Chloroauric acid, tetrahydrofuran (THF), nitric acid, sodium chloride, disodium hydrogen phosphate dodecahydrate, sodium dihydrogen phosphate, urea, Nafion solution, sodium hydroxide (NaOH), potassium ferricyanide, potassium ferrocyanide trihydrate, and potassium chloride were purchased from Aladdin Biological Technology Co., Ltd., China. Silver nitrate, sodium dodecyl sulfate (SDS), and 0.05 μm alumina suspension were purchased from Alfa Aesar. Dimethylamine borane complex (DMAB) and carboxylated PS microspheres were purchased from Energy Chemical. Urease from jack bean (200-300 U/mg) was purchased from Shanghai Macklin Biochemical Co., Ltd. Ethanol was purchased from Sinopharm Chemical Reagent Co., Ltd., China. Glassy carbon electrodes (GCE, 3 mm diameter) were purchased from Yueci Electronics, China. Screen-printed carbon electrodes (SPCE, 4 mm diameter) were purchased from Changsha Sanjun Electronics Co., Ltd., China. Artificial tear solution was purchased from Dongguan Chuangfeng Automation Co., Ltd., China. Deionized water was used throughout all experiments.

In a clean Petri dish, add 10 mM AgNO3 solution, then slowly add an appropriate amount of 1 wt% PS sphere solution (deionized water:ethanol = 1:1). After 5 min to allow complete assembly of the PS spheres, add 1 drop of 2 wt% SDS solution. The Petri dish was sealed with plastic wrap and punctured with small holes. Subsequently, 2 g of finely ground DMAB powder and 2 mL of 6 M HNO₃ solution were placed in the lower compartment of a desiccator. The Petri dish containing the PS self-assembled film was carefully placed on the upper stage of the desiccator. After 24 h of reaction, PS@Ag nanonet films were obtained^[20].

A mixed solution of 3 mM HAuCl₄ and 150 mM Na₂HPO₄ was prepared as the reaction solution. The as-prepared PS@Ag film was transferred to this reaction solution using a hydrophilic glass slide and left to react at room temperature for 30 min. Subsequently, the film was loaded onto a glass slide or a pre-cleaned glassy carbon electrode and dried. The film was then immersed in THF for approximately 1 h to remove the PS spheres and subsequently soaked in saturated NaCl solution to dissolve AgCl, yielding the AuAg NBs.

The GCE electrodes were polished with 0.05 μm alumina suspension and rinsed alternately with deionized water and ethanol 2-3 times. For comparison, a 10 μL aliquot of the prepared AuAg NPs dispersion was drop-cast onto the GCE surface, followed by urease immobilization^[21]. The loadings of AuAg NBs and AuAg NPs were approximately 9.4 and 8.3 μg cm^-2, respectively, ensuring comparable material loadings for subsequent comparison.

Different volumes of urease solution (5, 7.5, 10, and 15 μL; 10 mg mL^-1) were dropped onto the electrode surface to optimize the urease loading. After the immobilized urease had fully dried, 2 μL of 0.5 wt% Nafion solution was cast onto the modified electrode to protect the electrode surface and eliminate potential interferences. The electrode was rinsed with 0.01 M phosphate buffer solution (PBS) at each step to remove excess enzyme and Nafion.

Scanning electron microscopy (SEM) was performed using a Hitachi S-4800 microscope (Hitachi High-Technologies Corporation, Japan) at 3 kV. Transmission electron microscopy (TEM) was performed on a JEM 2100 microscope (JEOL Ltd., Japan) at 200 kV. X-ray diffraction (XRD) patterns were collected using a D8 FOCUS diffractometer (Bruker AXS GmbH, Germany). Ultraviolet-visible (UV-vis) spectra were recorded on a Cary 500 spectrophotometer (Varian Australia Pty Ltd., Australia). The Fourier-transform infrared (FTIR) spectra of the materials were recorded using a Nicolet iS10 FTIR spectrometer (Thermo Fisher Scientific, USA). X-ray photoelectron spectroscopy (XPS) measurements were performed using a K-Alpha X-ray photoelectron spectrometer (Thermo Fisher Scientific, USA) with Al Kα radiation and the C 1s peak at 284.8 eV as an internal standard.

