Authors: Qi Wang (Department of Materials Science and Engineering, The Ohio State University, Columbus, OH, 43210, USA), Jinghua Li (Department of Materials Science and Engineering, The Ohio State University, Columbus, OH, 43210, USA; Chronic Brain Injury Program, The Ohio State University, Columbus, OH, 43210, USA)
Categories: Review, cardiac stimulation, flexible electronics, neuromodulation, nongenetic modulation, optoelectronics
Source: Small Methods
Authors: Qi Wang, Jinghua Li
Precise modulation of excitable tissues—including neurons and cardiomyocytes—is essential for both understanding physiological functions and developing advanced therapies for neurological and cardiac disorders. Conventional modulation techniques such as electrical stimulation, pharmacological intervention, and optogenetics, face limitations in terms of invasiveness, spatiotemporal resolution, and/or requirement for genetic modulation. Optoelectronic interfaces based on light‐matter interaction have emerged as promising alternatives. These platforms offer wireless, nongenetic modulation capabilities with high spatiotemporal resolution and minimal invasiveness and risks of infection. Here, a summary of recent advances in nongenetic optoelectronic modulation strategies is presented. Aspects such as material selection and processing, device designs, working principles, and fabrication techniques are discussed. Then, key characterization methodologies, including benchtop assessments and validation within the living systems are discussed. Alongside the discussion, representative applications across in vitro and in vivo models of cardiac and central/peripheral nervous systems are highlighted. Finally, future directions and clinical opportunities, aiming to provide a thorough reference for the continued development of this field for both fundamental research and next‐generation therapeutic applications are explored.
Many vital physiological functions—such as neural information processing, cardiac rhythm, and muscle contraction—depend on the precise generation, propagation, and response to electrical signals in excitable tissues.^[^
^1^ , ^2^
^]^ These tissues primarily include neurons in the central and peripheral nervous systems, cardiomyocytes in the heart, and both skeletal and smooth muscle cells.^[^
^3^ , ^4^
^]^ Although diverse in structure, function, and developmental origin, these excitable cells and tissues share common electrophysiological they generate action potentials through dynamic changes in membrane voltage, enabling rapid and controlled physiological responses.^[^
^5^ , ^6^
^]^ Accordingly, the ability to modulate their electrical activity with high spatial and temporal precision is essential for understanding physiological functions, building disease models, and developing novel therapeutic strategies.^[^
^7^ , ^8^ , ^9^
^]^
In the nervous system, neurons communicate through electrical impulses and synaptic transmission, enabling the integration of external stimuli, regulation of internal states, and generation of complex behaviors.^[^
^10^ , ^11^
^]^ External modulation of neuronal activity (i.e., neuromodulation) has become a widely used tool in both basic neuroscience and translational engineering.^[^
^11^ , ^12^ , ^13^
^]^ It allows for targeted control of specific neuronal populations, supports functional mapping of brain circuits, and enables investigation of causal relationships between neuronal firing patterns and perception, memory, or behavior.^[^
^14^ , ^15^
^]^ Clinically, neuromodulation has been applied to a range of neurological and psychiatric disorders, including Parkinson's disease, epilepsy, depression, and chronic pain, offering effective interventions particularly for patients who are refractory to conventional pharmacological treatment.^[^
^16^ , ^17^ , ^18^
^]^
Meanwhile, excitable cells in nonneuronal systems—such as the heart and skeletal muscle—also rely on precise electrical regulation to maintain rhythmic contraction and coordinated motor output.^[^
^19^ , ^20^ , ^21^
^]^ In the heart, synchronized depolarization of cardiomyocytes is critical for effective blood ejection, and disruptions in electrical signaling can lead to life‐threatening arrhythmias.^[^
^22^
^]^ In skeletal muscle, action potentials triggered by motor neuron input drive voluntary movement, and disruptions in this process can cause various neuromuscular disorders.^[^
^23^ , ^24^
^]^ In recent years, researchers have increasingly extended the conceptual and technological framework of neuromodulation to a broader class of excitable tissues, with promising developments in cardiac rhythm management, muscle function restoration, and bioelectronic therapies.^[^
^25^ , ^26^
^]^
Currently, the most widely adopted excitable tissue modulation techniques fall into three primary electrical stimulation, pharmacological modulation, and optogenetics.^[^
^13^ , ^27^ , ^28^
^]^ Electrical stimulation is the most mature form of neuromodulation and broader bioelectrical intervention.^[^
^29^
^]^ It delivers controlled electrical currents via implanted or surface electrodes to alter cellular membrane potentials and elicit action potentials.^[^
^30^ , ^31^
^]^ Clinically approved applications include deep brain stimulation for Parkinson's disease, spinal cord stimulation for chronic pain, and vagus nerve stimulation for epilepsy and depression, as well as cardiac pacing and defibrillation for arrhythmia management, and peripheral nerve or muscle stimulation for motor function restoration.^[^
^18^ , ^29^ , ^32^
^]^ Its key advantages include rapid, reversible effects and real‐time control over stimulation parameters.^[^
^33^ , ^34^
^]^ However, pixelated electrical implants involve time‐consuming fabrication processes and are limited in their achievable spatial resolution compared with optical methods.^[^
^31^ , ^35^ , ^36^
^]^ Its broader application is also constrained by several other factors, including the need for bulky external equipment to deliver electrical signals, the invasive nature of surgical implantation procedures, and the risk of chronic inflammation or tissue damage at the electrode–tissue interface with prolonged use.^[^
^36^ , ^37^
^]^
Pharmacological modulation regulates electrical activity in excitable tissues by administering bioactive compounds that influence neurotransmitter levels, receptor activity, or ion channel function.^[^
^38^ , ^39^ , ^40^
^]^ It is a mainstay in the treatment of a wide range of neurological and psychiatric disorders, including epilepsy, depression, anxiety, and Parkinson's disease, and is also employed in cardiology for anti‐arrhythmic therapies.^[^
^38^ , ^41^ , ^42^
^]^ Drugs targeting dopaminergic, serotonergic, γ‐aminobutyric acid‐ergic, and glutamatergic systems have demonstrated clinical efficacy and are routinely used in both acute and chronic settings.^[^
^38^ , ^42^
^]^ Pharmacological approaches offer broad accessibility, well‐established delivery methods, and systemic or local effects depending on formulation.^[^
^38^ , ^43^
^]^ However, major limitations low temporal resolution, as drug effects typically develop and subside over minutes to hours; poor spatial specificity, since systemically delivered agents influence multiple tissues or regions simultaneously; and the risk of systemic side effects or tolerance, which can compromise safety and reduce long‐term efficacy.^[^
^44^ , ^45^ , ^46^ , ^47^ , ^48^ , ^49^
^]^
Optogenetics enables precise modulation of neuronal activity by genetically introducing light‐sensitive ion channels, such as channelrhodopsins or halorhodopsins, into specific neuronal populations.^[^
^28^ , ^50^ , ^51^
^]^ Upon light exposure, these channels open or close, allowing precise control of neuronal firing. The technique offers millisecond‐level temporal resolution and cell‐type specificity, making it one of the most powerful tools for dissecting neural circuits and causally linking neuronal activity to behavior.^[^
^28^
^]^ However, its application requires genetic modification of target cells, typically through viral vectors or transgenic methods, which raises ethical and safety concerns in clinical contexts.^[^
^52^ , ^53^
^]^ These limitations have hindered its direct translation to human therapies, despite its experimental success.^[^
^53^ , ^54^
^]^
These limitations have prompted the exploration of nongenetic strategies for modulating excitable tissues. Among them, optoelectronic stimulation using semiconductor‐based devices has emerged as a promising approach.^[^
^55^ , ^56^ , ^57^
^]^ By converting light into electrical signals through optoelectronic effects with an optical power level similar to those used for optogenetics, these devices modulate the surrounding ionic environment at the biotic/abiotic interface, thereby altering the membrane potential in a range of excitable cells without requiring gene delivery or systemic drug administration.^[^
^58^
^]^ This strategy offers several key wireless operation, high spatial resolution, and minimal invasiveness.^[^
^59^ , ^60^
^]^ Moreover, these devices are compatible with flexible, ultrathin, or bioresorbable substrates, making them well‐suited for implantation.^[^
^61^ , ^62^
^]^ As such, optoelectronic modulation of excitable tissues provides a compelling platform for both basic neuroscience research and the development of next‐generation therapeutic technologies, particularly in contexts where conventional methods face technical or ethical constraints.^[^
^63^ , ^64^
^]^
In this review, we present an overview of the current advances in nongenetic, optoelectronic modulation of excitable tissues. We begin by discussing material selection and device architecture, followed by a summary of their working principles and fabrication strategies. Key characterization methods are then summarized, including electrochemical performance testing through benchtop experiments and cellular‐level electrophysiological and optical validation. Finally, we highlight representative in vivo applications across the nervous, cardiac, and peripheral motor systems, demonstrating the translational potential of these technologies for future therapeutic development.
