M-H]⁻ calculated C₁₂H₈BrN₂O₃Re 492.93 m/z, found 492.70. ¹H NMR (300 MHz, DMSO-d₆, ppm): δ = 9.22 (dd, 1H; Ar CH), 8.70 (dd, 1H; Ar CH), 7.99 (m, 1H; Ar CH), 7.68 (m, 4H; 2 Ar CH and 2 NH₂), 6.95 (d, 1H; Ar CH). ¹³C{¹H} NMR (300 MHz, DMSO-d₆, ppm): δ = 154.62 (C Ar), 145.92 (C Ar), 141.37 (C Ar), 139.71 (C Ar), 130.03 (C Ar), 129.89 (C Ar), 128.66 (C Ar), 127.03 (C Ar), 123.90 (C Ar) carbonyl quartenary carbon positions not observed.

3. Yield: 55 mg (88%). IR (CO stretch, cm⁻¹): 2016 (s), 1890 (m), 1868 (s). ESI MS (Positive mode) [M-Cl-H]⁺ calculated C₁₃H₁₀N₂O₃Re 429.024 m/z, found 429.031. ¹H NMR (300 MHz, DMSO-d₆, ppm): δ = 9.12(d, 1H; Ar CH), 8.57 (d, 1H; Ar CH), 7.54 (m, 4H; 2 Ar CH and 2 NH₂), 6.88 (d, 1H; Ar CH), 2.52 (s, 3H; CH₃). ¹³C{¹H} NMR (300 MHz, DMSO-d₆, ppm): δ = 153.58 (C Ar), 144.49 (C Ar), 141.08 (C Ar), 138.85 (C Ar), 138.67 (C Ar), 131.39 (C Ar), 129.76 (C Ar), 125.62 (C Ar), 123.86 (C Ar) and 21.31 (CH₃) carbonyl quartenary carbon positions not observed.

4. Yield: 56 mg (92%). IR (CO stretch, cm-1): 2016(s), 1890 (m), 1871(s). ESI MS (negative mode) [M-Br-H]- C₁₃H₁₀N₂O₃Re calculated 427.01 m/z, found 427.02. 1H NMR (300 MHz, DMSO-d₆, ppm): δ = 9.13 (dd, 1H; Ar CH), 8.57 (dd, 1H; Ar CH), 7.54 (m, 4H; 2 Ar CH and 2 NH₂), 6.92 (d, 1H; Ar CH), 2.53 (s, 3H; CH₃). ¹³C NMR{¹H} (300 MHz, DMSO-d6): δ = 198.05 (CO), 197.07(CO), 192.94 (CO), 153.82 (C Ar), 144.70 (C Ar), 141.15 (C Ar), 138.88 (C Ar), 138.74 (C Ar), 131.48 (C Ar), 129.77 (C Ar), 125.72 (C Ar), 123.98 (C Ar) and 21.38 (CH₃).

Synthesis of 5–10. The synthesis of 5 is representative for the conditions used for the generation of compounds 6–10 but using an alternate rhenium starting material (Re(CO)₅Br) or aminophenanthridine. Re(CO)₅Cl (0.17 mmol), an equivalent amount of (4-amino-2-methyl)phenanthridine (0.17 mmol), and 8 mL of THF were mixed together and allowed to reflux for 18 hrs. After cooling to room temperature, the mixture was filtered, washed with THF, hexanes, and Et₂O. Afterwards, the resulting off-white powder was dried in a vacuum oven. Yield: 46 mg (52 %). IR (CO stretch, cm⁻¹): 2021(vs), 1913(s), 1869(vs). ESI MS (positive mode) calcd C₁₇H₁₁N₂O₃Re 478.0328 m/z, found 478.0543. ¹H NMR (300 MHz, DMSO-d₆, ppm): δ = 9.76 (s, 1H), 8.89 (d, 1H), 8.59 (s, 1H), 8.53 (d, 1H), 8.11 (t, 1H), 7.91 (t, 1H), 7.70 (br d, 1H, N-H), 7.54 (s, 1H), 6.91 (br d, 1H, N-H), 2.63 (s, 3H, CH₃). ¹³C{¹H} NMR (500 MHz, DMSO-d₆, ppm): δ = 197.86 (CO), 196.83 (CO), 192.81 (CO), 156.93(C₆), 141.18 (C_Ar), 139.37 (C_Ar), 138.90 (C_Ar), 133.87 (C_Ar), 131.49 (C_Ar), 130.29 (C_Ar), 129.20 (C_Ar), 129.05 (C_Ar), 126.33 (C_Ar), 125.49 (C_Ar), 122.66 (C_Ar), 121.00 (C₃), 21.12 (CH₃).

