Introduction

The proposed reaction mechanisms for the oxidation of organic substrates with the Fenton reagent[155] (a mixture of ferrous ions and hydrogen peroxide) can roughly be divided into two groups. The first group regards Fenton chemistry as the production of free hydroxyl radicals by the metal catalyzed decomposition of the peroxide [156]

$$Fe^{2+} + H_2O_2 \rightarrow Fe^{3+} + OH^{-} + OH\mbox{\.{}}$$ (59)

The other mechanisms involve the formation of a highly reactive iron-oxo complex such as the ferryl ion ([Fe $^{\mathrm{IV}}$O]$^{2+}$) as the oxidative intermediate.[157] The reactions are of industrial interest for their applications in waste water treatment and paper bleaching, but also in biological processes are their oxygen activation abilities recognized. Recently, much attention has been given to iron containing biological molecules that efficiently catalyze the oxidation of organic substrates, such as the antitumor drug bleomycin[158], cytochrome oxydase[159,160,161,162] and methane monooxygenases[163,164]. Biomimetically designed ligand environments of iron complexes[132,131] are studied to optimize industrial catalysts, but also other chelated[165,166] and un-chelated[167,136,139] iron(II)/H$_2$O$_2$ complex in aqueous solution are studied as well as oxidations by bare iron-oxo species in the gas phase[168,169,170] in order to reveal the reaction mechanisms. Despite the numerous studies over more than 60 years, the controversy remains as there have not yet been definitive experiments to distinguish between the proposed alternatives. The main experimental difficulty, the extremely short life times of the reaction intermediates, is not a problem for computer simulation methods. A second difficulty is the sensitivity of Fenton chemistry to the reaction conditions, such as the pH, the metal ligands (chelating agents), and the nature of organic substrates, which complicate the development of a microscopic model. We have therefore first restricted ourselves in a previous study to the mechanism of the basic reaction between hydrated Fe(II) and hydrogen peroxide in vacuo[145]. Subsequently, the influence of other parameters on the mechanism will be the subject of study, starting with the solvent effect of aqueous solution in the present study. A preliminary account has appeared in ref. bernd3.

In the gas-phase study, we have analyzed the reaction of H$_2$O$_2$ with Fe$^{2+}$ from the point of view of the energetics and the electronic and geometric structure using Density Functional Theory (DFT) calculations. As the primary step in the proposed models usually is the coordination of the hydrogen peroxide to iron,

$$L_6Fe^{II} + H_2O_2 \leftrightarrow L_5Fe^{II}H_2O_2 + L$$ (60)

we started from a configuration of H$_2$O$_2$ coordinated to a pentaaqua iron(II) complex. It turned out that the complexed hydrogen peroxide easily dissociates but the formation of free OH. radicals is from an energetic point of view very unlikely. Instead, the formation of the ferryl ion ([(H$_2$O)$_5$Fe $^{\mathrm{IV}}$=O]$^{2+}$) plus water is exothermic by $\Delta E_0$ = 28 kcal/mol.[145]

In practice however, these reactions take place in aqueous solution and the solvent effects are expected to be an important factor in the balance between the competing reactions. Already in our gas phase study we have observed that adding a single water molecule could facilitate the formation of the ferryl ion from the Fenton reagent by lowering the reaction barrier of the proton transfer process. In the present study, we introduce the effects of a complete water solution environment and report the oxidizing intermediates formed in the reaction of Fe$^{2+}$ and hydrogen peroxide in water at $T$=300K predicted by computer simulation. To be able to model correctly the active role of the solvent water molecules in the Fenton reaction, the method of choice for this study is the Car-Parrinello (CP) method.[49] The CP method applies classical molecular dynamics (MD), computing the inter- and intramolecular forces from the electronic structure determined quantum mechanically with Density Functional Theory (DFT) in an efficient way. The CP method is therefore regarded as an ab initio (DFT) molecular dynamics (AIMD) method. This method has proven to be a very valuable tool to study at a microscopic scale structure and dynamics of water[50,51], (metal-) ions in water[117,52,53,172] and simple chemical reactions in aqueous solution.[144,54,55] The full strength of AIMD becomes apparent in the simulation of bond breaking and making as e.g. a proton or OH. radical jumps through an aqueous solution; something that is practically impossible to model with classical force fields.

This paper is structured as follows: in the next section (section 5.2), we start with the computational details. Then, in order to asses the accuracy of the Car-Parrinello method in the description of the solvent structure around an iron complex, we compare structural properties of FeCl$_2$ in water with experimental data (section 5.2.1). The presentation of our results (section 5.3) begins with static DFT calculations of the energetics for the elementary reactions to produce the OH. radical and the ferryl ion with the Fenton reagent in vacuo, in section 5.3.1. Our main results, the reaction mechanisms of iron(II) and H$_2$O$_2$ in aqueous solution at room temperature, are presented in sections 5.3.2, 5.3.3 and 5.3.4. We first show the spontaneous formation of iron(IV)oxo species in an AIMD simulation when we start from H$_2$O$_2$ coordinated to pentaaqua iron(II), via an iron(IV) dihydroxo intermediate, with and without solvent molecules playing an active role in the mechanism (sections 5.3.2 and 5.3.3, respectively). In section 5.3.4, we also include the coordination process, by starting from separated reactants in water. A summary and conclusions are given in section 5.4.