Hydrovoltaic-Driven Self-Powered Sensors for IoT and Healthcare Applications: A Review

The mechanism of hydrovoltaic sensor is governed by humidity dependent water adsorption, hydration, and ion transport within the active material. In general, the sensing layer is designed to contain abundant hydrophilic functional groups, such as hydroxyl, carboxyl, or amine groups, which can interact with water molecules through hydrogen bonding and facilitate ion dissociation and transport. When the surrounding RH increases, water molecules are progressively adsorbed onto the surface and penetrate into the internal micro/nanochannels of the active layer. This hydration process can induce swelling, enhance ionic dissociation, and ion migration, thereby altering the local ion concentration, interfacial potential, and EDL structure.

These changes directly modulate the spontaneous direct current (DC) output of hydrovoltaic devices. At low humidity, limited water adsorption restricts ion dissociation and charge transport, resulting in a relatively weak voltage or current response. As humidity increases, the expanded hydration network provides continuous pathways for ion migration and strengthens charge separation at the solid-liquid interface. Consequently, the generated voltage and current can increase systematically with RH that can be converted into measurable electrical signals without an external power supply. By calibrating the relationship between RH and electrical response, hydrovoltaic devices can function as self-powered humidity sensors [57-59].

Lei et al. proposed a sandwich structure potentiometric humidity sensor (Fig. 2(a))[60]. The device was fabricated with reduced graphene oxide (rGO), graphene oxide (GO), and foamed metal nickel (Ni_fm). The abundant oxygen-rich functional groups in the GO layer facilitate hydration. Density functional theory (DFT) calculations showed that hydration in GO significantly reduces the adsorption energy of protons, thereby enhancing H⁺ mobility and ionic conductivity. In addition, asymmetrical architecture of rGO/GO/Ni_fm induces electrical potential difference, which can serve as a galvanic cell altering solid and liquid electrolytes.

Fig. 2.

Hydrovoltaic humidity sensors. (a) Schematic of sandwich structure potentiometric humidity sensor. (b) Sensing performance of the sensor under humid conditions. Adapted from Ref. [60].

As a result, the device generated an open-circuit voltage (V_OC) of up to 0.77 V at 90% RH and exhibited rapid response and recovery times of 0.8 and 2.4 s, respectively (Fig. 2(b)). The sensor also showed linear sensitivities of 5 mV/% RH in the 20-70% RH range and 20 mV/% RH in the 70-90% RH range. Moreover, the device maintained a stable voltage response over more than 1500 humidity cycles and high humidity exposure for over 600 s. These results demonstrate that the asymmetric rGO/GO/Ni_fm architecture can serve as a self-powered platform for humidity monitoring and respiration-related moisture sensing.

Li et al. introduced a polydopamine (PDA) based moisture power generator as a humidity-responsive platform for self-powered wearable sensing [61]. A porous PDA film with micropores of approximately 1 μm in average diameter was formed on a PET substrate by spray printing, and a gradient PDA (g-PDA) structure with an asymmetric distribution of hydroxyl groups was subsequently introduced through electrochemical polarization. Upon moisture absorption, the oxygen-rich functional groups in the g-PDA film undergo ionization, generating an asymmetric ion distribution between mobile H⁺ ions and locally confined anions bound to the PDA skeleton. The hydroxyl group gradient promotes directional H⁺ migration, resulting in a potential difference and voltage generation.

Based on this mechanism, the g-PDA device exhibited rapid moisture response and high output performance. When the RH changed from 5% to 55%, the voltage reached a stable state within 0.22 s and recovered within 0.26 s after moisture removal. In addition, the output voltage increased from approximately 0.14 V to 0.52 V as ΔRH increased from 30% to 90%, demonstrating its capability for self-powered humidity sensing. The device also retained approximately 98% of its initial voltage after 1000 bending cycles at 120˚, confirming its mechanical durability. These results suggest that g-PDA based hydrovoltaic devices can serve as flexible and self-powered wearable humidity sensors capable of monitoring local moisture variations at the skin-device interface.