All electrochemical measurements were performed using the Bio-Logic electrochemical workstation (SP-50e, BioLogic Science Instruments SAS, France), with an Ag/AgCl electrode as the reference electrode and a Pt plate as the counter electrode. Cyclic voltammetry (CV) curves were recorded in 0.01 M PBS containing 10 mM [Fe(CN)₆]^3-/4-. Electrochemical impedance spectroscopy (EIS) curves were measured in 10 mM [Fe(CN)₆]^3-/4- and 0.1 M KCl as the supporting electrolyte, with a frequency range of 0.1-10⁵ Hz and an amplitude of 5 mV. Open-circuit potential (OCP) tests were carried out by detecting the potential changes induced by different concentrations of urea in 0.01 M PBS. For non-continuous urea-addition experiments, the potential response was recorded 300 s after each addition. The linear calibration measurements were performed under stirring at 500 rpm. For linear calibration measurements, the response values for the GCE/AuAg NBs/urease and SPCE/AuAg NBs/urease electrodes were extracted at ~120 s and 60 s, respectively. The selected response time may influence the calibration slope and, consequently, the calculated sensitivity and limit of detection (LOD). All measurements were carried out at room temperature. Electrodes were stored in PBS solution at 4 °C when not in use.

Density functional theory (DFT) calculations were carried out using the Vienna Ab Initio Simulation Package (VASP)^[22]. The exchange-correlation interactions were described within the generalized gradient approximation (GGA) using the Perdew-Burke-Ernzerhof (PBE) functional. The interactions between the ionic cores and valence electrons were treated using the projected augmented wave (PAW) method^[23], with a plane-wave kinetic energy cutoff of 450 eV. Partial occupancies of the Kohn-Sham orbitals were treated using the Gaussian smearing method with a smearing width of 0.05 eV. The electronic self-consistent field (SCF) calculations were considered converged when the total energy change was less than 1 × 10^-5 eV, while the structural optimization was terminated when the residual force on each atom was below 0.02 eV Å^-1. A vacuum layer of 18 Å was introduced perpendicular to the slab surface to eliminate interactions between periodic images. For geometry optimization, Brillouin-zone integrations were performed using Γ-centered k-point meshes of 2 × 2 × 1 for the AuAg NPs and 1 × 2 × 1 for all other models. Long-range van der Waals interactions were taken into account using Grimme’s D3 dispersion correction scheme^[24]. The adsorption energy (E_ads) was calculated according to Equation (1):

(1) $$ \mathrm{E}_{\text {ads }}=\mathrm{E}_{\text {tot }}-\mathrm{E}_{\text {slab }}-\mathrm{E}_{\text {mol}} $$

where E_tot, E_slab, and E_mol represent the total energy of the adsorption structure, the energy of the substrate (AuAg NPs, AuAg NBs, or PPy with a degree of polymerization of 2), and the energy of the molecule (deprotonated Gly-His dipeptide), respectively. The simplified Gly-His dipeptide model cannot fully capture the complex three-dimensional structure, adsorption orientation, and solvent effects of intact urease. Therefore, the calculated adsorption energies are used primarily to compare relative local adsorption tendencies on different substrates.

The overall schematic diagram of the synthesis of AuAg NBs is shown in Figure 1. In step I, the PS sphere self-assembled film was immersed in the AgNO₃ solution, and Ag⁺ was slowly reduced by BH₃ gas for 24 h to form elemental Ag, yielding a long-range-ordered PS@Ag film [reaction Equation (2)]. In step II, the PS@Ag film was placed on the surface of the mixed reaction solution of HAuCl₄ and Na₂HPO₄, where a galvanic replacement reaction occurred. During the reaction, most of the metallic Ag was replaced by Au, while a small amount of AgCl formed as a by-product [reaction Equation (3)]. In step III, following the principle of like dissolves like, PS@AuAg was immersed in THF to remove the PS spheres. The AgCl by-product can be removed by soaking the film in saturated NaCl solution^[25]. Finally, AuAg NBs were obtained.

