Comparative study of catalytic activities among transition metal-doped IrO2 nanoparticles.

Catalytic activities of transition metal-doped IrO2 nanoparticles (TM-IrO2 NPs; TM = Cr, Mn, Fe, Co, or Ni) are compared for various oxidation reactions such as electrochemical oxygen evolution reaction (OER), gas-phase photo-oxidation of thiol function group, and CO oxidative conversion. Here, we discovered a series of TM-IrO2 catalysts have a common activity trend for these oxidation reactions, and their activities are closely related with modified electronic states of IrO2, strongly affected by the types of the transition metal across the periodic table. For all oxidation reactions, Cr- and Mn-IrO2 achieved the highest oxidation catalytic activity, and sequentially decreased activities were obtained with Fe, Co, and Ni doped IrO2. For instance, the highest OER activity was achieved by Cr or Mn doping exhibiting the smallest overpotential η = 275~230 mV at 10 mA/cm2, while Ni-IrO2 showed rather larger overpotential (η = 347 mV) even compared with non-doped IrO2 (η = 314 mV). Scanning transmission X-ray microscopy and high-resolution photoemission spectra of TM-IrO2 indicated dopant metals modified the Ir-O interaction and thus increasing oxygen vacancy defects in IrO2. Strongly positive correlation was observed between the catalytic activities and vacancy states. The amount of defect related signals was observed the most for Cr- or Mn-IrO2, less so for Fe- or Co-IrO2, and unnoted for Ni-IrO2 compared with bare IrO2. Based on these catalytic activities and surface spectroscopic analysis results, vacancy defects induced by doping in TM-IrO2 NPs are proposed to contribute to enhance the oxidation activities.

Iridium oxide (IrO 2 ) has been demonstrated as an active catalyst material in impressively wide ranges of applications for electrocatalyst [1][2][3][4] , catalyst [5][6][7][8] , photocatalyst, etc [9][10][11] . For instance, IrO 2 has excellent electrocatalytic activities for oxygen evolution reaction (OER) in batteries or water-electrolysis and considered as a benchmarking material to compare the catalytic activities of others [12][13][14] . Defects or even amorphous structure on the IrO 2 surface have been proposed as the active sites of electrochemical OER. In addition to the electrochemical oxidation reaction, IrO 2 nanoparticles have been recently reported to display photocatalyst capability, in particular, to split water under UV irradiation through the excitation from the O-p band to the Ir-d band 15,16 .
Along with mechanistic studies of IrO 2 catalyst 17,18 , attempts to enhance their catalytic activities have been tried via alloying or inserting dopants such as metals or anions into these nanoparticles (NPs) [19][20][21][22][23] . In IrO 2 , transition metal dopants can access various oxidation states which can disturb the bonding between Ir and O and consequentially modulate the electronic/chemical states of IrO 2 . Therefore, it can be hypothesized that these modified electronic states vary the intrinsic catalytic activity. Although the active sites of the IrO 2 catalysts can be different depending on the catalytic reactions due to the different chemicals and reaction pathways, similar strategies such as doping or alloying have been successfully demonstrated to improve activity of IrO 2 . This can be valid if there is a related active factor of various reactions that controls catalytic activity on the IrO 2 surfaces.
To understand the possible relation of the active species and catalytic activity of IrO 2 for various reactions, we inserted various transition metal ions (TM + ) into IrO 2 NPs to enhance their own catalytic activities and then systematically compared the catalytic activities of these transition metal-doped IrO 2 (TM-IrO 2 ) NPs for different types of oxidation reactions. In particular, the cost-effective transition metals such as Cr, Mn, Fe, Co, and Ni  24 . Therefore, the TM-IrO 2 NPs were characterized by using scanning electron microscopy (SEM), Raman spectroscopy, and scanning transmission X-ray microscopy (STXM) and the influence of the dopant types on the electronic structures was studied. Then, the catalytic activities of the TM-IrO 2 NPs were assessed for the OER in an aqueous solution by taking electrochemical measurements, for the photocatalytic oxidation of 2-aminothiophenol (2-ATP) in a gas phase media by performing high-resolution photoemission spectroscopy (HRPES), and for the thermal CO conversion to CO 2 by measuring mass spectroscopy. We observed there is a common catalytic activity trend for all these different oxidation reactions. Espeically, highly enhanced catalytic activity was achieved when Cr or Mn was doped in IrO 2 NPs in which modified electron states showed largest generation of vacancy defects proposing as a main contributor.

