CO and O₂ Binding to Pseudo-Tetradentate Ligand-Copper(I)-Complexes with a Variable N-Donor Moiety: Kinetic/Thermodynamic Investigation Reveals Ligand Induced Changes in Reaction Mechanism

Abstract

The kinetics, thermodynamics, and coordination dynamics for O₂ and CO 1:1 binding to a series of pseudo-tetradentate ligand-copper(I)-complexes (^DLCu^I) to give Cu^I/O₂ and Cu^I/CO product species are reported. The ^DLCu^I series possess an identical tridentate core structure where the cuprous ion binds to the bispicolylamine (L) fragment. ^DL also contains a fourth variable N-donor moiety {D = benzyl (Bz); pyridyl (Py); imidazolyl (Im); dimethylamino (NMe₂-); tert-butylphenyl pyridyl (TBP); quinolyl (Q)}. The structural characteristics of ^DLCu^I-CO and ^DLCu^I are detailed, with X-ray crystal structures reported for ^TBPLCu^I-CO, ^BzLCu^I-CO, and ^QLCu^I. Infrared studies (solution and solid-state) confirm that ^DLCu^I-CO possess the same four-coordinate core structure in solution with the variable D moiety ‘dangling’, i.e. not coordinated to the copper(I) ion. Other trends observed for the present series appear to derive from the degree to which the D-group interacts with the cuprous ion center. Electrochemical studies reveal close similarities of behavior for ^ImLCu^I and ^NMe₂LCu^I (as well as for ^TBPLCu^I and ^QLCu^I), which relate to the O₂-binding kinetics and thermodynamics. Equilibrium CO binding data (K_CO, ΔH°, ΔS°) were obtained by conducting UV-visible spectrophotometric CO titrations, while CO binding kinetics and thermodynamics (k_CO ; ΔH^‡, ΔS^‡) were measured through variable temperature (193 K – 293 K) transient absorbance laser flash photolysis experiments, λ_ex = 355 nm. Carbon monoxide dissociation rate constants (k_−CO) and corresponding activation parameters (ΔH^‡, ΔS^‡) have also been obtained. CO binding to ^DLCu^I follows an associative mechanism with the increased donation from D leading to higher k_CO values. Unlike that seen in previous work, the K_CO values increased as the k_CO and k_−CO values declines; the latter decreased at a faster rate. By using the ‘flash-and-trap’ method (λ_ex = 355 nm ; 188 K – 218 K), the kinetics and thermodynamics (k_O2 ; ΔH^‡, ΔS^‡) for O₂ binding to ^NMe₂LCu^I and ^ImLCu^I were measured and compared to ^PyLCu^I. A surprising change in the O₂ binding mechanism was deduced from the thermodynamic ΔS^‡ values observed, associative for ^PyLCu^I but dissociative for ^NMe₂LCu^I and ^ImLCu^I; these results are interpreted as arising from a difference in the timing of electron transfer from copper(I) to O₂ as this molecule coordinates and a tetrahydrofuran (THF) solvent molecule dissociates. The change in mechanism was not simply related to alterations in ^DLCu^II/I geometries or the order that O₂/THF coordinate. The equilibrium O₂ binding constant (K_O2 ; ΔH°, ΔS°) and O₂ dissociation rate constants (k_−O2 ; ΔH^‡, ΔS^‡) were also determined. Overall the results demonstrate that subtle changes in the coordination environment, as occurs over time through evolution in nature or through controlled ligand design in synthetic systems, dictate to a critically detailed level the observed chemistry in terms of reaction kinetics, structure and reactivity, and thus function. Results reported here are also compared to relevant copper and/or iron biological systems and analogous synthetic ligand-copper systems.

Introduction

A detailed understanding of copper(II/I) protein active site properties and their 1:1 small molecule (CO, O₂, NO) binding characteristics is of importance and has inspired a diverse array of research efforts for biochemists and synthetic bioinorganic chemists alike.^1-6 Research advances through synthetic modeling have helped to gain important insights into copper coordination and reactivity through generation of ligands with precisely optimized formulations and through study of their ligand-copper/small-molecule chemistry. In this context, however, the ability to directly monitor the formation, structures, and subsequent reactivity of Cu^I/O₂ derived primary species, i.e. mononuclear copper(II)-superoxo complexes formulated as (ligand)Cu^II-O₂^•−, is relatively rare.^7-15 Indeed, such species are implicated as key active-site entities in the (bio)chemistry of dioxygen activating copper enzymes.^16-19

Time resolved laser flash spectroscopy has been widely used to investigate small molecule interactions (CO, O₂, NO, etc.) with natural and synthetic hemes.^20-36 By taking advantage of the extremely short time scale resolution of such photolytic methods, detailed mechanistic information has been achieved. For example, a wealth of knowledge has been gained about both hemoglobin and myoglobin by monitoring photoinitiated CO dissociation from heme-CO in the presence of O₂.³⁷ Also, in the heme-copper protein cytochrome c oxidase (CcO; Figure 1), the characterization of photodriven CO transfer (as a surrogate for O₂ and potentially NO³⁸) from heme_a3-CO to Cu_B suggested that the Cu_B site was a “doorway” for small molecules into and out of the heterobimetallic active site.^3,27 Assuming a common O₂ mechanistic pathway and with some indirect evidence existing,^39-41 a Cu^II_B-O₂^•− intermediate is assumed to precede any heme_a3/O₂ interactions; a putative peroxo-bridged CcO intermediate that was crystallized is shown in Figure 1.

Figure 1.

Overview of reduced copper containing protein active-site O₂ and CO interactions, highlighting the heme_a3-Cu_B active-site for O₂ binding/reduction of mitochondrial cytochrome c oxidase (CcO), the arthropodal and molluscan hemolymph O₂-carrier Hc, and the mammalian neuropeptide hormone producing monooxygenase PHM.

A key aspect for accomplishing such photoinitiated processes is the high binding affinity of heme systems for CO (K_CO > K_O2), as exploited by the “flash-and-trap” experiment pioneered by Gibson and coworkers.^32,33 Similar studies involving CO and O₂ have not been accomplished for the well-characterized type-3 binuclear copper proteins hemocyanin (Hc; Figure 1), tyrosinase (Tyr; widespread monooxygenase, converting phenols to o-catechols and o-quinones), and catechol oxidase (o-catechol → o-quinone), mainly due to the opposite behavior, a high stability (K_O2 > K_CO) of their 2:1 copper-dioxygen adduct, a dicopper(II)-μ-η²:η²-(side-on)-peroxo protein species.^3,16,42,43

Early work established that only one small molecule, CO or O₂, binds per dicopper active site, presumably to Cu_B, and that the close proximity of the two copper centers (Cu_A and Cu_B) enables fast electron-transfer (et) for the two-electron reduction of O₂.^3,16,42,43 However, even following photoexcitation and subsequent dissociation of O₂ from oxy-Tyr⁴⁴ and/or oxy-Hc,^45,46 a Cu^II_B-O₂^•− adduct has not been detected and instead only implicated through theoretical modeling.^16,47 Never-the-less, Hirota, Bubacco, and coworkers recently used flash photolysis and complementary K-edge X-ray absorption spectroscopy (XAS) measurements to show that the O₂ binding rate constant (k_O2) for a deoxy-Hc was dependent on the copper(I) coordination geometry.⁴⁵ In addition, the thermodynamic data (ΔH^‡, ΔS^‡) they obtained were comparable to those reported for mononuclear synthetic model compounds, which emphasizes the important relationship between synthetic bioinorganic chemistry and metallobiochemistry.

In peptidylglycine α-hydroxylating monooxygenase (PHM), the active-site contains two uncoupled copper centers (Cu…Cu ∼ 11 Å; Cu_A ≡ Cu_H; Cu_B ≡ Cu_M). A copper(II)-superoxo species, which is suggested by some to be the key intermediate that initiates substrate oxidation reactions, has been characterized by X-ray crystallography.¹⁸ This “end-on” η¹ -bound (to Cu_M) “precatalytic” copper-dioxygen adduct is shown in Figure 1. Interestingly, CO binds the catalytic (Cu_M) site as well as to the electron-transfer (Cu_H) center, the latter only if substrate is present.^48,49 Key Cu^I/O₂ 1:1 species are also implicated in the enzymatic reactions carried out by copper amine oxidase (CAO)^50-52 and galactose oxidase.¹⁹

As mentioned, CO heme protein active site interactions have been widely investigated. In addition, the use of CO for investigation of copper proteins provides tremendous insights and fundamental information.³ The occurrence of detectable and significant variations in active site copper ligation within a given protein has been noted and is also present in proteins from different species or organisms. The ν_CO values of carbonmonoxy-copper proteins are highly variable, even when equally coordinated by three imidazolyl N-donors. For example, ν_CO values corresponding to the Cu^I-CO adducts of CAO (2061 – 2085 cm^-1),⁵³ the Cu_H-CO site of PHM (2062 cm^-1),⁴⁸ and the Cu_B-CO sites of CcO (2053 – 2063 cm^-1),^54-56 nitrite reductase (NiR; 2050 cm^-1),⁵⁷ and Hc-CO (2043 – 2063 cm⁻¹),⁵⁸ exhibit large differences. Such variations in ν_CO for N₃Cu^I -CO adducts indicates subtle changes in the local active site environment (e.g., coordination geometry changes, dielectric of the medium, etc.) that alter the copper ion's electron density and thus back-donation to the ligated carbon monoxide. This likely relates to or explains the diverse O₂ and/or NO reactivity amongst various copper proteins.

