Green Synthesis and Modification of RuO2 Materials for the Oxygen Evolution Reaction

Ion exchange method as a green synthesis route is proposed to prepare hydrous ruthenium oxide nanoparticles (H-RuO2). Calcination of H-RuO2 at 350°C resulted in the crystalline rutile RuO2 nanoparticles (C-RuO2). Treatment of H-RuO2 with 20 vol% ammonium hydroxide solution under microwave irradiation and calcination at 350°C resulted in a highly electrocatalytic active crystalline RuO2 nanoparticles (A-C-RuO2). Electrocatalytic performances of H-RuO2, C-RuO2 and A-C-RuO2 for the oxygen evolution reaction in 0.50 mol L−1 H2SO4 medium are evaluated and compared. Improved performances towards the oxygen evolution reaction are observed for A-C-RuO2 when compared to C-RuO2. Based on XRD, TEM, XPS and Raman characterizations performed on all the specimens, it is deduced that the physicochemical properties (crystallinity, mean crystallite size, level of hydrous rutile content) are varied for A-C-RuO2 when compared to C-RuO2. Structure-property correlation has been established to describe the higher electrocatalytic activity of A-C-RuO2.


INTRODUCTION
Renewable energy sources such as hydroelectric, wind, solar and tidal powers have been improving constantly, despite these energies have supplied instabilities (International Agency of Energy, 2019; Shariatzadeh et al., 2015). Electrochemical energy storage and conversion technologies are therefore gaining importance due to the growing electric energy demand without impacting the greenhouse gas emission (Badwal et al., 2015;Carmo and Stolten, 2019). Electrodes (anode and cathode) constitute the important aspect of all electrochemical technological processes (water electrolysis, fuel-cell, redox batteries, fuel production from CO 2 , etc.). More developmental work on electrode materials, also known as electrocatalysts, is monumental because the electrodes determine the efficacy, feasibility, cost-effectiveness and durability of electrochemical processes (Walsh, 2019). In particular, the oxygen evolution reaction (OER) is a challenging reaction with high activation overpotential and high potential conditions, leading to strongly oxidizing operating conditions for the anode materials. Many different materials have been investigated for this reaction (Wang et al., 2018;Laha et al., 2019;Retuerto et al., 2019;Chen et al., 2020).
In this context, ruthenium oxide (RuO 2 ) has become a very important material owing to its potency as supercapacitor (Zheng et al., 1995;Liu et al., 1997;Over, 2012) and as electrocatalyst for the OER (Trasatti, 2000;Carmo et al., 2013). RuO 2 is a versatile oxide material used in several electrochemical processes, in the view of its high thermal and chemical stabilities, metallic type conductivity (i.e., low resistivity), highly reversible redox property and long-life cycle. Hence, RuO 2 coated materials were used as anodes in many electrochemical industrial processes like cathodic protection, metal recovery, photoelectrochemical water oxidation, electroplating, electrowinning, electroflotation, electrosynthesis, cathodic protection, etc. (Cardarelli et al., 1998;Trasatti, 1999;Bock et al., 2000;Shibli et al., 2004;Katsaounis et al., 2007). Importantly, RuO 2 exhibits as a corrosion-resistant electrode at low overpotentials for the OER in water electrolysis cells. The electrocatalytic activity, selectivity and stability of rutiletype RuO 2 catalysts for the water splitting reaction depend on the surface and bulk structure of the material, and further on the synthetic method as well as on the synthesis precursors (Pelegrino et al., 2002).
Microwave assisted "instant method" (Devadas et al., 2011;Audichon et al., 2017a) and microwave assisted "polyol method" (Audichon et al., 2017b) were reported by our group for rapid synthesis of RuO 2 . Synthesized RuO 2 was further investigated for its structure to tune the electrocatalytic activity towards the water electrolysis reaction or the capacitance for capacitor application (Devadas et al., 2011;Audichon et al., 2017a;Audichon et al., 2017b). In the case of the "instant method," it was observed that an impurity phase containing lithium ruthenium oxide or lithium carbonate was formed in the synthesis process (Devadas et al., 2011;Audichon et al., 2017a). Moreover, the use of ethanol-water mixture was needed to centrifuge and to separate precipitant formed. In the case of the polyol method, the reaction was carried out in ethylene glycol as organic solvent, with formation of glycolate and other organic compounds as by-products Audichon et al., 2017b).
