Stability of PEDOT:PSS‐Coated Gold Electrodes in Cell Culture Conditions

Poly(3,4‐ethylenedioxythiophene) doped with polystyrene sulfonate (PEDOT:PSS) is widely used as a coating on microelectrode arrays in order to reduce impedance for both in vitro and in vivo electrophysiology. In many applications, electrode performance of months to years is desired; yet, there are few studies to date that examine the long‐term stability of conducting polymers and their devices. Here, the stability of PEDOT:PSS microelectrodes is examined over a period of four months in cell culture media enriched with fetal bovine serum. The electrochemical impedance remains constant for most electrodes throughout the study, and only small changes in the structure of functional electrodes are observed. The results demonstrate that PEDOT:PSS electrodes show adequate stability for a variety of in vitro electrophysiology applications in toxicology, drug development, tissue engineering, and fundamental studies of electrically active cells and tissues.

10% fetal bovine serum and therefore, introduced the possibility of protein biofouling. Moreover, we used PEDOT:PSS that is deposited using spin coating from a commercial dispersion after addition of a crosslinker, a process that is consistent with large-scale manufacturing. Recently, this material formulation has demonstrated viability for clinical translation in first-inhuman studies. [8,23] To the best of our knowledge, no one has reported on the long-term performance of these solution-deposited PEDOT:PSS electrodes, despite their manufacturability and clinical relevance.
The stability of PEDOT:PSS-coated electrodes was investigated by performing electrochemical impedance spectroscopy (EIS) over the four-month incubation period in cell culture media. EIS is commonly used to determine electrode impedance, a measure that predicts the potential for high quality biological recordings. Importantly, EIS is performed with small voltages (10 mV) and therefore, the measurements do not have a substantial effect on the electrodes. The electrodes were part of a flexible array fabricated on parylene C with standard microfabrication processes as reported elsewhere. [24] The array contained 64 gold electrodes with a diameter of 20 µm and coated with the PEDOT:PSS dispersion (Figure 1a).
Incubation in cell culture media necessitated sterility in order to avoid bacterial or fungal growth that could confound results. Significant bacterial and fungal biofouling would not be observed in typical in vitro or in vivo applications. Therefore, we needed to develop a setup for sterile and stable EIS. Introducing external electrodes (e.g., Ag/AgCl, Pt) into the electrolyte for EIS created a source of potential contamination. Instead, a permanent counter electrode (CE) was incorporated. The setup consisted of a glass substrate with a large gold CE and the microelectrode array (working electrode, WE) was glued in place near it. In this two-electrode configuration, the CE was almost half a million times larger than the WE, ensuring that all the applied potential dropped at the WE/electrolyte interface.
A silicone well was fixed onto the glass substrate to confine the media and immerse all electrodes ( Figure 1b). Furthermore, the setup was designed while considering sterilization methods. These components brought together are all compatible with steam autoclave, a clinically used sterilization method. Previous work has shown that autoclaving PEDOT:PSS electrodes has little impact on electrical properties. [25] Immediately after autoclaving, a sterile petri dish lid was fitted on top of the well to establish a closed system that ensures sterility. Finally, in order to have stable and quick connections throughout the study, the CE electrode was soldered with a wire, and the electrode array was connected to a zero insertion force (ZIF) clip soldered with 10 wires (Figure 1c). Ten of the 64 electrodes were randomly chosen for EIS measurements based on the ZIF clip connection. The complete setup was incubated in cell culture conditions (37 °C and 5% CO 2 ) and was only removed to perform EIS and to refresh media. While some studies have utilized standard protocols [26] for accelerated aging by incubating devices at higher temperatures, physiological temperature was used here in order to avoid protein denaturation that could impact the nature of electrode fouling. Additionally, the conditions used here were identical to the ones used in in vitro electrophysiology, hence relevant to application. Before the start of the experiment, the electrode array was washed with deionized water to remove excess PSSand any low molecular weight compounds for stable baseline impedance measurement. All EIS measurements were conducted in cell culture media, and it was found that the spectrum did not greatly differ from the spectrum in phosphate-buffered saline solution ( Figure S1, Supporting Information).
