Phenylamino Derivatives of Tris(2-pyridylmethyl)amine: Hydrogen-Bonded Peroxodicopper Complexes

Abstract

A series of copper complexes bearing new 6-substituted tris(2-pyridylmethyl)amine ligands (L^R) appended with NH(p-R-C₆H₄) groups (R=H, CF₃, OMe) were prepared. These ligands are electronically tunable (ΔE_1/2 = 160 mV) and Cu^I(L^R)⁺ complexes react with oxygen to form hydrogen bonded (trans-1,2-peroxo)dicopper species.

Graphical abstract

Copper-containing oxygenase and oxidase enzymes are an important class of metalloenzymes whose diverse O₂ activation pathways facilitate a wide variety of biological functions.¹ Although challenging to study in the native enzymes, the study of O₂ binding and activation by copper within easily modifiable synthetic systems has greatly expanded our understanding of these metalloenzymes,² and the first crystallized Cu-O₂ adduct was the (trans-1,2-peroxo)dicopper complex of the tris(2-pyridylmethyl)amine (tpa) ligand.³ In the nearly 30 years since that structure, modifications to tpa and similar ligand frameworks have been targeted to elucidate the key factors responsible for O₂ binding and activation. Although critical to the function of many metalloenzymes,⁴ the impact of secondary coordination sphere hydrogen bonding (H-bonding) interactions on copper-oxygen species is not well understood.⁵ For example, H-bonding interactions have been demonstrated to either stabilize⁶ or destabilize⁷ Cu-O₂(H) adducts. For this reason, synthetic systems bearing tunable second sphere H-bonding groups provide a facile means to study their influence on Cu-O₂ binding and activation.

Our lab recently introduced the tris(6-hydroxyl-2-pyridylmethyl)amine (H₃thpa) ligand that incorporates three –OH groups within the secondary sphere of tpa (Fig. 1A).⁸ The pendent –OH groups are potent H-bond donors to metal-coordinated substrates and facilitate proton and electron transfer reactions.⁹ However, these complexes underwent facile formation of dinuclear copper species in the presence of weak bases, which limited investigations of subsequent reactivity. To overcome this limitation, we targeted pendent phenylamino groups as less acidic, sterically protected H-bond donors (Fig 1A).¹⁰ In contrast to previously reported –NHCO^tBu substituted tpa variants, the phenylamino group provides both steric and electronic flexibility that allows them to act as highly tunable H-bond donors. In this communication we present a series of aniline-appended tripodal ligands that feature highly directed H-bonding interactions of varied acidity and highlight the design concept by demonstrating H-bond dictated O₂ reactivity.

Figure 1.

Cu^ICl complexes derived from H₃thpa and L^R (A). Cyclic voltammetry of 1^H, 1^CF3 and 1^OMe ((B) 0.1 M NBu₄PF₆ CH₂Cl₂). Crystal structure of 1^H ((C) 30% ellipsoids, H atoms not involved in H-bonding omitted for clarity).

The ligand tris(6-phenylamino-2-pyridylmethyl)amine (L^H) was prepared via a Buchwald-Hartwig coupling of tris(6-bromo-2-pyridylmethyl)amine (Br₃tpa) with aniline, Pd(OAc)₂, and rac-BINAP. The addition of CuCl to L^H in THF afforded the yellow complex Cu(L^H)Cl (1^H) in 59% yield after 48 h. A crystal of 1^H suitable for X-ray diffraction was grown by layering pentane over a concentrated toluene solution at −30°C (Fig 1C). The solid-state structure revealed C₃-symmetry (R-3 space group) with directed H-bonds from the pendent –NHPh units to the Cl ligand (N–Cl: 3.311(4) Å). These H-bonds are significantly longer than the previous Cu(H₃thpa)Cl complex bearing pendent –OH groups (N_avg–Cl: 3.048 Å) consistent with weaker H-bond interactions.⁸ In addition, the Cu–N_axial and Cu–Cl bonds in 1^H, at 2.252(4) Å and 2.3398(14) Å respectively, are shortened compared to Cu(H₃thpa)Cl (2.283(2) Å and 2.5661(6) Å respectively). Overall, the phenylamino groups in 1^H provide a sterically protected pocket for the Cl ligand while allowing for further electronic tuning by modifying the identity of the aniline.

