Energy-conversion nano/micromaterial based retinal prostheses for vision restoration

Abstract

Retinal degenerative diseases result in progressive and profound visual impairment. Retinal prosthesis implantation is a promising strategy for retinal degenerative disease treatment. Conventional retinal prostheses based on electrode arrays have limitations such as low resolution and poor biocompatibility. This review focuses on retinal prostheses using energy-conversion nano/micromaterials, including photovoltaic, piezoelectric, upconversion, and photothermal materials. These materials can convert external energy into neural stimulation signals, enabling wireless or self-powered operation. For instance, photovoltaic materials offer high sensitivity and resolution; piezoelectric materials can harness ultrasound for non-invasive neural stimulation; upconversion materials assist humans in distinguishing multiple spectra of near-infrared light; and photothermal materials can stimulate neurons through near-infrared light with better tissue penetration. In addition, the nano/microscale structure of the retinal prosthesis strengthens the physical and chemical properties and provides more sensitive neuronal signal transmission. Although remarkable progress has been made, challenges like achieving high and stable resolution, ensuring long-term biocompatibility, and optimizing the material–neural tissue interface remain. Future research should focus on developing novel nano/microstructured materials, innovative device designs, and a better understanding of the bio-interface to restore more natural visual function for patients with retinal degeneration.

---

Ruiying Li

Ruiying Li is an undergraduate student in the Ophthalmology program at Capital Medical University, expecting to graduate in 2027. Her research focuses on applying bibliometric and bioinformatic methods to investigate ophthalmic diseases. She has first-authored several papers, including bibliometric analyses on cuproptosis and retinal detachment, which are scheduled for publication in 2026. Her interests lie at the intersection of ophthalmology, data science, and advanced nanomaterials.

Yueyang Shang

Yueyang Shang is presently studying for his undergraduate degree at the Forth Clinical School of Capital Medical University (Beijing Tongren Hospital), majoring in clinical medicine, ophthalmology and optometry medicine. His research interests focused on constructing piezoelectric nanoparticles for wireless neurostimulation and myopia treatment.

Tianyu Gao

Tianyu Gao is a postgraduate student at the School of Pharmacy, Capital Medical University. Her current research interests focused on designing and constructing functional nanomaterials, including piezoelectric nanogenerators and so on, for bioimaging, cancer therapy, and neurostimulation.

Xinmiao Lan

Xinmiao Lan obtained a master's degree from the Faculty of Dentistry, The University of Hong Kong in 2015 and a Ph.D. degree from the same faculty in September 2019. She finished her postdoctoral fellowship under the guidance of Prof. Richard Yu-xiong Su at the University of Hong Kong. Currently, she is an Associate Professor at Capital Medical University and an Honorary Associate Professor at the University of Hong Kong. Her research interests are nanoparticles, advanced nanomaterials, chemotherapy, immunotherapy, and cancer therapy.

Peijian Feng

Peijian Feng is presently an associate professor at the School of Pharmaceutical Sciences, Capital Medical University. She received her Ph. D. degree in polymer physics and chemistry from Nanjing University, in 2020. Her current research interests are focused on designing and constructing functional nanomaterials, including fluorescence nanoprobes, biomimetic nanoassemblies, and piezoelectric nanogenerators, for bioimaging, cancer therapy, and neurostimulation.

---

1. Introduction

Vision is one of the most important human senses. The retina receives light signals, converts them into neural electrical signals, and transmits them to the cerebral cortex. Degenerative retinal diseases have caused millions of people to lose vision. Numerous treatments for degenerative retinal diseases like optogenetics, stem cell therapy, cell replacement, and the use of retinal prostheses have prospects for vision restoration in basic research or clinics.¹ Optogenetics and retinal prosthesis implantation are more effective for advanced retinal degenerative disease with photoreceptor cell loss. By introducing photoreceptor proteins (channelrhodopsin variants) to retinal cells, current optogenetic methods enable the light-sensitive character of retinal ganglion cells (RGCs) or degenerative bipolar and amacrine cell networks (photoreceptor cells).^2,3 They are beneficial for vision function restoration from a long-term perspective; however, potential irreversible side effects and complex translational paths remain significant hurdles for broad clinical adoption.⁴ Retinal prostheses are a family of implantable devices developed to mimic photoreceptor cells, catching light and producing electrical stimulation to surviving neural networks.⁵

Retinal prostheses based on multielectrode arrays (MEAs), such as Argus I,⁶ Argus II,⁷ IRIS II,⁸ and Alpha-IMS,⁹ as well as the subretinal photovoltaic implant (PRIMA),^10,11 are among the most established retinal prostheses evaluated in clinical trials. They capture images with camera-equipped glasses and convert them to electric signals, allowing the inner retina to perceive light signaling.¹² However, this direct electrical stimulation approach faces persistent challenges, including resolution limited by electrode size and density, long-term biocompatibility issues within the delicate retinal environment, and reliance on complex external hardware.^13,14 Consequently, achieving high-acuity resolution, ensuring long-term functional stability, and improving biocompatibility remain significant goals for next-generation visual prostheses. Addressing these limitations, recent studies have explored novel materials and designs, such as conformable liquid-metal based microelectrode arrays designed for improved tissue integration,¹⁵ bio-integrated solutions using living cells as biosensors,¹⁶ and bioinspired self-driven systems simulating retinomorphic functions.¹⁷ As highlighted in recent reviews, the development of novel designs and microfabrication techniques for photovoltaic implants, which directly convert light into localized neural stimulation, is at the forefront of addressing these fundamental challenges.¹⁸ Alongside these pioneering systems, newer-generation devices based on organic semiconductor polymers, exemplified by POLYRETINA,¹⁹ are demonstrating significant advancements in creating flexible, high-density interfaces.

The goal of a retinal prosthesis is to mimic the complex functions of the retina, involving light sensing and the transmission of electrical signals to neurons. Energy-conversion materials refer to substances that can transform one form of energy (such as light, heat, mechanical, or chemical energy) into another, providing new thinking and means for vision restoration by leveraging the unique advantages of these materials.²⁰ Photovoltaic, piezoelectric, upconversion, and photothermal materials represent typical platforms in this background,²¹ and have achieved notable success in light sensing and neuronal stimulation.²² These materials act as photodetectors and neural modulators, aiming to restore vision with minimal damage to the surrounding tissue. In the nano- and microscale regimes, the physical and chemical properties of materials are predominantly dictated by their morphology, size, and composition. These structural characteristics act as crucial determinants, intricately influencing how materials interact with their environment, respond to external stimuli, and exhibit unique behaviors.^23,24 Energy-conversion nano/micromaterials offer a fundamental shift, presenting benefits such as intrinsic flexibility for superior tissue conformity,²⁵ high spatiotemporal responsiveness, and the potential for creating high-resolution neural interfaces.²⁶ Critically, many of these approaches enable self-powered or wireless operation, mitigating the need for external power supplies and transcutaneous wiring.²⁷

As shown in Fig. 1, in this review, by reviewing the operating principle, composition design, and application of different novel retinal prostheses based on energy-conversion nano/micromaterials, we hope it is valuable for the development of more effective and precise retinal prostheses for vision restoration.

Fig. 1 Scheme of novel retinal prostheses based on energy-conversion nano/micromaterials.

2. Basic principles of visual imaging

2.1 Normal retinal physiology and signal processing

The retina is the most ingeniously constructed part of the eye and consists of neuronal cells and Müller glial cells for homeostasis maintenance.^28,29 Retinal neuronal cells include photoreceptors, bipolar cells, retinal ganglion cells, horizontal cells, and amacrine cells. Photoreceptors can be broadly categorized into rods and cones. The basic principle of vision is to receive light and convert it to electrical signals by photoreceptor cells to generate action potentials (APs) on nerve cell membranes. Cones are responsible for color vision, while rods are more sensitive to dim light. The visual pigment in rod cells is called rhodopsin, which consists of opsin and retinol. Under low light conditions, the breakdown of rhodopsin generates a G protein-coupled receptor. This receptor activates phosphodiesterase via guanylate binding protein, leading to the breakdown of cyclic guanosine monophosphate (cGMP, a second messenger). As a result, this process inhibits the cGMP-gated cation channel. The outermost layer of the retina is the retinal pigment epithelium (RPE) cells, which protect photoreceptor cells by extending pseudopodial protrusions to envelop photoreceptors, as well as absorbing optical signals misused by photoreceptors.

The hyperpolarized receptor potential in photoreceptor cells modulates action potentials in retinal ganglion cells (RGCs) via bipolar cells. At this crucial synaptic stage, the visual signal is fundamentally segregated into parallel ON and OFF pathways, a cornerstone of visual processing that allows for the separate encoding of light increments and decrements.³⁰ These pathways are established by two functional subclasses of bipolar cells: depolarized (ON) cells that signal increases in light, and hyperpolarized (OFF) cells that signal decreases, which together form the basis for contrast perception. These parallel streams then project to a diverse population of over 30 distinct RGC subtypes, each specialized for extracting specific visual features.³¹ This intricate visual signal processing-segregating signals into ON/OFF pathways via bipolar cells and relaying them to specialized RGC subtypes-is further sculpted by amacrine cells, which provide vital inhibitory modulation through neurotransmitters like γ-aminobutyric acid (GABA) and glycine at the postsynaptic membrane. Additionally, the dendrites of numerous RGCs exhibit “self-avoidance” traits, a design that minimizes overlap and reduces the potential for redundant inputs.

The transmission of visual information from the retina to the cerebral cortex is a highly complex and dynamic neurocomputational process, far exceeding a simple relay of light. Instead of acting as a passive camera, the retina actively processes the visual scene to encode it into a precise temporal sequence of action potentials.³² This encoding is handled by over 30 diverse RGC subtypes, each functioning as a specialized feature detector tiled across the visual field, extracting and communicating specific information such as object motion, contrast edges, and fine spatial details.³³ Crucially, the final output signal is not static, but shows a series of precisely timed electrical spikes whose frequency and temporal structure are constantly adjusted. This dynamic regulation, involving intricate feedback and feedforward circuits from amacrine and horizontal cells, enables the visual system to adapt to vast changes in ambient light levels (gain control) and highlight salient features in the environment.³⁴

2.2 Basic principles for the development of retinal prostheses

Each region of the retina plays a vital role, and conditions such as age-related macular degeneration (AMD) and retinitis pigmentosa (RP) are among the leading causes of blindness. In these diseases, the primary pathology is the loss of photoreceptor cells. However, a significant portion of the inner retinal neurons, such as bipolar and ganglion cells, often remain functionally viable. This provides a critical therapeutic window, forming the foundational principle for the development of retinal prostheses,³⁵ which aim to bypass damaged photoreceptors and directly stimulate these surviving neural circuits, a strategy successfully employed by established clinical systems.^9,14

However, the clinical path for retinal prostheses based on direct electrical stimulation is shadowed by fundamental physiological limitations from the complex biology of the diseased retina. A primary challenge is the non-selective activation of ON and OFF pathways, which merges the opponent signals essential for interpreting contrast.³⁶ This is compounded by the generation of non-physiological spiking patterns; instead of replicating the rich temporal codes of natural vision, electrical stimulation typically evokes highly synchronized, artificial bursts across large groups of neurons with unnatural temporal precision.³⁷ Critically, this artificial signal is delivered to a biological substrate that is itself a source of noise. The remodelled degenerated retina develops pathological hyperactivity and network oscillations, which create a noisy background that directly impairs stimulation efficiency.³⁸

The design and optimization of retinal prostheses need to closely adhere to the natural working principles of the retina, achieving biomimetic simulation in light sensing, signal modulation and ganglion cell output, to more efficiently replace damaged retinal function and help blind patients restore meaningful visual perception.

3. Mechanism of energy-conversion nano/micromaterials for neurostimulation and retinal prostheses

Energy-conversion nano/micromaterials have made significant progress in energy science,^39–41 photocatalysis,^42,43 and biomedical applications.^44,45 For example, in environmental detoxification, photocatalysts are capable of absorbing photons and generating reactive oxygen species (ROS) for pollutant degradation.⁴⁶ Recently, energy-conversion nano/micromaterials have received much attention in biomedical applications, especially in neurostimulation and therapy.⁴⁷ Based on the basic principles of visual imaging in the natural retina, light-active nanomaterials are optimum candidates for the core components of retinal prostheses. Numerous semiconductor materials are found suitable for photodetection, absorbing photons, and outputting electrical signals for neurostimulation. Among these, perovskite materials, particularly in the form of colloidal nanocrystals (PNCs), are rapidly emerging as a leading platform for next-generation retinal prostheses.⁴⁸ This device exhibits high photovoltaic performance with a maximum external quantum efficiency of 90%. PNC-based devices not only demonstrate high photoresponsivity and ultrafast response times essential for capturing visual information but also leverage their intrinsic ionic migration properties to enable reconfigurable photoresponsivity. This remarkable feature allows for built-in, low-energy image preprocessing functions, such as contrast enhancement and adaptive imaging, directly at the hardware level, mimicking the computational capabilities of biological retinal networks. In this part, the mechanism of energy conversion in photovoltaic, piezoelectric, upconversion and photothermal materials for neurostimulation and retinal prostheses has been discussed.

