Smartphone-Based Surface Plasmon Resonance Sensor for Glyphosate Detection: Different pH and Concentrations

Glyphosate [(N-phosphonomethyl) Glycine], one of the most worldwide commercialized herbicides, has provoked many debates about its carcinogenic effects. Here, a smartphone-based surface plasmon resonance (SPR) sensor is proposed for glyphosate detection using different pH and concentrations. CuO nanoparticles have been added to glyphosate samples, diluted in ultrapure water solutions, to enhance its detection. An increase of sensitivity was observed in acidic solutions reaching a dilution of 10-8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{-8}$$\end{document} (v/v), which is equivalent to 5·10-7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$5\cdot 10^{-7}$$\end{document} ppm. This novel smartphone-based SPR device for glyphosate detection besides presenting very high sensitivity; it has also favorable features such as easy handling, portability, and real-time analysis.


Introduction
Glyphosate [(N-phosphonomethyl) Glycine] is a herbicide widely used in crop control [1,2], being degradable by soil microbes and binding ability to soil colloids [3]. However, some recent studies have pointed out harmful effects for human health and environment [4]. In recent years, more than eighty methods have been proposed. They are normally expensive and slow [3].
Surface plasmon resonance (SPR) principle is one of these methods, where advantageous features, like high sensitivity, real-time response, high accuracy, immunity to electromagnetic interference, and attractive cost [5][6][7], are attractive for glyphosate detection [8,9]. On the other hand, different approaches have been proposed to enhance SPR sensor performance parameters, using different materials or techniques. One of these approaches uses smartphone to excite the surface plasmons on the sensor chip and detect the output image. It is a good alternative compared to the conventional devices, having ease of operation and portability.
Recently, a smartphone-based Surface Plasmon Resonance (SPR) device was applied for glyphosate detection [10,11], using a polymeric prism as the optical substrate. Here, we use this setup to make experiments. The sensor uses a red input source generated on the smartphone screen and the front camera to detect the output image. The experimental setup is presented here considering the prismbased glyphosate detector [10,11] and different samples of glyphosate diluted in ultrapure water with different pH values. The sensing scheme relied on the interaction between silver nanoparticles produced by laser ablation in an aqueous sodium citrate solution. This forms a stable colloid with the analyte during the measurement time interval. The Raman spectra of the samples reached a detection limit of 1.7 ppm of glyphosate in water. The glyphosate dilution ranged from 10 −2 (v/v) to 10 −8 (v/v), and our device detected all range. Moreover, sensitivity increased at acidic solution when compared to a basic one.

Smartphone-Based SPR Sensor
The used SPR-based platform is depicted in Fig. 1, with a polymeric structure adapted for a Galaxy SIII (Samsung). The transducer is the DOCE (Diffractive Optical Coupling Element) or VIR biochip [3]. Light emitted by the smartphone display impinges the DOCE surface at 90 degrees ( Fig. 1). The entrance wavelength is = 670 nm. The flow cell which is installed at the sensing spot.
The app developed in [10,11] to acquire and process the data was recently improved using Java language. It generates the image that provides the light source in the desired wavelength, controls the acquisition of the image reflected by the transducer, records and processes the information, showing the calculated values on the screen. It is possible to perform signal acquisition and processing in a range of 700ms.
A photo of the test platform is presented in Fig. 2, where all elements are evidenced. The samples circulate through the sensor using a syringe -flow-cell set, having a cleaningrinsing process after each analysis. Beams of light reach the
Samples were prepared by dilution, with GlyphotalTR ® 10 −2 (v/v) the most concentrated, which represents 5.0 ppm. The nanoparticles were also diluted ( 1 g∕100 L ) in each dilution of GlyphotalTR ® . H 2 SO 4 and NaOH solutions were used to perform pH changes.