6. Yield: 30 mg (50%). IR (CO stretch, cm⁻¹): 2023 (vs), 1911 (s), 1866 (vs). ESI MS (positive mode) calcd C₂₀H₁₇N₂O₃Re 520.0747 m/z, found 520.0884. ¹H NMR (300 MHz, DMSO-d₆, ppm): δ = 9.79 (s, 1H), 9.05 (d, 1H), 8.67 (s, 1H), 8.55 (d, 1H), 8.13 (t, 1H), 7.93 (t, 1H), 7.79 (s, 1H), 7.72 (br d, 1H, N-H), 6.68 (br d, 1H, N-H), 1.49 (s, 9H, -C(CH₃)₃. ¹³C{¹H} NMR (500 MHz, DMSO-d₆, ppm): δ = 197.80 (CO), 196.82 (CO), 192.79 (CO), 157.26 (C₆), 151.60 (C_Ar), 141.02 (C_Ar), 139.26 (C_Ar), 133.88 (C_Ar), 131.79 (C_Ar), 130.34 (C_Ar), 129.02 (C_Ar), 126.33 (C_Ar), 125.90 (C_Ar), 125.19 (C_Ar), 122.89 (C_Ar), 117.31 (C₃), 35.16 (C(CH₃)₃, 31.10 (CH₃). Crystals suitable for X-ray diffraction were grown by vapor diffusion of Et₂O into a DMF solution.

7. Yield: 48.5 mg (61%). IR (CO stretch, cm⁻¹): 2024 (vs), 1874 (br vs). ESI MS (positive mode) calcd C₁₇H₉F₃N₂O₃Re 533.0123 m/z, found 533.0100. ¹H NMR (300 MHz, DMSO-d₆, ppm): δ = 10.01 (s, 1H), 9.21 (s, 1H), 9.15 (d, 1H), 8.64 (d, 1H), 8.19 (t, 1H), 7.99 (m, 2H and N-H), 7.88 (s, 1H), 7.29 (br d, 1H, N-H). ¹⁹F NMR (300 MHz, d₆ - DMSO): δ = −60.34 (s, 3F, -CF₃). ¹³C{¹H} NMR (500 MHz, DMSO-d₆, ppm): δ = 197.55 (CO), 196.45 (CO), 192.51 (CO), 160.49 (C₆), 143.15 (C_Ar), 142.89 (C_Ar), 134.77 (C_Ar), 131.52 (C_Ar), 130.79 (C_Ar), 130.14 (C_Ar), 128.59 (C_Ar), 128.32 (C_Ar), 126.58 (C_Ar), 126.22 (C_Ar), 123.92 (C_Ar), 123.86 (CF₃), 119.69 (C₃).

8. Yield: 16 mg (40%). IR (CO stretch, cm⁻¹): 2015 (vs), 1866 (br vs). ESI MS (negative mode) calcd C₁₇H₁₂BrN₂O₃Re 557.9572 m/z, found 557.9568. ¹H NMR (300 MHz, DMSO-d₆, ppm): δ = 9.76 (s, 1H), 8.88 (d, 1H), 8.57 (s, 1H), 8.51 (d, 1H), 8.09 (t, 1H), 7.87 (m, 2H and N-H), 7.53 (s, 1H, 3), 6.96 (br d, 1H, N-H), 2.62 (s, 3H, CH₃). ¹³C{¹H} NMR (500 MHz, DMSO-d₆, ppm): δ = 197.45 (CO), 196.32 (CO), 192.31 (CO), 157.12 (C₆), 141.16 (C_Ar), 139.53 (C_Ar), 138.90 (C_Ar), 133.87 (C_Ar), 131.48 (C_Ar), 130.23 (C_Ar), 129.19 (C_Ar), 129.04 (C_Ar), 126.39 (C_Ar), 125.43 (C_Ar), 122.67 (C_Ar), 121.03 (C₃), 21.11 (CH₃).

