ZnO-nanostructured electrochemical sensor for efficient detection of glyphosate in water

ZnO-nanostructured electrochemical sensor for efficient detection of glyphosate in water Zorica Novakovic a , Zorka Z. Vasiljevic b , Maria Vesna Nikolic b , Nenad B. Tadic c , Tijana Djordjevic d , Marko Radovic a , Ivana Gadjanski a , Sneˇ zana Papovi´ c e , Filip Vlahovic f , Dalibor Stankovic g , Jasmina Vidic h,* a University of Novi Sad, BioSense Institute, Dr Zorana Djindjica 1a, 21000 Novi Sad, Serbia b University of Belgrade - Institute for Multidisciplinary Research, Kneza Viseslava 1, 11030 Belgrade, Serbia c University of Belgrade, Faculty of Physics, Cara Dusana 13, 11000 Belgrade, Serbia d Institute of Pesticides and Environmental Protection, Banatska 31b, 11080 Belgrade, Serbia e University of Novi Sad, Faculty of Science, Trg Dositeja Obradovica 3, 21000 Novi Sad, Serbia f Institute of Chemistry, Technology and Metallurgy, National Institute of the Republic of Serbia, University of Belgrade, Njegoseva 12, 11000 Belgrade, Serbia g Faculty of Chemistry, University of Belgrade, Studentski trg 12-16, 11000 Belgrade, Serbia h Universit´ e Paris-Saclay, INRAE, AgroParisTech, Micalis Institute, UMR 1319, 78350 Jouy-en-Josas, France ARTICLE INFO Keywords: Pesticides Eco-friendly ZnO nanoparticles Sensor Environmental water Computational modeling ABSTRACT Glyphosate is a widely used broad-spectrum herbicide for controlling grassy weeds, despite having potential health hazards. Herein, we report on a solid-state electrochemical sensor based on ZnO nanoparticles (ZnO NPs) for on-site detection of glyphosate. Accordingly, ZnO NPs was drop-cast on the surface of a disposable screenprinted carbon electrode. Eco-friendly ZnO NPs of only 7 nm crystallite sizes were obtained by green sol-gel synthesis using lemon (Citrus limon) waste aqueous extract as the green reducing and capping/stabilizing agent and Zn nitrate precursor as evidenced by scanning electron microscopy (SEM), transmission electron microscopy (TEM), X-ray diffraction and diffuse reflectance. SEM confirmed successful electrode functionalization with the synthesized nanoparticles. Under laboratory conditions in acetate buffer (pH 5), the sensor demonstrated excellent selectivity and sensitivity, with a detection limit of 0.648 µM, a wide linear detection range (0.5 µM to 7.5 mM), and a rapid detection time of 30 min. When tested in river water, the sensor achieved a detection limit of 0.96 µM using differential pulse voltammetry. It also exceptionally tolerated interference from similar organophosphorus compounds and ions commonly found in river water. The excellent detection performance of the sensor was attributed to the strong coordination interactions between Zn atoms and phosphonate/carboxylate groups that are enhanced by a hydrogen bond at acidic pH, as determined by chemical calculations. This disposable sensor offers a cost-effective, efficient, and environmentally friendly solution for monitoring glyphosate in water systems. 1. Introduction Glyphosate, N-(phosphonomethyl)glycine, is one of the most effective and the most widely used non-selective herbicide, accounting for nearly 72 % of global pesticide use [1]. Glyphosate persists in the environment after penetrating underground water reservoirs and, therefore, drinking water. Human exposure to glyphosate has become, hence, an everyday occurrence in almost all populations. Initially characterized as ’virtually nontoxic’ to animals and humans by Monsanto, which first commercialized it as Roundup® in 1974, glyphosate was classified as a probable carcinogen by the International Agency for Research on Cancer in 2015. Moreover, glyphosate is suspected of impacting the human endocrine, nervous, cardiovascular, and reproductive systems, it can compromise liver function, cause renal impairment, skin and eye irritation, and gastrointestinal problems [2–7]. The US Environmental Protection Agency guidelines allow a maximum of 4.1 μ M of glyphosate in drinking water, whereas The European Union (EU) guidelines allow a maximum of 0.6 nM [8,9]. * Corresponding author. E-mail address: [email protected] (J. Vidic). Contents lists available at ScienceDirect Talanta Open journal homepage: www.sciencedirect.com/journal/talanta-open https://doi.org/10.1016/j.talo.2025.100481 Received 28 April 2025; Received in revised form 11 May 2025; Accepted 24 May 2025 Talanta Open 12 (2025) 100481 Available online 25 May 2025 2666-8319/© 2025 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). Because of the legislations, and potential hazardous effects, a robust and easy-to-use analytical method for glyphosate quantification in biological and environmental samples is in rising demand. Detecting and quantifying glyphosate is challenging due to its low volatility, high solubility in water, and lack of chromophoric groups [10]. Thus far, different analytical approaches have been proposed including chromatography [11], spectroscopic methods [12–14], nuclear magnetic resonance [15], and mass spectrometry [16]. To achieve a good selectivity at low concentrations and high sensitivity, these techniques are often associated with separation methods such as gas or liquid chromatography and require a derivation step [17]. Moreover, they are costly, require time-consuming procedures and highly trained operators, and therefore cannot be used for routine on-site analysis. Consequently, to replace bulk instruments, different electrochemical sensors for