Investigation to Minimize Electrochemical Impedance Spectroscopy Drift

Abstract

Electrochemical impedance spectroscopy (EIS) is a technique used to characterize physiochemical processes, especially in the field of biosensors. However, EIS has been known to have reproducibility issues due to an inherent drift. When taking repeated measurements of the same exact solution using EIS, impedance measurements have an increasing trend which can be detailed by a linear slope. The reported EIS drift ranges from 0.11 to 5.5 Ω/min. We studied the drift to assist with future data interpretation and model fitting. We discovered the cleanliness and treatment of the working electrode effects EIS drift, and the minimization of the drift can occur by rinsing the working electrode in-between repeated runs.

Introduction

Electrochemical impedance spectroscopy (EIS) is used for the study of batteries, fuel cells, corrosion energy applications and sensors due to the ability to characterize physiochemical processes.^1–4 During EIS, an alternating current potential is applied to the electrochemical cell, and sinusoidal current response is measured; the output of this method can be displayed as a Nyquist plot.

Previous reports included Nyquist plot shifts, higher charge transfer resistance with repeated measurements, but focused on the necessary control experiments needed to verify the appropriate signal.^8–12 As an example of biosensors, EIS has been used to determine nitrates in soil and protein binding events of antibodies.^5–7 Small charge transfer amounts, at rates in which drift is reported, can be incorrectly mistaken as significant data.^6,8 In other words, drift can be often interpreted as data when using serial dilutions used to detect analytes in solution.^8,9 Further, research is limited to minimize and understand EIS drift to only a few key seminal papers.¹⁴

Our study focuses on understanding how to minimize the Nyquist plot shifting. Since this is one directional, increasing impedance (or charge transfer resistance) over time, we call this an EIS drift. The reason for the EIS drift is not clear, but some theories include a ferricyanide reduction film or chloride film developing on the surface of the working electrode which could etch the surface.^11,13 Experiments for this theory were on a gold or platinum working electrode, and our experiments further expand this theory for a glassy carbon electrode.

We present our progress on experimentally minimizing EIS drift based on the previously proposed theories. Further, we present possible methods of reducing EIS drift through varying electrode polishing, base solution type, and electrode treatment. Using the data from these experiments, determinations were made for the most efficient ways to minimize drift.

Materials and Methods

Chemical and Reagents

Unless specified, all chemicals were purchased from Fisher Scientific and of analytical reagent grade. Phosphate buffer saline (PBS) solution was made in the laboratory and kept a pH of 7.4 using standard protocols. Ultra-high purity (UHP) waster was taken from a Millipore 3Q Direct UV system and recorded at 18.2 M MΩ.

Electrode preparation

A glassy carbon rod working electrode was used (BASi, 3 mm diameter) and polished with 0.05 μM alumina polish on an alumina polishing pad for 1-minute switching between clockwise and counterclockwise direction every 15 seconds before each serial dilution. The initial polishing protocol rinsed the electrodes with UHP water before and after polishing. The updated polishing protocol rinsed the electrodes with UHP water and methanol before and after polishing, and polished using a wetted alumina polishing pad.

Electrochemical Experiments

All electrochemical experiments were carried out at ambient room temperature in a faraday cage (Gamry Instruments VistaShield). A three-electrode system was used consisting of an Ag/AgCl/ 3M NaCl reference electrode (BASi), a platinum wire auxiliary electrode (BASi), and the glassy carbon working electrode (GCE).

EIS measurements were obtained using a Reference 600+ (Gamry Instruments) potentiostat. Unless otherwise specified, all experiments are N=3, with each independent run a new electrode polish/replica and not a rescan. Averages and standard deviations are reported for all cases. Raw data is available by request.

Before each experiment was performed, the solution was sonicated for 30 seconds. Unless specified, EIS was done using 20 mM [Fe(CN)⁶]^3−/4−(1:1) in PBS (FFCN).

EIS frequency scan range of 100 kHz to 0.1 Hz (with 10 points per decade), the DC offset was 0 V versus open circuit potential (E_oc), and the AC voltage was 5 mV.

Nyquist Analysis

EIS was analyzed using Gamry Echem Analyst, and the data was fit to a constant phase element with diffusion equivalent circuit shown in Figure 1.¹⁵ We analyzed the data to report only the charge transfer resistance, R_CT. None of the other variables were found to consistently drift linearity, and therefore, for simplicity, only this resistive measurement is reported to monitor the drift. The change in charge transfer resistance between a certain R_CT and the initial R_CT, (R_CT-R_CT0) was used to analyze a linear fit using Excel.

Figure 1.

Randle’s with a constant phase element (CPE) with diffusion equivalent circuit was used for analysis and fit with Gamry Analyst. Only the R_ct was used and reported as relevant.

In data analysis, as previously reported, the R_ct at the zero point was not considered because the difference in R_ct was significantly larger between 0 and 10 minutes when compared to that of other times.¹⁴ The R_ct versus time was fitted to detail the drift as linear fit from 10 to 50 minutes for easier comparison of the graphs.

Results and Discussion

Initial Experiments

An example of the raw data displayed as a Nyquist plot is shown in Figure 2; the drift can be visualized between individual measurements over time. Numerical values, such as linear slope of the normalized R_ct value from 10–50 min measurements, is simply easier to evaluate the magnitude of the drift.

Figure 2.

