The work, published in Biomedical Microdevices, centers on a class of devices known as AlGaN/GaN heterojunction field-effect transistors, or HFETs. These transistors exploit a remarkable property of the aluminum gallium nitride and gallium nitride pair: when the two crystal layers are stacked, a two-dimensional electron gas forms spontaneously at their interface, creating an extraordinarily dense, highly mobile sheet of charge carriers. That property has made GaN-based devices famous in power electronics and radio-frequency amplifiers, but it also makes them exquisitely sensitive sensors of electrical potential, which is precisely what a neural probe needs.

Traditionally, optogenetic experiments rely on metal microelectrodes to eavesdrop on neurons while an optical fiber or LED delivers stimulation. The problem is that light striking a metal electrode, or the surrounding tissue and electrolyte, generates photoelectric artifacts: spurious voltages that can dwarf the faint millivolt-scale spikes neurons produce. These artifacts arise from photovoltaic effects at the electrode surface, photoelectrochemical reactions in the electrolyte, and light-induced currents in the recording circuitry. The result is a recording channel that goes temporarily blind every time the stimulation light turns on, obscuring the neural response the experimenter most wants to see.

The Sun Yat-sen team, led by corresponding author Baijun Zhang of the State Key Laboratory of Optoelectronic Materials and Technologies, took a different route. Instead of metal electrodes, they built their neural probe around AlGaN/GaN HFETs, which offer high sensitivity to extracellular potential changes along with the biocompatibility needed for implantable devices. The transistor architecture also brings a crucial advantage: because the sensing happens at a transistor gate rather than at a direct metal-electrolyte junction, the device can be configured in ways that metal electrodes simply cannot.

The core of the innovation lies in pairing two different flavors of transistor at the tip of the probe. The first, called the recording HFET or R-HFET, has a bare gate region exposed to the extracellular environment. It is highly sensitive to changes in potential at the electrode-tissue interface and serves as the primary channel for collecting neuronal signals. The second, the differential HFET or D-HFET, is nearly identical in structure but its gate is covered with a silicon dioxide passivation film. That insulating layer renders the D-HFET largely deaf to extracellular potentials, meaning it picks up essentially none of the neural activity.

Here is the elegant part: while the D-HFET cannot hear the neurons, it remains just as sensitive as the R-HFET to the light stimulation signal coming from the co-integrated LED photoelectrode. Both transistors sit at the probe tip, experiencing the same optical environment, the same LED pulses, and the same artifact-generating conditions. The only signal the D-HFET registers is the artifact itself. By carefully adjusting the drain-source bias of the D-HFET and then subtracting its output from the R-HFET signal, the researchers can cancel the optical artifact while preserving the genuine neuronal spikes that only the R-HFET detected.

This differential scheme is a clever inversion of a known vulnerability. AlGaN/GaN transistors are, in fact, notoriously light-sensitive: sub-bandgap photons can trap and release electrons at surface states and in the buffer layers, causing shifts in threshold voltage and persistent photoconductivity, effects that have plagued GaN photodetector and transistor designers for years. Prior work in the GaN community has documented these trap-related optical effects extensively, and passivation layers such as silicon dioxide and aluminum oxide have long been used to tame surface states. The Chinese team turned that liability into an asset, engineering a deliberately light-sensitive reference channel whose noise mirrors the noise in the recording channel.

To validate the approach, the researchers carried out simulated biological experiments in phosphate buffered saline, a standard electrolyte that mimics the ionic environment of extracellular fluid. The integrated photoelectrode probe, combining the LED stimulation element with the paired HFET recording elements, demonstrated artifact reduction of 90 percent. That level of suppression means the neuronal signal, which would otherwise be buried under a light-induced transient, can be clearly separated and read out even during active optical stimulation. For optogenetics experiments, where the most interesting neural dynamics often occur within milliseconds of a light pulse, this timing window is exactly where clean data matters most.