glyphosate detection have been proposed in recent years [17]. Electrochemical devices can be easily miniaturized into portable kits that require only a few processing steps and minimum sample manipulation, thus rendering them ideal for undemanding on-site detection [18–20]. For instance, electrochemical signals obtained by glyphosate inhibiting biochemical reactions or binding to biomolecules (enzymes, aptamers or antibodies) allow its label-free quantitative detection [21–23]. Electrochemical platforms have also been proposed that enable glyphosate recognition using specific polymers or nanoparticles. For this, metal-organic frameworks (MOF) [24,25], covalent organic framework (COF) [18], molecularly imprinted polymers (MIP) [26], quantum dots [27], or conductive polymers (CPol) [28] were used to enhance sensor sensitivity. Computer modeling indicated that glyphosate can make stable complexes with such materials at the minimum binding free energy through ionic and hydrogen bonds, hydrophobic interactions, and van der Waals forces [17,18]. Furthermore, metals, such as Ca, Mg, Fe, Cu, Mn, and Zn, and their metal oxide nanoparticles (NPs) have been efficiently applied for electrochemical detection of glyphosate in water or food samples [29–31]. Phosphonate and carboxyl groups in glyphosate form coordination bonds with metal cations, while the amine group can contribute to metal complexation at certain pHs [17]. Though providing specific glyphosate detection, each of the above materials has its drawbacks. The focus and challenge of the current research is to use low-toxic glyphosate-binding nanomaterials to produce a stable and homogenous electrode surface, but to avoid passivation or produce nano-waste needed to be treated, and to understand their interaction mechanisms. ZnO nanoparticles (NPs) are a promising candidate for this purpose. Most semiconductor materials can effectively reduce complex organic pollutants [32], and molecular modeling has identified zinc as the most stable element for forming tetrahedral and octahedral complexes with glyphosate [33]. ZnO is a biocompatible semiconductor with high electron mobility, high diffusion coefficient, elevated electric point, wide band gap (3.37 eV), excellent electron transfer ability, and high exciton binding energy. Importantly, ZnO NPs of tunable morphology and properties can be obtained using eco-friendly synthesis methods that do not need harsh reaction conditions but use plant, microbe, and fruits extracts as the reducing/capping agent in the synthesis process [34,35]. For instance, citruses are popular fruits, and their consumption, especially after juice production, leaves waste that can be used in the green sol-gel synthesis of ZnO NPs, as we recently demonstrated [36]. These versatile characteristics make eco-friendly ZnO NPs an excellent option for flexible working electrode surface modification with consistent reproducibility. Moreover, ZnO in form of doped NPs or nanocomposites was successfully applied for electrode modifications in order to enhance their analytical properties [35,37–44] and to enable detection of various compounds such as fungicides [45], drugs [46], herbicides [47,48]. Nevertheless, the efficiency of pure ZnO NPs of sizes <10 nm, that are expected to have a high reactivity due to the high size-to-volume ratio still has to be tested. In this work, a commercial screen-printed carbon electrode (SPCE) was modified by drop casting with ZnO NPs of an average crystallite size of 7 nm and average particle size of 9.53 nm obtained by a green sol-gel method using Citrus limon aqueous extract. Immobilization of such small-sized ZnO NPs through their interaction with carbon provided a larger electroactive surface showing a high affinity to bind glyphosate specifically in both laboratory conditions and in river water. Furthermore, we provide a detailed structural and morphological characterization of the ZnO NPs and computer modeling of their interactions with glyphosate. Compared to other electrochemical platforms for glyphosate detection reported in the past few years, our device exhibits superior sensitivity, a shorter detection time, and a wider dynamic range, indicating its potential effectiveness in developing a safety test for glyphosate detection. 2. Experimental 2.1. Materials and reagents Potassium ferricyanide K 3 [Fe(CN) 6 ], potassium ferrocyanide K 4 [Fe (CN) 6 ], potassium chloride, zinc nitrate hexahydrate (Zn(NO 3 ) 2 ⋅6H 2 O reagent grade, purity ≥98 %), sodium acetate, acetic acid, sodium hydroxide, sodium nitrate, calcium chloride, sodium sulfite, mesotrione (70 %), carbendazim (97 %) and irgasan (97.0–103.0 %) were purchased from Sigma Aldrich (Saint Quentin Fallavier, France). SPCEs (DropSens DRP-110) were purchased from Metrohm (France). Dimethylformamide, DMF (Prolabo, France) was of analytical grade. A stock solution of PBS (10x) was purchased from VWR Prolab (Rosny-sur-Bois, France), while 0.1 M acetate buffer, pH 5, was prepared by mixing 0.06 M CH 3 COONa and 0.03 M CH 3 COOH, and the pH was adjusted using 5 M NaOH. The technical substance (Glyphosate acid), Batch No 230,711–2A, manufactured by Jiangsu Good Harvest - Weien Agrochemical Co., Ltd., China, was used in this study. Glyphosate stated purity of 97.7 % was determined by the manufacturer through a five-batch analysis, as detailed in the accompanying Certificate of Analysis (CoA). Purity was additionally assessed in the laboratory using the high performance liquid chromatography system (Shimadzu Prominence) equipped with a photodiode array (PDA) detector, using an ion exchange column (Zorbax Eclipse XDB C18), and UV detection at 195 nm, and external standardization. The laboratory results corroborated the purity value indicated in the CoA, and for subsequent analytical procedures, the purity value presented in the CoA was adopted. 