Sparked ZnO nanoparticles-based electrochemical sensor for onsite determination of glyphosate residues

ZnO-NPs were synthesized by the sparking method in an atmospheric system at room temperature as shown in figure 1. The four pairs of pure zinc wires (purity 99.97%) with a diameter of 0.38 mm were used as anodes and cathodes of sparking tips. First, the experiment started with cleaning the glass substrate by sonication in deionized water, acetone, propanol, and deionized water and dried at room temperature, respectively. Next, the sparking tips of pure zinc wires were set over the 3 ml of deionized water contained in the glass substrate for a distance of 2 mm. The distance between the sparking wires of the anode and cathode was set at 1 mm. Then, a constant voltage of 7.90 V and a current of 3.87 A were set in the sparking machine to form the ZnO-NPs for 3 h. It should be noted that these specific voltage and current values were chosen based on the circuit design of the sparking machine, which incorporated 25 nF capacitor [34]. This configuration was intended to create a strong electric field (internal electric field approximately 10 kV cm⁻¹) capable of sparking the Zn tips (melting temperature of ∼419.5 °C, deposition rate of ∼0.7 nm spark⁻¹, 3 s spark⁻¹ under normal atmosphere) [31]. The 3 h timeframe was selected to ensure the production of a sufficient quantity of ZnO-NPs in water, enabling the deposition of over 100 electrochemical sensors in a single synthesis process. Finally, the ZnO-NPs suspension was loaded in a test tube screw cap for further use at room temperature.

Zoom In Zoom Out Reset image size

The modification of the SPAgE sensor with ZnO-NPs is displayed in figure 2. First, 3 ml of ZnO-NPs suspension was sonicated for 30 min. Next, 75 μl of the PDDA solution was added into the ZnO-NPs suspension and further sonicated for 30 min. It should be noted that PDDA acts as the binder between ZnO-NPs and SPAgE to enhance the adhesion property only. Based on CV test, the PDDA did not show any significant peak and effects on glyphosate sensing as demonstrated in figure S1 in the supporting Information. Then, 3 μl of ZnO-NPs/PDDA solution was drop-casted onto the surface of the bare working electrode (Area = 7.065 mm²) three times and dried at room temperature to get the ZnO-NPs/PDDA/SPAgE sensor.

Zoom In Zoom Out Reset image size

The structural morphology of the ZnO-NPs was characterized by high magnification and high-resolution transmission electron microscopy (TEM and HR-TEM, JEOL model: JEM-3100 (HR)). The crystallographic structures of ZnO-NPs were analyzed by x-ray diffraction (XRD, model: Bruker D8 Advance) and a diffractometer with Cu Kα radiation (λ = 1.54 A°) in 2θ range of 20–80°. The chemical and valence states of the ZnO-NPs were examined by x-ray photoelectron spectrometry (XPS, model: Axis Ultra DLD, Kratos analytical, Manchester UK).

To investigate the electron transfer reaction mechanism, the CV, DPV, and CA measurements were conducted with a Sensit Smart potentiostat from PalmSens BV. The performance of CV was analyzed for glyphosate by applying a potential range of −0.8 V to +0.8 V and a scan rate of 50 mVs⁻¹ in 10 mM PBS as an electrolyte solution. Data were collected and processed using PSTrace software. All electrochemical measurements were carried out by covering the surface electrode of the sensor with 50 μl of samples and were repeated three times. In addition, a fresh sensor was used to prevent contamination for each measurement.

Figures 3(a)–(e) shows the formation of ZnO nanoplate due to the agglomeration of ZnO-NPs causing a long time of sparking process. As shown in figure 3(e), the lattice spacing of the crystalline structure of the ZnO-NPs is 0.27 nm which corresponds to the (100) plane of the hexagonal wurtzite ZnO crystal [35, 36]. The EDX elemental mapping (figures 3(b) and (c)) and EDX (figure 3(f)) of the ZnO-NPs structures confirm the main elements and good distribution of Zn and O atoms with the atomic ratio of 36.09:63.91 (Zn:O) for synthesized ZnO-NPs. Therefore, it can be verified that ZnO-NPs were successfully prepared by our proposed sparking method.