Semiconductors that convert optical energy to electrical signals are central to optoelectronic modulation devices. A wide range of semiconductor materials has been developed for optoelectronic applications, including silicon‐based,^[^
^61^ , ^62^ , ^65^
^]^ compound (e.g., III–V) semiconductors,^[^
^66^ , ^67^ , ^68^
^]^ organic semiconductors,^[^
^59^ , ^69^
^]^ perovskites,^[^
^70^ , ^71^ , ^72^
^]^ and dye‐sensitized materials.^[^
^73^ , ^74^ , ^75^
^]^ However, for in vivo applications, strict requirements on biocompatibility, fabrication compatibility, and mechanical flexibility significantly limit the choice of materials.^[^
^76^
^]^
Among the available options, silicon‐based and organic semiconductors have become the most widely used, mainstream platforms. Silicon offers excellent chemical stability and bioresorbability, high photoresponse, fast charge transport, compatibility with scalable microfabrication techniques, and tunable electronic properties through controlled doping.^[^
^76^ , ^77^
^]^ Despite its inherently high Young's modulus, silicon can achieve mechanical flexibility when engineered into ultrathin layers, because the flexural rigidity of a material scales with the cube of its thickness.^[^
^78^
^]^ An alternative strategy includes engineering the monolithic Si membrane into microstructures—such as serpentine, mesh, or porous architectures—that can effectively relieve and localize strain.^[^
^60^ , ^65^
^]^ These structural engineering approaches substantially lower the apparent bending stiffness without altering the intrinsic modulus of silicon, thereby enabling conformal integration with soft and dynamic biological tissues.^[^
^61^ , ^62^
^]^ Despite its advantages, the relatively high Young's modulus of silicon poses challenges for seamless integration with soft, deformable, and dynamic tissues and organs. Organic semiconductors, such as perylenetetracarboxylic diimide (PTCDI, n‐type), metal‐free phthalocyanine (H2Pc, p‐type),^[^
^59^
^]^ and poly(3‐hexylthiophene) (P3HT, p‐type),^[^
^69^
^]^ possess intrinsic flexibility, can be solution‐processed at low temperatures, and exhibit acceptable stability in physiological environments.^[^
^59^ , ^64^
^]^ These advantages enable their integration into soft, biocompatible interfaces for optoelectronic modulation. Beyond these platforms, colloidal quantum dots (QDs) have recently emerged as an additional class of photoactive materials for biointerfaces.^[^
^79^
^]^ Their size‐dependent bandgaps enable tunable absorption from the visible to near‐infrared, and their large absorption cross‐sections support efficient carrier generation under low irradiance.^[^
^80^
^]^ Cadmium‐ and lead‐free compositions (e.g., InP/ZnS,^[^
^81^
^]^ AgBiS2)^[^
^82^
^]^ exhibit improved biocompatibility and aqueous stability compared with CdSe,^[^
^83^
^]^ highlighting their potential as a complementary materials platform for neural interfaces.
Optoelectronic modulation is achieved through the directed migration of photogenerated charge carriers at the semiconductor–electrolyte interface, where changes in the interfacial electric field reshape the local ionic distribution and thereby regulate the membrane potential of excitable cells such as neurons and cardiomyocytes (Figure
1a,b).^[^
^56^ , ^57^
^]^ Taking neurons as an example, in the resting state the membrane potential is maintained at ≈–70 mV by the asymmetric ion distribution (Na⁺, K⁺, Cl^−^) and selective channel permeability.^[^
^59^
^]^ When a p⁺n‐structured device is illuminated, the interface adjacent to the neuron accumulates negative charge, attracting cations from the extracellular space and reducing the transmembrane potential difference. Once the membrane potential approaches the threshold (about –55 mV), voltage‐gated Na⁺ channels open, Na⁺ rapidly enters the cell, and an action potential is triggered.^[^
^84^
^]^ This depolarization propagates along the axon, followed by delayed K⁺ efflux that repolarizes the membrane and restores the resting state. By contrast, illumination of an n⁺p device leads to positive interfacial charging, which attracts anions or repels cations, thereby increasing the potential difference across the membrane and inducing hyperpolarization. Under these conditions, action potential propagation is blocked, resulting in an inhibitory effect.^[^
^85^
^]^
![Figure 1: Design, Fabrication, and Characterization of Optoelectronic Modulation Devices for Excitable Tissues. a–b) Schematic illustrations of physiological mechanisms underlying photomodulation, including a) excitation and b) inhibition. c–e) Representative energy band structures at different interface structures, including c) silicon PN junctions, d) heterojunctions, and e) semiconductor–solution interfaces. f–h) Fabrication methods for forming optoelectronic interfaces in various formats, including f) ion implantation (Reproduced under terms of the CC‐BY license.^[^
^89^
^]^ Copyright 2023, The Authors, published by AAAS), g) evaporation, and (h) solution‐based methods (Reproduced with permission.^[^
^60^
^]^ Copyright 2024, Springer Nature). i–j) Schematic illustrations of the two predominant characterization methods, along with representative features of the resulting i) Droplet‐based electrolytic recording, and j) Micropipette‐based transient recording. Reproduced with permission.^[^
^90^
^]^ Copyright 2019, Springer Nature.](SMTD-10-e01371-g003.jpg)
These excitatory and inhibitory outcomes arise from the underlying interfacial processes. At the onset of illumination, photo‐capacitive (non‐Faradaic) mechanisms dominate.^[^
^55^ , ^56^
^]^ Band bending formed by Fermi level equilibration drives photogenerated carriers to the interface,^[^
^86^
^]^ where they electrostatically attract counterions to form an electric double layer. The associated displacement current produces a sharp transient modulation, but because it rapidly decays, it cannot sustain prolonged effects. With continuous illumination, photo‐Faradaic processes gradually become electrons or holes are injected across the interface into the electrolyte and participate in redox reactions.^[^
^55^ , ^56^
^]^ This sustained charge transfer further modifies the interfacial potential—electron injection produces negative charging and promotes depolarization, while hole injection produces positive charging and favors hyperpolarization. Upon cessation of illumination, accumulated charges dissipate, counterion distributions relax back toward equilibrium, and the membrane potential recovers close to its resting level, reflecting the reverse process of optoelectronic modulation. Because the capacitive current is transient and mainly occurs as a sharp displacement peak at the onset of illumination before rapidly decaying, it cannot sustain prolonged modulation. To overcome this limitation, enhancing the Faradaic contribution is often necessary to provide continuous charge transfer. In practice, pseudocapacitive materials such as RuO2,^[^
^87^
^]^ MnO2,^[^
^88^
^]^ and Poly(3,4‐ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS)^[^
^63^
^]^ are widely employed not only on return electrodes but also as coatings on the stimulation interface itself, where they increase charge‐storage capacity, facilitate efficient redox‐mediated injection, and improve long‐term stability under physiological conditions.
Despite structural variations among different device designs, the fundamental operational mechanism remains photoexcitation induces charge carrier separation and interfacial accumulation, which in turn modulates local ionic distributions and transmembrane potentials. The nature of the resulting stimulation—whether excitatory or inhibitory—is determined by the direction of photogenerated charge transfer at the semiconductor–electrolyte interface.^[^
^61^
^]^ When electrons migrate toward the electrolyte and holes remain in the semiconductor bulk, this typically leads to membrane depolarization and excitatory stimulation. Conversely, when holes move toward the electrolyte and electrons are retained in the bulk, the result is membrane hyperpolarization and inhibition of excitability. This charge transfer direction is governed by the type of semiconductor material (p‐type or n‐type) and the interfacial band alignment. Figure 1c–e summarize three representative device configurations that employ this principle through different material and structural strategies. The first category comprises p–n junction devices, in which a depletion region forms between p‐type and n‐type semiconductors, creating a built‐in electric field that drives carrier separation. By tuning the doping gradient and junction polarity, these devices can be engineered to direct carrier accumulation at the semiconductor–electrolyte interface, enabling either excitatory or inhibitory responses. Examples include conventional p⁺–n and n⁺–p silicon junctions.^[^
^61^
^]^ The second category consists of heterojunction devices, composed of stacked semiconductors with differing bandgaps, crystal structures, or work functions. These configurations introduce energy‐level asymmetries at the interface that facilitate directional charge separation. Representative architectures include organic heterojunctions such as P3HT (p‐type) and PTCDI (n‐type),^[^
^59^
^]^ as well as junctions between crystalline and porous silicon,^[^
^60^ , ^65^
^]^ where differences in surface states and dielectric properties induce interfacial band offsets despite identical elemental composition. The third category involves direct semiconductor–electrolyte interfaces, where band bending arises from work function mismatch. In p‐type semiconductors, downward band bending drives photogenerated electrons toward the electrolyte, promoting depolarization and excitatory stimulation.^[^
^69^
^]^ In contrast, n‐type semiconductors exhibit upward band bending, which directs holes to the electrolyte, enhancing membrane polarization and producing inhibitory effects. Across all configurations, the ability to control the direction and spatial localization of interfacial charge transfer ultimately determines the nature of the neural response.
