A Comparative Study of the Catalytic Behaviour of Alkoxy-1,3,5-Triazapentadiene Copper(II) Complexes in Cyclohexane Oxidation

The mononuclear copper complexes [Cu{NH=C(OR)NC(OR)=NH}2] with alkoxy-1,3,5-triazapentadiene ligands that have different substituents (R = Me (1), Et (2), nPr (3), iPr (4), CH2CH2OCH3 (5)) were prepared, characterized (including the single crystal X-ray analysis of 3) and studied as catalysts in the mild oxidation of alkanes with H2O2 as an oxidant, pyridine as a promoting agent and cyclohexane as a main model substrate. The complex 4 showed the highest activity with a yield of products up to 18.5% and turnover frequency (TOF) up to 41 h−1. Cyclohexyl hydroperoxide was the main reaction product in all cases. Selectivity parameters in the oxidation of substituted cyclohexanes and adamantane disclosed a dominant free radical reaction mechanism with hydroxyl radicals as C–H-attacking species. The main overoxidation product was 6-hydroxyhexanoic acid, suggesting the presence of a secondary reaction mechanism of a different type. All complexes undergo gradual alteration of their structures in acetonitrile solutions to produce catalytically-active intermediates, as evidenced by UV/Vis spectroscopy and kinetic studies. Complex 4, having tertiary C–H bonds in its iPr substituents, showed the fastest alteration rate, which can be significantly suppressed by using the CD3CN solvent instead of CH3CN one. The observed process was associated to an autocatalytic oxidation of the alkoxy-1,3,5-triazapentadiene ligand. The deuterated complex 4-d32 was prepared and showed higher stability under the same conditions. The complexes 1 and 4 showed different reactivity in the formation of H218O from 18O2 in acetonitrile solutions.


Introduction
Selective transformation of inactive C-H bonds into functional groups is a challenging class of reactions with great potential for fine organic synthesis, late stage drug functionalization and other fields [1][2][3][4][5]. Due to the high inertness of C-H bonds, especially sp 3 ones, their activation often requires harsh conditions and/or strong oxidizing agents [6]. The stability of coordination compounds (catalysts) under the catalytic conditions is a critical parameter for achieving high efficiency of the catalytic system [7]. In the case of alkane functionalization, C-H-attacking species may be active enough to react with a catalyst causing its degradation and limiting the turnover numbers (TONs). Nature found a solution to this problem by continuous regeneration of the catalysts (enzymes). However, such an approach could be hardly implemented in the artificial catalytic systems, thus leaving the preparation of stable and robust catalysts as an obvious alternative.
One of the principal approaches is the use of all-inorganic compounds with no organic ligands and no groups that can be easily oxidized [8][9][10][11]. However, in most cases, the use of organic ligand is

Crystal Structure of 3
The single crystal X-ray analysis reveals three features of the mononuclear structure where the Cu(II) atom is in a distorted square-planar coordination environment with two monoanionic 1,3,5-triazapentadienato species acting as N,N-chelators and forming two six-membered Cu metallacycles ( Figure 1). The Cu-N bond distances are almost equal (from 1.929(5) to 1.935(5) Å) and the cis N-Cu-N angles are close to 90 • varying from 87.8 (2) • to 96.0(2) • (Table 1). In contrast to the structures of complexes 1, 2, 4 and 5 [23], the N 4 environment around the copper atom is not completely planar with two ligands rotated at 24.2 • to each other.
The complex molecules are joined together by a set of N-H···O and N-H···N hydrogen bonds, involving the nitrogen atoms from C=NH imine groups and the n-propoxy oxygen atoms of the ligands, thus forming 2D supramolecular sheets (

Crystal Structure of 3
The single crystal X-ray analysis reveals three features of the mononuclear structure where the Cu(II) atom is in a distorted square-planar coordination environment with two monoanionic 1,3,5triazapentadienato species acting as N,N-chelators and forming two six-membered Cu metallacycles ( Figure 1). The Cu-N bond distances are almost equal (from 1.929(5) to 1.935(5) Å) and the cis N-Cu-N angles are close to 90° varying from 87.8(2)° to 96.0(2)° (Table 1). In contrast to the structures of complexes 1, 2, 4 and 5 [23], the N4 environment around the copper atom is not completely planar with two ligands rotated at 24.2° to each other.
