The active site for the water oxidising anodic iridium oxide probed through in situ Raman spectroscopy

Abstract

The structure of anodic iridium oxide (IrO_x) under water oxidation was explored using in situ Raman spectroscopy and theoretical calculations. Isotopic substitution experiments and theoretical calculations confirmed formation of an Ir=O species during oxygen evolution reaction.

---

The oxygen electrode remains a central challenge in implementing a hydrogen-based economy.^1,2 IrO_x remains one of the most important catalysts for water oxidation reaction.³ Dimensionally stable anodes (DSA) are composed of primarily IrO₂.^4–8 These materials require being treated through activation protocols before they can be implemented. IrO_x materials were shown to be very active by Mallouk et al.⁹ In the alkaline media, Mallouk et al. proposed that these materials are unstable and form molecular species that show oxygen activity.¹⁰ In fact, the activity shown by these materials in alkaline electrolytes was much higher than in acidic or neutral electrolytes. We have recently characterized the active IrO_x material through ex situ Raman spectroscopy on anodic IrO_x grown on an iridium foil.¹¹ Attempts to understand the mechanism of OER on IrO_x has encouraged operando studies using XPS and XAS.^12,13 These investigations focused on an Ir-centered approach to understanding the nature of the active catalyst. XPS studies observed an additional Ir 4f peak at higher binding energy (+0.7 eV) during OER and suggested an Ir⁵⁺ species forming during the catalytic event.¹² Minguzzi et al. suggested the formation of both Ir³⁺ and Ir⁵⁺ during OER. But both these studies do not compare the shifts to standard Ir-containing compounds. High oxidation states (+5 or greater) are only known for fluorine-containing molecules amongst the stable Ir compounds or in IrO_x molecular species isolated within a noble gas matrix. Reier et al. and Pfeiffer et al. observed an additional feature at 529 eV in the O-Kedge NEXAFS spectra of highly OER active Ir–NiO_x and amorphous IrO_x respectively. Using DFT calculations, they have suggested this to be an electrophilic oxygen species present in these materials.^14–16 By the nature of their inquiry, these studies cannot clarify the exact chemical nature of the IrO_x active site which is possible through vibrational spectroscopy. Mo et al. carried out in situ Raman spectroscopy on IrO_x/Au system but essentially used XAS data for interpretation of results.¹⁷ Frei and coworkers have investigated IrO_x clusters as OER catalysts using rapid scan FTIR under visible light.¹⁸ They reported detection of a transient Ir–OOH species with O–O vibration being detected at 830 cm⁻¹. This work goes beyond work reported by Pavlovic et al. and uses in situ Raman spectroscopy to understand the behavior of these materials under conditions of OER.¹¹ Specifically we focus on active site that forms in situ. We suggest a Ir=O species, with five coordinated square pyramidal Ir as the active site for water oxidation and suggest a mechanistic pathway for product formation and replenishment of the active site.

Cyclic voltammogram of IrO_x materials in different electrolytes is well known (Fig. S1, ESI†). All the potentials reported in this paper are referenced to the reversible hydrogen electrode (RHE). The material was electroprecipitated at 1.5 V onto a Au substrate as shown by Mallouk et al.⁹ In our previous publication we have shown that the anodic Iridium oxide is essentially a hydrous material with OH and H₂O groups playing important role in its overall structure.¹¹ Experiments of IrO_x/Au were conducted in an in situ Raman cell, using Biologic VSP-300 potentiostat. Raman spectra were collected with Ocean Optics QE65 pro spectrometer using 785 nm Laser source. The Laser intensity was maintained at 500 mW per 0.1 mm² at the sample (collection time of 10 s). The potential was scanned from 0.4–1.8 V, at steps of 0.1 V.

