Ion–backbone accessibility enables unity doping efficiency in organic electrochemical transistors

Abstract

Organic electrochemical transistors are commonly benchmarked using volumetric capacitance (C*); however, this metric does not distinguish between Faradaic and non-Faradaic contributions and therefore does not directly quantify the density of electronically active charge carriers that governs device-relevant doping capacity. Here we introduce effective volumetric capacitance (C_eff* = C* · η), where η denotes doping efficiency, as a metric for electronically effective volumetric doping. Using this metric, we show ionophilic side chains promote high ionic uptake but also sequester ions away from the conjugated backbone, lowering doping efficiency. By contrast, side-chain removal increases ion access to electrochemically addressable backbone sites, enabling doping efficiencies approaching unity. The resulting materials exhibited both enhanced volumetric charge density and improved charge transport, yielding transistors with transconductance among the highest reported. These results establish a practical design rule for organic mixed conductors by showing that electronically effective doping, rather than ionic uptake alone, governs transistor operation.

Organic electrochemical transistors (OECT) are benchmarked by volumetric capacitance, which is limited for device-relevant doping capacity. Here, the authors present effective volumetric capacitance as a metric, showing that electronically active doping, rather than ion uptake alone, is important for OECTs.

Introduction

Organic electrochemical transistors (OECTs) have emerged as a promising class of electronic devices uniquely suited for bioelectronics, owing to their high biocompatibility^1,2, soft mechanical properties^3,4, and efficient signal transduction^5,6. Their operation relies on volumetric doping mediated by mixed ionic–electronic conduction in polymeric semiconductors⁷, enabling exceptionally high transconductance (g_m) under low-voltage operation—an attribute difficult to achieve with conventional inorganic-based technologies.

To capture the distinctive physics of OECTs, Bernards et al. introduced a theoretical framework modeling transconductance (g_m) as⁸:

$$g_m=\mu C^{*}\left(V_G-V_{th}\right)\left(Wd/L\right)$$ 1

where μ is the charge carrier mobility, C* is the volumetric capacitance, and W, d, and L denote the channel width, thickness, and length, respectively. V_G and V_th denote gate voltage and threshold voltage, respectively. While this model has been instrumental in guiding material and device design, it implicitly assumes that a one-to-one correspondence between accumulated ions and electronic charge carriers⁹—a premise that has remained largely unchallenged. C* is generally extracted from electrochemical impedance spectroscopy (EIS) at low frequencies¹⁰ or from chronoamperometry measurements¹¹; however, these techniques capture not only Faradaic contributions from redox reactions, which directly reflect the doping capacity of the conjugated polymer, but also non-Faradaic processes such as ion diffusion and electric double-layer (EDL) formation^12–14, which do not contribute to polaron generation. As a result, the extracted C* may not fully represent the actual doping capacity of conjugated polymers, especially when doping efficiency (η)—the ratio of charge carriers generated to ions accumulated—is far from unity. For example, polymers such as polyvinyl alcohol (PVA) and polyacrylonitrile (PAN), both being electrochemically inactive (η = 0) and not resulting in any actual charge carrier generation, can exhibit high C* values exceeding 200 F cm⁻³¹⁵. This example illustrates a limitation of using C* as a key figure of merit for accurately representing the device-relevant doping capacity of channel materials.

Importantly, the current use of C* to characterize doping capacity may inadvertently guide materials optimization toward maximizing ionic uptake without fully capturing electronically effective charge generation. For example, a great deal of effort has been devoted to incorporating various oligo ethylene glycol (OEG) side chains to take advantage of their high ionophilicity, enabling C* values exceeding 250 F cm^{–39,16–20}. While OEG side chains enhance ion uptake/release processes, ensure solution processability, and enable the operation of OECT under aqueous conditions, a high C* alone does not necessarily justify the use of OEG side chains if the observed high C* primarily arises from non-Faradaic processes rather than the desirable Faradaic processes. Moreover, OEG side chains inevitably compromise film crystallinity and substantially reduce charge carrier mobility (μ)^9,21,22, thereby posing a significant trade-off that limits further improvements in device performance^20,23,24. More fundamentally, predicting the net efficacy of side-chain engineering remains highly elusive at the molecular level. Specific side-chain modifications critically dictate the solid-state molecular packing and resulting charge-transport properties²⁵, while simultaneously governing the intricate spatial segregation and penetration dynamics of solvent molecules and dopant ions within the polymer matrix²⁶. To address these limitations, a new figure of merit that can accurately capture the true doping capacity of mixed conductors is required.

Here, we identify effective volumetric capacitance (Cₑ_ff* = C* · η) as a device-relevant metric for OECT channel optimization, because it captures electronically effective doping rather than total ionic uptake. Using an operando spectro-electrochemical approach focused on the Faradaic process, C_eff* quantifies the electrochemically generated charge carriers, enabling more meaningful comparison across different materials. This new figure of merit, Cₑ_ff*, enabled us to clarify how different side chains influence the OECT performance in a series of structurally related thiophene-based polymers. Our collected results allow us to identify three critical factors—side-chain presence, side-chain chemical nature (e.g., alkyl vs. OEG), and ion polarity (hydrophilic vs. hydrophobic)—that govern Cₑ_ff* and η by modulating ion–polymer backbone distance, as supported by molecular dynamics (MD) simulations. Interestingly, we discovered that the nature of side chains (alkyl vs. OEG) can, in some cases, exert only a minor impact on C_eff* and η—a finding that challenges the prevailing view. Notably, under the maximized ion–backbone accessibility enabled by a conjugated polymer with cleavable side chains, POET-T2 reaches C_eff* of 260 F cm⁻³ with η approaching unity accompanied by enhanced charge-carrier mobility, yielding normalized transconductance and [μC*]_OECT values among the highest reported to date. These results establish Cₑ_ff* as an intrinsic materials parameter governing OECT performance and provide actionable guidance for optimizing channel materials in OECTs.

Results

Thin-film structural and charge-transport characteristics

To isolate how side chains influence thin-film structure and charge transport prior to electrochemical doping, we constructed a structurally matched polymer platform as shown in Fig. 1a. Alkyl (P3HT), OEG (p(g₃2T-T)), side-chain-cleavable (POET-T2), and co-polymer (POET-T2₅₀-stat-p(g₃2T-T2)₅₀) were prepared, all sharing an identical polythiophene backbone. For the side-chain cleavage process, an acid-triggered cleavage method was employed, consistent with our previous work²⁷.