Zoom In Zoom Out Reset image size

The XRD pattern of ZnO-NPs was recorded in the 2θ range of 20–80° as displayed in figure 4(a). It exhibits peaks positioned at 33.05°, 34.51° and 36.37° corresponding to (100), (002), and (101) ZnO planes (JCPDS card with no. 01-075-1526), respectively. The presence of these peaks confirms a typical hexagonal wurtzite structure of synthesized ZnO-NPs via the sparking method. The crystallite size of ZnO-NPs was calculated to be 27.20–55.60 nm corresponding to the peaks positioned at 33.05°, 34.51° and 36.37° using the Scherrer equation by equation (1). Therefore, the average crystallite size is found to be 39.77 nm. It should be noted that Williamson–Hall (W–H) plot can be also used to identify the crystallite size. However, when considering previous research, it was found that Scherrer's formula provided a more fitted estimation of the crystallite size for the synthesized ZnO-NPs in comparison to the W–H analysis [37]. Therefore, the Scherrer equation was chosen as the preferred method for this study to determine the crystallite size by using equation (1).

$$\begin{eqnarray}&&D=\displaystyle \frac{k\lambda }{\beta \,\cos \,\theta },\end{eqnarray} \tag{ 1 }$$

Zoom In Zoom Out Reset image size

where $D$ is the average crystallite size, $k$ is the constant value of 0.9, $\lambda$ is the wavelength of $Cu\,K\alpha$ line of $1.5405\,{\rm{{\rm A}}}^\circ ,$ $\beta$ is the full width at half maximum (FWHM) and $\theta$ is the Bragg's diffraction angle.

To understand the formation mechanism of ZnO-NPs via the sparking method, the high voltage (E_app) was applied to pure zinc wires. At the end of the zinc wire, the high temperature and pressure will be generated by the bombardment of electrons and ions in a very short time. This phenomenon leads to melting the zinc wire at sparking tips and also oxidizing in the air to form the ZnO. Then, the droplets of ZnO-NPs fall from the sparking tips into the water. A fast change of temperature by water can cause condensation to form the ZnO nanoplate as displayed in the TEM image (figure 3(a)). In addition, the applying energy (E_app) to metal wires and the synthesis time can affect particle size. Traiwatcharanon et al [38] reported that the size of the synthesized particle was ∼23 nm for a sparking time of 45 min. In this work, the average size of nanoparticles increases significantly to ∼40 nm for a sparking time of 3 h. The effect can be explained by equation (2) [39].

$$\begin{eqnarray}&&{E}_{app}={E}_{loss}+m{C}_{p}\left({T}_{m}-T\right)+m{L}_{f},\end{eqnarray} \tag{ 2 }$$

where ${E}_{app}$ is input energy that heats the zinc wires, ${E}_{loss}$ is energy loss, $m$ is the mass of particles, ${C}_{p}$ is heat capacity, ${T}_{m}$ is melting point temperature and ${L}_{f}\,$is the heat of the fusion of metallic tips.

From equation (2) , the right-hand side term is the energy required for bringing the material from room temperature $\left(T\right)$ to the vapor phase. Therefore, the energy was separated into the first term energy loss $\left({E}_{loss}\right)$ to an environment which was expressed in equation (3) and the second term is the melting energy for zinc wires from room temperature $\left(T\right)$ to the melting point $\left({T}_{m}\right)$ and the last is the enthalpy of melting. The mass of particles can be expressed by equation (3).

$$\begin{eqnarray}&&{E}_{loss}=\,\pi {r}^{2}\sigma \tau {T}_{b}^{4}+2\pi r\tau k\left({T}_{b}-T\right),\end{eqnarray} \tag{ 3 }$$

where $r$ is the radius of the spark or the radius of the circular heated spot, $\tau$ is spark duration, $m$ is the mass of particles, $k$ is the thermal conductivity of the zinc wire, ${T}_{b}\,$is boiling point temperature and $\sigma$ is Stefan's constant $5.67\times {10}^{-8}\,{\mathrm{Wm}}^{-2}{{\rm{k}}}^{-4}.$

From equation (3), the right-hand side term is heat loss of the heated spot by radiation and thermal conduction, respectively. Part of Stefan's constant will not be considered due to the metal heat radiation emitted less than the black body. The effect of time can be described by the agglomeration state. The nanoparticle can grow bigger with increasing time. At a longer time, the temperature will increase resulting in the size of nanoparticles growing and agglomerating.