Two ligand derivatives featuring electronically distinct, yet sterically similar H-bond donors were prepared. 4-Trifluoromethylaniline and 4-methoxyaniline afforded ligands L^CF3 and L^OMe respectively, which were metalated with CuCl to provide Cu(L^CF3)Cl (1^CF3) and Cu(L^OMe)Cl (1^OMe). The electronic influence of each ligand variant was interrogated by analysis of the Cu^I/II redox couple (Fig 1B). Complex 1^H exhibits a reversible Cu^I/II redox couple at −470 mV vs Fc (CH₂Cl₂; 0.1 M NBu₄PF₆). This value is shifted to more negative potentials relative to the –OH complex Cu(H₃thpa)Cl (−365 mV vs Fc), consistent with increased electron releasing properties of L^H than H₃thpa. The Cu^I/II redox couple in 1^OMe features the most reducing Cu center at −510 mV vs Fc. The electronically deficient 1^CF3, however, exhibits a Cu^I/II redox couple at −350 mV vs Fc, which highlights the electronic tunability of the L^R ligands with simple substitution on the parent aniline. To contextualize these electrochemically obtained values with a well-defined metric, the potential difference of 1^CF3 (120 mV) and 1^OMe (−40 mV) from the parent aniline (1^H) were plotted against Hammett values (p-substitution) of 0.54 and −0.27 respectively (see SI). The electrochemical shifts show a good correlation (R² = 0.98) with these Hammett values, indicating that they may be used to provide a predictive measure ligand field effects within the tpa scaffold.

The electronic variability in the L^R ligand series was also evident by ¹H NMR and IR spectroscopy. All three complexes are C₃-symmetric by ¹H NMR spectroscopy (CD₂Cl₂) and show a significantly downfield shifted –NH resonance consistent with H-bonding interactions between the –NH and the Cl ligand. The –NH peak appears at δ=9.88 in 1^H whereas it shifts downfield (δ=10.17) in 1^CF3 and upfield (δ=9.72) in 1^OMe. The magnitude of the two shifts, +0.29 and −0.16 respectively, are again consistent with the expected proportion based on Hammett parameters (R² = 0.99, see SI). The weakening of the –NH bond upon H-bonding is also evident by infrared spectroscopy (CH₂Cl₂) where the –NH stretching frequency shifts from 3431 cm⁻¹ for L^H to 3223 cm⁻¹ in 1^H. However, the value of the –NH stretch for 1^CF3 and 1^OMe does correlate directly with the Hammett value. The –NH stretch in 1^OMe shifts to lower energy (3220 cm⁻¹) consistent with a stronger H-bond interaction, while the –NH stretch in 1^CF3 falls between 1^H and 1^OMe at 3221 cm⁻¹. The IR data demonstrate the difficulty of assigning trends in a complex system where electronic character of the metal and M-X unit is highly coupled to both H-bond donor strength and H-bond acceptor strength.