9. Yield: 16 mg (52%). IR (CO stretch, cm⁻¹): 2023 (vs), 1913 (s), 1864 (vs). ESI MS (negative mode) calcd C₂₀H₁₈BrN₂O₃Re 600.0042 m/z, found 600.0001. ¹H NMR (300 MHz, DMSO-d₆, ppm): δ = 9.78 (s, 1H, 6), 9.05 (d, 1H), 8.66 (s, 1H), 8.53 (d, 1H), 8.11 (t, 1H), 7.88 (m, 2H and N-H), 7.77 (s, 1H), 6.92 (br d, 1H, N-H), 1.47 (s, 9H, -C(CH₃)₃. ¹³C{¹H} NMR (500 MHz, DMSO-d₆, ppm): δ = 197.37 (CO), 196.30 (CO), 192.26 (CO), 157.44 (C₆), 151.59 (C_Ar), 140.98 (C_Ar), 139.39 (C_Ar), 133.86 (C_Ar), 131.76 (C_Ar), 130.27 (C_Ar), 129.01 (C_Ar), 126.39 (C_Ar), 125.88 (C_Ar), 125.11 (C_Ar), 122.90 (C_Ar), 117.37 (C₃), 35.15 (C(CH₃)₃, 31.09 (CH₃).

10. Yield: 32 mg (65%). IR (CO stretch, cm⁻¹): 2021 (vs), 1874 (br vs). ESI MS (negative mode) calcd C₁₇H₉BrF₃N₂O₃Re 611.9289 m/z, found 611.9197. ¹H NMR (300 MHz, DMSO-d₆, ppm): δ = 10.04 (s, 1H), 9.22 (s, 1H), 9.16 (d, 1H), 8.64 (d, 1H), 8.21 (t, 1H), 8.12 (br d, 1H, N-H), 8.03 (t, 1H), 7.91 (s, 1H), 7.32 (br d, 1H, N-H). ¹⁹F NMR (300 MHz, d₆ - DMSO): δ = −60.31 (s, 3F, -CF₃). ¹³C{¹H} NMR (500 MHz, DMSO-d₆, ppm): δ = 197.65 (CO), 196.45 (CO), 192.49 (CO), 161.16 (C₆), 143.66 (C_Ar), 143.57 (C_Ar), 135.27 (C_Ar), 132.00 (C_Ar), 131.23 (C_Ar), 130.64 (C_Ar), 129.08 (C_Ar), 128.81 (C_Ar), 127.15 (C_Ar), 126.67 (C_Ar), 123.93 (C_Ar), 123.87 (CF₃), 120.26 (C₃).

Computational Experimental

ORCA version 4.1.2 [15,16] was used for single point and TD-DFT calculations, to explicitly account for both scalar relativistic effects and spin-orbit coupling. It has been previously shown that SOC can have significant effect in the observed optical spectra of Re(I) carbonyl complexes of N-heterocyclic ligands [17,18]. Simulating SOC-corrected spectrum within the TDDFT framework can be approximately accounted for by allowing the single-triplet transitions to borrow intensity from singlet-singlet transitions. This can be expressed as a linear combination of scalar singlet-singlet and singlet triplet excitations, with oscillator strength of the spin corrected spectrum expressed as in Equation 1:[18]

$$f_{soc}=\frac{\langle {\psi }_S\left|{\widehat{\text{h}}}_{SOC}\right|{\psi }_T\rangle }{E_T-E_S}{f}_S$$ (1)