To confirm the chemical compositions and valence states of the synthesized ZnO-NPs, x-ray photoelectron spectroscopy (XPS) was used as displayed in figures 4(b)–(d). The XPS survey scan (figure 4(b)) observes the dominant characteristic of Zn and O peaks as major components with no obvious impurity in ZnO-NPs. The devolution of the Zn 2p peak (figure 4(c)) indicates two symmetric peaks at binding energies of 1021 eV and 1044 eV corresponding to the Zn 2p_3/2 and 2p_1/2, respectively. Based on the binding energy of Zn 2p, the separation peak was 23 eV indicating that the chemical valence of the oxidation state at the surface of the ZnO-NPs was +2 (Zn²⁺) [40, 41]. The deconvolution of O 1s core level peak (figure 4(d)) which were subdivided into four asymmetric peaks at a binding energy of 530 eV, 531 eV, 532 eV, and 533 eV, respectively. The O 1s state consists of three specific binding energy peaks at low binding energy state peak (LP), middle binding energy peak (MP), and higher binding state peak (HP) corresponding to O_{Lat (I)}, O_{Vac (II),} and O_{Ads (III)}, respectively [42]. Therefore, the binding energy of 530 eV can be attributed to the lattice oxygen anions $\left({{\rm{O}}}_{2}^{-}\right)$ in the oxidation state on the wurtzite structure of ZnO. The 531 eV peak associated with the oxygen vacancies or defects in the lattice ${{\rm{O}}}_{x}\,\left({{\rm{O}}}_{2}^{-},\,{{\rm{O}}}^{-}\right),$ and the 532 eV peak is related to the adsorbed oxygen (chemisorbed oxygen) at the surface [43–50]. It can be observed the initial O 1s peak has FWHM as 2.94 eV after the fit curve it decreases to 1.43 eV for each peak. The possible reason is the chemisorbed oxygen (O_Ch) on the Zn surface. For the higher binding energy than LP, the MP and HP peaks are associated with oxygen vacancies and surface hydroxide groups. From the XPS results, it is clear that the ZnO-NPs were successfully formed and hydroxide groups may enhance the interaction for the adsorption of glyphosate molecules.

To further investigate the sensing materials on SPAgE electrodes, the XRD patterns of ZnO-NPs/SPAgE and ZnO-NPs/PDDA/SPAgE are shown in figure S2 in the Supporting Information. The ZnO-NPs maintain their wurtzite phase after coating onto SPAgE electrodes. Addition of PDDA binding agent (ZnO-NPs/PDDA/SPAgE) does not exhibit any noticeable peaks that would indicate a significant impact on the phase and structure properties of the main ZnO-NPs sensing material. However, PDDA is still present on SPAgE electrode since N bonding occurs based on XPS analysis as shown in figure S3 in the Supporting Information. It should be emphasized that PDDA has no significant effects on glyphosate sensing (figure S1 in the Supporting Information). Its purpose is solely to improve the bonding between ZnO-NPs and SPAgE. Therefore, ZnO-NPs remain the primary sensing material for glyphosate detection.

To obtain the high sensitivity of the ZnO-NPs/PDDA/SPAgE sensor for the detection of glyphosate, the parameters including pH of the supporting electrolyte solution (PBS solution) and scan rate should be studied. The peak of oxidation current $\left({I}_{pa}\right),$ reduction current $\left(\,{I}_{pc}\right),$ oxidation peak potential $\left({E}_{pa}\right)$ and reduction peak potential $\left({E}_{pc}\right)$ can be directly extracted from CV experiments based on a standard technique [51].