In the present study, a simple, effective and economic synthesis route for the large-scale synthesis of RuO 2 nanoparticles is proposed based on ion-exchange method. In this ion-exchange method, the principles of green chemistry (time, energy and atom savings, no organic solvent, no organic precursor, safe and environmental-benign process) are being realized for the synthesis of materials (Anastas and Eghbali, 2010). The possibility to improve the electrocatalytic activity of RuO 2 was also investigated by treating under microwave irradiation the ion-exchange derived product with ammonium hydroxide. All the specimens were evaluated for the OER in acidic medium; their electrocatalytic activity was analyzed and compared in light of their physicochemical characteristics (compositions, structure and crystallinity).

Materials
All experiments were carried out using ultrapure water (18.2 MΩ cm) obtained from a Milli-Q Millipore system. The anion exchange resin was obtained from Amberlite IRA910 in the chloride form (15-60 mesh, ion exchange capacity of 3.8 meq/g dry weight). Ruthenium chloride (RuCl 3 .xH 2 O, 99.9% purity, Ru 38 wt%) and ammonium hydroxide (NH 4 OH, 28-30%, ACS reagent grade) were obtained from Alfa Aesar and Sigma Aldrich, respectively. Nafion solution (5 wt% in aliphatic alcohols) from Aldrich was used as received. All the electrochemical measurements were carried out with the electrolyte prepared from suprapur H 2 SO 4 obtained from Merck and electrolyte was de-aerated by bubbling nitrogen of U quality from l'Air Liquide.
Synthesis of RuO 2 Nanoparticles Using Anion Exchanger 5 g of as received anion exchange resin were washed using ultrapure water and transformed into hydroxide form by agitating in NaOH solution (2.0 mol L −1 , 100 ml) for 18 h followed by repeated washing with ultrapure water till the pH of washing water becomes neutral. Hydroxide form of anion exchanger was then closely packed in a column and an aqueous solution containing RuCl 3 (0.010 mol L −1 , 50 ml) was passed through the column with flow of 1 ml/min. The contents at the end of the column were collected in a conical flask. A part of the contents in the conical flask was dried at 80°C to obtain hydrous ruthenium oxide nanoparticles (H-RuO 2 ). A part of H-RuO 2 was put aside and a part was calcined at 350°C (heating rate of 5°C/min under air atmosphere) to obtain the sample denoted as C-RuO 2 . The remaining part of the contents in conical flask was transferred to a microwave reactor chamber and treated under microwave irradiation (impulsion power of 400 W) with 20 vol% ammonium hydroxide solution in a microwave oven MARS 5 (CEM Corporation). The reactor chamber in the microwave was attached to a water condenser in order to avoid water evaporation and temperature was maintained at 60°C. After 1 h of microwave irradiation, the mixture in reactor was centrifuged. Finally, the obtained product was calcined at 350°C (heating rate of 5°C/min under air atmosphere) and the obtained sample was denoted as A-C-RuO 2 . The designed synthesis process is shown in Figure 1.

Physicochemical Characterizations
All the XRD Patterns were recorded on a Brucker D8-Advance X-ray diffractometer with a copper anode (K α 1.5405 Å) powered at 40 kV and 40 mA. The in situ XRD analysis on H-RuO2 was performed by heating the sample from room temperature to 350°C under air atmosphere. Kanthal holder was used in order to perform the heat treatment with a heating rate of 10°C min-1 between each set temperature, where the XRD patterns were recorded. XRD measurements were carried out in the 2θ range from 10°to 70°in step mode of 0.065°and a fixed acquisition time of 5 s/step. The diffraction peaks were fitted with Lorentzian-Gaussian functions using Fityk software (Fityk, 2010) and the lattice parameters of C-RuO 2 were calculated using the linear least square method, considering the rutile structure of RuO 2 .
