Improving the Stability of DSA Electrodes by the Addition of TiO₂ Nanoparticles

Titanium substrates (15.5 × 17 mm) were pre-treated by etching in 10% hydrochloric acid at room temperature for 1 h, before rinsing with deionized water and drying at room temperature. The precursor solution was prepared by mixing a solution of 0.02 M IrCl₃ (99.9%, American Elements) in deionized water with a suspension of 0.02 M TiO₂ nanoparticles (Degussa P25 99.9%) in isopropanol. P25 TiO₂ nanoparticles is a standard material found in applications (especially photocatalysis) and is characterized as mixture of anatase and rutile (75–84% anatase) with a BET surface area of approximately 49 m² g⁻¹, and average crystallite sizes of 25–29 nm (anatase phase) and 55–85 nm (rutile phase).^22,23 The size of the TiO₂ nanoparticles used in this work are larger than those used to stabilize PbO₂ films (approx. 5 nm),²¹ but are smaller than the SiO₂ particle-like film used to stabilize IrO₂ films (approx. 500 nm).¹⁵ The mixture of IrCl₃ and TiO₂ nanoparticles was ultrasonicated (Kudos 100W Ultrasonic bath) for at least 5 minutes to ensure that the TiO₂ nanoparticles were well-dispersed. This suspension was typically stable for at least 2 hours without evidence of settling or separation. To coat the titanium substrates, the suspension was sprayed over the Ti substrates which were heated at 150ᵒC using a small airbrush (Blackridge Professional) with a nozzle size of 0.2 mm. The electrode was coated with 6 layers, with each layer annealed in air at 450ᵒC for 5 min and the final layer annealed for 1 h. Electrodes containing 0, 10, 25 and 40 wt% TiO₂ nanoparticles were prepared in this way and in all cases the total oxide loading was determined to be 0.35 ± 0.06 mg cm⁻².

The surface morphology of the prepared anodes was investigated using a JEOL JSM 7000F field emission, high-resolution scanning electron microscope (at 15 kV) and EDS analysis performed with the JEOL energy dispersive X-ray analysis system.

The electrochemical behavior of the prepared anodes were assessed in 0.5 M H₂SO₄ electrolyte using a three-electrode cell system controlled by a Gamry Interface 1000 potentiostat. Electrochemical impedance spectroscopy (EIS) was used to determine the resistance between the working and reference electrode. The EIS was conducted in hybrid mode, with a DC current of 0.6 mA with an AC amplitude of 10 mV between 100 kHz and 0.1 Hz. Accelerated lifetime tests were performed on the TiO₂ nanoparticle + IrO₂ anodes at 50°C using a large stainless steel cathode (16 cm²) and 2 M sulphuric acid as the electrolyte. In these tests, the anodes were operated at 1 A cm⁻² until the cell potential reached 10 V at which point the anodes were deemed to have failed.

Scanning electron microscopy revealed that the surface structure and the morphology of fabricated electrodes vary considerable as a function of the TiO₂ nanoparticles: IrO₂ ratio (Figure 1). As expected, the layer of pure IrO₂ exhibits the cracked mud structure commonly observed for DSA electrodes.^1,2,17 For the electrode layer containing 10 wt% TiO₂ nanoparticles, a similar morphology to the pure IrO₂ anode is found, but with much smaller cracks across surface of the electrode. Quite different morphology is observed for the electrodes containing 25 and 40 wt% TiO₂ nanoparticles, with these showing pronounced roughness and a particle-like structure. While the surface morphology of electrode layer containing 25 wt% TiO₂ nanoparticles is uniform on the ∼10 μm scale (as determined by EDS mapping), the electrode containing 40 wt% TiO₂ nanoparticles has an uneven distribution of large agglomerates, with some regions of the substrate poorly by the IrO₂-TiO₂ composite (marked by a dashed region on the figure). EDS analysis of this electrode revealed that the poorly coated regions had Ti:Ir atomic ratios of 25:2, whereas the coated regions had Ti:Ir atomic ratios of 4:1. This suggests that the poorly coated regions were still covered with IrO₂, but as the Ti ratio was very high, it suggests that the coating in these regions were very thin. It must be noted that in all cases, the EDS revealed higher than expected Ti:Ir ratios (compared to the layer composition) as the a significant fraction of the Ti signal would have originated from the underlying Ti substrate due to the penetration depth of EDS.