The –NHPh appended tpa derivatives provide a tunable framework for studying O₂ binding to Cu and the resulting H-bonded (trans-1,2-peroxo)dicopper complexes. The complex [Cu^I(L^H)]BAr’ (BAr’= [B(C₆F₅)₄]⁻) was prepared by mixing L^H and Cu(MeCN)₄BAr’ in CH₂Cl₂ under an inert atmosphere. ¹H NMR spectroscopy was used to verify formation of a C₃-symmetric Cu(I) complex. Cooling a CH₂Cl₂ solution of [Cu^I(L^H)]BAr’ to −70°C and introducing dry O₂ afforded the (trans-1,2-peroxo)dicopper complex [(Cu(L^H))₂(O₂)][BAr’]₂ (2^H). The reaction was confirmed by a color change in solution (colorless to brown) and the appearance of a band at 457 nm (ε= 3000 M⁻¹cm⁻¹) in the electronic absorption spectrum, assigned as the oxygen to copper ligand-to-metal charge transfer (LMCT) band. This contrasts with the unsubstituted complex [(Cu(tpa))₂(O₂)][PF₆]₂ which exhibits a primary LMCT band at 525 nm (ε= 11500 M⁻¹cm⁻¹) and a shoulder at 590 nm (ε= 7600 M⁻¹cm⁻¹).³ In 2^H we propose that the directed H-bonding interactions to the peroxide unit reduces peroxide electron donation into Cu. This effect weakens the Cu–O bond and results in a hypsochromic shift of the LMCT. Furthermore, the hypochromic shift and loss of the shoulder band in 2^H was also observed by Masuda and co-workers in a similar H-bonded system and was postulated to be due to decreased rotational freedom of the peroxo unit.^5f Complex 2^H was subjected to additional spectroscopic characterization.¹¹ The EPR (X-band) spectrum for 2^H is silent, consistent with other S=1 integer spin (trans-1,2-peroxo)dicopper complexes. Upon warming to room temperature, solutions of 2^H convert to a green, EPR active species, suggesting decomposition to monomeric Cu(II) byproducts. The UV-Vis and EPR data were corroborated by solid-state characterization of 2^H.

A crystal of 2^H suitable for X-ray diffraction was grown in CH₂Cl₂ in a −80° freezer under an atmosphere of oxygen over 3 days (Fig 2B), constituting the first crystallographically characterized H-bonded (trans-1,2-peroxo)dicopper complex. For each ‘Cu(L^H)’ in 2^H only one –NHPh group engages in H-bonding with the proximal oxygen of the peroxo unit (N5–O1: 2.723(7) Å) while the other two –NHPh groups engage the distal oxygen (N6–O2: 2.859(7) Å and N7–O2: 2.851(7) Å). Previous examples of H-bonded Cu-O₂ species have shown that H-bonding to the proximal oxygen generally increases stability of the species while H-bonding to the distal oxygen decreases stability.^5g 2^H contains both proximal and distal H-bonds and stable in solution at −70°C for at least 8 hours, consistent with a net stabilizing effect to the O–O unit. The weakening of the Cu–O bond covalency in 2^H, observed by UV-Vis, was also corroborated by the solid-state data. The Cu–O bond at 1.902(3) Å is elongated compared to the unsubstituted complex [(Cu(tpa))₂(O₂)][PF₆]₂ (1.852(4) Å). The O–O bond in 2^H (1.477(5) Å) is also elongated compared to the parent tpa complex (1.433(9) Å). Although the O–O bond might be expected to shorten as a result of H-bonding interactions with the peroxide,^{5e, 5f} additional structural factors may be responsible for the bond elongation. The steric profile provided by the –NHPh groups on L^H limit the possible interatomic distance between Cu centers. The Cu^…Cu distance in 2^H (4.691(1) Å) is 0.3 Å longer than that observed for [(Cu(tpa))₂(O₂)][PF₆]₂ (4.358(3) Å). Despite these steric considerations, stability may also be augmented by π-π interactions between the pendent –NHPh groups and the opposing pyridine rings (π-π distance: 3.4–3.6 Å).

Figure 2.

Reactivity of Cu(I) complexes toward O₂ (A). Crystal structure of 2^H ((B) 30% ellipsoids, some atoms represented in wireframe, H atoms not involved in H-bonding and BAr’ counteranions omitted for clarity).