Raman spectra (Stokes spectra) were obtained at room temperature using a HR UV 800 confocal scanning spectrometer (Horiba Jobin Yvon) equipped with a Peltiercooled charge coupled device (1,152 × 298 pixels). The Raman scattering was excited using a 632.8 nm excitation wavelength supplied by an internal, air-cooled, He-Ne laser (Melles Griot) through an Olympus high-stability BXFM microscope coupled confocally. Laser power delivered at the specimen (ca. 20 mW Frontiers in Energy Research | www.frontiersin.org October 2020 | Volume 8 | Article 571704 without filtering) could be monitored via a filter wheel with optical densities of 0.3, 0.6, and 1. The backscattered radiation was collected using the same microscope with a ×100 Olympus objective (numerical aperture 0.9). An 1,800 grooves mm -1 holographic grating was used, and the confocal hole aperture was 300 mm, resulting in an instrumental bandwidth of 6.3 cm-1 and a spectral resolution of 0.5 cm-1 in the 50-1,000 cm -1 Raman shift range. The spectrometer was calibrated using the G25 phonon of diamond Si (Fd-3m) resulting in an accuracy of less than 0.5 cm -1 . LabSPEC v.5 software was used to perform both acquisition and data processing. H-RuO 2 , C-RuO 2 and A-C-RuO 2 nanoparticle morphologies were examined using a transmission electron microscope (JEOL 2100, UHR, 200 kV) and the qualitative elemental analysis of the specimens was made by EDS analyzer equipped with a LaB 6 filament. The mean particle size and size distribution were determined by measuring the féret's diameter of 250 isolated particles using ImageJ free software (Rasband, 2009). X-ray photoelectron spectroscopy (XPS) core level spectra were collected using a VG ESCALAB three MKII spectrometer using a monochromatized Mg K α radiation (1,253.6 eV). The source was operated at 300 W (15 kV and 20 mA). The data were collected at room temperature and the operating pressure in the analysis chamber was set below 8 × 10 -9 Torr. The core level spectra were recorded with a resolution of 50 meV and sample analysis covered a surface of 2 × 3 mm.

Electrochemical Measurements
The electrocatalytic activity for the OER was evaluated in a standard three-electrode electrochemical cell fitted with a 3 cm 2 glassy carbon plate as auxiliary electrode and a reversible hydrogen electrode (RHE) as reference electrode, which was connected through a Luggin capillary to the working electrode compartment. The anode was made from an ink consisting in a suspension of the prepared materials (10 mg) in a mixture of 1,000 µl ultrapure water and 100 µl Nafion solution. The working electrode was prepared by dipping 5 µl of the ink on a polished glassy-carbon electrode (0.071 cm 2 geometric surface area), leading to a catalyst loading of 45.5 µg catalyst on the electrode. The applied potential controlled by a VoltaLab PGZ 402 potentiostat and the outputs were monitored by a computer. The electrolyte used was a 0.50 mol L -1 H 2 SO 4 solution in ultrapure water saturated with nitrogen for 15 min prior to the measurements. Nitrogen atmosphere was maintained in the electrochemical cell throughout the measurements. The linear scan voltammograms were recorded at a scan rate of 0.010 V s -1 . Chronoamperometric curves were recorded at +1.600 V vs RHE for 60 min, under the similar experimental conditions of linear voltammograms.

Green Synthesis of Hydrous RuO 2
Ion-exchange method adopted in this work leans on a soft hydrolysis route with aid of hydroxyl anions (OH − ) from anion exchanger packed in column. Ion exchange method has been used for the batch synthesis of the colloidal precursors of hydroxide and oxide materials of Fe, Al, Cr, Zr and Ti with aid of anion exchangers (Vertegel et al., 1995;Tretyakov et al., 1997;Shafran et al., 2005). To the best of our knowledge the synthesis of ruthenium hydroxide/oxide materials using anion exchanger enabling hydrolysis of a ruthenium salt is not reported. Dynamic/column mode and static/batch mode of operations were used in ion exchange synthesis route (Pelletier et al., 2000). In the column mode, a solution of a metal ion precursor is forced through a column containing a densely packed ion exchange resin. The formed metal hydroxide leads to a stable colloidal phase, which can stabilize the concentration gradient along the column due to variation of the degree of exchange. The slow release of hydroxyl ions from the surface of the resin gives kinetic control of the hydrolysis reaction when compared to hydrolysis of metal ions in conventional solution-based methods. In this study, the contact of RuCl 3 solution with OH form of Amberlite IRA910 anion exchanger leads to the formation of Ru(OH) 3 precipitate. Van Muylder and Pourbaix (1963) and Music et al. (Music et al., 2002) proposed that Ru(OH) 3 was not stable in the presence of air and oxidized into hydrous RuO 2 . Dmowski et al. (Dmowski et al., 2002) examined the hydrous RuO 2 by atomic pair density function (PDF) method and pointed out that hydrous RuO 2 is a composite of anhydrous rutile-like RuO 2 nanocrystals dispersed by boundaries of structural water associated with Ru-O bonds.