The fundamental transduction mechanism of hydrovoltaic strain sensors is governed by the geometric deformation of internal conductive pathways and the subsequent variations in electrical resistance under physical elongation. For instance, when tensile stress is applied to an elastic fabric coated with carbon nanomaterials, the internal micro/nanochannels undergo elongation, accompanied by localized cracking and interfacial slippage. These structural alternations directly lead to an increase in the internal electrical resistance of the device, which scales proportionally with the applied strain. The elevated resistance intrinsically affects the DC power generation dynamics driven by water molecule diffusion and ion transport.

Specifically, mechanical stretching locally amplifies the moisture concentration gradient and expands the effective evaporation area, which can result in an augmented V_OC. Conversely, the short-circuit current (I_SC) exhibits an immediate decline due to its acute sensitivity to the increased internal resistance. By continuously monitoring these distinct variations in the spontaneously generated power, the magnitude of the applied mechanical strain can be accurately quantified, thereby enabling the device to function as a highly sensitive, self-powered strain sensors [62-64].

Luo et al. introduced a highly stretchable fiberform hydrovoltaic generator as a self-powered strain sensing platform (Fig. 3(a))[65]. The device, denoted as MC-FHPG, was fabricated by loading oxidized multiwalled carbon nanotubes (o-MWCNT) and oxidized carbon black (o-CB) onto discarded mask strap fibers. The mask strap substrate provides microscale capillary channels between yarns, while the o-MWCNT/o-CB network forms nanoscale water diffusion pathways and conductive networks on the fiber surface. When water is supplied to one end of the fiber, capillary driven water transport along the charged carbon surface induces an EDL at the solid-liquid interface and generates a spontaneous hydrovoltaic current. This current provides the electrical signal for strain sensing without an external power source.

Fig. 3.

Hydrovoltaic strain sensors. (a) Schematic illustration of MC-FHPG strain sensor. (b) Strain-dependent current response and sensing performance of MC-FHPG. Adapted from Ref. [65].

The strain sensing performance of the MC-FHPG was evaluated by monitoring the variation in the current under tensile deformation (Fig. 3(b)). The relative current response can be divided into three distinct linear regions across the 0-100% strain range, corresponding to gauge factors of -3.68, -0.93, and -0.14 for 0-15%, 15-45%, and 45-100% strain, respectively. This negative current originates from increase of the internal resistance induced by strain, which suppresses I_SC during stretching. Repeated stretching tests at 10%, 20%, and 40% strain produced reproducible current modulations with strain-dependent amplitudes, confirming the capability of the device to resolve different deformation levels. Moreover, the device exhibited fast dynamic response and recovery times of 108 and 129 ms under a transient strain of 1%, respectively. As a proof-of-concept demonstration, the MC-FHPG was mounted on a finger joint, where progressive finger bending from 15˚ to 90˚ produced a corresponding decrease in the output current. This result highlights the applicability of fiberform hydrovoltaic generators as a stretchable self-powered strain sensors for wearable motion tracking.

Wen et al. reported a self-gradient hydrogel-based moisture-induced electric generator (SHMEG) as a climate-adaptive stretchable platform for self-powered wearable sensing (Fig. 4(a))[66]. The device was fabricated by applying a pre-gel solution composed of polyacrylamide (PAM), glycerol, and NaCl onto a CB-loaded knitted fabric, followed by self-diffusion and thermal gelation to form a concentration-gradient hydrogel structure. The CB-loaded fabric provides a negatively charged and porous interface for ion adsorption, while the hydrogel maintains hygroscopicity, moisture retention, ionic conductivity, and mechanical stretchability. The spontaneous formation of ion and water concentration gradients establishes an EDL gradient at the hydrogel-carbon interface. This interfacial gradient promotes directional ion transport and enables sustained hydrovoltaic output.

Fig. 4.

