Chlorine evolution reaction electrocatalysis on RuO₂(110) and IrO₂(110) grown using molecular-beam epitaxy

Chlorine is an essential building block in the materials and chemical industries. With annual production exceeding 45 × 10⁶ tons, chlorine production consumes over 150 TWh of electricity. Therefore, improving the chlorine production cost and efficiency offers a pathway to reduce the energy requirement to contribute to the societal pursuit of sustainability.¹ In this work, we study the chlor-alkali process, the primary chlorine production method, with the goal of uncovering the mechanism of the anodic reaction, i.e., chlorine evolution reaction (CER, 2Cl⁻ → Cl₂ + 2e⁻).² 

Decades of research have led to the development of dimensionally stable anodes (DSAs), which are used to support the CER in the chlor-alkali process³ and in some redox flow batteries.^4,5 These DSAs contain RuO₂ and IrO₂ as the electroactive surfaces, with RuO₂ as the more active while IrO₂ as the more stable.^6–10 Using density-functional theory (DFT) calculations, Hansen et al. have suggested that the CER proceeds via the Cl⁻ electroadsorption on the O-covered rutile surfaces.¹¹ Few years later, Exner et al. have identified that OCl_ad is the thermodynamically stable adsorbate on RuO₂ when the solvent effects are taken into account in the DFT calculations.¹² Their studies have indicated that the OCl_ad stabilization energy can serve as an activity descriptor to predict the effectiveness of a CER electrocatalyst and further use this result to explain why RuO₂ is more active than IrO₂. Nevertheless, there is still a gap in the comparison between theory and experiment; most measurements have so far been conducted on polycrystalline surfaces, where there can be different terminations, structural defects, or even phases within the same material.^13–15 Inspired by the recent studies by Rao et al.¹⁶ and Exner et al.,¹⁷ whose studies have pioneered the use of single-orientation rutile thin films as a model system for studying the molecular details of electrocatalysis, we use well-defined (110) surfaces of RuO₂ and IrO₂ grown using molecular-beam epitaxy (MBE) to systematically compare the CER electrocatalysis on RuO₂ vs. IrO₂. We further do a reaction-order measurement to provide the molecular details on the CER mechanism on RuO₂(110) to test the previously proposed Volmer-Heyrovsky mechanism.^18,19

Early investigations of the CER on well-defined surfaces can be traced to the work of Guerrini et al.,²⁰ whose Tafel slope measurement result on RuO₂ single crystals was subsequently analyzed by Over and colleagues to reveal the detail of the CER mechanism.^17,21 By comparing the Tafel slope at different overpotentials, they have suggested that the CER on RuO₂(110) occurs via the Volmer–Heyrovsky mechanism on O-covered RuO₂. This finding is consistent with the recent X-ray work that has shown that the RuO₂(110) surface in 0.1M HClO₄ is covered by O_ad at the potential range where the CER occurs using X-ray scattering.¹⁶ However, because the X-ray analysis was done in a chloride-free electrolyte, it is unclear if the conclusion would be the same in the presence of Cl⁻. We have recently reported the surface chemistry of IrO₂(110)²² and RuO₂(110)²³ films grown using MBE as a function of potential. In this study, we use these model surfaces to study how the presence of Cl⁻ can affect the surface chemistry of IrO₂ and RuO₂. Our approach utilizes cyclic voltammetry (CV) to determine how the surface adsorbates change with potential on single-termination RuO₂(110) and IrO₂(110). Our finding confirms the theoretical prediction and shows that Cl⁻ does not affect the OH_ad group and only interact with the O_ad-covered rutile surface. We further measure the CER activity in different Cl⁻ concentrations to extract the reaction order. Although our reaction-order result is different from the literature,²⁴ we find that the Volmer-Heyrovsky mechanism can explain both the reaction order we observe and the results in the literature. The difference is due to the impact of the Cl⁻ activity on the OCl_ad coverage, which influences the Tafel slope and the reaction order in a concentration-dependent manner.

40 monolayers thick RuO₂(110) and IrO₂(110) were grown on TiO₂(110) substrates by using MBE. A quartz crystal microbalance was used to calibrate the Ru and Ir fluxes for the MBE growth. Distilled ozone was used as an oxidant for the growth of all films in a background pressure of 1 × 10⁻⁶ Torr. The substrate temperature was 300 °C and 350 °C for IrO₂ and RuO₂, respectively. In situ reflection high-energy electron diffraction (RHEED), low-energy electron diffraction (LEED), and X-ray diffraction (XRD, Rigaku SmartLab) were used to confirm that the IrO₂ and RuO₂ films were phase-pure, epitaxial, and of high structural quality.