where the numerator is the SOC matrix element, the denominator is the difference in energy of the coupled triplet and singlet state, and ƒ_S is the oscillator strength of the singlet state. Solvent and dispersion effects were accounted for using the solvation model based on density (SMD; solvent=CH₃CN) [19] and Grimme’s D3 dispersion correction with the Becke-Johnson damping (D3BJ) [20]. The ground state geometries of the complexes were optimized using Gaussian 16, Rev. B.01 [21]. employing the PBE0 [22] functional and def2-SVP [23,24] basis set on non-metal atoms and the corresponding ecp basis set for Re (SMD-PBE0-D3(BJ)/def2-SVP). Crystal structure coordinates were used as input coordinates and frequency calculations were performed to confirm all optimized structures were at a minimum.

To probe the effects of HF exchange we employed two functionals: the modified global hybrid B3LYP* (HFX=15%) [25,26] and the range-separated hybrid CAM-B3LYP (HFX=19–65%) [27,28]. The ‘resolution of identity with chain-of-sphere’ approximation was used in all single point and TD-DFT calculations (RIJCOSX; intaccx: 4.01, 4.01, 4.34; gridx 1,1,2) [29]. Scalar relativistic and SOC effects were included using the two-component zeroth order regular approximation (ZORA) [30] as implemented in ORCA. The ZORA all-electron relativistic basis sets, def2-TZVP was used for all non-metal elements (H, C, N, O and Cl) while the SARC-TZVP was used for the heavy element Re [24] with their corresponding auxiliary basis sets [31–35]. SCF convergence and grid criteria were set to TightSCF and grid5, respectively, while the grid criteria for the final energy evaluation was set to FinalGrid6. The first 50 excited singlet and triplet-excited states were considered, which produced 200 SOC-states. Scalar-only and SOC-corrected spectra were simulated using Multiwfn version 3.6 software and fragment contribution to the MOs were generated using the same program [36,37] using the Hirshfeld partition method [38]. All structures and MOs were generated using Avagadro [39].

Results and Discussion

In the rhenium carbonyl halide complexes studied, both (8-amino)quinoline and (8-amino-6-methyl)quinoline bind to the metal ion in the expected bidentate fashion, as neutral ligands. Both bromide and chloride complexes were fully characterized, including via X-ray crystallography for the (8-amino)quinoline structures (Figure 2). We were not able to fully structurally elucidate the (8-amino-2-methyl)quinoline-supported 3 due to crystal quality issues, but were able to confirm connectivity and the coordination modes are identical to those seen in 4. There are two equivalents of complex in the asymmetric unit of compound 1 and 4; only one is shown in the figure. These complexes exhibited the expected geometry around the metal center, with a facial arrangement of carbonyl ligands. Table 1 lists selected bond lengths and angles for compounds 1, 2 and 4. Interestingly, similar metal-nitrogen bond distances are observed for both the N-heterocyclic nitrogen [1: 2.196(8) and 2.174(8) Å; 2: 2.179(4) Å; 4: 2.180(5) and 2.186(5)] and the exocyclic NH₂ [1: 2.206(7) and 2.197(7) Å; 2: 2.197(5) Å; 4: 2.198(5) and 2.191(5)], which is notable considering the differences in the Lewis basicity of these two donor sites. The N-Re-N angles are acute and identical within error for all three complexes [1: 76.5(3)° and 76.6(3)°; 2: 76.63(16)°; 4: 76.63(19)° and 76.82(19)°]. The Re-C, Re-X, and C-O bond lengths and angles are typical for Re(CO)₃ compounds with bidentate, nitrogenous ligands.

Figure 2:

The structures of 1, 2 and 4 with 35% thermal ellipsoids. Hydrogen atoms and select atom labels have been omitted for clarity.

Table 1:

Selected bond lengths (Å) and angles (°) for compounds 1, 2 and 4. Where two sets of values are given, two molecules were found in the asymmetric unit.