To realize the aforementioned junction structures, commonly used fabrication strategies include ion implantation, thin‐film deposition, and solution‐based processing. The ion implantation method is typically implemented on a silicon‐on‐insulator (SOI) substrate, where the doping type and concentration within the device layer are precisely controlled to form a PN junction. Subsequently, the buried oxide layer is removed via HF etching, releasing a flexible PN junction membrane.^[^
^62^
^]^ A typical fabrication workflow on SOI substrates is shown in Figure 1f. This freestanding structure can then be integrated into soft substrates through transfer techniques or adapted into complex 3D configurations using 3D printing.^[^
^89^
^]^ Another approach involves thermal evaporation or related physical vapor deposition techniques, where p‐type and n‐type materials are sequentially deposited to form well‐defined heterojunctions (Figure 1g). For example, p–i–n silicon diode junctions can be fabricated via chemical vapor deposition by depositing intrinsic and n‐type silicon layers onto a p‐type silicon‐on‐insulator substrate.^[^
^90^
^]^ In another case, organic heterojunctions composed of phthalocyanine (p‐type) and PTCDI (n‐type) are formed through sequential thermal evaporation to create nanocrystalline bilayers.^[^
^59^
^]^
Another strategy involves solution‐based selective etching techniques to define functional interfaces within a single material system (Figure 1h). For example, porous silicon (Por‐Si) structures can be prepared through a combination of metal‐assisted chemical etching and stain‐etching, leading to the formation of nanoporous and nonporous regions within p‐type silicon.^[^
^60^ , ^65^
^]^ The nanoporous regions exhibit a wider bandgap due to quantum confinement effects and possess distinct band structures and carrier transport characteristics compared to the nonporous regions, thereby resulting in different electronic properties. Subsequent post‐etching treatments, such as nitric acid oxidation and oxygen plasma exposure, further modify the surface chemistry by introducing thin silicon oxide layers within the porous domains. These treatments enhance the hydrophilicity, interfacial stability, and charge separation efficiency, while also suppressing surface recombination. The combined structural and chemical contrasts between the regions lead to junction‐like band alignment, enabling the formation of functional interface regions with diode‐like behavior without the need for multiple material layers.^[^
^60^
^]^ In parallel, colloidal QDs provide another solution‐processed approach for constructing optoelectronic biointerfaces. They are typically synthesized via wet‐chemical routes, followed by film deposition through techniques such as spin coating or drop casting. For instance, InP QDs have been processed into thin films on glass substrates by spin coating, enabling bidirectional modulation of neuronal membrane potentials, while AgBiS2 QDs have been assembled into pseudocapacitive electrode films through solution deposition, achieving stable photostimulation with good biocompatibility.^[^
^83^
^]^ These fabrication routes illustrate how QDs can be integrated into device architectures using scalable, low‐temperature solution processing, complementing the more established silicon‐ and organic‐based systems.
To evaluate the basic performance of the fabricated devices, it is essential to measure their light‐induced voltage and current responses. Two primary characterization approaches are commonly droplet‐based electrolytic recording and micropipette‐based transient recording, with detailed configurations and mechanisms described below.
Droplet‐based electrolytic recording, also referred to as the electrophotoresponse (EPR) method,^[^
^59^ , ^61^
^]^ is suitable for devices with a backside electrode. A droplet of electrolyte (e.g., phosphate‐buffered saline (PBS)) is placed on the surface of the device, and the open‐circuit voltage and short‐circuit current are measured between a reference electrode (typically Ag/AgCl) inserted into the droplet and the backside electrode of the device. In this configuration, the device forms a closed circuit with the external measurement system, where the current loop is closed through the reference electrode and the electrolyte droplet. Upon illumination, photogenerated electron–hole pairs are separated, and either electrons or holes are accumulated at the interface with the electrolyte, depending on the polarity of the junction. This charge accumulation leads to the formation of one electric double layer at the electrolyte–semiconductor interface and another at the reference electrode, acting as a pair of capacitors in series. These capacitive layers initially drive a displacement current due to charging, leading to a transient current spike. As the double layers become saturated, the capacitive current diminishes rapidly, leaving behind a much smaller steady‐state Faradaic current arising from interfacial redox reactions. During this process, the voltage increases and stabilizes, reflecting the open‐circuit photovoltage generated by the built‐in electric field of the device under illumination. Importantly, although the current drops significantly after initial charging, the photovoltage remains at a relatively stable level, maintained by the charge stored in the double layers. Upon cessation of illumination, a reverse transient current is observed as the capacitors discharge, and the voltage rapidly decays toward baseline without polarity reversal. The total amount of charge transfer is obtained by integrating the photocurrent over time, and this quantity depends on both the light intensity and the duration of illumination. This method enables rapid and straightforward validation of photoelectrode quality under well‐controlled conditions. The typical voltage and current response profiles for this setup are shown in Figure 1i.
It should be noted that, while the droplet‐based method allows for quick assessment of the performance and photocharge density, it does not reflect the actual configuration when the device is in contact with biotissues. To better mimic in vivo operation, a micropipette‐based transient recording method can be employed. In this configuration, the entire device is immersed in PBS, and a specific area on its surface is locally illuminated. A fine‐tipped microelectrode (e.g., a glass micropipette) is positioned just above the illuminated region, and voltage or current is recorded between this microelectrode and a distant reference electrode placed in the solution and treated as ground.^[^
^60^ , ^90^
^]^ Unlike the droplet‐based method, where the measurement system completes the entire external circuit, the micropipette‐based approach passively samples a localized portion of the internal current loop formed within the device–electrolyte system. Both the illuminated region and the backside electrode (exposed to solution) establish electric double layers with the electrolyte, enabling ionic and electronic transport within a self‐contained circuit. The microelectrode acts solely as a probe, capturing localized electrical activity with a high spatial resolution but reduced signal magnitude. Specifically, upon illumination, photogenerated carriers accumulate at the semiconductor–electrolyte interface, attracting counter‐ions from the electrolyte and forming electric double layers according to the mechanism described above. This process generates a transient capacitive current, followed by a smaller steady‐state Faradaic current, similar to that observed in the droplet‐based configuration. When the light is turned off, the relaxation of the double layers induces a reverse transient current. In this setup, the recorded voltage does not reach a steady plateau but instead exhibits a time‐dependent profile shaped by capacitive charging and ionic redistribution. The accumulation of photocarriers and the associated migration of counter‐ions result in local electric field screening, reducing the potential difference detected by the micropipette. Consequently, the voltage signal closely follows the transient current dynamics, in contrast to the more stable photovoltage observed in the droplet‐based method.
In addition to droplet‐based and micropipette‐based configurations, the three‐electrode potentiostat is a classical electrochemical characterization method that provides complementary insights. In this setup, the device under study is employed as the working electrode together with a reference and a counter electrode, enabling precise control of the applied potential and accurate measurement of the resulting currents. This configuration not only allows the characterization of photocurrent and photovoltage, but also enables deeper probing of device–electrolyte interactions. For example, electrochemical impedance spectroscopy (EIS) can decouple charge‐transfer resistance from interfacial capacitance. A lower interfacial resistance generally indicates more efficient charge transfer, while the frequency‐dependent response reflects the device's ability to operate across different time scales, including high‐frequency regimes relevant to rapid stimulation. Cyclic voltammetry (CV), on the other hand, provides information on charge storage mechanisms, distinguishing capacitive behavior from Faradaic processes. The presence and reversibility of oxidation and reduction peaks offer critical insights into whether interfacial reactions are stable, reversible, or prone to degradation. Together, EIS and CV enable a comprehensive evaluation of interfacial charge dynamics and stability, thereby serving as a powerful complement to droplet‐ and micropipette‐based recordings.
Beyond characterizing the photoelectrical properties of the devices at the electrolyte interface, it is essential to verify their biological efficacy through cell culture‐based experiments. These studies directly assess whether the generated electrical signals can effectively modulate the activity of excitable cells such as neurons and cardiomyocytes. At the cellular level, commonly used experimental approaches for the evaluation of modulation outcomes can be roughly grouped into two electrophysiological methods, which record voltage or current changes using microelectrodes, and optical imaging techniques, which monitor intracellular calcium dynamics or membrane potential changes using fluorescent probes. The following section briefly introduces details about the two strategies.
