Resolving the Mechanism for H2O2 Decomposition over Zr(IV)-Substituted Lindqvist Tungstate: Evidence of Singlet Oxygen Intermediacy

The decomposition of hydrogen peroxide (H2O2) is the main undesired side reaction in catalytic oxidation processes of industrial interest that make use of H2O2 as a terminal oxidant, such as the epoxidation of alkenes. However, the mechanism responsible for this reaction is still poorly understood, thus hindering the development of design rules to maximize the efficiency of catalytic oxidations in terms of product selectivity and oxidant utilization efficiency. Here, we thoroughly investigated the H2O2 decomposition mechanism using a Zr-monosubstituted dimeric Lindqvist tungstate, (Bu4N)6[{W5O18Zr(μ-OH)}2] ({ZrW5}2), which revealed high activity for this reaction in acetonitrile. The mechanism of the {ZrW5}2-catalyzed H2O2 degradation in the absence of an organic substrate was investigated using kinetic, spectroscopic, and computational tools. The reaction is first order in the Zr catalyst and shows saturation behavior with increasing H2O2 concentration. The apparent activation energy is 11.5 kcal·mol–1, which is significantly lower than the values previously found for Ti- and Nb-substituted Lindqvist tungstates (14.6 and 16.7 kcal·mol–1, respectively). EPR spectroscopic studies indicated the formation of superoxide radicals, while EPR with a specific singlet oxygen trap, 2,2,6,6-tetramethylpiperidone (4-oxo-TEMP), revealed the generation of 1O2. The interaction of test substrates, α-terpinene and tetramethylethylene, with H2O2 in the presence of {ZrW5}2 corroborated the formation of products typical of the oxidation processes that engage 1O2 (endoperoxide ascaridole and 2,3-dimethyl-3-butene-2-hydroperoxide, respectively). While radical scavengers tBuOH and p-benzoquinone produced no effect on the peroxide product yield, the addition of 4-oxo-TEMP significantly reduced it. After optimization of the reaction conditions, a 90% yield of ascaridole was attained. DFT calculations provided an atomistic description of the H2O2 decomposition mechanism by Zr-substituted Lindqvist tungstate catalysts. Calculations showed that the reaction proceeds through a Zr-trioxidane [Zr-η2-OO(OH)] key intermediate, whose formation is the rate-determining step. The Zr-substituted POM activates heterolytically a first H2O2 molecule to generate a Zr-peroxo species, which attacks nucleophilically to a second H2O2, causing its heterolytic O–O cleavage to yield the Zr-trioxidane complex. In agreement with spectroscopic and kinetic studies, the lowest-energy pathway involves dimeric Zr species and an inner-sphere mechanism. Still, we also found monomeric inner- and outer-sphere pathways that are close in energy and could coexist with the dimeric one. The highly reactive Zr-trioxidane intermediate can evolve heterolytically to release singlet oxygen and also decompose homolytically, producing superoxide as the predominant radical species. For H2O2 decomposition by Ti- and Nb-substituted POMs, we also propose the formation of the TM-trioxidane key intermediate, finding good agreement with the observed trends in apparent activation energies.


Mechanism 1a
Hereinafter, M(OH)2M is dimer {ZrW5}2, MOH -monomer ZrW5 existing in chemical equilibrium with M(OH)2M (Eq. S1), and MOOH is monomeric hydroperoxo complex For Mechanism 1a, we assume that MOOH is not stable and disappears due to reverse transformation into MOH and/or reaction with H2O2. Therefore, These approximations (in combination with the mass-balance equation S4) allow us to find all POM form (M(OH)2M, MOH, and MOOH) concentration: The reaction rate is: If we assume that the major part of POM exists in the MOH form, we can simplify the right part of Eq. S6 by using the Taylor series: The reaction rate after simplification is: (S10) Therefore, Mechanism 1a is anticipated to give second reaction order in H2O2 when [H2O2] is small and first reaction order in H2O2 when [H2O2] is high.

Mechanism 1b
For Mechanism 1b, we assume that MOOH is relatively stable and exists in chemical equilibrium with MOH (Eq. S11). (S11) Using this assumption in combination with mass-balance equations we can find all POM form (M(OH)2M, MOH, and MOOH) concentrations.
The reaction rate before simplification is: After simplification (by using the Taylor series): (S13) Hence, we can conclude that Mechanism 1b also gives second reaction order in H2O2 when [H2O2] is small and first reaction order in H2O2 when [H2O2] is high.

Mechanism 2b
For Mechanism 2b, we assume that MOOH and MOOH(H2O2) are relatively stable and exist in chemical equilibriums with MOH and MOOH, respectively.
The rate of H2O2 decomposition has two contributions: MOOH decomposition (step 3, WMOOH) and MOOH(H2O2) decomposition (step 5, WMOOH(H2O2)): Step 3 contribution to the reaction rate: Step 5 contribution to the reaction rate: The reaction rate is therefore:

Mechanism 3a
For Mechanism 3a, we assume that MOOH is not stable and prone to reverse transformation to MOH and/or decomposes to give peroxide degradation products. [ Using this approximation in combination with mass-balance we can find all POM form (M(OH)2M, MOH, and MOOH) concentrations.
The reaction rate after simplification (by using the Taylor series): Therefore, Mechanism 3a gives first reaction order in H2O2 when [H2O2] is small and zero reaction order when [H2O2] is high.