The complex molecules are joined together by a set of N-H···O and N-H···N hydrogen bonds, involving the nitrogen atoms from C=NH imine groups and the n-propoxy oxygen atoms of the ligands, thus forming 2D supramolecular sheets (

Catalytic Oxidation of Cyclohexane with H 2 O 2
Cyclohexane is a widely used model substrate for C-H oxidation studies due to its suitable bond dissociation energy, easy identification and quantification of reaction products and importance of this process in industry. Thus, we investigated the potential of 1-5 for the mild oxidation of cyclohexane with hydrogen peroxide (Scheme 2). Along with the catalyst, we applied 10 equiv. of pyridine promoter, which is known to enhance the activity of transition metal catalysts, among them copper complexes [6,38,39].

Catalytic Oxidation of Cyclohexane with H2O2
Cyclohexane is a widely used model substrate for C-H oxidation studies due to its suitable bond dissociation energy, easy identification and quantification of reaction products and importance of this process in industry. Thus, we investigated the potential of 1-5 for the mild oxidation of cyclohexane with hydrogen peroxide (Scheme 2). Along with the catalyst, we applied 10 equiv. of pyridine promoter, which is known to enhance the activity of transition metal catalysts, among them copper complexes [6,38,39]. Accumulations of oxygenates (cyclohexanol and cyclohexanone) with time in the cyclohexane oxidation using 1 and 2.5 eq. of H2O2, catalysed by 1-5 (0.5 mol %), in the presence of pyridine (5 mol %) are depicted in Figure 4. The curves for complexes 1, 2 and 5 exhibit linear dependences at the initial period (Figure 4a). These complexes are least-active, showing equal initial reaction rates W0 of 1.8 × 10 −7 M s −1 (for 2.5 eq. of H2O2). After 30 min, the reaction rates slightly increase until 1.2 × 10 −6 M s −1 at 1 h (for 5, Figure 4a). After this period, all the accumulations undergo sharp accelerations, reaching a plateau after 2 h ( Figure S5). Complex 3 shows the higher W0 of 1.7 × 10 −6 M s −1 (for 2.5 eq. of H2O2). In contrast to 1, 2 and 5, complex 3 exhibits a rather short lag period of 20 min, then reaching the reaction rate of ca. 1 × 10 −5 M s −1 at 30 min time ( Figure 4a).
The catalytic behaviour of complex 4 was found to be dependent on the time between dissolution of the complex and initiation of the reaction (addition of H2O2). With this time minimized, the accumulation curve is linear in the 20-min period (W0 = 1.1 × 10 −5 M s −1 ), then showing a gradual decay of the reaction rate with the maximum concentration of oxygenates of 0.021 M which corresponds to the yield of 10.4% based on cyclohexane (Figure 4a). When complex 4 was kept for 2 min in acetonitrile (at 50 °C) prior to the addition of the oxidant, the accumulation curve appeared to be of non-linear character (Figure 4a) with the initial reaction rate (W0 = 9.9 × 10 −6 M s −1 ) very close to that for the catalytic system without pre-treatment of the complex 4 in acetonitrile.
For comparative purpose, the catalytic properties of copper nitrate were tested using similar conditions with 0.5 mol% loading of Cu(NO3)2. The choice of copper nitrate was governed due to its solubility in acetonitrile and stability in solution, while the commonly-used copper chloride is known to have complex behaviour, forming polynuclear Cl-bridged species in solution [40] (the same observation can be made for iron chloride also, basing on its non-linear W0 vs. [FeCl3]0 dependence) [41]. The accumulation of the reaction products for Cu(NO3)2 is linear in the first 40 min time with W0 = 1.9 × 10 −6 M s −1 , then reaching a plateau ( Figure 4a). Hence, the initial reaction rate exhibited by Cu(NO3)2 is considerably lower than that for complex 4, but higher than for all other complexes.
The accumulations for a 2.5 times lower amount of H2O2 (0.2 M; 1 eq.) discloses the drastic difference between complexes 3, 4 and 1, 2, 5 ( Figure 4b). While the latter show the negligible activity (W0 from 5 × 10 −8 to 8.5 × 10 −8 M s −1 ) with slow acceleration, becoming notable only after 1 h, catalysts 3 and 4 demonstrate a rapid increase of their reaction rates with the time (Figure 4b). Accumulations of oxygenates (cyclohexanol and cyclohexanone) with time in the cyclohexane oxidation using 1 and 2.5 eq. of H 2 O 2 , catalysed by 1-5 (0.5 mol %), in the presence of pyridine (5 mol %) are depicted in Figure 4. The curves for complexes 1, 2 and 5 exhibit linear dependences at the initial period ( Figure 4a). These complexes are least-active, showing equal initial reaction rates W 0 of 1.8 × 10 −7 M s −1 (for 2.5 eq. of H 2 O 2 ). After 30 min, the reaction rates slightly increase until 1.2 × 10 −6 M s −1 at 1 h (for 5, Figure 4a). After this period, all the accumulations undergo sharp accelerations, reaching a plateau after 2 h ( Figure S5). Complex 3 shows the higher W 0 of 1.7 × 10 −6 M s −1 (for 2.5 eq. of H 2 O 2 ). In contrast to 1, 2 and 5, complex 3 exhibits a rather short lag period of 20 min, then reaching the reaction rate of ca. 1 × 10 −5 M s −1 at 30 min time ( Figure 4a).