Fig. 1. Thin-film characteristics of side-chain-cleavable polymer.

a Schematic illustration of the side-chain cleavage process and molecular structures of the polymers investigated in this work. b Fourier-transform infrared (FT-IR) spectra of POET-T2. c π–π stacking and lamellar d-spacing plots of POET-T2 and co-polymer films. d Comparison of GIWAXS-derived g-factors for polymers along the (010) π–π stacking direction. e UV-vis absorption spectra, and f energy band diagrams of POET-T2. g Box plots representing the saturation field-effect mobility (μ_sat) of each polymer. Data are presented as mean values from at least six independent measurements. Error bars represent ±1 s.e.m., and whiskers indicate the minimum and maximum values.

We first studied POET-T2, since its cleavable side chains would allow us to address a crucial yet challenging question due to the inaccessibility of structurally matched polymer systems without side chains: are side chains beneficial for mixed ionic-electronic conduction? POET-T2 offered a unique opportunity to answer this question as it can cleave its side chains at room temperature after film formation to yield POET-T2-COOH, enabling direct before-and-after comparisons. Figure 1b displays Fourier-transform infrared (FT-IR) spectra of POET-T2 before and after side-chain cleavage via acid-catalyzed cleavage (e.g., triflic acid, TfOH). The clear emergence of broad absorption bands associated with O–H and C=O groups of the carboxylic acid, along with the disappearance of characteristic C–H stretching bands, confirms the effective removal of side chains from the conjugated polymer. A primary advantage of the side-chain-cleavable system is the selective elimination of electrically insulating alkyl segments within the film, preserving π–π stacking²⁸. This structural modification was validated by grazing incidence wide-angle X-ray scattering (GIWAXS), presented in Fig. 1c. Analysis of the line-cut data (Supplementary Fig. 8) reveals that the lamellar d-spacing notably decreases from 14.48 to 10.04 Å following side-chain cleavage, whereas the π–π stacking distance remains consistent at 3.62 Å. Furthermore, the partial side-chain removal in the co-polymer (a 50:50 random co-polymer from POET-T2 and p(g₃2T-T2)) resulted in an approximately proportional reduction in the lamellar d-spacing as expected.

The more compact packing resulted in approximately a 10% reduction in film thickness (Supplementary Fig. 9), consistent with our previous work²⁷. To assess the degree of order in π–π stacking within the films, we extracted the g-factor from the GIWAXS line-cuts (Fig. 1d)²⁹. As anticipated, p(g₃2T-T) exhibited a high g-factor of 20.8%, reflecting weak π–π interactions associated with the intrinsically low crystallinity of OEG-based polymers^21–24. By contrast, both POET-T2 and the co-polymer displayed decreased g-factors after side-chain cleavage, with cleaved POET-T2 showing values below 10%, indicating that π–π order is preserved and modestly improved upon side-chain removal. As depicted in Fig. 1e, the absorbance intensity remained mostly unchanged after cleavage, exhibiting only a slight red shift and a minor shoulder peak associated with increased effective conjugation length and local aggregation³⁰. Considering that local aggregation in polythiophene films typically enhances charge transport³¹, we conclude that the side-chain cleavage introduced favorable morphological characteristics without compromising the intermolecular packing. Furthermore, negligible differences in the energetics of POET-T2 after side-chain cleavage were observed, as summarized in Fig. 1f and Supplementary Table 3.

To evaluate the intrinsic charge-carrier mobility decoupled from ion uptake/release processes, field-effect transistors (FETs) were fabricated for all studied polymers (see Methods for details), and their saturation mobilities (μ_sat) are summarized in Fig. 1g. P3HT exhibited a μ_sat of 0.007 ± 0.0005 cm²V^–1 s^–1, whereas the OEG-polymer, p(g₃2T-T), showed approximately 1000 times lower mobility, as expected²². Notably, POET-T2 and the copolymer showed significantly increased μ_sat after side-chain cleavage, reflecting morphology-induced structural advantages, with the mobility of POET-T2 reaching 0.01 ± 0.0011 cm²V^–1s^–1. Cleaved POET-T2 even demonstrated higher mobility than P3HT, a well-studied benchmark polythiophene-based semiconductor³². This trend of OFET mobility aligns with our prior studies on charge transport of side-chain-cleavable polymers^27,28, where the side-chain-cleaved POET-T2 film exhibited a high Hall mobility of up to 1.00 cm²V^–1 s⁻¹²⁷. Although these measurement techniques alone may not fully explain 3-dimensional charge transport characteristics in OECTs, the increased μ following cleavage suggests that employing side-chain-cleavable polymers represents a promising alternative to OEG-based counterparts, particularly for retaining high charge carrier mobility.

Determination of doping capacity (Cₑ_ff*)

Volumetric capacitance (C*) is widely used to estimate the electrochemical doping capacity of polymers^9–11. Based on chronoamperometry measurements (Supplementary Figs. 13, 14), we found that side-chain cleavage in POET-T2 results in an 85% increase in C*, from 134 to 247 F cm⁻³. This enhancement is attributed to the removal of steric hindrance from bulky alkyl chains, as well as the formation of ionophilic COOH groups which facilitate ionic uptake processes. This hypothesis is further supported by EIS measurements of the POET-T2 films (Fig. 2a, b). Upon side-chain cleavage, the charge transfer resistance (R_ct) decreased by approximately 50%, from 1.99 to 1.03 kΩ cm², indicating a lower energetic barrier for ion exchange at the polymer–electrolyte interface^33,34. In addition, ionic conductivity increased across the entire frequency range, and the phase angle magnitude rose in the high-frequency region, suggesting more capacitive behavior and faster ionic dynamics within the device³⁵.