Results and Discussion
Characterization of transition metal doped IrO 2 nanoparticles. First, Raman spectra of bare IrO 2 NPs and the five different TM-IrO 2 NPs (see Fig. 1) were obtained and the major Raman shifts at ~561 (E g ), 723 (B 2g ), and 752 cm −1 (A 1g ) are well associated with typical rutile IrO 2 25 . Compared with the non-doped IrO 2 NPs, TM-IrO 2 NPs was confirmed to have no significant changes in these Raman shifts, except additional minor peaks. These minor peaks are considered as dopant-related signals because they were observed near a Raman shift similar to that of its metal oxide bonds (Cr 2 O 3 : 680, 612, MnO: 644, Fe 2 O 3 : 612, Co 2 O 3 : 651, and NiO: 569 cm −1 ) [26][27][28][29][30] . In addition, SEM images were also obtained to determine the surface morphology of IrO 2 and TM-IrO 2 NPs which have mostly similar round partice shapes and small protrusion on the surface. Scanning transmission electron microscope (STEM) images and energy-dispersive X-ray spectroscopy (EDS) mapping images shows Ir and doped transition metal were uniformly distributed over the nanoparticles (Supporting Information Fig. S1). Figure 2 shows the X-ray absorption spectroscopy (XAS) results stacked in STXM images of the IrO 2 samples, black regions. We focused on the O K-edge spectra to characterize the changes of the oxygen states as the result of the doping. Peaks at 530~535 eV are generally observed due to the excitation of electrons from the O 1 s state to the O 2p -metal d hybrid state; peaks at 540~545 eV, were by excitation from the O 1 s state to the unoccupied p state; and the peak at about 528.8 eV (denoted as e) was observed in the formation of an amorphous or defect structure on the surface 31 . Specifically, non-doped bare IrO 2 NPs have four main peaks at 530.1, 532.9, 539.5, and 544 eV (denoted as a, b, c, and d, respectively), which are matched with those of typical rutile IrO 2 structures 31,32 . Notably, although all of the TM-IrO 2 NPs have O K-edge features at similar position, peak shape and relative intensity near 530 eV was dramatically modified depending on the types of the dopants. In detail, both of the Cr-IrO 2 and Mn-IrO 2 NPs had particularly low intensity of peak a, and the broaden peak around 530 eV due to strong appearance of an additional peak (peak e) at a photon energy of 529 eV (Fig. 2b,c). These differences with dopants can be understood by alteration of the hybridization between the O 2p and metal d orbitals. The pre-edge signal at ~529 eV is proposed to attribute to iridium vacancies or O 2p hole states by the presence of O − species in rutile IrO 2 31 . The hole states created by Cr or Mn doping would be expected to cause an increase of the resonance at peak e and consequent decrease at peak a. On the other hand, compared to the non-doped IrO 2 NP, Fe-IrO 2 and Co-IrO 2 NPs show changes near 529~530 eV but much less changes than Cr-IrO 2 or Mn-IrO 2 (Fig. 2d,e). In other words, decrease in the intensity of peak a was smaller and increase of peak e was also smaller in the case of Fe or Co doped ones compared with Cr or Mn doped ones. Lastly, the peak shapes of Ni-IrO 2 turns to be very similar to those of IrO 2 NPs. The O K-edge spectra revealed more generation of the vacancies states when Cr or Mn was incorporated into IrO 2 . Changes of the spectra was large with the transition metals on the left side of the periodic table and decrease with the order of Fe, Cu, and Ni across the periodic table. Note that the transition metal-oxygen bond formation energy levels are greater for those metals on the left side of the periodic table than for those on the right side, and such stronger bonds may have stabilized the vacancies in our IrO 2 NPs containing doped Cr or Mn 33 . Therefore, the result of O K-edge spectra can classify the dopants into the following three: (i) Cr and Mn which changes the O states the most, (ii) Fe and Co which changes in a milder amount, (iii) Ni which changes least.
