Hydrogen peroxide (H₂O₂) has found wide applications in industry. For example, it is a selective oxidant for the production of fine chemicals such as epoxides, bleaching agent in paper pulp production, a disinfectant, and a water purification agent. Industrial production of H₂O₂ is mainly by liquid phase autooxidation of anthraquinone, which is a free-radical process [1]. H₂O₂ can also be produced by autooxidation of other organic compounds, and an obsolete industrial process involved autooxidation of 2-propanol [1, 2].

The formation of H₂O₂ in the liquid phase accompanying catalyzed oxidation of organics has been reported nearly two decades ago. H₂O₂ was detected by titration during palladium (Ⅱ) acetate-catalyzed oxidation of terminal alkenes to ketones in an organic solvent [3, 4]. Its formation as the initial production of dioxygen reduction was implied in the Pd acetate-catalyzed 2-octanol oxidation from the stoichiometric consumption of molecular O₂ and 2-octanol [5]. H₂O₂ formation during aqueous phase oxidation of glycerol catalyzed by Au/C [6] and oxidation of CO catalyzed by Au/TiO₂ [7] and Au/Calgon [8] have been reported. Although these reports demonstrated that it is technically feasible to produce H₂O₂ by aqueous phase oxidation of organics or CO, the need to separate the organics and the soluble catalysts from the mixture can be expensive and poses a hurdle to commercialization.

Direct catalytic oxidation of H₂ with O₂ to form H₂O₂ in the vapor phase would be the most desirable, and it has been demonstrated in laboratories [9-11]. However, the process has not been commercialized due to concern for safety since efficient production involves H₂ and O₂ present as an explosive mixture. This concern for safety could be mitigated by using another reductant instead of H₂, such as organics. In this article, we offer a perspective of the feasibility of vapor phase production of H₂O₂ by simultaneous catalytic oxidation of organics by examining postulated reaction mechanisms and deriving from them what would be essential properties of a solid catalyst.

With the exception of direct formation of H₂O₂ from H₂ and O₂ and from autooxidation of anthraquinone, all available mechanistic information for H₂O₂ production are derived from investigations of liquid phase oxidation of organic compounds where the emphasis is on the mechanism of the organic reaction. Formation of H₂O₂ in the anthraquinone autooxidation process involves first the hydrogenation of anthraquinone to anthrahydroquinone. The H atom of the OH group in anthrahydroquinone is abstracted by an O₂ molecule to form a ∙O₂H radical (Scheme 1). Abstraction of another H atom from another OH group by the ∙O₂H radical completes the formation of H₂O₂ and regenerates the anthraquinone [12]. For the H₂O₂ formation during alcohol or aldehyde oxidation catalyzed by metal complexes such as Pd acetate, O₂ is activated on the metal complex by forming a peroxy ligand (Scheme 2). Insertion of the proxy into a metal-hydride bond forms a hydroperoxy ligand [13, 14]. Hydrogenation or hydrolysis of the metal-hydroperoxy bond results in H₂O₂ formation [15].

The mechanism of H₂O₂ formation during aqueous Au-catalyzed oxidation of glycerol and CO can be inferred from the isotopic labelling of the organic products when ¹⁸O-labeled O₂ or H₂O was used in the reaction. In glycerol oxidation, ¹⁸O was found in the glyceric acid product only when H₂¹⁸O was used but not with ¹⁸O₂ [6]. In contrast, in the Au-catalyzed CO oxidation-assisted formation of peroxide which was subsequent used to form propylene epoxide, ¹⁸O-labelled propylene epoxide was observed exclusively when ¹⁸O₂ was used but not with H₂¹⁸O [7]. These results are consistent with the mechanism shown in Scheme 3 [6]. O₂ is activated by associative adsorption, possibly as a peroxy species not unlike that in Scheme 2. It is then hydrogenated to form hydroperoxy, and eventually to H₂O₂. It should be noted that in Scheme 3, the catalytic cycle is not complete until the adsorbed hydroxyl (^*OH) is removed from the surface. In the CO oxidation-assisted process, the hydrogen source is H₂O, and the reaction of CO with ^*OH, similar to steps in the formation of adsorbed formate in the water gas shift reaction (CO+H₂O = CO₂ +H₂), completes the catalytic cycle and accounts for the appearance of labelled CO₂ when H₂¹⁸O was used but not with ¹⁸O₂ [7]. In the glycerol oxidation-assisted process, the removal of ^*OH was proposed to be accomplished by acquiring an electron to form hydroxyl anion (OHˉ) that desorbs into the liquid phase [6].