Patch‐clamp based recording, regarded as the gold standard in research on electrophysiology, offers both high‐fidelity recording and precise control of membrane potential. Experiments are typically conducted in a bath solution, where an Ag/AgCl reference electrode is placed at a distal location to maintain a stable baseline potential.^[^
^90^
^]^ A second Ag/AgCl electrode is inserted into a glass micropipette filled with an intracellular‐like solution, which is carefully manipulated to form a high‐resistance seal (i.e., gigaseal) with the target cell membrane. This configuration enables researchers to record rapid changes in membrane potential with high temporal resolution, while also allowing for the injection or withdrawal of current to precisely control the cell's electrophysiological state.
In the context of optoelectronic modulation (mostly in the neuroscience field), the application of patch‐clamp techniques can be broadly categorized into two experimental strategies, corresponding to fine control of membrane potential and dynamic modulation of neuronal firing activity. The first approach involves holding the membrane potential at defined levels to examine whether optical stimulation induces depolarization or hyperpolarization, and whether such changes can elicit action potentials—thereby evaluating the device's capacity to modulate neuron excitability. The second approach entails driving neurons into repetitive firing through continuous depolarizing current injection, followed by optical stimulation to assess its inhibitory effect on action potential generation. Together, these strategies form the foundational experimental framework for assessing the modulation function of optoelectronic devices.
A representative study applies optoelectronic silicon diodes to modulate neuronal excitability in cultured dorsal root ganglion (DRG) neurons (Figure
2a).^[^
^61^
^]^ The device is based on a ≈2 µm‐thick single‐crystalline silicon membrane doped via ion implantation to form either p⁺–n or n⁺–p junctions. Transferred onto a flexible PET substrate and modified with gold nanoparticles, the junction structure establishes a built‐in electric field that supports polarity‐dependent charge accumulation under illumination, enabling bidirectional modulation of membrane potential. Under dark conditions, neurons grown on glass, p⁺n‐type devices, and n⁺p‐type devices exhibit comparable resting membrane potentials, ≈ –50 mV. To evaluate subthreshold responses, the holding potential is set to –65 mV. During 5‐second continuous illumination, neurons on p⁺n‐type devices exhibit depolarization, whereas neurons on n⁺p‐type devices show hyperpolarization, both returning to baseline after light cessation. The amplitude of membrane potential changes increases with light intensity and eventually reaches saturation.
![Figure 2: Representative categories of cell‐based in vitro characterization methods for optoelectronic modulation. Four commonly used techniques are summarized by their respective principle, experimental setup, and evaluation a) patch‐clamp recording (Reproduced with permission.^[^
^61^
^]^ Copyright 2022, Springer Nature), b) MEA recording (Reproduced under terms of the CC‐BY license.^[^
^89^
^]^ Copyright 2023, The Authors, published by AAAS), c) calcium imaging, and d) membrane potential imaging (reproduced with permission.^[^
^60^
^]^ Copyright 2024, Springer Nature).](SMTD-10-e01371-g008.jpg)
To assess the ability to evoke action potentials, the holding potential is adjusted to –45 mV. Under 5‐second illumination, neurons on p⁺n‐type devices generate spikes, with onset latency decreasing from 3.5 to 0.6 s and spike frequency increasing from 3 to 55 Hz as light intensity rises from 0.6 to 2.1 W cm^−2^. This delayed excitation is associated with the presence of a slowly inactivating potassium current that initially counteracts depolarization and gradually diminishes, allowing the membrane potential to accumulate toward threshold. This mechanistic interplay explains the observed dependence of onset latency on light intensity, where lower intensities require longer illumination to elicit spiking. In essence, the excitatory mode reflects the initiation of a new action potential, which necessarily involves a threshold‐crossing process shaped by ionic conductance and therefore exhibits a latency. For inhibition studies, DRG neurons cultured on n⁺p‐type devices are first activated by constant depolarizing current injection to induce repetitive firing. Continuous laser illumination at 532 nm is then applied during the ongoing current‐induced firing for durations of 5, 10, 20, or 30 s. Under these conditions, spike frequency progressively declines with increasing light intensity; complete suppression of firing occurs at ≈1.2 W cm^−2^, and this fully inhibited state can be consistently maintained for up to 60 s when the intensity is increased to 1.5 W cm^−2^. Here, hyperpolarization emerges almost immediately after illumination, as photoinduced currents rapidly alter the balance of ionic fluxes and shift the membrane potential in the negative direction. Unlike excitation, inhibition does not require the gradual build‐up toward threshold, but instead acts on pre‐existing trains of action potentials, blocking their propagation. At subthreshold intensities, this blocking is probabilistic—only a fraction of spikes is suppressed—whereas stronger illumination produces more reliable and sustained inhibition. Thus, while excitation corresponds to the generation of new spikes through a cumulative process, inhibition represents the interruption of ongoing activity through an intensity‐dependent, near‐instantaneous mechanism. Due to its high temporal resolution and precise control over membrane potential, patch‐clamp recording is particularly well‐suited for mechanistic studies of cell–device interactions. However, its low throughput, technical complexity, and limited scalability present challenges for evaluating population‐level responses or performing long‐term functional assessments.
To address these limitations, microelectrode arrays (MEAs) offer an alternative strategy for recording extracellular electrical activity across cell populations. MEAs provide a scalable and noninvasive platform for recording extracellular electrical activity from multiple sites simultaneously. In a typical configuration, cells are cultured on a substrate integrated with MEAs. Each electrode detects local field potential generated by nearby excitable cells, capturing extracellular voltage changes associated with membrane depolarization and repolarization. In electrically coupled tissues such as cardiomyocyte monolayers, excitation initiated at a local site propagates to nearby locations, and signals recorded at different electrodes display temporal delays that reflect the direction and speed of signal propagation.
A representative experiment involves the integration of µ‐solar cells at the center of an MEA platform, where human induced pluripotent stem cell–derived cardiomyocytes (hiPSC‐CMs) are cultured directly above (Figure 2b).^[^
^89^
^]^ These µ‐solar cells are ultrathin, freestanding photovoltaic devices fabricated using microfabrication and photolithography techniques. Each device consists of a silicon‐based p–i–n junction patterned on a silicon‐on‐insulator wafer and released by selectively etching the buried oxide layer, yielding square‐shaped devices ≈80 µm in size. When exposed to pulsed light, they generate localized electric fields that stimulate adjacent cardiomyocytes, enabling precise optical control of cardiac activity. Under pulsed green light stimulation (532 nm, 100 ms, 1.2 Hz), the spontaneous beating rate of the hiPSC‐CMs increases from ≈57 beats per minute to around 72 beats per minute, demonstrating effective light‐driven cardiac modulation. In this experiment, four rounds of photostimulation are performed with the stimulation site remaining fixed. Each round consists of 10 minutes of periodic light pulses (at a frequency of 1 or 1.2 Hz), followed by a 2‐min rest period. During the sequential rounds of stimulation, the resulting activation maps captured by MEAs show progressive synchronization of beating activity, reflected by decreased activation delay and reduced variation in R–R intervals. These observations confirm the capability of the optoelectronic interface to modulate both rhythm and synchronization across multicellular networks. MEA‐based recording enables real‐time monitoring of population‐level responses across large areas and supports long‐term measurements under physiological conditions. However, they are limited in spatial coverage and often require physical contact with the tissue. In contrast, optical imaging techniques provide indirect but highly sensitive readouts of cellular excitation by visualizing physiological indicators such as intracellular calcium concentration or membrane voltage. These methods offer several distinct advantages, including high spatial and temporal resolution, large‐area imaging capability, and compatibility with dynamic biological systems. As a result, optical techniques are particularly valuable for mapping activation patterns across large cell populations and for real‐time visualization of excitation dynamics during photostimulation.
Calcium imaging is a widely used optical technique for evaluating cellular excitation by visualizing intracellular calcium dynamics. In excitable cells, membrane depolarization triggers the opening of voltage‐gated calcium channels, leading to calcium influx from the extracellular space. This rise in intracellular calcium concentration initiates downstream processes such as neurotransmitter release in neurons or contraction in cardiomyocytes. By using fluorescent calcium indicators that change their emission properties upon binding Ca^2^⁺, this method enables real‐time visualization of excitation patterns in both single cells and multicellular networks.
A representative study demonstrates the use of calcium imaging to assess light‐induced modulation in neonatal rat ventricular cardiomyocytes cultured on Por‐Si optoelectronic platforms (Figure 2c).^[^
^60^
^]^ The Por‐Si platform used here is based on a previously reported design, in which a leadless, flexible device is fabricated from p‐type crystalline silicon. The material is patterned into a nanoporous/nonporous heterostructure via stain etching—a chemical oxidation‐based process—and treated with oxygen plasma to enhance its photoelectrochemical response. The resulting 2 µm‐thick membrane is highly conformable and biocompatible, capable of converting low‐power light into localized electrical stimulation without the need for electrodes or external wiring.^[^
^65^
^]^ In the study shown in Figure 2c, the platform is further refined by incorporating self‐limiting stain etching, nitric acid oxidation, and optimized plasma treatment. These modifications yield a high‐surface‐area nanoporous heterojunction that exhibits bipolar photoelectrochemical behavior, characterized by sub‐millisecond (0.3 ms) discharging dynamics and localized current polarity reversal. These effects arise from asymmetric carrier photogenerated holes diffuse broadly through the crystalline matrix, while electrons remain confined within the nanoporous regions. This spatial separation enables simultaneous anodic and cathodic processes on adjacent regions of the same device, supporting fast, localized, and spatially addressable stimulation.
