Robust and flexible organic electrochemical transistors enabled by electropolymerized PEDOT

High-performance OECTs based on electropolymerized small-anion doped PEDOT

The main challenge in fabricating electropolymerized OECTs is depositing the channel material onto insulating substrates. This can be addressed by short-circuiting metal source-drain electrodes, spaced by a moderately small gap (typically less than 50 µm), and using them as a single working electrode during electropolymerization⁴³. This setup enables the formation of a continuous channel by bridging the insulating gap between the electrodes with the electropolymerized material. Other approaches include using alternating current to promote fiber formation along the direction of the electric field, eventually creating dendritic networks^44,45, and precoating the channel area with solution-processed PEDOT^46,47.

We electropolymerized PEDOT:ClO₄ films onto short-circuited source-drain electrodes patterns to fabricate OECTs (Fig. 1a) using varying concentrations (1, 5, 100 mM) of the dopant LiClO₄. All devices exhibited p-type behavior and operated in depletion mode, as expected for PEDOT-based OECTs¹⁸. Notably, electrolyte concentrations of 5 mM and 100 mM both resulted in complete channel coverage, yielding OECTs with similar electrical characteristics (Supplementary Fig. 1). In contrast, a 1 mM concentration was insufficient to bridge source and drain electrodes, as no significant current was detected during electrical measurements. We consistently used high concentration of 100 mM in this study, as it is commonly employed in electrochemical systems for its high ionic conductivity and reduced ionic migration⁴⁸.

The morphology of PEDOT:ClO₄ films electrodeposited at charge densities of 20 (Fig. 1c, f), 30 (Fig. 1d, g), and 60 (Fig. 1e, h) mC/cm² was investigated using scanning electron microscopy (SEM) and atomic force microscopy (AFM). As expected, electropolymerization initiated at the Au electrodes and progressively extended across the channel area, with growth occurring both laterally and vertically relative to the electrode surface. This behavior aligns with established nucleation-growth mechanisms, resulting in a progressive increase in surface roughness (root mean square roughness, R_q) at longer deposition times⁴⁹. At an early stage (20 mC/cm², 40 s), the film primarily formed on the source and drain electrodes, with limited extension into the adjacent channel regions. The resulting morphology displayed an R_q of 8 nm (Fig. 1c, f). Increasing the deposition charge (30 mC/cm², 60 s) led to full channel coverage, yielding a granular network on the Au electrodes (R_q = 21 nm, Fig. 1d, g; Supplementary Fig. 2a) and densely packed aggregates across the channel (Supplementary Fig. 2b). Further increasing the deposition charge (60 mC/cm², 120 s) led to continued film growth atop the existing PEDOT layer, producing a rougher granular morphology across both the electrodes and channel (R_q = 38 nm, Fig. 1e, h; Supplementary Fig. 2c). Although the morphological image does not show a complete connection between the source and drain electrodes at low deposition charge (20 mC/cm², Fig. 1c, f), localized conductive pathways may lead to devices showing a low source-drain current and a limited performance in terms of transconductance (0.1 mS) and ON/OFF ratio (2 × 10¹) (Supplementary Fig. 3a, d). When the channel was fully covered, at 30 mC/cm² (Supplementary Fig. 3b, e) and 60 mC/cm² (Supplementary Fig. 3c, f), the devices displayed similar performance with high transconductance (above 10 mS), and an ON/OFF ratio around 1 × 10³, despite minor changes in morphology (Supplementary Fig. 2). Electrodeposited OECTs obtained with other small dopants (Supplementary Fig. 4), such as LiBF₄ and TEABF₄, frequently used for PEDOT electrodeposition, showed similar transconductance (~10 mS) and ON/OFF ratios (~4 × 10²), indicating that various combinations of small anion dopants and accompanying cations do not significantly impact the electropolymerization process, consistent with previous findings⁵⁰. Overall, our results show that electrodeposition produces highly reproducible devices with performance comparable to state-of-the-art electropolymerized OECTs (Supplementary Table 1).

