Amplifier-Assisted Organic Electrochemical Neurons

• Amplifier-assisted organic electrochemical neurons are hybrid circuits that merge op-amp feedback with OECT nonlinearities to convert ionic signals into self-sustained spiking behaviors.
• These systems exploit volumetric electrochemical doping and interfacial ion exchange to achieve high transconductance and biologically relevant timing in aqueous environments.
• Integrating nonlinear device physics with circuit-level feedback, they enable practical biointerfaces and neuromorphic platforms for sensing and artificial neural networks.

Amplifier-assisted organic electrochemical neurons are hybrid bioelectronic circuits in which electrolyte-gated organic electrochemical transistors (OECTs) are combined with amplification stages to convert ionic, chemical, or membrane-voltage perturbations into nonlinear electronic dynamics, including self-sustained oscillations and conductance-based spiking. In the most explicit formulation, the architecture merges a feedback oscillator stage implemented with an operational amplifier and resistive feedback with a self-sustained negative-resistance oscillator stage implemented by an OECT. In a broader materials-and-systems sense, the topic also includes interface engineering strategies that strengthen aqueous ionic-to-electronic transduction, and biointerfaces that couple organic semiconductors directly to living neurons or other excitable tissues. Across this literature, the central technical problem is consistent: how to obtain large transconductive gain, biologically relevant timing, and chemically meaningful operation in water while preserving the soft-matter, mixed ionic-electronic character that distinguishes organic bioelectronics from conventional solid-state neuromorphic hardware (Rivera-Sierra et al., 1 Aug 2025, Harikesh et al., 2022, Bischak et al., 2019, Antognazza et al., 2012, Kleemann, 2021).

1. Hybrid circuit concept and device basis

The defining architecture is a hybrid oscillator in which the operational amplifier provides inversion and gain, while the OECT supplies the nonlinear active element. In the model introduced for amplifier-assisted organic electrochemical neurons, the membrane voltage $V_{\mathrm{mem}}$ is fed into an op-amp resistive network, the op-amp output generates an inverted amplified gate signal $V_G$, that gate biases the sodium-like OECT branch, and the resulting sodium current feeds back into the membrane node. The gain is written as $A = 1 + R_f/R_{\mathrm{in}}$, and the gate voltage is written as $$V_G = A V_0 - (R_{\mathrm{in}}/R_f)V_{\mathrm{mem}}$$. This formulation is technically significant because it makes clear that the op-amp does not by itself create the spiking dynamics; rather, it shapes membrane-voltage feedback so that the OECT is driven into the regime where its nonlinear ionic-electronic response can sustain oscillations. A recurrent misconception is therefore that the amplifier is the source of excitability. The cited modeling states the opposite: the source of self-sustained oscillation is the negative differential resistance or negative transconductance regime of the OECT sodium branch, while the amplifier sets the polarity and magnitude of the coupling needed to reach that regime.

This circuit-level view depends on the distinct physics of OECTs. Unlike conventional organic thin-film transistors, OECTs are low-voltage, electrolyte-gated, volumetric devices in which ions penetrate the semiconductor film and modulate bulk conductivity by electrochemical doping or de-doping. The thesis literature emphasizes that this volumetric operation underlies unusually large transconductance, high volumetric double-layer capacitance, and built-in temporal dynamics arising from ion motion. In bioelectronic terms, the OECT is therefore both a transducer and an amplifier: small ionic perturbations can produce large changes in electronic current, which is precisely the condition required for neuron-like front ends and mixed ionic-electronic feedback loops (Rivera-Sierra et al., 1 Aug 2025, Kleemann, 2021).