The electronic tuning provided by L^CF3 and L^OMe regulates the energy of the O to Cu LMCT. [Cu(L^CF3)]BAr’ and [Cu(L^OMe)]BAr’ readily bind O₂ at −70°C to form the analogous (trans-1,2-peroxo)dicopper complexes [(Cu(L^CF3))₂(O₂)][BAr’]₂ (2^CF3) and [(Cu(L^OMe))₂(O₂)][BAr’]₂ (2^OMe). While the series of complexes (2^R) are all brown in solution, the LMCT band shifts as a function of the electronic character at the metal.¹² 2^CF3 features a 7 nm hypsochromic shift of the LMCT, while 2^OMe features a 3 nm bathochromic shift from the parent 2^H. The observed shifting of the LMCT contrasts with previously reported substituted tpa ligands, where the addition of 4-OMe groups to the pyridine backbone had no effect on the LMCT.¹³ In 2^R, the shift of the LMCT bands is in accordance with the ligands’ respective electronic (Hammett) parameter (R² = 0.99, see SI) and may be a product of both the Cu effective charge and H-bond donor strength. The oxidation potential of the halide-free [Cu^I(L^R)]BAr’ complexes, obtained by square-wave voltammetry, provided an additional description of the electronic character at Cu. [Cu(L^H)]BAr’ in MeCN (0.1M NBu₄PF₆) exhibits an irreversible oxidation event at +110 mV vs Fc. The associated L^CF3 and L^OMe Cu(I) complexes feature redox potentials shifted by +90 mV and −20 mV from L^H, respectively. Importantly, these values are significantly more positive than [Cu(tpa)]PF₆ (E_ox = −386 mV vs Fc, see SI). O₂ binding to [Cu(L^CF3)]BAr’ at potentials as high as +200 mV vs Fc contrasts with most known Cu(I)-tpa variants that exhibit diminished O₂ reactivity at more positive potentials.¹⁴ These data indicate that the Cu(I) centers in [Cu(L^R)]BAr’ are only modestly reducing and might be anticipated to engage in very weak binding to O₂. To account for the facile O₂ reactivity, we propose that the H-bonding groups on L^R provide additional stabilizing interactions for O₂ binding.

A sterically identical ligand to L^H with no H-bond donors was prepared in order to determine the role of H-bonding on the formation of the (trans-1,2-peroxo)dicopper complexes. The ligand tris(6-phenoxy-2-pyridylmethyl)amine (tpa^OPh) containing pendent phenylether groups was prepared via an Ullmann coupling of phenol and Br₃tpa. When tpa^OPh and Cu(MeCN)₄BAr’ were dissolved in CH₂Cl₂ and cooled to −70°C the resulting complex [Cu(tpa^OPh)]BAr’ did not react with O₂. ¹H NMR spectra of [Cu(L^H)]BAr’ and [Cu(tpa^OPh)]BAr’ in CD₂Cl₂ revealed an almost identical chemical shift of the methylene protons, at δ=4.07 and 4.05 respectively, consistent with a similar electronic environment at copper.^{15 1}H NMR spectra of [Cu(L^H)]BAr’ and [Cu(tpa^OPh)]BAr’ exhibit C₃-symmetry at both 25°C and −80°C, however, at −80°C the methylene proton peak on [Cu(tpa^OPh)]BAr’ broadens by 16.8 Hz, consistent with a fluxional process. This observation of dynamic ligand binding may further contribute to the lack of O₂ reactivity with tpa^OPh.¹⁶ Although steric and electronic properties of [Cu(L^R)]BAr’ would suggest that formation of 2^R is unfavorable, these hindrances were overcome by introducing favorable hydrogen bonding interactions.

In conclusion, we have introduced a new variation on the tpa framework that incorporates highly tunable –NHPh groups in the secondary sphere. These groups act as H-bond donors while providing steric protection that can be used to isolate H-bonded Cu^ICl complexes. Cu(I) complexes bearing the new ligands react with O₂ to form H-bonded (trans-1,2-peroxo)dicopper complexes whose spectroscopic characteristics are dictated by the ligand electronics.

Figure 3.

UV-Vis overlay of 2^H, 2^CF3, 2^OMe (1:1 CH₂Cl₂/acetone, −70°C), and the reaction of [Cu(tpa^OPh)]BAr’ with O₂ at −70°C. Inlay shows λ_max of O to Cu LMCT in 2^R.