The significance extends beyond a single device demonstration. Integrated optogenetic probes that both stimulate and record in a single implant are the core tools of modern circuit-level neuroscience, allowing researchers to manipulate and monitor the same neuronal population in freely moving animals. Earlier efforts to combat photoelectric artifacts have included conductive shielding layers, as in double-sided sapphire optrodes, and careful materials engineering. The differential HFET approach adds a new weapon: an active, transistor-based cancellation scheme that can be tuned electrically through the drain-source bias, offering flexibility that passive shielding cannot match.

The broader GaN biosensing literature also supports the choice of platform. AlGaN/GaN high electron mobility transistor sensors have been used to detect potassium ions, cardiac troponin in physiological samples, and even SARS-CoV-2 spike proteins and virions, thanks to the sensitivity and chemical robustness of the two-dimensional electron gas. Applying the same transistor technology to neural recording, and solving its Achilles heel of optical sensitivity through differential pairing, suggests a versatile device platform that could serve multiple sensing modalities on a single probe. The work was supported by the Guangdong Basic and Applied Basic Research Foundation, the Science and Technology Plan of Guangdong Province, and joint funding from the National Natural Science Foundation of China and the Macao Science and Technology Development Fund.

Subject of Research: AlGaN/GaN heterojunction field-effect transistor neural probes that suppress optical artifacts from integrated LEDs in optogenetics

Article Title: AlGaN/GaN heterojunction field-effect transistors for suppressing optical artifacts from integrated light-emitting diodes

Article References: Cao, X., Ding, Y., Yang, X., Zhao, W., Li, X., Wen, Y., Li, Y., Huang, X., Li, Z., Weng, J., & Zhang, B. (2026). AlGaN/GaN heterojunction field-effect transistors for suppressing optical artifacts from integrated light-emitting diodes. Biomedical Microdevices, 28(4), Article 68. https://doi.org/10.1007/s10544-026-00851-9

Image Credits: AI Generated

DOI: 10.1007/s10544-026-00851-9

Keywords: optogenetics, AlGaN/GaN HFET, neural probe, optical artifacts, LED photoelectrode, differential recording, two-dimensional electron gas, biomedical microdevices, neural signal acquisition, SiO2 passivation, gallium nitride, signal-to-noise ratio

The work, published in Biomedical Microdevices, extends a long-running effort to build what the group calls an artificial placenta: a lung assist device assembled from arrays of single oxygenator units, each a microfluidic chip fabricated from polydimethylsiloxane, the transparent silicone elastomer beloved by microfluidics engineers. PDMS is easy to mold, gas-permeable and optically clear, but in contact with blood it is profoundly thrombogenic. When plasma proteins adsorb onto its hydrophobic surface, they undergo conformational changes that activate the coagulation factors, platelets and complement proteins that normally patrol the vasculature. In a device whose channels are measured in tens or hundreds of micrometers, even a small clot can occlude flow, degrade gas exchange and shed emboli into the patient’s circulation.

The McMaster strategy centers on a molecule with an unusual pedigree. Antithrombin is the body’s natural brake on coagulation, a serine protease inhibitor that neutralizes thrombin and factor Xa. Heparin accelerates this inhibition dramatically, but heparin immobilized on biomaterial surfaces has historically underperformed, because the pentasaccharide sequence that activates antithrombin must be presented in a specific orientation and because bound heparin alone cannot catalyze inhibition without recruiting antithrombin from plasma. To sidestep these limitations, Anthony Chan, John Brash and colleagues developed a covalent antithrombin-heparin complex in which the two molecules are permanently linked, preserving the catalytic machinery in a single, surface-tethered unit. Earlier studies showed that such complexes, when coated onto flat PDMS using polydopamine as an adhesive layer, could render the material blood-compatible.

Polydopamine itself is a piece of bioinspired chemistry borrowed from marine mussels, which anchor themselves to rocks in churning surf using adhesive proteins rich in the amino acid DOPA. When dopamine is oxidized under mildly alkaline conditions, it polymerizes into a thin, conformal film that adheres tenaciously to virtually any surface, from metals to polymers, through a combination of covalent and noncovalent interactions. In the new study, the team flowed a polydopamine solution through the microchannels of the oxygenator units, then introduced the antithrombin-heparin complex under flow as well, allowing the coating to build up uniformly inside the tortuous three-dimensional geometry that flat-surface experiments cannot fully replicate.