Fig. 2. (Spectro-)electrochemical properties of polymers.

a Nyquist plot and b Bode plot of electrochemical impedance spectroscopy (EIS) for POET-T2, measured before and after side-chain cleavage. A 50 mV AC signal was applied with a DC offset of –1 V. c Schematic illustration highlighting the different current generation mechanisms between Faradaic and non-Faradaic processes. d Voltage-dependent absorption spectra of POET-T2 films before and after side-chain cleavage. N, P, and BP denote neutral, polaronic, and bipolaronic absorption, respectively. e Degree of doping plotted as a function of applied voltage, extracted from the absorption spectra. V₀ represents the baseline potential corresponding to the fully de-doped state of the polymer film. f Comparison of C*, effective volumetric capacitance (Cₑ_ff*), and doping efficiency (η) across various polymers. Measurements were conducted using [EMIM][TFSI]-based gel electrolytes to focus on ion-backbone interactions by excluding the contribution of water uptake in aqueous electrolytes. All measured in a vertical device configuration (PEDOT electrode (WE)/gel electrolyte/POET-T2/ITO electrode (CE)).

As illustrated in Fig. 2c, C* contains not only Faradaic redox reactions (electrochemical doping) but also non-Faradaic processes—referred to as C_dl in Fig. 2c—such as EDL formation and ion diffusion^12–14. These non-doping processes still generate capacitive charging current during measurement in C* (e.g., via EIS or chronoamperometry), leading to an overestimation of the true doping capacity. Therefore, an increased C* does not necessarily translate to higher doping capacity and thus higher performance in OECTs. This distinction becomes clearer when comparing the channel conductivity equation and the geometry-normalized transconductance equation:

$$\sigma \left(S{cm}^{-1}\right)=\mu ne$$ 2

$$g_{m,norm}\left(S{cm}^{-1}\right)=\mu C^{*}\left(V_G-V_{th}\right)=\mu Q^{*}$$ 3

Here, σ and e denote the channel conductivity and elementary charge, respectively. When adopting the Bernards model in OECTs, the total number of ions accumulated within polymer film (Q* = C*V) is assumed to be equal to the number of generated free charge carriers (Q* = ne). However, an increase in C* does not correspond to a proportional increase in free charge density, as the non-Faradaic process could contribute to C* as well. To explicitly evaluate the doping capacity of materials, we propose the use of effective volumetric capacitance, C_eff*, to represent the amount of charge carriers generated per unit voltage.

During electrochemical doping, ion injections reduce the number of neutral units and increase the number of doped units of a polymer, a transformation that can be directly observed through changes in the absorption spectra (Fig. 2d). Notably, side-chain-cleaved POET-T2 exhibits complete quenching of the neutral absorption band (400–700 nm) at potentials less than –0.8 V, whereas pristine POET-T2 retains approximately 70% of its neutral absorbance at –0.8 V. According to the Beer–Lambert law, absorbance in the neutral region is linearly proportional to the number of neutral units. Thus, by performing multi-peak fitting to separate polaronic absorption, we were able to quantify the degree of doping as a function of applied voltage (Fig. 2e; see Supplementary Figs. 15, 16 and Supplementary Note 1 for detailed procedures)³⁶. Side-chain-cleaved polymers exhibited a higher doping rate than their uncleaved counterparts, as evidenced by the steeper slope of the curve. This slope can be employed to calculate the C_eff* via the following equation:

$$degreeofdoping=\frac{N_d}{N_i}=\frac{Q^{*}\eta }{N_{i}e}=\frac{C^{*}V\eta }{N_{i}e}=\frac{{C_{eff}}^{*}}{N_{i}e}V$$ 4

where N_i and N_d denote the number of initial neutral units, and doped units, respectively. η is defined as the ratio η = C_eff*/C*, representing the fraction of generated charges relative to the total accumulated charges. The slope of curves in Fig. 2e directly represents the doping capacity, which is identical to the C_eff*/(N_ie). Since N_i is constant, we can then quantify not only the C_eff*, but also the efficiency factor η by combining it with C* obtained from chronoamperometry.

The extracted values of C*, C_eff*, and η for various polymers are summarized in Fig. 2f and Supplementary Table 4. As anticipated, OEG-based polymer p(g₃2T-T) exhibited the highest C* of 332 F cm^–3 due to its high ionic affinity; however, its C_eff* reached only 177 F cm^–3 accompanied by a low doping efficiency of 0.534, showing a marginal improvement in doping capacity compared to alkyl-based polymer P3HT (C_eff* = 150 F cm^–3, η = 0.664). This implies that nearly half of the injected ions remain electrochemically inactive, likely confined near the OEG side chains without contributing to doping. In contrast, side-chain-cleavable polymers—POET-T2 and its co-polymer—showed a marked increase in both C* and C_eff* with side-chain cleavage. Notably, cleaved POET-T2 exhibited an η of unity, indicating that all injected ions effectively doped the polymer without ion loss to non-Faradaic pathways. As a result, cleaved POET-T2 exhibited the highest C_eff* among the polymers in this series, reaching 260 F cm^–3, representing a 600% increase over the uncleaved counterpart (37 F cm^–3). Comparing p(g₃2T-T) and cleaved POET-T2 suggests that maximizing the doping efficiency η, rather than increasing C* alone, is a more effective strategy for achieving a high doping capacity in conjugated polymers. Partial side-chain cleavage in the case of the co-polymer also led to a noticeable increase in C_eff* and η from before cleavage. These values (167.7 F cm^–3, 0.589) were similar to those obtained from the p(g₃2T-T), suggesting a non-linear correlation between composition and doping behavior. These results indicate that the presence or absence of side chains critically governs the doping characteristics of polythiophenes, rather than the polarity of the side-chain (e.g., alkyl vs. OEG) as widely assumed in the literature⁹.

The exceptional doping efficiency of POET-T2 can be largely attributed to the elimination of sterically hindering side chains, and the origin of this behavior is analyzed in the following section. It is also worth noting that C_eff* and η are sensitive to the polarity of electrolytes. Compared with the case in which [EMIM][TFSI] is used as an electrolyte, p(g₃2T-T) exhibited higher C_eff* and η values when paired with a hydrophilic ionic liquid ([EMIM][OTf]) or an aqueous NaCl electrolyte, as summarized in Supplementary Tables 5 and 6. These findings indicate electrolytes strongly dominate volumetric doping and should be carefully selected depending on the nature of side chains to maximize C_eff* and η.