Results of the in situ Raman experiment are shown in Fig. 1. More details of the experiments are provided in the ESI.† Peaks at 225 and 324 cm⁻¹ originate from the Au surface. The peaks from 450–780 cm⁻¹ originate from iridium oxygen vibrations, where peak assignments have been used from Pavlovic et al.¹¹ Electrochemically, four regions (R) in the IrO_x CV can be clearly identified. R1(E < 0.4 V), R2 (0.4 < E < 0.7 V), R3 (0.7 < E < 1.2 V) and R4 (E < 1.4 V). Whereas R1 corresponds to a colorless compound rich in Ir³⁺, R2 represents the region of first oxidation wave from Ir³⁺↔ Ir⁴⁺. At the beginning of the experiment at OCP, there is a certain peak structure (465, 527, 600, 702 cm⁻¹), which can be attributed to the Ir–O–Ir stretch vibrations of μ-oxo type from the basal plane of edge sharing IrO₆ octahedra.¹¹ As soon as the potential of 0.4 V is applied, a sharp peak at 520 cm⁻¹ is formed. This is the most dominant peak δ in region R1. Region R1 has no ε peak which appears in R2. The ε peak loses intensity in R3. R4 shows significant drift of peaks γ → ε towards higher frequencies. At potentials 1.3 V and higher, AuO_x is formed (Fig. S4, ESI†). Thus peaks in R4 have substantial AuO_x contribution to them. The final peaks observed at OCP in the treated materials are composed of both IrO_x and AuO_x peaks. The peak η at 771 cm⁻¹ is seen in R4. This peak disappears as soon as the applied potential is switched off. This peak does not exist in experiments done on a bare Au surface and thus must be assigned to IrO_x material under conditions of water oxidation. Isotopic substitution using D₂O and H₂O¹⁸ as synthesis and reaction media were also carried out. Results have been tabulated in Table 1. Fig. 2 shows effects of isotopic substitution on η. In the experiment that has 50 : 50 (H₂O : H₂O¹⁸), only two peaks of nearly equal intensity were visible. Thus this peak could be assigned to Ir=O vibration (as opposed to any other species that contained O–O, or OH vibration). If O movement is the primary oscillator in this vibration, then one can expect a shift of upon O¹⁸ substitution. Deuterium substitution resulted in almost no shift of the peak (Δf ∼ 5 cm⁻¹). For Ir–O–OH containing groups, the shift reported by Frei et al. due to deuteration was 30 cm⁻¹.¹⁸ Small shifts due to deuteration are possible through altered hydrogen bonding contributions from the solvent. Vibrations where the motion of the hydrogen atom explicitly contribute to the vibrational mode show significant alteration of frequency upon deuteration due to the large mass change . Fig. S5 (ESI†) shows the results from various experiments with combinations of H₂O, D₂O, and H₂O¹⁸ containing electrolytes. Fig. 2, shows η peak at OER potentials in variously labeled electrolytes. Fig. S5(d) (ESI†) shows the results of experiments when material was synthesized in H₂O¹⁸ and tested in H₂O.¹⁶ Peaks γ to ε begin to split into two sets of peaks in the beginning of R3 indicating that material exchanges oxygens with the electrolyte even at potentials as low as 0.7 V. Even the position of η peak only shows the contribution from O¹⁶ isotope.

Fig. 1 (a) In situ Raman spectroscopy of IrO_x/Au system in 0.1 M NaOH. The peaks labeled by greek letters originate from Ir–O stretch vibrations. Rest of the peaks can be assigned to Au substrate (ESI,† Section 2). η peak is only visible during the process of OER and disappears with removal of applied potential. Peak assignments (in greek letters) have been borrowed from Pavlovic et al.¹¹ (b) CV of IrO_x/Au in 0.1 M NaOH (c) Average currents from in situ experiments in various electrochemical media.

Table 1 Peaks obtained in the Raman spectra of IrO_x/Au labeled at 0.8 V. The peaks tend to move with applied potential and thus their exact location can be described only at a certain potential. The Ir–O stretching assignments have been borrowed from Pavolovic et al.¹¹

Peak labels	H2O	D2O	H2O^18	1 : 1 (H2O : H2O^18)	H2O^18 tested in H2O
a Collected at 1.7 V.
γ	476	475	423	454, 482	430, 458, 480
δ	527	554	476	541, 482	554
ε	595	602	554	605, 552	608
η^a	771	767	712	722, 771	—

---

Fig. 2 Raman spectra of the η peak in various isotopically labeled electrolytes. The peaks are background corrected and normalized (details in Experimental section).