3.1 Photovoltaic effect

The photovoltaic effect, a phenomenon characterized by voltage generation upon illumination of a semiconductor material,^49,50 primarily operates through photogeneration and separation of charge carriers. In retinal prostheses, this effect may interact with supporting mechanisms such as photothermal conversion or capacitive coupling to enhance device functionality.⁵¹ Semiconductor-based nano/micromaterials have emerged as the predominant choice for photoelectric neural interfaces due to their exceptional capabilities in both photonic energy harvesting and precise retinal ganglion cell (RGC) activation.^5,52 As illustrated in Fig. 2a and b, the neurostimulation process involves three critical phases: photon absorption inducing electron–hole pair generation, carrier migration to material interfaces under built-in electric fields, and extracellular charge accumulation at electrode–neuron junctions, thereby enabling wireless neural activation without external power sources.^53,54 Current advancements in photovoltaic retinal prostheses leverage structural engineering to optimize key performance parameters: (1) bandgap modulation through quantum confinement effects enables spectral tuning to match retinal photoreceptor absorption profiles;³⁷ (2) hierarchical nanostructuring increases light-trapping capacity while maintaining cellular-scale spatial resolution;⁵⁵ and (3) surface plasmon resonance enhancement in noble metal-decorated semiconductors boosts photon capture efficiency.⁵⁶ Notably, subretinal implants based on inorganic silicon photodiode arrays, such as the PRIMA device, have demonstrated clinical viability in human trials, achieving visual acuity restoration of 20/460 to 20/550 in patients with atrophic age-related macular degeneration.⁵⁷ As a subretinal implant, PRIMA is strategically placed to more directly stimulate the inner nuclear layer, representing a significant engineering contribution towards potentially achieving more selective activation of ON and OFF pathways compared to epiretinal devices. Separately, prostheses based on organic photovoltaic polymers are showing significant promise in preclinical models, though they have not yet reached the same stage of clinical validation.

Fig. 2 The mechanism of energy conversion nano/micromaterials for neurostimulation and retinal prostheses. (a and b) Photovoltaic effect.⁵³ Figures (a and b) have been reproduced from ref. 53 with permission from Ridvan Balamur et al., CC BY; copyright: 2024. (c and d) Piezoelectric effect.⁵⁸ Figures (c and d) have been reproduced from ref. 58 with permission from Wiley-VCH GmbH; copyright: 2023. (e) Photochemical upconversion effect (PUC).⁵⁹ This figure has been reproduced from ref. 59 with permission from American Chemical Society; copyright: 2018. (f) Photothermal effect.⁵⁶ This figure has been reproduced from ref. 56 with permission from American Chemical Society; copyright: 2025.

3.2 Piezoelectric effect

The piezoelectric effect is mainly attributed to the asymmetry of the crystal structure or molecular chain of the piezoelectric materials, which leads to the mismatch of positive and negative charge centers in the crystal. When the piezoelectric material is subjected to external stress, the distance between the positive and negative charge centers changes, and the dipole moment of the crystal structure changes along the direction of stress, thus showing the change of surface free charge on a macroscopic level.⁶⁰ Piezoelectric materials have the unique capability to convert mechanical energy into electrical energy and are widely used in neural stimulation.^61,62

In particular, polymer piezoelectric materials, with their excellent flexibility and biocompatibility, have the potential to respond to light when combined with other materials.^63–65 As shown in Fig. 2c, the device consisting of azobenzene-containing liquid crystal polymers and ferroelectric polymers can convert light energy into electrical signals through the light-induced stress generated in the linear liquid crystal polymer (LLCP) layer under light irradiation. As illustrated in Fig. 2d, through the change in dipole density in the poly(vinylidene fluoride-co-trifluoroethylene) (P(VDF-TrFE)) layer, the detector can not only convert light energy but also mimic the retina's sensing of light, directly transmitting electrical signals to neural cells. Ultrasonic retinal prosthesis is a type of novel retinal prosthesis based on the piezoelectric effect. As a kind of mechanical wave, ultrasound is widely used in biomedicine, with high energy, high penetration, and high precision, and can penetrate human tissues without significant attenuation to tissues.⁶⁶ Ultrasonic waves induce the deformation of piezoelectric materials, generating electrical signals. These signals subsequently open the voltage-gated channels of cellular calcium ions, triggering a cascade of intracellular events.⁶² Therefore, the ultrasonic piezoelectric effect can be used as an effective means of retinal neuron stimulation.

3.3 Photochemical upconversion effect (PUC)

Upconversion is a photo-physical process in which high-energy photons are generated from a material excited by low-energy photons. It mainly includes two mechanisms: multi-photon absorption and energy transfer. Multi-photon absorption refers to the process where a material absorbs multiple low-energy photons, causing electrons to transition to higher energy levels and subsequently emit high-energy photons. Energy transfer upconversion involves a sensitizer absorbing low-energy photons and transferring the energy to an activator, which then emits high-energy photons after gaining sufficient energy.⁶⁷ These upconversion materials, typically composed of a host matrix (e.g., NaYF₄) co-doped with rare-earth lanthanide ions, exhibit unique luminescence properties including high tissue penetration depth, low autofluorescence background, and high resistance to photobleaching, making them a compelling alternative to traditional fluorophores for biological applications.⁶⁸ The main challenge for their clinical translation, however, lies in ensuring long-term biocompatibility and mitigating potential cytotoxicity, which is often associated with the release of fluoride and lanthanide ions. Consequently, surface modification strategies, such as coating with a silica shell or functionalizing with polymers like polyethylene glycol (PEG), are critical for improving their colloidal stability in physiological environments and reducing adverse biological effects.⁶⁹ Building on these principles, various research efforts have explored upconversion materials for neural stimulation and visual activation. As shown in Fig. 2e, upconversion nanoparticles (UCNPs) can absorb low-energy near-infrared light (NIR) and emit high-energy visible light. This makes them ideal light-converting materials for optogenetics because NIR light has better tissue penetration capabilities, allowing for the regulation of neuronal activity from outside the body. Optogenetics involves introducing light-sensitive protein genes into neurons for expression using viral vectors and other methods. These neurons are then activated or inhibited by specific wavelengths of light. The role of upconversion materials is to convert long-wavelength, highly penetrating NIR light into short-wavelength light, controlling the emission wavelength within the excitation range of the light-sensitive proteins on the target neuronal membranes, thereby achieving wireless control of neuronal activity.⁵⁹

3.4 Photothermal effect

The photothermal effect refers to the phenomenon in which materials absorb light energy and convert it into heat energy.⁷⁰ It has important applications in many fields, especially in artificial retinal prostheses and neural signal regulation. The photothermal effect refers to the process by which a material converts light energy into heat. This process involves the interaction between photons and the atoms or molecules within the material. When photons are absorbed, they excite the electrons in the material, causing them to transition to higher energy levels. As these excited electrons return to their ground states, they release energy in the form of heat, thereby increasing the temperature of the material. This is the basic principle of the photothermal effect.⁵⁶ Commonly used photothermal materials include gold nanorods (AuNRs),⁷¹ liquid metal-based materials,⁷² indocyanine green,⁷³etc. They exhibit unique properties in photothermal neural regulation, providing new directions for the treatment of related diseases. As shown in Fig. 2f, by using near-infrared laser irradiation, AuNRs absorb light energy and convert it into heat energy, causing rapid changes in surrounding temperature and activating retinal neurons. Specifically, when a near-infrared laser is irradiated onto AuNRs bound to retinal neurons, the surface plasmon resonance of AuNRs generates local thermal effects. This thermal effect activates temperature-sensitive ion channels (such as Transient Receptor Potential Vanilloid 1 (TRPV1) channels), causing changes in the membrane potential of neurons and generating action potentials, thereby achieving the transmission of neural signals.

4. Application of energy-conversion nano/micromaterials for retinal prostheses

4.1 Photovoltaic materials

4.1.1 2D semiconductor materials. 2D semiconductor materials such as graphene and transition metal dichalcogenides (TMDs) show excellent optoelectronic features, flexibility, and biocompatibility,^74,75 which mark a new era for optoelectronic devices.^76–78 Notably, their naturally thin structure enables efficient light absorption and interaction.^79,80 Studies have demonstrated near-unity absorption in monolayer TMD and universal absorptance quantification.^81,82 For example, single-layer TMDs possess a direct bandgap, which is essential for converting photons into electrical signals with minimal losses.⁸³ Importantly, recent research shows that structural engineering, such as rotational misalignment, can induce direct bandgap characteristics even in few-layer TMDs like bilayers, enhancing their optoelectronic utility.⁸⁴ This has led to significant improvements in photocurrent generation for retinal applications.⁸⁵ The atomic-scale thickness of 2D semiconductor materials provides flexibility and biocompatibility, making them suitable for implantable devices,⁸⁶ highlighting remarkable mechanical resilience, with materials withstanding millions of bending cycles, and enhanced biological integration, including improved neuron adhesion.⁸⁷ The flexibility helps conform to the curvature of the eye, reducing stress and strain.⁸⁸ This conformity is supported by findings that these materials can match the Young's modulus of retinal tissue, significantly reducing mechanical mismatch.

The CurvIS array is an optimized 2D heterostructure device consisting of MoS₂ and graphene.⁸⁹ MoS₂ absorbs photons, producing electron–hole pairs and initiating a photocurrent, which is then amplified and processed into electrical pulses designed to mimic healthy photoreceptor signals. These pulses stimulate the optic nerves via a microelectrode array. The design combines MoS₂'s superior photo-absorption and fracture strain alongside graphene's flexibility and strength. The ultrathin 51 nm structure allows it to conform to hemispherical surfaces like the human eye's curvature without mechanical fractures. Fig. 3a illustrates the device design and material layering. A key feature is the CurvIS array's ability to avoid infrared interference, processing only visible light. This simplifies the design, enhances signal clarity, and improves energy efficiency, making it ideal for compact, implantable optoelectronic applications. Fig. 3b illustrates the imaging capabilities, showing clear capture of objects (like Sigma) and demonstrating IR blindness, distinguishing it from conventional silicon photodetectors.

Fig. 3 2D material-based optoelectronic devices for retinal prostheses. (a) The CurvIS array, a MoS₂/graphene nanostructure, exemplifies the integration of flexibility and high sensitivity in optoelectronic devices, designed to fit the eye's curvature for an improved interface with the neural tissue.⁸⁹ (b) CurvIS array's dual capability: left, a high-resolution capture of the sigma symbol demonstrates precise visual reproduction, right: confirms IR blindness, ensuring image clarity amidst fluctuating infrared light.⁸⁹ Figures (a and b) have been reproduced from ref. 89 with permission from Changsoon Choi et al., CC BY; copyright: 2017. (c) Showcasing the range of wavelengths that can be absorbed by transition metal dichalcogenides, this graph highlights the potential for precise photon-to-electron conversion in artificial vision, vital for prosthetics that can adapt to different lighting conditions.⁹⁰ This figure has been reproduced from ref. 90 with permission from Na Li et al., CC-BY-NC; copyright: 2023. (d) A schematic comparison between the human visual processing pathway and the 2D MIR optoelectronic retina, emphasizing the latter's capability to emulate natural perception and information encoding through innovative materials science.⁹¹ (e) Depicting the rapid and sensitive response of the b-AsP/MoTe₂ heterostructure to both MIR and NIR stimuli, this graph demonstrates the material's potential for fast, bias-free signal processing in neuromorphic computing.⁹¹ Figures (d and e) have been reproduced from ref. 91 with permission from Fakun Wang et al., CC BY; copyright: 2023.

Semiconductors have tunable bandgaps that can be adjusted through external electric fields or layer stacking, as described in Fig. 3c.⁹⁰ A 2D optoelectronic retina is composed of black phosphorus (b-AsP) and MoTe₂ achieved with the use of a van der Waals (vdWs) heterostructure. The b-AsP is used as the mid-infrared (MIR) photosensitive layer owing to its narrow bandgap and MoTe₂ with an appropriate bandgap of ∼1.0 eV serves as the NIR sensitizer. The b-AsP/MoTe₂ vdW heterostructure is integral to the simultaneous processing of MIR signals, representing a significant advancement by overcoming limitations, like bulkiness and limited efficiency, found in traditional infrared machine vision systems.⁹¹ From the conceptual framework depicted in Fig. 3d, the photovoltaic and photothermoelectric effects are identified as the main mechanisms for perceiving NIR and MIR illumination, respectively. The device's key innovation is its capacity to detect MIR illumination intensity, encode it into a spike train for neuromorphic computing, and enable a spiking neural network (SNN) to successfully conduct digit classification tasks with over 96% accuracy. This accomplishment highlights the device's potential in advanced neuromorphic computing applications. The device shows outstanding sensitivity, with a high MIR detectivity of approximately 9.6 × 10⁸ Jones (cm Hz^0.5 W⁻¹) and a fast NIR response rate of about 600 nanoseconds. Additionally, the device works effectively without requiring an applied electrical bias. The NIR and MIR photoresponses at varying illumination intensities are presented in Fig. 3e. The device exhibits bias-free, rapid infrared response through an all-optical excitation mechanism with stochastic NIR sampling. The study utilizes the distinct attributes of 2D vdW heterostructures and combines perception and encoding capabilities, thereby enabling advancements in IR machine vision systems. Further developments in optical neuromorphic functionalities are predicted, which may have potential usage in high-speed, low-latency, and energy-efficient systems.