Devices with similar steady-state characteristics (Fig. 2) were obtained using the larger polymeric dopant polystyrene sulfonate (PSS⁻), both through electrodeposition (g_m~10 mS, ON/OFF ratio ~6 × 10²) and inkjet printing from a commercial suspension (g_m~9 mS, ON/OFF ratio ~7.4 × 10²) (Table 1). Notably, the drain current (I_D) saturated at ~V_G = 0 V for PEDOT:ClO₄ OECTs (Fig. 2a, b, and Supplementary Fig. 5a, d), but at a slightly negative V_G for the electropolymerized (Fig. 2d, e, and Supplementary Fig. 5b, e) and printed (Fig. 2g, f, and Supplementary Fig. 5c, f) PEDOT:PSS OECTs. This indicates that, at V_G = 0 V, PEDOT:ClO₄ films are fully doped, while PEDOT:PSS films are partially doped. This is further supported by UV-Vis spectroscopic measurements, which are discussed in detail in the following section. Therefore, PEDOT:ClO₄ devices can operate with lower energy consumption. These findings demonstrate that the ClO₄⁻ dopant promotes a fully doping state in PEDOT films, likely due to the more efficient insertion and redistribution of the smaller dopant ions during electrodeposition. Remarkably, all devices retained similar performance after 4 months of storage in ambient conditions (Fig. 2b, e, h).

Despite their similar steady-state performance, the devices exhibited different stability under cyclic pulsed V_G application. To assess cycling stability, we monitored current responses during gate voltage pulses (V_G = 0.8 V, pulse width = 5 s) over time. For PEDOT:ClO₄ OECTs, the current (I_D) remained nearly constant, with only a 2% decrease after 1000 cycles (Fig. 2c). Electrodeposited PEDOT:PSS-OECTs showed a roughly 20% decline in I_D (Fig. 2f). A larger I_D drop of about 50% was observed in printed devices (Fig. 2i). The reduced cycling stability of PEDOT:PSS is primarily attributed to the greater swelling of its hydrophilic PSS chains^51,52,53. In inkjet-printed devices, this effect is more pronounced due to the excess PSS required to stabilize commercial PEDOT:PSS dispersions. The higher PSS content increases film hydrophilicity, leading to enhanced swelling in aqueous environments and reduced operational stability. This observation is consistent with our previous findings: solution-processed PEDOT:PSS films showed a significant weight increase after 10 min of water immersion, whereas PEDOT:ClO₄ films exhibited minimal change⁵⁴. Beyond swelling, differences in channel/electrode interface quality, arising from distinct fabrication methods, may also contribute to the observed performance variations. Overall, electropolymerized OECTs demonstrated superior stability compared to printed devices, particularly when the small dopant anion ClO₄⁻ was used. While inkjet printing is widely used for fabricating flexible electronic devices, the advantages of electropolymerization underscore its promising potential for the development of flexible OECTs.

Electropolymerized PEDOT:ClO₄ on parylene substrate for flexible and lightweight OECTs

Flexible and lightweight OECTs were fabricated by electropolymerizing PEDOT:ClO₄ films onto source-drain electrodes patterned on a thin (10 μm) parylene layer deposited on a glass substrate. At the end of the fabrication process, the parylene layer was peeled off from the glass to yield ultrathin flexible devices (Fig. 3a). Compared to conventional flexible substrates such as polyimide or polyethylene terephthalate (PET), ultrathin parylene films significantly reduce overall device weight and allow conformability to small-radius curved surfaces, facilitating mechanical flexibility testing. The ability of a material to conform to curved surfaces is governed by its bending stiffness, which is proportional to the Young’s modulus (E) and the cube of its thickness (t³). For a given material, where the Young’s modulus is constant, reducing the thickness significantly lowers the bending stiffness, allowing it to bend more easily and conform to surfaces with smaller radii of curvature. In the flat state, the flexible devices exhibited a transconductance of 7.6 mS and an ON/OFF ratio of 2.5 × 10² (Fig. 3b), slightly lower than those of the rigid counterparts, likely due to variations in PEDOT film growth influenced by differences in gold electrode surfaces across substrates (Supplementary Fig. 6). Importantly, the transfer characteristics remained nearly unchanged under mechanical deformation applied either parallel (Fig. 3c) or perpendicular (Fig. 3d) to the channel length, demonstrating robust flexibility in both directions. After recovery from bending, the devices retained a transconductance of 7.4 mS and an ON/OFF ratio of 2.3 × 10², confirming stable performance after mechanical strains. Further evaluation of mechanical durability and long-term reliability in future will be important for advancing their practical application in flexible electronics.

Electrochromic behavior of PEDOT:ClO₄ films during doping/dedoping monitored by UV–Vis spectroelectrochemistry