Computational simulations of ion injections and interactions

To rationalize the experimentally observed differences in C_eff* and η, we performed density functional theory (DFT) calculations and molecular dynamics (MD) simulations to elucidate how side chains modulate ion accessibility to the conjugated backbone at the molecular level. DFT revealed that the side-chain-cleaved POET-T2 and its co-polymer exhibit greater stabilization in binding energy upon oxidation (ΔΔE_bind) compared to other polymers, reaching –53.2 kcal mol^–1, indicating a stronger polaron-anion interaction (Fig. 3a). This stabilization arises from the removal of bulky side chains, which enables closer electrostatic interaction between the counter-ion (e.g., TFSI anion) and the oxidized backbone³⁷. By contrast, p(g₃2T-T) displayed the least negative ΔΔE_bind, indicating less effective stabilization of the polaron by the TFSI anion. We attribute this to the interference of its ionophilic OEG side chains, which compete with direct anion–backbone interactions and thereby reduce the polaron stabilization. In addition, an alkyl-chain-based polymer, P3HT, showed negative ΔΔE_bind comparable to cleaved POET-T2; this result agrees with previous reports indicating that alkyl side chains are highly compatible with TFSI anions³⁸.

Fig. 3. Predictions of dopant ion interaction and coordination with polymers by computational methods.

a DFT-predicted ensemble-averaged binding energy difference (ΔΔE_bind) between polaron backbone-anion complex and neutral backbone-anion complex. b Schematic illustration showing decreased backbone-ion distance with side-chain cleavage. c Snapshots of MD simulations consisting of a polymer film with [EMIM][TFSI] electrolyte reservoir, invoking electrolyte permeation by applied electric field. The polymer backbones are shown in purple, and [TFSI] anions are drawn as green spheres. The thiophene monomers with assigned polaron are highlighted in yellow. [EMIM] cations are omitted. The crystalline system with cleaved POET-T2 is presented as an example. d Histograms of distances between the dopant anions and backbones in the amorphous and crystalline regions of the thin films. e Fraction of dopant anions effectively doped on polymer backbones (η_cal), and charge carrier density derived from MD simulations using the iterative doping protocol under Boltzmann-weighted probability model. The reported values are calculated as a weighted average of the results from amorphous and crystalline domains, assuming equal volume fractions for both phases to represent the total morphology of the polymer films.

While ΔΔE_bind explains the effect of side chains on the energetic cost of polymer backbone oxidation per individual doped molecular complex, the polarity and presence of side chains also dictate the number of dopant ions that can physically approach the conjugated backbone (Fig. 3b). Crucially, the degree to which side chains govern this accessibility is intrinsically linked to the structural order of the polymer packing, ranging from amorphous to crystalline domains. To systematically assess the impact of such morphological variations, we performed extensive MD simulations for both amorphous and crystalline systems. We modeled each polymer film immersed in [EMIM][TFSI] using an iterative doping and relaxation protocol that explicitly captures the concerted nature of ion injection, polaron formation, and structural relaxation (Fig. 3c, Supplementary Fig. 24; see Methods and Supplementary Note 2). This dual-domain approach allowed us to verify whether the side-chain-induced ion-sequestration effect is a universal phenomenon regardless of the structural order. The resulting ion distributions reveal distinctive differences in both crystalline and amorphous domains: for side-chain-cleaved POET-T2, the backbone–anion distance histogram shows a sharp peak at shorter distances (~4.6 Å) than in other systems, confirming more facile ion access to the backbone of cleaved POET-T2 (Fig. 3d). In contrast, in p(g₃2T-T), the backbone–ion distance distribution is broad and extends beyond 6.0 Å, whereas anions preferentially coordinate to the ionophilic OEG side chains (side chain–ion peak distance ~4.6 Å; Supplementary Fig. 25), effectively sequestering the ions away from doping sites. This result explains why p(g₃2T-T), even with the highest C*, exhibits a relatively low η of 0.534 (Fig. 2f). Thus, side-chain cleavage removes steric barriers and undesirable electrostatic competition, ensuring intimate ion–backbone contact.

The decisive impact of this enhanced accessibility of counter-ions to the conjugated backbone is directly reflected in the predicted doping metrics in the MD simulations. We applied a physically guided thermodynamic probability model in the iterative doping protocol to estimate the doping efficiency (η_cal) and charge carrier density, by considering the energetic cost of backbone ionization, explicit atomistic reorganization, and the electrostatic stabilization of the generated polaron from the MD-derived ion distributions (Fig. 3e; see Methods for details). This analysis revealed a notable increase in both charge carrier density and η_cal in side-chain-cleaved polymers. Specifically, in side-chain-cleaved POET-T2, the charge-carrier density increased from 0.54 to 1.19 e nm^–3, yielding the highest η_cal value among the samples and reaching 0.74, whereas p(g₃2T-T) exhibited a lower η_cal of 0.56. Cleaved POET-T2 showed a greater enhancement in charge-carrier density and η_cal than the co-polymer, which was designed to undergo partial side-chain cleavage.

In summary, our computational results reveal a synergistic mechanism: the remarkable performance of the cleaved polymer arises from a combination of a favorable energetic landscape for electrochemical doping and, more critically, a dramatic enhancement in ion accessibility that ensures nearly every available ion can participate in the doping process.

Translation of doping efficiency into OECT performance

We next examined whether the distinctions revealed by C_eff* and η translate directly into improved device-level performance. For this purpose, we fabricated conventional lateral-transport OECTs using the side-chain-cleavable polymer POET-T2 and gel electrolytes, as described in Methods. The device architecture and an optical micrograph are shown in Fig. 4a. The transfer characteristics of POET-T2-based devices before and after side-chain cleavage are shown in Fig. 4b, c, respectively. Upon side-chain cleavage, a notable enhancement in the on-current (I_on) was observed—from 2.22 to 8.32 mA at V_D = − 0.5 V and V_G = − 0.8 V—along with a discernable suppression of hysteresis which typically originates from constricted, slow, and unpredictable ion diffusion⁵. Notably, the I_on of the cleaved POET-T2 is comparable to that of vertical-transport-type OECTs^5,6,39,40, which rely on ultra-short channel lengths (under 100 nm) to boost I_on and g_m. Because the off current in OECTs is largely governed by ionic drift currents⁴¹, the increase in C* after side-chain cleavage led to an approximately threefold rise in the off current (summarized in Supplementary Table 8); yet the off current remained substantially lower than that of p(g₃2T-T), which displayed the largest C*. This highlights that suppressing off current requires minimizing excessive increases in C*, which can be achieved through increasing doping efficiency.