Theoretical calculations using density functional theory were carried out to better understand the IrO_x system (Fig. S6–S8, ESI†). Specifically, calculations were performed to understand the identity of the Ir=O species. A trimeric IrO₆ unit was used to construct a model system for calculation. It was found that an Ir–O unit on the central Ir atom resulted in the formation of a square pyramidal central Ir atom. [Ir⁴⁺–Ir⁴⁺–Ir⁴⁺] trimeric IrO_x cluster was used for computation (Fig. 3). The Ir–O vibration was computed to be at 829 cm⁻¹. Furthermore it was found that the vibrations could be significantly modulated through hydrogen bonding near the “Ir–O” e.g. the Ir–O vibration was reduced to 766 cm⁻¹ from 829 cm⁻¹ by placing a H₂O molecule near the Ir–O moiety. Besides, the relative intensity of this vibration was found to be very strong compared to the other vibrations within this molecule. We believe at the optimum potential some of the [IrO₆]_n species turn into the OER active Ir=O species that removes any further oxidative strain from the material. Even though present in much less quantity, the extremely high Raman activity of the Ir–O stretch vibration makes it possible for the active species to be detected within a matrix of [IrO₆]_n containing material. The 771 cm⁻¹ peak can also be detected in acidic media but has a much smaller intensity (the OER activity of the material in acidic and neutral media is much less than alkaline media (Fig. S2, ESI†)). This is likely because the material is highly condensed in the acidic and neutral media and not many exposed active sites containing Ir=O species form, which gets reflected in the overall activity of the material. Ir–OOH species (Fig. S7, ESI†) were also computed. The frequency of vibration of O–O stretch was 715 cm⁻¹, besides the O–O stretch Raman intensity was much smaller. In light of our experimental and computational investigations we assign the observed 771 cm⁻¹ vibration to Ir=O stretch of a square pyramidal Ir site. The frontier orbitals of IrO_x were calculated and have been visualized in Fig. 3. The SOMO (Fig. 3(b)) shows that Ir–O unit is the most reactive part of the molecule with a major part of the molecular orbital localised in this region. The frontier orbitals are anti-bonding in nature between Ir-d and O-p with significant localization on O atom making it a suitable target for reaction. We believe this oxygen may be responsible for making the important O–O bond during the formation of molecular O₂ from water (Fig. 4). M–O species responsible for oxidizing water are common in literature.¹⁹ In case of Mn-containing Photosystem II catalyst, the O–O bond formation either happens through a nucleophilic attack at the Mn–O center (Mn⁵⁺–O, Mn⁴⁺–O or Mn⁴⁺–oxyl) species or by an oxyl–oxyl radical coupling mechanism.^20–22 But the exact nature of the species is debatable. It has been suggested in the literature that an electrophilic oxygen at the active site is essential to make reactions like O–O bond formation and C–H bond activation, possible.^16,23,24 Wang et al. and Concepcion and coworkers suggested an O–O bond formation using nucleophilic attack on Ru=O site.^25,26 Using DFT based theory, Mai et al. suggested that C–H activation essentially consists of two parts, an Fe⁴⁺=O species which was a precursor to Fe³⁺–O˙⁻ oxyl species that carried out the hydrogen atom transfer.²³ Similarly, using DFT calculations Yang et al. have argued that Ru⁵⁺=O can act as a precursor to Ru⁴⁺–O˙⁻ oxyl species which carries out water O–O bond formation. For the IrO_x cluster shown in Fig. 3, the total Löwdin charge at the oxygen atom amounts to only −0.12 (as compared to −0.35 for the bridging oxo-s) and the Löwdin spin population at the central Ir and oxo amounts to 47% and 29%, respectively. Semantically, although this charge distribution would be compatible with a metal-oxyl (O˙⁻) species, the spin distribution rather indicates a much lower spin population at the oxygen (29%) than expected for a radical species (∼100%). This is the result of a somewhat covalent Ir–O π-interaction that involves the spin carrying d_xz orbital (t_2g in octahedral symmetry) at Ir (ESI†). We believe what we see in our experiment, is the structure that is precursor to the reactive oxyl species as proposed by Mai and Yang et al.^23,24 The observation of a reactive oxyl radical species under experimental conditions remains difficult, specifically at room temperature. Herlihy et al. observed an oxyl species on TiO_x system using in situ ultrafast infrared spectroscopy.²⁷ Based on our data, we claim an Ir–O species that can in principle act as a precursor to a highly electrophilic oxyl species that reacts with water (or dissociated water) to form an O–O bond.

Fig. 3 Theoretical calculations were carried out on IrO_x species containing with “Ir–O” moiety. The calculations resulted in an Ir–O species that was essentially square pyramidal at the Ir center. (Top) Ir⁴⁺ containing species with the central atom having the Ir–O bond. The oxygen atom has been labeled with a O* (Top). (bottom) SOMO of the species.

Fig. 4 Proposed active species responsible for water oxidation in IrO_x materials (alkaline media). Suggested mechanism of oxidation of water and regeneration of the active site is shown. The oxo species is likely to go through an oxyl type species before the O–O bond is formed.

The scaffold enabling the OER is derived from a highly hydrous [IrO₆]_n species (of the type IrO_x(OH)_y(H₂O)_z), where each Ir atom is octahedrally coordinated by oxygens. Whereas the colour and vibrational spectroscopy of most of this material can be traced using ex situ spectroscopy, the formation of an active a square pyramidal “Ir–O” species is detectable only through in situ spectroscopy under applied potential. We propose that this Ir–O species is the active site within the molecule which is open to attack from H₂O molecules leading to formation of O₂ through formation of an O–O bond as shown in Fig. 4. The OER chemistry is likely to proceed primarily in the ligand sphere.

The authors thank BMWi for supporting this work through grant 03ESP106D (EKOLYSER). Open Access funding provided by the Max Planck Society.