Specifically, cells are incubated with the cell‐permeant fluorescent Calcium Indicator Cal‐520 Acetoxymethyl Ester (Cal‐520 AM), which passively diffuses into the cytosol and is cleaved by intracellular esterases to become fluorescent. Upon pulsed green light stimulation (532 nm, 50 ms, 1 Hz), a rapid and synchronized increase in fluorescence intensity appears across the monolayer, indicating coordinated calcium influx and cell excitation. Concurrently, the beating frequency increases, and the calcium signals exhibit enhanced synchronization and localized propagation patterns, confirming that the optoelectronic platform precisely modulates tissue‐level electrophysiological behavior.
Despite the advantages, calcium imaging exhibits a temporal delay due to its indirect coupling to membrane voltage dynamics. Even kinetically optimized calcium indicators, such as GCaMP6f, show rise times of ≈50 ms and decay constants of ≈140 ms in response to a single action potential.^[^
^91^ , ^92^
^]^ These kinetics are insufficient to capture rapid voltage transients occurring on the millisecond or sub‐millisecond scale. To overcome this limitation, membrane potential imaging provides a complementary strategy, directly visualizing voltage changes using voltage‐sensitive dyes or genetically encoded voltage indicators with temporal resolutions in the range of 0.2–2 ms.^[^
^93^
^]^ Membrane potential imaging utilizes voltage‐sensitive dyes, such as FluoVolt, which integrate into the cell membrane and exhibit fluorescence changes in response to shifts in transmembrane potential. As a complementary approach to address the limited temporal resolution of calcium imaging, the study also employs membrane potential imaging using the voltage‐sensitive dye FluoVolt (Figure 2d).^[^
^60^
^]^ Cardiomyocytes cultured on Por‐Si platforms are stained with FluoVolt and imaged under 488 nm excitation and 500–568 nm emission detection. Pulsed green light (532 nm, 50 ms, 1 Hz) induces localized fluorescence increases, indicating membrane depolarization in the illuminated region. A distinct spatial gradient in signal intensity is observed, with the strongest responses at the illumination center, highlighting the spatial confinement and precision of optoelectronic modulation.
Together, these four methods—patch‐clamp recording, microelectrode arrays, calcium imaging, and membrane potential imaging—offer complementary tools for validating the cellular effects of optoelectronic neuromodulation. Electrical techniques provide direct, high‐resolution measurements of voltage or current, while optical methods enable wide‐field, noninvasive visualization of excitation dynamics. Selection of a specific approach depends on experimental requirements related to spatial resolution, throughput, and the nature of the targeted physiological responses.
Beyond initial validation at the cellular level, further investigations using animal models are essential to evaluate the modulatory performance of optoelectronic devices under complex physiological conditions and to explore their potential applications in functional biological systems. Compared to benchtop testing and cell‐based experiments, studies at the tissue and in vivo levels more closely reflect real‐world application scenarios and present significantly greater complexity. Devices must meet multiple requirements, including biocompatibility, mechanical flexibility, conformability, sufficient signal‐to‐noise ratio, and long‐term stability, in order to interface reliably with soft tissue and maintain functional performance.
On the other hand, the experimental system itself introduces additional challenges, such as precise implantation and fixation of the device, control over illumination area and intensity, and attenuation of light through varying tissue depths. Furthermore, the presence of multisource physiological signals may compromise the stability and reliability of experimental data, complicating the interpretation of results.
Based on the experimental configuration and tissue utilized, existing studies can be broadly categorized into two ex vivo tissue models (e.g., brain slices,^[^
^90^
^]^ excised hearts^[^
^65^
^]^) and in vivo experiments.^[^
^59^
^]^ These can be further grouped by target region into four major application brain modulation, cardiac regulation, limb motor control, and other organ or neural systems (e.g., facial nerves).^[^
^62^
^]^ These studies not only verify device functionality across diverse biological contexts but also provide critical insights for future clinical translation.
In animal studies, functional evaluation methods generally fall into three categories. The first involves direct recording of local physiological signals during stimulation, analogous to cellular‐level electrophysiology. These measurements typically use electrodes to capture action potentials or local field potentials on tissue surfaces. The second approach assesses physiological or behavioral responses at a higher level, such as changes in electrocardiogram (ECG) signals—particularly the QRS complex, a waveform segment composed of the Q, R, and S waves that collectively represent ventricular depolarization—along with heart rate or limb movement, to determine whether stimulation elicits the intended functional outcome. This method is commonly applied in studies involving cardiac or motor system modulation. The third category focuses on post‐stimulation tissue analysis. Techniques such as immunohistochemical staining, tissue sectioning, or quantification of neurotransmitter levels are used to evaluate the efficacy, safety, or underlying mechanisms of stimulation. For example, researchers may examine the expression of neuronal activation markers in the stimulated region or assess whether chronic implantation leads to inflammation or structural damage. These methods are particularly valuable for long‐term evaluation and mechanistic investigations.
As the central component of the nervous system, the brain governs sensation, behavior, emotion, and cognition, making it a primary target for evaluating and applying optoelectronic neuromodulation devices. Compared to other regions, brain‐focused studies emphasize the modulation of specific neural circuits or subregions, offering high spatial selectivity and experimental flexibility. Brain slice models are commonly used in early‐stage evaluations, as they preserve local circuit structures and allow precise targeting of nonsurface neuronal populations. In ex vivo experiments, devices are typically interfaced with acute brain slices and excited using external light sources, while neural activity is recorded using patch‐clamp or extracellular electrodes placed near the target neurons within the slice. In contrast, in vivo experiments, devices are typically placed on the brain surface and stimulated with external light sources, while neural activity is recorded using single or multichannel electrode arrays targeting surface or deep brain structures.
In a representative study (Figure
3a,b), researchers develop a flexible optoelectronic device based on a p–i–n heterojunction formed on a p⁺‐doped single‐crystalline silicon substrate.^[^
^90^
^]^ The top layers, consisting of nanocrystalline intrinsic and n‐type semiconductors, are deposited via chemical vapor deposition, and the entire stack is patterned with engineered microstructures to improve conformal contact with neural tissue and spatial control of illumination. The thin and flexible configuration facilitates integration with the curved surface of the brain while maintaining effective light‐to‐electric conversion. In addition to these demonstrations, systematic investigations across 16 representative silicon device architectures compare photothermal, capacitive, and Faradaic photoresponses, revealing how surface properties, doping conditions, and lateral dimensions govern the balance of these responses. The results indicate that device structure and size directly determine the dominant intrinsic nanocrystalline silicon nanowires, owing to nanoscale confinement effects, exhibit pronounced photothermal responses and are well suited for precise modulation at the intracellular or subcellular level; p–i–n multilayered silicon membranes are dominated by capacitive currents, with negligible thermal and Faradaic contributions, enabling gentle and controllable stimulation at the single‐cell or small‐tissue level; and, building on this, decorating the surface of p–i–n devices with metal nanoparticles (e.g., Au, Ag, Pt) significantly enhances the capacitive response while introducing a more pronounced Faradaic component, with Au demonstrating the most stable and effective performance. This latter configuration is precisely the form employed in the subsequent in vivo cortical experiments. Overall, capacitive‐dominated responses are regarded as the safest and most efficient route for tissue stimulation, whereas metal‐decorated p–i–n devices, while maintaining capacitive dominance, deliver enhanced stimulation strength and are thus particularly suited for in vivo neural modulation.
