GaN nanowire arrays were grown on planar Si (111) substrates using plasma-assisted molecular beam epitaxy (MBE). The process began with annealing the Si (111) substrates to eliminate any residual oxide. For n-type and p-type doping, Si and Mg were employed, respectively. The Mg-doping concentration was varied by regulating the Mg effusion cell temperatures (T_Mg). Samples p-n GaN, p⁺-n GaN, and p⁺⁺-n GaN nanowires correspond to Mg fluxes in the range of 1 × 10^-10-1 × 10^-9 Torr.

The p-n GaN nanowires sample was first cleaned with acetone and deionized water to eliminate organic contaminants before photodeposition. Subsequently, the cleaned sample was immersed in a solution containing 8 mL of deionized water, 2 mL of NaIO₃ solution (0.2 M L⁻¹), and 30 µL of CoNO₃ (0.2 M L⁻¹), and 30 µL of NiNO₃ (0.2 M L⁻¹) solutions. The sample was then illuminated for 30 min using a Hg lamp (Tanon UV-100).

Initially, the as-grown GaN nanowire samples were divided into multiple pieces of the desired dimensions. To establish ohmic contacts, an In-Ga eutectic alloy (Alfa Aesar) was applied to the Si substrate following the removal of the SiO₂ layer. The coated region was then connected to a copper sheet using silver paste (SPI Supplies). Finally, epoxy was employed to encapsulate the nanowires for insulation and packaging, leaving the front-facing nanowires exposed for photodetection.

The as-fabricated photoelectrode, counter electrode (Pt mesh), and reference electrode (Ag/AgCl) were immersed in an electrolyte solution of 0.01 M PBS. The incident light for the measurements was generated by LEDs with wavelengths ranging from 265 to 365 nm, and the corresponding light intensities were calibrated using an optical power meter (Newport Model No. 2936 R). For the demonstration of glucose sensing, deionized water, and a glucose standard solution were chosen as the electrolytes. The corresponding photoresponse performance of our devices was evaluated by an electrochemical workstation (CHI 660E). The corresponding average photocurrent was obtained by integrating the current over time.

The morphologies of the GaN nanowire samples were characterized using a scanning electron microscope (Hitachi SU8220 system, operating at 3 kV) and a transmission electron microscope (JEOL-2100F system, operating at 200 kV). Elemental mapping via energy-dispersive X-ray spectroscopy was performed using a 26FEI Talos F200X device at 200 kV. High-resolution transmission electron microscopy (HRTEM) results were obtained using a JEM-ARM 200 F instrument at 200 kV. X-ray photoelectron spectroscopy (XPS) measurements were conducted on a Thermo Scientific K-Alpha XPS instrument equipped with an Al Kα source (hν = 1253.6 eV), operating at 15 kV. TRPL spectra were measured by the TPL-300 (Time-Tech Spectra) with a 230-nm pulsed laser. Kelvin probe force microscopy (KPFM) measurements were conducted using an atomic force microscope (AFM) (NTEGRA SPECTRA, NT-MDT). The probe employed in the measurements is a PtIr-coated tip. Monochromatic lights were provided by a xenon lamp equipped with a monochromator (IHR320, Horiba JY). The detailed test method followed the previous work [23].

The construction of a high responsivity and stable photoanode is crucial in the pursuit of PEC-type photosensors across a wide range of applications [24]. The n-type semiconductors are usually used as photoanodes in PEC-type photosensors, where the photodetection process encompasses charge generation, separation, and charge transfer in the PEC reactions. Specifically, as shown in Fig. 1a and c, when the n-GaN nanowire is in contact with the electrolyte, the surface band bends upward at equilibrium [25]. Upon illumination, photogenerated holes migrate toward the nanowire/electrolyte interface, driving the water oxidation reaction (i.e., oxygen evolution reaction, OER). Electrons migrate to the external circuit, finally contributing to the generation of photocurrent. From the perspective of semiconductor band structure engineering, the band bending degree of the n-GaN surface determines the carrier separation capability of the whole nanowire [26]. However, the carrier separation within mono-doped (e.g., n-type) semiconductors is usually inefficient, leading to limited overall device efficiency [27]. Here, to address this issue, we propose and design a GaN p-n junction that can pump photogenerated holes transporting toward the nanowire surface under light illumination. Specifically, as shown in Fig. 1b, a GaN p-n junction can be formed by p-type doping for the top segment of the nanowire while maintaining the bottom segment as n-type doping. As depicted in Fig. 1d, the formation of the p-n junction within the nanowire could generate a significant built-in electric field, leading to a large band bending at the p-GaN/n-GaN interface, which acts as a strong driving force for the separation of photogenerated carriers [26].

