Optimization and electrochemical characterization of RF-sputtered iridium oxide microelectrodes for electrical stimulation

Nichiden Anelva, SPF-210H sputtering equipment was deployed to deposit iridium oxide thin films, as shown in figure 1. All films were deposited on the silicon wafer with Ti adhesion layer by the reactively sputtered process in Ar/O₂ plasmas. The Ti adhesion layer with 50 nm thickness was sputtered with 10 sccm (standard cubic centimeters per minute) Ar flow under the chamber pressure of 2.4 Pa. Before the deposition process of the iridium oxide, an Ar sputter process was performed to remove the potential oxide layer on the Ti adhesion layer. The RF-power worked at 13.56 MHz, 150 W and was coupled with the target electrode via the manual impedance matching network. The manual impedance matching network could adjust the reflected power from 0.05 to 1 of the RF-power, and the plate current was related to the reflected power. When the reflected power was kept at 10 W, the plate current was in the range of 110 to 120 mA. The system was continually cooled with circulating water.

Zoom In Zoom Out Reset image size

Prior to deposition process, the sputtering chamber was evacuated to 2×10⁻⁴ Pa by an oil-sealed mechanical pump and a vertical turbo molecular pump. The wafer was not artificially heated to prevent the occurrence of the carbonization of the photoresist. The flows of Ar and oxygen were controlled by the gas mass flow control meters. A 10 sccm Ar flow was introduced, and the chamber pressure was adjusted to 2.4 Pa by the main valve. At the same RF-power, the amount of the sputtered iridium atoms is controlled by the Ar flow. The constant argon flow creates the constant amount of the sputtered iridium atoms. The oxidization of the sputtered iridium atoms is controlled by the flow of the oxygen, and the oxygen flow was selected from 5 to 40 sccm using a mixture of Ar and O₂ (from 1:0.5 to 1:4). While the oxygen flow was turned on, the main valve was kept at the position of 2.4 Pa, and the sputtering pressure was from 2.5 to 5.4 Pa.

Figure 2 shows the fabrication process of the iridium oxide microelectrodes. A 5 µm thick positive photoresist (PR) layer was spun on the silicon wafer (3 inches in diameter, 100 crystal orientation), as shown in figure 2(a). The PR (AZ4330) was spun at 1500 r min⁻¹ (30 s) and baked at 50 and 90 °C for 1 h, separately. Next, the PR was patterned by the ultraviolet (UV) light exposure and developed (figure 2(b)). Then a Ti adhesive layer and iridium oxide (IrO_x) thin film were sputtered and patterned by a lift-off process (figures 2(c) and (d)). 5 µm thick Parylene-C film was deposited on the silicon wafer by the chemical vapor deposition process (figure 2(e)). A 10 µm thick PR was spun (figure 2(f)), patterned (figure 2(g)) and stored at 65 °C for 1 h in a vacuum baking oven. At last, the microelectrode site and bonding pad were formed by a reactive ion etching system (Nextral 100, Reactive Ion Etcher, Alcatel, France) (figure 2(h)). The fabricated iridium oxide electrode was shown in figure 3.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The morphologies of the samples were observed by a field-emission scanning electron microscope (SEM, Carl Zeiss Ultra55), as shown in figure 3. The diameter of the microelectrode site is 100 ± 0.5 µm (mean±sd n = 8). Its corresponding surface area is calculated as 7850 ± 79 µm². Atomic force microscopy (AFM, Nanoscope III, Digital Instr.) was also performed to obtain the surface roughness of the samples. The CSC of the iridium oxide microelectrodes was calculated by the cyclic voltammetry (CV). The CV was measured in a physiological saline solution (0.9% NaCl) by the standard three electrodes system with an Ag/AgCl reference and a Pt-sheet counter electrode.

The applied potential for activation was swept from −1.0 to 1.0 V at 0.05 Hz for 100 cycles [24]. After activation, six cycles were collected for each CV curve, with a potential range of −0.6 to 0.8 V to keep the voltage in the 'water window'. The 'water window' is a voltage range that neither hydrogen nor oxygen is produced from the water in the measurement. The CSC of the iridium oxide microelectrodes was measured under the scan rate of 100 mV s⁻¹ [16]. Electrochemical impedance spectroscopy (EIS) was measured at the same electrochemical cell at the same time. The amplitude of sinusoidal signal is 10 mV and its frequency range varies from 0.1 to 100 kHz. The image of the fabricated iridium oxide electrode is shown in figure 3. To improve the validity of comparing electrical impedance based on changes in the sputtering process, we have utilized a custom developed tool (figure S1, Supplementary data are available from stacks.iop.org/JMM/24/025015/mmedia) to ensure consistent contact area for each of the tested microelectrodes. Voltage transient measurements were used to evaluate the voltage following property and stability by a current-controlled stimulation pulse. The voltage transient was recorded by the configuration with three electrodes, including large-area Pt-sheet return electrode and a noncurrent-carrying Ag/AgCl reference electrode.

