Copper-assisted oxidation of catechols into quinone derivatives

Catechols are ubiquitous substances often acting as antioxidants, thus of importance in a variety of biological processes. The Fenton and Haber–Weiss processes are thought to transform these molecules into aggressive reactive oxygen species (ROS), a source of oxidative stress and possibly inducing degenerative diseases. Here, using model conditions (ultrahigh vacuum and single crystals), we unveil another process capable of converting catechols into ROSs, namely an intramolecular redox reaction catalysed by a Cu surface. We focus on a tri-catechol, the hexahydroxytriphenylene molecule, and show that this antioxidant is thereby transformed into a semiquinone, as an intermediate product, and then into an even stronger oxidant, a quinone, as final product. We argue that the transformations occur via two intramolecular redox reactions: since the Cu surface cannot oxidise the molecules, the starting catechol and the semiquinone forms each are, at the same time, self-oxidised and self-reduced. Thanks to these reactions, the quinone and semiquinone are able to interact with the substrate by readily accepting electrons donated by the substrate. Our combined experimental surface science and ab initio analysis highlights the key role played by metal nanoparticles in the development of degenerative diseases.

The Fenton or Haber-Weiss reactions consist of redox processes involving metal cations, as illustrated in Fig. S1 in the case of hydrogen peroxide and Cu cations.
The mechanism of oxidative DNA damage induced by catechols has been investigated in the literature (see, e.g. Ref. 1). In this mechanism, a catechol is oxidised into a semiquinone and a quinone with the generation of H2O2. In this chemical pathway, Cu 2+ are reduced into Cu + , the latter contributing to the oxidative damage of DNA (Fig. S2). This mechanism is very different from Haber-Weiss or Fenton reactions, because it requires the reduction of Cu 2+ into Cu + while Haber-Weiss or Fenton are based on the oxidation of Cu + into Cu 2+ . This mechanism is a priori impossible in the case of Cu nanoparticles because copper atoms are in the Cu(0) form, which cannot be reduced.!  Fig. 1 of the main text, the dehydrogenation of the HHTP molecule may occur in a way that breaks the three fold symmetry of the molecule, but still altering all three starting catechol groups. Such a situation is depicted in Fig. S3. In such a case, the charge reorganisation within the semiquinone gives it a monoradical character, and not a triradical character as in the case of Fig. 1 of the main text.

Monoradical form of the semiquinone
One may wish to seek for STM evidences of one form or the other (Fig. 1 of the main text versus We first tested several possible binding sites for the semiquinone form of HHTP on Cu(111) in a (5×5) Cu(111) unit cell. In these different phases, the center of the central C ring of the molecule lies: a. on bridge sites of Cu(111) (Fig. S4a), b. on hollow fcc sites of Cu(111) (Fig. S4b), c. on hollow hcp sites of Cu(111) (Fig. S4c), d. directly atop a Cu atom (Fig. S4d).
The difference in total energy of these configurations with respect to the most stable configuration We also performed simulations of the STM images for the two enantiomers of the semiquinone forms, and find no obvious difference (Fig. S5). [112] Electron diffraction data for the fully dehydrogenated molecules ⟨ 112 ⟩ Figure S9 shows a RHEED pattern acquired for fully-dehydrogenated HHTP molecules on Cu(111). In addition to the Cu-related streaks, which were present before molecule deposition, a pattern of streaks with a ×1/5 spacing compared to the spacing between Cu streaks is observed.
The streaks are labeled according to their diffraction order with respect to the molecular lattice (gray numbers in parenthesis) or with respect to the copper lattice (orange numbers in parenthesis).!  which is consistent with a half-dehydrogenation of the molecules (Table S1). The binding energy of the other components, corresponding to C atoms not bound with O atoms, are hardly affected by the kind of fit (with altogether three or four components), see Table S1.

Procedure for the analysis of the XPS data
XPS data were fit with a code written for Igor (developed by Francesco Bruno, ALOISA beamline, Laboratorio TASC, CNR-IOM).
Additional O 1s core level spectra  The experimental data (black curves) are fit (red curve) with several components, each corresponding to chemically inequivalent O atoms.
After having deposited a multilayer of HHTP molecules onto Cu(111) at room temperature, we increased the sample temperature step-by-step, and acquired O 1s core level spectra at each step ( Fig. S11). Please note that the spectrum measured here at 373 K has a very intense high binding energy component, much stronger than the one observed in Fig. 2 of the main text for a room temperature deposit. This is not a contradiction, since the spectrum measured here (Fig. S11) concerns a multilayer deposit of HHTP molecules, where most molecules are fully hydrogenated (see analysis of Fig. S12 below), while Fig. 2 concerns a sub-monolayer deposit of HHTP molecules, where all molecules are already half-dehydrogenated at room temperature.
In Fig. S11, we observe a progressive decrease of the intensity of the high binding energy component, starting from 450 K, until it vanishes (530 K). At least a significant fraction of the decrease of the intensity versus temperature in this binding energy range arrises from the desorption of molecules not directly in contact with the substrate. More directly informative concerning the chemical modification of the molecules in contact with Cu(111), the lower binding energy peak increases in intensity with temperature (starting from 450 K) and shifts to lower binding energy.
Eventually, at the highest temperatures, the absence of signal points towards a desorption or decomposition of the molecules that were in contact with Cu(111).
In Fig. S12, we compare the O 1s core level spectra for HHTP molecules deposited on Cu (111) Figure S13a shows two spectra measured at the K edge of C, with the polarisation of the electric field parallel (s polarisation) and perpendicular (p polarisation) to the surface. Several peaks/features are observed.
A first comparison to previously published measurements on simpler related compounds (benzene, phenols and quinones 5-7 ) allows assigning part of these peaks/features. The shoulder at 284.9 eV and the peak at 288.9 eV observed with the p polarisation are also found in benzene, where they have been ascribed to transitions from the C 1s core level to the first, second and third excited π* states (Fig. S13b) We suggest that the four latter correspond to chemically inequivalent C atoms ( The former peak (283.7 eV) presumably relates to the molecular charge distribution, being reminiscent of observations in p-benzoquinone, which were interpreted as a signature of a transition to a low-energy molecular C=C-C-O hybrid π* orbital. 5 Measurements with s polarisation reveal broad features, centred at 292 eV, 296 eV and 303 eV, that are characteristic of electronic transitions to states, 8 which are located in the plane of the rings and are usually used to determine the orientation of the molecules. The high intensity of these features, together with the non-vanishing π transitions, are consistent with the bowl-shape molecular geometry, with the center of the molecules being parallel to the surface.
In any case, it is noteworthy that we observe a strong anisotropy in the spectra when changing light polarisation, a clear signature of an almost"planar" bonding on the surface. 8,9 σ * Structure of the starting HHTP molecule in the gas phase  Figure S14 shows the DFT-optimised structure of the HHTP molecule in the absence of a substrate (gas phase). The C-C and C-O bond lengths are indicated on the figure. In the central ring, the bond lengths are similar and globally close to the 140 pm value of benzene, pointing towards an aromatic character of the central ring. This also seems to be the case in the peripheral rings, although the bond length dispersion is slightly larger there.!