Fig. 4. Mixed ionic–electronic conduction characteristics of all-solid-state OECTs.

a Schematic illustration and optical microscope image of the device architecture. Transfer characteristics of OECTs with b pristine and c side-chain-cleaved POET-T2 films, using [EMIM][TFSI]-based gel electrolyte. d Geometry-normalized transconductance (g_m,norm) and subthreshold swing (SS) of POET-T2 OECTs. Output characteristics of e pristine and f side-chain-cleaved POET-T2 OECTs. g Transconductance (g_m) and h rise time (τ_r) plots as a function of device geometry, controlled via film thickness and channel length. i Long-term on/off stability of POET-T2 OECTs under repeated gate pulses (0/–0.6 V, 0.125 Hz) with a constant drain voltage of –0.5 V. j Box plots of g_{m, norm} for various polymers (dashed lines are guides to the eye). Data are presented as mean values from at least six independent measurements. Error bars represent ±1 s.e.m., and whiskers indicate the minimum and maximum values. k Comparison of on/off current ratio and [μC*]_OECT value of POET-T2 with previously reported OECTs operated at |V_D | ≤ 0.5 V^{3,11,45,47–61}. Intrinsic material properties were considered; processing-based studies (e.g., mechanical stretching) were excluded.

The geometry-normalized transconductance (g_{m, norm}) provides a more direct figure of merit for evaluating device performance⁴². As shown in Fig. 4d, the g_{m, norm} of cleaved POET-T2 reached 276.4 S cm⁻¹ at V_D = − 0.5 V and V_G = − 0.6 V, underscoring its excellent signal transduction capability. Compared to the uncleaved film, which exhibited g_{m, norm} of approximately 98.5 S cm⁻¹, this corresponds to an increase of over 150%. The increase is ascribed to improved charge mobility and doping efficiency enabled by the removal of non-conductive, bulky side chains.

The output characteristics exhibited a more pronounced enhancement upon side-chain cleavage, with the saturation current increasing nearly sixfold from 1.15 to 6.5 mA at V_G = − 0.8 V (Fig. 4e, f), consistent with the trend in C_eff* (Fig. 2f). To ensure that this improvement truly originates from side-chain cleavage rather than the acidic treatment itself, we conducted control experiments on P3HT (see Supplementary Fig. 28). As a result, acid-treated P3HT displayed 60% decrease in both on-current and g_{m, norm}, which confirmed that the performance enhancement can indeed be attributable to side-chain removal. Furthermore, our DFT calculations and in situ Raman spectroscopy (Supplementary Fig. 29) show that deprotonation of the COOH group in cleaved POET-T2 is negligible and does not affect the electronic properties of the backbone.

Volumetric doping in OECTs allows geometry-dependent modulation of g_m and response time (Fig. 4g, h). Both pristine and cleaved POET-T2 exhibited linear g_m scaling with Wd/L, with the cleaved polymer showing a steeper slope corresponding to higher g_{m, norm}. At higher Wd/L over 2 μm, both cleaved POET-T2 and pristine POET-T2-based devices reached saturation, presumably due to the contact resistance effects^40,42.

As the rise time (τ_r) of the OECT is generally considered a limiting factor for operational speed⁴³, we plotted the τ_r values across various device geometries (Fig. 4h). τ_r scaled nearly linearly with channel volume⁴⁴, and side-chain-cleaved POET-T2-based devices demonstrated more than a twofold improvement in operational speed, in agreement with their higher double-layer charging frequency (f_dl) and higher phase angle observed in EIS (see Fig. 2a, b). Although the absolute operational speed (~1 Hz) is still lower than that of previously reported OECTs (>100 Hz)⁴³, it is reasonable to assume that further optimization of the device scaling and electrolyte could substantially enhance the transient characteristics.

Furthermore, side-chain removal markedly improved device stability, as shown in Fig. 4i. Cleaved POET-T2 exhibited nearly constant on–off characteristics over extended operation, with an on current coefficient of variation (CoV) of 0.0894, whereas the pristine counterpart displayed unstable behavior and occasionally failed to turn on properly, resulting in a much higher CoV of 0.496. Given that OECT degradation is closely linked to structural strain and morphological fatigue arising from repeated ion injection and extraction⁴⁵, the enhanced ionic conductivity and reduced R_ct following side-chain cleavage mitigate these stresses by facilitating more efficient and reversible ion transport, thereby improving long-term operational stability.

Figure 4j presents the g_{m, norm} of the polymers investigated in this study (details in Supplementary Fig. 32; summarized in Supplementary Table 8). Overall, the co-polymer also exhibited an increase in g_{m, norm} reaching 140 S cm⁻¹, but cleaved POET-T2 demonstrated the highest value reaching 270 S cm⁻¹, while conventional polymers appeared limited to approximately 150 S cm⁻¹. In addition, the trend in g_{m, norm} closely mirrors that of doping efficiency, η (see Fig. 2f). Because [μC*]_OECT remains the standard OECT benchmarking metric, we calculated [μC*]_OECT from the maximum transconductance, and cleaved POET-T2 exhibited a [μC*]_OECT value over 870 F cm⁻¹V⁻¹s⁻¹, underscoring its exceptional performance.

As summarized in Supplementary Tables 9 and 10, when operated in either a polar ionic liquid ([EMIM][OTf]) or an aqueous NaCl solution, the device metrics varied with the electrolyte. This anion-dependent performance variation is attributed to the intrinsic chemical properties of ions (e.g., steric compactness, dipole moment, and energetics) on doping characteristics such as polaron stabilization⁴⁶. Nevertheless, side-chain-cleaved POET-T2 still showed more than a twofold enhancement in both g_{m, norm} and [μC*]_OECT compared to the uncleaved counterpart, with g_{m, norm} reaching values above 330 S cm⁻¹ in the [EMIM][OTf] electrolyte. Consequently, cleaved POET-T2 represents among the highest [μC*]_OECT values reported to date for OECTs^{3,11,45,47–61} (Fig. 4k).

To further evaluate the generality of the C_eff* and η framework, we extended our analysis to conjugated polymers with different backbone chemistries, as summarized in Supplementary Tables 6 and 10. In a thienothiophene (TT)-based side-chain-cleavable polymer (POET-TT), we again observed a marked increase in η after side-chain cleavage, approaching unity. However, the device-level performance of POET-TT remained lower than that of the polythiophene-based POET-T2 system. This distinction highlights that maximizing C_eff* and η alone does not automatically translate to higher OECT performances. Since g_{m, norm} is also critically determined by µ—backbone engineering must be employed as a distinct design lever to optimize µ alongside C_eff* and η for superior device performance.