To verify the above design, we conducted open-circuit potential (OCP) tests to investigate free carrier dynamics and measure the photovoltage of the n-GaN and p-n GaN nanowires, which were grown under the same conditions except for the p-doping of the upper segment of p-n GaN nanowires (Fig. S1a and b). The difference between OCP under dark/light conditions (ΔOCP) is usually used to characterize the photovoltage (V_ph). As depicted in Fig. S1c, the OCP of the n-GaN nanowires shifts to negative potentials under illumination, confirming its electron-conducting n-type properties [28]. Similarly, the OCP of the p-n GaN nanowires also exhibits a negative shift, indicating that the dominant holes migrate toward the surface in the p-n GaN nanowires, consistent with that of the n-GaN nanowires. Importantly, as shown in Fig. 1e, the measured photovoltage of p-n GaN nanowires (0.49 V) increases by 172% compared to that of n-GaN nanowires (0.18 V), consistent with the band design in Fig. 1a-d. Essentially, the p-n GaN nanowires exhibit a larger photovoltage compared to n-GaN nanowires, contributing to a stronger driving force for carrier separation and subsequent photocurrent generation.

As illustrated above, by introducing the p-GaN segment to form a p-n junction, the photovoltage can be prominently improved with the photogenerated holes effectively pumped toward the nanowire surface. However, the downward surface band bending of the p-GaN segment notably forms a potential barrier that might hinder the migration of photogenerated holes toward the electrolyte [29]. Meanwhile, since the downward surface band bending of the p-GaN segment in contact with the electrolyte might increase with the rise of the p-doping concentration level [30], we attempted to employ a doping modulation scheme to optimize the surface band structure of the p-GaN segment, trying to maximize the hole migration efficiency to the electrolyte. As illustrated in Fig. 1f, when a high p-doping concentration is applied, the significant downward bending of the surface band may create a high potential barrier, impeding the migration of photogenerated holes to the electrolyte. This might result in a small photocurrent. Conversely, at lower p-doping levels, the potential barrier may be substantially reduced due to a moderate band bending, allowing photogenerated holes to effectively migrate to the electrolyte and then engage in reactions, thereby generating a larger photocurrent.

To verify the influence of p-doping on the surface band of the GaN nanowire and its corresponding photoresponse, three types of p-n GaN nanowires were grown on silicon substrates using the molecular beam epitaxy (MBE) technique, with varied p-doping levels by employing different concentrations of p-type dopants during the growth of the p-GaN segment. Si and Mg were employed as donor and acceptor dopants, respectively, for n-type and p-type doping (details see Experimental Section). The p-n GaN nanowires were denoted as p-n GaN, p⁺-n GaN, and p⁺⁺-n GaN nanowires, corresponding to the nanowires with incremental Mg-doping concentrations.