Mechanism 3b
For Mechanism 3b, we assume that MOOH is relatively stable and exists in chemical equilibrium with MOH.
Using that approximation in combination with mass-balance allows us to find all POM form (M(OH)2M, MOH, and MOOH) concentrations.
The rate of reaction after simplification (by using Taylor series):  [ The reaction rate is therefore: Mechanisms 2b, 3a, 3b, and 4 give 1-0 reaction order in [H2O2] and first reaction order in [POM].
Thus, based only on the observed reaction orders in the reactants, we cannot distinguish between these mechanisms. In order to choose the most probable mechanism, we have used double reverse coordinates 1/W0 -1/[H2O2].
Mechanisms 2b and 4 have the same expression for the reaction rate (compare Eqs. S24.1 and S35.1): where A, B, and C depend on the specific mechanism (2b or 4) and [M(OH)2M]0. Therefore, for mechanisms 2b and 4, the reaction rate function becomes polynomial in double reverse coordinates (Eq. 37): On the other hand, mechanisms 3a and 3b have another expression for the reaction rate (see Eqs. S26 and S28): where A, B, and C depend on the mechanism type (3a or 3b) and [M(OH)2M]0.
Therefore, for mechanisms 3a and 3b, the reaction rate function becomes linear in the double reverse coordinates (Eq. 39): The kinetic modelling study made also possible to estimate the value of K4 (mechanism 2b) as ca.
20. On the other hand, K2 was impossible to determine because its value is determined by the position of the minimum of the polynomial, which we fit to the experimental data in inverse coordinates ( Figure S6). To accurately determine the minimum, it is necessary to have experimental data points in both branches of the polynomial. However, at positive K2, the minimum is in the negative region along the 1/[H2O2] axis; therefore, it is impossible to obtain experimental points for the second branch of the polynomial, since they correspond to negative H2O2 concentrations. Therefore, the value of K2 cannot be determined with a good accuracy within the framework of our model.    The following parameters were used for EPR spectrum simulation (C): g1 = 2.0375; g2 = 2.0121; g3 = 2.0053; g1Strain = 0.0051; g2Strain = 0.0041; g3Strain = 0.0049.

Optimization of α-terpinene oxidation
The optimization of reaction conditions was carried out for the α-terpinene oxidation in order to obtain a maximum yield of ascaridole. Double increase of the catalyst amount (H2O2/Zr = 25 vs 50) led to the increase of α-terpinene conversion and product yield as well as reduced the reaction time (Table S1, compare entries 1 and 2). Note that at such molar ratio of H2O2 to Zr epoxidation of cyclohexene was negligible. ¡Error! Marcador no definido. When a more concentrated H2O2 reagent (50 or 77% instead of 30%) was used, α-terpinene conversion and ascaridole yield increased (Table S1, entries 3 and 4). This indicates that less amount of H2O in the reaction medium is more favorable for the formation of the endoperoxide. However, α-terpinene oxidation with anhydrous urea-hydrogen peroxide adduct was not successful: the ascaridole yield reached only 25% (Table S1, entry 5). Further optimization of the reaction conditions was performed using 50% H2O2 as more accessible and safer oxidant. The concentration of H2O2 in the reaction mixture turned out to be a crucial factor for the α-terpinene oxidation ( Figure S7). Once the oxidant was added in one portion, the stoichiometric amount of H2O2 (i.e., 2 equiv. to substrate) was optimal. With 1 equiv. of oxidant, the yield of ascaridole reached only 32%, while the increase of H2O2 concentration up to 3 equiv. also reduced the yield down to 43% in comparison with 62% for 2 equiv. of H2O2. However, when 3 equiv. of the oxidant was added dropwise into the reaction mixture, endoperoxide yield increased significantly and attained 88% ( Figure S7). In all cases the selectivity to ascaridole was 84-88%.
Under the same conditions, with the dropwise addition of the oxidant, TME produced the corresponding hydroperoxide (Scheme 2, main text) with a 70% yield.     Figure S11. Gibbs free-energy path of the monomerization and its 1 st H2O2 activation. All freeenergies are in kcal·mol -1 and are relative to the starting dimeric structure Ad.  Figure S12. Free-energy profile (kcal·mol -1 ) for the evolution of monomeric Zr-trioxide intermediate D to produce singlet oxygen through a water mediated process where trioxidane is formed. All energies are relative to the initial dimeric specie Ad. Figure S13. Alternative reaction pathways for the second H2O2 activation by Zr-dimeric species.
Gibbs free energies with respect to species Ad are given in kcal·mol -1 .       H2O2 and H2O (0.08 M, 0.20 M and 0.66 M, respectively) were set to reproduce the experimental conditions. Entries 1-3 and 4-6 evaluate the impact of the water content on the reaction rate and overall path distribution. Reaction rates (rrel.) were estimated as the time needed to consume 90% of the initial H2O2 and were normalized by the fastest one (Entry 3).