Quantifying what sticks to the inside of a sealed microfluidic device is a technical challenge in its own right. The researchers solved it by radiolabelling the antithrombin-heparin complex, which allowed them to measure surface density directly: the coated units carried 0.21 plus or minus 0.05 micrograms of the complex per square centimeter. More importantly, the coating proved durable. When the modified devices were perfused with flowing blood for two days, 76 percent of the bound complex remained on the surface, a stability figure that matters enormously for any device intended to support a neonate for days or weeks. A coating that leaches away within hours would offer only fleeting protection and might itself become a source of embolic debris.

Surface density alone does not guarantee function, so the team also measured whether the immobilized heparin retained its biological activity. Their assay exploited the fact that active heparin binds antithrombin from plasma with high affinity. Devices coated with the antithrombin-heparin complex captured 47.78 plus or minus 10.63 nanograms of antithrombin per square centimeter from plasma, roughly four times the 11.56 plus or minus 4.58 nanograms per square centimeter measured on devices coated with polydopamine alone. That fourfold difference demonstrates that the covalent complex presents heparin in a catalytically competent configuration, effectively turning the entire blood-contacting surface of the device into an anticoagulant reactor that continuously neutralizes thrombin as blood passes through.

The functional consequences were visible at the macroscopic scale. When plasma was perfused through the modified units for one hour, the devices resisted clotting, whereas unmodified or polydopamine-only controls showed the fibrin deposition and flow obstruction characteristic of biomaterial-triggered coagulation. Just as critically, the researchers verified that the coating did not compromise the device’s primary job. Oxygen permeability, the property that allows the thin PDMS membranes to transfer gas between an oxygen supply and the blood, was unchanged by the surface treatment. That dual requirement, anticoagulant function without degraded gas exchange, has been the stumbling block for many previous hemocompatibility strategies, including polyethylene glycol layers and zwitterionic coatings, which can delaminate or alter transport properties over time.

The clinical context gives the work its urgency. Neonatal extracorporeal membrane oxygenation, or ECMO, remains an anticoagulation enigma, as pediatric intensivists have described it, because the systemic heparin required to keep circuits patent exposes infants, whose hemostatic systems are immature, to serious bleeding risks including intracranial hemorrhage. Ventilator-induced lung injury adds another layer of harm for preterm babies, whose alveoli can be damaged by the very pressures meant to keep them alive. A lung assist device whose internal surfaces actively inhibit clot formation could reduce the systemic anticoagulation burden, and the artificial placenta concept envisions pumpless microfluidic oxygenator arrays that could support preterm neonates with far less trauma than conventional extracorporeal circuits.

What distinguishes the new study is the translation from flat substrates to functional devices under realistic flow conditions. Coating chemistry that works on a flat coupon frequently fails inside a microchannel, where flow profiles, channel aspect ratios and surface-to-volume ratios conspire to produce uneven films. By performing both the polydopamine deposition and the complex immobilization under flow, the team showed that the strategy scales to the device level, a prerequisite for assembling the single oxygenator units into the integrated arrays that would constitute a clinical lung assist device. The authors note that the results demonstrate a previously developed modification strategy can be translated from flat PDMS substrates to microfluidic units, providing device-level anticoagulant function without measurably compromising membrane oxygen permeability under the conditions tested.

Challenges remain before the technology reaches the neonatal intensive care unit. The one-hour plasma clotting resistance and two-day stability experiments, while encouraging, must be extended to longer durations, whole blood and ultimately animal models, and the group has already explored pairing the antithrombin-heparin coating with immobilized tissue plasminogen activator to add fibrinolytic activity to the anticoagulant function. Regulatory pathways for combination products that blend a device with a pharmacologically active surface will also demand careful scrutiny. Still, the study offers a compelling proof of concept: by borrowing the adhesive tenacity of a mussel and the catalytic elegance of a natural anticoagulant complex, the researchers have shown that the surfaces of life-supporting microdevices can be engineered to fight the clotting response they provoke, bringing the artificial placenta one step closer to the bedside of the tiniest patients.

