A resurrection of the Haber–Weiss reaction

arising from Zhao et al. Nature Communications https://doi.org/10.1038/s41467-020-20071-w (2020)

As Elimelech et al.¹ point out, there is a legitimate need for a method that produces singlet dioxygen (¹Δ_gO₂) efficiently, because this species plays a role in a research fields like environmental science and biochemistry. They¹ describe a flow-through filtration process whereby singlet dioxygen is generated electrochemically. The mechanism proposed by Elimelech et al.¹ for its production is based on the reduction of hydrogen peroxide by superoxide, the infamous Haber–Weiss reaction, and is therefore incorrect. Furthermore, the evidence for formation of singlet dioxygen is questionable.

The mechanism, shown in Fig. 4 of that publication¹, starts with reduction of O₂ at the cathode to O₂^•− and H₂O₂. H₂O₂ is, of course, also formed by the spontaneous disproportionation of O₂^•−. Subsequently, ¹Δ_gO₂ is proposed to be produced via reaction 1:

$${{{{{{{\rm{O}}}}}}}_{2}}^{{{\bullet- }}}+{{{{{{\rm{H}}}}}}}^{+}\,{+\,{{{{{\rm{H}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{2}\,{\to }\,^{1}{\Delta }_{{{{{{\rm{g}}}}}}}{{{{{{\rm{O}}}}}}}_{2}+{{{{{{\rm{HO}}}}}}}^{{{\bullet }}}+{{{{{{\rm{H}}}}}}}_{2}{{{{{\rm{O}}}}}}$$

(1)

This reaction became known as the Haber–Weiss reaction (with ¹Δ_gO₂ or³Σ_g⁻O₂) and was proposed in 1931². It proceeds with a rate constant of at best 1 M⁻¹s^{−1 3}, and cannot compete⁴ with the rapid and spontaneous⁵ disproportionation of O₂^•−.

Elimelech et al. cite my 1976 publication⁶ in support for formation of ¹Δ_gO₂. Indeed, I wrote that reaction (1) is thermodynamically possible. However, the standard Gibbs energies of formation of HO^• and O₂^•− have been determined more accurately since then⁷, with the result that reaction 1 with O₂ in the singlet state is thermodynamically not possible^3,8.

Elimelech et al. report that insignificant amounts of HO^• were detected. This should have led them to reject Reaction 1, because for every ¹Δ_gO₂ also one HO^• is formed. Could it be that the terephthalate concentration these authors used to detect HO^• was insufficient? If ¹Δ_gO₂ was detected with micromolar concentrations of furfuryl alcohol, then HO^• should also have been seen, given that the rate constants of ¹Δ_gO₂ and HO^• with furfuryl alcohol and terephthalate are similar, that terephthalate was present in millimolar concentrations, and that unlike HO^•, ¹Δ_gO₂ is quenched in water at a rate of 2.7 • 10⁵ s^{−1 9}. Although irrelevant at this stage, the notion that HO^• reacts with terephthalate to yield hydroxyterephthalate is incorrect. Instead, an adduct is formed that needs to be oxidised to yield hydroxyterephthalate.

The authors base their conclusion that ¹Δ_gO₂ is formed also on the reaction of the latter with 2,2,6,6-tetramethylpiperidine, but do not discuss the possibility that this compound may be oxidised at the anode and then yields 2,2,6,6-tetramethyl-4-piperidinol-N-oxyl after reaction with O₂¹⁰. Formation of any products from ¹ΔgO₂ would have been enhanced in D₂O where ¹ΔgO₂ lives much longer, or decreased by addition of a quencher, such as 1,4-diazobicyclo[2.2.2]octane.

Thermodynamically, a simpler route to ¹Δ_gO₂ could be the oxidation of O₂^•− to ¹Δ_gO₂ at the anode, because the electrode potential of the couple ¹Δ_gO₂(aq)/O₂^•− is +0.81 V¹¹. In contrast, the spontaneous disproportionation of O₂^•− is not an alternative, as the yield of ¹Δ_gO₂ varies from not detectable to extremely low, as recently reviewed⁸.

If we assume that is ¹Δ_gO₂ formed, we may ask: how much? Given the consumption of furfuryl alcohol, the rate constant of the reaction of this compound with ¹Δ_gO₂, the quenching rate constant for ¹Δ_gO₂ in water⁹, and the flow rate, one arrives at a low nanomolar steady-state concentration of ¹Δ_gO₂. This calculation also shows that only 1.8% of all ¹Δ_gO₂ reacts with furfuryl alcohol.

Experimental conditions and nomenclature need to be discussed too. Elimelech et al¹. use scavengers at a single concentration, with the exception of terephthalate which was used at two concentrations. These experiments do not prove that ¹Δ_gO₂, O₂^•−, H₂O₂, and HO^• are formed, because such a non-dose-dependent approach only gives an indication. The authors refer to these scavengers as specific for a particular species. Since HO^• reacts with nearly everything at high rates, such terminology is inappropriate. Indeed, N₃⁻ quenches ¹Δ_gO₂, but it also reacts with HO^•, as do p-benzoquinone and catalase. It would have been beneficial if the authors had consulted the Solution Kinetics Database of the National Institute of Standards and Technology¹² for the relevant rate constants. The description of ¹ΔgO₂ as “possessing an empty π* orbital”¹ is simplistic¹³. When electrofiltration was carried out to remove substances, temperature and pH were not mentioned. The word “quenching” is used as a synonym of “scavenging”, which is incorrect¹⁴. The term ROS for Reactive Oxygen Species, although widespread, is misleading as neither O₂^•− nor H₂O₂ are reactive, as argued before¹⁵. A study of the thermodynamics and kinetics of reactions of small, short-lived, oxygen-containing species will illustrate this.

In summary, it is not impossible that Elimelech et al. produced singlet dioxygen, but certainly not via the oxidation of O₂^•− by H₂O₂.