However, several challenges persist, including the necessity for further research to enhance the long-term stability and biocompatibility of the devices and to mitigate potential risks such as inflammation or rejection. Additionally, there is a need to integrate ultrathin materials into deposited tissues while ensuring that their properties remain unaltered during mechanical deformations.¹⁵ Beyond device integration, significant scalability hurdles persist in the manufacturing of 2D heterostructure materials for widespread clinical and commercial use. The key bottlenecks include the achievement of uniform, wafer-scale thickness and quality, in addition to the management of defect densities that exhibit an increase with increasing production scale.⁹² Although chemical vapour deposition (CVD) is a leading synthesis method, maintaining temperature and precursor uniformity for large-area industrial processing remains a major challenge.⁹³ Furthermore, the process of transferring delicate 2D material layers from a growth substrate to a final device substrate is a critical step that frequently introduces contamination and defects, thereby hindering yield and reproducibility at industrial scales.⁴ Recent advancements in alternative methods, such as adhesive wafer bonding and atmospheric pressure CVD using nanoparticle precursors, show promise in overcoming these limitations. However, achieving cost-effective, high-volume production with consistent quality control is essential for future translation.⁹⁴ Subsequent research will continue to concentrate on the optimisation of the structure and design parameters of 2D semiconductors.¹³

4.1.2 Organic semiconductor polymers. Organic semiconductor polymers, with their excellent biocompatibility, flexibility, and efficient photoelectric conversion properties, provide a low-trauma, high-resolution innovative solution for biomimetic visual repair.⁹⁵ Recent advancements in organic semiconductor polymers known as POLYRETINA, a high-density epiretinal device, set a new visual acuity standard.^19,96,97

Fig. 4a shows the setup and interface. POLYRETINA is stretchable, insertable via a small incision, conforms to the eye, has 2215 pixels for capacitive charge injection, offers visual acuity around 20/600, and has a broader FOV (field-of-view).⁹⁸In vivo tests in minipigs showed reliability and safety (Fig. 4b).⁹⁹ Yu et al. designed a bionic artificial retina with a photoelectric device based on a PCBM (phenyl-C61-butyric acid methyl) ester mixture film for RGB (red–green–blue) detection. A pyramid-shaped microarray enhances nerve contact for local activation and improves color recognition accuracy.¹⁰⁰ Wang and colleagues developed a monopolar chip to address crosstalk degradation.¹⁰¹ Using computational steering and distributed pulsing, it achieved visual acuity correlating with pixel pitch and significantly reduced crosstalk. To counter the photo-responsivity decrease in smaller pixels, T. W. Huang et al. created a wireless implant with vertical p–n junctions in a 3D structure, eliminating oxidative stress and maintaining a consistent threshold for smaller pixels.¹⁰² The Okayama University-type retinal prosthesis (OUReP) film, made of a dye-coupled polyethylene film (Fig. 4c), is inserted subretinally and shows significant visual function improvement in primates.^103–105 Other studies confirm that OUReP produces electrical power from light and stimulates neurons.¹⁰⁶ Maya-Vetencourt et al. created a fully organic prosthesis (P3HT, poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS), and silk fibroin) tested on rats with retinitis pigmentosa (Fig. 4d), showing sustained visual acuity enhancement up to 180 days.¹⁰⁷ Recent research also highlights promising approaches using metallic nanoparticles integrated with organic photovoltaic materials, as shown in Fig. 4e. Rahmani and Eom demonstrated enhanced organic photovoltaic-based retinal prostheses by incorporating plasmonic silver nanoparticles (AgNPs) into a cathode-modified structure.¹⁰⁸ Embedding a monolayer array of spherical AgNPs in the cathode electrode achieved plasmonic enhancement, significantly increasing device efficiency (10% current density boost and doubled efficiency at low light), particularly at low intensity (0.26 mW mm⁻²). Ongoing innovation in organic semiconductor polymers holds significant promise for overcoming limitations and developing next-generation retinal prostheses with enhanced stability, higher resolution, and natural visual perception, bringing hope for more effective vision restoration. A study by Airaghi Leccardi et al. showcased an advancement in the use of organic polymer materials, specifically P3HT, to create a photovoltaic interface for NIR neural stimulation.¹⁰⁹ This involved optimizing the interface's nanostructure, properties, and robust response at NIR wavelengths. Comparing bulk heterojunction (BHJ) compositions, P3HT: [6,6] phenyl-C61-butyric acid methyl ester (PC₆₀BM) performed well at 565 nm but less efficiently at NIR. Conversely, poly[2,6-(4,4-bis-(2-ethylhexyl)-4H-cyclopenta[2,1-b;3,4-b′]dithiophene)-alt-4,7(2,1,3-benzothiadiazole)] blended with PC60BM (PCPDTBT:PC₆₀BM) displayed an increased photoelectric response at 730 nm, proving more efficient for NIR stimulation and becoming the preferred choice for NIR-responsive retinal prostheses.

Fig. 4 Integrated organic semiconductor polymer-based materials in retinal prosthesis technology. (a) The bioelectronic setup integrating organic photovoltaics for neural stimulation.¹⁹ This figure has been reproduced from ref. 19 with permission from Laura Ferlauto et al., CC BY; copyright: 2018. (b) The restoration of neural activity in minipigs through POLYRETINA implantation.⁹⁹ This figure has been reproduced from ref. 99 with permission from Paola Vagni et al., CC BY; copyright: 2022. (c) Introducing the dye-coupled film as a pivotal component for the Okayama University-type retinal prosthesis.¹⁰⁵ This figure has been reproduced from ref. 105 with permission from IOP Publishing; copyright: 2021. (d) A multi-layered organic prosthetic device, detailing its structural composition and placement within the retina.¹⁰⁷ This figure has been reproduced from ref. 107 with permission from Springer Nature; copyright: 2017. (e) Scheme of implanted AgNP enhanced organic photovoltaic-based retinal prosthesis in the retina space.¹⁰⁸ This figure has been reproduced from ref. 108 with permission from Ali Rahmani et al., CC-BY-NC; copyright: 2024.

4.1.3 Semiconductor polymer nanoparticles. Recent research has indicated the significant potential of semiconductor polymer nanoparticles, notably P3HT nanoparticles, in the field of retinal prostheses. These nanoparticles can be internalized by biological entities and, upon exposure to light, generate charges that trigger neuronal activity.¹¹⁰ In 2013, Ghezzi et al. conducted one of the earliest studies demonstrating the potential of P3HT in retinal prostheses. The research examined the impact of a single-component organic film made of P3HT in inducing neuronal firing upon illumination. The experiment showed that this bio-organic interface reinstated light sensitivity in explants of rat retinas affected by light-induced photoreceptor degeneration. These findings indicate that all-organic devices could have a significant role to play in sub-retinal prosthetic implants in the future. Building on this foundation, the investigation of P3HT nanoparticles revealed a new approach to enhancing visual function by stimulating retinal ganglion cells directly, deviating from the sole focus on improving photoreceptor survival. This shift to nanotechnology emphasized P3HT's exceptional ability for light-induced neural stimulation, presenting a hopeful avenue for retinal regeneration. In 2020, Maya-Vetencourt et al. found that subretinal injection of P3HT polymer nanoparticles can significantly increase visually evoked potential amplitude and rescue visual function in rats. Further research indicates that the visual rescuing effect of P3HT is achieved via direct stimulation of retinal ganglion cells instead of enhanced photoreceptor survival.¹¹¹

Maya-Vetencourt et al. and S. Francia et al. explored a different approach using fully organic P3HT nanoparticles (NPs) to design a retinal prosthesis.^111,112 These P3HT NPs were synthesized via a method designed to preserve their structural properties (Fig. 5a). Upon subretinal injection, the NPs were found to distribute and directly contact second-order retinal neurons (Fig. 5b). The proposed mechanism involves light-activated P3HT NPs generating an electric potential (Fig. 5c), capable of capacitively depolarizing neuronal membranes to activate them and potentially restore physiological signals. Importantly, histological studies supported the NPs’ safety and biocompatibility in aged rats, with no significant tissue damage or inflammation observed (Fig. 5d). The efficacy was demonstrated through both electrophysiological (visually evoked potentials, Fig. 5e) and behavioral analyses (Fig. 5f and g), showing restored visual responses and improved light sensitivity in treated rats with degenerative retinas. This work highlights the potential of organic semiconductor NPs to directly interface with and restore function to damaged retinas.

Fig. 5 Nanotechnological advancements in retinal prosthetics using P3HT nanoparticles. (a) Illustrating P3HT nanoparticles, crucial for their function in light-induced neural stimulation.¹¹¹ (b) The interaction of P3HT NPs with retinal neurons, essential for activating visual pathways. (c) Visualizing the electrostatic potential mapping, detailing the charge distribution essential for P3HT NP function.¹¹² (d) Transversal retinal sections post-NP injection, confirming the biocompatibility and retention of NPs.¹¹² (e) Detailing the setup for recording visually evoked potentials, indicating the functional recovery of visual responses.¹¹² (f) Comparing pupillary light reflex measures, demonstrating the therapeutic potential of P3HT NPs.¹¹² (g) Quantification of visual function recovery across rat models, validating the efficacy of P3HT nanoparticle treatment.¹¹² This figure has been reproduced from ref. 112 with permission from S. Francia et al., CC BY; copyright: 2022.

Moreover, polymer nanoparticles fabricated from semiconducting and conducting materials such as P3HT and PEDOT:PSS exhibit a unique microstructure. As demonstrated in this work by Tullii G. et al., their architecture, where conducting islets are dispersed within the semiconducting matrix, leads to a significant enhancement in charge dissociation and electron transfer efficiency.¹¹³ As a result, the generation of the photocurrent is boosted by approximately one order of magnitude, optimizing the overall performance of these materials in optoelectronic applications.

Despite these advancements in organic photovoltaic materials, limitations persist. These include the need to improve visual acuity and field of view, address long-term biocompatibility, and refine technology for broader applications. However, ongoing innovations in organic photovoltaics offer promising solutions. Future research should prioritize optimizing the prosthetic–neural tissue interface, enhancing device resolution and stability, and expanding functionality across varying lighting conditions to provide more effective solutions for retinal degenerative conditions.

4.1.4 Inorganic semiconductor nano/micromaterials. The working mechanism of inorganic semiconductor materials in artificial retina is mainly based on their light absorption and photoelectric conversion characteristics. For example, titanium dioxide (TiO₂) nanomaterials undergo valence band electron transitions to the conduction band under excitation by ultraviolet and partially visible light, forming photo-generated charge carriers. These charge carriers separate and migrate under the action of the electric field inside the material, thereby achieving photoelectric conversion.¹¹⁴ The generated electrical signals need to communicate effectively with the remaining neurons in the retina. TiO₂ nanomaterials have advantages such as chemical stability, low cost, and good biocompatibility. The gold nanoparticle coated titanium dioxide nanowire array developed by Zheng's team has a diameter of approximately 100 nm and a length of 2 μm, similar in size to photoreceptors (Fig. 6a and b).¹¹⁵ This array is capable of converting ultraviolet, blue, and green light into a photocurrent, activating retinal ganglion cells in visually impaired mice with retinal degeneration, with a light intensity threshold as low as 10 μW mm⁻². Doping and surface modification of TiO₂ nanomaterials can further enhance their light absorption and photoelectric conversion properties. Studies have demonstrated that gold nanoparticle-decorated titania nanowire arrays, when subretinally implanted in blind mice, can restore visual responses in retinal ganglion cells and improve behavioral vision. Further work with improved nanowire arrays has confirmed their efficacy in both blind mice and monkeys, showing enhanced performance in behavioral tests and long-term stability after subretinal implantation (Fig. 6c and d).¹¹⁶

Fig. 6 Applications of inorganic semiconductor nano/micromaterials in artificial retinal prostheses. (a) Schematics comparing the structures of a healthy retina and a degenerative retina where the photoreceptor layer is replaced by Au-TiO₂ (gold-titanium dioxide) nanowire arrays acting as artificial photoreceptors.¹¹⁵ (b) Side-view scanning electron microscopy (SEM) image showing the organized structure of the Au-TiO₂ nanowire arrays. Scale bars: 2 μm.¹¹⁵ Figures (a and b) have been reproduced from ref. 115 with permission from Jing Tang et al., CC BY; copyright: 2018. (c) Visual testing setup and task, specifically the Visual Guided Saccades (VGS) behavioral task, used to evaluate vision restoration in monkeys implanted with nanowire arrays.¹¹⁶ (d) Fundus photographs illustrating the stable subretinal implantation of Au-TiO₂ nanowire arrays within the eyes of monkeys.¹¹⁶ Figures (c and d) have been reproduced from ref. 116 with permission from Ruyi Yang et al., CC BY; copyright: 2023. (e) Schematic diagram depicting the microinjection process and intended integration of ZnIn₂S₄/NGQD microflowers into the subretinal space of a degenerative retina, emphasizing size matching with natural photoreceptors.¹¹⁷ (f) Scanning electron microscopy (SEM) image showing the distinct hierarchical microflower morphology of the as-prepared ZnIn₂S₄/NGQD particles.¹¹⁷ (g) DAPI (4′,6-diamidino-2-phenylindole) Quantum dots -stained vertical cross-sections of different mouse retinas (wild-type, degenerative Rd10, and Rd10 implanted with ZnIn₂S₄/NGQD microflowers or glass microstructures), illustrating retinal layer structure and nanoparticle localization.¹¹⁷ (h) Representative electrophysiological recordings showing light-evoked firing responses of retinal ganglion cells (RGCs) from control and nanoparticle-implanted degenerative mouse retinas, indicating functional recovery.¹¹⁷ The figure (d–h) have been reproduced from ref. 117 with permission from Mo Yang et al., CC BY; copyright: 2024.

