The suspensions of α-Ni(OH)₂ and the unprecedented Ni_1-x-yV_xCe_y-LDH NPs (x = y = 0.05 or 0.10) were prepared as reported previously (Assis et al., 2020; Azeredo et al., 2021) by dissolving the total of 4.82 mmol of the metal acetates (or metal chloride, in the case of vanadium ions) in 25 ml of glycerin at 50°C. After the dissolution of the metal salts, 9.64 mmol of KOH dissolved in 25 ml of anhydrous n-butanol was added dropwise into the stirred solution at room temperature.

Fluorine-doped tin oxide (FTO) glass plates were carefully washed with isopropanol and water, dried in air, and in the sequence modified by spin-coating the suspensions of α-Ni(OH)₂ and Ni_0.9V_0.05Ce_0.05-LDH NPs, at 2,500 rpm. These electrodes were then dried under vacuum at 80°C for 120 min. Such modified FTO electrodes were evaluated using CV, EIS, and galvanostatic charge-discharge (GCD). GCD curves were registered using larger modified FTO glass electrodes (geometric area ∼4 cm²) prepared as discussed earlier by spin-coating, applying a suitable volume of NP suspension (containing 3–8 mg) on the surface, and heating to 80°C.

The fabrication of the screen-printed electrode (SPE) is shown in Scheme 1. A cheap, flexible, and commercially available polyethylene terephthalate foil was used as a substrate to compose the SPEs. Ag/AgCl conductive ink was used to provide the reference electrode and the contact pads of the three electrodes to promote conductivity. Graphite ink loaded with 25 weight% of Ni_0.9Ce_0.05V_0.05-LDH NPs was screen-printed to build the working (WE) and counter (CE) electrodes. The fabrication process was simplified using the same ink in WE and CE, especially because similar results were observed for the SPE using a CE printed only with graphite ink (Scheme 1A). After that, the ink layer was dried at 85°C for 10 min in a Breville oven. A “blue tape” (adhesive) was used to define the active area of the electrodes, as well as to insulate the wire connections. The final dimensions of the SPE sensor were 3 × 1 cm (length vs. width). The surface area of the WE was ∼12 mm², and the dimensions of the entire sensor were 3 × 1 cm (length vs. width), as shown in the design drawing (Scheme 1B) and photography (Scheme 1C) of the electrochemical analytical devices. The electrochemical measurements were carried out after properly connecting each of the electrodes (WE, reference electrode, and CE) to a potentiostat.

The samples were characterized by XRD in a Bruker D2 Phaser equipment with a Cu Kα source (λ = 1.5418 Å as the source wavelength, voltage = 30 kV, current = 15 mA, step = 0.05°), in the 2θ range 5–70°. The samples were prepared by depositing the materials on glass slips and vacuum drying at 80°C. TEM and STEM images of the nanocomposites were obtained in JEOL JEM-2100 FEG equipment with an applied accelerating voltage of 200 kV. The samples were prepared on lacey carbon copper grids (Ted Pella) by dispersing 3 μl of an NP suspension diluted in DI-water. The relative amounts of nickel, vanadium, and cerium were measured by inductively coupled plasma optical emission spectrometry (Arcos/Spectro) using the solutions obtained by dissolving the concentrated dispersions of Ni_1-x-yV_xCe_y-LDH with a 0.1 mol L⁻¹ nitric acid solution.

CV, GCD, and EIS measurements were carried out using an EcoChemie Autolab PGSTAT30 potentiostat/galvanostat and a conventional three-electrode cell constituted by an Ag/AgCl (3.0 mol L⁻¹ KCl) reference, a platinum plate counter, and a modified FTO WE. For the batch injection analysis (BIA) experiments, amperometric detection was chosen, being the modified electrode positioned at the bottom of the three-dimensional-printed electrochemical BIA cell (Squissato et al., 2017; Rocha et al., 2018a). To perform the injections, a battery-powered Eppendorf electronic micropipette (Multipette^® stream) was used, which allows injections with volumes and rates between 10 and 1,000 µl and flow rates from 28 to 345 μl s⁻¹. The injections mentioned earlier were carried out in a wall jet configuration (inverted position in relation to the working electrode with a distance of approximately 2 mm) (Rocha et al., 2018b). Electrochemical impedance spectra were recorded in the range from 0.01 to 50,000 Hz, modulating the frequency of a sinusoidal potential wave (amplitude = 10 mV) superimposed to a DC potential.