Microscopy and Nanoscopy Analysis of the Glyphosate-CuO Interaction
SEM images, performed by an electronic microscope (Carl Zeiss EVO MA10) associated with an Energy-Dispersive Spectroscopy (EDS) (Oxford Instruments SDD detector), analyzed the interaction of Glyphosate with nanocrystals (NCs) according to pH changes. The compositional analysis was performed using the INCA Tools for Measurements software (INCA V7.2 Service Pack.13.2, ETAS GmbH, Stuttgart, Germany). In the SEM images, a carbon was attached to a double-stick tape on aluminum stubs and coated with gold in a sputter coating apparatus. The complex was diluted to a concentration of ( 10 −2 v/v). AFM images were also performed by the SPM 9600, Shimadzu, a high-resolution scanning probe microscopy. A 3 L of GlyphotalTR ® -CuO was poured into ultrapure water [ 10 −4 (v/v)] solution on a sheet of mica.

Absorbance Spectrum Analysis of Glyphosate Samples
A Shimadzu UV-1800 spectrophotometer was used to verify the characteristics of the glyphosate through the UV-Visible range, where a quartz cuvette with an optical path of 10 mm was used.

Experimental Results
Glyphosate samples at dilutions of 10 −2 (v∕v) (usually used for soil applications), 10 −6 v∕v and 10 −8 (v/v) (hardly detected by other devices [12]) in ultrapure water have been used. To measure the glyphosate enhancement signal and the consequent measurement of its affinity, ZnO, Ag 2 O, and CuO nanoparticles were added.
The results for copper oxide (CuO) pointed out to be the most efficient nanoparticle for amplifying the absorbance signal in all dilutions [10]. On the other hand, dilutions of 10 −4 and 10 −6 (v/v) showed no significant absorbance in the UV range [10]. This demonstrates the need to investigate new techniques for detecting glyphosate at lower dilutions, Fig. 3 Investigation of the influence of pH 1-acidic in UPultrapure water on the absorbance spectrum of glyphosatecontaining substances at dilutions of 10 −2 , 10 −4 , 10 −6 , and 10 −8 (v/v) and CuO-oxide nanoparticles enabling the identification and analysis of glyphosate in food, for example. On the other hand, our experiments showed that changes in pH also increase the absorbance level, allowing significant detection for the lowest dilution ( 10 −6 (v/v), which was not found when we used only CuO nanoparticles [10].
Investigations of the effect of pH 1 (acidic) in the analysis of UV-visible spectroscopy in dilutions of 10 −2 , 10 −4 , 10 −6 , and 10 −8 (v/v) of glyphosate with CuO nanoparticles in ultrapure water are shown in Fig. 3.
It was possible to identify, in Fig. 3, absorbance spectra of pesticides in all acidic dilutions. For dilutions up to 10 −6 (v/v), the absorbance of glyphosate with the nanoparticles exhibited the same behavior, demonstrating at very low concentrations, the equipment has similar sensitivity. For dilutions of 10 −4 (v/v), there is an increase in the intensity of the spectrum, and for dilutions of 10 −2 (v/v), the spectrum presented the same profile. However, the displacement of the characteristic peak to the lower energy region results in an increase in the concentration of the species, which reduces the average distance between the molecules to the point of modifying the absorption capacity of the pesticide.

Performance Analysis
To evaluate the performance of the proposed system, an average of 50 samples of the signal energy, for each concentration value used, was obtained by the smartphone ( E R ). The total time to perform the experiment for each concentration value is 35s. UV-Vis analysis was used as a standard reference. The peaks of the obtained absorbance signal ( Max Abs ) were analyzed for each concentration and the results are shown in Fig. 4. Comparing these results with those of a conventional instrument (UV-Vis) the same behavior was observed.
To analyze the response of the portable platform to glyphosate solutions of acidic pH, the same methodology applied to the neutral solution was used. CuO nanoparticle solutions were sequenced for dilutions of 10 −8 , 10 −6 , 10 −4 , and 10 −2 (v/v) of Glyphosate and the results with acidic pH are shown in Fig. 5.
It can be seen in Fig. 5 that the energy ( E R ) increases when the dilution is less than 10 −4 (v/v). This behavior was also observed by UV-Vis experiments. For values from 10 −8 to 10 −6 (v/v), the sensor response is even more pronounced than the UV-Vis approach. When the transition occurs from 10 −4 to 10 −2 (v/v), there is a change in the dynamics of the output signal, with a decreased E R value. Although this behavior is not desired, the same response was verified in the maximum absorbance value obtained by UV-Vis. This result comes from the fact that acidic pH has amplified the sensor response, which means that solutions with very high concentrations may be outside the sensor measuring range.
The change to acidic pH allowed to measure dilution of 10 −8 (v/v), which would not be possible with a solution of neutral pH. This result proves that the amplification of the response signal occurs due to the change from neutral pH to acidic pH. To assess the effects of pH on the sensor response, the sensitivity of the instrument to the dilution of the solution was determined using the following: where, for the proposed system, the output signal, y, is the energy value, E R , and for the UV-Vis, y is the maximum absorbance value, Max Abs . The results obtained are shown in Fig. 6.
Therefore, when using an acidic solution, the sensor showed a higher sensitivity than the neutral solution, when a dilution of 10 −4 (v/v) was considered. This increase in sensitivity was verified for both UV-Vis and the proposed SPR sensor. Nevertheless, when the dilution was of 10 −2 (v/v), there was a decrease in the sensitivity of the proposed platform and UV-Vis. This occurred to the extrapolation of the response range of the used equipment.