![Figure 3: Evaluation and application of nongenetic optoelectronic neuromodulation in the central nervous system. A flexible p–i–n silicon device is used for a) localized activation of neuronal circuits in brain slices via light‐induced synaptic responses and b) dose‐dependent cortical spiking and contralateral forelimb motor control in vivo. Reproduced with permission.^[^
^90^
^]^ Copyright 2022, Springer Nature. c) A bioresorbable silicon diode device enables polarity‐dependent excitation or inhibition of cortical activity. Reproduced with permission.^[^
^61^
^]^ Copyright 2022, Springer Nature.](SMTD-10-e01371-g007.jpg)
In the ex vivo experiments, 250 µm‐thick coronal brain slices are prepared to preserve local neural circuitry and are placed in a perfusion chamber containing artificial cerebrospinal fluid. More broadly, whole‐cell recordings in acute brain slices are known to demand stringent conditions. Typical protocols involve rapid preparation under ice‐cold, low‐Ca^2^⁺ cutting solutions, followed by recovery in oxygenated artificial cerebrospinal fluid (ACSF) (95% O2/5% CO2) at 32–34 °C for about 30–60 min before recording. Borosilicate glass micropipettes with a tip diameter of ≈1 µm and a resistance of 3–6 MΩ are commonly used to achieve stable giga‐seals (>1 GΩ), and recordings are performed in ACSF maintained near physiological pH (≈7.3–7.4) and osmolarity (≈300 mOsm). The slices are then positioned directly on top of the device, and light stimulation is applied vertically from above using 473 nm blue light pulses (1 ms duration, 2 mW power, ≈57 µm spot diameter). Whole‐cell patch‐clamp recordings are performed to assess postsynaptic neuronal activity. The recording electrode is positioned on a neuron located away from the illuminated site, and the membrane potential is held at –70 mV to capture excitatory postsynaptic currents (EPSCs). Upon light exposure, short‐latency EPSCs are observed without direct action potential generation, indicating that upstream neurons are activated by the device and that synaptic transmission propagates the signal to the recorded neuron. The EPSC amplitude increases with light intensity, demonstrating a dose‐dependent relationship. These findings verify the device's ability to modulate localized neural circuits and support its application in functional brain mapping.
Subsequent in vivo experiments are conducted in anesthetized mice to further evaluate device performance. The optoelectronic device is placed on the dorsal surface of the motor and somatosensory cortex, and 473 nm blue light (5 mW power, 216 µm spot diameter, 100 ms pulse width) is delivered via a laser scanning system. A 32‐channel linear microelectrode array is inserted at a 30° angle beneath the stimulation site to record neural activity across cortical layers, covering depths from 200 to 900 µm. Light stimulation reliably evokes time‐locked spiking activity across multiple channels, with the strongest responses observed in superficial layers near the illumination center and gradually propagating to deeper regions. The firing rate scales with light intensity, indicating spatially precise and repeatable activation of cortical circuits.
In addition to electrophysiological recording, behavioral experiments are conducted to evaluate the ability of cortical stimulation to induce specific motor responses. The device is placed over the right forelimb motor cortex, and high‐frame‐rate video is used to monitor movement of the animal's limbs. Upon light stimulation, consistent and rapid flexion–extension movements of the contralateral (left) forelimb are observed, and stimulation of the left motor cortex induces mirrored movements on the right side. These responses are highly synchronized with the onset of light pulses and exhibit strong trial‐to‐trial repeatability. As the light intensity increases, both the angular amplitude and the probability of limb movement increase, following a clear dose–response trend. These results confirm the potential of the device for motor control and support its future application in brain–machine interfaces, neural prosthetics, and rehabilitation technologies.
In another representative study from the same work described in Figure 2a, researchers evaluate the neuromodulatory performance of the thin‐film silicon optoelectronic diode in the mouse cerebral cortex (Figure 3c).^[^
^61^
^]^ In vivo experiments are conducted by placing the device directly onto the dorsal surface of the somatosensory cortex in anesthetized mice. A 473 nm blue laser is applied vertically above the device with a pulse duration of 1 second. A 32‐channel linear microelectrode array is inserted at an oblique angle beneath the stimulation site, covering a depth of ≈800 µm to record neural activity across cortical layers.
When the p⁺n configuration is used, light stimulation reliably evokes robust spiking activity across multiple channels. The responses are characterized by short latency, temporal synchrony, and progressive propagation from superficial to deeper cortical layers. The firing rate increases with light intensity, indicating a clear dose‐dependent response. In contrast, the n⁺p configuration is applied after pre‐activating cortical neurons with α‐amino‐3‐hydroxy‐5‐methyl‐4‐isoxazolepropionic acid, which induces continuous spontaneous firing. Subsequent light stimulation significantly suppresses neural activity, with consistent inhibitory effects across trials. Control experiments using p‐type silicon films without junction structures do not produce comparable effects, confirming that the modulation originates from the built‐in electric field of the device rather than from photothermal artifacts.
As a vital organ responsible for driving biological activity, the heart sustains blood circulation through rhythmic contractions. Its electrophysiological behavior is highly sensitive to externally applied electric fields, making it both an important target for research and a representative model for evaluating optoelectronic stimulation. However, its mode of excitation fundamentally differs from that of isolated cells or peripheral nerves. In cell‐level experiments, excitation occurs once the membrane potential of an individual cell crosses the threshold, whereas cardiac stimulation depends on collective dynamics at the tissue level. Local depolarization must recruit a sufficient number of cardiomyocytes via gap junction coupling to initiate a propagating excitation wave that drives coordinated contraction. This synchronous recruitment requirement, together with the need to sustain conduction across the myocardium, makes effective activation of the heart intrinsically more demanding than single‐cell or peripheral nerve stimulation. Moreover, unlike peripheral nerves, where individual axons can conduct action potentials independently, cardiac excitation requires continuous wave propagation through coupled cells, which elevates the stimulation threshold. The longer action potential duration and refractory period of cardiomyocytes further narrow the temporal window for modulation, adding to the mechanistic complexity of cardiac control. These anatomical and electrophysiological characteristics also facilitate systematic and reproducible validation of optoelectronic stimulation strategies.
At the tissue level, the ex vivo heart model (the Langendorff setup) serves as a widely used platform for assessing the outcomes of optoelectronic stimulation. It retains the complete 3D structure and intrinsic electrophysiological properties, offering more realistic insights into device–tissue interactions compared to cellular models. Compared to in vivo experiments, it minimizes physiological interferences, allowing precise control over illumination parameters and synchronized electrical recordings. Nevertheless, its limitations include dependence on external perfusion to maintain tissue viability, restricted experimental duration, and the absence of autonomic nervous regulation and hemodynamic feedback. It is thus better suited for short‐term, localized functional validation and is widely adopted for mechanism analysis and parameter optimization.
Among all evaluation methods, the recording and analysis of electrophysiological signals play a central role due to their direct measurement capability and the nongenetic, label‐free nature. Predominant techniques under this category can be classified as ECG or EGM. ECGs are recorded via surface electrodes and reflect the time‐dependent electrical potential generated by integrated activity across the entire heart. They feature standardized waveforms—such as the P wave, QRS complex, and T wave—that are indicative of global cardiac rhythm and conduction. In contrast, EGMs are recorded through electrodes in direct contact with the myocardial surface or interior, capturing localized signals with higher amplitude and temporal precision. EGMs are well‐suited for analyzing local depolarization dynamics, pacing thresholds, and conduction delays.
Figure
4a shows a representative example of the nanoporous/nonporous silicon heterostructure platform discussed in Figure 2c, here applied for functional cardiac stimulation in an ex vivo model.^[^
^65^
^]^ In this experiment, the device is attached to the left ventricle of a rat heart maintained in a Langendorff perfusion system. This system uses retrograde aortic perfusion of oxygenated buffer under constant pressure to sustain coronary circulation and rhythmic contractions in the absence of autonomic innervation. The device attaches firmly to the moist myocardial surface via capillary force, without adhesives or mechanical fixation, and maintains stable contact during perfusion. Light is delivered using lasers at 532 nm (green) or 808 nm (near‐infrared), with a pulse width of 10 ms and a frequency of 4 Hz. Cardiac electrical activity is recorded using needle electrodes placed in the ventricular tissue to monitor epicardial signals. Below the optical stimulation threshold, the recorded signals display only artifacts. Above the threshold, the ventricular signals immediately synchronize with the laser pulses, indicating effective myocardial depolarization and successful overdrive pacing. Threshold testing shows that 532 nm stimulation requires 4.0 mW mm^−2^, whereas 808 nm requires 8.7 mW mm^−2^ to achieve a comparable response. This difference arises from the higher absorption coefficient of silicon at 532 nm, which enhances carrier generation and electrochemical injection efficiency. In contrast, lower absorption at 808 nm necessitates higher light intensity to achieve similar effects.