Scanning electron microscope (SEM) images of the p-n GaN nanowires are shown in Fig. S2. These images display a uniform distribution of the nanowires on Si substrate, exhibiting nearly consistent diameters and lengths of the three types of nanowires with different Mg-doping concentrations, and their filling factors are very close at 0.76, 0.77, and 0.80 for the p-n GaN, p⁺-n GaN, and p⁺⁺-n GaN nanowires, respectively. To confirm the formation of p-n GaN nanowires with varied p-doping levels and assess the impact of different p-doping levels on the surface band bending, we conducted Mott-Schottky tests [31]. As shown in Fig. 2a, the Mott-Schottky curves reveal a similar inverted “V-shape” with distinct slopes. The positive slope of the curve corresponds to the n-type conductivity of the material, while the negative slope indicates the presence of p-GaN [32]. The Mott-Schottky diagram provides valuable information by analyzing the linear region of the fitted curve, where the slope reflects the relative magnitude of the donor concentration, and the intercept represents the flat-band potential [33]. The magnitude of the positive slope in the Mott-Schottky curves directly reflects the relative donor concentration in the material, with smaller slope values indicating higher donor concentrations. As a result, the donor concentration is the highest in p-n GaN nanowires, while p⁺⁺-n GaN nanowires have the lowest donor concentration. This can be attributed to the compensation of the background n-type dopants by the Mg acceptor, and the compensation effect becomes larger with increasing p-doping concentration [25]. Significantly, the flat-band potential of p-n GaN nanowires (− 0.94 V vs. Ag/AgCl) exhibits a positive shift relative to p⁺-n GaN (− 1.39 V vs. Ag/AgCl) and p⁺⁺-n GaN nanowires (− 1.44 V vs. Ag/AgCl), indicating a decrease in the bending of p-GaN segment band edges [34].

To further investigate the impact of different p-doping levels on the surface band bending, we carried out contact potential difference (CPD) measurements using Kelvin probe force microscopy (KPFM) in air to directly probe the work function level. Figure S3a-c shows the surface potential scans in a 2 × 2 μm² region of three types of p-n GaN nanowires under dark conditions. The average CPD values for p-n GaN, p⁺-n GaN, and p⁺⁺-n GaN nanowires are 268.1, 97.6, and − 175.1 mV, respectively, as depicted in Fig. 2b (extracted from Fig. S3a-c). Here, the CPD is defined as (Φ_tip −  Φ_surface)/ −  e, where Φ_tip and Φ_surface represent the work functions of the Kelvin probe tip and the nanowire surface, respectively, and e denotes the elementary charge [35]. The Φ_surface can be calculated by the function: Φ_surface =  eV_CPD +  Φ_tip. As the p-doping concentration rises, the work function of the p-n GaN nanowires progressively increases, resulting in a closer alignment of the Fermi level with the valence band. Furthermore, to characterize the location of the valence band maximum relative to the Fermi level, XPS was employed to measure the valence band spectra of the nanowires, and the corresponding results are depicted in Fig. S4. In Fig. 2b, the relative positions of the valence band maxima in p-n GaN, p⁺-n GaN, and p⁺⁺-n GaN nanowires were determined as 1.53, 1.36, and 1.12 eV, respectively, by extrapolating the linear region of the XPS valence band spectra in Fig. S4. Based on the analysis of CPD and XPS valence band spectra, the surface Fermi level of the p-GaN segment decreases as the p-doping concentration rises, causing an increase in the downward band bending at equilibrium in contact with the electrolyte, which is consistent with the Mott-Schottky results. As mentioned above, the downward bending of the surface band forms a potential barrier that hinders the migration of photogenerated holes to the electrolyte. Hence, by reducing the surface band bending at the p-GaN segment, it might decrease the hole transport resistance and thereby enhance the photoresponse.