Zoom In Zoom Out Reset image size

We propose that the mechanism for these observed changes in morphology due to the addition of nanoparticles is related to differences in thermal expansion coefficient of composite and substrate and the improved drying characteristics of the precursor suspension. It is reported that the differences in thermal expansion coefficient of the layer and the substrate during the thermal decomposition method may be linked to the formation of cracks,⁹ presumably through induced mechanical stress on the coating. In our case, IrO₂ has a smaller thermal expansion coefficient (5 × 10⁻⁶ / °C²⁴) than Ti (1 × 10⁻⁵ / °C²⁵) which would explain the observed cracking, whereas the thermal expansion coefficient of TiO₂ (9 × 10⁻⁶ / °C^26,27) is closer to Ti, and thus the presence of the TiO₂ nanoparticles may reduce some mechanical stress. It is also possible that the composite has stronger mechanical strength as observed in some metal-ceramic composites.²⁸ We also speculate that the initial drying and decomposition of the precursor suspension plays a role in determining the final morphology, with many investigations revealing that the drying of suspensions leads to complex and interesting structures.^29,30 It is also likely that the presence of the TiO₂ nanoparticles provides the IrCl₃ (the IrO₂ precursor) nucleation sites from which a network of IrO₂ crystallites can grow.

Cyclic voltammetry of the fabricated anodes in 0.5 M H₂SO₄ solution revealed similar electrochemical behavior to the traditional IrO₂ DSA electrodes,^31–33 which suggests that the surface electrochemistry is most likely dominated by iridium species^34,35 (Figure 2).

Zoom In Zoom Out Reset image size

As expected, the anodic charge (a measure of the electrochemically active surface area^1,36,37) was found to decrease with the addition of TiO₂ nanoparticles due to the decreasing mass of IrO₂ in the active layer (Figure 3). This suggests that the TiO₂ nanoparticles do not help improve the utilization of the IrO₂ phase within the composite layer, unlike the addition of 10 wt% Sb-doped SnO₂ nanoparticles to IrO₂ which did increase the overall electrochemically active surface area.¹⁷ In this case, by calculating the anodic charge on an IrO₂ mass basis, it is found that the specific active area of the IrO₂ decreased upon additional of 10 wt% TiO₂ nanoparticles before increasing with further TiO₂ nanoparticle additions (Figure 3). We propose that the initial decrease in IrO₂ utilization is due to the reduced crack size, which would restrict the transport of electrolyte into the active layer.³⁶ Interestingly the anodic charge on an IrO₂ mass basis measured at 200 mV s⁻¹ is higher for the anode containing 40 wt% TiO₂ nanoparticles compared to the pure IrO₂ anode (Figure 3), which suggests that a great proportion of the Ir atoms are easily accessible³⁶ in this anode, presumably due to the particle-like morphology this anode exhibits. However, it is important to note that the TiO₂ used in this work has limited electrical conductivity, and it is likely that for thicker films, this will reduce the utilization of the IrO₂ due to ohmic losses between the film and the titanium substrate.

Zoom In Zoom Out Reset image size

The electrocatalytic performance of fabricated electrodes toward oxygen evolution reaction was investigated by linear sweep voltammetry in 0.5 M H₂SO₄ with a scan rate of 1 mV s⁻¹ (Figure 4). This shows that the onset of the oxygen evolution reaction begins at around 1.2 V vs. SCE for all electrodes, with Tafel slopes between 75–85 mV at low overpotentials, before curving upwards to Tafel slopes of 150–190 mV at higher overpotentials. The lower Tafel slope is similar to that report for IrO₂ in KOH³⁸ and the 60–90 mV reported on some Ir_xTi_(1-x)O₂ anodes in acid.³⁹ These slopes are significantly higher than that normally reported for IrO₂ in acidic electrolytes,^40–42 which may suggest that the OER is operating via a different mechanism on these anodes. This could be due to the low IrO₂ loading on the anodes used in this work (total oxide loading = 0.35 mg cm⁻²), with some evidence suggesting that low total loadings of the active phase can increase the Tafel slope for OER.^39,43,44 Regardless of this, very little difference is found between the anodes, with the difference in potential between the most active anode (containing 0 wt% TiO₂ nanoparticles) and least active anode (containing 40 wt% TiO₂ nanoparticles) only around 30 mV at useful current densities (>10 mA cm⁻²). This shows that the OER performance can be largely maintained even with the addition of up to 40% TiO₂ nanoparticles, with any small loss in performance likely only due to fewer IrO₂ sites within the anode. We expect this loss in performance could be minimized by increasing the anode's total oxide loading as the quantity of TiO₂ nanoparticles within the layer increases, to ensure that the IrO₂ loading remains approx. constant.