The precipitate obtained in the present study is free of any other impurities and no separation step is needed. Moreover, the spent anion exchangers can be reused after treatment with NaOH solution, which suggests the elimination of waste streams during the synthesis process.
Physicochemical Characterization of Hydrous and Crystalline RuO 2 Figure 2 illustrates the in-situ XRD patterns of H-RuO 2 calcined from room temperature to 350°C. As expected, only broad peaks are present, and no characteristic peak related to the crystalline rutile structure is identified for H-RuO 2 specimen. It is observed that H-RuO 2 specimen did not show any characteristic peak of crystalline RuO 2 even after calcination at 200°C, whereas the rutile type crystalline RuO 2 phase is observed at 350°C (JCPDS 88-0322). Similar behavior was observed for the samples prepared by the "instant method" with Li 2 CO 3 as hydrolyzing agent and crystallinity was obtained only after reaching the calcination temperature of 350°C (Devadas et al., 2011;Audichon et al., 2017a). Also, the diffractograms obtained on RuO 2 prepared by the "instant method" showed impurity diffraction peaks of lithium ruthenium oxide (Devadas et al., 2011;Audichon et al., 2017a) and/or lithium carbonate (Audichon et al., 2017a). In contrary, the present ion-exchange method (no excess hydrolyzing agents, no purification steps) resulted in the pure rutile phase RuO 2 structure. It was reported that well crystalline RuO 2 will be obtained only after calcination at high temperatures (Cruz et al., 2011;Tsuji et al., 2011). However, calcination of RuO 2 above 350°C may decrease its electrocatalytic activity towards the OER (Devadas et al., 2011;Audichon et al., 2017a). Hence, in the present study, C-RuO 2 was obtained after calcination of H-RuO 2 at 350°C for 1 h. Based on XRD investigation of calcined RuO 2 at different temperatures by Jirkovsky et al. (2006), it was mentioned that broadening of the diffraction peaks was directly related to the coherent domain size of crystallites for calcination temperatures lower than 500°C and internal strain/stress effects on broadening of diffracted peaks appeared only for calcination temperatures higher than 600°C. In this study, the calcination temperature is set at 350°C and the mean crystallite size determined using the Scherrer equation is representative of the average coherent domain size. The C-RuO 2 is found to have a mean crystallite size close to 23 nm, which is in comparison with the earlier reports (Devadas et al., 2011;Audichon et al., 2017a). The unit cell parameters are calculated using the linear least square method and results indicate that the C-RuO 2 crystal lattice has the unit cell values of a b 0.4483 nm and c 0.3102 nm. The TEM images for the H-RuO 2 show the formation of particles with homogeneous size distribution ( Figure 3A,B). The inset of Figure 3A depicts the size distribution histogram for the H-RuO 2 sample. The average particle size is found to be lower than 2.0 nm. The TEM micrographs of C-RuO 2 exemplify the growth/agglomeration of nanoparticles ( Figure 3C,D) with well crystallization and particle sizes in the range from 10 to 30 nm, which agrees with results obtained on RuO 2 nanostructures prepared by other synthesis methods (Natarajan et al., 2013;Audichon et al., 2014). Figure 4, Raman spectra of H-RuO 2 and C-RuO 2 specimens exhibit three bands; frequencies are obtained accurately by fitting the bands with a Gaussian -Lorentzian function. In the case of H-RuO 2 , the E g vibration mode is centered at 507 cm −1 ; A 1g and B 2g vibration modes are observed at 652 cm −1 and 685 cm −1 , respectively. According to Raman-active phonon frequencies predicted by group theory, the single crystal rutile RuO 2 is characterized by Raman bands located at 528, 646 and 716 cm −1 for E g , A 1g , and B 2g vibration modes, respectively (Mar et al., 1995). Thus, Raman spectra of H-RuO 2 leads to the conclusion that when RuCl 3 solution passes through the anion exchange column, the rutile RuO 2 phase is formed rather than an amorphous phase, conversely to the conclusion drawn from the XRD pattern. Also, Raman studies on H-RuO 2 agree with Dmowski et al. (Dmowski et al., 2002) studies suggesting that hydrous RuO 2 is a composite of anhydrous rutile-like RuO 2 nanocrystals dispersed by boundaries of structural water. C-RuO 2 is characterized by the Raman bands centered at 512, 630 