Then, electrochemical impedance spectroscopy (EIS) is used to investigate the electrical properties of the photoelectrode, which reflect the transfer efficiency of charge carriers in the photoelectrode. The previous research has consistently demonstrated that a smaller arc radius in the Nyquist plots is indicative of a lower charge transfer resistance (R_ct) within the photoelectrode [23]. Therefore, we conducted EIS tests under operating conditions (0 V vs. Ag/AgCl, 340-nm light irradiation). The p-n GaN nanowires exhibit the smallest semicircle diameter (Fig. 2c), confirming the presence of a minimal charge transfer barrier. Furthermore, tests were conducted at various operating voltages and fitted using the equivalent circuit illustrated in the inset of Fig. 2c. The fitted parameters are presented in Fig. 2d. R_s, R_bulk, and R_ct correspond to the solution resistance, bulk resistance of the material, and charge transfer resistance, respectively. The R_s values of the three types of p-n GaN nanowires are nearly identical, indicating the validity of the fitting results. The R_bulk shows a tendency to increase with the rise of p-doping concentration. This is reasonable because the effective donor concentration of nanowires decreases with increasing p-doping concentration, leading to overall conductivity decreases [36], which is consistent with the Mott-Schottky results. Importantly, the R_ct values increase with the rise of p-doping concentration. This result verifies that moderate downward surface band bending can reduce the charge transfer resistance, facilitating the migration of photogenerated holes to the electrolyte. Notably, the relations of the resistances among the three photoelectrodes are consistent under different voltages.

To further verify the photoresponse determined by the different surface bands, we evaluated the photocurrents of the photoelectrodes through I-t tests, as illustrated in Fig. 2e. The photocurrent decreases with increasing p-doping concentration, confirming that the severe surface band bending hinders the transfer of holes into the electrolyte, leading to a small photoresponse. Furthermore, we conducted surface photovoltage (SPV) tests using KPFM in air to directly examine the carrier migration behavior. Typically, SPV measurements enable the assessment of the migration capability of photogenerated carriers toward the nanowire surface, and a positive SPV indicates the dominant migration of photogenerated holes toward the nanowire surface [5]. Figure S3 displays the CPD mappings under both dark and 340-nm light irradiation. The SPV is calculated as the difference between the average CPD values measured under illuminated and dark conditions. The SPV value, extracted from Fig. S3 and displayed in Fig. 2f, decreases with increasing p-doping concentration. Thus, the SPV results are consistent with the i-t results, further supporting that moderate surface band bending promotes the migration of photogenerated holes to the nanowire surface, consequently leading to a high photoresponse.

Based on the preceding discussion, we have confirmed that a rational design of p-n junctions with proper p-doping promotes the migration of holes to the nanowire surface while optimizing the transfer of holes into the electrolyte, thereby enhancing the overall performance of the photoelectrode. However, as shown in Fig. 3a, the previous investigations have revealed that the surface states of pristine GaN nanowires can trap photogenerated carriers, thereby impeding surface photoelectrochemical reactions [37]. Additionally, the scarcity of active sites on the pristine GaN surface imposes limitations on the surface reaction rate, hindering the rapid consumption of the photogenerated carriers [38]. Consequently, carrier accumulation on the nanowire surface may compromise device performance [24]. As shown in Fig. S5, the as-constructed p-n GaN photoelectrode was tested under chopped light. During the test, the photocurrent exhibits a gradual decline, ultimately decreasing by 40.5% after 6200 s of continuous testing, suggesting a persistent deterioration in device performance. To improve the consumption of photogenerated carriers and enhance the stability of the device during long-term operation, surface modification for nanowires is highly desirable. The CoNiO_x is a bimetallic oxide known for its exceptional OER catalytic activity and is commonly employed as a hole transport layer to effectively enhance carrier extraction [39]. Herein, CoNiO_x co-catalyst was deposited on the p-n GaN nanowires by using a photodeposition process (details see Experimental Section).