2.2. ZnO NPs synthesis Spent lemon halves (1 kg) were collected from lemons (Citrus limon) purchased from a local store (Meyer lemon hybrid, imported from Eren Tarim Urunteleri Ltd, Mersin, Turkey). They were chopped finely and boiled in 3 L of water for 30 min. The pale-yellow pectin solution was filtered through filter paper, bottled, and stored in a fridge for further use. ZnO NPs were obtained by adding 2 g of Zn(NO 3 ) 2 ⋅6H 2 O to 42.5 mL of the lemon extract and mixed at 60 ◦C in a water bath for 1 h, followed by calcination of the obtained gel in chamber furnace at 400 ◦C for 1 h, with a heating rate of 10◦/min, representing a modification of the methods described by [36,49]. 2.3. Structural and optical properties of ZnO NPs In order to verify the formation of ZnO NPs, an X-ray diffraction pattern of the produced particles was measured on a Rigaku Ultima IV diffractometer (Tokyo, Japan), range 10◦−90◦, step 0.05 s and acquisition rate 1◦/min. The crystallite size (D XRD ) is determined from the measured diffraction pattern as the slope of the linear fit of the plot of the relation between the crystallite size, strain and peak broadening as [50]: Z. Novakovic et al. Talanta Open 12 (2025) 100481 2 (dhklβhklcosθ)2=K DXRD (d2 hklβhklcosθ)+( ε 2)2(1) where d hkl denotes the D-spacing between adjacent planes, (hkl) represent the Miller indices, β hkl denotes the full peak width at half-maximum intensity, θ represents the Bragg angle, K =0.75 represents the shape of the reciprocal lattice point value applied for spherical particles [50] and ε is the effective strain. To determine the optical bandgap for the synthesized nanopowder, the diffuse reflectance spectrum was measured on a Shimadzu UV-2600 device with an ISR2600 Plus Integrating sphere attachment (Kyoto, Japan) in the range 200–800 nm. 2.4. Transmission electron microscopy (TEM) TEM observations of the synthesized ZnO nanopowder were performed using a FEI Talos F200X electron microscope with an X-FEG source and a maximum accelerating voltage of 200 kV. Both conventional and high-resolution TEM (TEM/HRTEM) imaging was conducted to examine the morphology and crystal structure of the nanopowder, while the phase composition of the sample was analyzed using selected area electron diffraction (SAED) method. For spatially-resolved energy dispersive spectrometry (EDS) analysis, the scanning transmission electron microscopy (STEM) mode was used, utilizing four silicon detectors integrated into the Talos microscope. Electron-transparent samples for TEM examination were prepared by standard procedure, where the solid powder was first dispersed into ethanol and then ultrasonicated for 5 min. A droplet of the suspension was then placed on a carbon-supported Cu grid and allowed to dry in the air. 2.5. Scanning electron microscopy (SEM) Morphological analysis was conducted using an Apreo 2C HighResolution Scanning Electron Microscope (HRSEM, Thermo Fisher Scientific, Waltham, MA, USA). A droplet of dispersed and ultrasonicated ZnO nanopowder was applied to a carbon supported 32-mm aluminum stub, while SPCEs carrying ZnO NPs were mounted onto 32-mm aluminum stubs using carbon tape and characterized at 10 kV accelerating voltage and 0.1 nA probe current, employing both in-lens and Everhart–Thornley (ETD) detectors. Energy-dispersive X-ray (EDX) spectroscopy was performed under the same conditions to confirm the elemental composition. 2.6. Electrode modification For electrode functionalization, suspensions of ZnO NPs in DMF were prepared at concentrations of 0.1 mg/mL, 1 mg/mL, 10 mg/mL, and 20 mg/mL and subjected to 3 h of ultrasonication (Meditech Scientific, Clamart, France). The SPCE was functionalized using the drop-casting method at room temperature, covered with aluminium foil, for different times. For this, SPCEs were mounted onto 32 mm diameter aluminum stubs using carbon adhesive discs. 2.7. Electrochemical measurements Electrochemical measurements were conducted using a PalmSens4 potentiostat/galvanostat/impedance analyzer (PalmSens BV, Netherlands), controlled by the PSTrace voltammetric software (Version 5.9). Electrochemical measurements were performed in 5 mM K 3 [Fe (CN) 6 ]/K 4 [Fe(CN) 6 ] (1:1) as a redox probe with 0.1 M KCl as the electrolyte solution. Cyclic voltammetry (CV) measurements were performed in the potential range from −0.5 V to 1.2 V and at a scan rate of 0.1 V/s. Differential Pulse Voltammetry (DPV) was chosen as the method for the detection and quantification of glyphosate, with a potential ranging from −0.1 to +0.5 V, a pulse amplitude of 0.025 V, and a scan rate of 0.025 V/s. Electrochemical impedance spectroscopy (EIS) measurements were performed in the frequency range from 1 Hz to 100 kHz, with a potential amplitude of +10 mV in 5 mM K 3 [Fe(CN) 6 ]/K 4 [Fe (CN) 6 ] (1:1) as a redox probe in a 0.1 M KCl electrolyte solution. The electroactive surface area was calculated according to the Randles-Sevcik equation for reversible electrochemical process under diffusive control at 25 ◦C: Ip=2.69⋅105⋅A⋅C⋅n3/2⋅D1/2⋅ ν 1/2(2) where A is the electroactive area (cm 2 ), I p is the peak