3,3′,5,5′-Tetramethylbenzidine (TMB), Co (NO₃)₂.6H₂O, glucose, fructose, maltose, sucrose, ascorbic acid, and dopamine were purchased from Aladdin (Shanghai, China). Ruthenium bipyridine was purchased from Sigma-Aldrich (Shanghai, China). Hexamethylenetetramine, Na₂HPO₄, NaH₂PO₄, H₂O₂, acetic acid, sodium acetate, hydrochloric acid, and NaOH were purchased from Sinopharm Chemical Reagent Co., Ltd (Shanghai, China). All the chemical reagents were of analytical grade. Deionized water (18.2 MΩ cm at room temperature), obtained from a Milli-Q water purification system (Millipore Corp., MA), was used to prepared all of the aqueous solutions.

FT-IR spectra were obtained from the KBr slice with a Nicolet iS10 FT-IR spectrophotometer (Thermo Fisher Scientific, Shanghai). X-ray diffraction (XRD) patterns were recorded on a Bruker D8 Advance diffractometer (Germany) with a Cu Kα (1.5406 Å) radiation source. All pH measurements were performed with a Sartorius PB-10 digital pH meter (Shanghai, China). Transmission electron microscopy (TEM) and scanning electron microscopy (SEM) were performed with a HITACHI model HT7700 instrument operating at 80 kV accelerating voltage and a ZEISS Gemini 300 with OXFORD Xplore, respectively. The ultraviolet-visible (UV-Vis) absorption spectra were recorded by using a U3900 spectrophotometer (Hitachi, Japan). All fluorescence measurements were performed on a Hitachi F-7000 fluorescence spectrometer. The excitation wavelength was set at 496 nm, and the emission spectra from 550 to 750 nm were observed. The fluorescence intensity at 610 nm was used to evaluate the performance of the proposed strategy.

Co₃O₄ nanosheets were prepared in accordance with the method previously reported with a minor modification (Hwang et al., 2011). Under continuous stirring conditions, 1.3 g of Co(NO₃)₂·6H₂O and 0.6 g of hexamethylenetetramine were dissolved in 100 ml of deionized water, respectively. Then, the aforementioned two solutions were mixed together. The resultant solution was maintained at pH 10 with 1 M sodium hydroxide solution and was vigorously stirred for 2 h. Next, the resultant solution was transferred into a pressure-tight teflon-lined stainless-steel autoclave, and then, it was heated up to 110°C. After 15 h, the autoclave cooled down naturally to room temperature. Then, the obtained products were washed with water, ethanol, and acetone, in turn, and vacuum dried at 60°C for 4 h. At last, the dried products were calcined for 2 h at 200°C. The obtained products were studied for their morphological characteristics and the peroxidase mimetic activity test.

The Co₃O₄ nanosheets were added into HAc-NaAc buffer (100 mM, pH 4.5) in the presence of H₂O₂ by varying concentrations of TMB. The reactions were monitored in time at a wavelength 650 nm by using a U3900 spectrophotometer. The steady-state kinetic catalytic parameters were determined based on the Michaelis–Menten equation(Huang et al., 2018).

H₂O₂ detection was conducted as follows: different concentrations of H₂O₂ were introduced into the mixture of 20 μg/ml Co₃O₄ nanosheets and 0.5 mM TMB in HAc-NaAc buffer (100 mM, pH 4.5). The solution was incubated for 30 min at 37°C, and the absorbance of the solution was detected.

Glucose detection was conducted as follows: 5 μL of 50 mgmL⁻¹ glucose oxidase was added into 95 μL PBS (10 mM, pH 7.4) containing different concentrations of glucose, and then, the mixture was incubated for 30 min at 37°C. Subsequently, the abovementioned reaction solution was transferred into 400 μL acetate buffer (100 mM, pH 4.5) containing 0.5 mM TMB, 20 μg/ml Co₃O₄ nanosheets, and 0.5 mM Ru (bpy)₃ ²⁺. After incubating for another 30 min at 37°C, the reaction solution was monitored by using the fluorescence spectrophotometer. In the control experiments, fructose, maltose, sucrose, ascorbic acid, and dopamine were used instead of glucose, respectively.

For glucose determination in human serum samples, they were pretreated according to the literature (Peng and Weng, 2017) with a minor modification: 50 μL of the serum sample was diluted with 50 μL water and then were added into 900 μL solution containing 0.11 M Ba(OH)₂ and 0.0765 M ZnSO₄. The resultant solution was centrifuged for 10 min with 4,000 rpm. The supernatant was collected for the determination of glucose. All the experiments involving human beings were approved by the Ethics Committee Approval of China and operated in strict compliance with the Ethics Committee of Qingdao Agricultural University.