Metric	1	2	4
Re(1)-N(2) (amine) (Å)	2.207(7)	2.197(5)	2.198(5)
Re(2)-N(4)	2.199(7)	-	2.191(5)
Re(1)-N(1) (ring) (Å)	2.195(8)	2.179(4)	2.180(5)
Re(1)-N(1)	2.174(8)	-	2.186(5)
Re(1)-C(1) (Å)	1.889(9)	1.915(6)	1.915(6)
Re(1)-C(2)	1.931(11)	1.922(6)	1.918(7)
Re(1)-C(3)	1.923(9)	1.911(6)	1.898(7)
Re(2)-C(13) (Å)	1.917(9)	-	1.891(7)
Re(2)-C(14)	1.931(11)	-	1.916(7)
Re(2)-C(15)	1.893(9)	-	1.915(7)
C(1)-O(1) (Å)	1.162(11)	1.129(6)	1.152(7)
C(2)-O(2)	1.142(13)	1.154(7)	1.151(8)
C(3)-O(3)	1.153(12)	1.153(7)	1.145(8)
C(13)-O(4)	1.159(11)	-	1.156(8)
C(14)-O(5)	1.142(13)	-	1.152(7)
C(15)-O(6)	1.163(11)	-	1.155(8)
N(1)-Re(1)-N(2) (°)	76.5(3)	76.63(16)	76.63(19)
N(3)-Re(2)-N(4)	76.6(3)	-	76.82(19)

We were able to determine the solid-state structures of all the (4-amino)phenanthridine complexes, including both chloro and bromo variants. The structures of 5-10 are shown in Figure 3. The binding of the ligands in the (4-amino)phenanthridine complexes is identical to that seen in the (8-amino)quinoline compounds, and no significant structural deviations are observed. The structures of 8 and 9 each contain two molecules in the asymmetric unit. Table 2 shows the selected bond lengths and angles for compounds 5–10. The Re-N bond lengths do show occasionally some variation between the cyclic and exocyclic nitrogen atoms, for example, as can be seen in the values of 2.165(7) and 2.205(6) Å respectively for compound 7. These variations are in relative good agreement with those seen in the corresponding chloride complex 1. The bite angle of the chelate in each of the aminophenanthridine compounds (~76°) is identical to that seen in the (8-amino)quinoline complexes. The remainder of the coordination sphere in each of 5–10 (Re-Cl, Re-C, and C-O bond lengths and angles) are quite similar to those of the non-π expanded systems.

Figure 3:

The structures of 5–10 with 35% thermal ellipsoids. Hydrogen atoms and some atom labels have been omitted for clarity.

Table 2:

Selected bond lengths (Å) and angles (°) for compounds 5–10.

Metric	5	6	7	8	9	10
Re(1)-N(2) (amine) (Å)	2.200(3)	2.206(5)	2.196(4)	2.248(10)	2.198(5)	2.200(3)
Re(2)-N(4)	-	-	-	2.184(8)	2.184(4)	-
Re(1)-N(1) (ring) (Å)	2.182(3)	2.165(6)	2.173(4)	2.166(9)	2.180(4)	2.191(3)
Re(2)-N(3)	-	-	-	2.183(8)	2.178(4)	-
Re(1)-C(1) (Å)	1.925(3)	1.908(7)	1.920(5)	1.919(13)	1.917(7)	1.906(4)
Re(1)-C(2)	1.995(3)	1.932(9)	1.951(6)	1.916(12)	1.916(6)	1.895(4)
Re(1)-C(3)	1.918(3)	1.931(8)	1.922(5)	1.890(12)	1.912(7)	1.915(4)
Re(2)-C(18b)	-	-	-	2.058(11)	1.962(12)	-
Re(2)-C(19b)	-	-	-	1.925(11)	1.972(6)	-
Re(2)-C(20b)	-	-	-	1.919(11)	1.909(7)	-
C(1)-O(1) (Å)	1.158(4)	1.156(8)	1.145(6)	1.141(14)	1.148(7)	1.160(5)
C(2)-O(2)	1.064(4)	1.079(10)	1.073(6)	1.148(13)	1.141(7)	1.156(5)
C(3)-O(3)	1.157(4)	1.138(10)	1.147(6)	1.140(13)	1.155(7)	1.148(5)
C(21a/18b)-O(4)	-	-	-	0.940(11)	1.143(12)	-
C(22a/19b)-O(5)	-	-	-	1.154(13)	1.052(6)	-
C(23a/20b)-O(6)	-	-	-	1.148(12)	1.152(7)	-
N(1)-Re(1)-N(2) (°)	76.90(9)	76.88(14)	76.5(2)	76.3(3)	76.57(17)	76.23(12)
N(3)-Re(2)-N(4)	-	-	-	76.0(3)	76.49(16)	-