XPS was performed in a SPECS Phoibos 150 high-resolution hemispherical analyzer with multichannel detection using an Al K_α X-ray source (1,486.6 eV) with 180 W, operating at a 10^–8 mbar base pressure, in normal emission, and constant pass energy (60 eV for the survey and 25 eV for high-resolution spectra). The hydroxide films were deposited on the native silicon oxide of silicon wafers that were glued to the sample holder with conductive carbon tape. The samples were inserted in a load-lock chamber to degas for 12 h before transferring to the XPS chamber. The obtained data were fitted assuming Shirley-type background and Lorentzian peaks numerically convoluted with a normalized Gaussian function with a fixed full-width half maximum of ∼1.2 eV that describes the instrumental broadening.

In the last decade, the sol-gel method reported by Tower (Tower, 1924) has been extensively explored in the preparation of mono- and bimetallic hydroxides with a potential application, especially as electrochemical sensors (Azeredo et al., 2019; Azeredo et al., 2020) or electrodes for HSCs (Gonçalves et al., 2018a; Gomes et al., 2019). However, to the best of our knowledge, this is the first time that this method has been used to obtain multifunctional trimetallic hydroxides, reporting the unprecedented trimetallic NiVCe-based hydroxide (Figure 1A).

The relative amounts of nickel, vanadium, and cerium in the Ni_1-x-yV_xCe_y-LDH sol in the as-prepared materials were determined by inductively coupled plasma optical emission spectrometry. The relative Ni:V:Ce molar ratio measured for the Ni_0.9V_0.05Ce_0.05-LDH was 90.3:4.9:4.8, in good agreement with the expected values (90, 5, and 5%, respectively). In addition, the layered structure of Ni(OH)₂ and Ni_0.9V_0.05Ce_0.05-LDH NPs (Figure 1A) was confirmed by XRD measurements (Figure 1B). The Ni(OH)₂ NPs presented a peak at 10.5° (8.45 Å), indexed to the (003) plane assigned to the turbostratic alpha-phase [α-Ni(OH)₂], as reported in the literature (Gonçalves et al., 2018b) [JCPDS no. 38-0175, α-Ni(OH)₂·0.75H₂O]. Similarly, the presence of the basal reflections (003) in the Ni_0.9V_0.05Ce_0.05-LDH diffractogram at 10.5° (8.45 Å) evidences the formation of the LDH structure, where the apparent lower intensity and broadness of that diffraction peak are characteristics of a less crystalline material owing to rather smaller size nanoparticles (Azeredo et al., 2021). However, when the amount of vanadium ions precursor (VCl₃) is greater than five atom%, for example, in the case of bimetallic Ni_0.9V_0.1-LDH (Supplementary Figure S1A) and trimetallic Ni_0.8V_0.1Ce_0.1-LDH (Supplementary Figure S1B), the formation of KCl is noteworthy. In this case, a second phase indexed to the face-centered cubic KCl structure with peaks at 2θ = 28.37, 40.54, and 58.64°, which were indexed to (200), (220), and (400), respectively (Agarwal et al., 2017), was also recorded.