and 690 cm −1 for E g , A 1g , and B 2g vibration modes, respectively. The Raman features indicate a decrease of the linewidth for C-RuO 2 which can be attributed to well crystallization and structural ordering due to calcination, whereas the red shift of the peak locations compared to those of the single crystal rutile type RuO 2 is attributed to the nanoscopic nature of C-RuO 2 material (Chen et al., 2004). The observed peak positions of H-RuO 2 and C-RuO 2 are significantly different from those of characteristic bands of crystalline-RuO 2 , which is due to the influence of structural water bounded to nanocrystals on the Raman spectrum vibrational frequencies (Mo et al., 2001). XPS core level spectra for the C-RuO 2 sample is presented in Figure 5 and Ru 3d core level spectrum resembles to that of hydrous RuO 2 •xH 2 O material reported by Morgan (Mo et al., 2001). Based on the XPS data, different oxidation states for Ru are identified in C-RuO 2 specimen and these oxidation states are Ru IV (Ru 3d 5/2 peaks located at 280.6 eV with the corresponding satellites at 282.5 eV) and Ru VI−VIII (Ru 3d 5/2 peak located at 284.2 eV). The attribution of both XPS signals at 280.6 and 282.5 eV to Ru IV is based on results widely discussed in the literature (Chan et al., 1997;Kim et al., 1997;Mo et al., 2001;Over et al., 2002;Chang and Hu, 2004;Foelske et al., 2006;Morgan, 2015;Walsh, 2019), although the origin of the signal close to 282.5 eV differs depending on the authors. Recently, Näslund et al. (2014) explained that "the strong Coulomb interaction between valence electrons and the core hole produced in the photoionization process" was responsible of this peak (final-state screening effect). The XPS signal observed at 284.2 eV corresponds to RuO 4 surface species (Chang and Hu, 2004). These RuO 4 species are attributed to highly oxidized surface ruthenium, namely Ru VI−VIII , since RuO 4 is a volatile compound. The O 1s core level spectrum ( Figure 5B) presents three different XPS signals attributed to physisorbed water (H 2 O), oxygen involved in RuO 2 lattice as hydrous ruthenium oxide (OH − ) and oxygen in RuO 2 lattice (O 2-). Table 1 presents the atomic ratios calculated from XPS spectra for C-RuO 2 . Figure 6 presents the characterization data for the A-C-RuO 2 specimen. It can be seen from comparison of Figures 3, 6 that structural differences appeared between C-RuO 2 and A-C-RuO 2 . Based on observations of X-ray diffraction ( Figure 6A) and Raman scattering measurements ( Figure 6B), increased linewidth broadenings and shifts in the peak positions are noticed for A-C-RuO 2 when compared to XRD pattern/Raman spectrum of C-RuO 2 . The shift to higher angles of XRD peaks ( Figure 6A), the red shift of Raman peaks ( Figure 6B) and the increase of the linewidth broadening for A-C-RuO 2 can be correlated to changes in crystallinity, local disorder and mean crystallite size. The unit cell parameters for A-C-RuO 2 have values of a b 0.4478 nm and c 0.3097 nm. Contraction of lattice parameters for A-C-RuO 2 when compared to C-RuO 2 is correlated to the decrease of mean nanocrystallite size in A-C-  RuO 2 specimen (Leontyev et al., 2014). The mean crystallite size determined using the Scherrer equation for A-C-RuO 2 sample was found to be close to 11 nm, which is twofold smaller than the mean crystallite size of C-RuO 2 . In agreement with XRD results, the micrographs of the A-C-RuO 2 sample ( Figure 6D) represents the growth of nanoparticles with sizes similar to that of C-RuO 2 (i.e. in the range from 10 to 30 nm) ( Figure 3D). The smaller mean crystallite size measured on A-C-RuO 2 specimen than that measured on C-RuO 2 sample does not involve a smaller mean particle size, it only implies that the A-C-RuO 2 sample exhibits polycrystalline nanoparticles ( Figure 6E). The comparison of XPS data for C-RuO 2 ( Figure 5) and A-C-RuO 2 ( Figure 6C) evidences another anomaly in their microstructure. An increase of highly oxidized surface ruthenium species, an increase of oxygen in RuO 2 lattice and a decrease of the amount of hydrous ruthenium oxide (OH − ) is observed for A-C-RuO 2 . For comparison, the atomic ratios obtained from XPS analysis for C-RuO 2 and A-C-RuO 2 are presented in Table 1. Thus, ammonium hydroxide treatment under microwave irradiation on hydrous RuO 2 has influenced to limit the growth of nanocrystals of H-RuO 2 during calcination process, thereby resulting in lower crystallite sizes/lattice parameters of A-C-RuO 2 when compared to C-RuO 2 (a blank test of H-RuO 2 treatment under microwave irradiation without addition of  ammonium hydroxide was performed and the RuO 2 structure after calcination at 350°C was the same as that of C-RuO 2 ).