The catalytic behaviour of complex 4 was found to be dependent on the time between dissolution of the complex and initiation of the reaction (addition of H 2 O 2 ). With this time minimized, the accumulation curve is linear in the 20-min period (W 0 = 1.1 × 10 −5 M s −1 ), then showing a gradual decay of the reaction rate with the maximum concentration of oxygenates of 0.021 M which corresponds to the yield of 10.4% based on cyclohexane (Figure 4a). When complex 4 was kept for 2 min in acetonitrile (at 50 • C) prior to the addition of the oxidant, the accumulation curve appeared to be of non-linear character (Figure 4a) with the initial reaction rate (W 0 = 9.9 × 10 −6 M s −1 ) very close to that for the catalytic system without pre-treatment of the complex 4 in acetonitrile.
For comparative purpose, the catalytic properties of copper nitrate were tested using similar conditions with 0.5 mol % loading of Cu(NO 3 ) 2 . The choice of copper nitrate was governed due to its solubility in acetonitrile and stability in solution, while the commonly-used copper chloride is known to have complex behaviour, forming polynuclear Cl-bridged species in solution [40] (the same observation can be made for iron chloride also, basing on its non-linear W 0 vs. [FeCl 3 ] 0 dependence) [41]. The accumulation of the reaction products for Cu(NO 3 ) 2 is linear in the first 40 min time with W 0 = 1.9 × 10 −6 M s −1 , then reaching a plateau ( Figure 4a). Hence, the initial reaction rate exhibited by Cu(NO 3 ) 2 is considerably lower than that for complex 4, but higher than for all other complexes.
The accumulations for a 2.5 times lower amount of H 2 O 2 (0.2 M; 1 eq.) discloses the drastic difference between complexes 3, 4 and 1, 2, 5 ( Figure 4b). While the latter show the negligible activity (W 0 from 5 × 10 −8 to 8.5 × 10 −8 M s −1 ) with slow acceleration, becoming notable only after 1 h, catalysts 3 and 4 demonstrate a rapid increase of their reaction rates with the time (Figure 4b).
The dependences of the initial reaction rates on the oxidant concentration for 1-5 are depicted at Figure 5. The initial rates W 0 for catalysts 1, 2 and 5 are only slightly influenced by the H 2 O 2 excess. However, the higher H 2 O 2 concentration causes rapid acceleration of the reactions shortly after their beginnings ( Figure S6). Under these conditions, all the complexes show similar yields (18.5% for 4) and the highest turnover frequency (moles of product produced per mol of catalyst per a certain period) of  (Figure 4b). The initial reaction rate W 0 was found to be 5 × 10 −7 M s −1 , while the reaction rate at the quasi-linear period W lin is ca. five times higher (2.7 × 10 −6 M s −1 ). Considering the absence of the  The dependences of the initial reaction rates on the oxidant concentration for 1-5 are depicted at Figure 5. The initial rates W0 for catalysts 1, 2 and 5 are only slightly influenced by the H2O2 excess. However, the higher H2O2 concentration causes rapid acceleration of the reactions shortly after their beginnings ( Figure S6). Under these conditions, all the complexes show similar yields (18.5% for 4) and the highest turnover frequency (moles of product produced per mol of catalyst per a certain period) of 41 h −1 (for 4). The W0 vs.    The dependences of the initial reaction rates on the oxidant concentration for 1-5 are depicted at Figure 5. The initial rates W0 for catalysts 1, 2 and 5 are only slightly influenced by the H2O2 excess. However, the higher H2O2 concentration causes rapid acceleration of the reactions shortly after their beginnings ( Figure S6). Under these conditions, all the complexes show similar yields (18.5% for 4) and the highest turnover frequency (moles of product produced per mol of catalyst per a certain period) of 41 h −1 (for 4). The W0 vs.  For complex 4, a notable lag period is observed only for the lowest studied [H2O2]0 of 0.2 M (Figure 4b). The initial reaction rate W0 was found to be 5 × 10 −7 M s −1 , while the reaction rate at the quasi-linear period Wlin is ca. five times higher (2.7 × 10 −6 M s −1 ). Considering the absence of the lag period for [H2O2]0 > 0.2 M (Figures 4b, S6 and S7), one may conclude that the observed initial reaction rates when [H2O2]0 > 0.2 M correspond to the Wlin, but not to W0 (i.e., for complex 4 it is not possible to estimate true W0 values for [H2O2]0 > 0.2 M). This assumption is in accord with the nearly linear The structures of complexes 1-5 differ only by the type of substituents of the triazapentadiene ligands (Scheme 1). Assuming that the rate-limiting step is the reaction of a metal complex with hydrogen peroxide [15,[41][42][43][44][45][46], these substituents appear to be too distanced from the copper centre to sterically hinder this reaction. Considering the catalytic behaviour of 1-5, one may suppose that the complexes undergo gradual alteration in solution to form species in which catalytic activity is much higher than that of the original complexes. This hypothesis explains the observed acceleration of the reaction rates ( Figure 4 and Figures S5-S7) as well as the non-linear increase of the rates with an increase of the oxidant amount ( Figure 5). From this point of view, complexes 1, 2 and 5 appear to be more resistant to degradation, while complexes 3 and especially 4 are more prone to that.