Discussion

In summary, we establish C_eff* as a practical design metric for evaluating electronically effective doping and guiding material optimization in OECTs. Our investigation of a series of structurally identical conjugated polymers—yet with different side chains—revealed that the polarity of the side chains (e.g., alkyl vs OEG) exerts a smaller influence on C_eff* than the physical presence of side chains. By eliminating non-conductive and sterically hindering side chains, cleaved POET-T2 approaches unity doping efficiency while simultaneously improving charge transport, thereby addressing the trade-off between charge transport and doping capacity. Our computational modeling provided a clear atomistic rationale for enhancing C_eff* and η, demonstrating that side-chain cleavage ensures intimate and unobstructed access for dopant ions to the polymer backbone, while conventional designs with long ionophilic side chains intercept dopant ions, diverting them away from the backbone. As a result, we demonstrated OECTs with ultra-high g_{m, norm} and [μC*]_OECT enabled by side-chain-cleaved POET-T2. Although maximizing doping efficiency is not universally required for all OECT application spaces (e.g., synaptic devices), the C_eff* framework provides a practical metric demonstrating that high-performance OECT operation is governed by electronically effective doping rather than ionic uptake alone. This enables more meaningful comparisons between materials and more reliable benchmarking of transistor performance, a distinction that cannot be captured by a C*-only perspective.

Methods

Materials

Poly(3-hexylthiophene) (P3HT), poly[2,5-bis(3-alkylthiophen-2-yl)thieno[2-b]thiophene] (PBTTT), poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP), chloroform, acetone, and triflic acid (TfOH) are purchased from Sigma-Aldrich. 1-Ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([EMIM][TFSI]) and 1-ethyl-3-methylimidazolium triflate ([EMIM][OTf]) are purchased from Tokyo Chemical Industry (TCI). Poly[2-(3,3′-bis(2-(2-(2-methoxyethoxy)ethoxy)ethoxy)-[2,2’-bithiophen]-5-yl)thieno[3,2-b]thiophene] (p(g₃2T-TT)) is purchased from Ossila. Poly(3,4-ethylenedioxythiophene): poly(styrenesulfonate) (PEDOT:PSS) solution (PH1000) was purchased from Heraeus. All chemicals were used without further purification. For synthesis, 5,5’-bis(trimethylstannyl)-2,2’-bithiophene (T2 monomer) was purchased from Sigma-Aldrich and further purified through recrystallization in methanol before polymer synthesis. 2-bromothiophene-5-carboxylic acid was purchased from Ambeed Chemical and used without further purification.

Synthesis of p(g₃2T-T)

5,5’-dibromo-3,3’-bis(2-(2-(2-methoxyethoxy)ethoxy)ethoxy)-2,2’-bithiophene (195.91 mg, 1.00 Eq, 302.13 μmol), 2,5-bis(trimethylstannyl)thiophene (123.80 mg, 1.00 Eq, 302.13 μmol) and tetrakis(triphenylphosphine)palladium(0) (13.966 mg, 0.04 Eq, 12.085 μmol) were charged into a 10 mL microwave reactor vial. The mixture was evacuated and refilled with argon for three cycles before the addition of anhydrous toluene (2.2 mL) and DMF (0.22 mL) under an argon stream. The reaction was heated to 200 °C and held in a CEM microwave reactor (300 W) for 30 min. After the polymerization, the crude polymer was dissolved in hot tetrachloroethane and precipitated into hexanes. The collected polymer was extracted via a Soxhlet extractor with ethyl acetate and collected in chloroform. The polymer solution was concentrated, reprecipitated, and collected via filtration and dried under vacuum as a dark solid (123 mg, 71.3%). M_n = 33k, M_w = 49.5k, Đ = 1.5 determined by GPC in DMF at 55 °C.

Synthesis of POET-T2

Detailed information on the synthesis and purification of bis(2-methylhexan-2-yl) 5,5’-dibromo-[2,2’-bithiophene]-4,4’-dicarboxylate is described in our previous paper²⁷. Stoichiometric imbalance was utilized to prepare a low-molecular-weight chloroform-soluble fraction of POET-T2. bis(2-methylhexan-2-yl) 5,5’-dibromo-[2,2’-bithiophene]-4,4’-dicarboxylate (128.1 mg, 1.00 Eq, 210.5 μmol), 5,5’-bis(trimethylstannyl)-2,2’-bithiophene (98.37 mg, 0.95 Eq, 0.2000 mmol) were added to a 5 mL microwave vial. Vacuum and argon were cycled 3 times, and 1 mL of o-xylene was added. The reaction was heated to 100 °C and kept stirring for 48 hours. The polymer was precipitated in methanol and purified by Soxhlet extraction with methanol, dichloromethane, and chloroform. The chloroform-soluble fraction was concentrated and precipitated in methanol and collected via filtration to give POET-T2 (106 mg, 172 μmol, 81.9%). M_n = 19.6k, M_w = 57.5k, Đ = 2.94 determined by GPC in DMF at 55 °C.

Synthesis of co-polymer (POET-T2₅₀-stat-p(g₃2T-T2)₅₀)

bis(2-methylhexan-2-yl) 5,5’-dibromo-[2,2’-bithiophene]-4,4’-dicarboxylate (60.84 mg, 1 Eq, 0.1000 mmol), 5,5’-dibromo-3,3’-bis(2-(2-(2-methoxyethoxy)ethoxy)ethoxy)-2,2’-bithiophene (64.84 mg, 1 Eq, 100.0 μmol), 5,5’-bis(trimethylstannyl)-2,2’-bithiophene (98.37 mg, 2 Eq, 200.0 μmol) and Pd G3 PtBu₃ (9.15 mg, 8 mol%) were added to a 5 mL microwave vial. Vacuum and argon were cycled 3 times, and 1 mL of o-xylene was added, and the reaction was heated to 100 °C and kept stirring for 24 hours. The polymer was precipitated in methanol and purified by Soxhlet extraction with acetone, ethyl acetate, and chloroform. The chloroform-soluble fraction was concentrated and precipitated in methanol and collected via filtration to give POET-T2₅₀-stat-p(g₃2T-T2)₅₀ (65 mg, 51 μmol, 51%). M_n = 25.0k, M_w = 40.5k, Đ = 1.62 determined by GPC in DMF at 55 °C.