(2) $$ \mathrm{BH}_{3}+6 \mathrm{AgNO}+3 \mathrm{H}_{2} \mathrm{O} \rightarrow 6 \mathrm{Ag} \downarrow+\mathrm{H}_{3} \mathrm{BO}_{3}+6 \mathrm{HNO}_{3} $$

(3) $$ \qquad \quad 3 \mathrm{Ag}+\mathrm{HAuCl}_{4} \rightarrow \mathrm{Au}+3 \mathrm{AgCl}+\mathrm{HCl} $$

Figure 2A shows an SEM image of a self-assembled PS-sphere monolayer, with the spheres measuring approximately 300 nm in diameter. Figure 2B shows the PS@Ag film, which has a large-area, long-range-ordered honeycomb structure. The Ag particles (approximately 5 nm) are densely arranged, providing a suitable template for the subsequent galvanic replacement of Ag at the original sites. Figure 2C shows the morphological characterization of the AuAg NBs, presenting an ordered periodic arrangement with large nanobowls as the repeating units. Particle size distribution analysis shows an average large-nanobowl diameter of 303 nm and a uniform size distribution. The base of each large nanobowl consists of loosely arranged small particles that form a hollow bowl-like structure. We refer to these structures as small nanobowls. Therefore, AuAg NBs exhibit a hierarchical structure where small nanobowls are nested within large nanobowls. Figure 2D indicates that the content of Au is significantly higher than that of Ag, which is consistent with the energy-dispersive X-ray spectroscopy (EDS) results [Supplementary Figure 1]. As shown in Figure 2E, an EDS linear scan on a single small AuAg nanobowl showed similar periodic variations in the Au and Ag signals. The peaks and valleys indicate the hollow structure of a single small nanobowl. Judging from the intensity, the peak of Au is stronger than that of Ag, indicating a relatively higher content. Figure 2F presents an EDS line scan across three adjacent large nanobowls, and it is found that the curves of Au and Ag elements show periodic variations similar to those of the small nanobowls. These results indicate that Au and Ag are spatially co-distributed throughout both the small and large nanobowl structures at the spatial resolution of EDS, supporting the formation of the hierarchical bimetallic nanobowl architecture. It should be noted, however, that EDS provides information on elemental distribution but cannot alone confirm atomic-scale alloy formation. Therefore, the formation of the AuAg alloy is further corroborated below by complementary characterization techniques.

In the experiments shown in Supplementary Figure 2, the HAuCl₄ concentration was maintained at 3 mM while the Na₂HPO₄ concentration was varied to observe its effect on the morphology of the product. Supplementary Figure 2A and B show SEM images of AuAg NBs at different magnifications when the concentration of Na₂HPO₄ is 10 mM. The small nanobowls are randomly arranged on the substrate without forming large nanobowl structures. When the concentration of Na₂HPO₄ is 50 mM, Supplementary Figure 2C reveals that the overall AuAg NBs are more ordered than those prepared at 10 mM Na₂HPO₄, and Supplementary Figure 2D displays the presence of large nanobowl contours, but the bowl walls are damaged. At 75 mM Na₂HPO₄, Supplementary Figure 2E shows an overall ordered arrangement, and Supplementary Figure 2F shows a few defects in the large nanobowl contours, but the bowl walls are intact. When the concentration of Na₂HPO₄ is 150 mM, Supplementary Figure 2G presents a large area of ordered arrangement structure, and in Supplementary Figure 2H, the edges of the large nanobowls are clearly visible. This indicates that the concentration of Na₂HPO₄ plays an important role in the preparation of the ordered arrangement structure of AuAg NBs. In the experiments shown in Supplementary Figure 3, the HAuCl₄ concentration was reduced to 0.3 mM to decrease the rate of the substitution reaction. By changing the concentration of Na₂HPO₄, the regulatory effect on the product structure can be more clearly revealed. At a low HAuCl₄ concentration, Na₂HPO₄ concentrations of 1 mM, 2.5 mM, 5 mM, and 7.5 mM each produced periodically arranged AuAg NBs, but the structures at the bases of the large nanobowls differed markedly. Supplementary Figure 3A shows that the contours of the small nanobowls at the base of the bowl are indistinct; only the overall large-nanobowl structure can be observed. Supplementary Figure 3B shows the initial appearance of the small-nanobowl contours, Supplementary Figure 3C shows clear small-nanobowl contours, and Supplementary Figure 3D shows distinct small-nanobowl contours at the bottom with large gaps between them.