Zoom In Zoom Out Reset image size

The accelerated life tests of the fabricated anodes showed that while the lifetime of the IrO₂ anode containing no TiO₂ nanoparticles was only 17 h, the addition of 10 wt% and 25 wt% TiO₂ nanoparticles increased this by two and ten times respectively (Figure 5). The lifetime for the anode containing 40 wt% TiO₂ nanoparticles was much shorter at 2.7 h. This important result indicates that simply by adding an optimum amount of inert nanoparticles into a DSA layer can significantly increase the service life of these anodes. We interpret the short lifetime of the anode containing 40 wt% TiO₂ nanoparticles as being related to the poor coverage of the film over the titanium substrate which would have led to rapid passivation and possibly mechanical failure of the particle-like coating. Indeed, SEM analysis after this electrode had failed showed large regions of anode were no longer coated, suggesting that at least part of the failure mechanism was due to erosion of the coating. Given that the density of IrO₂ (11.67 g cm⁻³) is much higher than TiO₂ (4.25 g cm⁻³)^4,5 for the anode containing 40 wt% TiO₂ nanoparticles, the 60 wt% IrO₂ in the layer only corresponds to around 35 vol% IrO₂ which might not be enough to act as an effective binder for the TiO₂ nanoparticles.

Zoom In Zoom Out Reset image size

The deactivation of DSA electrodes is generally assumed to be due to erosion, corrosion and/or passivation of the underlying substrate.^6,46–48 Erosion occurs when the all or part of the active layer detaches from electrode substrate due to poor adhesion of the coating or mechanical stress, resulting from bubble formation at the cracks and pores of the coating. This differs to corrosion, where the active species are anodically oxidized to a soluble or volatile species during the OER.⁴⁹ Passivation is caused by the formation of insulating layer on the (typically) titanium substrate and is widely known as final stage of DSA deactivation, and often accelerates the failure of the anodes following the partial detachment or corrosion of the coating.^6,50 In this work, as the TiO₂ nanoparticles are unlikely to influence the corrosion rate of the active IrO₂ phase, the improved lifetime of the anodes containing 10 and 25 wt% TiO₂ nanoparticles must rise from improved mechanical strength and/or reduced electrolyte penetration through the layer leading to reduced substrate passivation. As the addition of 10 wt% TiO₂ nanoparticles resulted in noticeably smaller cracks in the layer compared to the pure IrO₂ anode, it is reasonable to propose that this anode has an improved lifetime due to less electrolyte penetration and substrate passivation. For the anode containing 25 wt% TiO₂ nanoparticles, no cracked-mud structure was found and instead the layer exhibited a particle-like morphology. While this layer is porous, the coverage of IrO₂ over the substrate was even, which suggests that electrolyte penetration to the substrate may be impeded. We also propose that the composite layer may have improved mechanical strength (similar effect to that found in other nanoparticle composites^18–20) which reduces the erosion rate under strong oxygen evolution reaction rates.

This study investigated the effects of addition of TiO₂ nanoparticles on service lifetime and electrochemical behavior of IrO₂ anodes. The accelerated service lifetime test show that the addition of TiO₂ nanoparticles into the oxide matrix enhances the durability of the anodes, with 10 wt% and 25 wt% TiO₂ nanoparticle additions increasing the service lifetime of the electrodes by 2x and 10x respectively. When 40 wt% TiO₂ nanoparticles were added, the anodes had much lower lifetimes due in part to poor coverage of the coating on the titanium substrate. Importantly, the simplicity of embedding the oxide nanoparticles into the thermally prepared IrO₂ layers should be compatible with standard industrial DSA production methods and maybe further optimisable to improve both service lifetime and electrochemically active surface area.

The authors thank Nathan Burns for the preparation of the anodes and conducting the accelerated lifetime tests reported in this work.

Aaron T. Marshall 0000-0002-3530-7251