Synthesis of POET-TT

To a 10 mL flask, Bis(2-methylhexan-2-yl) 5,5’-dibromo-[2,2’-bithiophene]-4,4’-dicarboxylate (60.84 mg, 1 Eq) and 2,5-bis(trimethylstannyl)thieno[3,2-b]thiophene (46.58 mg, 1 Eq) and Pd G3 PtBu₃ (4.58 mg, 8 mol%) were dissolved in 1 mL o-xylene under inert atmosphere. The reaction was heated to 90 °C and kept stirring for 24 hours. The polymer was then precipitated in methanol and purified by Soxhlet extraction with methanol, dichloromethane, and chloroform. The thimble was then removed from the extractor and placed in a flask of chlorobenzene at 90 °C to extract the polymer from the thimble within 30 min. This fraction was then concentrated, precipitated in methanol, and collected via filtration to give POET-TT (41.1 mg, 70 μmol, 70%). M_n = 11.0k, M_w = 19.25k, Đ = 1.75 determined by GPC in DMF at 55 °C.

Side-chain cleavage

Acid-triggered side-chain cleavage is done by soaking the polymer-coated substrate for 5−10 min in 22.6 mM TfOH in acetonitrile solution at room temperature.

Film analysis

Glass substrates, ITO-coated glasses, and silicon wafers are sonicated with D.I. water, acetone, and isopropyl alcohol for 15 min each. All polymers were dissolved in chloroform at a concentration of 10 mg mL⁻¹ and stirred at 60 °C for at least 12 hours. The polymer films were prepared by spin-coating the solutions onto pre-cleaned substrates at 1500 rpm, 60 seconds. FT-IR spectra were taken on a Bruker Optics Hyperion 1000 with Tensor 27. Absorption spectra were obtained with a Jasco V-770 spectrophotometer, with voltage applied to the sample with a Keithley 2636B source-meter unit. Absorbance was calculated using the equation A = log(I₀/I), where I₀ and I represent the incident and transmitted light intensities, respectively. Detailed setups are depicted in Supplementary Fig. 12. The film thicknesses were recorded by a profilometer (KLA-Tencor D−100). Cyclic voltammetry (CV) and EIS measurements were done with a CHI660D Electrochemical Workstation. CV measurements were done in a 0.1 M tetrabutylammonium hexafluorophosphate in acetonitrile solution. GIWAXS measurements were performed with a high-resolution synchrotron X-ray beam source, located at the 3 C beamline of the Pohang Accelerator Laboratory (Republic of Korea), with a grazing incidence angle of approximately 0.1°. Intermolecular d-spacing was calculated using Bragg’s Law. Raman spectra were obtained with a Bruker SENTERRA dispersive Raman microscope.

DFT calculations

Density functional theory (DFT) calculations were performed to evaluate how the polymer structure affects the energetic stabilization of a polaron by a dopant anion. All DFT calculations were conducted using Gaussian 16 software⁶² with the ωB97XD functional⁶³ and the 6-31 + G(d) basis set. We calculated the binding energies for the molecular complexes of polymer backbone and [TFSI] dopant anions extracted from our molecular dynamics (MD) simulations (see the next section, MD simulations, for details). These complexes consisted of a [TFSI] anion and a truncated polymer backbone motif, which is a trimer of thiophene rings with corresponding truncated side chains. For each polymer system, 100 complex geometries were randomly sampled from the MD trajectories when polymers were uncharged and when positively charged polarons were assigned to the central thiophene ring. The selections were done for the molecular complexes where the center-of-mass (COM) distance between the central thiophene ring and the [TFSI] anion was less than 8.5 Å. The terminal ends of the backbone trimers were capped with hydrogen atoms. The side chains on the central monomer were truncated and capped as follows: methyl (-CH₃) for alkyl chains, methoxy (-OCH₃) for OEG chains, and methyl ester (-COOCH₃) for the POET-T2 before cleavage. The full carboxylic acid (-COOH) group was retained for the cleaved POET-T2. Single-point energy calculations were then performed on each of the 100 sampled geometries for both the charge-neutral and the positively charged (polaron) states of the complex. The final reported stabilization energy (ΔΔE_bind) represents the difference between the ensemble-averaged binding energies of the polaron-anion complexes and the neutral backbone-anion complexes.

MD simulations

Molecular dynamics (MD) simulations were conducted using OpenMM 8.0 software⁶⁴ on NVIDIA GeForce RTX 4090 GPUs. The polymers and [EMIM][TFSI] electrolytes were modeled with the OPLS-AA force field⁶⁵ in conjunction with compatible parameters from Do et al.⁶⁶ and Marcon and Raos⁶⁷. For statistical validity, three independent simulations were run for each polymer system.

Pristine polymer films were prepared following two distinct protocols to represent the amorphous and crystalline domains of the materials. For the amorphous phases of the polythiophene-based polymers, fully stretched polymer chains (36 monomers each) were randomly placed in a large simulation box. The system was then compressed into a cubic box of approximately 9 x 9 x 9 nm³ in the isothermal-isobaric (NPT) ensemble at 300 K and 1 atm. To ensure thorough structural relaxation, the resulting amorphous films underwent simulated annealing at 800 K for 100 ns, followed by a gradual quenching back to 300 K (cooling rate of 1 K/10 ps). A final 40 ns equilibration was performed at 300 K and 1 atm in the NPT ensemble.

For the polythiophene-based polymers, we also constructed crystalline systems for additional examinations distinct from amorphous systems. To produce ordered polymer film systems, the initial configurations utilized fully stretched polymer chains aligned along the x-axis, where covalent bonds were assigned across periodic boundaries to ensure long-range continuity. For P3HT, p(g₃2T-T), and POET-T2, each chain consisted of 36 monomers, while the co-polymers were modeled with 32 monomers per chain. To establish the three-dimensional architecture, 20 parallel chains were stacked along the y-axis to represent the π-π stacking direction, and the z-axis was designated for the lamellar orientation. During this initial setup, side chains were configured in a fully stretched, all-trans conformation. The resulting structures underwent density equilibration for 20 ns at 300 K and 1 atm in the NPT ensemble using an anisotropic barostat, which allowed each lattice dimension to relax independently to preserve crystalline symmetry. Finally, the number of lamellar layers was calibrated to achieve an equilibrated film thickness of approximately 9 nm along the z-axis.