Nyquist plot representation of the data from 0, 10, 30, and 50 min scans after all three (working, reference, and auxiliary) are in the FFCN solution using the initial polishing technique. The data was converted to a charge transfer resistance (Figure 3) using the detailed equivalent circuit (Figure 1).

The analyzed data from the Nyquist plot is displayed in Figure 3; electrodes used in these plots were based on the initial polishing technique. A drift of 0.50 Ω/min was observed which was our starting point to minimize. While seemingly insignificant, this small drift could be mistaken for minimal changes in the surface in the observation of heterogenous analyte adsorption to the surface.

Figure 3.

EIS measurements with the initial polishing technique. The slope is 0.50 Ω/min. (N=2)

Potentiostat Control

We used a control to confirm the measured drift was entirely based on a solution phenomenon, and not equipment error. Our potentiostat was connected to a dummy cell with a known R_ct of 3005 Ω to confirm this would hold consistent. Figure 4 confirms that no drift present, and therefore, all the drift observed in Figure 3 was specific to something occurring in the solution.

Figure 4.

Experiment using a dummy cell. There was a slope of 0 Ω/min which means that there was no drift and that the drift seen in Figure 2 is due to solution effects. (N=1)

Updated Polishing Technique

Drift was minimized by changing the electrode polishing technique to include rinsing the electrode with both water and methanol before and after polishing, and to polish using a wetted alumina polishing pad. This proved to lower drift during experiments. The result, Figure 5, reduced the drift to 0.15 Ω/min, indicating a cleaner surface improves the EIS drift of the electrode. Future experiments only used this updated polishing technique, comparing subsequent figures to Figure 5.

Figure 5.

The new polishing technique was used, and all electrodes were kept in solution. EIS measurements were conducted every 10 minutes for 50 minutes. The slope is 0.15 Ω/min, which is much lower than the data in Figure 2, indicating a cleaner surface is needed for minimal drift.

Electrode Removal Experiments

Experiments were the reference, auxiliary, and working electrodes were conducted to see if drift could be further minimized (Figures 6 and 7). Previous experiments kept the electrodes in the solution in-between runs; in these experiments the electrodes were removed in-between runs. When removed, they were rinsed and soaked in UHP water.

Figure 6.

The working electrode was removed between EIS measurements and was placed in UHP water in-between runs. The drift was 0.11 Ω/min, which is the lowest slope seen in our experiments, and lower than the 0.21 Ω/min standard.

Figure 7.

In-between runs the auxiliary and reference electrodes were removed and placed in UHP water and 3 M NaCl respectively. The slope is 0.21 Ω/min, higher than the initial 0.15 Ω/min standard

Compared to the initial test case (Figure 4), which had a drift of 0.15 Ω/min, the drift was decreased to 0.11 Ω/min when the working electrode was removed and rinsed (Figure 6). Removing the auxiliary and reference electrodes (Figure 7), resulted in an increase in drift of 0.21 Ω/min. These experiments indicate that the drift is primarily caused by a phenomenon at the working electrode, as the removal and rinsing of the working electrode in-between runs results in a decrease. We are unsure why removing the auxiliary and reference electrode results in a higher drift, but one potential explanation is due to potential contamination from the 3 M NaCl solution used for the reference electrode.

Changing the Testing Solution

We manipulated the testing solution by decreasing the [Fe(CN)₆]^3−/4− concentration to 1 mM in PBS (Figure 8) and testing [Fe(CN)₆]^3−/4− in UHP (Figure 9). The data for 1 mM [Fe(CN)₆]^3−/4− in PBS had much larger error bars than the previous experiment (Figure 4,0.15 Ω/min) and a much larger EIS drift of 5.5 Ω/min. This result is counterintuitive to previously published results, and we do not yet have a working theory on the increase of impedance with lower [Fe(CN)₆] ^3−/4− concentration.²

Figure 8.

The [Fe(CN)₆]^3−/4−content was decreased to 1 mM in PBS. The EIS drift was 5.5 Ω/min, which is much higher than the data seen in[ in Figure 2, indicating that lower concentration increases drift. The large error bars indicate that lower concentration also decreases repeatability.

Figure 9.

New polishing technique experiment with 20 mM FFCN in deionized water. The slope is 0.28 Ω/min which is lower than the data seen in Figure 2, but still higher than the data seen in Figure 4. This indicates that deionized water increases drift more than PBS.

The drift in 20 mM FFCN in UHP water (Figure 9) was determined to be 0.28 Ω/min. This was also counterintuitive to literature which predicted that a decrease in chlorine concentration would decrease the overall drift.^10,11

Conclusion

Our data indicates that EIS measurements in an aqueous solution have inherent drift; reported in this data with a range of 0.11 to 5.5 Ω/min. Within this study, we focused on standard laboratory practices to minimize EIS drift. Overall, the best way to minimize drift is by focusing on the working electrode. By polishing the working electrode with better techniques, drift is significantly lowered. In addition, the drift is further minimized to 0.11 Ω/min when removing the working electrode and rinsing the electrode with UHP water in-between EIS runs. Other tested factors either showed no changes or increasing changes to the EIS drift. Even though we are not yet able to completely remove the inherent EIS drift, this data did show that more focus needs to be put on the working electrode if the cause of the drift is to be found. Future work to further minimize drift will focus on using higher concentration of FFCN in PBS or to use oxygen plasma in an attempt to further clean the working electrode surface. Additional work will be needed to explain the data response observed in the increase when FFCN concentration is decreased and changing the PBS buffered solution to UHP.