Next, HRPES spectra of TM-IrO 2 NPs are compared in the O 1 s and Ir 4f regions (Supporting Information Fig. S2), having peaks centered at 529.7 eV and 61.7 eV, respectively, and changes of the spectra induced by new feature were observed indicating modification of the Ir-O interaction. The deconvolution of the O 1 s peaks of all the TM-IrO 2 NPs yielded two minor peaks: one at ~531.0 eV (denoted as O v ), which might be associated with the defective oxygen structure in IrO 2 , and the other at ~533 eV (denoted as TM-IrO x ; TM = Cr, Mn, Fe, Co, or Ni) 34,35 . Similarly, a new set of minor peaks generated in the Ir 4f spectra at a higher binding energy state, which may have been attributed to Ir 3+ . The Ir 3+ state has often been reported to be present in amorphous or anodized IrO x structures. Note that at both O 1 s and Ir 4f spectra, the intensities of the minor signals (i.e. O v and Ir 3+ ) were higher for Cr-or Mn-IrO 2 NPs than those for the other dopant-containing IrO 2 NPs, which provided a hint that these features are related to each other. These HRPES spectra indicate quite consistent dependence with those of the O K-edge XAS spectra.
Electrocatalytic OER activity of TM-IrO 2 NP. Electrocatalytic OER activities of TM-IrO 2 NPs were evaluated in the alkaline 1 M KOH electrolyte condition (Fig. 3), exhibiting distinctively different activities. Because the loaded catalyst amounts and the exposed area on the glassy carbon electrode were controlled to be same for the IrO 2 and TM-IrO 2 NPs, the trend of the mass activity (mA/mg) normalized by the catalyst loading was the same as that of the current density (mA/cm 2 ). Cr-IrO 2 and Mn-IrO 2 showed the greatest increases in OER activity, Fe-IrO 2 and Co-IrO 2 showed modest increases, and Ni-IrO 2 showed a rather decrease in activity, compared to the IrO 2 NPs, (Fig. 3(a)). At a current density of 10 mA/cm 2 , the over-potential (η) values of Cr-IrO 2 and Mn-IrO 2 were 267 mV and 279 mV, about 35 mV less than that of the IrO 2 NPs. It is notable the enhanced overpotentials of these TM-IrO 2 NPs are also comparable values with those of the high performing OER catalysts 36 (Supporting  Information Table S1). Meanwhile the η of Fe-IrO 2 and Co-IrO 2 were about 10 mV less than that of the IrO 2 NPs. Ni-IrO 2 required an η value 33 mV greater than that of the IrO 2 NPs to reach an OER activity of 10 mA/cm 2 ( Fig. 3(b)). Higher OER activities were achieved with the TM-IrO 2 having more defect-related features according to XAS spectra shown in Fig. 2. Therefore, we propose the vacancy defect states in TM-IrO 2 NPs to be the active sites displaying the higher electrochemical OER activities.
Photocatalytic oxidation activity. We determined the photocatalytic activities of 2-ATP oxidation by surface-sensitive S 2p core-level HRPES spectra acquired from the products of the exposure of 180 L of 2-ATP to oxygen and 365-nm-wavelength UV light in the presence of each type of TM-IrO 2 NPs (Fig. 4(a-f)). Three distinct 2p 3/2 peaks were observed, at 161.5, 162.9, and 168.6 eV, which corresponded to the thiol group (-SH; denoted as S1), the bound state (denoted as S2), and sulfonic acid (SO 3 H) (denoted as S3), respectively. Since sulfonic acid is formed as the result of the oxidation product of the thiol group 37,38 , and the unreacted one is observed at peak S1, we monitored the oxidation of 2-ATP by measuring the ratio of the intensity of S3 to that of S1. These oxidation also showed the same activity trends (Fig. 4(g)). Cr-IrO 2 or Mn-IrO 2 exhibited clear enhancements in photocatalytic activity compared with bare IrO 2 , and the conversion ratio decreased in the order of Fe-IrO 2 , Co-IrO 2 , IrO 2 , and Ni-IrO 2 . These results showed closely correlated activity trend between thiol oxidation and OER.