To simulate the OECT environment, the equilibrated pristine polymer films were placed in contact with an electrolyte reservoir along the z-axis at 300 K. The reservoir contained 3000 and 4000 [EMIM][TFSI] ion pairs for amorphous and crystalline systems, respectively. The combined system was first equilibrated for 50 ns at 300 K in the NP_zT ensemble, allowing the box dimension to fluctuate only along the z-axis. Subsequently, we implemented an iterative protocol of doping and relaxation to simulate the doping process invoked by an external electric field. This approach was necessitated by two primary physical considerations. First, electrochemical doping and ion injections are inherently concerted processes triggered by the external electric field. As ions permeate the polymer film, the local electronic state and subsequent electrostatic interactions evolve dynamically. Our iterative method aimed to capture these changes on-the-fly by updating the charge state in response to the instantaneous ion distribution. Second, this protocol explicitly accounts for the atomistic reorganization associated with polaron formation. By incorporating dedicated relaxation periods between doping steps, the polymer matrix and dopant ions are allowed to structurally equilibrate to the newly assigned charge states.

The iterative doping protocol consists of alternating cycles: an electric-field-driven infiltration step to facilitate ion ingress, followed by the assignment of polarons onto monomers situated near dopant ions according to a thermodynamic probability model, and a zero-field equilibration step to ensure local thermodynamic relaxation (see Supplementary Fig. 24). This iterative sequence continues until the internal doping density reaches a quasi-steady state. During the polaron assignment step, system charge neutrality was maintained by removing an equivalent number of electrolyte counter-ions from the simulation box. Details of the thermodynamic probability model and system-specific iterative protocol parameters are provided in Supplementary Information, Supplementary Note 2.

The calculated doping efficiency (η_cal) was evaluated as the ratio of the converged charge carrier density to the total injected ion density in the polymer film region, determined by statistical counting in the MD simulation trajectory once the iterative protocol reached a steady state. The presented values represent a weighted average of the results from amorphous and crystalline systems, assuming equal volume fractions for both phases.

To generate representative geometries of oxidized polymers for DFT calculations, we utilized an independent set of MD simulation trajectories produced by applying an electric field of 1.0 V/nm for 30 ns, followed by a quiescent equilibration run for 50 ns, and a 100 ns production run on the initially equilibrated amorphous polymer film/electrolyte layered systems. We explicitly modeled polarons as localized +1e charges on thiophene rings that were in proximity to a [TFSI] anion (cutoff distance <7.0 Å). System charge neutrality was maintained by removing an equivalent number of [EMIM] cations from the simulation box. These charged systems were equilibrated for 5 ns in the NP_zT ensemble, after which a 10 ns production run in the NVT ensemble was performed to sample the complex geometries.

We employed a Langevin thermostat with a friction coefficient of 1 ps^-1 and a Monte Carlo barostat to maintain the simulation temperature and pressure. The distance cutoff of 1.2 nm was applied for nonbonded Lennard-Jones and short-range electrostatic interactions. The long-range electrostatic interactions were computed by the particle mesh Ewald (PME) method. We used a timestep of 2 fs for the simulations at 300 K, and 1 fs for the simulated annealing at higher temperatures, while the covalent bonds with hydrogen atoms were kept rigid.

Device fabrication

For OFET fabrication, silicon wafers with a 100 nm SiO₂ layer were first treated with piranha solution. The substrates were then immersed in a mild octadecyltrichlorosilane (OTS) solution for 1 hour, followed by thermal annealing at 150 °C for 30 min. Polymer films were deposited by spin-coating the solutions onto the wafer at 1500 rpm for 60 s. Au electrodes (40 nm thick) were deposited by thermal evaporation using a shadow mask. Channel width-to-length ratio was 10.

For OECT fabrication, Au/Ti electrodes (36/4 nm) were deposited by thermal evaporation through a shadow mask onto pre-cleaned glass substrates. Polymer films were subsequently deposited by spin-coating (1500 rpm, 60 s) the prepared solutions (10 mg ml^‒1 in chloroform) onto the substrates, as described in the Film analysis section. No post-treatments, such as thermal annealing and patterning, were applied. Electrolyte solutions were prepared by dissolving PVDF-HFP in acetone at a weight ratio of 1:7 and stirring at 60 °C for at least 2 hours. Ionic liquids were then added to form PVDF-HFP/ionic liquid/acetone solutions with a weight ratio of 1:4:7. These solutions were then spin-coated (1500 rpm, 60 seconds) onto pre-cleaned glass substrates to form gel electrolytes approximately 10 μm thick. The dried electrolyte films were cut with a razor blade and transferred to the desired positions. For the top electrode, PEDOT: PSS (PH1000) was drop-cast onto a hydrophobic substrate and annealed at 80 °C for 12 hours. To enhance conductivity, 10 vol% of ethylene glycol was added to the PEDOT: PSS solution and stirred for 20 minutes prior to casting. The resulting solid-state electrodes were cut with a razor blade and pasted onto the device at the desired positions.

Device characterization

I–V characterizations including chronoamperometry were performed using a Keithley 2636B source-meter unit, with a scan rate of 50 mV s^‒1. Home-made software was utilized to control the instrument for measuring transfer and output characteristics, as well as response time and device stability. All measurements were performed under ambient conditions.

Author contributions

W.J.P. designed and performed most of the experiments. K.-J.J. designed and performed most of the computational simulations. J.S., J.N., and X.Z. synthesized and provided polymers. J.K. assisted in data curation and analysis. C.Y.S., W.Y., and D.S.C. supervised this work. All authors contributed to the discussion and manuscript preparation.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Funding

This research was supported by the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT, South Korea (2022M3H4A1A03067131, RS-2024-00406548, RS-2024-00345700, RS-2026-25538961). W.Y. acknowledges the support from the Office of Naval Research (MURI Award N00014-23-1-2001) and the National Science Foundation (NSF) under Award DMR-2210586.

Data availability

The authors declare that the main data supporting the findings of this study are available within the paper and its Supplementary Information files. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-75568-7.