Climate-adaptive hydrovoltaic strain sensors. (a) Fabrication and surface characterization of SHMEG (b) Climate-adaptive output and wearable sensing performance of the SHMEG. Adapted from Ref. [66].

Based on this mechanism, the SHMEG exhibited stable and climate-adaptive power generation performance (Fig. 4(b)). The device generated a sustained V_OC of approximately 0.75 V for 140 h and I_SC of 15 ̄μA at around 75% RH under room temperature. Due to the combined effects of glycerol and NaCl, the hydrogel achieved moisture retention and anti-freezing characteristics. In addition, its high stretchability of up to 300% enabled wearable sensing applications. During stretching and finger bending, the output current decreased with increasing internal resistance, allowing mechanical deformation to be tracked through the self-powered electrical signal. The device was further integrated with a wireless sensing system for real-time finger-motion monitoring, suggesting its potential as a climate-adaptive self-powered platform for wearable strain sensing.

Unlike strain sensors, which primarily quantify uniaxial tensile deformation, mechanical sensors can be designed to detect more complex stimuli, including pressure, bending, sliding, and tapping. This broader sensing capability is achieved by tailoring the device architecture so that mechanical inputs modulate liquid motion, water distribution, ion transport pathways, and interfacial contact. For example, pressure-induced deformation and displacement of droplets within confined microfluidic channels can alter the flow velocity and solid-liquid contact area, producing corresponding changes in the flow-induced voltage. These structural and functional design strategies enable mechanical stimuli with different intensity, frequency, and repetitive pattern to produce distinguishable self-generated electrical responses.

Hydrovoltaic pressure and motion sensing can therefore be broadly implemented through several strategies. Direct mechanical modulation of the liquid phase represents a primary strategy for hydrovoltaic pressure and motion sensing. In this configuration pressing, bending or sliding changes the position, shape, velocity, or contact area of liquid, resulting in interfacial charge redistribution and alter the generated voltage. Complex mechanical stimuli can be detected by coupling hydrovoltaic generation with an auxiliary transduction mechanism. Moisture-induced actuation can be monitored through the voltage generated during water redistribution, whereas mechanoluminescent materials convert pressure of sliding into optical signals that subsequently modulate the hydrovoltaic output. This approach enables hydrovoltaic devices to detect pressure and dynamic motion beyond conventional strain sensing [67-70].

Kong et al. presented a flexible self-powered pressure sensor based on graphene microfluidic channel (GMC) structure [71]. The device was fabricated by assembling monolayer graphene on PDMS and SiO₂ substrate to form a confined microchannel with inlet and outlet. Two graphene/PDMS layers and terminal electrodes defined the channel geometry, while the graphene surface provides a solid-liquid interface for droplet-based charge redistribution. When external pressure was applied, deformation of the flexible channel induced displacement and accelerated motion of the confined droplet along the graphene surface. This pressure-dependent droplet motion altered the charge distribution at the EDL and generated a corresponding flow-induced voltage, enabling the direct conversion of mechanical pressure into self-powered electrical signal.

Based on this mechanism, the GMC pressure sensor exhibited distinguishable voltage responses over a broad pressure range from 0.254 to 26.85 kPa. The relative voltage change showed two linear response regions, with sensitivities of 17.36 kPa^-1 in the low-pressure range of 0-7 kPa and 5.15 kPa^-1 in the high-pressure range of 13-26 kPa. The corresponding coefficients of determination were 0.997 and 0.991, respectively. Under loading and unloading at 2.138 kPa, the sensor exhibited response and recovery times of 36.4 and 239.2 ms. These results demonstrate that mechanically regulated droplet displacement within the flexible GMC can provide quantitative and rapid self-powered pressure sensing.