current (A), D is the diffusion coefficient of the redox probe (6.1 •10 −6 cm 2 /s for [Fe (CN) 6 ] 4− in solution), n is the number of electrons transferred, ν is scan rate (0.5 V/s was used) and C is the concentration of the redox probe (mol/cm 3 ). The roughness factor (R f ) of the electrode was estimated as the ratio between the electroactive surface area and its geometric surface area. The surface coverage, Γ, of the working electrode after BSA immobilization was calculated using the following equation Ip=n2F2A υ Γ 4RT (3) where I p is the Faradaic current, ν is the scan rate, A is the surface area of the working electrode (12.64 mm 2 ), n is the number of electrons transferred, F is Faraday’s constant, R is the molar gas constant (8.314 J/ mol⋅K), and T is the temperature (K). The heterogeneous rate constant was calculated according to the equation reported by Randiviir [51]: k∘=RT n2F2ARctC(4) where, R represents the universal gas constant, T is the absolute temperature (298 K), n denotes the number of electrons involved in the redox process (assumed to be 1), F is the Faraday constant, A corresponds to the geometric surface area of the electrode (0.071 cm²), and C is the bulk concentration of the redox probe. 2.8. Computational modelling The first-principles calculations of interaction ZnO NPs -glyphosate were derived within the Density Functional Theory (DFT) [52,53] framework. All relevant properties of the (solid-state) materials were calculated using the Amsterdam Density Functional (ADF) [54] periodic DFT code, BAND [55,56] (Version 2024.1). The computational analyses were executed utilizing the dispersion-corrected PBE-D3 [57] density functional approximation in conjunction with the polarized triple-ζ (TZP) [58] basis set with a small frozen core and good numerical quality [59]. Scalar relativistic effects were included for both core and valence electrons. The long-range dispersion interactions were considered by the DFT-D3 [60] method in all calculations. All periodic Energy Decomposition Analysis (pEDA) [61] calculations were performed with k-space sampling restricted to the Γ-point. In all cases, the surface of the material and glyphosate molecule in appropriate ionic form were considered and modeled as interactive fragments. 2.9. River water sampling, spiking and recovery calculation The applicability of the designed electrochemical nanobased sensing platform was confirmed by testing water samples from the Bievre River, located within Jouy en Josas, France, in the vicinity of agriculture farms (48◦45 ′ 50.8 ″ N 2◦10 ′ 43.6 ″ E). The water samples were spiked with glyphosate and incubated at room temperature under shaking for 1 h. Afterward, each sample was diluted in 0.1 M acetate buffer, pH 5, and tested electrochemically (three replicates for each glyphosate concentration). The recovery, expressed in %, was calculated according to the following equation: Z. Novakovic et al. Talanta Open 12 (2025) 100481 3 Recovery =Found/Addedx100 (5) 3. Results and discussion 3.1. Structure and morphology of synthesized ZnO NPs Pure phase zinc oxide was obtained via the synthesis process presented in Fig. 1a as all the measured X-ray diffraction peaks could be indexed to standard ZnO - JCPDS 36–1451, as shown in Fig. 1b. Elemental analysis and mapping by EDS of the obtained nanoparticles showed a homogenous distribution of Zn and O throughout the sample, as shown in Fig. S1. The Zn and O content was determined and corresponds to previous values obtained for ZnO [36] confirming the formation of pure phase ZnO. Structural refinement of the measured diffractogram was performed using the Rietveld method and GSAS II software package [62] in order to obtain the lattice parameters of the hexagonal wurtzite P6 3 mc crystalline lattice of ZnO. The determined parameters (with R wp of 0.115) were: a =b =3.2529(4) Å and c = 5.2168(7) Å, with a unit cell volume of 47.816(11) Å 3 . The distortion parameter (c/a) was calculated as 1.6037. These values are in accordance with values previously obtained for ZnO NPs [36,50]. The size-strain plot (SSP) method was applied to calculate the crystallite size. This method takes into account notable microstrain contribution reflected in peak broadening compared to highly crystalline ZnO [36,50]. The obtained plot and linear fit are shown in Fig. 1c enabling calculation of the crystallite size (D XRD ) using the Eq. (1) of 7.43 nm, the microstrain ( ε ) as 0.148 and the dislocation density (δ) as 0.01812. The crystallite size is small, and observation of the morphology (Fig. 1d) also shows small ZnO NPs grouped in the form of “flower petals”. The absorbance F(R) was determined from the Kubelka-Munk transformation of the measured diffuse reflectance spectrum. Maximum absorption can be noted around 345 nm, as shown in Fig. 1e. This represents a blue shift compared to bulk sample and has been noted before for ZnO NPs [63]. The direct optical bandgap for ZnO NPs of 3.23 eV was estimated from the Tauc plot, as shown in Fig. 1e. This value is within the range of values determined for different ZnO NP morphologies [36,63]. Recorded TEM images of ZnO nanoparticles (Fig. 2a, 2b and 2c) show that most particles are small, around 10 nm, but a few larger particles (~ 50 nm) can also be observed. The larger particles are more hexagonal in shape, while the smaller particles are more rounded. Similar morphologies were recently observed for ZnO nanoparticles obtained using the green synthesis method [36]. The