Na₂HPO₄ regulates the hierarchical morphology of AuAg NBs mainly by increasing the solution pH and modulating the replacement kinetics [Supplementary Table 1 and Supplementary Figure 4]. Both excessively slow and rapid reactions were unfavorable for the formation of the ordered hierarchical structure, while the NaOH control confirmed that the effect of Na₂HPO₄ was mainly related to pH regulation. Based on these observations, 3 mM HAuCl₄ and 150 mM Na₂HPO₄ at room temperature were selected as the optimized reaction conditions for the subsequent experiments.

Supplementary Figure 5 shows TEM images of AuAg NBs at different magnifications. Supplementary Figure 5A and B reveal that the large nanobowls are closely connected over a large area. Supplementary Figure 5C shows that the bottom of each large nanobowl is composed of interconnected small nanobowls. Figure 3A shows that the small nanobowls are approximately 30 nm in diameter. The contrast between the outer contour and the interior is consistent with a hollow structure of the small nanobowl. In the EDS elemental analysis of Supplementary Figure 1, the atomic percentage of Ag is 37.2%. According to Vegard’s law:

(4) $$ a_{\mathrm{Au}_{1-n} \mathrm{Ag}_{n}(\text {alloy})}=(1-n) a_{\mathrm{Au}}+n a_{\mathrm{Ag}} $$

where n denotes the Ag atomic fraction in the alloy, and a_Au and a_Ag refer to the lattice spacings of the pure Au (111) and Ag (111) planes, respectively^[26]. We can estimate the lattice spacing of the small AuAg nanobowl to be 0.235 nm. Figure 3B shows a high-resolution transmission electron microscope image of the dashed-box region in Figure 3A, where the lattice spacing is 0.235 nm, and the measurement results are consistent with Vegard’s law. Figure 3C shows multiple concentric diffraction rings corresponding to the (111), (200), (220), and (311) planes of the AuAg NBs. The XRD pattern of AuAg NBs is consistent with a face-centered cubic (fcc) crystal structure [Figure 3D], which is consistent with the selected area electron diffraction (SAED) results. This set of XRD peaks is consistent with the formation of a single-phase AuAg alloy. Compared with the Ag film, AuAg NBs show broader diffraction peaks. On the one hand, this is because the AuAg alloy nanobowls are composed of two substances with similar crystal structures, which broadens the diffraction peaks of the alloy. On the other hand, the small particle size of AuAg NBs also leads to peak broadening^[27].

Figure 3E shows the UV-vis spectra of the Ag film and the AuAg NBs. Both samples show broad absorption peaks, resulting from intense plasmon resonance in the aggregated nanoparticles. The AuAg alloy has only one strong absorption peak. Different ratios of Au and Ag will cause changes in the position of the absorption peak^[28]. The AuAg NBs have a higher Au content than Ag content; therefore, the absorption peak is around 500 nm. Figure 3F shows the FTIR spectra of AuAg NBs and AuAg NBs/urease. The AuAg NBs do not have a distinct absorption peak. The AuAg NBs/urease sample shows a strong, broad absorption band at 3,270 cm^-1, indicating the presence of O-H and NH₂ groups in the urease. The peak at 1,641 cm^-1 is attributed to the stretching vibration of C=O. The spectral band at 1,542 cm^-1 indicates the bending vibration of the N-H group. These results indicate that urease was successfully immobilized on the surface of the AuAg NBs.