^aCompound 8.

^bCompound 9.

For all ten compounds, the ¹H NMR spectra exhibit expected chemical shifts, with very downfield shifted NH₂ signals. The IR spectra for each compound exhibit strong bands resulting from the CO stretching vibrations which appear at ~2015–2025 for the symmetric stretch and ~1870–1910 for the asymmetric stretches. Between the (8-amino)quinoline and (4-amino)phenanthridine complexes, the most noticeable feature, albeit small, in spectroscopic features can be seen in their UV-visible spectra. For the quinoline complexes, the UV-visible spectra show what appear to be metal-to-ligand charge transfer (MLCT) bands at energies less than 400 nm (Figure 4). Moving to the phenanthridine ligand results in few changes to the UV-visible spectra. This is somewhat counterintuitive, as one might expect a significant red shift due to a stabilization of the π* system of the phenanthridine ligand, assuming that the metal d orbital energies are approximately the same in both sets of complexes. However, weak absorption bands above 500 nm are observable in the spectra of some of the phenanthridine complexes, notably in the CF₃ containing complexes 9 and 10.

Figure 4:

UV-Visible spectra of quinoline compounds 1–4 (top) and benzannulated phenanthridine analogs 5–10 (bottom).

Electronic Structure Investigation

To help interpret the transitions observed by electronic absorption spectroscopy, computational modeling was performed. The structures of compounds for which crystal structure data was available were optimized to validate a theoretical approach. Good agreement between DFT-derived geometric values and experimental ones was observed using SMD-PBE0-D3(BJ)/def2-SVP (Table S1). For instance, the optimized metal-ligand bonds deviate, on average, from those observed experimentally in the solid-state by 0.013 Å (1), 0.028 Å (5), and 0.019 Å (7). Satisfied with our ability to reproduce relevant structural metrics in silico, we then investigated the electronic structures of 3 and 5 in more depth by explicitly accounting for scalar relativistic effects using the ZORA approach and relativistic all-electron basis sets for all atoms. Focusing on this pair of molecules controls for substitution pattern of the heterocyclic fragment (R = CH₃ for both), allowing us to isolate and probe the impact of benzannulation.

Within the framework of the spin-free approximation, the electronic states can be classified as singlets (S = 0, M = 1) and triplets (S = 1, M = 3). Chemically relevant information can then be extracted, providing a molecular orbital character or the electron-hole pair distribution of the resulting SOC-excited states that form upon photoexcitation. Molecular orbital (MO) diagrams and orbital contour plots for the spin-free singlet ground states of 3 and 5 are shown in Figure 5. The three highest energy occupied MOs for both complexes (HOMO, HOMO-1 and HOMO-2) are predominantly localized on the d⁶ rhenium centers, with some (d-π*) back-bonding interactions with two of three carbonyl ligands evident for each MO. The HOMO and HOMO-1 additionally exhibit anti-bonding (d-p)π* character with the chloride ligand, as befits a π-donor substituent. The mixing of the filled ‘t_2g-like’ orbitals with ligand-based lone-pair and π*-orbitals become significant in complexes containing π-donor ligands (e.g. halides, amides, alkoxides, etc.) [40].