With this background, one of our goals is to develop an understanding of the kinetics and thermodynamics of copper(I) mediated small molecule interactions, specifically in this report with CO and O₂. Carbon monoxide is a good surrogate for dioxygen, even while possessing redox inactive binding characteristics. In addition, copper(I)-carbonyl adducts are typically more stable than their 1:1 copper-dioxygen counterparts and therefore easier to study. By gaining a deeper understanding of how the detailed nature of ligand environment affects CO binding to copper(I), as has been carried out in natural systems, reliable comparisons can be made to Cu^I/O₂ and potentially NO (bio)chemistries.

We previously reported on the CO photodissociation chemistry of a number of cuprous pyridylalkylamine compounds.⁵⁹ Effects upon systematic variations in the ligand framework were examined: (i) via changes in the electron-donating ability of TMPA {^PyL ; tris(2-pyridylmethyl)amine} through addition of 4-pyridyl substituents (R-^PyL ; R = N(CH₃)₂-, CH₃O-); (ii) by an increase in the chelate ring-size from 5-membered to 6-membered ligand coordination to copper as in PMEA {bis[(3-pyridyl)methyl]-2-(2-pyridyl)ethylamine} vs PMAP {bis[2-(2-pyridyl)ethyl]-(2-pyridyl)methyl-amine}; and (iii) with a change in the donor group moieties as in BQPA {bis(2-quinolylmethyl)(2-pyridylmethyl)amine}. The effort advanced our understanding of the coordination environment required for ligand-copper(I)-carbonyl photodissociation and rebinding of carbon monoxide; tridentate coordination from the tetradentate ligand moieties to the copper(I)-carbonyl center (overall four-coordinate) was required (Chart 1), in which one N-donor moiety was ‘dangling’, i.e. uncoordinated. Observed variations in kinetics and thermodynamics were shown to derive from particular changes in the copper complex electron-releasing properties, coordination geometry, and/or steric effects.

Chart 1.

The subject and direction of the work discussed herein examines the effect of changes in the dangling N-donor moiety within various ligand-copper(I)-complexes (^DLCu^I) on both 1:1 CO and O₂-binding kinetics and thermodynamics as well as CO and O₂ coordination dynamics. Variable temperature transient absorbance (TA) laser flash photolysis in tetrahydrofuran (THF) solvent has been employed. The ^DL ligand series consists of potentially tetradentate ligands with the same PY1 {L ; bis(2-pyridylmethyl)amine} core, however with variable N-donor moieties (D), see Chart 1. For the studies of CO binding, a tridentate ligand species ^BzL was also examined as a “standard” for comparison.²⁵ As will be described, drastic differences in Cu^I-CO and Cu^I-O₂ chemistry result from seemingly minor changes in the ^DLCu^I ligand framework.

Experimental

See Supporting Information for details concerning Materials and Methods, references to syntheses of previously published ligands and complexes along with synthetic procedures for new compounds.

Results

Synthesis of Ligand-Copper(I) (^DLCu^I) and Ligand-Copper(I)-Carbonyl (^DLCu^I-CO) Complexes

The ligands (^DL) used in the present study were previously reported and characterized in detail (Chart 1).^60-65 The ligand-copper(I) complexes as B(C₆F₅)₄⁻ salts (^DLCu^I) were straightforwardly prepared by addition of [Cu^I(MeCN)₄]B(C₆F₅)₄ to the appropriate ligand in deoxygenated Et₂O and isolated by slow precipitation in dry air-free pentane under an argon atmosphere.^8,60,64-66 The B(C₆F₅)₄⁻ counteranion was chosen to afford greater solubility in tetrahydrofuran (THF) solvent.

The presence of coordinated acetonitrile (CH₃CN) in the isolated copper(I) complexes of ^PyLCu^I, ^ImLCu^I, ^NMe₂LCu^I, and ^BzLCu^I was confirmed through elemental (C, H, N) combustion analysis and ¹H-NMR spectroscopy studies in deuterated nitromethane (CD₃NO₂).^8,60,64,65 The X-ray crystal structures of ^BzLCu^I and the ClO₄⁻ salts of ^NMe₂LCu^I and ^PyLCu^I were previously reported with each containing an exogenously derived CH₃CN molecule.^60,62,67 As expected for a tridentate ligand system, ^BzLCu^I binds CH₃CN the strongest as supported by the shorter Cu–N_nitrile bond distance of 1.900(4) Å versus a value of 2.038(2) Å for ^NMe₂LCu^I and 1.990(12) Å for ^PyLCu^I, the complexes with tetradentate N4 ligation. Acetonitrile does not coordinate to ^TBPLCu^I and ^QLCu^I based on the characterization methods described above and X-ray structural characterization, vide infra.

Ligand-copper(I)-carbonyl complexes (^DLCu^I-CO) were formed in situ through vigorous CO bubbling into dry THF solutions. ^TBPLCu^I-CO and ^BzLCu^I-CO were isolated as overall four-coordinate species, i.e., possessing a dangling ligand donor arm (vide infra), following dissolution of ^TBPLCu^I and ^BzLCu^I in CO saturated Et₂O and layering with pentane for slow diffusion.⁶⁸ In a previous study, ^PyLCu^I-CO was isolated as a five-coordinate species, i.e., with all ligand N donors coordinated.⁵⁹

X-ray Crystallography of ^TBPLCu^I-CO, ^BzLCu^I-CO, ^QLCu^I

ORTEP diagrams of ^TBPLCu^I-CO (A) and ^BzLCu^I-CO (B) are shown in Figure 2 and that for ^QLCu^I is given in Figure 3; selected bond lengths and angles for all are provided in the figure caption.⁶⁸ Both ligand-copper(I)-carbonyl structures display an overall four-coordinate geometry consisting of the copper(I) ion coordinated by the apical alkylamino nitrogen, two pyridyl donors, and the carbon from a coordinated CO molecule. The N_pyridyl–Cu–N_amine angles of ∼81° are rather severe in comparison to other structurally characterized four-coordinate copper(I) carbonyl complexes which have average N–Cu–N bond angles of ∼95°.^59,69-75 The Cu-C bond length is shorter (stronger) for ^TBPLCu^I-CO (1.801 Å) in comparison to that observed in ^BzLCu^I-CO (1.815 Å), possibly due to an indirect increase in overall electron density in the former complex, with the difference being a tert-butylphenyl (tbp) substituted pyridine vs. benzyl substituent on the L tridentate moiety, see Chart 1.

Figure 2.

ORTEP diagrams of A. ^TBPLCu^I-CO and B. ^BzLCu^I-CO; hydrogen atoms and the B(C₆F₅)₄ counteranion have been omitted for clarity. Selected bond lengths are: A. Cu–C 1.801(6) Å, C–O 1.124(6) Å, Cu–N_alkylamine 2.147(4) Å, Cu–N_{pyridine(avg)} 2.051(5) Å; and B. Cu–C 1.815(1) Å, C–O 1.123(2) Å, Cu–N_alkylamine 2.158(6) Å, and Cu–N_{pyridine(avg)} 2.048(3) Å. Selected bond angles are: A. C–Cu–N_{pyridine(avg)} 121.4(2)°, C–Cu–N_alkylamine 131.0(0)°, N_{pyridine(avg)}–Cu–N_alkylamine 81.0(9)°, N_pyridyl–Cu–N_pyridyl 109.6(0)°; and B. C–Cu–N_{pyridine(avg)} 121.95(7)°, C–Cu–N_alkylamine 129.94(7)°, N_{pyridine(avg)}–Cu–N_alkylamine 80.80(6)°, and N_pyridyl–Cu–N_pyridyl 108.94(6)°.⁶⁸

Figure 3.

ORTEP diagram of ^QLCu^I with hydrogen atoms and the B(C₆F₅)₄ counteranion omitted for clarity. Selected bond lengths are: Cu–N_pyridine(1) 1.988(2) Å, Cu–N_pyridine(3) 2.006(2) Å, Cu–N_{alkylamine(2)} 2.185(2) Å, and Cu–N_quinoline(4) 2.012(2) Å. Selected bond angles are: N_pyridyl(avg)–Cu–N_{alkylamine(2)} 82.84(8)°, N_pyridyl(1)–Cu–N_pyridyl(3) 123.63(8)°, N_pyridyl(1)–Cu–N_quinolyl(4) 119.95(8)°, N_amine(2)–Cu–N_quinolyl(4) 82.58(7)°, and N_pyridyl(3)–Cu–N_quinolyl(4) 111.70(8)°.⁶⁸

A tridentate (N₃) coordination mode for ^TBPLCu^I-CO is of interest due to the tetradentate (N₄) nature of ^TBPL; the 6-tbp-substituted pyridyl arm dissociates, i.e. is dangling, due to the increased steric constraints. However in the absence of CO all three pyridyl donors as well as the bridgehead alkylamine coordinate to the copper(I) ion, as supported by the known X-ray crystal structure of ^TBPLCu^I.⁶⁴ The coordination sphere of ^TBPLCu^I did not include a coordinated acetonitrile molecule as is generally observed for analogous ligand-copper(I) complexes, vide supra.