CO oxidation activity. In addition, we assessed the catalytic properties for thermal oxidation conversion of CO to CO 2 . The conversion reactions of each catalyst was monitored at various temperatures between 300 K and 550 K by using mass spectrometry (Fig. 5(a~f)). For all IrO 2 based catalyst samples, the conversion of CO to CO 2 steadly increased in the range 300 K to 450 K but rather decreased above 450 K probably due to desorption of the reactant chemicals on the catalyst surface before conversion. When we compare the CO oxidative conversion activity of the six catalysts at 450 K, again Cr-or Mn-doped IrO 2 NPs showed much enhanced converaion rate, and the conversion rates follow the same order at other reaction temperature as well. Overall, CO to CO 2 conversion using TM-IrO 2 catalysts exhibits the same order of the activity as electrochemical OER or photocatalytic 2-ATP oxidation.
Common oxidation activity trends. By characterizing oxidation activites related to the electronic structures of bare IrO 2 and five different TM-IrO 2 NP catalysts, we found that TM-IrO 2 catalysts had the same activity trends for three quite different reaction conditions (i.e. electrochemical oxidation of water in an aqueous condition, photocatalytic oxidation of 2-ATP in ultra-high vacuum condition, and thermal CO oxidation. For all oxidation reactions, five different TM-IrO 2 catalysts can be roughly divided into three groups in terms of the activity: (i) Cr-IrO 2 and Mn-IrO 2 which had similarly high enhancement in activity, (ii) Fe-IrO 2 and Co-IrO 2 which had less enhancement, and (iii) Ni-IrO 2 which had similar activity compared with non-doped IrO 2 . These results indicate there can be a common main contributor to catalytic activity, which is independent on the specific environmental conditions of each reation. In other words, changes in the electrical conductivity might appear important to determine electrocatalytic OER reaction activity but they are not expected to be critical for thermal CO oxidation. Or, doping in a photocatalyst material has been demonstrated to be effective to vary bandgaps or band positions affecting the photocatalytic acitivty, but it would not matter directly to electrochemical or thermal catalytic activity. Based on the spectroscopic surface analysis studies, a reasonable explanation is that the dopant metals disturb the Ir-O interaction and create oxygen vacancy defective states which are crucial to determine the catalytic activity of various oxidation reaciton. It was observed that electronic states of TM-IrO 2 were modified especially

Conclusion
In this study, transition metal doped IrO 2 nanoparticles were prepared and their catalytic activities were compared for three different types of oxidation reactions such as electrochemical oxygen evolution reaction, photocatalytic 2-ATP thiol oxidation reaction, and thermal CO conversion to CO 2 . Interestingly, Cr-IrO 2 or Mn-IrO 2 showed the most superior catalytic activity for all oxidation reactions followed by Fe-IrO 2 and Co-IrO 2 , and Ni-IrO 2 had similar or even decreased activity compared with non-doped IrO 2 NPs. In other words, these five TM-IrO 2 catalysts were classified into three group in terms of enhancement of the catalytic activity. For instance, the Cr-or Mn-IrO 2 exhibited decreased OER overpotential at 10 mA/cm 2 by 35~30 mV compared with non-doped IrO 2 , while the overpotential of Ni-IrO 2 rather increased by ~30 mV. These activity trends were also in good agreement with our STXM and HRPES measurements that showing Cr-IrO 2 and Mn-IrO 2 had the most changes in O K-edge or Ir 4f spectra related to the oxygen vacancy sites of IrO 2 . Meanwhile, Fe-or Co-IrO 2 had mild, and Ni-IrO 2 had smallest evolution of defect-related features compared with IrO 2 . These combined experimental results suggest that, among five transition metals, Cr-or Mn-dopant in IrO 2 are the most effective to generate defect structures by modifying the electronic/chemical states and thus contribute to enhanced activity for various oxidation reactions. were purchased from Sigma-Aldrich, and sodium hydroxide (NaOH) was obtained from Samchun. All aqueous solutions were prepared using double distilled water (DDW). The IrO 2 NP solution was prepared by modifying Wohler's method 39 . K 2 IrCl 6 (120 mg, 0.25 mmol) was dissolved in water in a flask to produce 100 mL of the corresponding aqueous solution, to which 1 mL of an aqueous NaOH solution (25 wt%) was added. The flask was capped with a foil, the solution in the flask mixed, and Material characterizations. Raman spectra were obtained by using a spectrometer (Horiba, ARAMIS) with an Ar + ion CW (514.5-nm-wavelength) laser. The morphologies of the samples were characterized by using a field-emission scanning electron microscope (FE-SEM, FEI Inspect F50) operating at an acceleration voltage of 10 kV. Element distributions of Ir and dopant metal were characterized by using a scanning transmission microscope, energy-dispersive X-ray spectroscopy (STEM-EDS, FEI Talos F200X). A scanning transmission X-ray microscopy (STXM) was performed at the 10A beamline of the Pohang Accelerator Laboratory (PAL). A Fresnel zone plate with an outermost zone width of 25 nm was used to focus the X-rays onto the TM-IrO 2 on the TEM grids. Image stacks were acquired using X-ray absorption spectroscopy (XAS) to extract the O K-edge spectra. High-resolution photoemission spectroscopy (HRPES) experiments were performed at the 8A1 beamline of the PAL with an electron analyzer (Physical Electronics, PHI-3057). The binding energies of the core level spectra were determined with respect to the binding energy (E B = 84.0 eV) of the clean Au 4 f core level for the same photon energy.