Figure 5:

Molecular orbital diagrams of the spin free singlet ground states of 3 and 5 (RIJCOSX-ZORA-SMD-B3LYP*-D3(BJ)/def2-TZVP+SARC-ZORA-TZVP//SMD-PBE0-D3(BJ)/def2-SVP). Orbital contour plots (isosurface = 0.05) are shown for the orbitals most relevant to the transitions observed by UV-Vis absorption spectroscopy.

The highest occupied molecular orbitals of 3 and 5 (Figure 4) lie at similar energies due to their similar character; the identity of the nitrogenous chelate does not appear to significantly perturb the energies of the HOMO and HOMO-1. The HOMO-3 (and HOMO-4 for 5) is comprised of the π system of the heterocyclic ligands and is higher in energy for the larger phenanthridinyl π-system. As expected, the LUMO is localized on the heterocyclic ligands, and heavily at the C=N sub-unit of each heterocyclic ligand. Phenanthridines exhibit more localized ‘imine-like’ character at the 6-position C=N compared with closely related heterocycles, due to the position of benzannulation maximizing aromatic character within the two flanking six-carbon rings [41]. Previously reported phenanthridine-containing complexes of lighter transition metal complexes have been shown to exhibit similar character to the LUMO, which engender rich photophysical properties to the complexes [12,13,42–46]. Accordingly, the LUMO of 5 is more localized at the C=N of the heterocycle compared with the LUMO of 3 (28% vs. 36%; see Tables S5 and S8).

Despite the difference in LUMO character between 3 and 5, the energies of the LUMO for both complexes are essentially the same (Figure 5) resulting in comparable HOMO-LUMO energy gaps. This is consistent with the experimental absorbance spectra of 3 and 5 which show similar spectral features and energies. A comparable effect was observed in Group 10 complexes of bis(quinolinyl)amides and their corresponding benzannulated, phenanthridine-containing congeners [12]. For the compounds here, in particular, both spectra contain a weak absorbance band covering the entire visible region of the electromagnetic spectrum with maximum at 535 nm and 538 nm (~2.31 eV) for 3 and 5, respectively. The effect of benzannulation seems to be more drastic with the energies of LUMO+1, where a significant stabilization of the energy of LUMO+1 in 5 compared to 3 (−1.7 eV for 5 vs −1.2 eV for 3). In both cases, the LUMO+1 is largely of N-heterocycle anti-bonding character, delocalized largely in the aryl fragment of the N-heterocycle. The increased in delocalization with more bonding character to the LUMO+1 of 5 could be a reason for the observed stabilization with respect to 3.

Spin-Free vs. Spin-Orbit Coupling Theoretical Absorption Spectra

Spin-free (SF-TDDFT) and spin-orbit corrected (SOC-TDDFT) time-dependent DFT (TDDFT) calculations were used to simulate UV-Vis absorption spectra and interpret the observed optical properties of the rhenium complexes. Explicit inclusion of spin-orbit coupling effects have been previously reported to drastically affect the excited state dynamics of complexes containing heavier elements [17,18]. Other recent work has concluded that SOC is not important to model the experimental absorption spectra of Re(I) carbonyl complexes, but does have an effect on the potential energy surface [47] and excitation energies [48]. For systems 3 and 5, we found that the inclusion of SOC and using a functional with minimal HF-exchange (B3LYP*, HF = 15%) in TDDFT calculations allowed us to more accurately reproduce the experimental absorption spectra of 3 and 5 (Figures S36 and S37).