particle size distribution was determined for both particle types separately and is shown in Fig. 2d and 2e. The average particle size for the dominantly present small particles was 9.53±3.14 nm, while the larger particles had an average size of 51.49±16.31 nm. Clear ring diffraction patterns were retrieved from TEM images, as shown in Fig. 2f. In accordance with Fig. 1. ZnO nanoparticles: (a) Flowchart of ZnO NPs synthesis. (b) X-ray diffraction pattern; (c) size-strain plot enabling calculation of the crystallite size and strain; (d) SEM image; (e) absorption spectrum with inset showing Tauc plot estimation of the direct band gap. Z. Novakovic et al. Talanta Open 12 (2025) 100481 4 Fig. 2. ZnO nanoparticles: (a, b, c) TEM images with increasing magnification; (d) small and large (e) particle size distribution; (f) SAED pattern and diffraction ring. Fig. 3. Schematic illustration of electrode modification using the simple drop-casting method to create a thin solid film of ZnO nanoparticles for glyphosate detection. SPCE was modified by drop-casting with a ZnO NPs suspension. Glyphosate was added to the surface, incubated and washed. The detection was performed by DPV. Z. Novakovic et al. Talanta Open 12 (2025) 100481 5 previous research [36], the discrete and bright diffraction spots can be attributed to the larger particles, while the more continuous diffraction spots and ring can be attributed to the smaller particles. The calculated pink concentric rings overlaying part of the SAED image represent the calculated crystalline planes of the ZnO wurtzite phase (in accordance with XRD). The lattice distance of 0.2618 was determined from the HRTEM image shown in Fig. S2 and it is in line with the calculated lattice distance of 0.2608 (from Rietveld refinement of the measured XRD diffractogram) for the (002) reflection plane of ZnO. These results enable correlation between the XRD, HRTEM and SAED and confirm the formation of pure phase ZnO of a petal-like architecture. 3.2. Biosensor fabrication and characterization We employed synthesized ZnO NPs as a recognition element in the label-free electrochemical sensor for glyphosate determination. Fig. 3 shows a scheme of the electrode modification and the sensor construction. ZnO NPs were immobilized on a carbon surface by drop casting to enable the crosslinking of glyphosate to the electrode surface. K 3 [Fe (CN) 6 ]/K 4 [Fe(CN) 6 ] redox probe was used to convert the binding of glyphosate to the functionalized electrode into an electrical signal. The SPCE was functionalized by drop casting with 4 µl of ZnO NPs in concentrations ranging from 0.1 mg/mL to 20 mg/mL. Cyclic voltammograms (CVs), obtained with 5 mM ferro/ferricyanide and 0.1 M KCl as a supporting electrolyte, showed distinct oxidation and reduction peaks, indicative of a reversible redox process (Fig. 4a). The immobilization of 0.1 mg/mL ZnO NPs decreased the current intensity compared to that obtained with the bare electrode, suggesting that the attached nanoparticles had fewer available active sites for the redox reaction. Conversely, at 10 mg/mL ZnO NPs the higher current intensity of peaks compared to those obtained with 0.1 mg/mL indicates an optimization of the density of active sites for redox reactions. The intensity of peaks remained similar with 20 mg/mL ZnO NPs. Therefore, to economize the material consumption, 10 mg/mL ZnO NPs was used for electrode functionalization. Using the Randles–Sevcik equation (Eq. (2)), the electroactive surface area of the 10 mg/mL ZnO NPs-modified electrode was estimated to be 6.6 mm 2 , with the R f of 0.53. For comparison, the bare electrode before NPs immobilization had an electroactive surface area of about 11.43 mm 2 and the R f of 0.91. This decrease in the electroactive surface area after functionalization confirms that the attached ZnO NPs had fewer available active sites for the redox reaction. Although the peak current in CV decreased with ZnO NPs immobilization, EIS showed that the resistance at the electrode/solution interface decreased indicating improved mass transfer (Fig. 4b). We further tested the incubation volumes of ZnO NPs to achieve optimal electrode surface coverage. A gradual decrease in current intensity and good peak-to-peak separation were observed via cyclic voltammetry after incubating the electrode with 2, 3, or 4 μ L of 10 mg/mL ZnO NPs (Supporting Information, Fig. S3a). This decrease in current with increasing volume of ZnO suggests the formation of a ZnO NP layer over the higher electrode surface, hindering electron transfer between the redox species and the electrode surface, and the peak-to-peak separation indicates sufficient mass transfer at the electrode for all tested volumes. The highest surface coverage, calculated using Eq. (3), was achieved with 4 μ L of ZnO NPs suspension, Γ ~ 80.71 % (Supporting Information, Fig. S3b). However, to ensure that ZnO NPs were confined to the working electrode and to minimize their leakage on other electrodes and excessive spillage, 3 μ L of ZnO NPs suspension was used in further experiments. This volume provided Γ ~ 71.05 %. Next, CV and EIS were employed to evaluate the effect of varying deposition times for ZnO NPs onto the SPCE. Curves were recorded for ZnO NPs-modified electrodes after 2 h, 4 h and overnight spontaneous Fig. 4. Functionalization of SPCE with ZnO NPs: (a) Optimization of ZnO concentrations (0.1 mg/mL, 1 mg/mL, 10 mg/mL, and 20 mg/mL) for electrode modification. Potential vs. Ag/AgCl (b) EIS responses obtained for different concentrations of ZnO NPs. (c) Time optimization of drop-cast 10 mg/mL ZnO. Potential vs. Ag/AgCl. (d) EIS responses obtained with different incubation times for 10 mg/mL ZnO NPs immobilization. (e) SEM image of bare SPCE and modified with 10 mg ZnO NPs during 4 h. (f) CV at scan rates from 0.1 to 1 V/s recorded using ZnO NPs-modified SPCE. (g) The dependence of 5 mM K 3 [Fe(CN) 6 ]/K 4 [Fe(CN) 6 ] anionic and cationic peak intensity on the square root of scan rate. (h) Sensor repeatability studied with 10 successive voltammetric responses of the modified electrode. (i) Sensor reproducibility studied with 9 ZnO NPs-SPCEs using DPV. All measurements were performed with 5 mM K 3 [Fe(CN) 6 ]/K 4 [Fe(CN) 6 ] in 01 M KCl. Z. Novakovic et al. Talanta Open 12 (2025) 100481 6 coating (Figs. 4c and 4d). The decrease in current obtained with increasing immersion time confirmed that ZnO NPs immobilization reduced the effective surface area available for electron transfer (Fig. 4c). The 2-hour immersion time resulted in labially bound ZnO NPs since the current responses were unstable and the ZnO NPs layer was easily washed away. In contrast, a 4-hour immersion time produced a stable ZnO NP film, achieving a balance between surface coverage and electrochemical activity. The observed shift of current peaks toward more positive and negative potentials suggests that 4-hour incubation enables ZnO NPs to gradually form compact aggregates, acting as a barrier that increases the energy required for electron transfer in redox reactions. Extending the immersion overnight did not significantly affect the current compared to the 4-hour immersion but increased the film thickness and reduced its conductivity. Together, these observations indicated that 4-hour immersion time was optimal, providing sufficient ZnO NPs loading while preserving a high surface area for efficient electron transfer. Moreover, EIS indicated that the charge transfer resistance was the lowest for 4 h incubation time (Fig. 4d). SEM micrographs further confirmed successful modification of the carbon electrode. Fig. 4e displays the working electrode area composed of randomly oriented micrometer carbon flakes before ZnO NPs immobilization. After 4-hour deposition of ZnO NPs, the electrode surface was uniformly coated with small ZnO nanoparticles as observed for the assynthetized material (Fig. 4e). To investigate the electrochemical reaction kinetics of the modified electrode, CV was performed as a function of the scan rate in the ferro/ ferricyanide and 0.1 M KCl. As shown in Fig. 4f, the position of the oxidation peaks shifted toward more positive potentials as the scan rate increased, whereas the reduction peak shifted toward more negative potentials. The peak-to-peak distance of the potentials of oxidation and reduction are 260 mV and 580 mV with scan rate 0.01 and 0.1 V/s, respectively. This indicates that the electron transfer rate becomes comparable to the mass transport rate at higher scan rates. In addition, both anodic and cathodic peak currents increase linearly with the square root of the scan rate (Fig. 4g) indicating that the oxidation and reduction processes of the redox mediator at the ZnO NPs-SPCE are diffusioncontrolled. In addition, to evaluate the kinetic reversibility of the redox probe, the heterogeneous rate constant was calculated according to the equation (Eq. (4)) for bare and ZnO modified SPCE. The k ⁰ values were derived from the charge transfer resistance (Rct) obtained via electrochemical impedance spectroscopy (EIS), based on the Randles equivalent circuit model (Supp. information S4). For a one-electron transfer process, the calculated rate constant increased from 1.875 × 10⁻⁵ cm/s for the bare electrode to 2.83 ×10⁻⁵ cm/s after modification with ZnO nanoparticles. The observed increase of the heterogeneous rate constant may originate from the increased electroactive surface area provided by the ZnO nanostructures, the presence of surface defects or functional groups facilitating electron tunneling, but also possibly the electrocatalytic role of ZnO in redox reactions. The sensor interface was evaluated for operational voltammetric response stability and reproducibility, which are always a concern when an electrode is functionalized with nanomaterials using simple dropcasting. First, the stability was assessed by conducting multiple cyclic voltammetry scans with a 5 mM K 3 [Fe(CN) 6 ]/K 4 [Fe(CN) 6 ] in 0.1 M KCl. CV was applied because it records the entire current response as a function of potential. The consistent intensity of both peaks across 10 consecutive scans demonstrated the exceptional stability of the ZnO NPs-layer when attached to carbon electrodes. (Fig. 4h). Then, differential pulse voltammetry (DPV) was used to assess the reproducibility of electrodes. For this, nine electrodes were produced and tested for their response with the ferro/ferricyanide couple in a KCl solution. DPV was used instead of CV because it is more sensitive and thus a stable base line would indicate sensor suitability for detection of low-concentrations of the analyte. Excellent reproducibility was obtained as all analyzed electrodes displayed a very similar current response (Fig. 4i). The signal variability was found to be ≤7.71 %. Finally, the stability of the sensors was tested during storage in humidity chambers placed in the fridge (4 ◦C) and exposed to air. The signals recorded at several time points over 4 