Preparation of IrO
Electrocatalytic OER performance. The electrocatalytic performances of the prepared nanoparticles were compared for the OER in 1 M KOH electrolyte. Each IrO 2 NP and TM-IrO 2 sample (10 mg) was dispersed in a solution containing Nafion (100 μL; 5wt %, DuPount) and either absolute ethanol (2 mL; Daejung, 99.9%) or 2-propanol (2 mL; Samchun, 99.5%) by ultra-sonicating the mixture for 30 min. The prepared IrO 2 NP or TM-IrO 2 catalyst ink samples were drop-cast on a clean glassy carbon electrode (Alfa Aesar), and the area density of the loaded nanoparticle catalyst was controlled to be 2.8 ± 0.4 mg/cm 2 . The catalyst nanoparticles deposited on the glassy carbon electrodes were masked to have an exposed area of 0.5 cm 2 . The electrochemical OER activities were measured in 1 M KOH (Aldrich, 85% purity), pH 13.7, by carrying out cyclic voltammetry at a scan rate of 10 mV/sec using a potentiostat (Ivium Technologies). A coiled Pt wire and an Hg/HgO (1 M NaOH, CH Instruments) electrode were used as a counter and reference electrode, respectively. All of the measured potentials were converted to being versus a reversible hydrogen electrode (RHE) using the equation To compensate for the potential drop resulting from solution resistance (R s ), electrochemical impedance spectra (EIS) were acquired in the frequency range 100 kHz -0.1 Hz, and a typical R s of ~5-10 Ω was obtained in 1 M KOH electrolyte.
Photocatalytic oxidation performance. 2-Aminothio-phenol (C 4 H 4 SHNH 2 , 2-ATP) was purified by carrying out turbo pumping prior to dosing it onto the IrO 2 NPs and the five TM-IrO 2 samples. The dosing was carried out by using a direct dozer controlled by means of a variable leak valve. The resulting samples were then irradiated with UV light (λ = 365 nm, VL-4.LC. Tube 1 × 4-Watt, Vilber Lourmat) through a quartz window of a vacuum chamber. The pressure of the chamber was maintained at 10 −6 Torr during dosing, and the number of exposed molecules was defined by the dosing time in seconds, i.e., with one Langmuir (L) corresponding to one second of dosing under 10 −6 Torr. Thermal CO oxidation. The synthesized catalyst was dispersed in an aqueous solution (0.05 mg/10 mL) which was spin-coated to load on a copper substrate. Then, the catalyst loaded substrate was transferred into the ultra high vacuum (UHV) chamber for CO oxidation reaction. CO gas was flew into the chamber until the partial pressure became 10 −6 torr, and the same amount of O 2 gas was added. The catalyst was heated up to reaction temperature (300, 400, 450, or 550 K), and UV lamp was turned on through the view-port of UHV chamber to activate O 2 . The CO oxidation reaction was conducted for 1 hour and the amount of the generated CO 2 was measured by a Hiden RC 301 mass spectroscopy (mass range ~300 amu). The reaction was repeated by five times at each condition and the average conversion rate was taken.