Similar structural characteristics have been reported based on X-ray crystallographic analysis of the copper(I) complex and copper(I)-carbonyl adduct of BQPA.^59,63,76 For [Cu^I(bqpa)]B(C₆F₅)₄, all four N-donor moieties coordinate to the copper(I) ion and one quinolyl donor arm dissociates upon coordination of CO or triphenylphosphine (PPh₃).⁶³ As shown in Figure 3, the X-ray crystal structure of ^QLCu^I is four-coordinate with both pyridyl donors, the quinolyl donor, and the bridgehead alkylamine coordinated to the cuprous center. An X-ray crystal structure of ^QLCu^I-CO has not been obtained but IR spectroscopic data supports a tridentate ligand coordination in which the quinolyl arm is dangling, vide infra.

Infrared Spectroscopy (ν_CO: Solution and Solid State)

The solution (THF) and solid-state (Nujol mull) ν_CO values of the ligand-copper(I)-carbonyl complexes (^DLCu^I -CO) were determined in order to gauge the ligand-N-donor ability of the variable D-moiety as well as to elucidate potential differences in coordination number and/or the overall geometry. As described above, in part, the CO stretching frequencies (ν_CO) of synthetic copper(I)-carbonyl complexes and carbonmonoxy- copper proteins are found to be in the range ν_CO = 2035 – 2137 cm^-1. ^3,73,77-82 Notably, the characteristic ν_CO values of copper(I)-carbonyl proteins coordinated by three imidazolyl N-donors are usually Δν_CO = 20 – 40 cm^-1 lower than synthetic species with three N-donors. ^3,79 This suggests that copper protein histidine imidazole groups are exceptionally strong donors and relative to most synthetic copper(I)-carbonyl complexes, this leads to weaker C–O (but stronger Cu-C) bonds.

Solution-State IR Spectroscopic and Structural Properties

As discussed in the Introduction (Chart 1), previous work established that an equilibrium mixture of four- (⁴ ν_CO = 2090 cm^-1) and five- (⁵ν_CO = 2077 cm^-1) coordinate copper(I)-carbonyl isomers exist for ^PyLCu^I-CO in THF solvent, Table 1 and Chart 1. Similarly, an equilibrium mixture of ^ImLCu^I-CO isomers exist in THF, ⁴ν_CO = 2087 cm^-1 and ⁵ν_CO = 2063 cm^-1. The lower energy shift of Δ⁵ν_CO = 14 cm^-1 for ^ImLCu^I-CO isomer in comparison to ^PyLCu^I-CO indicates that the imidazole in ^ImL is a stronger donor for copper(I) than the pyridine in ^PyL. However, the hard aliphatic amine group within ^NMe₂LCu^I-CO does not ligate at all in THF solution, as indicated by the single CO stretch at 2090 cm^-1.

Table 1.

Carbonyl Stretching Frequencies for ^DLCu^I-CO Complexes and Cyclic Voltammetry Data for ^DLCu^I Complexes.

Compound	Neat solid	THF solution	E1/2b	ΔE b
ImLCuI-CO	2097, 2035(sm)	2087, 2063	−445	110
NMe2LCuI-CO	2097	2090	−445	85
PyLCuI-CO	2077	2091, 2074(sh)	−410	75
QLCuI-CO	2090	2091	−325	110
TBPLCuI-CO	2091	2092	−325	115
BzLCuI-CO	2093	2093	−225	155

^aAbbreviations are as follows: sm = small, sh = shoulder.

^bElectrochemical measurements are of copper(I) complexes (not carbonyl) in deoxygenated CH₃CN. Potentials are reported versus the Fc⁺/Fc redox couple and are rounded to the nearest 5 mV.

Consistent with their X-ray crystal structures, ^TBPLCu^I-CO and ^BzLCu^I-CO are four-coordinate structures in THF solution with ⁴ν_CO values of 2091 cm^-1 and 2093 cm^-1 respectively, see Table 1. THF solution data for ^QLCu^I-CO reveals the same N₃Cu^I-CO coordination based on the ⁴ν_CO value. Overall, the CO stretching frequencies for all four-coordinate ^DLCu^I-CO solution complexes are approximately equal, ⁴ν_CO(avg) = 2091 cm^-1, indicating that the dangling arm throughout the series must be the D-donor moiety, as depicted in Chart 1.

Solid-State IR Spectroscopic and Structural Properties

The ⁴ν_CO values of the isolated ^ImLCu^I-CO and ^NMe2LCu^I-CO species were both centered at 2097 cm^-1 in Nujol mull spectra suggesting that the imidazole or aliphatic dimethylamine donor moieties do not coordinate in the static solid-state structure. Since these values are higher than the solution state ⁴ν_CO values, the ^DL ligation may be somewhat more two-coordinate in nature, i.e., perhaps the aliphatic amine nitrogen is less strongly bound than for the other cases (Chart 2). In the solid state, ^PyLCu^I-CO has an overall five-coordinate structure.⁵⁹

Chart 2.

For ^ImLCu^I-CO, an additional very low energy and intensity CO stretch at 2035 cm^-1 is observed in the solid-state. Such a low ν_CO value suggests a coordination environment with anionic ligand donors, or three strong non-chelating N donors, as is known elsewhere.^3,78 In previous work, the electron-donating ability of the four-coordinate R- ^PyL ligand series was increased by introduction of para-substituents such as R = N(CH₃)₂- and CH₃O- (Chart 2)⁵⁹. Here, ν_CO values shifted to lower energy by 14 to 28 cm^-1, e.g., ν_CO = 2049 cm^-1 for [Cu^I(NMe₂- ^PyL)(CO)]⁺ in THF. To explain the very low 2035 cm^-1 ν_CO value, we suggest that the dangling imidazolyl moiety of ^ImLCu^I-CO may coordinate to a different ^ImLCu^I complex, forming a dimeric structure, see Chart 2 below. Pyridyl π–π stacking may facilitate formation of such a structure, as has been observed elsewhere for copper(I) structures.^83-85

Electrochemical Studies

Copper(II/I) redox chemistry was examined through cyclic voltammetry measurements carried out on the ligand-copper(I)-complexes (^DLCu^I). Quasi-reversible single electron-transfer processes (i_pc/i_pa ≈ 1) with peak-to-peak separations between ΔE = 75 – 155 mV were observed (Table 1). The Cu^II/Cu^I redox potentials, i.e. electron transfer, are largely affected by structural alterations prompted by changes in oxidation states as well as steric hindrance induced by ligand variations. Overall, the trend of Cu^II/I half-wave potentials (E_1/2) were similar to the trend of the ⁴ν_CO values of ^DLCu^I-CO species in THF, see Table 1. Ligand-copper-complexes with more negative E_1/2 values favor a higher oxidation state (Cu^II) and suggest a more donating ligand system. For example, the formally tridentate ^BzLCu^I, has the highest redox potential of −225 mV, hence favoring a lower oxidation state.

The same E_1/2 values were measured for ^QLCu^I and ^TBPLCu^I at −325 mV and their peak-to-peak separation was also about the same, ΔE_avg = 115 mV. These results suggest that the steric influence of the quinoline donor and 6-tbp-subsituent are very similar. The same E_1/2 values were also measured for ^ImLCu^I and ^NMe₂LCu^I at −445 mV, which was surprising because of their different static copper(II) structures. The X-ray crystal structure of [Cu^II(^NMe₂L)(Cl)]²⁺ reveals a distorted square pyramidal coordination geometry, τ = 0.26.^62,86 The τ structural parameter is 0.00 for a perfect square pyramidal structure and l.00 for a perfect trigonal bipyramidal coordination geometry.^86,87 By contrast, [Cu^II(^IML)(CH₃CN)]²⁺ displays a trigonal bipyramidal static structure, τ = 0.86.^65,86

Since [Cu^II(^PyL)(CH₃CN)]²⁺ has an almost perfect trigonal bipyramidal static structure, τ = 0.96,^67,86 the same degree of structural rearrangement would be expected to occur upon oxidation of ^ImLCu^Iand ^PyLCu^I. Both of their cuprous-acetonitrile structures are assumed to be very similar based on their related ^DLCu^I-CO ⁴ν_CO values. However, the E_1/2 values are quite different for ^ImLCu^I (107 mV) and ^PyLCu^I (76 mV) suggesting that the in situ copper(II) structures are different. The E_1/2 value of ^ImLCu^I (and ^NMe₂LCu^I) in comparison to ^PyLCu^I (−410 mV) is 35 mV more negative, therefore the latter is easier to reduce.

CO Equilibrium Binding and Thermodynamic Parameters (K_CO: ΔH°, ΔS°)

The equilibrium CO binding constants (K_CO, Scheme 1) and corresponding thermodynamic parameters (ΔH°; ΔS°) were determined by variable temperature monitoring of THF solvent reactions of CO with ^DLCu^I to form ^DLCu^I-CO.⁵⁹ Throughout the entire series (Table 2), K_CO increases are accompanied by decreasing (more favorable) enthalpies but less favorable (decreasing) reaction entropies. This behavior is typical, as enhanced binding can be described as being “tighter”. UV-visible spectral data representative of the conversion of ^TBPLCu^I to ^TBPLCu^I-CO are shown in Figure 4; data corresponding to all other species are found in the Supporting Information.

Scheme 1.

Table 2.