Perovskite nanomaterials have excellent optoelectronic properties, such as a high light absorption coefficient, adjustable bandgaps, and long carrier diffusion length. A hemispherical all-inorganic CsPbI₃ nanowire array was developed to construct an artificial retina, which can generate current without external bias and achieve a self-working mode. Through carefully designed hybrid nanostructures, it also endows the retina with filter-free color imaging capabilities.¹¹⁸ Quantum dots (QD) are zero-dimensional materials with quantum confinement effects, and their optoelectronic properties can be precisely controlled by changing their size, composition, and surface ligands. ZnIn₂S₄/nitrogen-doped graphene quantum dot (NGQD) micro-flowers (MF) with 0D/3D heterostructures were constructed by the hydrothermal method, simulating the size of natural photoreceptor cells (2–5 μm) and a flexible nano-petal structure, providing a high specific surface area and photoactivation ability. The light absorption range (extended to 800 nm) and photocurrent conversion efficiency (3.75 times higher than that of pure ZnIn₂S₄) were optimized through quantum dot coupling (Fig. 6e and f). In the P66 Rd10 retinal degeneration mouse model, the photoresponsive recovery effect of retinal ganglion cells (RGCs) was evaluated through multi-electrode array (MEA) and patch clamp recording (Fig. 6g and h).¹¹⁷ Notably, recent exploration extends to novel metal-free inorganic semiconductors, such as hollow sphere graphitic carbon nitride nanoparticles (hg-C₃N₄), which have been shown to restore light sensitivity in preclinical models of blindness by acting as a leadless, injectable opto-nanobiointerface.¹¹⁹ This approach represents a promising direction for enhancing long-term biocompatibility by avoiding the potential toxicity associated with heavy metal-containing quantum dots or perovskites.

The utilisation of inorganic semiconductor nano/micromaterials in the field of artificial retina has yielded promising results; however, numerous challenges persist. With regard to material stability, it has been demonstrated that certain materials are susceptible to degradation and corrosion in physiological environments, which can have a detrimental effect on long-term performance. A case in point is perovskite nano/micromaterials, which are sensitive to humidity and temperature. Furthermore, there is a necessity to enhance biocompatibility, given that long-term implantation has the potential to induce inflammation and immune responses. Specifically, the introduction of inorganic nanoparticles has been demonstrated to activate retinal microglia, the primary immune cells of the eye. These cells then release a cascade of pro-inflammatory cytokines, such as TNF-α and IL-1β, which have the potential to perpetuate chronic neuroinflammation and compromise the prosthesis's function.¹²⁰ For instance, recent studies utilising human retinal organoids have demonstrated that lead-based perovskite nanoparticles can induce developmental neurotoxicity by triggering endoplasmic reticulum stress and inflammatory pathways.¹²¹ Consequently, the comprehension and regulation of the nano/micro-biointerface are imperative. Surface modification strategies, such as coating nanoparticles with biocompatible layers like silica or integrating them into less reactive composite structures, have shown promise in reducing these inflammatory responses by limiting direct cellular interaction and subsequent immune activation.¹¹⁷ In the future, there is a necessity to develop new inorganic semiconductor nanomaterials and explore material systems with superior performance and biocompatibility. In terms of clinical application, the following aspects should be noted: the expansion of the scale of clinical trials; the conducting of in-depth evaluations of safety and effectiveness; and the continuous improvement of materials and implantation technologies. These strategies will promote the widespread application of inorganic semiconductor nanomaterials in the field of artificial retina and bring more hope to visually impaired patients.

Beyond the active semiconductor layers, the choice and architecture of the electrode material are equally critical for the overall performance and longevity of a retinal prosthesis. The electrodes are the essential interface for delivering stimulation to neurons. In this regard, advanced materials like 3D pyrolytic carbon are emerging as a superior alternative to traditional planar electrodes. A clear research progression has validated their potential: initial studies established the excellent electrochemical feasibility of 3D pyrolytic carbon, demonstrating a high charge storage capacity suitable for photovoltaic prostheses.¹²² Subsequent work addressed key engineering challenges by developing novel selective passivation and laser ablation techniques, which effectively reduce electrical crosstalk and enable the creation of high-fidelity, spatially precise stimulation sites.¹²³ Most recently, these advancements culminated in the direct demonstration of effective, high-amplitude stimulation of retinal neurons ex vivo, confirming that 3D pyrolytic carbon electrodes outperform their 2D counterparts and represent a robust platform for developing next-generation neural interfaces.¹²⁴

4.2 Piezoelectric materials

Piezoelectric materials have the unique capability to convert mechanical energy into electrical energy and are widely used in neural stimulation. Chen et al. fabricated a high-resolution photoelectric detector nanoarray from a piezoelectric-azobenzene polymer mixture using nanoimprint lithography (Fig. 7a and b). In this system, the azobenzene molecules undergo trans–cis isomerization when exposed to light, generating light-induced stress, which is then converted into electrical signals by the piezoelectric polymer. The photoisomerization process of azobenzene dyes is rapid, taking only nanoseconds, meeting the requirements for transient responses to light signals during visual formation. Each nanodot can identify the incident light direction. Thousands of these integrated units, with sensing sizes smaller than those of photoreceptors, achieve unprecedented image resolution (25 000 PPI (pixels per inch) vs. human retina's 5000 PPI), high sensitivity, color recognition, instantaneous response (<50 ms), and 3D vision detection.¹²⁵ Yu's team utilized this principle to develop a smart photoelectric detector, consisting of a P(VDF-TrFE) layer and a linear liquid crystal polymer (LLCP) layer. As shown in Fig. 7c and d, the GCamp6-GFP (genetically encoded calcium indicator 6-green fluorescent protein) N2a cells incubated on a P(VDF-TrFE)-LLCP film show significant fluorescence changes. Fig. 7e schematically depicts the interface between the artificial photoreceptor (based on piezoelectric nanomaterials) and degenerated mouse retina, replacing damaged natural photoreceptors. Electrophysiological responses of retinal ganglion cells are recorded in the figure on the right. Red arrows mark stimulated spikes induced by the artificial photoreceptor, while green arrows indicate spontaneous activity. These results demonstrate the artificial photoreceptor's ability to guide neuronal signals even in a degenerated retina, simulating natural signal transduction. The combined action of piezoelectric materials and azobenzene molecules enables efficient light-to-electrical signal conversion readable by neural cells. This approach is promising for artificial retinas, requiring no external light reception or power source, making it highly convenient and efficient.⁵⁸

Fig. 7 Advanced photoresponsive retinal device components and responses. (a) Schematic illustration of the layered structure of the high-resolution photoelectric detector array based on polymer blends, designed for directional light sensing.¹²⁵ (b) 3D topographic image of the nanodot array surface, showing structures with sizes comparable to those of natural photoreceptors, highlighting the design for efficient light interaction.¹²⁵ Figures (a and b) have been reproduced from ref. 125 with permission from WILEY–VCH Verlag GmbH & Co. KGaA, Weinheim; copyright: 2016. (c) Cellular responses (measured from the intracellular calcium change, ΔF/F₀) of cells cultured on different polymer components (LLCP and P(VDF-TrFE)) under illumination.⁵⁸ (d) Fluorescence images demonstrating light-induced activation of cultured neurons on the artificial photoreceptor surface before and after illumination.⁵⁸ (e). (Left) Interface schematic of the artificial photoreceptor with a degenerated retina. (Right) Electrophysiological recording of retinal ganglion cells showing stimulated spikes induced by the artificial photoreceptor.⁵⁸ This figure has been reproduced from ref. 58 with permission from Wiley–VCH GmbH; copyright: 2023.

An ultrasonic retinal prosthesis, which can convert the transmitted ultrasound waves into electrical energy, is used to electrically stimulate retinal neurons through integrated electrodes. Correspondingly, induced action potentials will be transmitted through the optic nerve to the central visual pathway to generate visual perception. This type of ultrasonic retinal prosthesis is also facilitated by piezoelectric materials. Fig. 8a illustrates a potential configuration for an ultrasonic retinal prosthesis system, showing glasses with integrated components for ultrasound delivery to the eye and signal relay to the visual cortex.¹²⁶ Prior research by Badadhe et al. demonstrated safe modulation of retinal neural activity across 0.5–43 MHz, providing a foundation for ultrasound vision restoration.¹²⁷ Lo et al. focused on precise stimulation using high-frequency ultrasound to target specific retinal areas, crucial for activating functional neurons in degeneration.¹²⁸ Advancing these principles, Jiang et al. innovated a wireless ultrasound-induced retinal stimulation array. Fig. 8b provides a schematic view of their system, detailing the stacked structure including the US transducer, a flexible circuit board with electrodes, and piezoelectric pixels, along with examples of generated voltage outputs. Unlike wired prostheses (e.g., Argus II), this wireless approach uses ultrasound for energy/data transmission, enhancing coupling and reducing safety hazards. The array utilizes high-efficiency Pb(Mg_1/3Nb_2/3)O₃-PbTiO₃ (PMN-PT) single crystals. Fig. 8c shows the in vivo experimental setup for retinal stimulation in a mouse model using this ultrasonic array, depicting the device positioned on the eye, the ultrasound transmitter, and components for monitoring. In vivo tests using this setup demonstrated neural activity corresponding to patterns, highlighting artificial vision potential.¹²⁶Fig. 8d provides an example of a spatially defined pattern of neural activation generated by ultrasonic stimulation, visualized as a heatmap indicating intensity distribution across a spatial area.

Fig. 8 Ultrasound stimulation for retinal prostheses. (a) Schematic diagram of a proposed ultrasonic retinal prosthesis system including external glasses and an implantable flexible device.¹²⁶ (b) Schematic view detailing the layered structure of a wireless ultrasound-induced retinal stimulation array device.¹²⁶ (c) Experimental setup for ex vivo retinal stimulation using an ultrasonic array in a mouse retina model.¹²⁶ (d) Spatially defined pattern of neural activation generated by ultrasonic stimulation, visualized as a heatmap.¹²⁶ Figures (a–d) have been reproduced from ref. 126 with permission from Laiming Jiang et al., CC BY; copyright: 2022. (e) Schematic diagram illustrating the in vivo experimental setup for neuron activity mapping in the visual pathway using the ultrasonic device.²⁷ (f) Simulated and measured ultrasound field patterns demonstrating spatial resolution, with indicated spot size and estimated visual acuity.²⁷ (g) Graph showing the amplitude of neural responses evoked by ultrasound at different center frequencies and frame rates, demonstrating temporal resolution capabilities.²⁷ Figures (e–g) have been reproduced from ref. 27 with permission from Gengxi Lu et al., CC BY; copyright: 2024.

In parallel, other studies explore different targets. Gong et al. used low-frequency ring-transducer ultrasound stimulation (LRUS) for visual cortex (VC) stimulation, bypassing damaged optic nerves, and showing faster VC responses than light. Qian et al. explored extraocular ultrasound retinal stimulation, demonstrating noninvasiveness and safety in rat models. Their work achieved spatial/temporal resolutions comparable to those of FDA-approved prostheses and generated precise visual patterns.¹²⁹

Beyond neurostimulation, Sun's research on material-level artificial intelligence (AI) demonstrates integrating sliding ferroelectricity within nanotubes for a programmable photovoltaic effect. This advances piezoelectric applications and shows potential for AI/programming by integrating sensing, memory, computation, and power at the material level.¹³⁰ Building upon the potential of non-invasive ultrasound, recent advancements explore novel approaches. Lu et al. introduced a fully noninvasive, imaging-guided ultrasonic retina prosthesis (U-RP) utilizing a customized 2D ultrasound microarray capable of simultaneous imaging and stimulation.²⁷ This system, depicted in Fig. 8e, places the ultrasound probe directly on the eye surface for efficient and non-invasive delivery of ultrasound waves. Integrating real-time 3D imaging guidance and auto-alignment, the system dynamically generates and precisely steers arbitrary 2D ultrasound patterns onto the retina. Significant performance advancements were demonstrated, including a spatial resolution achieving an estimated visual acuity better than 20/400 (Fig. 8f) and a temporal resolution of 15 Hz (Fig. 8g). The study suggested acoustic radiation force (ARF) as the primary biophysical mechanism and confirmed comprehensive safety.