Without added H₂, another hydrogen source is needed to hydrogenate O₂. It would be ideal if H₂O can serve as the hydrogen source. However, direct oxidation of H₂O with O₂ (Eq. (1)) is thermodynamically uphill (ΔG_Eq.(1)⁰ = 116.7 kJ/mol). Thus a source of energy is needed to drive the reactions forward.

(1)

Light is the most commonly utilized energy source for H₂O₂ and ·OH production, and photocatalytic processes have been heavily studied and reviewed. Known photochemical methods generally have very low overall energy utilization efficiencies, since the theoretical (i.e. minimum) light energy required for this two-electron (i.e. two photon) process of Eq. (1) is 0.6 eV per photon, but a typical photocatalytic process utilizes UV light (e.g. 3.1 eV at 400 nm) that corresponds to a theoretical maximum energy efficiency of 20%. Even for process under development that utilizes visible light (e.g. yellow light of ~2.14 eV at 580 nm), the theoretical efficiency is only 29%. The actual energy efficiency is significantly less than 10% due to less than perfect quantum efficiencies and energy losses in the transfer of excited electron to surface species, as well as losses incurred with successive steps in the reaction sequence leading to peroxide formation. Furthermore, these processes demand reactors designed for maximum light exposure, efficient mixing, and thermal control.

Another source of energy that can be supplied to effect Eq. (1) is energized electrons from applied electrical potential. Although electrocatalysis is well established, process economics, energy utilization efficiency, and process complexity are barriers to their wide applications. Loss of energy efficiencies are due to the overpotential needed to activate the oxygen molecule, losses due to the competing reaction of electrolysis of water, and to the decomposition of the formed peroxide on the electrode surface.

Another approach is to make use of the widely practiced strategy in the chemicals and fuel processing industry, which is to couple reactions thermally, i.e. using heat released in one reaction to drive another independent reaction. This practice is highly inefficient, energy intensive, and costly for dilute streams since most of the energy is expanded in heating up very large quantities of diluent instead of being utilized to advance the desired reaction.

On the other hand, coupling of reactions at the elementary steps would be more desirable and energy efficient. However, there are few systematic studies of such coupled reactions. Thus, they represent a potential area of fruitful research. Among the few literature examples of coupling chemical reactions to achieve production of H₂O₂ in an aqueous medium is the coupling of CO oxidation to CO₂ with the formation of H₂O₂ using a Au catalyst mentioned earlier [8]. ¹⁸O-isotope labelling experiments indicated that the oxygen atoms in the peroxide are derived from the O₂ molecule, whereas those in CO₂ are derived from CO and H₂O [7]. Although the CO utilization efficiency is very low, as a majority of CO is oxidized to CO₂ without being coupled to water oxidation, the molecule serves as the sacrificial reductant and its oxidation to CO₂ provides the thermodynamic driving force.

In addition to CO, it is useful to consider what other compounds could serve as sacrificial reductants. Of particular interest would be organic compounds that could be oxidized to useful/valuable products and co-produced with H₂O₂. To this end, one can use thermodynamics to gauge the feasibility of the coupled reactions as to whether the organic oxidation reactions can provide sufficient thermodynamic driving force to effect H₂O₂ formation from H₂O and O₂. Table 2 shows the ΔG°_rxn of oxidation of different classes of organics. Compared with ΔG°_Eq.(1) of 116.7 kJ/mol for Eq. (1), it is clear that most oxidation reactions meet the minimum thermodynamic requirement, except epoxidation of alkenes. However, thermodynamics determines what can be achieved potentially without regard to reaction rates, and reaction mechanisms are needed for reaction kinetics discussion. Since coupling these reactions with H₂O₂ in the vapor phase has not been demonstrated yet, we will extrapolate the knowledge collected for the liquid phase reactions and postulate some feasible mechanisms for discussion, as illustrated below with alcohol oxidation.