The variation of electrode functionality with time was studied by examining the impedance at 1 kHz, the frequency corresponding to recordings of individual action potentials (single units) (Figure 2, Figure S2, Supporting Information). "Functional" electrodes were defined as those with an impedance less than 1 MΩ at 1 kHz and "non-functional" electrodes as those with any higher impedance values. Although this impedance threshold is generally regarded as a marker of electrode functionality, some have reported this is not always the case. [27,28] At the beginning of the study, seven of the ten selected electrodes were functional and had an average impedance of 27.8 kΩ ± 0.6 at 1 kHz (Figure 2a). Five of these seven electrodes (1,2,5,7,8) were functional throughout the entire period and showed low and stable impedance (25.5 kΩ ± 3.4 average impedance at 1 kHz at day 112; Figure 2b). After day 35, electrodes began to show slightly more variation in impedance. Interestingly, one of the non-functional electrodes became functional early in the study at day 14 and remained so for the rest of the period (electrode 9, Figure 2c). The other two non-functional electrodes (3 and 10) were non-functional from the start and remained so throughout the entire experiment. Two electrodes (4 and 6) fluctuated between functional and non-functional ( Figure 2c). When high impedances were observed, we ensured that there was no debris on top of the electrodes by forcefully pipetting the media to generate flow and observed the same impedance again. These impedance fluctuations, referred to as "blinking," have been reported before by others, in vitro [29] and when implanted in the brain. [30] Altogether, at any given timepoint six to seven of the ten electrodes were functional (Figure 2d).
Adv. Mater. Technol. 2020, 5,1900662  Electrode array is embedded in parylene C. B) Electrode array (working electrode, WE) glued to a glass slide with a large PEDOT:PSS-coated gold electrode (1 cm × 1.5 cm) serving as the counter electrode (CE). A silicone well was placed on top in order to confine the media around each. The three components adhered together were autoclaved for sterility. C) A sterile petri dish lid was fitted on the silicone well to maintain sterility with a closed system. The electrode array device was connected with a ZIF clip, soldered with wires for connection. The gold CE was also soldered with a wire. The entire setup was incubated in a cell culture incubator and was removed for EIS timepoints.
The impedance spectra of functional and non-functional electrodes were compared to that of a bare gold (no PEDOT:PSS) electrode and that of an open circuit (WE disconnected) to determine delamination and connection failures, respectively ( Figure S2a, Supporting Information). A functional, PEDOT:PSS-coated electrode displayed a typical spectra consisting of a relatively flat, resistance-dominated region and a negatively sloped, capacitance-dominated region. [31] A gold electrode of the same diameter had a higher impedance that was capacitance dominated. The open circuit spectra showed higher impedance than the gold electrode, and furthermore, displayed erratic impedance values at 100 Hz and below. By examining each individual electrode, all non-functional EIS measurements matched that of the open circuit configuration. Therefore, we believe that when electrodes were non-functional, including at the start or later in the study, this was due to connection issues. More specifically, we believe these connection issues arose at the ZIF clip-parylene device interface. Manual alignment between ZIF clip pins and device contact pads as well as these rigid pins exerting significant pressure on a thin, flexible substrate could have caused poor contact. Reliable connections for flexible devices remain a pressing challenge in the field, and from our experience here, we suggest chemical bonding, instead of mechanical means, as well as the use of alignment tools and automated processes to improve this aspect.
The long-term behavior of the functional electrodes was further evaluated by examining the average impedance spectrum (Figure 3a). Generally, neural electrophysiological recordings are performed for capturing local field potentials and single action potential activity, which correspond to 10 and 1000 Hz frequencies, respectively. At both frequencies, impedance of functional electrodes slightly varied with time (typically 5-10% of the initial value) (Figure 3b). Slight impedance increases at 1000 Hz were observed up until day 21 when the average impedance was approximately 9% higher than the initial value. After day 21, the impedance began to decrease for the remaining period. The final (day 112) average impedance values of functional electrodes were approximately 17% and 8% lower than the initial value at 10 Hz and 1000 Hz, respectively. In summary, the small standard deviations among the functional electrodes indicated homogeneous properties.