Electrocatalytic Activity
The electrocatalytic activity of the H-RuO 2 and C-RuO 2 samples was evaluated by recording the current-potential curves of the OER in 0.50 mol L −1 H 2 SO 4 electrolyte at the quasi-stationary potential scan rate of 0.010 V s −1 . It is very difficult to give an accurate evaluation of the real surface area of the catalyst due to the absence of faradaic process in the supporting electrolyte and to the non-proportional increase of the capacitance with the real surface area (Devadas et al., 2011). Moreover, it was previously observed that the capacitive current measured could vary independently on the surface area for RuO 2 materials with particle sizes from two to almost 100 nm (Devadas et al., 2011;Audichon et al., 2017a). Hence, considering the large difference of OH species present at the surface of the C-RuO 2 and A-C-RuO 2 samples, as shown by XPS measurements, the determination of capacitance may not reflect accurately the difference of electrochemical active surface area. The determination of the BET porosity and surface area suffers the same problem. The confined water amount in the materials has a huge influence on adsorption of N 2 on internal RuO 2 domains (Yoshida et al., 2013) However, according to XPS measurements, C-RuO 2 and A-C-RuO 2 catalysts display significantly different amounts of water, which will therefore lead to inaccuracies in the determination of porosity and surface area by BET, and even to discrepancies in the evaluation of the active surface areas by both these different methods (Siviglia et al., 1983). Hence, because all electrode are loaded with the same mass of catalytic material, the mass specific activity is used for the representation of polarization curves. As shown in Figure 7, H-RuO 2 is found to exhibit the oxygen evolution with an onset potential at ca. +1.40 V vs RHE. After +1.500 V vs RHE, a steep decrease of current values is noticed for H-RuO 2 . Such behavior was also observed with hydrous RuO 2 nanomaterials synthesized in our previous investigations (Devadas et al., 2011;Audichon et al., 2017a;Audichon et al., 2017b), which is due to structural instability at higher potentials. C-RuO 2 shows the onset potential of oxygen evolution at ca. +1.42 V vs RHE and is found to be active at higher potentials ( Figure 7). The chronoamperometric study of the OER on C-RuO 2 at E +1.600 V vs RHE is shown in Figure 7 inset and a stable mass specific activity plateau of 22 A g −1 is reached after 10 min. These results infer the stability of C-RuO 2 for the OER in 0.50 mol L −1 H 2 SO 4 solution for the applied time period in comparison with H-RuO 2 material. A-C-RuO 2 exhibits improved performance towards the OER when compared to C-RuO 2 ( Figure 8A). At E +1.600 V vs RHE,  the mass specific activities of C-RuO 2 and A-C-RuO 2 were found to be 41 and 57 A g −1 , respectively. Table 2 compares the mass activities for these two catalysts with recent results in literature. C-RuO 2 and A-C-RuO 2 catalysts match well with rutile RuO 2 nanoparticles (Laha et al., 2019). The addition of transition group metals helps enhancing the catalytic activity Wu et al., 2020). Note that such binary or ternary materials could be achieved with the ion exchange method on resin discussed in the present work. Figure 8B illustrates the Tafel plots for the OER on C-RuO 2 and A-C-RuO 2 . The curves exhibit linear and parallel parts at lower potentials or overpotentials with slopes of ca. 69 mV decade −1 . The Tafel slope of 60 mV decade −1 was reported for (100) and (110) RuO 2 surfaces (Castelli et al., 1986;Stoerzinger et al., 2014). This indicates that the mechanism of water splitting is likely the same on both catalysts at low overpotentials. According to Antolini (Antolini, 2014), a Tafel slope of 60 mV decade −1 corresponds to the following mechanism, with step 1 as rate determining step: S-OH * ads → S-OH ads (2) either S-OH ads → S-O ads + H + + e − (3) and/or S-OH ads + S-OH ads → S-O ads + H 2 O + S (4) 2S-O ads → O 2(g) + 2S ( 5) where, S represents an active site, S-OH p ads and S−OH ads possess the same chemical structure but different energy states.