We used UV/Vis spectroscopy to monitor the stability of the coordination compounds in the solution. The UV/Vis spectra of 1-5 in acetonitrile solutions are similar, showing a strong absorption band from 287 to 300 nm and a weak absorption at 492 nm ( Figure 6). The spectra of 1-3 and 5 do not undergo changes within 120 min, as evidenced by the intensities of the characteristic absorptions at 492 nm (Figure 6b). The spectra of 1 in the presence of pyridine revealed a slightly lower absorption with no significant changes upon time (Figure 6b). In contrast, the spectra of 4 is evidence for its gradual alteration and degradation in solution (Figure 7a).
The UV/Vis spectra of complex 4 (1 × 10 −3 M) in acetonitrile as a function of time are depicted in Figure 7a. After a short lag period, the peak at 492 nm disappears, while two other absorptions at 428 and 695 nm become visible (Figure 7a). The intensities of these two bands undergo rapid growth showing maximum at 60 min time, then decreasing until the background level after 4 h time. Visually, these changes correspond to the change of the solution colour from red to green (4 red → 4 green ) and then to the colourless one with a cloudy precipitate. We were interested in the study of if and how these changes depend on complex 4's concentration and other conditions.
The increase of the intensity of the 428 nm band is quasi-linear in the ca. 30-50 min range, allowing to calculate its increase rate W 428 (4. much higher than that of the original complexes. This hypothesis explains the observed acceleration of the reaction rates (Figures 4, S5-S7) as well as the non-linear increase of the rates with an increase of the oxidant amount ( Figure 5). From this point of view, complexes 1, 2 and 5 appear to be more resistant to degradation, while complexes 3 and especially 4 are more prone to that.
We used UV/Vis spectroscopy to monitor the stability of the coordination compounds in the solution. The UV/Vis spectra of 1-5 in acetonitrile solutions are similar, showing a strong absorption band from 287 to 300 nm and a weak absorption at 492 nm ( Figure 6). The spectra of 1-3 and 5 do not undergo changes within 120 min, as evidenced by the intensities of the characteristic absorptions at 492 nm (Figure 6b). The spectra of 1 in the presence of pyridine revealed a slightly lower absorption with no significant changes upon time (Figure 6b). In contrast, the spectra of 4 is evidence for its gradual alteration and degradation in solution (Figure 7a).
The UV/Vis spectra of complex 4 (1 × 10 −3 M) in acetonitrile as a function of time are depicted in Figure 7a. After a short lag period, the peak at 492 nm disappears, while two other absorptions at 428 and 695 nm become visible (Figure 7a). The intensities of these two bands undergo rapid growth showing maximum at 60 min time, then decreasing until the background level after 4 h time. Visually, these changes correspond to the change of the solution colour from red to green (4 red → 4 green ) and then to the colourless one with a cloudy precipitate. We were interested in the study of if and how these changes depend on complex 4's concentration and other conditions.