Co₃O₄ nanosheets were synthesized through a hydrothermal method, using Co(NO₃)₂·6H₂O and hexamethylenetetramine as precursor materials, as shown in Figure 1A. The XRD pattern of the synthetic Co₃O₄ nanosheets is shown in Figure 1B. It is obvious that all the characteristic peaks are keeping in with the reported data (JCPDS Card No. 42-1467). The IR spectrum of the Co₃O₄ nanosheets is presented in Figure 1C. The two distinct adsorption peaks at 566 cm⁻¹ and 668 cm⁻¹ are ν1 and ν2, respectively, stretching vibrations of the metal–oxygen bonds which confirm the formation of Co₃O₄ (Hwang et al., 2011). The broad absorption bands at 3,420 cm⁻¹ and the peak at 1,624 cm⁻¹ are due to O-H stretching and the bending vibration of the adsorbed water at the surface. The band at 1,134 cm⁻¹ is corresponding to the Co-OH bending vibration. The synthesized nanosheets were also examined by SEM and TEM, which are shown in Figures 1D,E. It is obvious that the synthesized products are hexagonal nanosheets in large quantities. The stacking densities of the nanosheets are very high even though most of them are attached with each other through their surface. Most of them reveal a hexagonal shape with an internal angle of ∼120°; however, there are also some deformed hexagonal-structured nanosheets among them. The nanosheets are about 200 ± 10 nm in diagonal.

The colorless TMB solution can be oxidized into a blue solution by H₂O₂ in the presence of a catalyst. Therefore, TMB and H₂O₂ were used as the reaction substrates to evaluate the peroxidase-like activity of the Co₃O₄ nanosheets. The absorbance of oxTMB (the oxidation product of TMB) increased with time in the mixture of Co₃O₄ nanosheets, H₂O_2, and TMB (Figure 2A).

It can be seen from Figure 2B that the mixture of Co₃O₄ and H₂O₂ have no absorption at 650 nm (curve 1). It was also obvious that the oxidation reaction of TMB cannot take place in the presence of Co₃O₄ or H₂O₂ alone (curves 2 and 3). As a result, TMB was oxidized by H₂O₂ in the presence of as-synthesized Co₃O₄ nanosheets (curve 4). The absorbance of the mixture at wavelength 650 nm obviously increased with the increasing concentration of the Co₃O₄ nanosheets from 0 to 20 μg ml⁻¹ which are shown in Figure 2C. As a comparison, the activity of the earlier published peroxidase-like nanomaterials MnO₂ and 2D Co-MOF (Wang et al., 2022) have also been examined, which are shown in Supplementary Figure S1. To further investigate the catalytic activity of the Co₃O₄ nanosheets, the steady-state kinetic parameters were examined for the reaction between H₂O₂ and TMB, as shown in Supplementary Figure S2. The results demonstrated that the catalysis of Co₃O₄ nanosheets showed typical Michaelis–Menten curves (Supplementary Figure S2A). The curves were fitted to the Lineweaver–Burk plots (Supplementary Figure S2B). According to the plots, the Michaelis–Menten constants (K _m) and the maximum initial reaction rates (V _max) were calculated to be 0.42 mM and 5.7 × 10⁻⁷ M⋅ s⁻¹, respectively, which are superior to the previously published peroxidase-like materials (Lin et al., 2014a; Lin et al., 2014b; Huang et al., 2018). In a word, Co₃O₄ nanosheets can facilitate the oxidation of TMB in the presence of H₂O₂. So the content of H₂O₂ can be detected based on this catalytic reaction principle.

As H₂O₂ was the primary catalyzed reaction product of glucose oxidase, the glucose content can also be detected using the Co₃O₄ nanosheets as the catalyst (Illustrated as Scheme 1). Additionally, the colorimetric product oxTMB can effectively quench the fluorescence emitted by Ru (bpy)₃ ²⁺ due to the IFE. Thus, glucose also can be indirectly detected by the fluorescence method which can not only improve detection sensitivity but also avoid the false positives caused by a single colorimetric signal response (Shin et al., 2020; Munzi et al., 2021; Wan et al., 2021).

The absorbance spectra of TMB under different conditions were measured to prove the feasibility of the sensing strategy. As shown in Figure 3A, a very weak absorption was obtained in the presence of TMB independently (curve a). There is a slight increase of absorption compared with TMB in the mixture of Co₃O₄ and TMB (curve b), which might be a weak catalytic effect of Co₃O₄ to TMB. The biggest absorption signal (curve c) was obtained when H₂O₂ was added into the aforementioned solution.

Figure 3B shows fluorescence spectra of Ru (bpy)₃ ²⁺ under different experimental conditions. As shown in Figure 3B, the biggest fluorescence signal (curve a) was obtained in the presence of Ru(bpy)₃ ²⁺ independently. However, the fluorescence intensity significantly decreased when an amount of the Co₃O₄ nanosheets, TMB, and H₂O₂ were added into the aforementioned solution (curve b). It is worth noting that the maximum emission wavelength is blue-shifted slightly. The experimental results demonstrated that a certain amount of oxTMB was produced in the solution, which quenched the fluorescence emitted by Ru(bpy)₃ ²⁺ due to the IFE . The fluorescence quenching mechanism of Ru(bpy)₃ ²⁺ by the Co₃O₄ nanosheets raised from the IFE process are demonstrated in Figure 3C. The reduced fluorescence intensity has a quantitative relationship with the concentration of H₂O₂. Thus, the content of glucose can also be detected indirectly by this fluorescence sensing strategy.