Fig. 1 shows three possible mechanisms of alcohol oxidation to aldehyde coupled to H₂O₂ formation. Mechanism A is analogous to Scheme 2 in which an adsorbed peroxy species is formed when O₂ is adsorbed onto a catalyst, which is then hydrogenated to H₂O₂ while the alcohol is dehydrogenated to aldehyde. Although this process is thermodynamically feasible (ΔG_rxn⁰ ≈ ‒70 kJ/mol), it requires the unlikely formation of adsorbed H atom by cleaving a C‒H bond in the alcohol in the presence of O₂ (or adsorbed O). Instead, it should be more likely that breaking of such a C‒H bond is facilitated by adsorbed O via formation of a surface ^*OH, as in Mechanism B.

In mechanism B, O₂ is adsorbed as a surface peroxy species, which is hydrogenated by H transfer from an adsorbed H₂O (Ⅵ → Ⅶ). The presence of H₂O was found to be necessary in H₂O₂ formation during methanol oxidation using a Au catalyst, implying its involvement in the reaction mechanism [7]. A variation of this step is that the H is transferred from an adsorbed hydroxyl (^*OH). Further hydrogenation from another hydroxyl results in H₂O₂. Leaving behind an adsorbed O (Ⅷ). This adsorbed O is consumed in the oxidative dehydrogenation of the alcohol (Ⅷ → Ⅸ → Ⅴ).

In Mechanism C, O₂ is activated by adsorption onto a surface oxygen vacancy (υ in X), which significantly weakens the O─O bond. Cleavage of this bond and insertion of its O into a surface ^*OH results in a surface peroxy (Ⅻ), which is hydrogenated to H₂O₂. The O atom that is left in the vacancy is consumed in the alcohol oxidation (ⅩⅢ → XIV → X) to regenerate the vacancy.

Additional experimental data will be needed to test the validity of these mechanisms. Nonetheless, at this point, one may use these postulated mechanisms to develop guidelines for catalysts that might be suitable for the generation of H₂O₂ during oxidation of an organic compound. Upon examining the two more likely mechanisms B and C, a common feature becomes apparent: the need to remove the O atom left on the catalyst surface in the process of H₂O₂ formation. In mechanism B, this residual atom is an adsorbed O (Ⅷ), and in mechanism C, it is a lattice O (ⅩⅢ), and they are removed by reaction with the sacrificial organic reductant. Thus, the minimum ΔG_rxn⁰ of the oxidation reaction of the sacrificial reductant must be sufficient to compensate not only the ΔG_Eq.(1)⁰ but also the energy (ΔG_^*O⁰) associated with removal of the residual O (denoted ^*O). Fig. 2 show this minimum ΔG_rxn⁰ needed, which is equal to the sum of ΔG_^*O⁰, the kinetic barrier for the removal ^*O (i.e. that barrier to form ^*O^TS) ΔG_barrier⁰, and ΔG_Eq.(1)⁰.

Fig. 3 summarizes these considerations for the case of removal of adsorbed O, plotting the sum of ΔG_^*O⁰, ΔG_barrier⁰, and ΔG_Eq.(1)⁰ versus ΔG°_rxn for the reaction of the sacrificial reductant. The dotted horizontal line at 116.7 kJ/mol is the thermodynamic minimum required to overcome the ΔG_Eq.(1)⁰. Any energy needed to overcome the kinetic barrier (ΔG_barrier⁰) and to break the surface-oxygen bond (ΔG_^*O⁰) will be addition to that. The three sloping lines represent three ΔG_barrier⁰ of 10, 30, and 50 kJ/mol, respectively, and various values of ΔG_^*O⁰. A point on these lines corresponds to the situation when:

ΔG°_rxn = ΔG_Eq.(1)⁰ + ΔG_barrier⁰ + ΔG_^*O⁰

Thus, points to the left of these lines correspond to cases when the energy of oxidation of the sacrificial reductant is insufficient to overcome the (ΔG_Eq.(1)⁰ + ΔG_barrier⁰ + ΔG_^*O⁰) requirement for catalytic H₂O₂ production for that catalytic system. On the other hand, points to the right of these lines correspond to catalytic systems that are feasible. This analysis shows that for practical purposes, it is difficult to use hydrocarbon molecules effectively as sacrificial reductant for H₂O₂ production. Alcohols and aldehydes should be more effective. The analysis also provides guidelines for the choice of catalysts by providing constraints for strength of adsorption of O. In general, catalysts with low binding energy of O is preferable. Thus, the weak binding energy of O on Au makes Au a good catalyst choice. Other candidates include various metal oxides such as Fe, Cu, Mn, Zn, and Cr oxide, according to the data in Table 3. The conclusion that catalysts of weaker surface–oxygen bond is desirable is similar to that arrived for electrochemical production of H₂O₂ [16].