3.2.1. Effect of pH

The electrocatalytic performance of the modified SPAgE sensor with ZnO-NPs was examined by CV using 10 mM of PBS solution containing the glyphosate of 100 μM at a scan rate of 50 mV s⁻¹. The influence of electrolyte solution on the redox (oxidation and reduction) reaction of glyphosate was investigated by varying the pH values of PBS solution from 5.0 to 9.0 as displayed in figure 5(a). As figure 5(b), it can be explained in two main points: firstly, it was observed that a gradual increase in the cathodic peak current $\left({I}_{pc}\right)$ and reached maximum current at pH 6.0. After that, the pH value was continuously increased until pH 9.0 the cathodic peak current began to decrease because the molecules of Gly coordinate with Zn²⁺ leading to the Gly-Zn²⁺ complex and the chelating ligands of molecules of amino acids. Secondly, it was observed that the reduction in peak potential $\left({E}_{p}\right)$ shifts to more negative values with the increase of pH value due to the intervention of protons between Gly and ZnO on the working electrode [52]. The protons can participate in the reduction reaction. Therefore, pH 6.0 was selected to be an optimum parameter for this study.

Zoom In Zoom Out Reset image size

In addition, the linear regression equations of reduction peak potential $\left({E}_{pc}\right)$ and pH values were found to be:

$$\begin{eqnarray}&&{E}_{pc}\left(mv\right)=-0.0226\left(pH\right)+0.0214\,;{R}^{2}=0.986.\end{eqnarray} \tag{ 4 }$$

3.2.2. Effect of Scan rate

To describe the kinetics of the ZnO-NPs/PDDA/SPAgE sensor reactions, the CV was recorded at different scan rates over the range from 10 to 50 mV s⁻¹ in PBS solution (pH 6.0) containing 100 μM of glyphosate as displayed in figure 6(a). It can be seen that (I) The peak of oxidation/reduction current $\left({I}_{pa}\,{\rm{and}}\,{I}_{pc}\right)$ increases as a function of scan rates. (II) The oxidation peak potential $\left({E}_{pa}\right)$ tends to shift to the positive direction value. At the same time, reduction peak potentials $\left({E}_{pc}\right)$ also shift to more the negative direction value. In other words, it can be observed that the peak-to-peak separation $\left({\rm{\Delta }}{E}_{p}\right)$ increases (from 39 to 86 mV) with scan rate that indicates a quasi-reversible process and implies slow electron-transfer kinetics in the system.

Zoom In Zoom Out Reset image size

The corresponding plot of the ${I}_{pa}\,{\rm{and}}\,{I}_{pc}$ between scan rates is displayed in figure 6(b). It indicates that both reactions exhibit good linearity with over-increasing scan rates as regression coefficient (R²) 0.971 and 0.990. The linear relation between the cathodic peak current and square roots of scan rates as displayed in figure 6(c), the regression equation, and the correlation coefficient can be expressed as followed

$$\begin{eqnarray}&&{I}_{pc}\left(\mu {\rm{{\rm A}}}\right)=-261.77{v}^{1/2}\left({\rm{mV}}/{\rm{s}}\right)-45.340;{R}^{2}=0.9957.\end{eqnarray} \tag{ 5 }$$

This indicates that the electrochemical processes in the ZnO/PDDA/SPAgE sensor occur by the diffusion-controlled process. Moreover, the plot obtained by the plotting between the relationship logarithm of scan rates and cathodic peak current is given by equation (6).

$$\begin{eqnarray}&&\mathrm{log}\,{I}_{pc}\,\left(uA\right)=\,-0.4701\left(\mathrm{log}\,\nu \right)-2.3967\,;\,{R}^{2}=0.9935.\end{eqnarray} \tag{ 6 }$$

According to the theory, the slope value should be 0.50 which is related to the diffusion-controlled process. From this result, the slope is 0.47 which is close to the theoretical value. In addition, the relation between logarithms at scan rates and ${E}_{pc}$ (Tafel plot) is expressed as

$$\begin{eqnarray}&&{E}_{pc}\left(V\right)=-0.1655\left(\mathrm{log}\nu \right)-0.0946;{R}^{2}=0.9979.\end{eqnarray} \tag{ 7 }$$

Based on CV experiments with various pH values and scan rates, the reduction potentials of glyphosate are in the range of −0.26 to −0.42 V. It should be noted that the reduction potential of glyphosate is influenced by multiple factors including pH, temperature, scan rate, the choice of sensing material, and the specific electrode employed in the electrochemical measurement. Different studies on glyphosate detection can give the different reduction potentials [53–55].