Subject of Research: Covalent antithrombin-heparin surface modification of PDMS microfluidic oxygenator units for anticoagulant function in neonatal lung assist devices

Article Title: Surface modification of microfluidic oxygenator units with an antithrombin-heparin (ATH) covalent complex for enhanced anticoagulant function

Article References: Li, S., Sandejas, D., Saraei, N., Dabaghi, M., Atkinson, H. M., Fusch, G., Rochow, N., Fusch, C., Selvaganapathy, P. R., Chan, A. K. C., Brash, J. L., & Sask, K. N. (2026). Surface modification of microfluidic oxygenator units with an antithrombin-heparin (ATH) covalent complex for enhanced anticoagulant function. Biomedical Microdevices, 28(3), Article 60. https://doi.org/10.1007/s10544-026-00842-w

Image Credits: AI Generated

DOI: 10.1007/s10544-026-00842-w

Keywords: antithrombin-heparin complex, microfluidic oxygenator, polydopamine, PDMS surface modification, blood compatibility, neonatal respiratory distress, artificial placenta, lung assist device, anticoagulant coating, hemocompatibility, Biomedical Microdevices, ECMO

The centerpiece of the new design is a spherical capsule robot whose internal magnetic ring is fully suspended within the shell, a configuration inspired by the dynamic balance principles of the gyroscope. In conventional magnetically driven capsule endoscopes, the internal magnet is rigidly fixed, which means external magnetic fields directly couple torque and force into the capsule body in ways that can be difficult to predict and control, particularly when the capsule must roll along irregular, lubricated, and deformable tissue surfaces. The suspended magnetic ring decouples much of this interaction. Because the ring can rotate freely on its suspension, the capsule as a whole can maintain stability even as the internal magnetic element spins rapidly under the influence of an external driving field. This underactuated structure, meaning the robot has fewer controlled inputs than degrees of freedom, is deliberately exploited rather than treated as a drawback, allowing the same hardware to support two distinct operating modes: a stationary all-around observation mode in which the capsule pivots in place to survey its surroundings, and a rolling locomotion mode that transports it across the GI lining.

The external actuation scheme is equally central to the advance. The researchers drive the capsule using what they call a spatial universal rotating magnetic field, or SURMF, a rotating field whose axis can be oriented arbitrarily in three-dimensional space. Earlier work by the same group established the orthogonal transformation operation theorem for such fields, providing the mathematical machinery to generate a uniform rotating field pointing in any desired direction. When the SURMF axis is manipulated, the suspended magnetic ring inside the capsule experiences a rotating torque that, through the rotating magnetic coaxial effect of the suspended magnet, translates into controlled rolling of the outer shell. In essence, the external field spins the internal moment, and the geometry of the suspension converts that spin into locomotion, much as an internal rotor drives a spherical rolling robot. This arrangement affords what the authors describe as orthogonal decoupling of magnetic moments, flexible and efficient posture control, and good stability, qualities that have proven elusive in prior capsule designs that relied on wobbling external magnets, legged mechanisms, or inchworm-like extensile structures.

Yet the elegance of the rotating coaxial approach conceals a subtle problem, and resolving that problem constitutes the core contribution of the new paper. When the capsule rolls under SURMF actuation, the actual orientation of the internal magnetic moment does not always coincide with the nominal orientation commanded by the external field. A slip angle arises between the intended and realized moment directions, and this slip produces two cascading errors: a magnetic moment orientation deviation, which distorts the torque applied to the capsule, and a motion path deviation, which causes the robot to drift away from its planned trajectory across the GI wall. For a device whose entire value proposition is precision, the ability to inspect a specific lesion or hold a fixed viewing position in the stomach, such uncommanded drift is unacceptable. The researchers therefore derived a complete deviation model of the decoupled rotating coaxial magnetic moment, capturing analytically how the slip angle emerges from the interplay of magnetic torque, suspension dynamics, and the resistance of the tissue-contact interface.