Figure 3G shows the XPS survey spectrum of AuAg NBs. Figure 3H and I indicate that the main elemental components of the AuAg NBs are Ag and Au. After the formation of the AuAg alloy, both Ag 3d peaks shift toward lower binding energies [Supplementary Figure 6], whereas the Au 4f_7/2 peak shifts to a higher binding energy relative to pure Au (83.96 eV)^[29]. This phenomenon arises from the charge compensation mechanism associated with d-electron transfer from Au to Ag in the AuAg alloy, further supporting the formation of the AuAg alloy^[30,31].

Based on the combined evidence, the homogeneous spatial distribution of Au and Ag revealed by EDS, the single set of fcc diffraction features, the lattice spacing consistent with the expected AuAg solid solution observed by high-resolution transmission electron microscopy (HRTEM), and the opposite binding-energy shifts of Au 4f and Ag 3d in XPS collectively support alloy formation rather than a phase-separated physical mixture of Au and Ag.

Figure 4A illustrates the mechanism for detecting urea using AuAg NBs. Urease is immobilized on the AuAg NBs and catalyzes urea hydrolysis into NH₄⁺, HCO₃^-, and OH^-, thereby changing the OCP. Supplementary Figure 7 shows that urease is mainly distributed on the outer surface and inside the large AuAg nanobowls, and is also observed on the smaller nanobowls. This open structure facilitates urea transport in the electrolyte to the immobilized urease.

The potential response is affected by factors such as enzyme concentration, pH, and electrode type^[13,32]. In Supplementary Figure 8A, the redox peak potentials of [Fe(CN)₆]^3-/4- are around 0.29 V and 0.16 V. Following urease immobilization, the peak current decreases, owing to the electron-insulating biomacromolecular layer that obstructs interfacial charge transfer. Consequently, the charge transfer resistance (Rct) value experiences a pronounced increase to 1,155 Ω [Supplementary Figure 8B], which further indicates the successful fabrication of the GCE/AuAg NBs/urease electrode [Supplementary Table 2]. We investigated the effects of enzyme loading and PBS pH. As shown in Supplementary Figure 9A, the potential response of the GCE/AuAg NBs electrode significantly increases after loading 10 μL of urease. Further increasing the loading volume to 15 μL only yields a marginal improvement. Considering both sensing performance and cost efficiency, loading 10 μL of urease is optimal. Supplementary Figure 9B indicates that the potential response is the greatest at pH 7.4. Either excessively acidic or alkaline conditions will decrease enzyme activity, resulting in a reduction in the potential change. In conclusion, the optimal test conditions for the GCE/AuAg NBs/urease electrode are a urease loading of 10 μL and a PBS pH value of 7.4, and subsequent tests were conducted under these conditions. The GCE/AuAg NBs electrode without urease showed only a negligible ~1 mV response to 800 μM urea [Supplementary Figure 10], indicating that AuAg NBs themselves do not catalyze urea hydrolysis to generate a potentiometric response.