At higher overpotentials, the Tafel slope values for the OER on C-RuO 2 and A-C-RuO 2 are found to be ca. 166 and ca. 190 mV decade −1 , respectively. Based on the mechanistic studies for the OER on RuO 2 , IrO 2 and RuO 2 @IrO 2 core-shell nanocatalysts, it was reported that different Tafel slopes at high overpotentials were characteristic of the electrocatalytic material, whereas similar Tafel slopes (60 mV decade −1 for both RuO 2 and IrO 2 ) were found at low overpotentials (Ma et al., 2018). Thus, in the present study, varying Tafel slope value of C-RuO 2 and A-C-RuO 2 at high overpotentials corroborates the differences in their physicochemical characteristics.
At last, chronoamperometric curves of OER on C-RuO 2 and A-C-RuO 2 at E +1.600 V vs RHE are compared in Figure 8 (inset). A higher mass activity for A-C-RuO 2 when compared to C-RuO 2 was recorded, confirming the linear scan voltammetry measurements. The mass specific activity reaches a plateau of 29 A g −1 (activity gain of 32% compared with C-RuO 2 ) which also demonstrates the high stability of A-C-RuO 2 for the OER in 0.50 mol L −1 H 2 SO 4 solution for the applied time period.

DISCUSSION
DFT simulations have pointed out that under electrochemical conditions for the OER, two types of adsorption sites exist for RuO 2 electrocatalyst (Rossmeisl et al., 2007;Hansen et al., 2010). The first one forming a bridge between two adjacent Ru atoms along the [001] direction is electrocatalytically inactive. The second one denoted as coordinatively unsaturated site, involving one of the Ru-O bonds, is found to be electrocatalytically active (Rossmeisl et al., 2007;Hansen et al., 2010). DFT calculations have also identified a peroxo group formed by two adjacent octahedral oxygen sites that favors the rate determining intermediates for the OER (Rossmeisl et al., 2007;Hansen et al., 2010) and such peroxo group formation can be minimized by doping of RuO 2 (Petrykin et al., 2010). Ma et al. proposed that OER rate on RuO 2 was limited by OOH formation on the O ads and OER activity of RuO 2 @IrO 2 was limited by O ads (Ma et al., 2018). In the present study, A-C-RuO 2 material appears less ordered and possibly a part of the hydroxyl group present in the catalyst was converted into oxygen group that are coordinated to Ru atom. Plausibly, nitrogen doping in RuO 2 can be anticipated and its effect may be the microstructural changes of A-C-RuO 2 specimen, thereby exhibiting the enhanced electrocatalytic activity for the OER when compared to C-RuO 2 . However, EDX and XPS analyses of A-C-RuO 2 did not show any evidence of the presence of nitrogen either because of its low concentration or of its absence. Indeed, heat treatment at 350°C under pure N 2 atmosphere of as-prepared RuO 2 from the "instant method" also induced changes in crystallinity, crystallite size and morphology of the final material compared with the same materials heat treated under air, while nitrogen was not detected (Audichon et al., 2017b). The treatment by ammonium hydroxide under microwave irradiation of the H-RuO 2 from the anion exchange method also affects the microstructure of the final material, which highlights the importance of the role of nitrogen atoms on RuO 2 microstructure even if not detected.