For a comprehensive chemical characterization of the formed trimetallic hydroxide NPs, XPS analysis was used. Supplementary Figure S2 demonstrates the presence of the expected elements resulting from synthesis. Because the pure elements, Ni, V, and Ce, can produce multi- and complex phases depending on the synthesis conditions, it is worth studying the synthesis product from the pure elements for the sake of comparison. Supplementary Figure S3 displays high-resolution XPS from Ni 2p, V 2p, and Ce 3d from 100% Ni, V, and Ce NPs samples. In the case of Ni 2p (Supplementary Figure S3A), the fitting procedure reveals a single-phase consistent with lamellar Ni(OH)₂ that has strong components at 857.1 and 874.7 eV related to Ni 2p_3/2 and Ni 2p_1/2, respectively. The line shape, satellite structure, and peak positions are in good agreement with previous results from the literature (Xiong et al., 2017). Supplementary Figure S3B displays V 2p core-levels. Unfortunately, it has an important coincidence with O 1s satellites peaks promoted by the Al K_α3,4 radiation. Despite that, the comprehensive fitting procedure, which has constrained the relative peak position and peak intensity, has been able to correctly untangle the information from V 2p. Taking into account the narrow linewidth and peak position at 517.5 eV for the V 2p_3/2, one concludes for a single-phase consistent with V⁵⁺, suggesting the formation of V₂O₅ (Prześniak-Welenc et al., 2016). Finally, Supplementary Figure S3C displays high-resolution XPS for Ce 3d core levels, which displays a complex structure of multiplets. There are four main feature peaks named ν₀ and ν' (882.8 and 886.5 eV) associated to Ce 3d_5/2 and μ₀ and μ' (901.3 and 905.1 eV) associated with Ce 3d_3/2. The relative peak position, as well as peak intensity, are well-known characteristics for the correct identification of Ce³⁺ or Ce⁴⁺ phases as demonstrated by Allahgholi et al. (2015). Furthermore, only the Ce⁺⁴ phase shows a characteristic plasmon structure around 917 eV. Therefore, our data are consistent with solely Ce³⁺. Moreover, the line shape is also consistent with particle sizes between 1.4 and 2.4 nm, as previously reported by Allahgholi et al. (2015), which were also confirmed by TEM measurements (Supplementary Figure S4).

As we know the results from the single elements NPs, now in Figure 2, we can qualitatively compare the results for Ni 2p and Ce 3d core-levels for two trimetallic samples, Ni_0.9V_0.05Ce_0.05-LDH and Ni_0.8V_0.1Ce_0.1-LDH NPs. For the Ni, as a major element in the compound, the Ni 2p contributions are almost coincident for the three samples. Ce 3d_5/2 has partial overlap with the Ni 2p_1/2. Despite that, we can identify ν₀, ν', μ_0, and μ' peaks at almost the same position displayed in the pure Ce NPs. We might wonder if the NPs are formed by separated phases of Ni and Ce NPs instead of layered hydroxides formation. If separated phases are the case, we should be able to use the same peak positions, line widths, and relative peak intensities to fit the complex structure. Supplementary Figure S5 shows our attempt to fit the spectrum of Ni_0.8V_0.1Ce_0.1-LDH NPs using such constraints. The result demonstrates that extra structures are presented, suggesting the formation of a trimetallic hydroxide.

The formation of Ni_0.9V_0.05Ce_0.05-LDH NPs can be confirmed in the TEM/high-resolution transmission electron microscopy (HRTEM) images [Figures 3A, B, the NPs were designedly prepared by depositing them on graphene oxide (GO) nanosheets], showing spherical shape with 4–6-nm diameter and no significant signs of agglomeration. In addition, the HRTEM images shown in Figure 3C are consistent with NPs with a high degree of crystallinity, as supported by the presence of reflection fringes assigned to the (200) diffraction planes of NiO (JCPDS no. 73–1523). At this moment, it is important to note that the dehydration process and consequent in situ formation of the respective metal oxide NPs are quite common for this type of material, facilitated by the local heating and the high vacuum in the sample chamber. On the other hand, in contrast to TEM images, STEM images have even smaller NPs on GO, indicating that local heating can also induce the coalescence of nanoparticles, as confirmed by STEM-bright field (transmitted electrons, BF, Figure 3D) and STEM-dark field images (scattered electrons in high-angle annular dark-field-HAADF, Figures 3E, F).

The electrochemical behavior of Ni_0.9V_0.05Ce_0.05-LDH and some bimetallic (Ni_0.95Ce_0.05-LDH and Ni_0.95V_0.05-LDH), as well as for α-Ni(OH)₂ analog modified FTO electrodes, were evaluated by CV. The CVs were recorded in the range of 0.0 to +0.6 V vs. Ag/AgCl (3.0 mol L⁻¹ KCl), at a scan rate of 50 mV s⁻¹, using 1.0 mol L⁻¹ KOH solution as the background electrolyte. For all electrodes, the CVs exhibited a well‐defined pair of waves associated with the Ni²⁺(OH)₂/Ni³⁺(OOH) process (Figure 4). In addition, for the α-Ni(OH)₂, the anodic peak potential (i _ap) started to shift to more positive potentials, and the peak intensity decreased relatively fast after 50 continuous CVs (Figure 4A), typical of the conversion process of the nanomaterial from alpha to β-Ni(OH)₂ (Gonçalves et al., 2017). In contrast, the Ni_0.9V_0.05Ce_0.05-LDH exhibited a stable i _ap and a small potential shift during the 100 redox cycles (Figure 4B). This enhanced stability, also observed for bimetallic nickel/cerium LDH NPs (Gonçalves et al., 2017), probably is a result of the structural disorder induced to α-Ni(OH)₂ by cerium (III) cations (Gonçalves et al., 2017).