UV–Vis spectroelectrochemistry has been widely used to study changes in doping state of conducting polymers during electrochemical doping and dedoping processes⁵⁵. For our study, we monitored the UV–Vis absorption spectra of electrodeposited PEDOT:ClO₄ films in real time while applying cyclic linear voltage sweeps between −1.0 and 0.4 V, effectively capturing the dedoping (Fig. 4a) and redoping (Fig. 4b) processes. Consistent with prior studies on PEDOT films⁵⁶, we observed a characteristic absorbance band in the 400–700 nm range, which corresponds to the π–π* transitions of PEDOT neutral chains. Additionally, a second band appeared in the near-infrared range (700–1100 nm), associated with polaron absorption. The bipolaron states, which are typically located in the far infrared range, were not captured in our spectra. To analyze these changes more closely, we monitored selected wavelengths over time (Fig. 4c lower panel): the 621 nm curve (blue line), representing neutral chain signatures, increased significantly as the potential shifted from 0.4 to −1.0 V (Fig. 4c top panel). Simultaneously, the 801 nm (pale red line) and the 1000 nm (red line) curves, corresponding to the polaron band, decreased, consistent with the dedoping of the polymer film. During the redoping process, within the polaron absorption region (800 nm and 1000 nm), a unique behavior was observed near the highly doped state (around 0.4 V). The 800 nm peak displayed a dip, while the 1000 nm peak reached maximum intensity. In contrast, the 621 nm signal continued to decrease. These features suggest that, while the number of neutral chains decreases during redoping, the polarons do not increase to their maximum concentration but instead transition into other species, possibly bipolarons. This observation aligns with the idea that, under highly doped conditions, some polarons convert to bipolarons, indicating complex dynamics in the electronic states of PEDOT under different electrochemical conditions.

Similar results were observed for films electrodeposited from a 5 mM solution (Supplementary Fig. 7), indicating that dopant concentration has minimal impact on the electrical modulation of electropolymerized PEDOT films. This finding aligns with the comparable performance observed in OECTs, reinforcing that variations in dopant concentration do not significantly influence the electronic properties and functionality of these films.

UV–Vis spectroscopic comparison of PEDOT:ClO₄ and PEDOT:PSS at 0 V bias highlights distinct differences in their doping states (Supplementary Fig. 8). PEDOT:PSS displays a pronounced absorption band between 600 and 800 nm (black shaded area), while PEDOT:ClO₄ exhibits stronger absorption beyond 900 nm (red shaded area). The latter is indicative of increased bipolaron formation, as previously discussed, and reflects a higher doping level in the electropolymerized PEDOT:ClO₄ film.

Different ionic injection/ejection behaviors for PEDOT:ClO₄ and PEDOT:PSS measured by EQCM-D

To further investigate the doping and dedoping mechanisms in the devices, we employed Electrochemical Quartz Crystal Microbalance with Dissipation (EQCM-D) to monitor in situ mass changes of the channel materials under applied bias in aqueous NaCl. The changes in resonance frequency (Δf/n) and dissipation (ΔD) across multiple harmonics were recorded during cyclic linear potential sweeps, allowing us to gain insight into the dynamic mass uptake and release associated with electrochemical modulation. This approach provides valuable information on how mass transport correlates with the electronic state changes within the channel materials during the doping and dedoping processes, for both rigid and flexible devices.

For quantitative mass analyses, the mass variation (Δm) at the sensor surface or within the deposited film can be determined from the recorded oscillation frequency change (Δf) across different overtones orders (n) using the Sauerbrey’s equation (Eq. 1):

where Δf/n is the change of frequency at nth overtone. C = Z_q/2f₀² is the calculation constant (or sensitivity factor) of the quartz crystal, where f₀ is the fundamental resonance frequency, and Z_q is the acoustic impedance of the quartz crystal (~8.8 × 10⁵ g/cm²s). For a 5 MHz crystal sensor, C equals 17.7 ng/cm²Hz⁵⁷.

According to the Sauerbrey equation, an increase in Δf/n corresponds to a mass loss, and a decrease corresponds to a mass gain. This equation is typically applicable to rigid materials, such as metal coatings, in air or liquid⁵⁸. For polymer coatings in contact with liquid media, the Sauerbrey model remains valid if the dissipation change is minimal (ΔD ≈ 0); otherwise, viscoelastic models are preferable⁵⁸. It has been shown that for polymer films thinner than 100 nm, the Sauerbrey model reliably reflects mass changes (Δm)^40,41. When dissipation changes are small, the frequency shifts (Δf/n) overlap across all measured harmonics, indicating that the film behaves as a rigid layer firmly attached to the sensor surface⁵⁹. In our experiments, all electrodeposited films showed negligible shifts of Δf/n and ΔD over a complete cycle, spanning voltage sweeps from −0.8 to +0.8 V and back (Supplementary Figs. 9, 10). This behavior suggests minimal ionic accumulation or loss, indicating stable mass dynamics during the electrochemical modulation processes.