2. Interfacial ion translation for aqueous bioamplification

A central obstacle for amplifier-assisted operation in water is that many of the highest-mobility organic semiconductors are too hydrophobic to take up biologically relevant ions such as $\mathrm{Cl^-}$, $\mathrm{Na^+}$, or $\mathrm{K^+}$. The ion-exchange-gel OECT addresses this mismatch by placing an ion exchange gel between the aqueous electrolyte and a hydrophobic semiconductor channel. In the reported p-type OECT, the active layer is PBTTT, while the interfacial gel consists of PVDF-HFP and the ionic liquid BMIM TFSI. The mechanism is a two-step ion exchange pathway: under gate bias, chloride ions from the aqueous phase enter the gel, while $\mathrm{TFSI^-}$ ions from the gel migrate into PBTTT and compensate positive charge on the polymer backbone during p-doping. The semiconductor is thus gated from water without requiring the polymer itself to become hydrophilic. The authors explicitly describe this as exchanging aqueous $\mathrm{Cl^-}$ for $\mathrm{TFSI^-}$ at the interface, and the enhanced doping is localized to the polymer region in contact with the gel, indicating an interfacial mediation mechanism rather than a global modification of the aqueous electrolyte.

The reported performance changes are large enough to matter directly for organic electrochemical neurons and bioamplifiers. For PBTTT, adding the gel increases transfer current by more than four orders of magnitude at $V_G$0, and the transconductance rises from $V_G$1 to $V_G$2 for $V_G$3. For P3HT, the increase is about 24,000-fold, from $V_G$4 to $V_G$5. The intrinsic figure of merit $V_G$6 reaches $V_G$7 for PBTTT and $V_G$8 for P3HT. Ion injection is also accelerated: the dominant absorption decay constant for PBTTT drops from $V_G$9 to $A = 1 + R_f/R_{\mathrm{in}}$0, while P3HT improves from $A = 1 + R_f/R_{\mathrm{in}}$1 to $A = 1 + R_f/R_{\mathrm{in}}$2. In transistor switching, the turn-on and turn-off time constants are $A = 1 + R_f/R_{\mathrm{in}}$3 and $A = 1 + R_f/R_{\mathrm{in}}$4, respectively. The paper is explicit that these kinetics remain roughly $A = 1 + R_f/R_{\mathrm{in}}$5 slower than the fastest OECT systems based on more hydrophilic channel materials, so the gel should be understood as a substantial interfacial improvement rather than an absolute speed optimum.

Its relevance to amplifier-assisted neurons is demonstrated by biological recording. A PBTTT OECT with ion exchange gel was placed on a Venus flytrap together with a standard Ag/AgCl electrode. Upon mechanical stimulation of the trigger hairs, the OECT recorded extracellular action potentials simultaneously with the reference electrode. With the gel, the depolarization peak increased from $A = 1 + R_f/R_{\mathrm{in}}$6 to $A = 1 + R_f/R_{\mathrm{in}}$7, while the noise floor remained about $A = 1 + R_f/R_{\mathrm{in}}$8, and the signal-to-noise ratio rose from $A = 1 + R_f/R_{\mathrm{in}}$9 to $$V_G = A V_0 - (R_{\mathrm{in}}/R_f)V_{\mathrm{mem}}$$0. A plausible implication is that ion-exchange layers provide a practical route for turning hydrophobic, OFET-like polymers into aqueous bioamplifiers suitable for artificial-neuron front ends, neural probes, and mixed ionic-electronic transducers (Bischak et al., 2019).

3. Conductance-based organic electrochemical neurons

A second line of work realizes neuron-like dynamics directly at the device-and-circuit level by exploiting stable ion-tunable antiambipolarity in a mixed ion-electron conducting polymer. The conductance-based organic electrochemical neuron (c-OECN) is built from BBL, poly(benzimidazobenzophenanthroline), a rigid ladder-type conjugated polymer whose ordered backbone permits high electrochemical doping while preserving transport sufficiently to produce a reversible Gaussian transfer curve. In the OECT geometry with Au source/drain electrodes, an Ag/AgCl gate, and aqueous electrolyte, the BBL channel exhibits a current that rises, peaks at $$V_G = A V_0 - (R_{\mathrm{in}}/R_f)V_{\mathrm{mem}}$$1, and then decreases. The rising and falling flanks of that Gaussian-like transfer characteristic are interpreted as analogs of sodium-channel activation and inactivation. A second BBL OECT, configured differently and typically with a thicker film, provides the delayed potassium-like conductance. The circuit includes a membrane capacitor $$V_G = A V_0 - (R_{\mathrm{in}}/R_f)V_{\mathrm{mem}}$$2, effective batteries $$V_G = A V_0 - (R_{\mathrm{in}}/R_f)V_{\mathrm{mem}}$$3 and $$V_G = A V_0 - (R_{\mathrm{in}}/R_f)V_{\mathrm{mem}}$$4, and resistive or time-delay elements such as $$V_G = A V_0 - (R_{\mathrm{in}}/R_f)V_{\mathrm{mem}}$$5. The inverting amplifier converts membrane voltage into the gate sweep required to traverse the antiambipolar peak.