Kinetic and Thermodynamic Parameters for CO binding to ^DLCu^I in THF; Constants Reported at 298 K.^a

Compound	KCO (M-1)	ΔH°	ΔS°	kCO (M-1 s-1)	ΔH‡	ΔS‡	k-CO (s-1)	ΔH‡	ΔS‡
ImLCuI-CO	(2.4 ± 0.04) × 103	−31.4	−40.6	(2.8 ± 0.07) × 109	7.03	−40.5	(1.1 ± 0.04) × 106	38.5	0.1
NMe2LCuI-CO	(5.0 ± 0.05) × 103	−39.7	−62.5	(2.5 ± 0.05) × 109	9.07	−34.7	(5.0 ± 0.09) × 105	48.8	28
PyLCuI-CO	(1.2 ± 0.05) × 104	−35.9	−42.6	(1.9 ± 0.02) × 109	7.13	−43.3	(1.5 ± 0.04) × 105	43.0	−0.7
QLCuI-CO	(2.2 ± 0.2) × 104	−39.2	−48.5	(9.7 ± 0.1) × 108	5.66	−53.9	(4.4 ± 0.01) × 104	44.9	−5.3
TBPLCuI-CO	(1.4 ± 0.03) × 104	−40.4	−56.2	(5.2 ± 0.01) × 108	5.76	−58.7	(3.7 ± 0.01) × 104	46.2	−2.5
BzLCuI-CO	(5.6 ± 0.1) × 104	−46.0	−63.6	(5.0 ± 0.01) × 108	0.74	−75.9	(9.0 ± 0.03) × 103	46.8	−76

^aUnits of kinetic and thermodynamic values are as follows: K_CO, M^-1; k_CO, M^-1 s^-1; k_–CO, s^-1; ΔH, kJ mol^-1; ΔS, J mol^-1 K^-1. See Supporting Information for Van't Hoff and Eyring plots.

Figure 4.

Spectrophotometric titration of CO to ^TBPLCu^I in THF at room temperature. The inset is a Van't Hoff plot of the variable temperature K_CO data collected from −30 °C to +20 °C.

CO Photodissociation and Rebinding Kinetics

The kinetics and thermodynamics of CO binding to ^DLCu^I were measured through variable temperature (20 to −80°C) transient absorption (TA) laser flash photolysis experiments.^15,59 Absorption difference spectra, Abs{[^DLCu^I] – [^DLCu^I-CO]}, calculated from steady-state absorption experiments were in agreement with observed transient data collected within the range λ_mon = 325 – 700 nm; see Figure 5 for ΔA spectra corresponding to photodissociation of CO from ^TBPLCu^I-CO. The subsequent rebinding of CO followed a first-order kinetic model. Bimolecular rate constants (k_CO) for ^DLCu^I-CO formation were calculated based on [CO] dependence studies.⁵⁹ Complementary activation (ΔH^‡; ΔS^‡) parameters were calculated by determination of k_CO at variable temperatures and through Eyring analysis.⁵⁹

Figure 5.

Transient absorption difference spectra observed after pulsed 355 nm excitation of ^TBPLCu^I-CO in THF at room-temperature under 1 atm of CO. CO photodissociation and the formation of ^TBPLCu^I followed by subsequent CO recombination is measured and shown at various delay times: 0 ns (black); 100 ns (red); 250 ns (blue); 500 ns (green); 1000 ns (purple). The inset is an Eyring analysis plot of the k_CO data collected at variable temperature (−70 °C to 10 °C).

In all cases, CO (re)binding to ^DLCu^I follows an associative mechanism as suggested by the k_CO associated negative reaction entropies (ΔS^‡) in the range −40 to −76 J mol^-1 K^-1, see Table 2. Furthermore, the nearly identical thermodynamic (ΔS°) and activation (ΔS^‡) entropies along the ^DL series of complexes suggest a late transition state; the rate determining step involves an intermediate that is structurally similar to the final carbonylated product. Following CO photodissociation, the dangling N-donor arm of ^DLCu^I-CO binds to the open coordination site of the cuprous ion and must dissociate before CO can rebind.

Thermal ^DLCu^I-CO Dissociation Kinetics

With K_CO and k_CO values determined experimentally, the thermal CO dissociation rate constants (k_−CO) and corresponding activation parameters (ΔH^‡, ΔS^‡) were calculated.⁵⁹ As expected, the k_−CO value increases with decreasing reaction enthalpies (ΔH^‡) and increasing entropic (ΔS^‡) values. The transition state largely correlates with the overall geometry change as related to the degree to which the dangling arm is or is not interacting with the cuprous-carbonyl upon Cu–C bond breakage. See below for further discussion.

Kinetics (k_O2; ΔH^‡, ΔS^‡) in THF of 1:1 Dioxygen Binding to ^DLCu^I following CO Photodissociation from ^DLCu^I-CO; D = Py, Im, NMe₂

The primary interaction (1:1) of O₂ with ^PyLCu^I, ^ImLCu^I, and ^NMe₂LCu^I to form ^DLCu^II-O₂⁻ was monitored and measured in THF solvent through the “flash-and-trap” method.¹⁵ The experiments were conducted in the presence of precisely determined concentrations of O₂ and CO within the temperature range of −55 °C to −85 °C on nanosecond and longer time scales. The ΔA spectra, Abs{[^NMe₂LCu^I/II-X] − [^NMe₂LCu^I-CO]} X = THF or O₂⁻, representing the two separate O₂-binding processes (k_fast, k_slow) for formation of ^NMe₂LCu^II-O₂⁻ are shown in Figures 6A-B; corresponding data for formation of ^ImLCu^II-O₂⁻ is given in the Supporting Information. As depicted in Scheme 6, the initial “fast” process (k_fast; Figure 6A) involves the competitive binding of both CO and O₂ with ^DLCu^I, either regenerating ^DLCu^I-CO or forming ^DLCu^II-O₂⁻, thus, k_fast is a combination of two rate constants, k_O2 and k_CO. Following generation of ^DLCu^II-O₂⁻, since K_CO ≫ K_O2, CO subsequently displaces coordinated O₂ to reform the initial ^DLCu^I-CO species (k_slow). The latter values, k_O2 (and K_O2 and k_−O2) derived from k_slow, are tabulated here and used for discussion.

Figure 6.

Absorption difference spectra [^NMe₂LCu^I/II-X − ^NMe₂LCu^I-CO] recorded after pulsed λ_ex =355 nm excitation of ^NMe₂LCu^I-CO in THF at 193 K under 1 atm O₂:CO (1:1) mixture. The spectra were recorded at various delay times: (A) 0 to 5 μs representing the conversion of ^NMe₂LCu^I to a mixture of ^NMe₂LCu^I-CO and ^NMe₂LCu^II-O₂⁻: black squares, 0 μs; red circles, 0.1 μs; green triangles, 0.25 μs; blue diamonds, 1 μs; magenta stars, 2.5 μs; orange circles, 5.0 μs. The inset is an absorption transient monitored at 425 nm with a superimposed first-order fit (in red), k_obs = 2.92 × 10⁵ s^-1 and (B) 0 ms to 20 ms representing the conversion from ^NMe₂LCu^II-O₂⁻ to ^NMe₂LCu^I-CO: black squares, 0 ms; red circles, 2.5 ms; green triangles, 5.0 ms; blue diamonds, 10 ms; orange circles, 20 ms. The inset is an absorption transient monitored at 425 nm with a superimposed first-order fit (in red), k_obs = 236 s^-1.

Scheme 6.

A full analysis of the O₂ binding kinetics and thermodynamics for formation of ^PyLCu^II-O₂⁻ was reported previously in THF solvent via the “flash-and-trap” method,¹⁵ and in EtCN through stopped-flow UV-visible spectroscopy.⁸ In the present work, analogous data for formation of ^ImLCu^II-O₂⁻ and ^NMe₂LCu^II-O₂⁻ in THF are discussed and compared to that for ^PyLCu^II-O₂⁻. Room temperature k_O2 values were obtained by extrapolation from the activation parameters obtained at low temperatures where experimental data could be collected. All data here are also compared to those obtained by Schindler, Zuberbühler and coworkers on the closely related tripodal tetradentate ligand complex [Cu^I(Me₆tren)]⁺ {tris(2-dimethylaminoethyl)amine},⁶² see Table 3. In fact, very fast reactions with k_O2 > 10⁷ M^-1s^-1 at 298 K have only been observed and determined for the ^DLCu^I complexes described here, which includes the previously studied ^PyLCu^I, and [Cu^I(Me₆tren)]⁺.^13-15

Table 3.

Comparison of the kinetics and thermodynamics for formation of ^DLCu^II-O₂⁻ (D = Py, Im, NMe₂) in THF and other ligand-copper(II)-η¹-superoxo species.

