Hydrogen Peroxide as Green Oxidant: Mechanistic Insight in the Direct Synthesis and Application

In recent years the need to lower the environmental impact of chemical processes and the reduction of the reliance of non-renewable raw materials have been in the focus of academic and industrial research to battle and overcome the challenges of pollution and environmental changes originating from climate change. Oxidation reactions play a critical role in this research due to their application in wastewater treatment, downstream processing and as a source of important product groups in the chemical industry including epoxides, ketones and aldehydes.

The choice of oxidant plays an integral part in steering conversion and selectivity as well as the need for catalysts. Classical organic and inorganic oxidants can lead to toxic by-products and raise the need for expensive down streaming processing. Molecular oxygen, if used as oxidant, avoids these problems however, the difficulty of activating the oxygen bond includes the requirement of catalytic materials including, often expensive, noble metals catalysts. As an alternative hydrogen peroxide provides a high oxidation potential and when used as oxidant only produces water as by-product of the reaction.

The production of hydrogen peroxide relies to a big part on the anthraquinone process in which anthraquinone is reduced and reoxidized using noble metal catalysts, producing hydrogen peroxide during the oxidation step. This process however is only viable in larger scale production due to its high cost. In contrast, the direct synthesis of hydrogen peroxide, produces hydrogen peroxide directly from hydrogen and oxygen using noble metal catalysts and low environmental impact solvents such as water and small aliphatic alcohols and offers a cheaper alternative for smaller production scale. Especially smaller alcohols have proven to increase the yield of the direct synthesis. This was classically explained by higher solubilities of the gases in small alcohols disregarding changes in the reaction order of the reactants. The first part of this study elucidates the changes in the reaction mechanism when performed in aliphatic alcohols in contrast to water.

The role of small aliphatic alcohols was studied using transient experiments in a down flow lab-scale trickle bed reactor using a commercial 5% Pd/C catalyst and methanol, ethanol and iso-propanol. During a solvent switch oscillatory pattern of the hydrogen peroxide concentration were observed after the change from water to the corresponding alcohol. A strong dependency of the pattern on the reaction temperature, concentration of the alcohol and chain length of the alcohol could be established. In addition to hydrogen peroxide, with a small delay corresponding dehydrogenation products of the alcohol were detected, undergoing a similar pattern. In complementary gas step experiment in which the fed gas was changed during the experiments, significant difference in the concentrations of hydrogen peroxide after the change were measured. After the change from hydrogen to oxygen and vice versa the concentration of hydrogen peroxide shows a peak. However, after a switch from oxygen the measured concentration was 100 fold higher than after the switch from hydrogen to oxygen.

In order to get deeper mechanistic insight in-situ FTIR ATR experiments, mimicking the solvent switch experiments in the trickle bed reactor using methanol, were performed. During those experiments the same oscillatory pattern of the hydrogen peroxide concentration was observed. The existence of hydroxyalkyl species on the catalyst surface, appearing in the same pattern, was confirmed. Additionally methoxy species were detected during the reaction.

Based on these observations, a new mechanism was proposed in which small alcohols can act as a cocatalyst by forming hydroxyalkyl species, which can act as hydrogen donors in the reduction of oxygen. These species are then regenerated by hydrogenation with the hydrogen present in the system. At low hydrogen concentrations, the regeneration is insufficient and the dehydrogenation product is formed.

The influence of the structure of the used alcohol further FTIR-ATR studies were performed using ethanol as solvent. In steady state experiments and solvent step experiments the same hydroxyl alkyl species as in the reaction with methanol were observed. To eliminate possible observer species and to decrease overlapping of neighbouring bands MES studies were applied revealing the existence of double bond structures indicating rearrangements taking place. From these observations a 2 step dehydrogenation is proposed for higher alcohols.

In the second part of the study the demonstrate the excellent properties of hydrogen peroxide as oxidant was used in the selective oxidation of glucose. Aldonic Acid, which are used in the cosmetic, pharmaceutical, food industry as well as the production of fine chemicals, are classically produced using oxygen and noble metal catalysts. To study the reaction 12 different catalyst were synthesised with varying support (Al2O3, TS-1, TiMWW), metal (Au, Pd, Fe, W), and nanoparticle size (Au) and tested in a lab-scale batch reactor.

From all combinations Au/Al2O3-catalysts depicted the best performance. The main influence of the support could be seen side reaction producing formic acid and levulinic acid. This side reaction was found to depict autocatalytic behaviour by forming performic acid. This autocatalytic behaviour was hindered on titanium silicates. Only with gold catalysts very high conversions and selectivities could be achieved, while with iron and palladium lead to strong decomposition of the glucose. Only tungsten oxide based catalyst depicted moderate conversion and selectivity.

The structure sensitivity analysis, revealed a maximum in rate for a nanoparticle size of around 6.5 to 7nm. This maximum is significantly higher than classically reported in literature for gold catalysts. The position of this maximum showed a high temperature dependence. Based on this, the apparent activation energy of 5 selected Au/Al2O3-catalysts with varying nanoparticle size was determined. The strong changes in the apparent activation energies were assigned to difference in the activation energy on edges and terraces. A theoretical model based on a Langmuir-Hinshelwood-Mechanism with a non-linear step and first order for both reactants was implemented and showed good fit with experimental data.