Electrical contacts were made using the same protocol as reported previously.²⁵ In brief, titanium wires were attached to RuO₂ and IrO₂ films using silver paint (Ted Pella, Leitsilber 200) and the samples were covered with epoxy (Omegabond 101) except for the RuO₂ and IrO₂ surfaces. All electrochemical characterizations were done in a three-electrode standard electrochemical cell (Pine) with a potentiostat (Bio-Logic). The reference electrode was an Ag/AgCl electrode (Pine), calibrated to the H₂ redox. A Pt wire was used as a counter electrode. All the potentials in this study were resistance-corrected potentials. The electrolyte/cell resistance was obtained from an impedance measurement by using the high frequency intercept of the real resistance. The 0.1M H₂SO₄ electrolyte was prepared by dissolving H₂SO₄ (99.999% purity, Sigma Aldrich) in deionized water (18.2 MΩ cm). KCl (99.999% purity, Sigma Aldrich) was added into 0.1M H₂SO₄ to prepare electrolytes with different KCl concentrations.

Figure 1 shows the low-energy electron diffraction (LEED) images and the CVs of RuO₂(110) and IrO₂(110) in 0.1M H₂SO₄. The LEED images [Figs. 1(a) and 1(b)] indicate that our oxide films contain only the (110) termination as anticipated for the epitaxial growth. The CV of RuO₂(110) in 0.1M H₂SO₄ shows two pairs of reversible peaks (∼1.1 V and ∼1.5 V vs. RHE), which are similar to the previous single-crystal RuO₂(110) result in H₂SO₄.^20,26 Using previously reported surface structural characterizations¹⁶ and our prior DFT calculation,²³ we assign the first peak as the OH_ad formation (H₂O_ad → OH_ad + H⁺ + e⁻) and the second peak as the O_ad (OH_ad → O_ad + H⁺ + e⁻) formation. In comparison to RuO₂(110), we observe only one pair of peaks in CV of IrO₂(110) in the potential window from 0.3 to 1.6 V vs. RHE. This observation agrees with our previously reported CV of IrO₂ in acidic electrolyte showing that the peak potential of O_ad formation is more positive than 1.6 V vs. RHE and that the onset of the oxygen evolution reaction (OER) occurs prior to the O_ad formation.²² It is important to note that the O_ad formation on IrO₂ occurs at a higher potential than RuO₂(110), consistent with the previous DFT calculation.^22,23

To gain insights into the CER, we examine the CV of RuO₂(110) and IrO₂(110) in the presence of Cl⁻. Figures 2(a) and 2(b) show the CVs of RuO₂(110) and IrO₂(110) in 0.1M H₂SO₄ with 100 mM KCl at a scan rate of 200 mV/s. We observe the anodic current at ∼1.4 V vs. RHE, which we assign to the CER. On the returning, backward scan direction, we observe two additional cathodic peaks at ∼1.4 V vs. RHE and <1.2 V vs. RHE. We will discuss the origin of these peaks in the next paragraph. We first focus on the CER electrocatalysis between RuO₂ vs. IrO₂. To obtain the CER activity in the absence of the capacitive background and the cathodic reduction peaks, we measure the CVs at a lower scan rate (10 mV/s). The obtained CVs show the CER current with an onset of 1.4 V vs. RHE for RuO₂ and 1.5 V vs. RHE for IrO₂, confirming that RuO₂(110) is more active for the CER than IrO₂(110) [Fig. 2(c), RuO₂ is two orders of magnitude more active than IrO₂]. Interestingly, the CER starts at the onset of the O_ad formation on RuO₂(110). This finding is consistent with the CER active site being the O-covered rutile surface, which takes higher, more oxidizing potential to form for IrO₂ than RuO₂, explaining why IrO₂ requires a higher overpotential.

We now discuss the origin of the cathodic peaks in the CV of RuO₂ in the presence of Cl⁻. There are two hints: we observe that this cathodic current depends on the cycling potential window and the KCl concentration (see Fig. S1 of the supplementary material). By changing the upper limit of the CV from 1.6 to 1.5 V vs. RHE while keeping the lower limit constant (at 0.3 V vs. RHE), the cathodic peaks disappear. In addition, the cathodic peaks increase with the KCl concentration. Based on these results, we assign the cathodic current to the chlorine reduction reaction (CRR) of recently formed Cl₂ that has not yet had time to diffuse into the electrolyte. We note that we cannot rule out the reduction peak which could also be due to the reduction of the OCl_ad intermediate. Future in situ spectroscopy work would be essential to uncover the nature of these redoxes.

The CER electrocatalysis on RuO₂(110) occurs before the OER even at 10 mM KCl [Fig. 3(a)]. IrO₂(110), however, has the opposite situation; the CER and OER electrocatalysis on IrO₂(110) occur at nearly the same potential. In fact, we cannot distinguish the CER and OER current on IrO₂(110) until the concentration of KCl exceeds 50 mM, at which Cl⁻ concentration IrO₂(110) begins favoring the CER [Fig. 3(b)]. Our observation agrees with the previous finding that a significant portion of the oxidation current is attributed to the OER when the Cl⁻ concentration is dilute, making IrO₂ highly tolerant to the Cl⁻ during the OER.^27,28