Oxygen flow and its partial pressure are two key parameters in the process of the reactively sputtered iridium oxide. The oxygen pressure under different oxygen flow could imply the target and film characteristics of the sputtering process, as shown in figure 4. It can be seen that the initial oxygen pressure is about 0.1 Pa only, which shows that almost all oxygen participates in the oxidation reaction of iridium. The oxygen pressure increases obviously when the oxygen flow arrives at 15 sccm. Based on the insert figure in the figure 4, the nonlinear relationship between oxygen flow and pressure implies that little oxygen takes part in the oxidation reaction, which is also called the 'target poisoning process'. During this deposition process, the initially produced iridium oxide is absorbed on the surface of the iridium target which prevents oxidation reactions. As a result, the accumulating oxygen exists in the sputtering chamber contributes to the increasing of the pressure rapidly. While the oxygen flow is larger than 20 sccm, there is a good linear relationship between the oxygen pressure and flow. It shows that the oxidation reaction proceeds and all sputtered iridium is transformed into the oxidation state.

Zoom In Zoom Out Reset image size

Figures 5 and 6 show SEM and AFM images of iridium oxide films under different oxygen flows, respectively. The morphology and surface microstructural parameters of these samples are summarized in table 1. Although the surface of the film at 10 sccm is very rough, the cross section SEM in figure S2 (Supplementary data are available from stacks.iop.org/JMM/24/025015/mmedia) shows that the film has a continuous layer structure. The iridium at 15 sccm oxygen flow cannot be oxidized further, which results in the surface roughness of 17.1 nm. Although a dendritic surface microstructure is developed at 25 sccm, the film obtains good adhesion property to the substrate. Finally, a dendritic surface with increasing aggregation is produced at 40 sccm oxygen flow. The aggregation is also validated by the AFM data shown in figure 6(d).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 7 shows CV curves of sputtered iridium oxide thin film microelectrodes under oxygen flows varied from 5 to 40 sccm. It can be seen that the electrochemical behavior of microelectrode highly depends on the oxygen flow and its pressure.

Zoom In Zoom Out Reset image size

For the case of low oxygen flow, only part of the iridium in the reactively sputtered thin film is under oxidation state. Since there is metallographic phase iridium existing in the film, the CV shapes are narrow and the redox peaks are broad. When the oxygen flow arrives at 25 sccm, all iridium is changed to the oxidation state and shows a wide CV shape. Meanwhile, the sharp peaks disappear due to the high electrochemical activity of the redox reactions, which results in large pseudo capacitance and a large CSC [1, 11–13, 25–30]. It is also observed from figure 5(c) (25 sccm) that this film can be easily permeated by water and ionic species, which contributes to higher charge delivery [16]. The maximum CSC of 36.147 ± 1.015 mC cm⁻² occurs at the oxygen flow of 25 sccm, as shown in figure 8.

Zoom In Zoom Out Reset image size

Along with the further increase of oxygen flow and pressure, the aggregation phenomenon of the film surface microstructure is obvious. The oxidation state iridium with high valence results from the superfluous oxygen flow and pressure. In this case, the film is electrochemically unstable. The iridium at the high valence oxidation state is transformed into the stoichiometric state during the activation process. Thus, the distinguished peaks in the CV curves are obtained. The aggregation of iridium oxide on the substrate shown in figure 5(d) leads to the decline of the CSC. CSC after 100 cycles of activation is measured and shown in figure 8. The measurements in figure 8 are taken from five samples (each oxygen flow) under the same process run. The maximum CSC occurs at the oxygen flow of 25 sccm. At that point the oxygen pressure is 1.8 Pa while the chamber pressure is 4.2 Pa (the ratio of O₂:Ar pressure is 3:4). It should be noted that the point is also the second point after the so-called target poisoning process. The 15 sccm point is the target poisoning point, and the 20 sccm point is the transition point. This means the maximum CSC occurs just at the start of the compound mode [16]. At this state, the reactively sputtered iridium oxide process consumes the least oxygen and the appropriate oxygen pressure to turn all the metallic iridium into the iridium oxide.