By comparing the TDDFT calculated spectrum to the experimental spectrum of 3 and 5, it can be seen that SOC is particularly important in modeling the low energy transitions of the spectrum. In both 3 and 5, a weak absorption band at ~ 2.31 eV can be observed. In comparison, spin-free spectra only show the higher energy absorption bands. In addition, the inclusion and the amount of HF-exchange have a profound effect in the energy of the lowest energy band. Within the spin-free approximation, modeling singlet-triplet excitations using global hybrid functionals with small amount of HF-exchange (HF < 30%) and GGA-functionals is typically less accurate [49]. This can be remedied using global hybrid functionals with increased HF-exchange or range separated hybrids [50] which highlights the difference in the electron-correlation requirement for triplet states [51,52]. Consistent with these findings, the energy of T₁ at the optimized ground state geometry is predicted to be higher in CAM-B3LYP compared to B3LYP*. In both cases, excitation to this low energy excited triplet state is forbidden (f = 0), consistent with the spin multiplicity conservation rule. The SOC-corrected TDDFT also shows a bathochromic shift in the absorbance spectra of 3 and 5 when using B3LYP* compared to CAM-B3LYP. However, only B3LYP* with SOC was able to reproduce the experimental spectra of both 3 and 5. This suggests that the electron-correlation for SOC-states in these complexes may be similar to spin-free singlet states, preferring hybrid functionals with low amounts of HF admixture.

We therefore pursued the calculations performed using the B3LYP* functional with SOC-correction for 3 and 5. In the case of SOC-corrected TDDFT, the calculated energies of the dominant transition in the lowest energy absorption band for 3 (2.27 eV) and 5 (2.28 eV) deviate experimentally (2.31 eV) by about 0.04 eV. Further analysis on the SOC-corrected spectra of 3 and 5 indicate that the broad peak at 2.31 eV is largely dominated by the SOC-state 2 for both 3 and 5. Composition analysis reveal that triplet states contribute significantly to these SOC-states (3: 0.56T₂ + 0.40T₄; 5: 0.82T₂ + 0.14T₅). Singlet state contribution to these SOC-states for 3 (S₂ ~ 1%) and 5 (S₁ ~ 1%) is minor resulting in the observed weak molar absorptivities (ε < 250 M⁻¹ cm⁻¹). The molecular orbital characters of the contributing spin-free excited triplet states indicate that this SOC-state has mixed π-π* and ML_quinCT in 3 (Table S4), and purely ML_phenCT in 5 (Table S7), indicating significant configurational interactions in 3 containing the smaller 6-methyl-(8-amino)quinoline ligand. The extent of the MLCT-IL configurational interaction depends on the complex structure and the nature of the ligands [53,54]. This indicates that there is a larger redistribution of electron density upon excitation at 2.31 eV for 5 than in 3, which could have significant effects in the excited state dynamics and reactivity of the complexes [53].

In conclusion, we have produced the first series of Re(CO)₃ (8-amino)quinoline and aminophenanthridine complexes starting from the corresponding pentacarbonyl halides. In all cases, the ligands act as bidentate nitrogenous chelates, and we see little variation between the (8-amino)quinoline and aminophenanthridine compounds. In addition, the Re-N bond lengths in these compounds are remarkably similar despite the heterocyclic and exocyclic nature of these nitrogen atom positions. Spectroscopically, we also observed very similar properties for both the (8-amino)quinoline and (4-amino)phenanthridine compounds, with the largest difference in the intensity of the lowest energy transitions in the visible region of the spectra of the benzannulated complexes. These could be successfully modeled only using SOC-TDDFT calculations, which revealed that, in spite of differences in the character of LUMOs of the aminoquinoline and aminophenanthridine compounds, the HOMO-LUMO gap remains largely unchanged, underscored by the similarities in their electronic absorption spectra. Notably, omitting SOC misses the additional π-π* of the lowest energy transitions in the quinolinyl-supported complex 3; the benzannulated phenanthridinyl analog, on the other hand, exhibits more purely MLCT character to these absorptions. A similar effect in benzannulated complexes of bis(quinolinyl)amide ligands was exploited to yield tune emission energy from complexes with isoenergetic absorption profiles [41]. We are presently investigating if a similar effect operates in emissive analogs of 1-10.

Highlights:

The first (8-amino)quinoline and (4-amino)phenanthridine complexes of Re(CO)₃ are presented.

Both ligands exhibit effectively identical structural features when bound to Re(CO)₃

The 4-amino)phenanthridine complexes show an increase in the intensity of the lowest energy metal to ligand charge transfer (MLCT) transitions,