This architecture maps directly onto the Hodgkin–Huxley picture. The membrane equation is written as

$$V_G = A V_0 - (R_{\mathrm{in}}/R_f)V_{\mathrm{mem}}$$6

with the biological reference currents

$$V_G = A V_0 - (R_{\mathrm{in}}/R_f)V_{\mathrm{mem}}$$7

In the organic circuit, input current charges $$V_G = A V_0 - (R_{\mathrm{in}}/R_f)V_{\mathrm{mem}}$$8, raising $$V_G = A V_0 - (R_{\mathrm{in}}/R_f)V_{\mathrm{mem}}$$9; the amplifier inverts and amplifies that signal so that the Na-OECT gate sweeps through $\mathrm{Cl^-}$0; the Na-OECT current then spikes and drives the depolarizing upstroke; the delayed K-OECT subsequently turns on and discharges the capacitor, producing repolarization and short hyperpolarization. The paper notes a secondary Na spike on the return sweep through the Gaussian curve, but states that it is quenched by the K-OECT. The resulting waveform is described as a tonic spiking action potential rather than a leaky integrate-and-fire pulse.

The reported performance places the device in a biologically relevant dynamical regime for organic electronics. OECT switching times are about $\mathrm{Cl^-}$1, the standard configuration spikes at about $\mathrm{Cl^-}$2, and by reducing or eliminating the external capacitor and relying on intrinsic OECT capacitance, the architecture reaches up to $\mathrm{Cl^-}$3, which the paper describes as nearing $\mathrm{Cl^-}$4. The circuit reproduces latency, subthreshold oscillations, integration, refractoriness, resonance, threshold variability, rebound spiking, accommodation, phasic spiking, phasic bursting, class 1 excitability, class 2 excitability, and class 3 excitability. Chemical modulation is integral rather than incidental: $\mathrm{Cl^-}$5, $\mathrm{Cl^-}$6, and $\mathrm{Cl^-}$7 shift the transfer curve differently from $\mathrm{Cl^-}$8 and $\mathrm{Cl^-}$9, while acetylcholine, dopamine, and GABA also modulate the response. The c-OECN can exhibit stochastic spiking near threshold, and GABA can inhibit spiking altogether. The maximum power is around $\mathrm{Na^+}$0 during a spike for the slow-neuron configuration, with energy per spike of roughly $\mathrm{Na^+}$1 for a slow $\mathrm{Na^+}$2-period neuron and about $\mathrm{Na^+}$3 for the faster $\mathrm{Na^+}$4 configuration, not including the inverter. The main limitations stated in the paper are equally relevant: the system remains an analogue emulation rather than a full biochemical reconstruction, the secondary Na spike is an artifact of the Gaussian transfer characteristic, and the energy per spike is still much larger than in optimized silicon or biological systems (Harikesh et al., 2022).

4. Nonlinear dynamical description and oscillation criteria

The most explicit formal treatment casts the amplifier-assisted OECT neuron as a two-variable nonlinear dynamical system:

$\mathrm{Na^+}$5

where $\mathrm{Na^+}$6 is membrane voltage, $\mathrm{Na^+}$7 is an internal state variable, and $\mathrm{Na^+}$8 is an external input or bias current. For the OECT oscillator, the chosen variables are $\mathrm{Na^+}$9 and the potassium-channel activation variable $\mathrm{K^+}$0. The potassium branch is simplified as $\mathrm{K^+}$1, while $\mathrm{K^+}$2 relaxes toward a steady-state sigmoid $\mathrm{K^+}$3 with relaxation time $\mathrm{K^+}$4. The sodium branch is modeled through an activation function $\mathrm{K^+}$5, where both $\mathrm{K^+}$6 and $\mathrm{K^+}$7 are sigmoids with thresholds $\mathrm{K^+}$8 and steepness parameters $\mathrm{K^+}$9. The membrane current balance is written as

$\mathrm{TFSI^-}$0

which makes explicit that the sodium branch contributes both a bias-dependent drive term and a membrane-voltage-dependent feedback term.