4.3 Upconversion materials

Upconversion nanoparticles (UCNPs) have garnered significant interest for biomimetic vision, as they can locally convert penetrating near-infrared (NIR) light into visible light at levels potentially sufficient to activate retinal photoreceptor cells. This capability stems from the discrete electronic energy structure of doped lanthanide ions, allowing precise tuning of emission wavelengths to match the spectral sensitivity of visual pigments or optogenetic tools. Various research efforts have explored UCNPs for neural stimulation and visual activation. For instance, the team led by Liu et al. has developed tricolor UCNPs with switchable emission wavelengths, enabling selective activation of different neuronal populations by matching the spectral characteristics of optogenetic proteins.¹³¹

Ma et al. introduced injectable photoreceptors based on UCNPs for activating NIR vision in mammals.¹³²Fig. 9a illustrates the endocytosis of these injected nanoparticles by retinal photoreceptor cells. Fig. 9b depicts the visual conduction pathway mediated by these NIR transducers, shown to be analogous to that of visible light perception. Mice injected with these upconversion nanoparticles exhibited a pupillary light reflex under NIR light, indicating that upconversion facilitates NIR light perception similar to visible light. Building upon these foundational studies, integrating UCNP function into a robust, implantable optical device represents a significant advancement. A recent study by Wang et al. reports the development of an implantable dual-functional intraocular lens (DF-IOL) designed to provide both enhanced vision under dim infrared conditions and self-adaptive protection from strong light.¹³³ This innovative device integrates UCNPs and In-phthalocyanine (InPcs) within a polymer matrix (Fig. 9c). The DF-IOL was engineered to achieve ultralow-threshold upconversion luminescence for efficient NIR-to-visible conversion and incorporated a nonlinear optical material for self-adaptive light limiting under strong illumination. Crucially, in vivo validation in rabbits confirmed the device's effectiveness (Fig. 9d). Fig. 9e presents representative electroretinogram (ERG) results, demonstrating a clear retinal electrical response to 980 nm NIR illumination, specifically in DF-IOL implanted eyes, indicative of successful upconversion and signal transduction. Furthermore, under strong 532 nm visible light, the ERG amplitude was significantly reduced in implanted eyes compared to those of controls, confirming the device's protective function against bright light-induced overstimulation.

Fig. 9 Upconversion nanomaterials for vision applications. (a) Schematic illustrating the sub-retinal injection of pbUCNPs into a mouse and their binding to photoreceptors to convert NIR light to visible light.¹³² (b) Illustration of the pbUCNP structure modified for photoreceptor binding.¹³² Figures (a and b) have been reproduced from ref. 132 with permission from Elsevier Inc.; copyright: 2019. (c) Composition and structure schematic of the dual-functional intraocular lens (DF-IOL) containing upconversion nanoparticles (UCNPs) and In-phthalocyanines (InPcs) within a polymer matrix.¹³³ (d) Photographs showing the fabricated DF-IOL and schematic of its microincision implantation procedure in a rabbit eye.¹³³ (e) Graph of electroretinogram (ERG) amplitude in control and DF-IOL implanted rabbits under visible and NIR light stimulation, demonstrating infrared sensing and strong light protection capabilities.¹³³ Figures (c–e) have been reproduced from ref. 133 with permission from Elsevier B.V.; copyright: 2025. (f) The scheme of near-infrared spatiotemporal color vision in humans enabled by upconversion contact lenses (UCLs).¹³⁴ Figure (f) has been reproduced from ref. 134 with permission from Elsevier Inc.; copyright: 2025.

It is worth noting that Xue's group developed a wearable contact lens that enables humans to achieve near-infrared spatiotemporal color vision. This contact lens has excellent optical performance, hydrophilicity, flexibility, and biocompatibility (Fig. 9f).¹³⁴ Researchers have synthesized three color orthogonal UCNPs with multi-wavelength conversion capability, which can convert near-infrared light of different wavelengths into visible light of the three primary colors of red, green, and blue. These UCNPs were integrated into poly(2-hydroxyethyl methacrylate) (pHEMA) contact lenses to prepare trichromatic upconversion contact lenses (tUCLs) for human near-infrared color vision. To solve the problem of UCLs being unable to achieve fine image perception, a wearable glass system consisting of three flat-convex lenses and a built-in flat UCL has been developed. This system enables participants to distinguish near-infrared moving gratings, with a spatial resolution threshold of approximately 65 cycles per degree (c per d), which is comparable to the normal visual spatial resolution threshold of humans. After wearing UCLs, mice can perceive visible light converted from near-infrared light and recognize the spatiotemporal information of near-infrared light. Humans wearing UCLs can accurately recognize near-infrared time information, such as Morse code, and distinguish near-infrared pattern images. This type of infrared sensing contact lens offers a novel strategy for controlled light conversion and modulation within biological contexts and advanced photonic devices, holding potential for clinical applications in visual enhancement or restoration. However, there are currently some limitations, such as the difficulty of detecting near-infrared information in natural environments and the inability of UCLs to achieve fine image perception. Nevertheless, this research lays the foundation for the application of human near-infrared spatiotemporal color vision and is expected to play a role in infrared information encoding and transmission, enhanced vision, and other fields.

4.4 Photothermal materials

Furthermore, recent research has explored other energy conversion modalities like photoacoustic and photothermal effects for retinal stimulation. For example, flexible polydimethylsiloxane (PDMS)/carbon films have been shown to efficiently convert near-infrared laser pulses into localized ultrasound, enabling high-resolution (56 µm) RGC stimulation with minimal temperature increase, offering a new pathway for minimally invasive retinal prostheses.¹³⁵ Similarly, surface-functionalized bipyramidal gold nanoparticles (BipyAu) have demonstrated enhanced photothermal conversion efficiency and good biocompatibility in corneal tests, showing potential for retinal stimulation application.¹³⁶

Plasmonic gold nanorods (AuNRs) are also being investigated for photothermal stimulation, where their plasmon resonance converts absorbed light energy into heat, leading to localised temperature increases that can stimulate neurons. However, the clinical application of this effect is contingent on a critical trade-off between photothermal efficiency and tissue safety. While localised heating is necessary to activate temperature-sensitive ion channels like transient receptor potential vanilloid 1(TRPV1) for neural stimulation, excessive temperature can lead to irreversible cellular damage, protein denaturation, and inflammation.¹³⁷ Research has demonstrated that even minimal elevations in temperature, in excess of physiological standards, can exert stress upon retinal tissue. This phenomenon is accompanied by substantial microglial activation, which has been observed even at temperatures as low as 38.7 °C. This has been identified as an early indicator of potential damage.¹³⁸ The achievement of therapeutic efficacy is contingent upon precise control to maintain temperatures within a narrow safe window. This is defined as a range sufficiently high for stimulation but below the threshold for heat-induced injury, which, in some cases, is just a few degrees above the baseline temperature.¹³⁹

A recent study by Nie et al. explored intravitreally injected plasmonic gold nanorods (AuNRs) for retinal stimulation.⁵⁶ Intravitreal injection, a less invasive procedure compared to subretinal injection, was used to deliver anti-Thy1 antibody-conjugated AuNRs into multiple layers of the degenerative retina (Fig. 10a). Using a custom system (Fig. 10b), these AuNRs were shown to effectively activate bipolar cells with patterned near-infrared (NIR) light, offering extensive coverage and high spatial resolution, and enabling targeted stimulation via antibody conjugation. Calcium imaging demonstrated highly localized neural activation, specifically within the laser-projected area in retinal explants (Fig. 10c and d). Importantly, in fully blind mice, this patterned NIR stimulation evoked electrocorticogram (ECoG) responses in the visual cortex, confirming that AuNRs enabled cortical activity in blind animals without causing retinal damage (Fig. 10e and f).

Fig. 10 Patterned NIR activation of retinal neurons by intravitreally injected plasmonic gold nanorods. (a) Schematic illustration of intravitreal injection of plasmonic nanorods into the degenerative retina. (b) Schematic of the custom system used for in vivo fundoscopy, laser projection, and ECoG recording. (c) Calcium imaging of retinal explants under different conditions (AuNR only, PBS + NIR, and AuNR + NIR), showing localized neural activation by AuNRs and NIR. (d) Representative calcium response traces from individual neurons within the laser-projected area, indicating light-induced activity. (e) Representative fundus images (pre- and post-stimulation) and ECoG recording from AuNR-injected blind mice, demonstrating cortical responses to NIR laser stimulation. (f) Graph showing ECoG amplitude from blind mice during visible and NIR laser stimulation, confirming NIR-induced cortical activity in blind animals.⁵⁶ Fig. 10 has been reproduced from ref. 56 with permission from American Chemical Society; Copyright: 2025.

Beyond this, Zhu et al. explored wireless and opto-stimulated flexible implants as artificial retinas using ferroelectric BiFeO₃-BaTiO₃/P(VDF-TrFE) composites that convert absorbed light to heat, and subsequently electrical signals via the pyroelectric effect, demonstrating a strong photoelectric response across a wide wavelength range.¹⁴⁰ These flexible materials hold promise for artificial retina applications due to their light-responsive properties and ease of preparation.

5. Discussion and future perspective

Degenerative retinal diseases pose a significant global health challenge, causing irreversible vision loss for millions. While various strategies, including optogenetics and stem cell therapy, hold promise, the use of retinal prostheses remains a leading approach for restoring functional vision by electrically stimulating surviving neural circuits in the degenerated retina. Traditional retinal prostheses based on electrode arrays, such as Argus II and PRIMA, have demonstrated proof-of-concept in clinical trials. Retinal prostheses based on nano/micro-energy-conversion materials also provide novel perspectives. However, retinal prosthesis design still faces numerous challenges.

5.1 The challenges in next-generation retinal prostheses

The limitation of electrical stimulation is a common issue encountered by both traditional retinal prostheses and nano/micromaterial based retinal prostheses. A long-standing issue is the non-selective activation of ON and OFF pathways. In a healthy retina, these two pathways work in opposition to encode contrast. Standard electrical stimulation, however, generates relatively diffuse electric fields that indiscriminately activate the anatomically intermingled ON and OFF cells, leading to a perceptual cancellation of signals. This issue is made worse by the production of non-physiological spiking patterns. These patterns are usually highly synchronized and artificial bursts, and they cannot copy the rich temporal codes that are part of natural vision.³⁶ A critical design imperative, therefore, is the selective activation of distinct RGC populations, particularly the functionally antagonistic ON (light increment) and OFF (light decrement) pathways. The failure of most current electrical prostheses to differentiate between these opposing signals leads to their indiscriminate, simultaneous stimulation, which results in the perceptual cancellation of contrast and severely degrades contrast sensitivity.¹³

Nevertheless, significant research in the neuroengineering community has explored advanced strategies to overcome this barrier. For instance, landmark studies have demonstrated that it is possible to achieve differential responses by precisely modulating the stimulus waveform. Amplitude-modulated high-frequency stimulation (HFS) has been shown to drive ON and OFF cells in opposing directions, a feat achieved by inducing localized membrane hyperpolarization near the electrode.^141,142 Similarly, carefully selecting the stimulus pulse duration can preferentially bias the response toward ON cells.¹⁴³

The clinical translation of retinal prostheses has also been hampered by the pathological state of the degenerated retina itself. Critically, in the diseased retina, neural circuits undergo significant remodeling. Recent evidence demonstrates that this leads to pathological, long-range correlations between ganglion cells, creating a strongly coupled network. As a direct consequence, a localized electrical stimulus evokes a widespread and non-focal response that spreads far beyond the target area. This aberrant network behavior, rather than the electrode size alone, likely explains the low-resolution vision in patients with prostheses.¹⁴⁴ This confluence of issues reveals that earlier development efforts, often focused on an approach of increasing electrode density, fell into an instance of “blind miniaturization” while largely ignoring the intrinsic coding logic of the retina.

Truly meaningful vision restoration requires the device to replicate or preserve this intricate temporal and spatial encoding pattern, not simply activate surviving neurons. Overall, the design of next-generation retinal prostheses still faces the significant challenges of achieving cell-type specificity, generating naturalistic neural codes, and overcoming the noisy environment of the diseased retina, all of which remain critical areas for future research.

5.2 Advantages and current limitations of energy-conversion nano/micromaterial based retinal prostheses

The primary advantage of the energy-conversion materials summarized in Table 1 is their potential to circumvent the “blind miniaturization” by reshaping the implant’s form and function entirely, rather than simply shrinking conventional electrodes. This is exemplified by the development of foldable, wide-field prostheses that conform to the eye's curvature, the shift to near-infrared (NIR) responsive polymers to avoid interfering with residual vision, and the conceptual leap to “liquid retinal prostheses” using injectable nanoparticles.^19,109,111