Discussion
Although most of the available equipment fails to detect glyphosate without pretreatment due to its high affinity and interaction with metal cations [13], the smartphone-based SPR device shown was able to detect the herbicide in a very low concentration. Even in Europe, with the greatest restriction on the use of glyphosate(1 × 10 −4 ppm), only a small number of test laboratories are able to detect this chemical [14], once Cu 2+ is widely used for agricultural purposes as a fungicide and is often added together with glyphosate and is present in fertilizers, sewage sludge, and other waste. When Cu 2+ is present, a drastic reduction in the concentration of free glyphosate is observed due to the formation of the complexes [15,16]. Therefore, this does not mean that glyphosate is not present in the environment, but it may not be detected. The herbicide and Cu 2+ are strongly adsorbed on goethite (a hydroxide mineral containing iron), and the complex formed is affected by the pH and the presence of the other.
There is a tendency for transitions and some trivalent metal ions, and divalent alkaline-earth to form 1:1 (e.g., Ca(II) and Cu(II) [17,18]) and 2:1 metal chelates with glyphosate in solution [19]. DFT molecular modeling methods have shown that there is a stability order for tetrahedral and octahedral complexes between metals and glyphosate as Zn>Cu>Co>Fe>Cr>Al>Ca>Mg [20].
The pH effect on the adsorption of Cu 2+ and other metals for variable charge adsorbents has been previously reported. Interesting developed work has revealed that the percentage of Cu 2+ adsorbed on an iron oxyhydroxide decreased by about 80% when the pH decreased from 6.00 to 4.75 [21]. In addition, a similar behavior was found for Cu 2+ adsorption on an oxisol soil [22].
Cu 2+ was also used to develop a simple, unlabeled colorimetric method based on the inhibition of the peroxidaselike activity of Cu 2+ that is suitable for detecting glyphosate [23]. Cu 2+ was also used in the voltammetric test, where the complex has two absorption bands of ultraviolet to visible light (range of 200 to 1000 nm) [24]. The authors concluded that glyphosate lacks a chromophoric group in its structure, absorbing light up to 200 nm [25]. However, considering Cu 2+ solution, the band at 231 nm is evident and results from the charge metal-glyphosate transfer, where the herbicide sends an electron. Probably in this process, the electrons on the amino group are involved [26]. The authors indicate that glyphosate is part of the Cu 2+ complex and occupies 3 of the 6 positions in a distorted octahedral (Jahn-Teller Distortion). This octahedron is completed by three water molecules. Glyphosate does not show electrochemical response on the HMDE electrode [27] unless it interacts with Cu 2+ by reduction to Cu [24].
The interaction was confirmed by Energy-Dispersive X-Ray Spectroscopy (EDS) analysis of the scanning electron microscope (SEM) images. Figure 7 shows the AFM of the Glyphosate with CuO at 10 −4 (v/v) in the pH 6 and Fig. 8 shows the SEM at 10 −2 (v/v) in three different pH.
At acidic pH, glyphosate interacts with CuO being stellar forms highlighted, as shown in Fig. 7A. It does not happen in neutral and acidic media as can be seen in Fig. 7B, C, respectively.
Combining the geometric shape, we also hypothesized that the interaction between glyphosate and CuO NCs promotes an orbital overlap capable of stimulating the electromagnetic field, which leads to an increase in plasmon. It is worthy mentioning that in the acidic medium we have protonated glyphosate in the amine functional group and excess H + ions in the medium where there are ideal conditions for electron transfer. We believe that these factors would promote P-type doping, considerably expanding the magnitude of the electromagnetic field, and allowing the detection of glyphosate at a dilution of 10 −8 (v/v), which is equivalent to 5 ⋅ 10 −7 ppm. Very sensitive chromatographic methods were used to reach detection limits of this magnitude, such as LC-SPE-ESI/MS/MS [28].
Some studies have shown that when the nanoparticles are in a star form, a surface-enhanced Raman spectroscopy (SERS) intensification effect occurs [29,30], and surfaceenhanced rFTIR [31]. SERS sensors are considered the best devices for non-destructive molecular analysis considering reliability, sensibility, and selectivity. Unfortunately, the applicability of SERS is quite limited, which is mainly due to the lack of highly sensitive SERS platforms with good stability and reproducibility. Because of this, the use of many SERS platforms has been coupled with metal nanoparticles.
The morphology acquired by some nanoparticles is the key application in SERS because the resonant excitation of plasmons can dramatically amplify the electric field near the surface of the nanoparticles, which is explored in the present device [32][33][34][35]. This strongly contributes to increasing the sensitivity of plasmon resonance to the local dielectric environment. Improvements in the electric field for excitation of plasmon resonances in a nanostar result from plasmon hybridization and the characteristics of the nanoparticle core, as already concluded by [29,36,37]. Thus, they concluded that tip plasmons comprised the low-energy nanostar-binding plasmons with a finite contribution from the core plasmons. The core/tip interaction of the plasmon increases the excitation of the binding plasmons, increasing the local electric field.