After four months of incubation, the electrode array was washed with deionized water and dried, and all of the 64 electrodes were examined by optical microscopy and scanning electron microscopy (SEM) (Figure 4, Figures S3-S5, Supporting Information). For comparison, images of an electrode directly after fabrication were taken as well. Forty-six out of the 64 electrodes (>70%) showed no visible signs of degradation, and these include the impedance-fluctuating electrodes (4,6,9). The PEDOT:PSS material was intact and covered the entire electrode area. Furthermore, nearly all of the examined surface area of the parylene C encapsulation did not show any apparent damage ( Figure S3, Supporting Information).
Of the electrodes that showed damage, four showed partial delamination of cracked PEDOT:PSS ( Figures S3 and S5, Supporting Information). Delamination of PEDOT:PSS electrodes has been reported before. [18] Delamination is especially believed to be a consequence of poor material binding between the conducting polymer and underlying substrate, [32] and therefore, physical [19,33] and chemical [34] means of adhesion are Adv. Mater. Technol. 2020, 5, 1900662 Figure 2. Expected functionality analysis of individual electrodes (N = 10) with time. Numbers correspond to different electrodes. Functional electrodes defined as those < 1 MΩ at 1 kHz. Three of the selected electrodes were non-functional at fabrication; two of these remained so throughout the study while one electrode became functional. Two electrodes varied between functional and non-functional throughout the study. A) All electrodes (10 in total) studied. B) Only electrodes with good and stable performance (five in total). C) Only electrodes with variable functionality (three in total). D) The number of functional electrodes over the course of the study. At any given timepoint, 6 to 7 of the 10 selected electrodes were functional.
recommended. In our case, we have used GOPS as a crosslinker [35] of the PEDOT:PSS and to help with substrate adhesion; although, it is not expected that there is chemical bonding between this polymer and metal.
Observed by light microscopy, the color of three electrodes changed considerably ( Figure S3a, Supporting Information). One electrode showed a ruptured PEDOT:PSS-parylene C interface ( Figure S5, Supporting Information), and one electrode showed some sort of PEDOT:PSS corrosion ( Figure S4, Supporting Information). Finally, nine electrodes appeared to have lost one of the two layers of deposited PEDOT:PSS as determined by both light microscopy and SEM ( Figures S3 and S5, Supporting Information). These material failures were clustered in the central part of the array, and thus, could be due to flaws from the fabrication process like contamination or inhomogeneous oxygen plasma treatment ( Figure S3, Supporting Information).
A potential fragile aspect of the electrodes is the PEDOT:PSSparylene C interface, where we observed slight separation between the materials, especially for the discolored electrodes ( Figures S4 and S5   Slight decreases in impedance were observed at mid (1000 Hz) and low (10) frequencies, corresponding frequencies of action potential (single-unit) and local field potentials, respectively. operation in aqueous conditions and significant PEDOT:PSS swelling, [36] it is unclear whether or not this separation would be present.
Among the functional electrodes (1,2,5,7,8), as determined by EIS above, there were no major signs of damage but a couple subtle differences were observed (Figure 4, Figure S4, Supporting Information). The aged electrodes appeared to have a slightly darker color, and the PEDOT:PSS coatings seemed to have increased in volume as well as surface roughness. Also, there were indications of some debris, visible along the periphery of the electrodes. Despite these slight differences, none of these visible changes had influenced the electrical properties. Aside from electrode 3, which showed delamination of the first PEDOT:PSS layer, the unstable (4,6,9) and nonfunctional (10) electrodes did not show any apparent signs of material damage, which supports our conclusions that nonfunctional electrodes were mostly due to connection issues ( Figure S4, Supporting Information).