Based on the evidence of pH and scan rates conditions, the quantitative analysis for the determination of glyphosate on the ZnO-NPs/SPAgE sensor was investigated with a varied concentration level from μM (0–700 μM) to mM (1 –5 mM) using the differential pulse voltammetry (DPV) technique as displayed in figure 7(a). The reduction peak currents of the ZnO-NPs/PDDA/SPAgE sensor decrease continually with increasing glyphosate concentration. This phenomenon may be attributed to the complex formation between glyphosate and Zn²⁺ resulting in the limiting electron transfer leading to a decrease in the peak current. At low concentrations (0–700 μM), the surface of the ZnO-NPs/PDDA/SPAgE sensor can well absorb glyphosate molecules. Meanwhile, glyphosate molecules were suffocated to absorb on the ZnO-NPs/PDDA/SPAgE sensor at high concentrations (1–5 mM). Therefore, they shows two linear ranges obtained from the current response versus different concentrations of glyphosate as shown in figure 7(b). At lower concentrations (0–700 μM), the regression equation is I_pc (μA) = 0.0082 × C_Gly(μM) – 14.5670 with R² = 0.9354. At high concentrations (1–5 mM), the regression equation is I_pc (μA) = 0.00085 × C_Gly(μM) − 9.4579 with R² = 0.9962. For the potential, it can be observed that the potential peak of ZnO-NPs/PDDA/SPAgE sensor in the absence of glyphosate (0 μM) is −0.36 V. As the concentration of glyphosate increases, there is a slight shift in the potential, eventually reaching a convergence point at −0.41 V. It is well known that more glyphosate molecules need more energy for reduction process, resulting in causing a progressive shift towards more negative values in the reduction potential at high glyphosate concentration. To compare the reduction potential from CV measurements at the same glyphosate concentration, the reduction potential obtained from the DPV technique is well agreement with the reduction potential conducted using the CV (∼ −0.4 V).

Zoom In Zoom Out Reset image size

Figure 7(c) shows the calibration curve between the changing peak currents (ΔI = I_{no Gly} – I_Gly) and the logarithm of glyphosate concentration. The regression equation is ΔI (μA) = 4.483 Log C_Gly (μM) − 6.877 with a correlation coefficient of R² = 0.9916. Moreover, the limit of detection (LOD) efficiency is 2.84 μM, which was determined according to following the equation (8)

$$\begin{eqnarray}&&LOD=\displaystyle \frac{3\times S{D}_{b}}{m},\end{eqnarray} \tag{ 8 }$$

where SD_b is the standard deviation of three blank measurements and m is the slope calculated from the calibration curve.

To compare the performance of the ZnO-NPs/PDDA/SPAgE sensor with previous works, the ZnO/PDDA/SPAgE sensor exhibits a wide range detection and low LOD for glyphosate detection as displayed in table 1.

The interference was investigated by a chronoamperometry technique (CA) as displayed in figure 8(a). The CA response was detected by adding other interferences of herbicides (10 mM) like glufosinate-ammonium (Glu-Am), paraquat (PQ), and butachlor-propanil (BP). The performance of the ZnO-NPs/PDDA/SPAgE was studied at a fixed reduction potential applied during measurement $\left({E}_{pc}\right)$ of −0.36 V during the time interval of 360 s. A new interference was added every 60 s in PBS solution (pH 6). After adding the interference as shown in figure 8(b), it was found that the current slightly decreases comparing with the original current of pure glyphosate. The percentage of recovered data (% Recovery) with other interference (glufosinate-ammonium, paraquat, butachlor-propanil) is in the range of 81.57%–95.64%. The results indicate that the ZnO-NPs/PDDA/SPAgE sensor exhibits a high selectivity to glyphosate with less effect on interference of popular herbicides.

Zoom In Zoom Out Reset image size

In order to assess the additional capabilities of the ZnO/PDDA/SPAgE sensors, a set of 30 sensors was manufactured and stored in a closed room at room temperature without controlling the humidity. Among these sensors, 3 sensors were utilized to detect 1 mM glyphosate while the remaining 3 sensors were were dedicated to detecting 5 mM glyphosate using the DPV technique. This experimental process was repeated every 2 days for a duration of 8 days. The average oxidation peak currents along with their corresponding error bars are depicted in figure S4 in the supporting information. Analysis of the results reveals no significant deviation in the current values. The RSDs were found to be below approximately 5%. As a result, these findings confirm the good repeatability, reproducibility, and stability of the ZnO/PDDA/SPAgE sensor.