Table 1 Energy-conversion nano/micromaterials for retinal prosthesis

Energy conversion type	Material class	Examples	Nano/microstructure	Key advantages	Current limitations	Clinical trial status	Ref.
Photovoltaic	2D semiconductors	MoS2, graphene (in CurvIS array), black Phosphorus (b-AsP), MoTe2	2D layers, ultrathin ∼51 nm structure	Flexible, biocompatible, high sensitivity, high resolution, efficient light absorption, tunable bandgap	Challenges in manufacturing scalability for wafer-scale production; poor long-term stability in physiological environments and potential immunogenicity	Preclinical (rodent models)	13, 15, 25, 74–91, 145 and 146
Photovoltaic	Organic semiconductors & nanoparticles	Conjugated polymers (e.g., P3HT, PEDOT: PSS, PCPDTBT), polymer films (OUReP), P3HT nanoparticles, AgNPs integrated with organic PV	Multilayer thin films, nanoparticles	Flexible, biocompatible, low trauma, high resolution, neural stimulation, efficient NIR/visible response	Suboptimal long-term stability in physiological saline; potential thermal effects at high irradiance; electrode crosstalk limiting spatial resolution	Preclinical (large animal/primate models)	19 and 95–113
Photovoltaic	Other inorganic semiconductor nanomaterials	TiO2, Au-TiO2 NW array, perovskite (CsPbI3) NW array, perovskite QDs, ZnIn2S4/NGQD microflowers (MFs/QDs), hg-C3N4 nanoparticles	Nanowires, MFs (2–5 µm), QDs, nanoparticles	High absorption, tunable bandgap, efficient conversion, biocompatible, self-powered operation, metal-free composition	For perovskites, potential cytotoxicity from heavy metal leakage (e.g., lead) causing developmental defects and apoptosis; poor stability in aqueous environments; for QDs, induction of specific cell death pathways (e.g., ferroptosis) in retinal cells	Preclinical (primate models)	134, 53, 112, 114–119, 121, 147 and 148
Piezoelectric	Piezoelectric materials & composites	P(VDF-TrFE), LLCP (azobenzene-containing liquid crystal polymer), PMN-PT, BiFeO3-BaTiO3/P(VDF-TrFE) composites	Films (∼20 µm thick), nanodots, single crystals	Convert mechanical/light energy to electrical signals, neural stimulation, flexibility, biocompatible, light-induced pyroelectric effect, ultrasound stimulation	Limited temporal resolution (∼15 Hz); risk of tissue heating from ultrasound energy; complex fabrication and alignment	Preclinical (rodent/ex vivo models)	27, 47, 58, 60, 62–65, 125–129 and 140
Upconversion	Upconversion nanoparticles (UCNPs)	Lanthanide-doped UCNPs (e.g., NaYF4), UCNP-integrated intraocular lens (DF-IOL), pbUCNPs, UCNP contact lenses (tUCLs)	Nanoparticles, implantable lens functionalized with UCNPs	Absorb NIR, emit visible, deep penetration, activate downstream effectors (optogenetics/neurons), protective function, spatiotemporal color vision	Low quantum efficiency (<1%); high excitation power required, posing potential thermal risk; limited fine image perception	Preclinical (implants); human testing (wearable lenses)	134, 59, 67 and 131–133
Photothermal	Gold/carbon nanomaterials & films	Bipyramidal gold nanoparticles (BipyAu), plasmonic gold nanorods (AuNRs), liquid metal, indocyanine green (ICG)	Nanoparticles, nanorods, films	Convert light to heat, localized heating, neural stimulation, less invasive pathway, plasmon resonance	Risk of thermal damage to adjacent tissue; challenges in precise temperature control at the cellular interface; long-term nanoparticle clearance and toxicity concerns	Preclinical (rodent models)	56, 70–73, 136, 137 and 149
Photoacoustic	Polymer films	Flexible PDMS/carbon films	Films	Convert light to ultrasound, high-resolution RGC stimulation, minimally invasive	Nascent research stage with limited preclinical data on safety and efficacy; key hurdles include ensuring long-term biocompatibility and stability, a common challenge for novel flexible bio-interfaces; optimizing photoacoustic conversion efficiency	Early-stage preclinical	25 and 135

---

Crucially, these new platforms represent fundamental shifts in the approach to selective neural stimulation. For instance, the Au-TiO₂ nanowire arrays, while still photoelectric, enable a high-resolution stimulation. Different functional RGC responses, such as transient versus sustained ON-cell activity, can be distinguished and characterized under this stimulation, demonstrating a clear focus on functional cell-type outcomes.¹¹⁶ The photoacoustic platforms take this a step further by changing the stimulation modality itself. By converting light to ultrasound, they employ a mechanical stimulus to perform high-resolution RGC stimulation, which conceptually bypasses the problem of electrical field spread in conductive tissue.^25,135 Thus, the development of these materials shows a clear and deliberate path toward addressing the challenge of selective stimulation. This progress leverages high-resolution electrical interfaces and alternative stimulation modalities.

Nevertheless, these novel platforms introduce their own distinct limitations. Challenges remain in energy conversion efficiency, which may require high light intensities, and the unresolved questions of long-term biocompatibility and stability for materials such as polymer blends or nanoparticles. Furthermore, even a perfectly localized stimulus must contend with the fundamental biological reality of interfacing with a pathologically remodeled and “noisy” neural network, a problem shared by all prosthetic technologies.

5.3 Future perspective

The most promising path forward, therefore, is not to discard the principles of neural engineering, but to empower them with new material platforms. The next generation of devices could leverage the spatial precision of nano/micromaterials to deliver the advanced stimulation patterns known to influence neural coding. For example, a high-density photovoltaic film could be designed to deliver the specific high-frequency waveforms or timed pulses that have shown promise in selectively engaging ON and OFF pathways. This synergy between material precision and functional stimulation could finally unlock the clinical potential of these advanced strategies.

Crucially, all future development must be validated in degenerated retinal models. As previously discussed, strategies that succeed in healthy tissue may fail in the rewired and hyperexcitable retina. The ultimate goal must be to develop “smart” interfaces, perhaps with surface functionalizations for cellular targeting or feedback mechanisms to adapt to and even suppress pathological activity. Ultimately, integrating advanced nano/micromaterials with smart control systems may unlock new possibilities. By bridging the gap between advanced materials and established neuro-engineering principles, the field can move beyond simply activating neurons and toward the ultimate goal of replicating the retina's neural code.

Author contributions

Ruiying Li: conceptualization, writing – original draft, writing – review & editing and visualization. Yueyang Shang: writing – original draft and writing – review. Tianyu Gao: writing – review & editing. Xinmiao Lan: conceptualization and writing – review. Peijian Feng: conceptualization, supervision and writing review – editing.

Conflicts of interest

The authors declare no conflict of interest.

Data availability

No primary research results, software or code have been included and no new data were generated or analyzed as part of this review.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (NSFC) project (grant no. 22305161 and 32301133), the Basic Research Funds for Beijing Municipal Universities (grant no. XJJS202518) and the R&D Program of Beijing Municipal Education Commission (grant no. KM202310025022).