In conclusion, we demonstrated that PEDOT:PSS-coated gold electrodes were largely stable in cell culture conditions for at least four months. When high impedance values were recorded (non-functional), connection issues were determined to be the main cause. To the best of our knowledge, this is the first report on the long-term stability of PEDOT:PSS electrodes prepared from solution using a scalable manufacturing technique. These results are of consequence to in vitro electrophysiology for toxicology, drug development, tissue engineering, and fundamental studies of electrically active cells and tissues. Such electrodes will be useful for long-term electrophysiology of cell and tissue cultures. The presence of cells in vitro is not expected to drastically change the results found here. These results are also promising regarding the potential of these electrodes for chronic use in implantable medical devices.

Experimental Section
Array Fabrication: The fabrication of the electrode array has been reported elsewhere. [24] Briefly, 2 µm parylene C was deposited by a SCS Labcoater 2 on a clean glass slide. Metal electrodes and connection leads were patterned using a lift-off process with a bi-layer of LOR5A resist and S1813 photoresist. Photoresist was exposed with a SUSS MBJ4 contact aligner. An adhesion layer of 10 nm chromium and 150 nm of gold was evaporated with a Boc Edwards thermal evaporator. After lift-off, a 2 µm parylene C insulation layer was deposited with 3-(trimethoxysilyl)propyl methacrylate (A-174 Silane) as an adhesion promotor. The outline of the probe was etched with an Oxford Plasmalab 80 Plus (reactive ion etcher) using lithographically patterned AZ9260. The electrodes were coated with PEDOT:PSS by the following process. A soap layer was spin-coated before the deposition of a 2 µm sacrificial layer of parylene C. AZ9260 was spin coated, exposed, and developed with AZ developer and subsequently, the parylene C was etched. A mixture of Heraeus Clevios PH1000 (aqueous colloidal dispersion of chemically polymerized PEDOT:PSS), ethylene glycol, dodecyl benzene sulfonic acid and (3-glycidyloxypropyl) trimethoxysilane was spin-coated two times (3000 and 1500 rpm) with a one minute bake in-between at 110 °C. The sacrificial parylene C layers were peeled-off followed by a 1 h bake at 140 °C. Finally, the device was washed in deionized water to remove excess low molecular weight compounds and to delaminate the device from the glass slide.
Setup Assembly: A gold counter electrode (CE, 1 cm × 1.5 cm) was patterned on the glass substrate by using a polyimide foil mask and thermal evaporation of chromium and gold (10 and 150 nm, respectively). The CE was coated with the same PEDOT:PSS formulation as the electrodes of the array to minimize the CE/electrolyte voltage drop. The flexible electrode array was supported by polyimide foil except at the tip. The electrodes were at the tip of the array and the tip was glued near the counter electrode to keep it in place throughout the experiment. The silicone well was secured to the glass by gluing it with polydimethylsiloxane (PDMS, Dow Corning Sylgard 184). After sterilization, a wire was soldered to the counter/reference electrode, and wires were soldered to a zero insertion force (ZIF) clip that was connected to the electrode array.
Array Sterilization and Incubation: The array, large gold electrode, and silicone well (PDMS, Dow Corning Sylgard 184) were sterilized by steam autoclave for 20 min at 121 °C and 220 kPa (no dry time). Media (Dulbecco's Modified Eagle Medium with phenol red and 10% fetal bovine serum) was refreshed once or twice weekly in a cell culture hood with aseptic technique. The device was incubated at 37 °C/ 5% CO 2 at all times, except for EIS.
Electrochemical Impedance Spectroscopy: EIS measurements were performed with an Autolab PGSTAT128N where the PEDOT:PSS-coated electrodes from the array were the working electrodes and the large 1 cm × 1.5 cm electrode was the counter electrode. A 10 mV sinusoidal voltage was applied at a frequency range of 1 to 100 000 Hz.
Scanning Electron Microscopy: Scanning electron micrographs of the electrode array were acquired with a Carl Zeiss Ultra 55 after depositing 5 nm of gold-palladium on the electrode array with a Gatan 682 Precision Etching and Coating System.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.