Zhang et al. presented a self-powered mechanoluminescent hydrovoltaic patch for multidimensional pressure and motion sensing [72]. The device consisted of a porous PAN/ZnS:Cu@TiO₂/Au active layer positioned between flexible PET films and carbon electrodes. Sweat or water vapor transported through the porous substrate generated a baseline hydrovoltaic output, while the mechanoluminescent layer converted externally applied bending, pressing, sliding and tapping into an additional optical stimulus. Under mechanical deformation, piezoelectric polarization in ZnS:Cu promoted carrier detrapping and green-light emission. The emitted light was subsequently absorbed by Au nanoparticles, generating energetic carriers that were transferred to TiO₂ and promoted photocatalytic reactions. This process disturbed the interfacial ion equilibrium and increased the hydrovoltaic potential, allowing mechanical stimuli to be converted into the voltage response.

Based on this coupled transduction mechanism, light stimulation increased both the voltage and current outputs of the patch, supporting the photo-assisted enhancement of hydrovoltaic generation. The mechanoluminescent-hydrovoltaic coupling enabled the patch to distinguish mechanical inputs through changes in signal amplitude and waveform to monitor motions at the neck, wrist, elbow, and knee. The device was further applied to Morse-code communication by assigning arm bending and finger tapping to distinct code elements. Voltage sequences corresponding to “HELP ME”, “DANGER”, and “SAFE” were successfully generated, demonstrating the potential of the self-powered wearable motion and pressure sensors.

Hydrovoltaic ion sensors utilize ion-dependent modulation of hydrovoltaic output in charged micro/nanochannels. In hydrovoltaic devices, a potential difference can be generated by asymmetric water transport, ion transport, and ion distribution at the solid-liquid interface. When an ion concentration gradient or asymmetric ion distribution is formed across the device, ion diffusion and migration contribute to charge separation, thereby modulating the generated voltage. In this process, micro/nanochannels serve as confined transport pathways that regulate both water flow and ion migration. Surface charges on the channel walls interact electrostatically with dissolved ions, leading to preferential transport of counterion and partial exclusion of coion. This ion selective transport can induce ion separation and alter the interfacial potential difference. Therefore, changes in ion concentration, ion species, valence, and ion transport resistance can be reflected in the generated V_OC and I_SC, enabling self-powered ion sensing.

In this context, the λ_D provides an important parameter for understanding ion sensing behavior. At low ion concentrations, λ_D is relatively large, and EDL overlap within confined nanochannels can be maintained, allowing strong ion selectivity and efficient charge separation. As the ion concentration increases, enhanced charge screening compresses the EDL and can weaken ion selectivity within the nanochannels. This reduced ion selectivity suppresses ion separation and interfacial potential formations, thereby decreasing the generated V_OC [75,77,78].

Wang et al. proposed a stretchable hydrovoltaic device utilizing silk fibroin (SF)-coated PU/PAN composite nanofibers (SF@PPN) for ion sensing (Fig. 6(a))[79]. In this device, mechanical deformation regulates the nanochannel size of SF@PPN over a wide range, from 150 nm to 1.25 μm. This strain-dependent modulation of nanochannel dimensions directly affects water flux and ion selectivity. In this system, the carboxyl and amide groups of SF formed hydrogen bonds with the PPN network. SF dip coating combined with oxygen plasma treatment shifted the zeta potential from -12 to -55 mV, indicating an increase in the negative surface charge of the channel walls. Mechanical strain induced fiber drafting and local densification of the SF layer, thereby regulating the pore dimensions relative to the λ_D. When the pore size became comparable or smaller than λ_D, EDL overlap was enhanced, promoting selective ion transport through the nanochannels results in the change of generated voltage.

Fig. 6.

Hydrovoltaic ion sensors. (a) Schematic illustration of the strain-programmable SF@PPN device. (b) Ion-selective transport mechanism and ion sensing performance of the SF@PPN sensor. Adapted from Ref. [79].

A decrease in nanochannel promotes EDL overlap and improves ion selectivity while simultaneously restricting water transport (Fig. 6(b)). Conversely, larger nanochannels allow higher water flux by reducing flow resistance but weaken EDL overlap and thus lower ion selectivity. The authors verified these characteristics through real-time diffusion experiments and V_OC-strain measurements, with in-situ XPS and TOF-SIMS analyses, thereby clarifying the correlation between nanochannel size, water flux, and ion selectivity.