The increase of active surface area could also be proposed to explain the improved activity of the A-C-RuO 2 catalyst in comparison to that of C-RuO 2 catalyst. But if XRD measurements indicated a smaller mean crystallite size for A-C-RuO 2 than for C-RuO 2 , TEM characterizations showed same particle size as for C-RuO 2 and polycrystallinity of A-C-RuO 2 nanoparticles. Therefore, both catalysts should display same active surface area. The hydroxyl content in the A-C-RuO 2 compound has decreased and that of lattice oxygen has increased as shown by XPS measurements, but the material is made less ordered than C-RuO 2 , as evidenced by Raman measurements. In the case of RuO 2 material, it was highlighted that the electronic conductivity is supported by the rutile-like nanocrystals, while proton conduction is facilitated by the structural water along the RuO 2 grain boundaries (Dmowski et al., 2002). Based on the electrocatalytic activity of RuO 2 for various applications, it was reported that the right combination of electronic and protonic conduction of RuO 2 is the most obvious parameter to tune the electrochemical activity of RuO 2 (Walsh, 2019). In another study, it was proposed that the better electrochemical activity for the OER required more interconnected RuO 6 octahedra (which is the structure of anhydrous rutile) as well as optimized structural water content (Trasatti and Lodi, 1981). Since the calcination temperatures of both samples are the same, the ammonium hydroxide treatment might lead to right balance between the number of available oxygen and proton sites (OH − species) on the catalytic surface of A-C-RuO 2 , also balancing the electronic and protonic conduction of the catalyst. Moreover, XPS measurements indicated that the Ru oxidation state was predominantly +IV, which makes the compound extremely electrochemically stable (Sugimoto et al., 2006). But finally, the role of nitrogen on A-C-RuO 2 specimen is still not well established and the physicochemical properties of nitrogen treated RuO 2 need to be further investigated for the better understanding of the electrocatalytic activity enhancement of ammonium hydroxide treated RuO 2 .

CONCLUSION
Anion exchanger assisted synthesis of hydrous RuO 2 nanocrystals (H-RuO 2 ) was reported. Ion-exchange method proposed in this study offers a promising way respecting most of the twelve green chemistry principles (Anastas and Warner, 2000) for synthesis of RuO 2 electrocatalysts. The crystalline RuO 2 was obtained by calcination of hydrous RuO 2 at 350°C (C-RuO 2 ), which is evidenced by XRD and Raman characterizations. All the calcined specimens in this work resulted in rutile structure of RuO 2 without any impurities. The ammonia treated H-RuO 2 under microwave irradiation (A-C-RuO 2 ) showed improved performances towards the OER when compared to C-RuO 2 . Tafel slope value of A-C-RuO 2 was found to be 60 mV/decade, which represents the hydroxyl radical adsorption on catalytic active sites as the rate-limiting step for the OER at lower overpotentials. XRD, TEM measurements indicated the decrease in crystallite sizes of A-C-RuO 2 (11 nm) when compared to C-RuO 2 (23 nm) specimen. XPS analysis of A-C-RuO 2 pointed out the increase of highly oxidized surface ruthenium species and decrease of the amount of hydrous ruthenium oxide. Raman spectral investigations of A-C-RuO 2 indicated the local structural disorder. Structure-property correlation was established to describe the higher electrocatalytic activity of ammonium hydroxide treated specimen (A-C-RuO 2 ) when compared to C-RuO 2 .
Overall, ion-exchange method as green synthesis route facilitates the formation of highly pure rutile phase RuO 2 electrocatalysts. Further improvement of RuO 2 electrocatalytic activity was realized by microwave assisted ammonium hydroxide treatment of ion-exchange derived hydrous RuO 2 . The approach proposed in this work is simple, environmentalfriendly and more cost effective than traditional ones for obtaining RuO 2 nanoparticles with the same activity. The ease to proceed the scale up to produce metal oxides and mixed oxides (RuIrO 2 , for example) could make this synthesis method technologically and industrially interesting.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.