The possible sensing performance of ZnO-NPs/PDDA/SPAgE sensor for glyphosate is shown in figure 9. The sensing mechanism can be separated into 2 phases as follow;

• Solid phase: at the surface of the electrode, electrons within solid material require higher energy relative to the vacuum level to overcome the surface barrier of the material to create free electrons. When the electrode surface is modified with ZnO-NPs/PDDA material, the Fermi energy level changes to an equal level. Afterward, electrons will transfer from higher to lower Fermi energy levels until the two systems reach to equilibrium and form a new Fermi energy level.
• Liquid phase: in general, glyphosate is a weak acid due to its three dissociation constants at pK₁ 2.27–2.35, pK₂ 5.58–5.89, and pK₃ 10.25–10.89. The solution of glyphosate contains three functional groups such as amines, carboxylates, and phosphonates which behave as donor groups. The pathway of protonation reaction of deprotonated glyphosate starts from one oxygen atom from phosphonates groups, nitrogen atom from the amine group, and oxygen atom from carboxyl, respectively. The phosphonates group is the most reactive to directly bond with the surface of the material.

Zoom In Zoom Out Reset image size

The work functions of silver and zinc were 4.7 eV and 5.3 eV, respectively. Contact of the ZnO-NPs/PDDA/Ag will induce electron flow from ZnO to Ag electrode because of the changing of Fermi energy level and formation of new a Fermi energy level. The interactions between ZnO-NPs/PDDA/SPAgE and ligand of glyphosate are controlled by the protonation/deprotonation process. In a solution under the influence of the electrostatic force, the negatively charged donor groups especially phosphonates carboxylates in glyphosate are adsorbed on the surface of positively charged Zn²⁺. The binding of glyphosate–Zn complexation is expressed in the form of aqueous; ZnL⁻ or ZnHL complexes (L represents the glyphosate ligand) disrupt electron mobility leading to a decrease in the response current. The decrease in current response is relative to the number of glyphosates absorbed onto the surface of the ZnO-NPs/PDDA/SPAgE sensor.

In addition, the dissociation of glyphosate depends on the pH value of the solution. The phosphonates groups can protonate to the ZnO in all pH. In the typical case of pH < pK_a1, the COOH group in glyphosate in form (I) dissociates to COO⁻ ion as shown in form (II). For pK_a1 < pH < pK_a2, the glyphosate molecules in form (I) disappeare and transform to form (II) and (III). In case of form (III), it was found that the proton-donor (P-OH) dissociates into the proton-acceptor (P-O⁻) and there is still one proton left which is protonated to nitrogen (NH²⁺). The interaction of the carboxylate group with the ZnO surface is discovered to be a weak reaction. Above the pH > pK_a3, the positive charge (the amino group) has remained protonation but the phosphonates and carboxylate groups will completely be dissociated and precipitated.

Considering the variation in permissible levels of glyphosate contaminants in drinking water across countries [63, 64], a concentration of 4 μM of glyphosate was intentionally added to real samples including green tea, corn juice and mango juice. These real samples were purchased from a supermarket in Thailand. The DPV technique was employed to measure the reduction peak currents of ZnO-NPs/PDDA/SPAgE sensors upon the real sample with 4 μM glyphosate contamination. The resulting currents were then utilized in a linear calibration curve (figure 7(b)) to determine the concentration of glyphosate. The analytical results are summarized in table 2. The percentage recoveries of glyphosate detection in the real samples vary from 103.8% to 126.8% with a % RSD between 3.5% and 5.1%. These values indicate a reliable estimation of glyphosate concentration in the real samples. However, in the case of corn juice, there appears to be a slight overestimation of the glyphosate concentration. This discrepancy could be attributed to additional ingredients such as creamer or milk in instant corn juice, potentially causing a higher percentage recovery compared to other real samples.