The SF@PPN device also exhibited a broad ion concentration sensing ranging from 10^-7 to 10^-1 M for NaCl solutions. As the ion concentration increased, the real-time voltage curve showed a series of progressively lower plateaus, with V_OC decreasing from approximately 3.3 V to 83 mV. However, a single V_OC value provides limited information for distinguishing sensing responses associated with different ionic species and concentrations. To overcome this limitation, the authors constructed multidimensional ion-response surfaces using strain-dependent V_OC values obtained at different ion concentrations. A convolutional neural network (CNN) was subsequently applied to extract subtle features from these response surfaces and classify six ionic species, including N^a+, K⁺, Mg²⁺, Ca²⁺, Al³⁺, and Fe³⁺. In parallel, Frechet-distance-based analysis was employed to compare the strain-dependent response curve of an unknown sample with reference curves within the response-surface dataset, enabling the simultaneous identification of ion species and concentration. This approach demonstrates that converting a single hydrovoltaic output into multidimensional response patterns, followed by algorithm-assisted analysis, can distinguish coupled sensing information that cannot be readily resolved from a single voltage signal. Such multidimensional signal-processing strategies may also provide a useful framework for separating target-specific responses from variations induced by environmental parameters, such as humidity, temperature and evaporation conditions.

Ge et al. presented a SF-modified nylon-6,6 nanofiber (NNF) in which the nanochannel size was systematically regulated by varying the number of SF dip coating cycles, thereby controlling the effective fiber diameter and interfiber spacing (Fig. 7(a))[80]. The introduction of carboxyl and amide groups from SF shifted the zeta potential from -10.9 to -38.4 mV, indicating an increase in the negative surface charge of the nanochannel walls. This surface and structural modification provided an optimized balance between EDL overlap and evaporation-driven water transport. The optimized SF@NNF-based device generated a high V_OC of 4.82 V in deionized water and exhibited a broad NaCl concentration sensing range from 10^-7 to 10⁰ M, with a maximum sensitivity of 1.37 V dec^-1 (Fig. 7(b)). The concentration-dependent voltage response was attributed to changes in the λ_D and electrostatic screening within the nanochannels. As the ion concentration increased, enhanced charge screening reduced ion selectivity and produced a corresponding decrease in V_OC. Based on this response, the device was integrated into a wearable configuration to monitor electrolyte levels in human sweat in real time. The sensor also exhibited distinct voltage responses to LiCl, NaCl, and KCl, as well as NaCl, MgCl₂, and AlCl₃, indicating that ion size and valence influence charge screening and ion transport within the nanochannels. These characteristics further demonstrated the application of the SF@NNF hydrovoltaic device to trace ion monitoring in marine salt-spray environments.

Fig. 7.

Hydrovoltaic ion sensors. (a) Fabrication and nanochannel regulation of SF@NNF ion sensor. (b) Ion sensing voltage response of the SF@NNF device. Adapted from Ref. [80].

Hydrovoltaic gas sensors detect gaseous analytes through changes induced in the aqueous phase or at the hydrated solid-liquid interface. When a gas reaches the sensing layer, it dissolves in the adsorbed water film or supplied liquid can alter the pH, ionic composition, conductivity, or surface chemistry through acid-base reactions, redox reactions, or adsorption. These changes affect the protonation state, surface charge, zeta potential, EDL structure, and ion transport within the device. The resulting modulation of evaporation driven charge separation is reflected in changes in V_OC and I_SC. In several systems, direct adsorption of gas molecules onto the active material can also alter its electronic or surface properties and contribute to the sensing response.