![Figure 4: Evaluation and application of nongenetic optoelectronic modulation in cardiac models. a) An optoelectronic interface based on a porous/nonporous silicon heterostructure enables light‐triggered ventricular pacing and synchronized activation in ex vivo Langendorff‐perfused rat hearts. Reproduced with permission.^[^
^65^
^]^ Copyright 2022, Springer Nature. b) A 3D‐bioprinted GelMA cardiac patch embedded with µ‐solar cells enables reversible heart rate modulation in vivo in rats. Reproduced under terms of the CC‐BY license.^[^
^89^
^]^ Copyright 2023, The Authors, published by AAAS. c) A porous/nonporous silicon heterostructure optoelectronic interface enables high spatiotemporal resolution cardiac activation across multiple models (mouse, rat, pig), supporting single‐ and multisite stimulation in both ex vivo and in vivo settings. Reproduced with permission.^[^
^60^
^]^ Copyright 2024, Springer Nature.](SMTD-10-e01371-g004.jpg)
Building on this single‐site demonstration, the study further explores the system's spatiotemporal control capabilities. Dual‐site pacing—where stimuli are delivered to both the left and right ventricles either simultaneously or sequentially—is widely used in cardiac resynchronization therapy (CRT) to restore synchronous activation in patients with conduction delays or dyssynchronous contraction. Clinically, CRT typically begins with biventricular pacing using simultaneous stimulation to quickly narrow the QRS complex and improve hemodynamic performance. However, complete synchrony is not always optimal. During therapy, clinicians often fine‐tune the interventricular delay to better match physiological activation patterns and maximize cardiac output. Despite its clinical efficacy, traditional CRT remains limited by the constraints of wired electrical leads, which reduce spatial flexibility, tissue compatibility, and dynamic programmability. This study employs two independent optoelectronic devices to simulate dual‐site pacing in the ex vivo heart. Each device is attached to either the left or right ventricular surface and stimulated independently using lasers of different wavelengths. Microelectrode array recordings reveal distinct activation patterns under different stimulation protocols. Single‐site pacing produces a unidirectional propagation of the activation wave. In contrast, simultaneous dual‐site pacing results in a synchronized activation pattern that shortens conduction time. Introducing a time delay between the two sites enables sequential ventricular activation, more closely resembling the intrinsic conduction sequence.
In vivo heart experiments provide a more realistic evaluation of device function under physiological conditions. Unlike the controlled ex vivo environment, the in vivo setting introduces signal interference, interface variability, and mechanical coupling, placing higher demands on device stability and adaptability. Building upon the µ‐solar cell–based optoelectronic interface described earlier (Figure 2b), a representative study extends this approach to a light‐responsive, 3D bioprinted cardiac tissue for leadless, noninvasive rhythm modulation in vivo (Figure 4b).^[^
^89^
^]^ The construct consists of a gelatin methacryloyl (GelMA) hydrogel matrix embedded with µ‐solar cells and pre‐seeded with neonatal rat cardiomyocytes (rCMs). It is implanted onto the right atrium of an anesthetized rat and covalently bonded to the epicardial surface using a small amount of transglutaminase powder, enabling stable attachment without adhesives or mechanical supports.
Upon pulsed light stimulation (532 nm wavelength, 2.8 mW mm^−^ ^2^ intensity, 40 ms pulse width, 5 Hz frequency, for 12 minutes), the heart rate gradually increased from ≈280 to 302 beats per minute (BPM), representing a 7.8% increase. Within three minutes after cessation of light stimulation, the rate returned to ≈290 BPM, demonstrating reversibility. Infrared imaging confirmed that the surface temperature rise remained below 2 °C, ruling out photothermal artifacts. In control experiments using GelMA constructs without µ‐solar cells, no change in heart rate was observed under identical light conditions, confirming the essential role of the embedded photovoltaic elements in driving the cardiac modulation.
In another representative study (Figure 4c), the optimized optoelectronic platform based on a porous/nonporous silicon heterostructure—previously described in Figure 2c—is applied to achieve precise cardiac pacing in multiple in vivo models.^[^
^60^
^]^ Specifically, in a rat model of acute myocardial ischemia induced by left anterior descending (LAD) artery ligation, the Por‐Si chip is placed directly on the left ventricle without adhesives. A 1 ms, 635 nm laser pulse at 0.94 mW mm^−2^ enables stable pacing at 360 BPM, and the rate extends up to 600 BPM. Increasing the pulse width to 10 ms lowers the threshold to ≈0.73 mW mm^−2^. The device also supports biventricular pacing, generating QRS waveforms that closely match natural sinus rhythm, confirming its ability to coordinate ventricular activation. Researchers apply targeted optical stimulation to 100 individually addressable sites on the Por‐Si device. ECG records the resulting QRS waveforms, each reflecting a distinct activation origin and propagation path. This setup enables high‐throughput and spatially precise control of cardiac excitation. The feasibility of closed‐thoracic operation is validated in a mouse model. After implantation and chest closure, transcutaneous laser stimulation (62.7 mW mm^−2^, 10 ms) produces stable pacing at 360 BPM. This intensity is significantly lower than conventional optogenetic transdermal thresholds, and no inflammation is observed 7 days post‐surgery, indicating strong biocompatibility and long‐term implant potential.
In a porcine model, the device is scaled to 4 cm^2^ and affixed to the epicardial surface of the left ventricle. Under open‐chest conditions, single‐point laser stimulation (20–30 mW mm^−2^) increases the heart rate from 71 to 120 beats per minute. Electrocardiographic recordings reveal significant QRS prolongation, indicating slowed ventricular depolarization and altered conduction pathways. Mapping with a 30‐channel microelectrode array shows that activation originates beneath the illumination site on the left ventricle and propagates sequentially toward the right ventricle, forming a nonphysiological activation pattern initiated from a localized epicardial site. Under normal conditions, cardiac excitation originates in the sinoatrial node and rapidly propagates through the His–Purkinje system to achieve synchronized ventricular depolarization. In contrast, optically induced activation bypasses this conduction network, relying instead on slower, cell‐to‐cell propagation across the myocardium. This conduction delay accounts for the observed QRS widening and highlights the device's capacity to reconstruct activation pathways independently of native pacemaking structures. Further multisite stimulation experiments demonstrate the spatial programmability of the platform. Optical pulses applied to different regions of the device produce distinct QRS waveforms and activation trajectories, reflecting region‐specific control over the site of initiation. In dual‐site pacing experiments, two laser beams are directed at the left and right ventricles, and the timing between pulses is precisely controlled. Simultaneous stimulation produces a fused QRS complex and shortens total activation time, mimicking the effects of clinical CRT. When a defined temporal delay is introduced between the two sites, two separate depolarization wavefronts emerge, producing split or broadened QRS complexes consistent with sequential ventricular activation. This delayed pacing more closely resembles physiological conduction and illustrates the platform's ability to achieve fine spatiotemporal modulation of cardiac excitation in a large‐animal model.
Compared to modulation studies targeting the brain and heart, optoelectronic stimulation in the motor system focuses more on activating peripheral nerves and skeletal muscles, particularly for functional intervention and rehabilitation of the lower limbs. In animal models, the lower limbs are responsible for weight‐bearing and locomotion, with clearly observable movement changes and well‐established evaluation systems, making them a widely adopted platform for validating device performance. Anatomically, the lower limb features well‐organized tissue structures, with target nerves such as the sciatic nerve exhibiting clear trajectories, relatively large diameters, and superficial locations, facilitating conformal device attachment and precise targeting. Motor responses are typically assessed through macroscopic indicators such as gait alterations, joint angles, and muscle contractions, often complemented by EMG to quantify activation intensity and timing, thus forming a comprehensive stimulation–response feedback framework.
In a representative study (Figure
5a), researchers develop a fully biodegradable and flexible optoelectronic neural interface that enables effective modulation of the sciatic nerve in Sprague‐Dawley rats.^[^
^62^
^]^ The device features an ultrathin p⁺n crystalline silicon junction with a 10 nm molybdenum layer directly deposited onto its surface. A 300 nm‐thick extended molybdenum electrode array, patterned separately, is laminated on a poly(l‐lactic acid)‐co‐poly(trimethylene carbonate) flexible substrate and contacts this interfacial layer, thereby establishing effective electrical connection to the underlying silicon junction. This multilayered design provides excellent mechanical compliance, electrical integration, and biocompatibility. The use of molybdenum enhances charge injection capacity by introducing pseudocapacitive behavior, enabling faradaic reactions through its multiple oxidation states, and reducing interfacial resistance. This improvement allows efficient nerve activation upon attachment to the sciatic nerve and stimulation with pulsed red light at 635 nm (10 Hz, 10 ms). EMG recordings of the gastrocnemius muscle and synchronized video tracking confirm a tight correlation between optical stimulation, electrophysiological response, and limb motion. Under a light intensity of 0.95 W cm^−^ ^2^, the Si/Mo device induces compound muscle action potential (CMAP) amplitudes up to 9 mV and limb displacements of 16 mm, outperforming Si/Au and unmodified silicon devices. Even with a reduced device diameter of 2 mm, clear muscle responses are observed, demonstrating the potential for high spatial resolution in neuromodulation.