To improve the detection sensitivity, the experimental conditions have been optimized. Absorption is dependent on the concentration of the chromogenic reagent, so the concentrations of TMB were initially optimized. The results are shown in Supplementary Figure S3 in SI which showed that the absorption increased gradually with the increase of concentration from 0.1 to 0.4 mM and then increased slightly and kept stable. To ensure the experimental results, therefore, 0.5 mM TMB was selected as the optimal concentration.

All the experimental results confirmed that the Co₃O₄ nanosheets revealed a peroxidase-like activity. It was noted that the catalytic activity of peroxidase was associated with the pH and temperature. Therefore, the optimal reaction conditions were investigated. The results demonstrated that Co₃O₄ nanosheets have the best catalytic activity in pH 4.5 at 37°C (Supplementary Figures S4, S5 in SI).

The fluorescence detection method in this strategy is based on the inhibition capability of the colorimetric reaction products on the fluorescence emitted by Ru (bpy)₃ ²⁺. Therefore, only Ru (bpy)₃ ²⁺ concentrations have been optimized to obtain the biggest difference of the fluorescence intensity in the presence of Co₃O₄, TMB, H₂O_2, and Ru (bpy)₃ ²⁺ compared with Ru (bpy)₃ ²⁺ independently. The difference of the fluorescence intensity gradually increased with increasing Ru (bpy)₃ ²⁺ concentrations, as shown in Supplementary Figure S6 in SI; the biggest difference was obtained at the concentration of 0.5 mM. Therefore, 0.5 mM Ru (bpy)₃ ²⁺ was selected as the optimal concentration for this procedure.

A colorimetric sensor for the detection of H₂O₂ was constructed based on the peroxidase-like activity of the Co₃O₄ nanosheets. Under the optimal conditions, the concentration-response curve of H₂O₂ to TMB at wavelength 650 nm is shown in Figure 4A. A good linear relationship was established between 2 and 200 μM (Figure 4B). The linear regression equation was ΔA = 0.0068 c + 0.036 with the correlation coefficient of 0.997 [ΔA: difference of absorbance and c: concentration of H₂O₂ (μM)]. The limit of detection for H₂O₂ was estimated to be 0.4 μM (based on the signal-to-noise ratio of 3). Similarly, the fluorescence response of the system was examined with different concentrations of glucose which demonstrated that the fluorescence intensity gradually decreased with the increase in the concentration of glucose (Figure 4C). These results verified that with the greater concentration of glucose, more H₂O₂ were produced, which inhibited the fluorescence of Ru(bpy)₃ ²⁺. The difference of the fluorescence intensity showed a good linear relationship with the concentration of glucose ranging from 0.02 to 2 μM (the inset of Figure 4D). The linear regression equation was ΔF = 253 c + 96.5 with the correlation coefficient of 0.997 [ΔF: difference of fluorescence intensity and c: concentration of glucose (μM)]. The limit of detection was estimated to be 5.0 nM (based on the signal-to-noise ratio of 3), which are superior to the reported methods (Supplementary Table S1).

To prove the anti-interference capability of the strategy for the detection of glucose, five other analogs, such as fructose, maltose, sucrose, ascorbic acid, and dopamine, were also determined by the fluorescence method, which are shown in Supplementary Figure S7 in SI. The results showed that the other analogs hardly interfered with the determination of glucose under this experimental condition. The stability of the assay system has also been investigated. Supplementary Figure S8 in SI showed the fluorescence response of the sensing system with the concentration of 0.2 and 1.5 μM glucose based on the six-time detections, respectively. The fluorescence intensity was relatively stable with the relative standard deviation of 2.61 and 3.93%, respectively.

H₂O₂ in artificial lake water, glucose in fruit juice, and blood samples were detected by colorimetric and fluorescence methods, respectively. The water sample was obtained from the artificial lake in Qingdao Agricultural University, and fruit juice was purchased from the local supermarket. They have been processed according to the literature studies (Wang et al., 2016a; Wang et al., 2016b). Blood samples, provided by the Hospital of Qingdao Agricultural University by collecting from healthy volunteers with informed consent, were processed according to the literature (Peng and Weng, 2017). In addition, the recovery tests were examined by adding a known concentration of the standard to the pretreated solution in real samples (Supplementary Tables S2, S3 in SI). The results demonstrated that the recoveries obtained ranged from about 99.2 to 106.6% and 94–108%, respectively.