The gas response is influenced by the solubility of the analyte, its reaction with the sensing surface, and its transport through the water layer. Porous and hydrophilic structures are generally employed to maintain water transport and provide sufficient interfacial area for gas interaction. Target-selective functional groups or receptors can improve both selectivity and sensitivity by promoting preferential analyte binding and enhancing the resulting modulation of the hydrovoltaic output. Depending on the device configuration, gaseous analytes can be detected directly at a hydrated sensing interface or first collected in water and subsequently analyzed through the resulting hydrovoltaic response.

Zhang et al. developed a photoelectrochemically enhanced hydrovoltaic formaldehyde (HCHO) sensor consisting of an NH₂-MIL-125/TiO₂ nanotube active layer and a porous cellulose fiber coated with PEDOT:PSS and rGO [81]. The negatively charged TiO₂ nanochannels supported evaporation driven proton transport and streaming potential generation, while the NH₂-MIL-125/TiO₂ heterostructure promoted photoinduced charge separation under illumination. The resulting photoelectrochemical water splitting reaction consumed interfacial H₃O⁺ during hydrogen evolution, perturbing the ionic equilibrium and enhancing the hydrovoltaic output. HCHO reacted with the amino groups of NH₂-MIL-125 through Schiff-base formation, altering the surface charge and photoelectrochemical activity and consequently decreasing both V_OC and I_SC.

The sensor exhibited a linear response to HCHO over the concentration range from 10^-8 to 10^-6 M, with a detection limit of 5.2 × 10^-9 M. The sensor showed negligible interference from methanol and formic acid, confirming its selectivity toward HCHO. The active layer also retained its morphology and crystal structure after exposure to HCHO-containing solutions, demonstrating finite structural stability. For practical analysis, HCHO released from filtered and unfiltered cigarettes was collected in water and quantified using the sensor. The unfiltered cigarette exhibited an HCHO concentration approximately twice that of the filtered cigarette, demonstrating the capability of the device to differentiate HCHO emission levels from practical gas sources.

Zhong et al. reported a rainwater evaporation-driven gas sensor based on a nanoporous carbon film deposited on an Al₂O₃ substrate with multiwalled carbon nanotube (MWCNT) electrodes [82]. The device was integrated with a water reservoir that collects rainwater and continuously supply it to the porous carbon layer through capillary transport. Evaporation driven water flow through the negatively charged nanochannels resulted in EDL formation and preferential transport of H₃O⁺ counterions, thereby generating a streaming potential between electrodes. Adsorption of gas molecules altered the zeta potential of the porous carbon, which subsequently modulated the ion distribution and hydrovoltaic output.

The sensor produced gas-dependent voltage response toward oxygen, ethanol, ammonia, methane, and carbon monoxide. The response to oxygen increased from approximately 5% at a concentration of 40% to 17.1% at 100%. In contrast, ethanol decreased the response by 3.27% at 200 ppm and 27.15% at 850 ppm, while ammonia produced decreased output of 7.23% at 50 ppm and 44.82% at 800 ppm. These gas-dependent responses enabled the recognition of different gas species and concentrations. The oxygen-dependent electrical output was further utilized for capacitor charging and LED operation, demonstrating a proof-of-concept visual indicator for real time air quality monitoring.

Molecular sensors selectively detect target analytes, including biomolecules, organic compounds, and other chemical species by converting specific molecular interactions into measurable electrical signals. In self-powered hydrovoltaic molecular sensors, water-mediated interfacial processes generate a spontaneous voltage and current. Specific interactions between target molecules and the sensing material can modulate this hydrovoltaic output, allowing the presence and concentration of molecular analytes to be determined without an external power source.

A molecular sensor generally consists of a recognition element and a transduction element. The recognition element selectively interacts with the target analyte through binding, adsorption, or chemical reactions and can include such as antibodies, enzymes, proteins, functional receptors, or metal-organic frameworks (MOFs). The transduction element, such as an electrode, semiconductor, conductive polymer, or porous carbon structure, converts the molecular recognition into a measurable signal. Interaction with the target molecule can alter the surface charge, interfacial potential, electrical conductivity, water adsorption, ion mobility, or mass of the sensing material. In hydrovoltaic molecular sensors, these physicochemical changes regulate interfacial charge separation and ion transport, resulting in analyte-dependent variations in the generated voltage and current.