The nullcline analysis gives the geometric criterion for oscillation. In the $\mathrm{TFSI^-}$1 phase plane, the $\mathrm{TFSI^-}$2 nullcline is the set where membrane voltage is instantaneously stationary, and the $\mathrm{TFSI^-}$3 nullcline is the set where the potassium activation variable is stationary. If the nullclines intersect at a stable fixed point, the circuit settles to rest. If the intersection lies in the negative-resistance region of the sodium channel and loses stability, the trajectory is repelled and a limit cycle appears. The paper summarizes self-sustained oscillation by four conditions: the $\mathrm{TFSI^-}$4 and $\mathrm{TFSI^-}$5 nullclines intersect only once, that intersection lies in the unstable negative-resistance regime, the equilibrium loses stability, and a limit cycle emerges. It also remarks that along the $\mathrm{TFSI^-}$6 nullcline, a positive slope $\mathrm{TFSI^-}$7 corresponds to the oscillatory region because the sodium branch is used in reversed operation. The onset mechanism is identified as a Hopf bifurcation, while variations in $\mathrm{TFSI^-}$8 produce behavior reminiscent of saddle-node or saddle-node on invariant cycle phenomena. Shorter $\mathrm{TFSI^-}$9 yields compact oscillations; longer $\mathrm{Cl^-}$0 allows larger-amplitude, sharper relaxation-like spikes. Parameter tuning through $\mathrm{Cl^-}$1, $\mathrm{Cl^-}$2, $\mathrm{Cl^-}$3, $\mathrm{Cl^-}$4, $\mathrm{Cl^-}$5, $\mathrm{Cl^-}$6, $\mathrm{Cl^-}$7, $\mathrm{Cl^-}$8, $\mathrm{Cl^-}$9, and $\mathrm{TFSI^-}$0 reshapes the nullclines and therefore the oscillation window. In this framework, design proceeds by bifurcation control rather than by loop gain alone, and the nullcline picture complements but does not replace classical Barkhausen, Nyquist, or Routh–Hurwitz reasoning (Rivera-Sierra et al., 1 Aug 2025).

5. Biointerfaces, photoactivation, and neuromorphic sensing contexts

Amplifier-assisted organic electrochemical neurons belong to a broader family of soft-matter neural interfaces in which organic semiconductors act as active transducers rather than merely passive coatings. A notable precursor is the hybrid bio-organic photovoltaic interface based on rr-P3HT:PCBM deposited on ITO-coated glass and overlaid with a poly-L-lysine layer and cultured primary rat embryonic hippocampal neurons. The device is operated in photovoltaic mode with zero applied bias, $\mathrm{TFSI^-}$1, and the preferred interpretation is a purely capacitive mechanism: photoexcitation produces charge separation in the polymer, charges accumulate at the polymer/electrolyte interface and form a Helmholtz double layer, a corresponding double layer forms at the neuron/electrolyte interface, and the extracellular potential redistribution depolarizes the neuronal membrane. The paper argues against a significant Faradaic mechanism because photocurrents are only a few hundred pA, no visible electrode degradation is observed, pH drift remains below $\mathrm{TFSI^-}$2 pH units, and water electrolysis would be expected only at much larger biases stated as higher than $\mathrm{TFSI^-}$3. Under $\mathrm{TFSI^-}$4 illumination at $\mathrm{TFSI^-}$5, targeted by a digital micromirror device to a region of interest around the soma, a $\mathrm{TFSI^-}$6 pulse evokes a single action potential and a $\mathrm{TFSI^-}$7 pulse evokes burst firing. Repeated stimulation with a $\mathrm{TFSI^-}$8 train of $\mathrm{TFSI^-}$9 pulses yields robust light-locked activation in $V_G$00 cells, while control neurons on glass/ITO substrates do not respond. The organic photodiode retains activity after $V_G$01 days of immersion in saline media, operates for more than $V_G$02 hours with negligible efficiency loss, and persists for more than one month.