The deviation model is not merely descriptive; it is corrective. By characterizing the slip angle as a function of the actuation parameters, the team developed a compensation scheme in which the commanded orientation of the SURMF axis is deliberately offset so that the realized magnetic moment lands exactly where the control algorithm intends. The corrected model was verified experimentally, and the authors report that slip angle compensation lays the foundation for precise and stable control of the motion path of the novel capsule. In practical terms, this means a clinician or automated controller can now specify a path across the stomach wall, or a fixed observation point in the colon, and trust that the capsule will follow it despite the inherently slippery, compliant environment of the GI tract. The significance of this control-theoretic groundwork extends beyond the specific prototype, as the rotating coaxial driving theory establishes a general framework that other magnetically actuated capsule designs could adopt.

The clinical motivation behind the work is substantial. Gastrointestinal cancers remain among the most lethal malignancies worldwide, and early detection through comprehensive endoscopic screening dramatically improves outcomes. Conventional endoscopy, while powerful, is invasive, uncomfortable, and requires sedation, factors that depress screening participation. Passive capsule endoscopy solved the comfort problem but sacrificed controllability, meaning blind spots persist and lesions in the stomach or colon, where the lumen is wide and folded, are frequently missed. Prototype active capsules have explored numerous solutions, including legged microrobots, inchworm mechanisms with extensible anchors, vibro-impact locomotion for colonoscopy, and self-propelled designs, each with tradeoffs in complexity, safety, energy consumption, and tissue compatibility. The underactuated spherical approach offers a mechanically simple alternative: no legs, no extending claws, no onboard propulsion, just a suspended magnetic ring and a spherical shell, with all the intelligence residing in the external field control.

The gyroscope-inspired suspension deserves particular attention as an engineering insight. A freely suspended rotor maintains its orientation and spin characteristics with remarkable stability because gyroscopic dynamics resist changes in the rotor’s angular momentum direction. By embedding a fully suspended magnetic ring in the capsule shell, the designers harness this passive stability to keep the internal actuator’s behavior predictable even as the shell tumbles, contacts tissue, and absorbs impacts. The elasto-hydrodynamic lubrication properties of the GI lining, well studied in tribology, mean that capsules rolling on mucus-coated tissue experience low friction and intermittent slip, precisely the conditions that wreck naive magnetic control schemes. The suspended-ring architecture combined with slip angle compensation directly confronts these conditions rather than assuming them away, which is why the authors emphasize that their deviation model was corrected and verified rather than simply proposed.

The published work includes extensive experimental validation across twelve figures in the main text, with supplementary materials documenting the dynamic behavior of the prototype in detail. The authors report 83 accesses to the article within days of publication, suggesting keen interest in the medical robotics community. The research builds on a decade of cumulative development by the group, including earlier demonstrations of petal-shaped capsule robots, dual-hemisphere capsule designs with highly integrated electronics, and dynamic tracking studies of magnetically navigated capsule robots, as well as foundational theory on the orthogonal transformation of spatial universal uniform rotating magnetic fields published in Science China Technological Sciences.

What remains before such capsules reach patients is the usual gauntlet of translational medical devices: integration of imaging optics and wireless telemetry into the spherical shell, in vivo validation in animal models, and eventually clinical trials. The current study focused on the actuation and control physics, deliberately establishing the theoretical foundation before piling on payload electronics. But the destination is clear and compelling: a patient swallows a smooth, spherical pill, lies comfortably while an external magnetic system steers it to every corner of the stomach, holds it steady while its camera pans a full circle around a suspicious polyp, and then walks out of the clinic without sedation or a tube down the throat. With the slip angle problem now modeled, corrected, and experimentally verified, that vision has moved measurably closer to reality, and the underactuated spherical capsule robot stands as one of the more elegant candidate architectures for making all-around GI disease screening an everyday reality.

The study was supported in part by the National Natural Science Foundation of China under grants 62173059 and 61773084, and the authors declare no competing interests.