Bian et al. constructed a self-powered electrochemical sensing platform based on a ZnO/Ce-MOF heterostructure [83]. Vertically aligned ZnO nanowire arrays were grown on a glass substrate and coated with amino-functionalized Ce-BTC to form a porous ZnO/Ce-MOF composite, after which a lincomycin (LIN)-specified DNA aptamer was immobilized on the surface. Upon contact with water, positively charged ZnO/Ce-MOF surface formed an EDL, while evaporation driven water transport through the nanowire and MOF nanochannels induced directional ion migration and generated a streaming potential. The vertically aligned ZnO arrays and porous Ce-MOF network provided continuous pathways for water and ion transport, thereby enhancing the hydrovoltaic output. In addition, the energy band alignment between ZnO and Ce-MOF generated a built-in electric field that promoted photocarrier separation under Xe lamp illumination. The resulting hydrovoltaic-photoelectric coupling increased the I_SC by approximately 10.9-fold compared with that generated solely by the hydrovoltaic effect.

Based on this enhanced electrical output, the ZnO/Ce-MOF platform was employed for the aptamer-based detection of LIN. Immobilization of the negatively charged aptamer reduced interfacial charge transfer, while subsequent formation of the aptamer-LIN complex introduced additional steric hinderance and further decreased I_SC. The sensor exhibited a linear response over the LIN concentration range from 10^-15 to 10^-9 M, with a detection limit of 0.2 fM. The non-target antibiotics, including norfloxacin, kanamycin, tetracycline, ciprofloxacin, and chloramphenicol, produced only negligible changes in I_SC. By contrast, a marked decrease in I_SC was observed for both LIN and the antibiotic mixture containing LIN, demonstrating the high selectivity of the aptamer-functionalized sensor toward LIN.

Zhang et al. reported an enantioselective hydrovoltaic sensing device based on a porous carbon film (PCF) deposited on a Kapton substrate [84]. The PCF consisted of randomly stacked carbon nanoparticles with diameters of approximately 50 nm, forming interconnected nanochannels with characteristic sizes of 100-200 nm. These porous channels supported capillary water transport and evaporation-induced streaming potential generation, enabling continuous electrical output without an external power source. The PCF surface was subsequently functionalized through sequential modification with (3-aminopropyl)triethoxysilane (APTES), glutaraldehyde (GA), bovine serum albumin (BSA), and o-phenylenediamine (OPD). BSA provided a chiral recognition environment with different binding affinities for L-arginine (L-Arg) and D-arginine (D-Arg), while OPD participated in the subsequent chemical amplification process. Following Arg adsorption, H₂O₂ treatment induced NO formation, which reacted with the amino groups of OPD to form benzotriazole derivates. This reaction reduced the positive surface charge of the sensing layer and modified the EDL structure and H₃O⁺ transport assisted with water flow. The difference in surface coverage between L-Arg and D-Arg produced distinct changes in surface charge, thereby enhancing the recognition through the hydrovoltaic voltage response.

Based on this recognition and amplification mechanism, the sensor generated distinguishable voltage responses to the two arginine enantiomers. Before chemical amplification, the difference in output voltage between L-Arg and D-Arg was approximately 0.13 V. Following the H₂O₂/OPD reaction, this voltage difference increased to 0.32 V, corresponding to an approximately 2.5-fold enhancement in enantiomeric recognition. The response to L-Arg reached approximately 68%, compared with 26% for D-Arg, while the responses to the other tested chiral interferents remained substantially lower than that of L-Arg. In addition, five independently prepared sensors exhibited closely distributed response values of approximately 7% under the same measurement condition, indicating finite device reproducibility. These results demonstrate that the combination of BSA mediated chiral recognition and surface charge amplification enable the selective conversion of enantiomer adsorption into measurable hydrovoltaic voltage signals.