Conclusions
A smartphone-based SPR sensor has been proposed here to detect glyphosate in aqueous solutions. The results indicated the viability of the proposed sensor, with advantages of portability, ease of use, operation in real time, which results in a more effective control of the use of this herbicide in agricultural crops.
The selection of the appropriate nanoparticle (CuO) greatly improves the response of the equipment, allowing to identify the presence of this herbicide even at dilutions in the order of 10 −6 (v/v).
Changing the pH of the solvent to acidic leads to an improvement in the signal response. This has been proven through increased sensitivity for smartphone-based sensors and UV-Vis approaches. This is compared with measurements made at neutral pH. The proposed smartphone system was able to measure the dilution of up to 10 −8 (v/v) of glyphosate when using an acidic solution. A greater variation was proposed in the proposed sensor approach when compared to UV-Vis, in the dilutions of 10 −4 (v/v). The results show that the proposed solution is then able to measure low levels of glyphosate concentration, similar to those found in soils ( 10 −2 (v/v)) and in foods ( 10 −6 (v/v)).
The best pretreatment technique and the optimal concentration of samples are under investigation, which can lead to a better optimization of the concentration and effect of the herbicide, with less impact on the environment.
Glyphosate forms chelates with Cu2+ ions (lewis acidbase covalent bond forming an adult -coordination compound). In Cu NPs there is no Cu2+ on the surface and, therefore, there is no covalent bond between Gly and Cu NPs, as Cu2+ and Gly occur in the soil. Therefore, we use the term interaction, since at acidic pH, the glyphosate molecule is protonated and positively charged and Cu NPs have negative zeta potential. Therefore, an electrostatic interaction occurs that has not yet been deeply studied. Author Contributions Anderson Valle was in charge of the glyphosate experiments and the initial writing of the manuscript; Kaline Ferreira aided with the experiments with Glyphosate and the tests considering different substances; Luiz Goulart was responsible for revising the text and coordinating the sending of glyphosate samples; Carmonizia Freire tested the smartphone, adapted the software and collected the results; Eliton Medeiros helped with the process of reviewing and writing the manuscript; Carlos Alberto de Souza Filho was responsible for creating and programming the software, analyzing the experiments carried out by the smartphone, writing and revising the manuscript; Rossana Cruz was one of those responsible for revising the text; Luciano Rodrigues collaborated in writing the manuscript, considering aspects of materials and methods and discussion; Cleumar Moreira was one of those responsible for writing and revising the manuscript.
Data Availability Not applicable.