References

1. I. Tochitsky, M. A. Kienzler, E. Isacoff and R. H. Kramer, Chem. Rev., 2018, 118, 10748–10773.
2. J. Kralik, M. van Wyk, N. Stocker and S. Kleinlogel, Commun. Biol., 2022, 5, 1116.
3. J.-A. Sahel, E. Boulanger-Scemama, C. Pagot, A. Arleo, F. Galluppi, J. N. Martel, S. D. Esposti, A. Delaux, J.-B. de Saint Aubert, C. de Montleau, E. Gutman, I. Audo, J. Duebel, S. Picaud, D. Dalkara, L. Blouin, M. Taiel and B. Roska, Nat. Med., 2021, 27, 1223–1229.
4. M. R. Islam, S. Afroj and N. Karim, ACS Nano, 2023, 17, 18481–18493.
5. G. Manfredi, E. Colombo, J. Barsotti, F. Benfenati and G. Lanzani, Annu. Rev. Phys. Chem., 2019, 70, 99–121.
6. L. Yue, P. Falabella, P. Christopher, V. Wuyyuru, J. Dorn, P. Schor, R. J. Greenberg, J. D. Weiland and M. S. Humayun, Ophthalmology, 2015, 122, 2545–2552.
7. J. D. Dorn, A. K. Ahuja, A. Caspi, L. da Cruz, G. Dagnelie, J. A. Sahel, R. J. Greenberg, M. J. McMahon and I. I. S. G. Argus, JAMA Ophthalmol., 2013, 131, 183–189.
8. R. M. Mirochnik and J. S. Pezaris, Mil. Med. Res., 2019, 6, 19.
9. K. Stingl, K. U. Bartz-Schmidt, D. Besch, C. K. Chee, C. L. Cottriall, F. Gekeler, M. Groppe, T. L. Jackson, R. E. MacLaren, A. Koitschev, A. Kusnyerik, J. Neffendorf, J. Nemeth, M. A. Naeem, T. Peters, J. D. Ramsden, H. Sachs, A. Simpson, M. S. Singh, B. Wilhelm, D. Wong and E. Zrenner, Vision Res., 2015, 111, 149–160.
10. Z. C. Chen, B.-Y. Wang, A. Kochnev Goldstein, E. Butt, K. Mathieson and D. Palanker, J. Neural Eng., 2022, 19, 055008.
11. M. B. Bhuckory, N. Monkongpitukkul, A. Shin, A. Kochnev Goldstein, N. Jensen, S. V. Shah, D. Pham-Howard, E. Butt, R. Dalal, L. Galambos, K. Mathieson, T. Kamins and D. Palanker, Nat. Commun., 2025, 16, 2820.
12. P. Falabella, H. Nazari, P. Schor, J. D. Weiland and M. S. Humayun, in Artificial Vision: A Clinical Guide, ed. V. P. Gabel, Springer International Publishing, Cham, 2017, pp. 49–63,  DOI:10.1007/978-3-319-41876-6_5.
13. K. Y. Wu, M. Mina, J.-Y. Sahyoun, A. Kalevar and S. D. Tran, Sensors, 2023, 23, s23135782.
14. L. da Cruz, J. D. Dorn, M. S. Humayun, G. Dagnelie, J. Handa, P. O. Barale, J. A. Sahel, P. E. Stanga, F. Hafezi, A. B. Safran, J. Salzmann, A. Santos, D. Birch, R. Spencer, A. V. Cideciyan, E. de Juan, J. L. Duncan, D. Eliott, A. Fawzi, L. C. Olmos de Koo, A. C. Ho, G. Brown, J. Haller, C. Regillo, L. V. Del Priore, A. Arditi and R. J. Greenberg, Ophthalmology, 2016, 123, 2248–2254.
15. W. G. Chung, J. Jang, G. Cui, S. Lee, H. Jeong, H. Kang, H. Seo, S. Kim, E. Kim, J. Lee, S. G. Lee, S. H. Byeon and J.-U. Park, Nat. Nanotechnol., 2024, 19, 688–697.
16. G. Li, J. Yang, Y. Wang, W. Wang and L. Liu, Nanoscale, 2018, 10, 21046–21051.
17. X. Luo, C. Chen, Z. He, M. Wang, K. Pan, X. Dong, Z. Li, B. Liu, Z. Zhang, Y. Wu, C. Ban, R. Chen, D. Zhang, K. Wang, Q. Wang, J. Li, G. Lu, J. Liu, Z. Liu and W. Huang, Nat. Commun., 2024, 15, 3086.
18. P. Kusumanchi, S. S. Keller and R. S. Davidsen, Micro Nano Eng., 2024, 25, 100289.
19. L. Ferlauto, M. J. I. Airaghi Leccardi, N. A. L. Chenais, S. C. A. Gilliéron, P. Vagni, M. Bevilacqua, T. J. Wolfensberger, K. Sivula and D. Ghezzi, Nat. Commun., 2018, 9, 992.
20. D. Zhao, R. Huang, J.-M. Gan and Q.-D. Shen, ACS Nano, 2022, 16, 19892–19912.
21. N. Muzzio, M. E. Martinez-Cartagena and G. Romero, Adv. Drug Delivery Rev., 2022, 190, 114554.
22. D. Zhao, Z. Cao, G. Zhong, L. Shi, J. Xie and J. Hao, Adv. Funct. Mater., 2025, 2506050.
23. S. Yang and X. Luo, Nanoscale, 2014, 6, 4438–4457.
24. B. Huang, M.-H. Cao, F.-D. Nie, H. Huang and C.-W. Hu, Def. Technol., 2013, 9, 59–79.
25. N. Goel and R. Kumar, Nanomicro Lett., 2025, 17, 197.
26. E. Joseph, M. Ciocca, H. Wu, S. Marcozzi, M. Ucci, K. Keremane, L. Zheng, B. Poudel, C. Wu, A. Camaioni, K. Wang, S. Priya and T. Brown, npj Biosensing, 2024, 1, 15.
27. G. Lu, C. Gong, Y. Sun, X. Qian, D. S. Rajendran Nair, R. Li, Y. Zeng, J. Ji, J. Zhang, H. Kang, L. Jiang, J. Chen, C.-F. Chang, B. B. Thomas, M. S. Humayun and Q. Zhou, Nat. Commun., 2024, 15, 4481.
28. D. Holmes, Nature, 2018, 561, S2–S3.
29. R. H. Masland, Nat. Neurosci., 2001, 4, 877–886.
30. T. Ichinose and S. Habib, Front. Ophthalmol., 2022, 2, 989002.
31. T. Baden, P. Berens, K. Franke, M. Román Rosón, M. Bethge and T. Euler, Nature, 2016, 529, 345–350.
32. D. Palanker, Cold Spring Harbor Perspect. Med., 2023, 13, cshperspect.a041525.
33. P. R. Stoddart, J. M. Begeng, W. Tong, M. R. Ibbotson and T. Kameneva, Front. Cell. Neurosci., 2024, 18, 1360870.
34. M. Castillo García and E. Urdapilleta, Phys. Biol., 2022, 19, 066006.
35. K. Nowik, E. Langwińska-Wośko, P. Skopiński, K. E. Nowik and J. P. Szaflik, J. Clin. Neurosci., 2020, 78, 8–19.
36. M. Carleton and N. W. Oesch, eNeuro, 2024, 11 DOI:10.1523/ENEURO.0110-24.2024.
37. D. K. Freeman, J. F. Rizzo 3rd and S. I. Fried, J. Neural Eng., 2011, 8, 035005.
38. C. Haselier, S. Biswas, S. Rösch, G. Thumann, F. Müller and P. Walter, PLoS One, 2017, 12, e0190048.
39. L. Fagiolari, M. Sampò, A. Lamberti, J. Amici, C. Francia, S. Bodoardo and F. Bella, Energy Storage Mater., 2022, 51, 400–434.
40. H. Yu, Y. Wang, X. Zou, J. Yin, X. Shi, Y. Li, H. Zhao, L. Wang, H. M. Ng, B. Zou, X. Lu, K. S. Wong, W. Ma, Z. Zhu, H. Yan and S. Chen, Nat. Commun., 2023, 14, 2323.
41. X. Chen, H. Qin, X. Qian, W. Zhu, B. Li, B. Zhang, W. Lu, R. Li, S. Zhang, L. Zhu, F. Domingues Dos Santos, J. Bernholc and Q. M. Zhang, Science, 2022, 375, 1418–1422.
42. C. Dai and B. Liu, Energy Environ. Sci., 2020, 13, 24–52.
43. W.-J. Ong, L.-L. Tan, Y. H. Ng, S.-T. Yong and S.-P. Chai, Chem. Rev., 2016, 116, 7159–7329.
44. C. Qian, P. Feng, J. Yu, Y. Chen, Q. Hu, W. Sun, X. Xiao, X. Hu, A. Bellotti, Q.-D. Shen and Z. Gu, Angew. Chem., Int. Ed., 2017, 56, 2588–2593.
45. P. Feng, Y. Chen, L. Zhang, C.-G. Qian, X. Xiao, X. Han and Q.-D. Shen, ACS Appl. Mater. Interfaces, 2018, 10, 4359–4368.
46. A. Kumar, G. Sharma, M. Naushad, A. A. H. Al-Muhtaseb, A. García-Peñas, G. T. Mola, C. Si and F. J. Stadler, Chem. Eng. J., 2020, 382, 122937.
47. A. Sood, M. Desseigne, A. Dev, L. Maurizi, A. Kumar, N. Millot and S. S. Han, Small, 2023, 19, 2206401.
48. Z. He, H. Duan, J. Zeng, J. Zhou, X. Zhong, Z. Wu, S. Ni, Z. Jiang, G. Xie, J. Y. Lee, Y. Lu, Y. Zeng, B. Zhang, W. B. Ying, Z. Yang, Z. Zhang and G. Liu, Sci. Adv., 2025, 11, eads2834.
49. J. Zhang, X. Su, M. Shen, Z. Dai, L. Zhang, X. He, W. Cheng, M. Cao and G. Zou, Sci. Rep., 2013, 3, 2109.
50. M. B. Prince, in Applied Solid State Physics, ed. W. Low and M. Schieber, Springer US, Boston, MA, 1970, pp. 223–234,  DOI:10.1007/978-1-4684-1854-5_9.
51. M. Han, S. B. Srivastava, E. Yildiz, R. Melikov, S. Surme, I. B. Dogru-Yuksel, I. H. Kavakli, A. Sahin and S. Nizamoglu, ACS Appl. Mater. Interfaces, 2020, 12, 42997–43008.
52. G. Simone, D. Di Carlo Rasi, X. de Vries, G. H. L. Heintges, S. C. J. Meskers, R. A. J. Janssen and G. H. Gelinck, Adv. Mater., 2018, 30, 1804678.
53. R. Balamur, G. O. Eren, H. N. Kaleli, O. Karatum, L. Kaya, M. Hasanreisoglu and S. Nizamoglu, Adv. Sci., 2024, 11, 2306097.
54. J. Tang, C. Wu, Z. Qiao, J. Pi, Y. Zhang, F. Luo, J. Sun, W. Dan and H. Fan, J. Mater. Chem. B, 2022, 10, 1601–1611.
55. X. Chen, C. Wang, E. Baker and C. Sun, Sci. Rep., 2015, 5, 11977.
56. J. Nie, K. Eom, H. M. AlGhosain, A. Neifert, A. Cherian, G. M. Gerbaka, K. Y. Ma, T. Liu and J. Lee, ACS Nano, 2025, 19, 11823–11840.
57. D. Palanker, Y. Le Mer, S. Mohand-Said, M. Muqit and J. A. Sahel, Ophthalmology, 2020, 127, 1097–1104.
58. B. Peng, X. Chen, G. Yu, F. Xu, R. Yang, Z. Yu, J. Wei, G. Zhu, L. Qin, J. Zhang, Q. Shen and Y. Yu, Adv. Funct. Mater., 2023, 33, 2214172.
59. X. Lin, X. Chen, W. Zhang, T. Sun, P. Fang, Q. Liao, X. Chen, J. He, M. Liu, F. Wang and P. Shi, Nano Lett., 2018, 18, 948–956.
60. Q. Xu, X. Gao, S. Zhao, Y.-N. Liu, D. Zhang, K. Zhou, H. Khanbareh, W. Chen, Y. Zhang and C. Bowen, Adv. Mater., 2021, 33, 2008452.
61. D. Zhao, P. J. Feng, J. H. Liu, M. Dong, X. Q. Shen, Y. X. Chen and Q. D. Shen, Adv. Mater., 2020, 32, e2003800.
62. T. Kim, H. J. Kim, W. Choi, Y. M. Lee, J. H. Pyo, J. Lee, J. Kim, J. Kim, J.-H. Kim, C. Kim and W. J. Kim, Nat. Biomed. Eng., 2023, 7, 149–163.
63. Z. Zhang, X. Yang, Y. Zhao, F. Ye and L. Shang, Adv. Mater., 2023, 35, e2300220.
64. X. Zheng, Y. Jia and A. Chen, Nat. Commun., 2021, 12, 4875.
65. F. Wang, Z. Song, X. Cai, K. Guo, X. Pan, C. Ren and B. Li, Nanoscale, 2025, 17, 6981–6992.
66. R. Martínez-Fernández, R. Rodríguez-Rojas, M. Del Álamo, F. Hernández-Fernández, J. A. Pineda-Pardo, M. Dileone, F. Alonso-Frech, G. Foffani, I. Obeso, C. Gasca-Salas, E. de Luis-Pastor, L. Vela and J. A. Obeso, Lancet Neurol., 2018, 17, 54–63.
67. J. Feng, J. Alves, D. M. de Clercq and T. W. Schmidt, Annu. Rev. Phys. Chem., 2023, 74, 145–168.
68. M. Wang, G. Abbineni, A. Clevenger, C. Mao and S. Xu, Nanomedcine, 2011, 7, 710–729.
69. I. A. Khan, T. Yu, M. Yang, J. Liu and Z. Chen, BME Front., 2025, 6, 0120.
70. G. M. Neelgund and A. Oki, Mater. Chem. Front., 2018, 2, 64–75.
71. J. M. Begeng, W. Tong, B. D. Rosal, M. Ibbotson, T. Kameneva and P. R. Stoddart, ACS Nano, 2023, 17, 2079–2088.
72. C. Wang, L. Zhou, C. Liu, J. Qiao, X. Han, L. Wang, Y. Liu, B. Xu, Q. Qiu, Z. Zhang, J. Wang, X. Zhou, M. Zeng, L. Yu and L. Fu, Nat. Commun., 2024, 15, 6362.
73. S. Zhuang, M. He, J. Feng, S. Peng, H. Jiang, Y. Li, N. Hua, Y. Zheng, Q. Ye, M. Hu, Y. Nie, P. Yu, X. Yue, J. Qian and W. Yang, Small Methods, 2023, 7, 2300848.
74. D. Nguyen, M. Valet, J. Dégardin, L. Boucherit, X. Illa, J. de la Cruz, E. del Corro, J. Bousquet, J. A. Garrido, C. Hébert and S. Picaud, Front. Neurosci., 2021, 15, 615256.
75. J. Pang, S. Peng, C. Hou, H. Zhao, Y. Fan, C. Ye, N. Zhang, T. Wang, Y. Cao, W. Zhou, D. Sun, K. Wang, M. H. Rümmeli, H. Liu and G. Cuniberti, ACS Sens., 2023, 8, 482–514.
76. F. T. Duvan, M. Cunquero, E. Masvidal-Codina, S. T. Walston, M. Marsal, J. M. de la Cruz, D. Viana, D. Nguyen, J. Degardin, X. Illa, J. M. Zhang, M. D. P. Bernícola, J. G. Macias-Montero, C. Puigdengoles, G. Castro-Olvera, E. Del Corro, S. Dokos, M. Chmeissani, P. Loza-Alvarez, S. Picaud and J. A. Garrido, Nanoscale Horiz., 2024, 9, 1948–1961.
77. H. Ji, Z. Song, A. Wu, Y.-C. Zou and G. Yang, Nanoscale, 2025, 17, 7574–7599.
78. Z. Dang, F. Guo, Y. Zhao, K. Jin, W. Jie and J. Hao, Adv. Funct. Mater., 2024, 34, 2400105.
79. S. Xiao, T. Wang, Y. Liu, C. Xu, X. Han and X. Yan, Phys. Chem. Chem. Phys., 2016, 18, 26661–26669.
80. J. H. Kim, H. S. Lee, G. H. An, J. Lee, H. M. Oh, J. Choi and Y. H. Lee, ACS Nano, 2020, 14, 11985–11994.
81. I. Epstein, B. Terrés, A. J. Chaves, V. V. Pusapati, D. A. Rhodes, B. Frank, V. Zimmermann, Y. Qin, K. Watanabe, T. Taniguchi, H. Giessen, S. Tongay, J. C. Hone, N. M. R. Peres and F. H. L. Koppens, Nano Lett., 2020, 20, 3545–3552.