Parameter	PyLCuI THF(a)	PyLCuI EtCN(b)	[CuI(Me6tren)]+ EtCN(b)	ImLCuI THF(c)	NMe2LCuI THF(c)
kO2, 193 K (M-1 s-1)	(1.5 ± 0.02) × 108	(3.8 ± 0.01) × 104	(1.8 ± 0.04) × 105	(1.8 ± 0.03) × 108	(6.9 ± 0.02) × 107
kO2, 298 K (M-1 s-1)	(1.3 ± 0.02) × 109	(5.8 ± 0.8) × 107	(1.2 ± 0.1) × 107	(3.4 ± 0.6) × 1010	(2.3 ± 0.4) × 1011
ΔH‡ (kJ mol-1)	7.62	31.6	17.1	23.4	32.1
ΔS‡ (J mol-1 K-1)	− 45.1	10	− 52	35.1	80.1
k-O2, 193 K (s-1)	240 ± 6	130 ± 1	0.62 ± 0.01	1600 ± 0.05	470 ± 0.02
k-O2, 298 K (s-1)	(1.3 ± 0.03) × 108	(1.5 ± 0.2) × 108	(7.7 ± 0.9) × 105	(1.9 ± 0.2) × 109	(2.3 ± 0.3) × 109
ΔH‡ (kJ mol-1)	58.0	61.5	62.0	64.5	66.5
ΔS‡ (J mol-1 K-1)	105	118	76	150	157
KO2, 193 K (M-1)	(6.5 ± 0.02) × 105	260 ± 4	(2.9 ± 0.04) × 105	(1.1 ± 0.03) × 105	(1.5 ± 0.06) × 105
KO2, 298 K (M-1)	15.4 ± 0.3	0.38 ± 0.02	15.5 ± 0.5	17.4 ± 0.8	100 ± 0.7
ΔH° (kJ mol-1)	− 48.5	− 29.8	− 44.9	− 41.1	− 34.4
ΔS° (J mol-1 K-1)	− 140	− 108	− 128	− 114	− 77.3

O₂ binding data was previously published and collected through

^(a)flash-and-trap techniques in THF,¹⁵

^(b)stopped-flow spectroscopy in EtCN.¹⁴,⁸⁵

^(c)Averaged values were determined from analysis of k_slow using the flash-and-trap methods discussed herein, see text for more details. Bimolecular k_O2 values calculated at additional temperatures are given in the Supporting Information, Table S2.

Discussion

CO Equilibrium Binding and Thermodynamic Parameters (K_CO: ΔH°, ΔS°)

The K_CO values for formation of ^DLCu^I-CO increase as the E_1/2 values become more positive and the ⁴ν_CO values shift to higher energy (Table 1 and Table 2). In previous work that examined the CO binding properties of copper(I)-carbonyl adducts of the R-^PyL ligand series where only electronic effects are present, the opposite trend was observed; thus, a ligand having better donor properties led to enhanced CO binding.⁵⁹ This suggests that steric constraints or structural differences induced by changing the dangling “D” N-donor moiety of ^DL significantly affect the K_CO values in the present systems.

We suggest that the differences in ΔS° values or degree of disorder in the reaction are affiliated in part with the extent that the dangling “D” N-donor moiety is interacting with the final ^DLCu^I-CO adduct. For example, the Cu^I-CO adduct of the purely tridentate ^BzL has the highest K_CO value of 5.6 × 10⁴ M^-1. This clearly derives from ^BzLCu^I-CO having the most favorable reaction enthalpy in the series, ΔH° = −46.0 kJ mol^-1 K^-1, since the reaction is the most unfavored entropically, ΔS° = −63.6 kJ mol^-1 K^-1. In more depth, formation of ^BzLCu^I-CO is the most ordered reaction, meaning the benzyl D-group does not coordinate or donate any significant electron density to the cuprous center and also does not hinder coordination of CO to ^BzLCu^I.

The K_CO values for CO binding to ^QLCu^I and ^TBPLCu^I are the highest of the tetradentate ligands studied, 2.2 × 10⁴ M^-1 and 1.4 × 10⁴ M^-1 respectively. Based on their solution and solid state ⁴ν_CO values, the ligand-copper(I)-carbonyl complexes of both are tridentate in nature, consistent with the X-ray crystal structure of ^TBPLCu^I-CO (vide supra). However, in comparison to ^BzLCu^I-CO, the coordination of CO to the cuprous ion of ^QLCu^I and ^TBPLCu^I must be somewhat hindered by their bulky “D” N-donor groups (i.e., quinolyl or ^tBu-pyridyl arms), resulting in their slightly lower K_CO values (Table 2).

At the low end of the range, reaction of ^ImLCu^I with CO results in the smallest K_CO value of 2.4 × 10³ M^-1 with the most favorable entropy, ΔS° = −40.6 kJ mol^-1 K^-1. We suggest that the dangling imidazole is more interacting with the cuprous-carbonyl center resulting in increased disorder. The bond enthalpy for formation of ^ImLCu^I-CO is the highest (ΔH° = −31.4 kJ mol^-1), consistent with the weakest Cu–C bond of the series.

In comparison to the reaction of CO with ^PyLCu^I, the K_CO value of 1.2 × 10⁴ M^-1 is approximately five times higher as a result of the slightly lower thermodynamic values, ΔH° = −35.9 kJ mol^-1, ΔS° = −42.6 kJ mol^-1 K^-1.⁵⁹ Such small differences in the thermodynamic parameters, yet large effect on the equilibrium CO binding constant may suggest that the dangling pyridyl arm of ^PyL is less interacting with the cuprous ion than is the imidazolyl moiety of ^ImL.

By contrast, the equilibrium binding constant of K_CO = 5.0 × 10³ M^-1 for reaction of CO with ^NMe₂LCu^I is lower. However, the highly negative ΔS° value of −62.5 kJ mol^-1 K^-1 is similar to that of ^BzLCu^I-CO suggesting that the aliphatic dimethylamino arm is barely interacting if at all with the cuprous ion of ^NMe₂LCu^I in situ. Such a characteristic is consistent with the absence of an observable ⁵ν_CO for ^NMe₂LCu^I-CO.

CO Photodissociation and Rebinding Kinetics

The rate constant for formation of ^DLCu^I-CO (k_CO) appears to be controlled by the degree to which the dangling “D” arm interacts with the cuprous ion during Cu–C bond formation. This conclusion comes about from the observation that decreasing ⁴ν_CO frequencies for ^DLCu^I-CO along with decreased complex E_1/2 values for ^DLCu^I, which both indicate a better donor ligand, correlate with an increase in k_CO values (Table 1 and Table 2). For example, ^ImLCu^I-CO is comprised of the most electron-donating N-donor system for copper(I) due to the presence of the imidazolyl donor moiety and this complex gives the highest rate constant, k_CO = 2.8 × 10⁹ M^-1 s^-1. The corresponding activation parameters for formation of ^ImLCu^I-CO, ΔH^‡ = 7.03 kJ mol^-1 and ΔS^‡ = −40.5 kJ mol^-1 K^-1, are very similar to those measured for ^PyLCu^I-CO, suggesting little differences in the CO coordination dynamics. The coordination of CO to ^ImLCu^I is depicted in Scheme 2.

Scheme 2.

Consistent with the near equivalent E_1/2 values determined for ^ImLCu^I and ^NMe₂LCu^I, ^NMe₂LCu^I-CO has a similar CO rebinding rate of 2.5 × 10⁹ M^-1 s^-1. However, the ΔS^‡ value affiliated with CO binding to ^NMe₂LCu^I is much more positive than the ΔS° value suggesting that the course of reaction is more disordered as a result of the dangling or very weakly interacting donor moiety. Unlike for ^ImLCu^I-CO and ^PyLCu^I-CO, a five-coordinate carbonyl species is not structurally favorable for ^NMe₂LCu^I-CO; in support of this supposition, note that only a ⁴ν_CO value was observed, vide supra. As a result, dissociation of the fourth ligand N-donor arm of ^NMe₂LCu^I followed by subsequent CO rebinding likely involves significant structural changes resulting in a higher ΔS^‡ value.

The parallel thermodynamic and activation parameters associated with CO binding to ^QLCu^I (9.7 × 10⁸ M^-1 s^-1) and ^TBPLCu^I (5.2 × 10⁸ M^-1 s^-1) suggests that the mechanism of reaction is the same for both (Scheme 3). Such similarities are consistent with their analogous ⁴ν_CO values and identical E_1/2 values. Since solvent (THF) does not coordinate to ^QLCu^I and ^TBPLCu^I, only the “D” N-donor must dissociate from the cuprous ion to form their respective ligand-copper(I)-carbonyl adducts.

Scheme 3.

Carbon monoxide rebinding to ^BzLCu^I, where a fourth endogenous N-donor is not present, results in the lowest k_CO value of 5.0 × 10⁸ M^-1 s^-1; note that the Cu^II/Cu^I compound exhibits the most positive E_1/2 value. The negligible enthalpy (ΔH^‡ = 0.74 kJ mol^-1) and highly negative entropy (ΔS^‡ = −75.9 J mol^-1 K^-1) of activation implies that the tridentate nature of ^BzL minimizes geometric changes in the reaction of CO with ^BzLCu^I and that the ‘dangling’ benzyl group does not hamper CO rebinding to the cuprous center. However, since the standard entropy (ΔS°) for formation of ^BzLCu^I-CO is greater than the activation entropy (ΔS^‡), it is possible that the CO molecule rebinds to the naked [Cu^I(^BzL)]⁺ species before THF transiently coordinates, resulting in a more ordered reaction and negligible enthalpy of activation (ΔH^‡) as shown in Scheme 3.