Complementarily, the robustness of the modified electrodes was evaluated during 1,000 consecutive GCD processes at 50 A g⁻¹ in the 0.0 to +0.60 V vs. Ag/AgCl (3.0 mol L⁻¹ KCl) range and monitoring the evolution of their specific charge as a function of the number of cycles (Figure 4C). In fact, similar to the results obtained by CV, the Ni_0.9V_0.05Ce_0.05-LDH/FTO electrode showed an increase in the specific charge up to the 100th cycle, retaining approximately 55% of the specific charge at the end of 1,000 cycles. In turn, the α-Ni(OH)₂/FTO electrode shows charge retention of approximately 37% as a result of its conversion to β-Ni(OH)₂, confirming the greater stability of the trimetallic hydroxide in relation to the GCD cycles.

Considering the application, for example, in HSCs, the positive electrode materials should possess high specific capacity with a good rate capability. Interestingly, the plots shown in Figure 4D indicates that the maximum specific charge for Ni_0.9V_0.05Ce_0.05-LDH/FTO is 740 C g⁻¹ at 10 A g⁻¹ (discharge process) and displays the charge retention of 68.7% at 100 A g⁻¹, demonstrating that the trimetallic NPs can be able to withstand discharge rates as high as 100 A g⁻¹. As expected, the α-Ni(OH)₂/FTO electrode showed the maximum specific capacity at the lowest applied current density (712 C g⁻¹ at 5 A g⁻¹), retaining only 55.4% of specific capacity at 100 A g⁻¹. Furthermore, the electrochemical signature of both electrode materials show GCD curves characterized by a plateau (Figures 4E, F), typical of battery-type materials (Gonçalves et al., 2020b).

The electrochemical behavior of the trimetallic Ni_0.9V_0.05Ce_0.05-LDH was compared with other materials based on bimetallic and trimetallic hydroxide described in the literature and listed in Table 1. For instance, the maximum specific capacity for Ni_0.9V_0.05Ce_0.05-LDH/FTO is comparable with trimetallic α-NiCoMn−OH with a flower-like structure reported by Zhu et al. (2019) (757 C g⁻¹ at 1 A g⁻¹) but with better charge retention capacity. In fact, the Ni_0.9V_0.05Ce_0.05-LDH comparatively presented high specific capacity at much larger current densities without any conductive‐based material, as shown in Table 1. It is also relevant to mention that this excellent performance at high energy densities may be related to nanostructuration that leads to fast redox kinetics due to improved surface area and short diffusion pathways of the electrolyte in the NPs (Gonçalves et al., 2016).

The CVs showing high currents at potentials above 0.5 V vs. Ag/AgCl (Figures 4A, B) and the smaller specific discharge capacity at 5 A g⁻¹ compared with 10 A g⁻¹ (aforementioned, Figure 4D) demonstrate that the trimetallic hydroxide presented in this work also provides electrocatalytic activity for OER and will be parsed as a proof of concept.

This electrocatalytic property can be easily analyzed and monitored by EIS. The Nyquist plot and Bode phase spectra of α-Ni(OH)₂ after successive CVs [CVs performed range in the 0.0 to +0.6 V vs. Ag/AgCl _(3.0 mol L ⁻¹ _KCl)] display two semicircles (Figure 5A). The first activated process was assigned to the Ni²⁺(OH)₂/Ni³⁺(OOH) electrochemical reaction (at high frequencies, 500–30,000 Hz), whereas the second one was associated with the OER catalyzed by the NPs (at lower frequencies, 0.1–100 Hz). It is interesting to highlight that the impedance of the OER (second) process decreased significantly after successive CVs, as confirmed by the decrease of the corresponding semicircle diameter and a simultaneous displacement of the associated peak in the Bode phase plot to higher frequencies (Figure 5B), as a result of the increased activation of a greater number of active sites and the improvement of the electrolyte diffusion. In comparison with pure α-Ni(OH)₂, the higher stability of the Ni_0.9V_0.05Ce_0.05-LDH NPs is confirmed by the unchanging of the diameter assigned to the Ni²⁺(OH)₂/Ni³⁺(OOH) electrochemical reaction at high frequencies and series cell resistance (R_s) (Figures 5C, D), as already noted in the CV experiments (Figures 4A, B). In fact, for the pure α-Ni(OH)₂ (Figure 5A), the R_s increases quite significantly and may be related to the inherent resistance of the material due to the transition from the alpha to the beta phase.