The morphology of the surface-modified p-n GaN nanowires was analyzed using SEM. Figure 3b shows that the nanowire array remains intact after surface modification. Transmission electron microscopy (TEM) was then employed to further characterize the distribution and microstructure of the surface-modified nanowires. As depicted in Fig. 3c, two overlapped nanowires are discernible, with their contours delineated by red and green dotted lines, respectively. The high-resolution transmission electron microscopy (HRTEM) image (Figs. 3d and S6) demonstrates the presence of an amorphous layer enveloping the nanowires, clearly outlined by the white dotted lines. The energy-dispersive spectroscopy (EDS) elemental mapping (Fig. 3e) performed in the corresponding region illustrates the spatial distribution of Ga, N, O, Co, and Ni elements. These results display the distribution of Co, Ni, and O elements around each nanowire, indicating the presence of CoNiO_x species on the surface of nanowires. Furthermore, the chemical state of the decorated Co and Ni species was characterized using XPS. The Co 2p spectrum (Fig. 3f) and Ni 2p spectrum (Fig. 3g) exhibit characteristic doublets consisting of 2p_3/2 and 2p_1/2 peaks due to spin-orbit coupling, along with shakeup satellite peaks. The characteristic peaks of Co 2p_3/2 and Co 2p_1/2, observed at 781.5 and 797.3 eV in Fig. 3f, respectively, along with their respective satellite peaks (786.1/802.9 eV), indicate the presence of Co²⁺ species [40]. Similarly, Ni 2p_3/2 and Ni 2p_1/2, observed at 855.9 and 873.6 eV in Fig. 3g, respectively, along with their respective satellite peaks (861.8/879.8 eV), confirm the presence of Ni²⁺ species [41]. Combined with the above analysis, we confirmed the successful coating of p-n GaN nanowires with an amorphous layer of Co²⁺Ni²⁺ oxides (referred to as p-n GaN/CoNiO_x nanowires).

To investigate the impact of CoNiO_x on photogenerated carrier dynamics, we conducted time-resolved photoluminescence (TRPL) spectroscopy on both p-n GaN and p-n GaN/CoNiO_x nanowires (Fig. 4a). The emission decay curves were fitted with double-exponential decay functions [42], and the detailed parameters for TRPL lifetime are found in Table S1. The average decay time constant (τ_ave) of the pristine p-n GaN nanowires is 0.387 ns, while the p-n GaN/CoNiO_x nanowires demonstrate a longer τ_ave of 0.875 ns, suggesting a slower photoluminescence decay rate compared to the pristine p-n GaN nanowires. This implies that CoNiO_x decoration suppresses photogenerated carrier recombination and enhances charge separation which might be attributed to the passivation of surface trap states [43]. Furthermore, we conducted OCP tests of p-n GaN nanowires and p-n GaN/CoNiO_x nanowires (Fig. 4b). Remarkably, due to a slow relaxation process, the OCP of p-n GaN nanowires exhibits a slow recovery process with a recovery rate of 57.4%, confirming the trapping effect of surface states on the pristine GaN nanowires [44], which is consistent with the results of TRPL. In contrast, the OCP of p-n GaN/CoNiO_x nanowires almost fully recovers to its initial level within 80 s after light off, reaching a recovery rate of 94.7%, demonstrating that the CoNiO_x decoration might effectively passivate the surface state, eliminate the trapping effect, and facilitate fast carrier transfer. In addition, the SPV was assessed using KPFM. Figure S7 illustrates the CPD of p-n GaN/CoNiO_x nanowires under dark and 340 nm light irradiation, respectively. As shown in Fig. 4c, the SPV of p-n GaN/CoNiO_x nanowires (766.8 mV) was significantly increased by 110.7% compared with the SPV of pristine p-n GaN nanowires (363.9 mV), indicating that the surface modification effectively enhances the photogenerated carrier separation and allows more photogenerated holes to migrate to the nanowire surface. In short, the TRPL, OCP, and SPV measurements reveal the exceptional capability of CoNiO_x as a hole transport layer in facilitating rapid charge extraction. Meanwhile, the possible suppression of carrier trapping could enable rapid carrier migration to the electrolyte to participate in the reaction instead of accumulating on the nanowire surface, thereby improving the photodetection behavior.