AuAg NPs were synthesized by the NaBH₄ reduction method^[21], with an average particle size of ~30 nm [Supplementary Figure 11]. They exhibited optical properties similar to those of AuAg NBs [Supplementary Figure 12]. To evaluate the enzyme-immobilization performance of the AuAg NBs, we also modified the GCE with a PPy layer for comparison^[33]. Supplementary Figure 13 shows the CV curves of different electrodes measured for electrochemically active surface area (ECSA) evaluation in the non-Faradaic region. The calculation method of ECSA is as follows:

(5) $$ \mathrm{ECSA}=C_{d l} / C_{s} $$

where the double-layer capacitance (C_dl) is defined as the ratio of the capacitive current ΔI to the scan rate, namely the slope of the straight line in Supplementary Figure 14. C_s represents the specific areal capacitance of the sample catalyst. Figure 4B compares the ECSA values of GCE/Ag, GCE/AuAg NPs, GCE/PPy and GCE/AuAg NBs. The larger ECSA of AuAg NBs indicates a greater electrochemically accessible surface area, which may increase the amount of immobilized urease and contribute to the enhanced potentiometric response. Figure 4C compares the UV-vis absorption spectra of rinsing solutions collected from different electrodes after washing. Urease exhibits a characteristic absorption peak at approximately 278.5 nm, with its absorbance being directly proportional to the concentration^[34]. Therefore, a lower absorbance signifies a lower urease content in the rinsing solution. The GCE/AuAg NBs/urease electrode exhibits the lowest absorbance, demonstrating its outstanding enzyme immobilization efficiency. Figure 4D compares the responses of different electrodes to 800 μM urea. The GCE/AuAg NBs/urease electrode delivers the most significant potential change, with its potential response enhanced by 135.5% compared with the GCE/AuAg NPs/urease electrode.

Supplementary Figure 15 shows the XPS survey spectra of urease and AuAg NBs/urease. As shown in Figure 4E, the high-resolution N 1s spectra of urease exhibited two characteristic peaks at approximately 399.85 eV and 401.91 eV, corresponding to amide nitrogen (NH-C=O) and protonated amino groups (-NH₃⁺), respectively. After immobilization on the AuAg nanobowls, the amide nitrogen peak shifted to a higher binding energy of 399.96 eV, and a new -NH₂ component emerged at 400.43 eV, indicating that the immobilization process altered the chemical environment of nitrogen atoms^[35-37]. Meanwhile, compared with the pristine AuAg alloy, the Au 4f_7/2 and 4f_5/2 peaks in AuAg NBs/urease both shifted toward lower binding energy [Supplementary Figure 16]. These coordination interactions enable stable immobilization of urease on the AuAg alloy.

To further elucidate the adsorption behavior of urease on AuAg alloy nanobowls, DFT calculations were performed to evaluate the adsorption energies of a simplified Gly-His model on AuAg NPs, PPy, and AuAg NBs [Figure 4F]. A Gly-His dipeptide was employed as a simplified model to represent the local interaction of urease with different substrates^[38]. The calculated adsorption energies indicate that the Gly-His model exhibits the most favorable adsorption tendency on AuAg NBs, followed by AuAg NPs and PPy.

Notably, the DFT calculations predict stronger adsorption on AuAg NPs than on PPy, whereas PPy experimentally exhibits greater urease retention [Figure 4C] and a larger potential response [Figure 4D]. This discrepancy mainly arises from the simplified computational model and the difference between local adsorption and whole-enzyme immobilization. The two-monomer PPy model may underestimate the adsorption capability of the experimental three-dimensional PPy network^[39]. In addition, urease immobilization is also influenced by interfacial accessibility, steric effects, surface morphology, and protein conformation and orientation. In this regard, the larger ECSA of PPy compared with that of AuAg NPs [Figure 4B] may provide more opportunities for urease immobilization.

Taken together, the hierarchical structure increases the electrochemically accessible surface area and facilitates urease loading, whereas the AuAg surface provides favorable interfacial interactions for urease immobilization. These coupled effects contribute to the enhanced potentiometric response of AuAg NBs.