To confirm proof-of-concept of electrocatalytic performance of the prepared NPs, linear sweep voltammetry (LSV) measurements were carried out in 1.0-M KOH electrolyte using a standard three-electrode system (Figure 6A). The Tafel slope shown in Figure 6B derived from LSV curves is also an important parameter to evaluate the electrocatalyst reaction kinetics, which was calculated by the plot of potential against log (J) (Ahn et al., 2021). The Ni_0.9V_0.05Ce_0.05-LDH electrocatalyst showed a Tafel slope of 47 mV dec⁻¹, whereas that for the α-Ni(OH)₂ was 46 mV dec⁻¹, Tafel slope values similar to those reported for other NiV-LDHs (Liu et al., 2019; Bao et al., 2020), NiCoMn-LDH (Liu et al., 2021), NiCoFe-LDH (Zhang et al., 2020), and NiFeV-LDHs (Dinh et al., 2018; Li et al., 2018; Wang et al., 2019a) based electrocatalysts, indicating that both electrocatalysts are promising for OER.

Ni_0.9V_0.05Ce_0.05-LDHs were synthesized (as described in the experimental section) and used in SPE modification aiming to improve the electrochemical detection of hydrogen peroxide (H₂O₂) found at mouthwash sample (purchased in a local market). Initially, to achieve an improvement in the response of the modified SPE, it was submitted to a well-known electrochemical treatment process using NaOH 0.5 mol L⁻¹ by first applying +1.4 V for 200 s, followed by the potential of −1.0 V for another 200 s (Rocha et al., 2020). After the electrochemical treatment mentioned earlier, an increase of around 30% in the reduction current of 15 mmol L⁻¹ H₂O₂ (background electrolyte: KCl 0.1 mol L⁻¹) was observed.

Subsequently, H₂O₂ was determined using the BIA technique with amperometric detection (Rocha et al., 2018b). The parameters that affect this technique, such as detection potential, volume injected, and dispensing speed, were carefully evaluated. The first parameter to be evaluated was the best working potential. This potential was chosen through hydrodynamic voltammetry (Supplementary Figure S6). In this experiment, potentials in the range between 0 and −1.0 V were evaluated. As can be seen in the Supplementary Figure S6, −0.8 V provided the highest current intensity and, in addition, the lowest standard deviation among all other potentials. Therefore, this potential was chosen for future measurements. After determining the better working potential, the other parameters related to the BIA technique were evaluated (volume and dispensing speed). Supplementary Figure S7 displays that when a volume of 150 μl and an injection speed of 120 μl s⁻¹ were used, greater current intensities were achieved, as well as lower standard deviations. These conditions were adopted in the following measurements.

After selecting the better parameters for performing the BIA technique, a study to establish a good linear range for the H₂O₂ detection in mouthwash liquids was carried out (Figure 7). Based on the result observed from six increasing H₂O₂ concentrations, a linear range up to 60 mmol L⁻¹ was obtained following the equation: i = −2.2–4.9C, where i is the current (in μA), and C is the concentration of H₂O₂ (in mmol L⁻¹). The coefficient of correlation (R) was calculated as 0.997. The relative standard deviation found for the concentrations of 5 and 20 mmol L⁻¹ were 7.6 and 5.7%, respectively, and a LOD value of 24.3 μmol L⁻¹ was estimated following the International Union of Pure and Applied Chemistry recommendations.

To show the versatility of the proposed material, the determination of H₂O₂ was performed in a real complex mouthwash sample. A concentration of 25.8 ± 0.3 mmol L⁻¹ H₂O₂ was found in the sample, a concentration that is very close to that reported by the fabricant (25 mmol L⁻¹) and closer to the expected value than the results related in previous work (Katic et al., 2019). Additionally, a recovery study was carried out using the standard addition method, achieving a satisfactory recovery value of around 98%. Table 2 summarizes the results obtained.