The discussion thus far focuses on the production of H₂O₂ as the primary product. For an economically viable industrial process, it is also necessary to produce H₂O₂ at high concentrations with minimal consumption of sacrificial reductant. These demand additional requirements on the catalyst not discussed here, which include catalysts to have low activity in decomposing H₂O₂ (into H₂O and O₂) but high selectivity in oxidizing the sacrificial reductant to a selective oxidation product (such as alcohol to aldehyde but no further). For a long time, researchers have investigated the desirable catalytic properties for selective oxidation of hydrocarbons, but none has addressed the coupled production of H₂O₂. Thus this can be a new and fruitful direction of research.

Fig. 3 identifies organic oxidation reactions that can be used to produce H₂O₂. The ΔG⁰ of the coupled processes is thermodynamically favorable. In principle, for these coupled processes, it is possible to also derive useful work from them. One such method is to couple these reactions electrochemically. Although such processes need to be in the liquid phase, they are included for discussion because of the potential high overall energy efficiency. Thus, oxidation of the reductant (e.g. alcohol) can occur at an anode where protons and electrons are produced (Eq. (2)). At the cathode, the electrons reduce O₂, and the reduced O₂ combines with protons to from H₂O₂ (Eq. (3)), and the overall reaction is the oxidation of the reductant with the formation of H₂O₂. The available work in the overall reaction can be captured as electrical work.

(2)

(3)

Although such an electrogenerative concept has been known for a long time and is the basis of fuel cell operations [17, 18], its potential for H₂O₂ generation has not been extensively studied.

Recently, there is emerging interest in using microbial bioelectrochemical systems (BESs) such as Microbial Fuel Cells (MFCs) and Microbial Electrolysis Cells (MEC) that use microorganisms as catalysts to oxidize organic compounds at the anode of the electrogenerative cell. Use of a microbial fuel cell to generate H₂O₂ was first reported about a decade ago using acetate oxidation to bicarbonate to provide the thermodynamic driving force [19]. Since then, other reactions have been used successfully, including glucose oxidation [20].

There are analyses to determine requirements for optimal electrocatalysts [21]. Monometallic, bimetallic, organic semiconductor [22], oxidic, and N-doped C electrodes are known to produce H₂O₂ rather selectively [21, 23]. The reactions at the electrode surfaces are:

(4)

(5)

where ^* denotes a surface adsorption site. Thus, the reaction rate depends on the binding strength of hydroperoxy (^*-OOH) [21]. For surfaces where the binding strength is too weak such as on Au and Pd/Au, formation of ^*-OOH is limiting; whereas for surfaces where the binding is too strong such as Pt and Pd, degradative dissociation of ^*-OOH limits H₂O₂ formation.

In this article, the feasibility of a vapor phase catalytic H₂O₂ production process by coupling the oxidation of water with the oxidation of an organic compound is analyzed. Consideration of the thermodynamics of the reactions involved provides constraints that determine which organic reactions are useful. By postulating feasible reaction mechanisms, we further discussed the challenges of identifying a suitable catalyst that would promote the reaction at an adequate rate. The result of the analysis identified metal-oxygen bond strength to be a useful parameter for high activity. In general, catalysts with stronger metal-oxygen bonds require more thermodynamically favorable (more negative ΔG°_rxn) organic oxidation reactions to couple with the H₂O oxidation reaction. Undoubtedly, this generalization is subjected to refinement by taking into account variations in parameters that are assumed constant or estimated with simplified assumptions, especially the various bond energies, but the generalization arrived here may serve as a starting point in search of suitable catalysts. The analysis also identified potentially fruitful research directions. One is to expand our understanding of desirable catalytic properties for selective oxidation of hydrocarbons to include the requirement for coupled production of H₂O₂, and another to understand the requirement to minimize H₂O₂ decomposition activity of a catalyst without negatively impacting its H₂O₂ formation activity.

The generous support by Northwestern University through a gift from Dr. Warren Haug is greatly appreciated.