The photodetection performance was initially assessed through i-t tests (340 nm, 0.1 mW cm⁻²). Figure 4d illustrates a significant enhancement in the photocurrent of p-n GaN/CoNiO_x nanowires (24.8 μA cm⁻²), increasing by 101.6% compared to pristine p-n GaN nanowires (12.3 μA cm⁻²). Subsequently, the spectral response was assessed by illuminating the samples with light ranging from 265 to 365 nm (Fig. 4e), revealing the improved performance of p-n GaN/CoNiO_x nanowires across the full spectrum. Notably, the p-n GaN/CoNiO_x nanowires exhibit an outstanding responsivity of 247.8 mA W⁻¹ at 340 nm. Then, the photocurrent of the p-n GaN/CoNiO_x nanowires was measured under various incident light intensities, as recorded in Fig. S8 and summarized in Fig. 4f. The photocurrent exhibits a near-linear relationship with the increase in light intensity, indicating a great linear photoresponse. The photocurrent of the p-n GaN/CoNiO_x nanowires was measured under different bias voltages using chopped LSV, as recorded in Fig. S9. Photocurrent gradually increased with the potential shift from − 0.15 to 0.55 V vs. Ag/AgCl. This is reasonable because the positive potential accelerates the transfer of holes to the nanowires’ surface. Moreover, surface reactions are more favorable under a more positive bias, ultimately contributing to a larger photocurrent. To unveil the photoresponse speed of the photosensor, the response time (t_res) and recovery time (t_rec) were evaluated and exhibited in Fig. S10. The device demonstrates a fast photoresponse speed with a response time (t_res) of 3 ms and a recovery time (t_rec) of 0.7 ms. Furthermore, the excellent stability of the p-n GaN/CoNiO_x nanowires was demonstrated by the long-term test of the photoelectrode, lasting for 27.5 h (Fig. 4g). Strikingly, no significant attenuation was observed during the test. The insets of Fig. 4g reveal that the photocurrent remains at 90% of its initial value, with a slight decrease from 0.136 to 0.123 μA. Notably, our device is constituted by the PEC photoanode. And the photoanode is usually unstable during operation due to sluggish dynamics and photocorrosion. Here, we demonstrated a stable PEC photoanode by decorating a layer of CoNiO_x on nanowires to improve device stability. This remarkable stability can be attributed to the CoNiO_x decoration on the nanowires, which acts as a physical protective layer while rapidly consuming the carriers and effectively protecting the nanowires from photocorrosion caused by carrier accumulation [45]. Figure 4h displays the comparison of our device performance with recently reported self-powered PEC-type photosensors, and the responsivity and stability of these devices are summarized in Table 1. Notably, in Table 1, we also summarized and included different GaN nanowires-based photoelectrochemical-type photosensors. Different types of GaN nanowires in the form of heterojunction with other materials have been formed and reported, such as GaN/AlGaN [46], GaN/Pt [47], GaN/Cu [8], GaN/Cu₂O [48], GaN/AuNC [49], GaN/CsPbBr₃ [50], and GaN/Ga₂O₃ [51]. Impressively, our device exhibits a high responsivity with remarkable device stability which is superior to that of other reported PEC-type photosensors, demonstrating its great potential for future sensitive, sustainable optoelectronic sensing systems.

From the above discussion, we have demonstrated the remarkable photodetection performance of the proposed photoelectrode composed of p-n GaN/CoNiO_x nanowires, featuring ultrahigh responsivity and long-term operational stability. In essence, the PEC photodetection process combines conventional semiconductor physical processes with chemical reactions occurring in the electrolyte. These chemical reaction processes encompass the conversion and generation of substances, offering a unique opportunity for the precise identification of chemical and biological substances [21]. Glucose, a fundamental energy source for cells, plays a crucial role in various physiological processes. The accurate and reliable quantitative determination of glucose is of great importance in clinical and biomedical applications [58]. Thus, to demonstrate the practical applicability of the devices, we constructed a PEC glucose sensor based on the p-n GaN/CoNiO_x photoelectrode as a proof-of-concept application, which is schematically depicted in Fig. 5a. Essentially, CoNiO_x possesses excellent catalytic properties and serves as a glucose biorecognition unit that may selectively oxidize glucose [59,60]. We first conducted the measurement using pristine p-n GaN nanowires. As depicted in Fig. S11, the photocurrent illustrates almost no change by the addition of glucose compared to that observed in the pure electrolyte. Then, we conducted the measurement using p-n GaN/CoNiO_x nanowires (Fig. 5a). Basically, when the photoelectrode was illuminated, electron-hole pairs were generated in the nanowires. Photogenerated holes transferred to the electrolyte and effectively oxidized glucose, converting it into glucolactone [61,62]. The photoresponse can be enhanced by the addition of glucose compared to that observed in the pure electrolyte (deionized water), confirming the catalytic role of CoNiO_x in the glucose detection process. Then, the correlation between glucose concentration and photoresponse can be calibrated, serving as a foundation for the detection of glucose levels.