weeks indicated consistent sensor performance (Fig. S5); these data show that the sensors are relatively stable for 120 days. 3.3. Glyphosate detection at different pHs To confirm the specific binding of glyphosate to the ZnO NPsmodified SPCE rather than the bare SPCE, cyclic and differential pulse voltammetry measurements were performed using 5 mM K 3 [Fe(CN) 6 ]/ K 4 [Fe(CN) 6 ] in 0.1 M KCl before and after incubating the electrodes with 1 mM glyphosate in acetate buffer. The voltammograms (Supporting Information, Fig. S6a) showed a reduction in redox current after glyphosate adsorption on the bare SPCE, indicating the formation of an electron transfer barrier. This suggests glyphosate adsorption onto the bare SPCE. However, after glyphosate binding to ZnO NPs, electron transfer between the redox probe and the electrode became more efficient. This phenomenon was further confirmed by DPV, where a significant increase in current was recorded after incubating the ZnO NPsmodified SPCE with 1 mM glyphosate, compared to the bare SPCE (Supporting Information, Fig. S6b). This suggests the arranged attachment of glyphosate onto the surface carrying ZnO NPs that facilitates electron transfer. To achieve strong interactions between glyphosate and ZnO NPs, it is crucial to optimize the pH. Glyphosate acts as a potent chelating agent, forming stable coordination complexes with divalent cations such as Zn, Mg, Ca, and Mn through interactions highly pH-dependent [18]. At acidic pHs, glyphosate carries a negative charge due to the ionization of its carboxyl and phosphonate groups, and has an increased affinity for binding with divalent cations [3]. ZnO, with a point of zero charge around 9.2, exhibits a positive surface charge at pH levels below this threshold, and the charge intensifies as the pH decreases [64,65]. The effect of pH on the electrochemical behavior of ZnO NPs-modified SCE incubated with 1 mM glyphosate solution was studied using DPV in 0.1 M acetate buffer, across a pH range from 4 to 9. The maximum peak current was observed at pH 5, indicating optimal electrochemical response (Fig. 5a). In contrast, the peak current decreases, as the pH increases suggesting that either glyphosate exhibits greater electrochemical activity in more acidic conditions on the ZnO NPs-modified electrode or that its binding ZnO NPs includes hydrogen bonds. A minor shift in the peak potential toward more positive values with increasing pH, following a linear trend, E (V) =0.037 pH +0.375, R² = 0.9695, indicates that the redox process of glyphosate is influenced by proton concentration, characteristic of proton-coupled electron transfer reactions [66]. Based on these results, pH 5 was selected as the optimal condition for detection experiments. We further optimized the time required for glyphosate to interact fully with the ZnO surface. The 30-minute incubation time yielded the highest current intensities compared to lower times tested (Fig. 5b). High incubation times were disregarded to maintain a fast-testing process. A SEM micrograph at 10,000x magnification revealed some coating over ZnO nanoparticles probably due to glyphosate molecules binding to ZnO nanoparticles (Fig. 5c), confirming the success of the 30-minute coordination. The association of glyphosate with ZnO NPs was confirmed by EDX spectroscopy that detected nitrogen atoms at the electrode surface (Fig. S7). Surprisingly phosphorous was not detected. This might rise from matrix effects that reduce its detectability causing phosphorous weak interactions with the electron beam. Together, these results show that ZnO NPs not only facilitate the selective interaction of the electrode with glyphosate but also enhance the electrochemical response, improving sensor specificity and increasing the electrode’s conductivity, ultimately leading to a higher current signal. 3.4. Glyphosate detection The electroanalytical potential of ZnO NPs was challenged for the Z. Novakovic et al. Talanta Open 12 (2025) 100481 7 direct detection of glyphosate, given that glyphosate is a chelating agent that can form complexes with Zn 2+ - ions. DPV measurements were performed after 30-min incubation of the ZnO NPs-modified electrodes with various concentrations of glyphosate (1 µM to 7.5 mM) in acetate buffer, pH 5 (Fig. 6a). A linear increase of peak intensity was obtained with a correlation coefficient (R 2 ) of 0.9816 (Fig. 6b). The limit of detection (LoD) was calculated using the formula (3) s/m, where ‘s’ represents the standard deviation of the blank solution and ‘m’ is the Fig. 5. Optimization of the electrochemical reaction onto ZnO NPs-modified electrode: (a) DPV curves for 1 mM glyphosate in 0.1 M acetate buffer with pH 5–9 (left panel) and dependence of the oxidation peak current and potential on pH (right panel). DPV parameters: potential range −0.1 to 0.4 V vs. Ag/AgCl, pulse amplitude 0.025 V.; (b) Differential pulse voltammograms obtained after the different time of incubation of glyphosate (left panel) and change in current intensity versus different times of incubation (right panel); (c) SEM micrograph presents attached glyphosate molecule onto ZnO nanoparticles at 1000x magnification. Fig. 6. Detection of glyphosate in buffer solution and in river water: (a) DPV obtained for sensing various concentrations of glyphosate in 0.1 M acetate buffer, pH 5; (b) The calibration curve was obtained by using data in (a); data points represent the mean ±SD of the three independent experiments; (c) DPV