Figure 5A shows the potential response of the GCE/AuAg NBs/urease electrode upon successive increases in urea concentration. Figure 5B indicates that the biosensor shows a clear linear response over 0.1-1 mM. The sensor exhibited a sensitivity of 10.6 mV·mM^-1 and a LOD of 6.7 μM. To further extend the working range toward higher urea concentrations, the potentiometric response was additionally evaluated over the range of 1-6 mM [Supplementary Figure 17A]. A second linear response region was observed from 1 to 6 mM, with a sensitivity of 2.27 mV·mM^-1 [Supplementary Figure 17B]. At 2 mM urea, the GCE/AuAg NBs/urease electrode exhibited a larger potential response than the GCE/Ag/urease and GCE/AuAg NPs/urease electrodes [Supplementary Figure 17C], further demonstrating the GCE/AuAg NBs/urease advantage at higher concentrations.

The sensor response time was further evaluated using t90, defined as the time required to reach 90% of the steady-state potential response. The t90 values under different urease loadings, pH values, and measurement conditions are summarized in Supplementary Table 3. Under the optimized conditions of a 10 μL urease loading and pH 7.4, the GCE/AuAg NBs/urease electrode exhibited a t90 of approximately 207.5 s, which was further reduced to 136.0 s by stirring at 500 rpm due to enhanced mass transport.

To evaluate the repeatability of the electrode, 20 consecutive uses were performed. The relative standard deviation (RSD) was calculated according to the following equation:

(6) $$ \mathrm{RSD}(\%)=\mathrm{SD} / \text { mean } \times 100 \% $$

where SD and mean represent the standard deviation and arithmetic mean, respectively, of the 20 mean response values. As shown in Figure 5C, the electrode exhibited good repeatability, with an RSD of 3.8%, and ~90% of the initial response was retained after the 20th measurement. Figure 5D compares the linear range and LOD of the proposed sensor with those of previously reported urea-sensing systems, showing that the present sensor has a comparatively low LOD. The storage stability was further evaluated over 15 days. After 15 days, the sensor response decreased by only 9.3% [Figure 5E]. Batch-to-batch reproducibility was assessed using three independently prepared GCE/AuAg NBs/urease electrodes at a fixed urea concentration of 800 μM [Supplementary Figure 18]. The inter-batch RSD was 3.1%, demonstrating good consistency among independently fabricated sensors. As shown in Supplementary Figure 19, the response to urea was substantially higher than that to the tested interferents, indicating good selectivity of the GCE/AuAg NBs/urease electrode under tear-relevant conditions.

The GCE/AuAg NBs/urease electrode was tested on artificial tears to verify its urea detection capability in complex matrices [Supplementary Figure 20A and B]. Artificial tears were diluted 1:40 with PBS solution. The potentiometric responses of diluted and undiluted artificial tear samples were compared [Supplementary Table 4]. The results indicated that dilution had a minimal effect on the sensor response. Recovery tests were performed by spiking the samples with urea at 500 μM and 800 μM, respectively. The recovery rates for the artificial tear samples were 93.8% and 95.7%. Using the same method, tests in tap water samples gave recoveries of 98.2% and 103% [Supplementary Figure 20C and D, Supplementary Table 5]. These recovery results provide proof-of-concept evidence for the feasibility of urea sensing in artificial tears. Compared with reported urea sensors, the GCE/AuAg NBs/urease sensor exhibited a low LOD of 6.7 μM, linear ranges of 0.1-1 and 1-6 mM, 15-day stability, and urease-based potentiometric detection in artificial tears under mild conditions [Supplementary Table 6]^[40-46]. However, a faster response and validation with real tear samples are still needed.

We integrated an SPCE into eye patches [Figure 5F], proposing a potential strategy for the application of AuAg NBs in DED diagnosis. Figure 5G displays the potential response of the modified SPCE with increasing urea concentration. A well-defined linear relationship was obtained over the urea concentration range of 100-900 μM [Figure 5H], with a sensitivity of 8.1 mV·mM^-1 and an LOD of 8.7 μM. These results demonstrate the feasibility of integrating the sensing system with SPCE for miniaturized urea sensing. However, for practical wearable applications, challenges such as reliable tear collection, appropriate sensor positioning, mechanical stability, and wearing comfort still need to be addressed.