A thick, porous transistor channel that stopped being slow

What the work claims

This is a primary materials-engineering paper reporting fabricated and electrically characterized devices, not a simulation or a review. Organic electrochemical transistors, OECTs, use a conducting-polymer channel whose electronic conductance is modulated by ions moving in from a contacting electrolyte under a gate voltage, a mechanism widely used to amplify weak ionic or biological signals into a measurable electrical output. The channel material tested here is PEDOT:PSS, the standard conducting polymer for this purpose, blended with polyethyleneimine (PEI) and processed through pH-induced phase separation into a porous film.

The authors' central, measured result: their best film, at a PSS to PEI ratio of 1 to 2, reaches a transconductance of 30 millisiemens and a response time of 13 milliseconds at an unusually low gate voltage of 0.05 volts, despite a dry channel thickness of 120 to 200 micrometers, roughly 100 to 1,000 times thicker than the sub-micrometer channels conventional OECTs use to stay fast. Conventional thin, optimized OECTs typically report transconductance around 5 millisiemens with response times in the tens of milliseconds. The porous, far-thicker film matches or beats that response time while delivering roughly six times the absolute gain, a combination the authors describe as mitigating, in their own words, rather than eliminating, the field's long-standing gain-speed trade-off.¹

How it works

Channel conductance in an OECT scales with width times thickness divided by length, so a thicker channel simply contains more electrochemically active material to modulate, which should raise transconductance. The catch is that ions from the electrolyte have to physically diffuse through however much of that thickness gets doped or dedoped, and in a dense film that diffusion front only reaches a thin layer near the electrolyte interface before the device has already responded; pushing further into a thick, dense film takes progressively longer, which is why thick, dense channels have historically been slow.

The authors break this coupling by controlling the film's internal architecture rather than only its thickness. PEDOT:PSS is mixed with additional PSS and branched PEI, cast, and immersed in an acidic coagulation bath. Protonation of the PEI amine groups drives electrostatic complexation with PSS's sulfonate groups, triggering aqueous phase separation into a porous, interconnected PEDOT:PSS:PEI matrix, all in a fully water-based process the authors contrast favorably with prior porous-OECT routes that needed freeze-drying or aerogel processing. Separately, treating the film with dimethyl sulfoxide (DMSO) as a co-solvent and dopant, then annealing, raises the electronic conductivity of the PEDOT-rich solid phase, reaching up to 15.65 millisiemens per centimeter.

These two effects are functionally distinct and the paper isolates them: pore connectivity governs how far an ion has to travel locally before it reaches doped material, while DMSO treatment and annealing govern how well the resulting charge then moves electronically through the continuous PEDOT-rich backbone. With an interconnected pore network, electrolyte can penetrate the full macroscopic thickness through short local paths rather than one long diffusion front from the surface, so most of a nominally 150-micrometer-thick film behaves, kinetically, like a much thinner one. The authors demonstrate this is genuinely a morphology effect and not just a conductivity effect with a controlled internal comparison: two film compositions (PSS to PEI ratios of 1 to 1 and 1 to 2) with comparable electronic conductivity but visibly different pore structure gave response times of 39 and 13 milliseconds respectively, with the more interconnected, better-preserved pore network in the 1 to 2 film producing both the higher transconductance and the faster response. All electrical measurements were carried out in 0.1 molar aqueous sodium chloride, an ionic strength within range of physiological saline.

Where a skeptic should push

The authors' own framing is the most important caveat: this mitigates the trade-off, it does not eliminate it. Their fastest device, at 13 milliseconds, lands within, not clearly below, the range the paper itself cites for conventional thin-film OECTs' response times, described as tens of milliseconds. The genuine advance is achieving that speed at far greater thickness and roughly six times the absolute transconductance, not achieving a fundamentally faster device than the best thin-film alternatives.

Geometry-normalized transconductance, gain divided by the channel's width-thickness-over-length ratio, tells a more sobering story: the devices reported here reach only 0.069 to 0.08 millisiemens per micrometer, well below literature values that frequently exceed 2 millisiemens per micrometer. In plain terms, part of the absolute gain advantage comes from simply building a much bigger device, not from extracting more amplification per unit of active material than the best optimized thin films already achieve. The authors are explicit that this reflects a non-optimized macroscopic geometry rather than a material limit, and that further tuning of the channel's width, length and thickness could close this gap, but as published, the efficiency-per-volume comparison favors thinner, denser prior designs.

Finally, the reported figure is a single-pulse, time-domain response time, not a frequency-domain characterization. Converting a 13-millisecond step response into an implied bandwidth is a rough, order-of-magnitude estimate this analysis makes, not a number the paper itself reports, and it should be read with that uncertainty attached. No neurons, tissue, or long-term stability data appear in the paper; this is bench-scale materials characterization in saline, not a demonstration of biological signal acquisition.

Where this speed fix lands on the spike band

Soft, solution-processable, biocompatible amplification right at the electrode-tissue interface is a genuinely attractive proposition for organoid recording, where a rigid CMOS front end is often the wrong mechanical and chemical fit for a three-dimensional, soft, actively growing tissue culture. A channel that can be made thick, and therefore mechanically robust and easy to process at scale, without sacrificing switching speed removes a real constraint that has previously pushed OECT designs toward thin, more fragile films. That is the specific opportunity here: it widens the practical design space for an in-electrode, first-stage amplification element built from a material class that is already the leading candidate for soft, wet, ionically active bioelectronic interfaces, using a scalable, fully aqueous fabrication route rather than one of the more delicate freeze-dried or aerogel processes the paper cites as prior art.

The corresponding hype-correction is just as concrete, and follows directly from the same 13-millisecond number. Extracellular action potentials carry meaningful signal energy up into the low kilohertz range and are conventionally sorted using sub-millisecond timing precision; a device whose step response takes over a hundred times longer than that is not, on this evidence, a candidate to replace a low-noise CMOS amplifier at the spike-detection front end. What this specific advance plausibly reaches is the population-level, burst-band and local-field-potential regime, frequencies well under a hundred hertz, which is itself a substantial share of what organoid MEA experiments actually report, given how much of spontaneous organoid activity is structured around population bursts rather than isolated, individually resolved spikes. Read narrowly, that is real progress: a softer, thicker, more manufacturable amplifying material for the slow band an array already needs to cover. Read broadly, as a step toward organic, in-situ replacement of the fast, spike-resolving front end, it would be an overreach the paper's own numbers do not support, and any product narrative that elides that distinction is worth pushing back on specifically because the underlying measurement is otherwise solid.

The bottom line

Established, and measured with an internal composition-matched control: pore-network connectivity, engineered through pH-induced aqueous phase separation, lets a 120-to-200-micrometer PEDOT:PSS:PEI film reach 30 millisiemens of transconductance and a 13-millisecond response time, a combination competitive with, though not clearly superior to, conventional sub-micrometer OECT channels, at the cost of lower gain per unit of active volume than the best optimized thin films. Hypothesis, not yet shown: that this specific material or process would perform the same way integrated onto an actual recording electrode, in contact with living tissue rather than a saline bath, over any meaningful duration, or that its true frequency response permits any signal faster than population-level activity. What would confirm the stronger reading: a frequency-domain characterization locating the actual bandwidth ceiling, and a demonstration on a real electrode with biological signal. What would break the spike-band extrapolation specifically: nothing further is needed, since the 13-millisecond figure the authors themselves report already places this device roughly two orders of magnitude too slow for individually resolved action potentials on its own terms.