performed with various concentrations of glyphosate in the river water. DPV parameters: potential range −0.1 to 0.4 V vs. Ag/AgCl, pulse amplitude 0.025 V. (d) The calibration curve obtained by using the current obtained in (c); data points represent the mean ±SD of the three independent experiments. Z. Novakovic et al. Talanta Open 12 (2025) 100481 8 slope of the linear calibration curve, resulting in a LoD of 648 nM, while the limit of quantification (LOQ) was determined as 3.3 times the LOD according to the definition by the International Union of Pure and Applied Chemistry (IUPAC) [67], resulting in a LOQ of 2.14 µM. A linear increase in current intensity was also obtained when acetate buffer was replaced with pH 5 phosphate buffer (Fig. S8) suggesting that phosphate ions did not prevent interactions of glyphosate with the ZnO surface. 3.5. Glyphosate detection in river water Glyphosate detection was conducted on local river water using differential pulse voltammetry, which again revealed a linear increase in current intensity after incubating the solution with increasing concentration of glyphosate on the modified electrode (Fig. 6c). The calculated LoD was 960 nM and the recovery rate of 82.46 % (Fig. 6d). The lower sensitivity in the river water compared to PBS was expected since river water is a more complex media containing particulates, organic matter, microorganisms, and unknown ions that can interfere with the analyte binding or signal generation. The developed electrochemical sensor for glyphosate detection demonstrated comparable detection performance with recently reported sensors, as depicted in Table 1. However, our sensor shows higher sensitivity compared to others and the measurement is completed in only 30 min. 3.6. Specificity of detection The selectivity of the proposed sensor was evaluated using DPV measurements in the presence of common interfering herbicides and ions typically found in river water. Organic and inorganic substances were tested at a 10:1 molar ratio relative to glyphosate. The first part compared the signal obtained with carbendazim, mesotrione, and irgasan (10 µM) with that obtained with 1 µM glyphosate (Fig. 7a–c). Standard solutions of each interferent were prepared in 0.1 M acetate buffer, pH 5 and DPV voltammograms were recorded after 30-minute incubation with these herbicides on ZnO-modified electrodes preincubated with glyphosate. The results indicate that ZnO exhibits a strong affinity for glyphosate, with minimal current variations observed after adding the interferents that failed in preventing glyphosate detection. In the second part, the impact of ions (NO₃⁻, Cl⁻, Ca²⁺, Na⁺, SO₃²⁻) on glyphosate binding was investigated. DPV voltammograms recorded before and after glyphosate incubation in the presence of these ions reveal that binding remains unaffected (Fig. 7d). Additional experiments highlight the sensor’s selectivity by comparing DPV responses when the ZnO NPs-modified electrode was incubated separately with different herbicides (Fig. S8). A significant current increase was observed only for glyphosate, while other herbicides caused a decrease, further demonstrating the sensor’s specificity. 3.7. Modeling glyphosate binding to ZnO NPs Calculation of basic properties of the bulk material was carried out utilizing the experimentally obtained hexagonal (space group P6 3 mc) unit cell of ZnO. The calculated Fermi energy is −5.59 eV, and the predicted band gap is 0.76 eV, which is far below the experimental value of 3.4 eV [68]. This inconsistency with the experiment arises from the well-known deficiencies of DFT, which do not account for the discontinuity in the exchange-correlation potential. General Gradient Approximation (GGA) functionals, such as PBE, are designed for ground-state properties, whereas the band gap is an excited-state property that standard DFT cannot accurately describe [69,70]. Although the calculated ZnO band gap is lower than the experimental value, the primary focus of this work remains unaffected, as it is centered on investigating the interaction of glyphosate with the surface of the material. The projected density of states (pDOS) of ZnO bulk Table 1 Comparison of detection performance with other potential glyphosate detection platforms. Nanomaterial Electrochemical method Matrice LoD Dynamic range Reference Niobium NPs g CV a DPV b water 3.07 µM 5.90 –172.30 µmol/L [81] BC-nZVI h CV DPV LSV c juice, milk 768.7 nM [82] GO i NPs CV DPV EIS d water 2 μ M up to 100 μ g/L [83] ZnO NPs CV DPV CA e green tea, corn, mango juice 2.84 μ M 0 μ M–5 mM [31] TiO 2 NTsj/AgNPs-rGO EIS water 3 µM 0.005 –50 mg/L [84] MWCNTs k /ZnO CV CA drinking water 1 μ M 1 – 10 μ mol/L [85] AuNPs-ZnONWs-MoS 2 NSs l –soybean, corn 3.53 μ M 0 – 80 μ M [86] CPE/Sg-OS/EG m CV SWV f soil 0.98 μ M 10 – 100 μ M [87] Au-SPE CV SWV tap water 2 μ M 0.3, 0.15 and 0.075 mM [88] Cu/GC n CV, DPV –31 μ M 70 – 800 μ M [88] ZnO NPs DPV, EIS river water 0.96 µM 0.5 µM – 7.5 mM This work a cyclic voltammetry. b differential pulse. c linear sweep voltammetry. d electrochemical impedance spectroscopy. e chronoamperometry. f square wave voltammetry. g nanoparticles. h magnetic biochar-nano zero valent iron (BC-nZVI); graphene oxide j nanotubes. k multi-walled carbon nanotubes. l gold nanoparticles-ZnO nanowires-molybdenum disulfide nanosheets. m carbon paste electrode/homoionic clay-octyltriethoxisilane/ethylene glycol. n glassy carbon surface. Z. Novakovic et al. Talanta Open 12 (2025) 100481 9