The increase of the intensity of the 428 nm band is quasi-linear in the ca. 30-50 min range, allowing to calculate its increase rate W428   The use of deuterated acetonitrile leads to a lower W428 rate of 1.9 × 10 −3 A min −1 and delay of the 428 nm maximum appearance (Figure 7b). The influence of acetonitrile is clearly seen on the plots of products accumulations in the course of cyclohexane oxidation performed in CH3CN and CD3CN The use of deuterated acetonitrile leads to a lower W 428 rate of 1.9 × 10 −3 A min −1 and delay of the 428 nm maximum appearance (Figure 7b). The influence of acetonitrile is clearly seen on the plots of products accumulations in the course of cyclohexane oxidation performed in CH 3 CN and CD 3 CN (Figure 8). While the reaction performed in normal acetonitrile shows acceleration with W 0 = 5.5 × 10 −7 M s −1 , the reaction in deuterated acetonitrile shows the constant reaction rate W = 2.2 × 10 −7 M s −1 . An elegant way to enhance the robustness of the catalyst is to protect its oxidation-sensitive C-H bonds by their deuteration [48]. The use of a fully deuterated 2-propanol as a solvent enabled the synthesis of the 4-d32 complex, containing deuterated i Pr-d7 moieties including the tertiary C-D bonds (Figure 8, inset). The UV/Vis kinetics of 428 nm absorption of 4-d32 in acetonitrile is depicted in Figure  7b. The presence of a 3 h lag period was accounted for 4-d32, after which the spectra similar to that for 4 appears. The W428 rate was found to be 4.2 × 10 −3 A min −1 . This value is also very close to that for normal complex 4 (4.7 × 10 −3 A min −1 ). A considerable lag period allows recrystallization of 4-d32 from acetonitrile to produce crystals with unit cell parameters (determined by the single crystal X-ray diffraction) equal to those for 4, confirming the retention of its integrity.
The accumulation of products (cyclohexanol and cyclohexanone) in the course of cyclohexane oxidation catalysed by 4-d32 is depicted in Figure 8. The reaction shows an acceleration with the initial reaction rate W0 (2.5 × 10 −7 M s −1 ) very close to that for the 4-CD3CN system (2.2 × 10 −7 M s −1 ). However, the acceleration rate exhibited by 4-d32 is much lower than that for 4, as expected for the higher resistance of 4-d32 towards oxidation.
Cyclohexyl hydroperoxide, CyOOH, was detected as a main reaction product of the cyclohexane oxidation using catalysts 1-5 ( Figure 9). The formation of large amounts of CyOOH is a sign for a free radical oxidation mechanism, where long-lived alkyl radials react with dioxygen to form alkyl hydroperoxides [6,24,49,50]. In the case of H2O2 oxidant, a free radical mechanism points to an involvement of hydroxyl radicals as attacking species [6,24,49,50]. Oxidation of methylcyclohexane (MeCyH) and adamantane catalysed by complexes 1 and 4 discloses low bond-and regioselectivities as well as the absence of stereoselectivity of these catalytic systems ( Table 2). Epimerization of cis-1,2dimethylcyclohexane (cis-DMCH) stereoconfiguration is a sign for the involvement of long-lived alkyl radicals [24,49,51]. These results are consistent with a free-radical mechanism where the H atom of the C-H bond is abstracted by non-selective HO· radicals, catalytically generated from H2O2 (Table  2) [6,24,49,50]. Larger oxygen-centred radicals, such as tBuO· one, result in similar discrimination between secondary and tertiary bonds of adamantane, but much higher bond selectivity in the oxidation of methylcyclohexane due to sterical hindrance of the R substitutes of the tertiary R3C-H bond ( Table 2). Higher selectivities are typical for catalytic systems oxidizing C-H bonds with highvalent metal-oxo (HVMO) species or metal-peroxide intermediates ( Table 2) [14]. For copper, As can be seen, the solution behaviour of complex 4 differs a lot from those for the other complexes. Looking at the structures of all the compounds, one may see the principal feature of complex 4: Each of the triazapentadiene ligands in 4 contain two tertiary C-H bonds (Scheme 1, Figure 1 and Figures  S1-S4). Since tertiary C-H bonds are typically more active than secondary and primary ones, we assigned the alteration of the structure of 4 in the solution to the activation of this bond with the formation (and consequent degradation) of new species. Considering that complexes 1-5 are known to catalyse aerobic oxidation of alcohols [23], oxidation of the C-H bond with dioxygen could be a probable explanation. To confirm such an assumption, we first studied the UV/Vis kinetics of complex 4 (1 × 10 −3 M) in acetonitrile in the presence of a large excess (0.27 M) of cis-1,2-dimethylcyclohexane (cis-DMCH). This substrate has easily-oxidizing tertiary C-H bonds [47] and in this way can act as a quencher of the reactive C-H-attacking species. As expected, no changes in the 428 (Figure 7b) and 695 nm absorptions occurred, only the 287 nm one showing a slight increase after 4 h. No cis-DMCH oxidation products were detected due to potentially low activity of such catalytic system and the presence of traces of the respective tertiary alcohols in the substrate. Further, when complex 4 was dissolved in degassed acetonitrile ([4] 0 = 1 × 10 −3 M; degassing was performed by freeze-pump-thaw cycling) and stirred under inert atmosphere, no changes in colour were observed during at least 