Conflicts of interest

There are no conflicts to declare.

References

1. J. A. Turner, Science, 2004, 305, 972–974.
2. R. Schlögl, ChemSusChem, 2010, 3(2), 209–222.
3. C. C. L. McCrory, S. H. Jung, J. C. Peters and T. F. Jaramillo, J. Am. Chem. Soc., 2013, 135(45), 16977–16987.
4. A. Hrussanova, E. Guerrini and S. Trasatti, J. Electroanal. Chem., 2004, 564, 151–157.
5. A. Damjanovic and M. K. Y. Wong, J. Electrochem. Soc., 1967, 114, 592–593.
6. E. Guerrini, H. Chen and S. Trasatti, J. Solid. State. Electrochem., 2007, 11, 939–945.
7. S. Trasatti, Electrochim. Acta, 1991, 36, 225–241.
8. S. Trasatti, Electrodes of Conductive Metallic Oxides, Part A and Part B, Elsevier, Amsterdam, 1980–1981.
9. Y. Zhao, N. M. Vargas-Barbosa, E. A. Hernandez-Pagan and T. E. Mallouk, Small, 2011, 7(14), 2087–2093.
10. Y. Zhao, N. M. Vargas-Barbosa, M. E. Strayer, N. S. McCool, Maria-Erini Pandelia, T. P. Saunders, J. R. Swierk, J. F. Callejas, L. Jensen and T. E. Mallouk, J. Am. Chem. Soc., 2015, 137(27), 8749–8757.
11. Z. Pavlovic, C. Ranjan, Q. Gao, M. van Gastel and R. Schloegl, ACS Catalysis, 2016, 6(12), 8098–8105.
12. H. G. Sanchez Casalongue, M. Ling Ng, S. Kaya, D. Friebel, H. Ogasawara and A. Nilsson, Angew. Chem., Int. Ed., 2014, 53, 7169–7172.
13. A. Minguzzi, O. Lugaresi, E. Achilli, C. Locatelli, A. Vertova, P. Ghigna and S. Rondinini, Chem. Sci., 2014, 5, 3591–3597.
14. T. Reier, Z. Pawolek, S. Cherevko, M. Bruns, T. Jones, D. Teschner, S. Selve, A. Bergmann, H. N. Nong, R. Schlögl, K. J. J. Mayrhofer and P. Strasser, J. Am. Chem. Soc., 2015, 137(40), 13031–13040.
15. T. Reier, H. N. Nong, D. Teschner, R. Schlögl and P. Strasser, Adv. Energy Mater., 2016, 1601275,  DOI:10.1002/aenm.201601275.
16. V. Pfeifer, T. E. Jones, J. J. Velasco Vélez, C. Massué, M. T. Greiner, R. Arrigo, D. Teschner, F. Girgsdies, M. Scherzer, J. Allan, M. Hashagen, G. Weinberg, S. Piccinin, M. Hävecker, A. Knop-Gericke and R. Schlögl, Phys. Chem. Chem. Phys., 2016, 18, 2292.
17. Y. Mo, I. C. Stefan, W. B. Cai, J. Dong, P. Carey and D. A. Scherson, J. Phys. Chem. B, 2002, 106, 3681–3686.
18. N. Sivasankar, W. W. Weare and H. Frei, J. Am. Chem. Soc., 2011, 133, 12976.
19. Hohenberger, Johannes, Ray, Kallol, Meyer and Karsten, Nat. Commun., 2012, 3, 720.
20. N. Cox, D. A. Pantazis, F. Neese and W. Lubitz, Acc. Chem. Res., 2013, 46, 1588–1596.
21. L. Rapatskiy, et al. , J. Am. Chem. Soc., 2012, 134, 16619–16634.
22. P. E. M. Siegbahn, Biochim. Biophys. Acta, 2013, 1827, 1003–1019.
23. B. K. Mai and Y. Kim, Inorg. Chem., 2016, 55(8), 3844–3852.
24. X. Yang and M. H. Baik, J. Am. Chem. Soc., 2006, 128(23), 7476–7485.
25. L. P. Wang, Q. Wu and T. Van Voorhis, Inorg. Chem., 2010, 49, 4543.
26. J. J. Concepcion, M. K. Tsai, J. T. Muckerman and T. J. Meyer, J. Am. Chem. Soc., 2010, 132, 1545.
27. D. M. Herlihy, M. M. Waegele, X. Chen, C. D. Pemmaraju, D. Prendergast and T. Cuk, Nat. Chem., 2016, 8, 549–555.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental details, cyclic voltammograms, Raman spectra in various electrochemical media and computational details. See DOI: 10.1039/c7cc05669a

---