82. G. Kim, B. Huet, C. E. Stevens, K. Jo, J.-Y. Tsai, S. Bachu, M. Leger, S. Song, M. Rahaman, K. Y. Ma, N. R. Glavin, H. S. Shin, N. Alem, Q. Yan, J. R. Hendrickson, J. M. Redwing and D. Jariwala, Nat. Commun., 2024, 15, 6361.
83. S. Sovizi, S. Angizi, S. A. A. Alem, R. Goodarzi, M. R. Rahmani Taji Boyuk, H. Ghanbari, R. Szoszkiewicz, A. Simchi and P. Kruse, Chem. Rev., 2023, 123, 13869–13951.
84. M.-Y. Tian, Y.-M. Gao, Y.-J. Zhang, M.-X. Ren, X.-H. Lv, K.-X. Hou, C.-D. Jin, H. Zhang, R.-Q. Lian, P.-L. Gong, R.-N. Wang, J.-L. Wang and X.-Q. Shi, Commun. Mater., 2024, 5, 188.
85. N. Li, Q. Wang, C. He, J. Li, X. Li, C. Shen, B. Huang, J. Tang, H. Yu, S. Wang, L. Du, W. Yang, R. Yang, D. Shi and G. Zhang, ACS Nano, 2023, 17, 991–999.
86. Y. Chen, Z. Fan, Z. Zhang, W. Niu, C. Li, N. Yang, B. Chen and H. Zhang, Chem. Rev., 2018, 118, 6409–6455.
87. S. Gou, S. Yang, Y. Cheng, S. Yang, H. Liu, P. Li and Z. Du, Int. J. Mol. Sci., 2024, 25, ijms25168615.
88. X. Zhang, H. Lee, Y. Zhang, T. S. Walmsley, D. Li, E. Levine and Y.-Q. Xu, ACS Appl. Bio Mater., 2022, 5, 305–312.
89. C. Choi, M. K. Choi, S. Liu, M. S. Kim, O. K. Park, C. Im, J. Kim, X. Qin, G. J. Lee, K. W. Cho, M. Kim, E. Joh, J. Lee, D. Son, S. H. Kwon, N. L. Jeon, Y. M. Song, N. Lu and D. H. Kim, Nat. Commun., 2017, 8, 1664.
90. N. Li, S. Zhang, Y. Peng, X. Li, Y. Zhang, C. He and G. Zhang, Adv. Funct. Mater., 2023, 33, 2305589.
91. F. Wang, F. Hu, M. Dai, S. Zhu, F. Sun, R. Duan, C. Wang, J. Han, W. Deng, W. Chen, M. Ye, S. Han, B. Qiang, Y. Jin, Y. Chua, N. Chi, S. Yu, D. Nam, S. H. Chae and Q. J. Wang, Nat. Commun., 2023, 14, 1938.
92. A. Quellmalz, X. Wang, S. Sawallich, B. Uzlu, M. Otto, S. Wagner, Z. Wang, M. Prechtl, O. Hartwig, S. Luo, G. S. Duesberg, M. C. Lemme, K. B. Gylfason, N. Roxhed, G. Stemme and F. Niklaus, Nat. Commun., 2021, 12, 917.
93. J. Kim, H. Seung, D. Kang, J. Kim, H. Bae, H. Park, S. Kang, C. Choi, B. K. Choi, J. S. Kim, T. Hyeon, H. Lee, D. H. Kim, S. Shim and J. Park, Nano Lett., 2021, 21, 9153–9163.
94. N. Aspiotis, K. Morgan, B. März, K. Müller-Caspary, M. Ebert, E. Weatherby, M. E. Light, C.-C. Huang, D. W. Hewak, S. Majumdar and I. Zeimpekis, npj 2D Mater. Appl., 2023, 7, 18.
95. G. Lanzani, G. Chiaravalli, E. Colombo, G. Manfredi, S. Di Marco, V. Vurro and F. Benfenati, Nat. Rev. Bioeng., 2024, 2, 829–848.
96. A. Ahnood, H. Meffin, D. J. Garrett, K. Fox, K. Ganesan, A. Stacey, N. V. Apollo, Y. T. Wong, S. G. Lichter, W. Kentler, O. Kavehei, U. Greferath, K. A. Vessey, M. R. Ibbotson, E. L. Fletcher, A. N. Burkitt and S. Prawer, Adv. Biosyst., 2017, 1, e1600003.
97. Y. S. Hsiao, Y. H. Liao, H. L. Chen, P. Chen and F. C. Chen, ACS Appl. Mater. Interfaces, 2016, 8, 9275–9284.
98. H. Lorach, G. Goetz, R. Smith, X. Lei, Y. Mandel, T. Kamins, K. Mathieson, P. Huie, J. Harris, A. Sher and D. Palanker, Nat. Med., 2015, 21, 476–482.
99. P. Vagni, M. J. I. Airaghi Leccardi, C. H. Vila, E. G. Zollinger, G. Sherafatipour, T. J. Wolfensberger and D. Ghezzi, Nat. Commun., 2022, 13, 3678.
100. Z. H. Yu, W. J. Chen, X. Liu, Q. Y. Xia, Y. N. Yang, M. Dong, J. H. Liu, H. J. Guan, C. Sun, F. D. Feng and Q. D. Shen, ACS Appl. Mater. Interfaces, 2020, 12, 28759–28767.
101. B. Y. Wang, Z. C. Chen, M. Bhuckory, T. Huang, A. Shin, V. Zuckerman, E. Ho, E. Rosenfeld, L. Galambos, T. Kamins, K. Mathieson and D. Palanker, Nat. Commun., 2022, 13, 6627.
102. T. W. Huang, T. I. Kamins, Z. C. Chen, B. Y. Wang, M. Bhuckory, L. Galambos, E. Ho, T. Ling, S. Afshar, A. Shin, V. Zuckerman, J. S. Harris, K. Mathieson and D. Palanker, J. Neural Eng., 2021, 18, abe6b8.
103. T. Matsuo, T. Uchida, J. Sakurai, K. Yamashita, C. Matsuo, T. Araki, Y. Yamashita and K. Kamikawa, Artif. Organs, 2018, 42, E186–E203.
104. K. Yamashita, T. Uchida, T. Matsuo, W. Wong and P. Sundaram, J. Neural Eng., 2021, 18, abf892.
105. K. Yamashita, T. Tanaka, T. Matsuo and T. Uchida, Biomed. Mater., 2021, 16, abe809.
106. F. Huang, J. Bladon, R. C. Lagoy, P. N. Shorrock Jr, M. Hronik-Tupaj, C. A. Zoto, R. E. Connors, W. G. McGimpsey, P. Molnar, S. Lambert, A. R. Rittenhouse and C. R. Lambert, Acta Biomater., 2015, 12, 42–50.
107. J. F. Maya-Vetencourt, D. Ghezzi, M. R. Antognazza, E. Colombo, M. Mete, P. Feyen, A. Desii, A. Buschiazzo, M. Di Paolo, S. Di Marco, F. Ticconi, L. Emionite, D. Shmal, C. Marini, I. Donelli, G. Freddi, R. Maccarone, S. Bisti, G. Sambuceti and F. Benfenati, Nat. Mater., 2017, 16, 681–689.
108. A. Rahmani and K. Eom, Front. Cell. Neurosci., 2024, 18, 1385567.
109. M. J. I. Airaghi Leccardi, N. A. L. Chenais, L. Ferlauto, M. Kawecki, E. G. Zollinger and D. Ghezzi, Commun. Mater., 2020, 1, 21.
110. C. Tortiglione, M. R. Antognazza, A. Tino, C. Bossio, V. Marchesano, A. Bauduin, M. Zangoli, S. V. Morata and G. Lanzani, Sci. Adv., 2017, 3, e1601699.
111. J. F. Maya-Vetencourt, G. Manfredi, M. Mete, E. Colombo, M. Bramini, S. Di Marco, D. Shmal, G. Mantero, M. Dipalo, A. Rocchi, M. L. DiFrancesco, E. D. Papaleo, A. Russo, J. Barsotti, C. Eleftheriou, F. Di Maria, V. Cossu, F. Piazza, L. Emionite and F. Benfenati, Nat. Nanotechnol., 2020, 15, 698–708.
112. S. Francia, D. Shmal, S. Di Marco, G. Chiaravalli, J. F. Maya-Vetencourt, G. Mantero, C. Michetti, S. Cupini, G. Manfredi, M. L. DiFrancesco, A. Rocchi, S. Perotto, M. Attanasio, R. Sacco, S. Bisti, M. Mete, G. Pertile, G. Lanzani, E. Colombo and F. Benfenati, Nat. Commun., 2022, 13, 3677.
113. G. Tullii, C. Bellacanzone, H. Comas Rojas, F. Fumagalli, C. Ronchi, A. Villano, F. Gobbo, M. Bogar, B. Sartori, P. Sassi, G. Zampini, G. Quaglia, L. Latterini, H. Amenitsch and M. R. Antognazza, ACS Appl. Mater. Interfaces, 2025, 17, 22434–22447.
114. J. Liang, X. Yu, J. Qiu, M. Wang, C. Cheng, B. Huang, H. Zhang, R. Chen, W. Pei and H. Chen, ACS Appl. Mater. Interfaces, 2023, 15, 9584–9592.
115. J. Tang, N. Qin, Y. Chong, Y. Diao, Yiliguma, Z. Wang, T. Xue, M. Jiang, J. Zhang and G. Zheng, Nat. Commun., 2018, 9, 786.
116. R. Yang, P. Zhao, L. Wang, C. Feng, C. Peng, Z. Wang, Y. Zhang, M. Shen, K. Shi, S. Weng, C. Dong, F. Zeng, T. Zhang, X. Chen, S. Wang, Y. Wang, Y. Luo, Q. Chen, Y. Chen and J. Zhang, Nat. Biomed. Eng., 2024, 8, 1018–1039.
117. G. Oudeng, S. Banerjee, Q. Wang, D. Jiang, Y. Fan, H. Wu, F. Pan and M. Yang, Small, 2024, 20, e2400300.
118. Z. Long, X. Qiu, C. L. J. Chan, Z. Sun, Z. Yuan, S. Poddar, Y. Zhang, Y. Ding, L. Gu, Y. Zhou, W. Tang, A. K. Srivastava, C. Yu, X. Zou, G. Shen and Z. Fan, Nat. Commun., 2023, 14, 1972.
119. M. Chen, C. Müller, Y. Zhang, L. Meng, A. Kalatanova, J. Madsen, K. Klompmaker, T. Jakobsen, A. Joergensen, A. Askou, G. Huang, W. Li, R. Davidsen, T. Bek, N. Hatzakis, T. Corydon, H. Leinonen, B. Tian and M. Dong, bioRxiv, 2024,  DOI:10.1101/2024.10.23.619813.
120. X. Chen, X. Wang, Z. Cui, Q. Luo, Z. Jiang, Y. Huang, J. Jiang, J. Qiu, Y. Li, K. Yu and J. Zhuang, Int. J. Biol. Sci., 2023, 19, 1791–1812.
121. C. Yang, Z. Du, L. Mei, X. Chen, Y. Liao, L. Ge, J. Kang, Z. Gu, X. Fan and H. Xu, J. Nanobiotechnol., 2025, 23, 144.
122. R. S. Davidsen, S. Hemanth, S. S. Keller, T. Bek and O. Hansen, Micro Nano Eng., 2019, 2, 110–116.
123. B. Rezaei, S. Saghir, J. Y. Pan, R. S. Davidsen and S. S. Keller, Micromachines, 2022, 13, 371.
124. P. Kusumanchi, J. G. Madsen, T. Bek, S. S. Keller and R. S. Davidsen, Biomed. Microdevices, 2025, 27, 7.
125. X. Chen, S. Pan, P. J. Feng, H. Bian, X. Han, J. H. Liu, X. Guo, D. Chen, H. Ge and Q. D. Shen, Adv. Mater., 2016, 28, 10684–10691.
126. L. Jiang, G. Lu, Y. Zeng, Y. Sun, H. Kang, J. Burford, C. Gong, M. S. Humayun, Y. Chen and Q. Zhou, Nat. Commun., 2022, 13, 3853.
127. J. D. Badadhe, H. Roh, B. C. Lee, J. H. Kim and M. Im, Front. Cell. Neurosci., 2022, 16, 971148.
128. P. A. Lo, K. Huang, Q. Zhou, M. S. Humayun and L. Yue, Micromachines, 2020, 11, mi11100929.
129. X. Qian, G. Lu, B. B. Thomas, R. Li, X. Chen, K. K. Shung, M. Humayun and Q. Zhou, BME Front., 2022 DOI:10.34133/2022/9829316.
130. Y. Sun, S. Xu, Z. Xu, J. Tian, M. Bai, Z. Qi, Y. Niu, H. H. Aung, X. Xiong, J. Han, C. Lu, J. Yin, S. Wang, Q. Chen, R. Tenne, A. Zak and Y. Guo, Nat. Commun., 2022, 13, 5391.
131. X. Liu, H. Chen, Y. Wang, Y. Si, H. Zhang, X. Li, Z. Zhang, B. Yan, S. Jiang, F. Wang, S. Weng, W. Xu, D. Zhao, J. Zhang and F. Zhang, Nat. Commun., 2021, 12, 5662.
132. Y. Ma, J. Bao, Y. Zhang, Z. Li, X. Zhou, C. Wan, L. Huang, Y. Zhao, G. Han and T. Xue, Cell, 2019, 177, 243–255.
133. D. Wang, C. Wang, Y. Zhang, W. Xun, D. Chen, Q. Cheng, L. Zhang, Y. Liu, G. Li, S. Wu, H. Zhou and Y. Wang, Chem. Eng. J., 2025, 505, 159737.
134. Y. Ma, Y. Chen, S. Wang, Z.-H. Chen, Y. Zhang, L. Huang, X. Zhang, F. Yin, Y. Wang, M. Yang, Z. Li, K. Huang, X. Fang, Z. Li, M. Wang, W. Liu, J.-N. Li, L. Li, H. Zhao and T. Xue, Cell, 2025, 188, 3375–3388.
135. A. Leong, Y. Li, T. R. Ruikes, J. Voillot, Y. Yuan, G. Chen, A. Facon, C. A. Chhuon, C. Joffrois, G. Tessier, M. Cornebois, J. Dégardin, J. Louise, J. X. Cheng, C. Yang, H. Moulet and S. Picaud, bioRxiv, 2024 DOI:10.1101/2024.09.03.611068 , preprint.
136. D. Alba-Molina, M. Cano, M. Blanco-Blanco, L. Ortega-Llamas, Y. Jiménez-Gómez, A. Gonzalez-Lopez, M. Perez-Perdomo, L. Camacho, J. J. Giner-Casares and M. Gonzalez-Andrades, J. Mater. Chem. B., 2025, 13, 25.
137. Y. Wang, R. Garg, D. Cohen-Karni and T. Cohen-Karni, Nat. Rev. Bioeng., 2023, 1, 193–207.
138. N. L. Opie, U. Greferath, K. A. Vessey, A. N. Burkitt, H. Meffin, D. B. Grayden and E. L. Fletcher, Invest. Ophthalmol. Visual Sci., 2012, 53, 7802–7812.
139. M. Céspedes Tenorio, C. A. Wattson Sánchez and D. S. Dumani, Front. Therm. Eng., 2022, 2, 1005117.
140. Y. Zhu, X. Liu, J. Ma, Z. Wang, H. Jiang, C. Sun, D. Y. Jeong, H. Guan and B. Chu, ACS Appl. Mater. Interfaces, 2024, 16, 48395–48405.
141. P. Twyford, C. Cai and S. Fried, J. Neural Eng., 2014, 11, 025001.
142. T. Guo, D. Tsai, C. Y. Yang, A. Al Abed, P. Twyford, S. I. Fried, J. W. Morley, G. J. Suaning, S. Dokos and N. H. Lovell, Front. Neurosci., 2019, 13, 413.
143. M. Im, P. Werginz and S. I. Fried, J. Neural Eng., 2018, 15, 036010.
144. J. Ahn, S. Cha, K. E. Choi, S. W. Kim, Y. Yoo and Y. S. Goo, Front. Cell. Neurosci., 2022, 16, 889663.
145. Y. Chen, F. Fang and N. Zhang, npj 2D Mater. Appl., 2024, 8, 17.
146. N. Rabiee, Trends Anal. Chem., 2025, 189, 118279.
147. W. Yang, S.-H. Jo, Y. Tang, J. Park, S. G. Ji, S. H. Cho, Y. Hong, D.-H. Kim, J. Park, E. Yoon, H. Zhou, S.-J. Woo, H. Kim, H. J. Yun, Y. S. Lee, J. Y. Kim, B. Hu and T.-W. Lee, Adv. Mater., 2023, 35, 2304533.
148. N. Zheng, T. Liao, C. Zhang, Z. Zhang, S. Yan, X. Xi, F. Ruan, C. Yang, Q. Zhao, W. Deng, J. Huang, Z. T. Huang, Z. F. Chen, X. Wang, Q. Qu, Z. Zuo and C. He, Adv. Sci., 2024, 11, e2406343.
149. D. Kim, J.-W. Lee, S. Kang, W. Hong, J. Lee, H. J. Kwon, D. J. Kang, L. Lee and H. Kang, Adv. Sci., 2025, 12, e2411725.

---