96 h. In case the complexes 1-5 can be oxidized with dioxygen and/or catalyse this process, the oxygen from O 2 is expected to appear in the reaction products, first of all water. We attempted to follow this process by dissolving complexes 1 and 4 (as the most stable and least stable ones) in acetonitrile clearly demonstrates that dioxygen can be activated by the complexes studied (with their possible degradation) and this process can be triggered by the nature of the complex. An elegant way to enhance the robustness of the catalyst is to protect its oxidation-sensitive C-H bonds by their deuteration [48]. The use of a fully deuterated 2-propanol as a solvent enabled the synthesis of the 4-d 32 complex, containing deuterated i Pr-d 7 moieties including the tertiary C-D bonds (Figure 8, inset). The UV/Vis kinetics of 428 nm absorption of 4-d 32 in acetonitrile is depicted in Figure 7b. The presence of a 3 h lag period was accounted for 4-d 32 , after which the spectra similar to that for 4 appears. The W 428 rate was found to be 4.2 × 10 −3 A min −1 . This value is also very close to that for normal complex 4 (4.7 × 10 −3 A min −1 ). A considerable lag period allows recrystallization of 4-d 32 from acetonitrile to produce crystals with unit cell parameters (determined by the single crystal X-ray diffraction) equal to those for 4, confirming the retention of its integrity.
The accumulation of products (cyclohexanol and cyclohexanone) in the course of cyclohexane oxidation catalysed by 4-d 32 is depicted in Figure 8. The reaction shows an acceleration with the initial reaction rate W 0 (2.5 × 10 −7 M s −1 ) very close to that for the 4-CD 3 CN system (2.2 × 10 −7 M s −1 ). However, the acceleration rate exhibited by 4-d 32 is much lower than that for 4, as expected for the higher resistance of 4-d 32 towards oxidation.
Cyclohexyl hydroperoxide, CyOOH, was detected as a main reaction product of the cyclohexane oxidation using catalysts 1-5 ( Figure 9). The formation of large amounts of CyOOH is a sign for a free radical oxidation mechanism, where long-lived alkyl radials react with dioxygen to form alkyl hydroperoxides [6,24,49,50]. In the case of H 2 O 2 oxidant, a free radical mechanism points to an involvement of hydroxyl radicals as attacking species [6,24,49,50]. Oxidation of methylcyclohexane (MeCyH) and adamantane catalysed by complexes 1 and 4 discloses low bond-and regioselectivities as well as the absence of stereoselectivity of these catalytic systems ( Table 2). Epimerization of cis-1,2-dimethylcyclohexane (cis-DMCH) stereoconfiguration is a sign for the involvement of long-lived alkyl radicals [24,49,51]. These results are consistent with a free-radical mechanism where the H atom of the C-H bond is abstracted by non-selective HO· radicals, catalytically generated from H 2 O 2 ( Table 2) [6,24,49,50]. Larger oxygen-centred radicals, such as tBuO· one, result in similar discrimination between secondary and tertiary bonds of adamantane, but much higher bond selectivity in the oxidation of methylcyclohexane due to sterical hindrance of the R substitutes of the tertiary R 3 C-H bond ( Table 2). Higher selectivities are typical for catalytic systems oxidizing C-H bonds with high-valent metal-oxo (HVMO) species or metal-peroxide intermediates ( Table 2) [14]. For copper, however, such a mechanism is less probable due to the necessity of stabilization of the Cu III HVMO compounds [52][53][54]. One may note that some N 2 -and N 3 -donor ligands known to stabilize Cu III -O/Cu II -O· species resemble the triazapentadiene ones [55][56][57][58]. compounds [52][53][54]. One may note that some N2-and N3-donor ligands known to stabilize Cu III -O/Cu II -O· species resemble the triazapentadiene ones [55][56][57][58]. The difference between accumulations of the reaction products in normal and deuterated acetonitrile (Figure 8) cannot be explained from the point of view of a free radical mechanism with hydroxyl radicals as attacking species. It is known that the reaction rate of HO· radical with acetonitrile is ca. 100 times lower (k4/k3 ~ 0.01) than with cyclohexane [59-61]: The difference between accumulations of the reaction products in normal and deuterated acetonitrile (Figure 8) cannot be explained from the point of view of a free radical mechanism with hydroxyl radicals as attacking species. It is known that the reaction rate of HO· radical with acetonitrile is ca. 100 times lower (k 4 /k 3~0 .01) than with cyclohexane [59][60][61]: The reaction product appearing from the attack of HO· to acetonitrile could be acetamide, detected in small amounts by means of GCMS analysis. One may expect that deuterated acetonitrile reacts with HO· radicals even slowly due to the D/H kinetic isotope effect [62]. Hence, if transformation of the complex 4 is caused by the hydroxyl radical attack, such a transformation should be faster in CD 3 CN solvent rather than in CH 3 CN solvent (i.e., k 4 /k 3 > k 5 /k 3 ). However, this contradicts with the observed accumulations (Figure 8), where complex 4 shows first order dependence in CD 3 CN with the reaction rate considerably lower than that exhibited in CH 3 CN solvent. Finally, since the 4 red → 4 green transformation can occur in the absence of H 2 O 2 (Figure 7), hydroxyl radical attack can be ruled out as a reason for this transformation.