We now focus our attention on the analysis of the CER electrocatalysis and its connection with the surface O_ad on RuO₂(110), where the contribution of the OER is minimal and that all the oxidative current can be assigned to the CER. We first investigate the Tafel slope, which contains information on the mechanism.^29–31 To analyze the Tafel slope, we subtract the background OER current (i.e., the CV in chloride-free electrolyte) to acquire the CER current (i_CER), as shown in Fig. 4(a). The Tafel slope of CER in 0.1M H₂SO₄ with 100 mM KCl is 40 mV/decade as the potential is below 1.46 V vs. RHE. We also observe that the slope varies gradually with the potential from 40 to 80 mV which agrees with the previous CER study on single-crystal RuO₂(110).^20,21 The increment of the Tafel slope has been attributed to the changing Cl_ad coverage, which also depends on the potential.¹⁰ Exner et al. have suggested that the rate limiting step of the CER switches from the Heyrovsky to the Volmer step as the applied potential increases.¹⁷ 

To understand the role of Cl adsorption in the mechanism, we measure the reaction order of the CER with respect to Cl⁻ at different potential values [Fig. 4(b)]. We find that the reaction order is greater than unity at fixed RHE reference and decreases as potential increases. We note that this finding is different from the previous study by Consonni et al., whose CER measurement on RuO₂(110) single crystals shows the reaction order of 1 with respect to Cl⁻ in highly concentrated NaCl solution.²⁴ 

We can rationalize the ∼1.8 reaction order observed in this work under dilute KCl solution and the unity reaction order result measured in concentrated NaCl solution by Trasatti and co-workers by recognizing that the CER mechanism depends on the Cl_ad coverage. We build on the previous result that CER proceeds via the Volmer–Heyrovsky mechanism, assuming that Cl⁻ electroadsorbs directly on O_ad

In this pathway, we can show that θ_OCl depends on the applied potential and Cl⁻ concentration, assuming a steady-state condition (see the supplementary material)

In this expression, k₁^f and k₁^b represent the forward and backward rate constants of the Volmer step while k₂^f and k₂^b represent the forward and backward rate constants of the Heyrovsky step. In this expression, the reaction order of the OCl_ad coverage is between 0 and 1 with respect to Cl⁻. In the limit of large overpotential and high Cl⁻ concentration [i.e., k₁^b ≪ k₁^f · a(Cl⁻) and k₂^b ≪ k₂^f], θ_OCl can be approximated as k₁^f/(k₁^f + k₂^f). Using this approximation, the CER rate can be expressed as

In this rate expression, the current density of the CER is directly proportional to the Cl⁻ concentration, indicating the reaction order of 1 with respect to Cl⁻. At a small overpotential and/or low Cl⁻ concentration, which is the experimental condition in this work, the assumption of constant θ_OCl is no longer valid, i.e., θ_OCl depends on the potential and the Cl⁻ concentration. This Cl adsorption behavior can explain why the reaction order is greater than 1. Since the reaction order of OCl_ad formation is between 0 and 1 with respect to Cl⁻, the reaction order of the CER with respect to Cl⁻ should lie in the range of 1-2, in agreement with our experimental observation in Fig. 4(b). Our measurement of the CER solidifies the Volmer-Heyrovsky mechanism at both low and high overpotentials, and high and low Cl⁻ concentrations. Our comparative analysis between IrO₂ and RuO₂ also suggests that the OCl_ad formation may serve as the activity descriptor for the CER.

In summary, we demonstrate the measurement of the CER activity on RuO₂(110) and IrO₂(110) grown on TiO₂(110) substrates by using MBE. RuO₂(110) was found to be more active for the CER, in agreement with the prior results on polycrystalline samples. In our measurement, RuO₂(110) is approximately two orders of magnitude more active than IrO₂(110). We attribute the lower CER activity of IrO₂(110) to the higher O_ad energy, which requires higher potential to form an O-covered surface for the electroadsorption of Cl⁻. We find that the reaction order of the CER on RuO₂ is potential-dependent and greater than 1 at lower potential and/or Cl⁻ concentration. This observation is consistent with the Volmer-Heyrovsky mechanism and reflects the influence of the Cl⁻ activity on the Cl_ad coverage. Our experimental results and the microkinetic analysis suggest the importance of the OCl_ad formation (Cl⁻ + O_ad ↔ OCl_ad + e⁻) in the CER electrocatalysis and that this parameter could be used to screen future CER electrocatalysts.

See supplementary material for a detailed derivation of microkinetic analysis and additional CVs showing cathodic current on RuO₂(110).

This work was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Award No. DE-SC-SC0018029. This work made use of the Cornell Center for Materials Research (CCMR) Shared Facilities, which are supported through the NSF MRSEC Program (DMR-1719875). D.-Y.K. is grateful for support from the Taiwan Government Scholarship to study abroad. H.P. acknowledges support from the National Science Foundation [Platform for the Accelerated Realization, Analysis, and Discovery of Interface Materials (PARADIM)] under Cooperative Agreement No. DMR-1539918. J.N.N. acknowledges the support from the NSF Graduate Research Fellowship under Grant No. DGE-1650441. This work was performed in part at the Cornell NanoScale Science & Technology Facility (CNF), a member of the National Nanotechnology Infrastructure Network, which is supported by the NSF (Grant ECCS1542081).