![Figure 5: Evaluation and application of nongenetic optoelectronic modulation of sciatic nerves in peripheral motor systems. a) A biodegradable Si/Mo optoelectronic interface enables red‐light‐triggered sciatic nerve activation and lower‐limb motion in rats. Reproduced under terms of the CC‐BY license.^[^
^62^
^]^ Copyright 2024, The Authors, published by Springer Nature. b) Polarity‐configured silicon diode optoelectronic interfaces enable bidirectional modulation of peripheral nerve activity for excitation and inhibition. Reproduced with permission.^[^
^61^
^]^ Copyright 2022, Springer Nature. c) An organic optoelectronic interface enables leadless, subcutaneous sciatic nerve stimulation with stable long‐term performance in vivo. Reproduced with permission.^[^
^59^
^]^ Copyright 2022, Springer Nature.](SMTD-10-e01371-g005.jpg)
To further expand the functional capabilities of optoelectronic interfaces, the same study described in Figure 2a applies the silicon diode platform for peripheral nerve modulation using the sciatic nerve model (Figure 5b).^[^
^61^
^]^ Devices with p⁺n and n⁺p configurations are directly attached to exposed sciatic nerves, and red laser light at 635 nm is used for both pulsed and continuous illumination. CMAP recordings and synchronized limb motion tracking are used to evaluate multimodal responses. In the excitation mode, 1 ms pulsed light (0.4 Hz) applied to the p⁺–n device evokes distinct CMAP signals in the gastrocnemius muscle, with light intensity increasing from 0.1 to 1.8 W cm^−^ ^2^ and amplitudes reaching up to 3 mV, accompanied by over 4 mm of hindlimb elevation. No comparable responses are observed under bare nerve illumination or in device‐free controls, indicating that the effect is driven by photogenerated charge injection rather than thermal heating. In the inhibition mode, electrical stimulation electrodes placed proximally (1 ms pulse, 4 Hz) induce repeated hindlimb motion, while the n⁺p device is attached distally and illuminated continuously for 10 s. A notable reduction in both CMAP amplitude and movement magnitude is observed within the 0.1–1.5 W cm^−^ ^2^ range, with an average inhibition of ≈10–15%. No such effect occurs with p⁺n devices or in controls, validating the feasibility of polarity‐controlled optical suppression of neural activity.
Although these studies demonstrate the feasibility of optoelectronic neuromodulation in lower‐limb nerves, they are largely limited to short‐term, acute experiments using direct laser irradiation. Such conditions do not meet the demands of stable, long‐term, and wireless stimulation in deep tissue environments. On one hand, light penetration through deep tissue remains limited and requires careful balancing of wavelength, power, and thermal effects. On the other hand, chronically implanted neural interfaces on moving tissue must simultaneously ensure mechanical conformity, biocompatibility, and sustained stimulation performance.
To address these challenges, a flexible organic electrolytic photocapacitor (OEPC) based on organic semiconductors is proposed and validated for long‐term stimulation of the sciatic nerve in freely moving rats (Figure 5c).^[^
^59^
^]^ The OEPC incorporates a pn heterojunction formed by H2Pc and PTCDI, encapsulated in a parylene C flexible substrate with a total device thickness of ≈5 µm, offering exceptional mechanical compliance and biocompatibility. The device features a ring‐shaped bipolar layout, with a central pn pixel serving as the cathode and a surrounding loop electrode composed of 10 nm gold and 30 nm indium tin oxide (ITO) acting as the anode. Upon deep‐red illumination, it induces bidirectional displacement currents through capacitive coupling. ITO encapsulation protects the gold electrode from corrosion in physiological conditions, ensuring long‐term conductivity and optical transparency.
For in vivo applications, the OEPC is designed with a self‐locking ring structure, where one end of the band‐shaped substrate includes a guiding hole and the other features interlocking teeth. An aluminum reinforcement layer is incorporated to maintain both flexibility and mechanical stability, securing firm attachment to the nerve surface. In acute tests, the OEPC is placed on the sciatic nerve and stimulated with pulsed red light at 638 or 660 nm (9.4 mW mm^−^ ^2^, 1 ms), successfully evoking CMAPs of ≈10 mV in the biceps femoris along with visible muscle contractions. The amplitude is highly responsive to variations in light intensity, pulse width, and pixel size, demonstrating excellent controllability and repeatability.
In chronic experiments, the OEPC is implanted subcutaneously at a depth of 10–15 mm and fixed on the sciatic nerve via the self‐locking mechanism. All animals exhibit detectable CMAP responses within 20 days post‐implantation, with about half retaining robust signals after 60 days, and some showing continued responses beyond 100 days. Under red light pulses at 700 mW (200–1000 µs), the device reliably activates the nerve without invasive procedures, and the CMAP latency remains stable over time, showing little dependence on light intensity or pulse duration. This contrasts with the patch‐clamp experiments on cultured DRG neurons, where onset latency decreased with stronger illumination. The difference reflects the distinct patch‐clamp directly measures single‐cell depolarization dynamics, in which light intensity determines how quickly threshold is reached, whereas CMAPs represent the summed muscle response to synchronous activation of many sciatic nerve fibers. Once threshold activation is achieved, latency is governed mainly by the stable conduction velocity of intact myelinated axons and therefore remains constant across stimulation parameters. Behavioral assessments reveal no adverse effects on gait or coordination, and von Frey mechanical testing along with histological staining confirms the absence of inflammation or neural damage, underscoring the long‐term biocompatibility and safety of the OEPC interface.
In summary, emerging optoelectronic interfaces overcome key limitations of conventional techniques such as electrical stimulation, pharmacological treatment, and optogenetics by enabling nongenetic, leadless, precise, and minimally invasive modulation. Their demonstrated efficacy in controlling neural, cardiac, and peripheral motor systems highlights their potential for broad application in both research and therapeutic contexts. In general, careful selection of materials is necessary, depending on the specific application scenarios and key performance metrics. Each semiconductor material presents distinct advantages and limitations. Silicon nanowires offer highly miniaturized structures and fine spatial resolution, but they face challenges related to confined device dimensions and limited scalability. Emerging flexible devices based on monocrystalline silicon are compatible with standard microfabrication techniques; however, their fast charge carrier diffusion can compromise spatial resolution. Amorphous silicon generally suffers from low energy conversion efficiency and poor stability, primarily due to numerous trap states, grain boundaries and defects within its disordered structure. Organic polymers offer excellent conformability but also have the issue of limited energy conversion efficiency. To maximize performance and address the trade‐off between efficiency and resolution, careful engineering of device interfaces and heterojunction structures can offer additional opportunities for optimization, as demonstrated in recent pioneering works.^[^
^60^
^]^ Equally important is the refinement of fabrication processes to allow scalable, low‐cost, and customizable device production. Additionally, all of these materials face challenges related to long‐term stability, which limits their suitability for sustained use. Moreover, beyond current microfabrication techniques, solution‐based processing, roll‐to‐roll manufacturing, and 3D biofabrication serve as emerging alternatives and may offer viable pathways to produce flexible, conformal devices in various formats tailored to patient‐specific anatomy and physiological needs. To better contextualize these advances, Table
1 summarizes representative studies highlighting diverse device architectures, material properties, operation parameters, and biological validation models.
Another issue related to this field is how to achieve precise, adaptable, and safe optical addressing, especially for modulation in deep regions. While one advantage of this method includes the fact that optics can act at a distance, the effects of heating and irradiation should not be overlooked. This challenge becomes even more pronounced in large animal models. Yet, in many reported cases, light delivery is still achieved by surgical exposure or direct implantation of optical sources, rather than true noninvasive penetration. Although red and near‐infrared light are increasingly used for their superior tissue penetration, most successful implementations still depend on direct illumination at relatively high power levels or the placement of devices in superficial locations. The required light intensity is typically much higher than that used in direct exposure settings. Potential solutions include the development of materials with enhanced responsiveness to low‐intensity near‐infrared light, the incorporation of implantable optical waveguides or micro‐scale light sources for localized illumination, and the use of upconversion nanoparticles that convert deep‐penetrating infrared light into shorter wavelengths capable of activating semiconductor interfaces. Hybrid approaches that integrate multiple physical modalities—optical, electrical, magnetic, or acoustic—may further expand the reach of nongenetic modulation technologies beyond current anatomical constraints.
At the system level, the integration of optoelectronic interfaces with real‐time feedback systems will enable dynamic, closed‐loop modulation, as demonstrated in recent pioneering studies.^[^
^94^
^]^ By incorporating biosensors, data processing units, and adaptive control algorithms, future devices could intelligently adjust stimulation parameters based on physiological feedback. Such systems would improve therapeutic precision and reduce off‐target effects, particularly in chronic bioimplants. Translation to clinical use will require systematic, comprehensive long‐term evaluation of device safety, biocompatibility, and reliability. Attention must also be given to regulatory pathways, ethical concerns, and patient acceptability, especially in the context of implantable and bioresorbable technologies. In conclusion, optoelectronic modulation offers a powerful and versatile approach to controlling excitable tissues. Continued interdisciplinary collaboration across materials science, photonics, bioelectronics, and clinical research will be essential to bring these emerging technologies from proof‐of‐concept to practical application in the treatment of neurological, cardiac, and neuromuscular disorders.
The authors declare no conflict of interest.