The presence of lag periods in the UV/Vis spectra of 4 suggests an autocatalytic nature of the 4 red → 4 green reaction, where the 4 green intermediate could catalyse further oxidation of the initial complex 4 (4 red ). An autocatalytic process of this type (involving the Cu I/II cycle) was recently described by Semenov et al. [63], where the reaction product catalysed the electron transfer. The 4 green intermediate could be responsible for the 4 red → 4 green transformation and the acceleration of the reaction rate of cyclohexane oxidation as well (Figures 4 and 8). From this point of view, the different behaviour of 4 in CH 3 CN and CD 3 CN solvents (Figures 7 and 8) can be explained by hampering the autocatalytic process in deuterated solvent, where the latter may act as a ligand.
The typical overoxidation products resulting from cyclohexane oxidation via HO· radicals are cyclohexanediols and hydroxycyclohexanones [41,[64][65][66][67][68]. After notable accumulation of these products, the C 6 ring cleavage compounds start to appear. However, in the case of catalysts 1-5, the overoxidation pattern is different from the expected one and, moreover, the amount of 6-hydroxyhexanoic acid is considerably higher than those for cyclohexanediols ( Figure 10). Therefore, while the main reaction mechanism is believed to proceed through the hydroxyl radical attack of a C-H bond, the pattern of the overoxidation products ( Figure 10) evidences for the presence of a minor oxidation mechanism of a different type. Since complexes 1-5 are active catalysts in the aerobic oxidation of alcohols [23], one may suppose that this process is responsible for the oxidation of cyclohexanediols affecting, as a result, the overoxidation pattern.
While cyclohexyl hydroperoxide is a relatively stable compound allowing its observation by means of gas chromatography [38,64,65,69,70], it was interesting to see if the other studied alkanes can also produce detectable alkyl hydroperoxides under the conditions of the experiment. To reach this aim, we compared the chromatograms recorded before and after the addition of solid PPh 3 (following the method developed by Shul'pin [15,50,71]) to the reaction samples ( Figures S9-S11). cyclohexanediols affecting, as a result, the overoxidation pattern.
For methylcyclohexane, a group of peaks appearing after the primary alcohol was attributed to methylcyclohexyl hydroperoxides ( Figure S9). The mass-spectrum of tertiary peroxide (1methylhydroperoxide) was found to be consistent with that reported earlier [72]. All hydroperoxide peaks completely disappear after treatment of the sample with PPh3. Oxidation of cis-DMCH affords the expected products (alcohols and ketones), as well as the typical by-products, such as 2,6octanedione ( Figure S10). Two group of peaks were detected, all having identical mass-spectra, with the differences in retention times expected for tertiary trans-and cis-products as well as the products of secondary carbons attack. Earlier, some of us reported similar mass-spectra of the products found in the course of cis-DMCH oxidation with the Cu9/HNO3/m-CPBA catalytic system (where Cu9 is a nonanuclear complex of copper with silsesquioxane and m-CPBA is meta-chloroperoxybenzoic acid) in the chromatograms recorded before quenching with PPh3 [73]. Hence, we can tentatively assign the peaks IV and V ( Figure S10) to the tertiary trans-and cis-DMCH hydroperoxides which can be reduced to the respective alcohols by using PPh3. In contrast to MeCyH and cis-DMCH, the analysis of adamantane oxidation products did not afford any peak which could be attributable to an adamantyl hydroperoxide ( Figure S11). For methylcyclohexane, a group of peaks appearing after the primary alcohol was attributed to methylcyclohexyl hydroperoxides ( Figure S9).