The EQCM-D responses for PEDOT:ClO₄ and PEDOT:PSS films deposited over a 30-s period showed negligible changes in ΔD, along with overlapping Δf/n values across different overtones (n = 3, 5, and 7) (Supplementary Fig. 9a, b), indicating that these films could be reliably analyzed for mass transport. The cyclic voltammograms (CV) for PEDOT:ClO₄ (Fig. 5c and Supplementary Fig. 11c) exhibited the typical shape of CV for PEDOT⁶⁰, with increasing current at faster scan rates, suggesting a pseudo-capacitive behavior. A similar CV profile was observed for PEDOT:PSS (Fig. 5e). The peak observed at −0.4 V, which diminishes upon degassing (Supplementary Fig. 12), is related to oxygen electrochemistry under normal operating conditions.

Before a full cycle sweep, the PEDOT films were held at −0.8 V to achieve a fully dedoped state, due to the injection of the Na⁺ ions from the electrolyte and, where feasible, the ejection of mobile dopant anions. During the subsequent voltage sweep from −0.8 to 0.8 V (redoping), electrolyte cations migrate toward the counter electrode, while anions migrate to the PEDOT working electrode and enter the film. The ionic movement occurs oppositely during the reverse sweep from 0.8 to −0.8 V (dedoping).

The mass changes for PEDOT:ClO₄ during CV were determined from the frequency responses at the 3rd overtone across different voltage scan rates: 20 mV/s (Supplementary Fig. 11a), 50 mV/s (Fig. 5a), and 100 mV/s (Supplementary Fig. 11b). Illustrations of the ionic distribution at the initial state (without voltage applied), during redoping (from –0.8V to 0.8 V) and during dedoping (from 0.8V to –0.8 V) are provided for PEDOT:ClO₄ (Fig. 5g) and PEDOT:PSS (Fig. 5e). These mass variations correlated closely with the voltage and current profiles at all scan rates, with only minor differences observed (Supplementary Fig. 11d), indicating that ionic movement is minimally impacted by scan rate. For PEDOT:ClO₄, redoping (Fig. 5a, pink shaded area, and Fig. 5d, pink line) begins with a decrease in Δm between −0.8 to −0.3 V due to the ejection of Na⁺ ions (Fig. 5g). Beyond −0.3 V, Δm sharply increases as ClO₄⁻ dopant anions (Fig. 5g), previously ejected during dedoping, re-enter the film possibly accompanied by Cl⁻ anions from the electrolyte. During the reverse sweep (dedoping), Δm continuously decreases (Fig. 5a blue shaded area, and Fig. 5d, blue line), indicating that ClO₄⁻ and Cl⁻ anions leave the film as Na⁺ ions are injected (Fig. 5g). PEDOT:ClO₄ samples deposited for 60 s and 120 s exhibited larger ΔD values and pronounced Δf/n splitting across different overtones n = 3, 5 and 7 (Supplementary Fig. 10), suggesting that the Sauerbrey equation is no longer applicable. It has been reported that higher-order shear waves are more sensitive to the material near the electrodes, while lower-order waves experience less damping in the film. Frequency shifts were smaller at higher-order overtones and larger at lower-order overtones, indicating that mass changes are more pronounced closer to the electrolyte. This observation differs from the case of acid-treated PEDOT:PSS, which has been reported to exhibit greater mass change near the substrate³⁹.

For PEDOT:PSS, during redoping, a substantial mass decrease was observed (Fig. 5b pink shaded area, Fig. 5f pink line), indicating that the process is primarily driven by the ejection of Na⁺ cations from the film, along with the injection of Cl⁻ anions (Fig. 5h). During dedoping, Δm remained stable from 0.8 to 0.4 V but increased between 0.4 and -0.8 V (Fig. 5b blue shaded area, Fig. 5f blue line). The stable Δm between 0.8 and 0.4 V suggests a mass balance between the slight ejection of Cl⁻ ions and the injection of Na⁺ ions. The subsequent mass increase indicates that the process is dominated by Na⁺ ion injection to compensate negatively charged PSS chain (Fig. 5h), consistent with the widely accepted understanding of the doping/dedoping mechanism in PEDOT:PSS. Unlike previous findings regarding spin-coated PEDOT:PSS films containing additives⁴⁰, our results demonstrate the participation of Cl⁻ anions in this process.

Therefore, EQCM-D enabled a direct evaluation of the doping/dedoping mechanisms in electropolymerized PEDOT films, revealing that small anion-doped PEDOT (PEDOT:ClO₄) differs from polyanion-doped PEDOT (PEDOT:PSS) due to the distinct involvement of dopant counterions. While EQCM-D offers valuable insights into mass transport during electrochemical processes, it presents limitations on resolving contributions from various ionic species, such as Na⁺ and ClO₄⁻, and from the solvent. As such, a more detailed understanding of mass transport will require the integration of advanced operando characterization techniques, such as XRF for substituted ions.