The minimum consumption of oxygen generates the minimum released heat during the exothermic reaction. At an oxygen partial pressure of 1.8 Pa (oxygen flow is 25 sccm); the amount of heat absorbed by the PR is acceptable to avoid the carbonization. Thus, the reactively sputtered iridium oxide thin film can be patterned by the lift-off process. There was no evidence that a baking temperature less than 65 °C and a baking time less than 1 h could prevent the carbonization of PR. In the standard lift-off process, the wafer was dipped into acetone and immersed into an ultrasonic bath until the PR was completely removed. After the sputtering process, the fully carbonized PR, partially carbonized PR and normal PR existed on the wafer at the same time. It should be noticed that the fully carbonized PR could not be removed by the lift-off process. Thus, the lift-off process was focused on the partially carbonized PR and normal PR. The higher the carbonization degree reaches, the longer the immersion duration. The measurements in figure 9 are taken from six samples (each oxygen flow) from the same process run. The immersion duration required to complete this process was impacted by the sputtering process parameters used for metal deposition (figure 9).

Zoom In Zoom Out Reset image size

For the O₂ flow less than 20 sccm, the film pattern can be well realized. However, the metallographic phase iridium exists in the film during the process, which results in the relatively poor electrochemical property. When the O₂ flow is larger than 20 sccm, all iridium is transformed into the oxidation state. But as the further increasing O₂ flow, the immersion duration gets longer. As a result, the defects of patterns will increase. Compared with the other sputtering conditions, the pattern under 25 sccm O₂ flow is realized at the lowest duration of ultrasonic agitation with 7.25 ± 0.49 min.

The obtained EIS under the oxygen flow from 5 to 40 sccm is shown in figure 10. The impedance of each curve varies in a smaller range and they are enclosed in a small envelope. It shows that there are different electrochemical activities of the reactively sputtered iridium oxide thin film for all curves. In particular, the impedance magnitude at the low-frequency (below 10 Hz) and high-frequency (over 10 kHz) range decreases when the oxygen flow varies from 5 to 20 sccm. After that, it shifts to increase around the 25 sccm oxygen supply. This result is in accordance with the changing tendency of the CSC shown in figure 8.

Zoom In Zoom Out Reset image size

In the mid-range-frequency (10 Hz–10 kHz), the changing tendency of the CSC cannot match the impedance. This is attributes to that the conductivity of the iridium oxide depends on the electrode potential. The electrochemical activity of the iridium oxide is not only affected by the deposition condition and electrochemical activation process, but also the working voltage [31–35]. Because the working voltage during EIS measurement is 10 mV, the electrochemical activity is in a smaller voltage range than the 'water window' region for the CSC measurement. Therefore, the changing tendency is not in accordance with the CSC in the mid-range-frequency.

The microelectrode impedance is dominated by the double layer capacitance (C_DL) at the low-frequency. The capacitance C can be calculated from |Z| = (ωC)⁻¹ at the frequency ω = 1 s⁻¹ [16]. Based on the impedance magnitude spectra in figure 10, the capacitances are calculated and shown in figure 11. The measurements in figure 11 are taken from five samples (each oxygen flow) from the same process run. It is obtained that the maximal double layer capacitance and CSC occur at the oxygen flow supply of 25 sccm and pressure of 1.8 Pa. The changing tendencies of the double layer capacitance and CSC are similar. It shows that the double layer capacitance could be employed as an indirect measurement of the electrochemical activity for the sputtered iridium oxide. In other words, lager C_DL means lager CSC.

Zoom In Zoom Out Reset image size

An example of the voltage transient in response to a biphasic, symmetric current pulse is shown in figure 12. The current pulse amplitude is 100 µA, and the pulse width includes 100 µs, 200 µs and 400 µs. Figure 12(a) shows the voltage transient of the 1 kΩ pure resistance. The response time is defined as the shift time from the negative phase to positive phase of the voltage response. The current pulses were applied by the Master-8 programmable pulse-generator (Master-8 8-channel programmable pulse-generator, two channels with ten-turn dial, Israel). The biphasic pulse was obtained by a negative pulse and a delayed positive pulse. The response time of the pure resistance was zero when the delayed time was equal to the width of the negative pulse.

Zoom In Zoom Out Reset image size

The figure 12(b) shows the voltage transient of the reactively sputtered iridium oxide thin film microelectrodes fabricated at the oxygen flow of 25 and 40 sccm. Firstly, the response time under the short pulse width of 100 and 200 µs is almost indiscernible. When the pulse width increases to 400 µs, the response time at 25 sccm arrives at about 80 µs. However, the response time of the film at 40 sccm is 140 µs, which is about 75% more than the one at the optimal condition. The testing results demonstrate that the sputtered iridium oxide thin film deposited at 25 sccm for the microelectrode shows relatively good electrical property and stability. It plays an important role to simplify the circuit design in the functional electrical stimulation system [36–39]. Secondly, the polarization of the iridium oxide thin film microelectrode fabricated at the optimal condition (25 sccm) is 35% lower than the one of 40 sccm. The lower polarization could result in the lower energy consumption, which is suitable for the implantable functional electrical stimulation system.