Within the OECT literature, the same theme appears in a different form. The thesis on novel concepts for organic transistors frames OECTs as bridges between biological ionic signaling and electronic computation, emphasizing that they combine low-voltage operation, volumetric electrochemical gating, large transconductance, and temporal dynamics set by ion motion. It states that OECTs can mimic the behavior of biological neurons and synapses and can be used in brain-inspired networks. The thesis also describes current-driven inverter-like OECT readout reaching sensitivities of at least $V_G$03, about $V_G$04 times better than inorganic transistor ion sensors, field-directed electropolymerization for hardware “synaptogenesis,” and delayed-feedback architectures containing a transimpedance amplifier, low-pass filter, delay line, derivative term, and amplifier/inverter that yield bifurcation sequences and chaos for gain values above about $V_G$05, with a fractal dimension around $V_G$06. Reservoir computing with such networks reaches about $V_G$07 accuracy on the Iris flower classification test set. Taken together, these results suggest that amplifier-assisted organic electrochemical neurons are best understood not as isolated oscillators but as elements within a broader soft neuromorphic stack that can include sensing, transduction, plasticity, feedback, and classification in the same material platform (Antognazza et al., 2012, Kleemann, 2021).

6. Limitations, interpretive issues, and design directions

Several technical limits recur across the literature. OECTs are intrinsically slow because ion motion limits speed, and this trade-off is directly visible in neuron-like and sensing applications. The ion-exchange-gel OECT improves switching and recording in water, but its kinetics remain $V_G$08 slower than the fastest OECTs based on more hydrophilic channel materials, and the recorded Venus flytrap signals broaden somewhat. The c-OECN attains biologically plausible timing for organic devices, yet its action potential remains an analogue emulation, not a biochemical reconstruction of ion-channel kinetics, and its energy per spike remains high relative to optimized silicon or biological systems. At the architecture level, the thesis points to device variability, stability of the ionic environment, limited integration density, cross-sensitivity in sensing, and the still-early state of complete neuromorphic architectures that include both synapses and true artificial neurons.

Interpretive issues are equally important. In the photoactivation literature, the exact balance between capacitive and any residual Faradaic contribution is inferred rather than directly measured, even though the preferred explanation is capacitive coupling through the electrolyte. In the oscillator literature, linear feedback criteria alone are insufficient because the operative mechanism depends on nonlinear current-voltage characteristics, internal state dynamics, and bifurcation structure. In the ion-gel literature, the enhanced aqueous gating does not mean that the underlying hydrophobic semiconductor has become intrinsically hydrophilic; the mechanism instead depends on interfacial ion translation, specifically the exchange of biologically relevant ions for large hydrophobic ions.

The design directions stated in the cited works are correspondingly concrete. For ion-gel OECTs, possible improvements include reducing polymer thickness, increasing the polymer/gel interfacial area through micro- or nanostructuring, and designing gels with higher ion mobility. For OECT neuromorphic systems more broadly, the thesis calls for better control of synaptic growth and plasticity, improved materials with lower power consumption and possibly normally-off behavior, more stable and scalable networks, and integration into edge-intelligent sensor systems. For amplifier-assisted OECT oscillators, the dynamical-systems formulation implies a design methodology based on control of nullclines, timescale separation, and bifurcation onset. A plausible implication is that future amplifier-assisted organic electrochemical neurons will be judged less by any single device metric than by how effectively they coordinate interfacial ion transport, nonlinear transfer characteristics, and circuit-level feedback in aqueous, chemically rich environments (Bischak et al., 2019, Rivera-Sierra et al., 1 Aug 2025, Antognazza et al., 2012, Harikesh et al., 2022, Kleemann, 2021).