Thermal ^DLCu^I-CO Dissociation Kinetics

As shown in Scheme 2 for CO dissociation from ^ImLCu^I-CO, the dangling arm ligates to the copper(I) ion following CO dissociation. The negligible activation entropy (ΔS^‡ ∼ 0 J mol^-1 K^-1) suggests that there is little overall geometry differences between the initial carbonylated complex and the transition state. Therefore, the increased donation offered by the slightly interacting imidazole in ^ImLCu^I-CO results in an overall decrease in the structural changes as well as a decrease in the activation barrier to cleave the Cu–C bond. ^ImLCu^I-CO has a k_−CO value of 1.1 × 10⁶ s^-1 with a corresponding reaction enthalpy of ΔH^‡ = 38.5 kJ mol^-1. Similar characteristics were exhibited by ^PyLCu^I-CO,^{59 Q}LCu^I-CO, and ^TBPLCu^I-CO for CO dissociation, all with near zero ΔS^‡ values (Table 2).

Dissociation of CO from ^NMe₂LCu^I-CO is characterized by ΔS^‡ = 28 J mol^-1 K^-1, a more favorable value than is observed for any other in the series (Table 2). Thus, Cu–C bond cleavage involves considerable structural rearrangement and the high ΔS^‡ value implies an early transition state structurally similar to the terminal copper(I)-carbonyl species. Such a mechanism would require the dimethylamino donor arm of ^NMe₂L to be predominantly in the unbound state when the Cu–C bond is broken and further suggests that coordination of solvent occurs prior to or concomitant with CO release, see Scheme 4. As already stated, it is unfavorable for the hard dimethylamino arm to interact with the soft cuprous-carbonyl center, as evidenced or supported by a lack of observation of ⁵ν_CO for ^NMe₂LCu^I-CO.

Scheme 4.

Conversely, dissociation of CO from ^BzLCu^I-CO is accompanied by a strikingly unique unfavorable activation entropy of ΔS^‡ = −76 J mol^-1 K^-1 that leads to the smallest k_−CO value (9.0 × 10³ s^-1) in the series. The almost identical ΔS^‡ values for the forward and reverse reaction of ^BzLCu^I/CO suggest that virtually no geometry change results and therefore the reaction is highly ordered, i.e. there is a late transition state.

Kinetics (k_O2; ΔH^‡, ΔS^‡) in THF of 1:1 Dioxygen Binding to ^DLCu^I following CO Photodissociation from ^DLCu^I-CO; D = Py, Im, NMe₂

An examination of activation parameters (ΔH^‡, ΔS^‡; Table 3) indicates there is a striking difference between those for formation of ^PyLCu^II-O₂⁻ compared to the other two copper(II) η¹-superoxo complexes, ^ImLCu^II-O₂⁻ and ^NMe₂LCu^II-O₂⁻. The binding of O₂ to ^PyLCu^I occurs with a relatively large and negative activation entropy (−45.1 J mol^-1 K^-1, Table 3) indicating an associative reaction mechanism is involved.¹⁵ This suggests O₂ binds prior to loss of the bound solvent molecule and then THF solvent de-ligates. Or, solvent may not even coordinate to the cuprous center, accounting for a simple (unhindered) association of dioxygen. As an example of the latter point, the tetradentate ligand-copper(I)-complex [Cu^I(Me₆-tren)]⁺ does not coordinate an exogenously derived nitrile ligand.¹⁴ As a result, an associative O₂ binding mechanism (ΔS^‡ = −52 J mol^-1 K^-1) resulted for formation of [Cu^II(Me₆-tren)(O₂⁻)]⁺ in EtCN, see Table 3. By extreme contrast, the positive activation entropies measured for ^NMe₂LCu^II-O₂⁻ (ΔS^‡ = 80.1 J mol^-1 K^-1) and ^ImLCu^II-O₂⁻ (ΔS^‡ = 35.1 J mol^-1 K^-1) indicate that the O₂-binding reaction follows a dissociative mechanism where de-ligation of solvent (THF) occurs prior to O₂ binding. Throughout the remainder of the text, this latter reaction mechanism is referred to as a dissociative process, however, we note that a purely dissociative mechanism would result in first-order kinetics. Therefore, it is perhaps more appropriate to describe this process as a dissociative interchange reaction, in which bond breakage dominates over bond formation. More detailed explanation and analysis of the dioxygen binding mechanisms is given below.

As discussed earlier in the text, the structurally characterized cupric-chloride species [^NMe₂LCu^II(Cl)]⁺ has a square pyramidal geometry (τ = 0.26),⁶² different than the trigonal bipyramidal structures exhibited by [^PyLCu^II(Cl)]⁺ (τ = 1.00),⁸⁸ [^PyLCu^II(CH₃CN)]²⁺ (τ = 0.96),⁶⁷ and [^ImLCu^II(CH₃CN)]²⁺ (τ = 0.86)⁶⁵.⁸⁶ Also, the complementary ligand-copper(I)-CH₃CN structures of the three aforementioned species ^DLCu^I have analogous trigonal bipyramidal geometries. Therefore, based on the assumption that the cupric-superoxo species are analogous in structure to their X-ray crystallographically characterized copper(II) adducts, the structural rearrangement that results upon redox changes in the ^NMe₂L ligand system would be much greater. Consistent with this supposition is the observed highly positive activation entropy ΔS^‡ = 80.1 kJ mol^-1 for the reaction of O₂ with ^NMe₂LCu^I. Also, the overall redox process, i.e. O₂-binding, results in a higher enthalpy of activation of ΔH^‡ = 32.1 kJ mol^-1. The more favorable activation enthalpy for O₂ binding to ^ImLCu^I (ΔH^‡ = 23.4 kJ mol^-1) and less favorable ΔS^‡ (35.1 kJ mol^-1) in comparison to ^NMe₂LCu^I is consistent with a smaller geometry change with copper redox state due to O₂-binding.

Solvent dependence (EtCN vs THF) of 1:1 Copper-Dioxygen Adducts

The dynamics of the O₂ binding reaction of ^PyLCu^I are largely influenced by solvent medium based on the analogous O₂ binding data collected through the “flash-and-trap” method in THF¹⁵ and stopped-flow UV-visible spectroscopy in EtCN,⁸ see Table 3. The entropy of activation (ΔS^‡) for formation of ^PyLCu^II-O₂⁻ in EtCN (ΔS^‡ = 10 J mol^-1 K^-1) versus THF (ΔS^‡ = −45.1 J mol^-1 K^-1) indicates that O₂ binding to ^PyLCu^I follows a dissociative versus associative pathway, respectively, depending on the solvent. To observe such a solvent effect is not an unreasonable expectation since nitrile solvents are strong Lewis bases and soft donors, therefore strong ligands for copper(I) ions. In some cases, nitrile solvents severely inhibit binding of dioxygen such as in the reaction of O₂ with ^BzLCu^I, where EtCN solvent prevents formation of the commonly observed O₂ adduct, characterized as a dicopper(III) bis-μ-oxo species in less coordinating solvents (THF, acetone, toluene).⁶⁰

Previously, Schindler and coworkers attempted to time-resolve the formation of the cupric superoxo species ^NMe₂LCu^II-O₂⁻.⁶² However, the formation was too fast to be measured by stopped-flow UV-visible spectroscopy, even in EtCN (coordinating solvent) at 183 K. Similarly, we have unsuccessfully attempted to measure the formation of ^ImLCu^II-O₂⁻ by stopped-flow methods in EtCN.⁶⁵ Conversely, the k_O2 value for O₂ binding to ^PyLCu^I in EtCN was easily measured (Table 3). These comparisons suggest that like the k_O2 measurements in THF, a much higher k_O2 value in EtCN is expected for formation of ^NMe₂LCu^II-O₂⁻ and ^ImLCu^II-O₂⁻ compared to ^PyLCu^II-O₂⁻. Such measurements for formation of ^NMe₂LCu^II-O₂⁻ and ^ImLCu^II-O₂⁻ in EtCN were not carried out, but the values of k_O2 reported here in THF through the present “flash-and-trap” experiment, indeed are very large (> 6.9 × 10⁷ M⁻¹ s⁻¹ at 193 K; see Table 3), above the limit of stopped-flow UV-visible spectroscopic methods (ms timescale).

THF is usually considered a weakly coordinating solvent for ligand-copper-species and has even been described as a non-coordinating solvent. However, THF contains an electron pair centered on the oxygen-atom potentially capable of binding to copper(I) or any Lewis acid. A series of ligand-copper(II)-peroxo complexes utilizing the tridentate ligand R-MePY2 {bis[2-(2-pyridyl)ethyl]methylamine; R = Cl-, H-, MeO-, NMe₂-} with 4-pyridyl substituents were previously reported to efficiently oxidize THF to 2-hydroxy tetrahydrofuran (THF-OH).⁸⁵ Substrate (THF) oxidation was proposed as an “inner-sphere oxidation” because of a likely pre-equilibrium step involving THF-binding to the copper-dioxygen adduct.⁸⁹ Also, in a detailed study conducted by Tolman and coworkers on the mechanism of formation for the β-diketiminate Cu^III-η²-O₂²⁻ adduct, an associative oxygenation pathway (Scheme 7) involving THF coordination (pathway B) was proposed based on comparisons to the same reaction in THF/MeCN mixtures (pathway A).¹³ Complementary theoretical calculations led to a proposed transition state that involved dioxygen and solvent bound to copper simultaneously, either THF or MeCN accordingly.

Scheme 7.