The mass-spectrum of tertiary peroxide (1-methylhydroperoxide) was found to be consistent with that reported earlier [72]. All hydroperoxide peaks completely disappear after treatment of the sample with PPh 3 . Oxidation of cis-DMCH affords the expected products (alcohols and ketones), as well as the typical by-products, such as 2,6-octanedione ( Figure S10). Two group of peaks were detected, all having identical mass-spectra, with the differences in retention times expected for tertiary transand cis-products as well as the products of secondary carbons attack. Earlier, some of us reported similar mass-spectra of the products found in the course of cis-DMCH oxidation with the Cu 9 /HNO 3 /m-CPBA catalytic system (where Cu 9 is a nonanuclear complex of copper with silsesquioxane and m-CPBA is meta-chloroperoxybenzoic acid) in the chromatograms recorded before quenching with PPh 3 [73]. Hence, we can tentatively assign the peaks IV and V ( Figure S10) to the tertiary transand cis-DMCH hydroperoxides which can be reduced to the respective alcohols by using PPh 3 . In contrast to MeCyH and cis-DMCH, the analysis of adamantane oxidation products did not afford any peak which could be attributable to an adamantyl hydroperoxide ( Figure S11). a Parameters 1 • :2 • :3 • (for methylcyclohexane and cis-1,2-dimethylcyclohexane) and 3 • :2 • (for adamantane) are normalized bond selectivities, corrected for the numbers of respective hydrogen atoms. The cis/trans ratio is the ratio of the tertiary alcohols having methyl groups in cis and trans positions, respectively. b Yield of products (cyclohexanol and cyclohexanone) based on the substrate. c turnover numbers, mols of products per mol of catalyst. d Pyridine. e Pyrazinearboxylic acid. f Tert-butylhydroperoxide. g Meta-chloroperoxybenzoic acid. h Zero value means that the primary alcohol was not detected or its amount is not reported. HL 1 = product of condensation of salicylaldehyde and 1-
All experiments were carried out in air, unless stated otherwise. UV/Vis spectra were recorded using Lambda 35 spectrometer (PerkinElmer, Waltham, MA, USA) in a 260-800 nm spectral range in the 1 cm length cells having 1.5 or 4 mL total volume. Elemental analyses for C, H and N were carried out by the Microanalytical Service of the Instituto Superior Técnico.

Crystallography
The X-ray diffraction data for 3 and 4-d 32 were collected using a D8 Quest diffractometer (Bruker, Germany) with graphite-monochromated Mo Kα radiation. Data were collected using phi and omega scans of 1 • per frame. Cell parameters were retrieved using Bruker SMART software and refined using Bruker SAINT on all the observed reflections. Absorption corrections were applied using SADABS 2016/2 [83]. The structure was solved by direct methods and refined against F 2 using the program SHELX-2018/3 [84]. The carbon atoms of two of four propyl groups (C3, C4 and C10, C11, C12) were modelled as being disordered over the two positions with site occupancies 0.55 (2)

Conclusions
In this work, we have studied the chemistry and catalytic behaviour of the copper coordination compounds [Cu{NH=C(OR)NC(OR)=NH} 2 ] with a bidentate alkoxy-1,3,5-triazapentadiene ligands having various substituents. Complex 3 was characterized by single crystal X-ray analysis. The coordination compounds show a moderate activity in the cyclohexane oxidation with hydrogen peroxide with the maximum observed reaction rate of 1.1 × 10 −5 M s −1 , corresponding to the TOF value of 41 h −1 . From the selectivity data, kinetic experiments and direct observation of cyclohexyl hydroperoxide, the principal reaction pathway appears to operate via a free radical mechanism involving long-lived alkyl radicals. However, there are indications for another minor mechanism of different type, which affects the overoxidation products pattern.
The kinetic data disclosed a complex solution behaviour of all the complexes revealing gradual alteration of their structures in solution. The complex 1 (R = Me) was found to be the most stable compound, while complex 4 (R = i Pr) showed the fastest alteration rate to form an intermediate showing high catalytic activity in cyclohexane oxidation. The lowest stability of complex 4 was associated to the presence of weak tertiary C-H bonds in its structure which are aerobically oxidized. Basing on the spectral and kinetic data, an autocatalytic nature of this alteration process was suggested. The use of CD 3 CN instead of CH 3 CN solvent enhanced the stability of 4 presumably by hampering the autooxidation reaction. Protection of complex 4 by deuteration of its i Pr substituents allowed higher stability of the complex under catalytic conditions. We expect that the results obtained within the present research would improve the understanding of copper-catalysed oxidation processes. Further research will be focused on the characterisation of the intermediate formed from the autooxidation of complex 4 and investigation of the respective reaction mechanism.