Figure 5b illustrates the relationship between the photocurrent and the concentration of glucose injected into the electrolyte. Glucose was continuously injected into the electrolyte while the solution was stirred, resulting in an enhancement of the device photoresponse with increasing glucose concentration. A linear correlation was established between the photocurrent and glucose concentration (Fig. 5c), and the regression equation (I in µA cm⁻² = 6.226 + 0.173 C_glucose in µmol L⁻¹) was derived. The linear relationship displays a high correlation coefficient (R²) of 0.997. Notably, the device demonstrates a remarkable sensitivity with a value of 0.173 µA µM⁻¹ cm⁻², along with a low detection limit (LOD) of 0.07 µM (S/N = 3). These results establish a solid foundation for the quantitative determination and detection of glucose. The glucose sensing performance of the p-n GaN/CoNiO_x was compared with those of reported glucose sensors and summarized in Table S2. Compared with other sensors, our device exhibits higher sensitivity and lower detection limit in the detection of glucose. These are attributed to remarkable device stability and excellent glucose catalytic activity. Moreover, compared with most sensors, our device operates under self-powered conditions, which exhibit greater application potential. Moreover, to evaluate the reproducibility, five p-n GaN/CoNiO_x photoelectrodes were prepared using identical conditions, yielding a calculated relative standard deviation (RSD) of 3.27% (Fig. 5d). Additionally, to assess the long-term stability of the PEC sensing system, the constructed photoelectrode was stored under dry conditions at room temperature. The photoresponse was recorded every 2 days, revealing that the photoelectrode maintained 91% of its initial photoresponse over a 20-day storage period (Fig. 5e). In Fig. S12, SEM images exhibit the morphology of the freshly fabricated p-n GaN/CoNiO_x nanowires and the morphology after undergoing 20 days of stability testing. Impressively, the nanowire structure remains nearly unchanged, demonstrating excellent corrosion resistance. Furthermore, we also assessed the selectivity of the glucose sensor. In real glucose sensing conditions, there are other compounds such as fructose, lactose, maltose, urea, and uric acid [63]. The presence of these coexisting compounds can introduce interference and impact the accuracy of glucose sensing. In Fig. S13, we initially added glucose to the electrolyte and observed an increase in current. Subsequently, we sequentially added fructose, lactose, maltose, urea, and uric acid to the solution, but no significant change in current was observed. When glucose was reintroduced to the mixed solution, the current once again exhibited a prominent increase. Figure 5f displays the currents obtained when adding different compounds, as well as the current difference (∆Current) each time a new compound is added (data are extracted from Fig. S13). The ∆Current caused by glucose is significantly greater than those of other interfering compounds at equivalent concentrations. Hence, the interference from these coexisting compounds is negligible, indicating the high selectivity of glucose sensing and verifying the selective catalytic effect of CoNiO_x on glucose detection. Finally, the constructed sensor was utilized to examine the glucose levels in human serum. Three human serum samples were obtained, and without any pretreatment, only the serum was diluted in electrolytes before testing. The glucose concentrations measured are shown in Table 2, which are acceptable compared with the values measured at the local hospital. These results provide compelling evidence for the reliability of the fabricated PEC sensor in accurately detecting glucose levels in real human blood serum, demonstrating its practical potential for PEC biosensing applications.