Proposed Mechanisms for O₂-Binding to the ^DLCu^I Complexes (D = NMe₂, Im, Py)

In the present work and as described above, positive ΔS^‡ values were measured for O₂ binding to ^ImLCu^I (ΔS^‡ = 35.1 J mol^-1 K^-1) and ^NMe₂LCu^I (ΔS^‡ = 80.1 J mol^-1 K^-1) in THF solvent, to the first approximation similar to that found for ^PyLCu^I in EtCN, but very different to that for ^PyLCu^I in THF (and [Cu^I(Me₆tren)]⁺ in EtCN), see Table 3. Such a drastic difference in the O₂-binding mechanism depending on the variable N-donor moiety of ^DL is surprising. We suggest changes in the electron distribution and/or coordination variations within the species upon O₂ ligation to Cu^I are responsible and it is not simply related to the order in which O₂ or solvent coordinate. More specifically, the transition state differences may be largely based on whether or not Cu^I → O₂ electron transfer (et) occurs before or after solvent dissociation with the coordination of O₂ always being the initial step.

As shown in Scheme 8, the negative k_O2 affiliated ΔS^‡ value (associative mechanism) for ^PyLCu^II-O₂⁻ in THF suggests a late transition state in which et occurs before or concomitant with solvent release. Conversely, the positive k_O2 affiliated ΔS^‡ value (dissociative mechanism) for O₂ binding to ^ImLCu^I in THF, ^NMe₂LCu^I in THF, and ^PyLCu^I in EtCN, suggests an early transition state in which et can occur only after solvent is released, see Scheme 8. Note that a computational analysis of the proposed mechanism has not been conducted and therefore the proposed internal electron density distributions or solvent positioning (ligated or not) cannot be conclusively supported. However, the similar dissociative processes for formation of the square pyramidal ^NMe₂LCu^II-O₂⁻ species in THF and the trigonal bipyramidal ^ImLCu^II-O₂⁻ and ^PyLCu^II-O₂⁻ species in THF and EtCN (respectively) again points out that the reaction mechanism is not solely dependent on geometry or solvent medium, but rather the degree of solvent and/or O₂-fragment interaction, i.e. electronic distribution within a given species.

Scheme 8.

Smirnov and Roth have proposed a two-step, inner-sphere electron-transfer mechanism for the oxidation of O₂^•− to O₂ by the copper(II)-complexes of ^PyL and TEPA (tris(2-pyridylethyl)amine) in DMF/THF mixtures (DMF = dimethylformamide).⁹⁰ Based on low-temperature stopped-flow kinetic experiments and competitive ¹⁸O₂ isotope effects, a pre-equilibrium step with formation of a Cu-O₂ species was suggested to form prior to O₂ release, see Scheme 9. Formal oxidation states were not emphasized (Cu^II-O₂⁻ versus Cu^I-O₂) for the intermediate species; therefore an electrostatic versus covalent Cu–O₂ interaction was not designated. However, the release of O₂ was designated as the rate determining step which would complete the overall et process, i.e. reduction of the copper(II) ion. Formation of a pre-equilibrium complex is analogous to our proposal of a pre-electron transfer intermediate step upon formation of ^DLCu^II-O₂⁻ (Scheme 9).

Scheme 9.

Temperature dependence (193 K vs. 298 K) of equilibrium O₂ binding constants (K_O2; ΔH°, ΔS°) and dissociation rates (k_−O2; ΔH^‡, ΔS^‡)

Dissociation of O₂ from ^DLCu^II-O₂⁻ has an early transition state (+ΔS^‡) for all species studied that involves the transfer of an electron (et) from O₂^•− to copper(II) followed by coordination of solvent (THF) and concomitant release of O₂. This Cu-O₂ bond breakage process is depicted for ^ImLCu^II-O₂⁻ in Scheme 10. The k_−O2 activation parameters corresponding to ^ImLCu^II-O₂⁻ are almost the same as those for ^NMe₂LCu^II-O₂⁻ suggesting the same O₂ release mechanism. However, for O₂ dissociation from ^PyLCu^II-O₂⁻, the corresponding activation parameters (Table 3) are of slightly higher values than the aforementioned species suggesting a larger geometric change upon dioxygen loss.

Scheme 10.

The O₂ binding constants (K_O2) for all ^DLCu^II-O₂⁻ complexes are of approximately the same magnitude (10⁵ M^-1) at 193 K, overall favoring formation of the cupric-superoxo species. As expected, the equilibrium shifts more towards formation of the solvento species at 298 K. The K_O2 values clearly show the solvent effect where as compared to THF, EtCN significantly inhibits O₂ binding to ^PyLCu^I. However, Me₆TREN is a more Lewis basic ligand than is ^PyL such that O₂ binding to [Cu^I(Me₆tren)]⁺ seems unaffected.

Summary of 1:1 Cu/CO and Cu/O₂ binding to ^DLCu^I and Related Cu/O₂ = 2:1 Complexes

In the present study, CO, as a redox-inactive O₂-surrogate, was examined in its interactions with ligand-copper(I)-complexes (^DLCu^I). Data obtained concerning the CO binding kinetics and thermodynamics in THF were compared to those for 1:1 O₂ binding to ^PyLCu^I, ^ImLCu^I, and ^NMe₂LCu^I, where end-on (η¹) coordination occurs. The formation of 1:1 Cu/O₂ adducts were not observed for ^QLCu^I, ^TBPLCu^I, and ^BzLCu^I under the same solution reaction conditions by laser flash photolysis of the CO adducts.⁹¹ Since these cuprous complexes in fact do react with dioxygen to form 2:1 Cu/O₂ species under steady state conditions,^60,63,64 it is likely that primary 1:1 Cu/O₂ binding does not compete kinetically (e.g., Δ_absorbance is too small) under the conditions employed, i.e., those used for all of the complexes examined in these studies.

In our previous reports,^15,59 the CO and O₂ binding kinetics of ^PyLCu^I in THF solvent were detailed, with one finding being that k_CO ≈ k_O2 (Table 2). A look at the highlighted portion of Table 4, with data added in from the present study on ^ImLCu^I and ^NMe₂LCu^I (i.e., with one altered N-donor arm), gives the impression that these latter complexes have a very similar behavior. However, a significant finding in this report (Tables 3 and 4) is that there is a major change in the mechanism of O₂-binding to ^ImLCu^I and ^NMe₂LCu^I in THF; the ligand binding occurs via dissociative rather than associative processes. Recall that the dissociative mechanism does occur for both CO and O₂ binding to ^PyLCu^I in the strong competing solvent EtCN. Electron-transfer (et) from copper(I) to O₂ seems the only factor possible to explain a change in mechanism. The CO binding behavior to the reduced copper(I) species for the ^DL series is essentially the same (Tables 1, 2). The (electrochemical study) derived E_1/2 values for ^ImLCu^I and ^NMe₂LCu^I are the same, but more negative than that for ^PyLCu^I, however the coordination geometries of the copper(II) complexes of ^ImL (∼ ^PyL) and ^NMe₂L are different (vide supra). Thus, we conclude that the et properties/kinetics of the O₂-interaction somehow leads to the change in mechanism observed. Solvent release is initiated by O₂-interaction with copper(I) (i.e., we suggest a Cu^I-O₂ transient forms) prior to et (Scheme 8) giving the observed ^DLCu^II-O₂⁻ product. To summarize in other words, the reaction mechanism is governed by the order that et from copper(I) to O₂ occurs; for ^PyL, et occurs before solvent is released, perhaps outer-sphere, but not so for ^ImL or ^NMe₂L. For ^ImL and ^NMe₂L, a dissociative interchange mechanism (vide supra) may be an appropriate description in which bond breaking (Cu^I-solv → Cu^I + solv) dominates over bond formation and et (Cu^I + O₂ → Cu^II-O₂⁻).

Table 4.

Comparison of CO and O₂ Bimolecular Rates and Binding Constants at 298 K for Select Ligand-Copper(I)-Complexes, Hemocyanins (Hc), and Selected Hemes.

Compound or Protein	KCO (M-1)	KO2 (M-1)	kCO (M-1 s-1)	kO2 (M-1 s-1)	k-CO (s-1)	k-O2 (s-1)
NMe2LCuI (THF)	4.9 × 103	100	2.5 × 109	2.3 × 1011	5.0 × 105	2.3 × 109
ImLCuI (THF)	2.4 × 103	17.4	2.8 × 109	3.4 × 1010	1.1 × 106	1.9 × 109
PyLCuI (THF)	1.2 × 104	15.4	1.9 × 109	1.3 × 109	1.6 × 105	1.3 × 108
PyLCuI (CH3CN)	220	0.38	5.9 × 107	5.8 × 107	2.7 × 105	1.5 × 108
Myoglobin (human)	2.6 × 107	(0.74–117) × 104	7.6 × 105	(1.4–25) × 107	0.022	22
Hemoglobin (human)	4.6 × 108	(2.9–48) × 105	4.6 × 106	(2.9 –22) × 107	0.009	13.1
Limulus Hc (arthropod)	(2.7 – 11.2) × 103	(2.6 – 5.4) × 105	(2 – 4.3) × 105	(1.3 – 1.9) × 106	38 – 75	2.4 – 7.5
Busycon Hc (mollusk)	220 × 103	1.8 × 105	7.7 × 105	(1.1 – 2.2) × 106	3 – 4	6.5 – 11.5

For all ligand-copper-complexes studied, K_CO is much larger than K_O2 primarily as a result of the lower CO dissociation rate constant (k_−CO < k_−O2). Breakage of the Cu-O₂ bond is more favorable than Cu-CO bond breakage even though the former involves an et step. The high k_−O2 values are a result of the highly favorable ΔS^‡ value attributed to a late transition state (which leads to O₂-release) and the structural rearrangement that occurs upon redox changes.

As emphasized, the copper(II)-η¹-superoxo species generated from ^PyLCu^I, ^ImLCu^I, and ^NMe₂LCu^I, are spectroscopically very similar. In fact, spectroscopically (and therefore structurally) similar Cu/O₂ = 2:1 adducts also form, dicopper(II)-μ-1,2-(end-on)-peroxo species [{(^DL)Cu^II}₂(O₂²⁻)]²⁺ analogous to the well characterized case for ^PyL (Scheme 11).^8,61,62,65 However, the conditions (i.e. temperature) under which the [{(^DL)Cu^II}₂(O₂²⁻)]²⁺ complexes are observed, and/or their relative stability, varies significantly. Thus, reaction dynamics, i.e., kinetics and thermodynamics, governing each reaction step likely vary significantly, as observed in the present study on the Cu/O₂ = 1:1 adducts. For example, the peroxo species of ^ImL (Figure 7A) can only be observed at −128 °C in 2-methyl tetrahydrofuran (MeTHF).⁶⁵ In THF at −80 °C, a Cu-O₂ intermediate is not observed.⁶⁵ Quite differently, formation of ^PyLCu^II-O₂⁻ can be detected on the bench-top in MeTHF at −128 °C before complete conversion to the peroxo species (Figure 7B), which is stable for weeks at −80°C in THF.⁶¹ Interestingly, for ^NMe₂LCu^I + O₂, a new as yet uncharacterized species is detected in MeTHF at −128 °C in addition to the peroxo adduct (Figure 7C).⁷⁶ The new UV-vis feature(s) may suggest a different type or additional intermediate forms on the pathway to produce [{(^NMe₂L)Cu^II}₂(O₂²⁻)]²⁺; further studies are required.

Scheme 11.

Figure 7.

UV-visible spectra in MeTHF at −128 °C representing O₂ + ^DLCu^I (dark blue) to yield: A, [{(^ImL)Cu^II}₂(O₂²⁻)]²⁺ (purple); B, ^PyLCu^II-O₂⁻ (green) and [{(^PyL)Cu^II}₂(O₂²⁻)]²⁺ (purple); C, [{(^NMe₂L)Cu^II}₂(O₂²⁻)]²⁺ (purple) and a new copper-dioxygen adduct of ^NMe₂L (brown).

Comparison to natural copper and heme systems

The kinetics and thermodynamics of CO and O₂ binding to the ligand-copper(I)-complexes (^DLCu^I) reported here are in line with studies carried out on natural and synthetic heme systems (Table 4). As said, carbon monoxide coordinates to the cuprous center (^DLCu^I) with a higher affinity than that of O₂ (K_CO > K_O2), allowing for extension of the “flash-and-trap” experiment as inspired by extensive studies on a variety of heme systems. Furthermore, like our synthetic species, K_CO is much higher than K_O2 for the porphyrinate systems as a result of the lower CO dissociation rate (k_−CO < k_−O2); Cu-O₂ bond breakage is more favorable than breakage of the Cu-CO bond. The bimolecular rate constants for the copper complexes (k_CO and k_O2) are much higher than typically exhibited by heme systems and the equilibrium binding constants are much lower. Therefore, the trends observed here are in line with conclusions drawn concerning gaseous small molecule binding at the active site of the heme-copper heterobimetallic enzyme cytochrome c oxidase (CcO) (vide supra; see Introduction). In CcO, where one can imagine some competition for gases by both the heme and copper, CO coordinates to the Cu^I_B site before thermally transferring to the Fe^II-heme_a3 site (Figure 1). We have previously constructed a model binucleating iron and copper containing ligand system as well as 1:1 heme/copper component systems that demonstrate this CO migration phenomenon, and most recently have additionally examined NO transfer from iron-to-copper in the latter 1:1 component system.^22,25

The interaction of CO and O₂ with hemocyanins (Hc) has also been extensively studied and the characteristics are unlike hemes. For Hc, lower equilibrium CO binding constants in comparison to O₂ (K_O2 > K_CO) has been attributed to the higher rate of CO dissociation in comparison to that for O₂ (k_−CO > k_−O2) see Table 4. This behavior most likely occurs because both copper(II) centers are involved in O₂-binding and release, whereas only one copper center binds CO; more structural rearrangements must occur in the release of O₂. Conversely, the high O₂ association rates (k_O2) are a result of the 2:1 Cu/O₂ binding nature of the binuclear copper protein, a characteristic that is not exhibited upon CO binding (1:1 Cu/CO binding), vide supra.

As stated in the Introduction, an experiment such as the flash-and-trap experiment cannot be utilized for the binuclear copper proteins because K_O2 > K_CO. Laser flash photolysis has been used by Hirota et al. to examine O₂ binding by tyrosinase (Tyr) from Streptomyces antibioticus.⁴⁴ Following O₂-photoejection from oxy-Tyr to form deoxy-Tyr, reformation of the starting dicopper(II)-μ-η²:η²-peroxo species was observed while an initial 1:1 Cu/O₂ adduct was not detected. In a more recent study, Hirota, Bubacco, and coworkers reported thermodynamic data for oxygenation of Hc to form the dicopper(II)-(side-on)-peroxo species.⁴⁵ Interestingly, the values were comparable to the superoxo model compounds measured here, suggesting that formation of the 1:1 adduct is the rate-determining step and that oxyHc is easily produced through a nearly simultaneous two et step. This experimental finding fits with Solomon and coworkers' prior computational studies on the O₂-binding process that occurs in Hc.⁹²

Summary and Conclusions

In the present report, we have detailed the kinetics, thermodynamics, and coordination dynamics of 1:1 Cu^I/CO and Cu^I/O₂ binding within tetradentate ligand-copper(I)-complexes with analogous coordination frameworks (Chart 2). The major findings are:

1. In the ligand-copper(I) carbonyl complexes (^DLCu^I), the variable N-donor arm of the synthetic ligand is dangling; all species possess the same four-coordinate core structure in solution where the cuprous ion binds to the bispicolylamine (PY1) tridentate chelate along with the CO molecule.

2. The fast kinetics of CO binding are measurably influenced by the exact nature of that uncoordinated donor arm. With change in ligand where K_CO increases, k_CO and k_−CO decrease; k_−CO decreases more substantially because the dangling donor ligand partially interacts (but to different extents) before it fully coordinates with the Cu(I) ion as CO dissociates.

3. In previous studies where the exact same PY1 ligand framework was present but with more electron releasing 4-pyridyl substituents, K_CO increased, as well as k_CO, very different from what was found here. In other words, ligand electronic variations previously led to the standard expected results, but here, (small) changes in the identity of D (the fourth donor, 3^rd ligand arm, Chart 1) led to a change in the relationship of K_CO to k_CO.

4. Using nanosecond laser flash photolysis to initiate gas (CO, O₂) reactions with copper(I) complexes, the unique finding of an extremely large rate constant for O₂ binding, as previously described for ^PyLCu^I, is shown to be a general phenomenon amongst D-PY1 ligand copper(I) complexes. Further, as seen before, k_O2 and k_CO values are quite similar to each other.

5. Changing one donor arm N-ligand from pyridyl to imidazolyl or to dimethylamino results in a change in the mechanism of O₂-binding to the respective copper(I) complexes, associative for ^PyLCu^I but dissociative for ^ImLCu^I and ^NMe₂LCu^I. We attribute this to aspects of the timing of et from copper(I) to O₂ as this molecule coordinates, in relation to the dissociation of the bound solvent molecule THF. Based on our other experiments, the change in mechanism is not simply related to a change in the ^DLCu^II/I structures or the order that O₂/THF coordinate.

6. The work discussed here establishes transient absorbance laser flash photolysis as an invaluable tool for determining fast small molecule (CO and O₂) binding kinetics and thermodynamics to synthetic copper complexes such as ^DLCu^I. Further, the ability to combine this utility with variable low-temperature control has enabled us to fully characterize 1:1 binding of O₂ to ^ImLCu^I and ^NMe₂LCu^I, which was previously unobtainable with other relatively fast physical methods such as low-temperature stopped flow kinetics/spectroscopy.

Overall the results further support the general finding that subtle changes in coordination environment, as occurs over time through evolution in nature or through controlled ligand design in synthetic systems, can dictate the observed chemistry in terms of reaction kinetics, structure and reactivity, and thus function. Many mononuclear and binuclear copper proteins exist that have an active site ligation of three imidazolyl donors per copper center, yet have very different functions, such as dioxygen transport in hemocyanin (Hc), dioxygen activation and phenol o-hydroxylation in tyrosinase (Tyr), electron-transfer (et) at the Cu_H site in peptidylglycine α-hydroxylating monooxygenase (PHM) and/or dopamine β-monooxygenase, or O₂-binding (and/or et) at Cu_B in cytochrome c oxidases (CcO), see Introduction. Therefore, by gaining a deeper understanding of how the coordination environment and/or subtle changes in the surroundings influence the reactivity and/or binding properties of the copper center may help in understanding protein active site ligand dynamics. Such information may also provide insight into secondary coordination sphere contributions that account for N₃Cu^I active site functional differences.