Colorimetric detection of glyphosate: towards a handmade and portable analyzer

GFT purification

GFT was obtained from a commercial sample (Roundup WG^®), containing GFT (720 g kg⁻¹), isopropylamine (72.5 g kg⁻¹) and other inert components (207.5 g kg⁻¹). In a fume hood, 2 g of the commercial sample were mixed with 15 mL of water in a beaker under magnetic stirring. After, 15 mL of sodium hydroxide solution (4 mol L⁻¹) was added dropwise and the solution was heated (80°C) until isopropyl amine vapor ceases. Then, at room temperature, 5.6 mL of hydrochloric acid (37%) was added dropwise under stirring and after 30 min the solution was dried under reduced pressure and a solid was obtained. The solid was boiled in ethanol, filtered, solubilized with minimum boiling water amount, decolorized with active charcoal, filtered again and evaporated until half water volume. The precipitate was filtered, washed with acetone and dried in a high vacuum pump. It was obtained 0.55 g of N-phosphomethylglicine hydrochloride, GFT salt (37% yield). ¹H NMR (200 MHz, D₂O, TMSP): δ=3.24 (duplet J_HP=12.6, 2H) and 3.93 (singlet, 1H). ³¹P NMR (80 MHz, D₂O, H₃PO₄) δ=8.7 (triplet J_PH=12.6) [23].

Kinetic measurements

The reactions were monitored by UV-vis spectroscopy (Agilent Cary 60) by following the absorbance between 200 and 700 nm, under controlled temperature with an ultra-thermostatic bath (SP Labor SP-152). Two experimental conditions were carried out: (i) Condition 1 (excess of GFT): the reaction started with the addition of an aliquot of 20 μL of a DNFB solution (1.5×10⁻³ mol L⁻¹ in ethyl acetate) in a 3 mL cuvette (1 cm optical path), giving a final concentration of 1×10⁻⁵ mol L⁻¹. The cuvette contained aqueous solution of GFT (5×10⁻² mol L⁻¹), buffered with bicarbonate (0.1 mol L⁻¹), at 25 C; (ii) Condition 2 (excess of DNFB): the reaction started with the addition of an aliquot of 30 μL of a GFT solution (1×10⁻³ mol L⁻¹ in water) and an aliquot 30 μL of a DNFB solution (0.1 mol L⁻¹ in ethyl acetate), in a 300 μL quartz cuvette (0.1 cm optical path), giving a final concentration of 1×10⁻⁴ mol L⁻¹ and 1×10⁻² mol L⁻¹ of GFT and DNFB, respectively. The media was buffered with bicarbonate (8×10⁻² mol L⁻¹) at 25°C. The hydrolysis reaction of DNFB was also followed for comparison purposes. The reaction started with the addition of an aliquot of 20 μL of a DNFB solution (1.5×10⁻³ mol L⁻¹ in ethyl acetate) in 3 mL quartz cuvette, giving a final concentration of 1×10⁻⁵ mol L⁻¹ in a buffered solution (pH 8, bicarbonate 0.1 mol L⁻¹), at 35–55°C range. The alkaline hydrolysis was evaluated similarly by varying the concentration of KOH at 25°C. The kinetic profiles were fitted using iterative least-squares software. In the case of reactions with excess of GFT, reactions followed single reaction pseudo-first order equations given in the literature[24]. In the case of reactions with excess of DNFB, two parallel reactions of pseudo-first order were observed (hydrolysis reaction of DNFB and reaction of GFT with DNFB) which were fitted with eq. 1.

where A_inf(1) is the infinite absorbance for the hydrolysis reaction of DNFB; k_H2O is the rate constant for the hydrolysis reaction of DNFB; A_inf(2) is the infinite absorbance for the reaction of DNFB with GFT; k_GFT is the rate constant for the reaction of GFT with DNFB; A₀ is the absorbance at the beginning of the reaction and t is the time of the reaction.

NMR measurements

The equipment used was an NMR Bruker DPX 200 operating with 4.7 tesla, 200 MHz for ¹H and 80 MHz for ³¹P, with 3-(trimethylsilyl) propionic-2,2,3,3-d4 acid, sodium salt (TMSP) and H₃PO₄ capillary as reference, respectively. The reactions were performed directly in the NMR tube, where GFT (1 mg) and DNFB (1 mg) were added to 500 μL of a D₂O buffer solution (bicarbonate 0.1 mol L⁻¹), pD 9 and consecutive spectra were obtained.

Colorimeter construction

The colorimeter construction was based on a previous report using Styrofoam, which was proposed as an experiment for graduate course in physical chemistry [25]. Herein, we improved this apparatus, as follows and applied for the analysis of the herbicide GFT. In a wood box (8 cm×6 cm×6 cm), two holes (1 mm) with 4 cm of height in both sides (A and B), to connect the LEDs, were made (Fig. 3; Side A and Side B). In the side A, two additional holes (1 mm) were made with height of 3 and 5 cm to connect a (100 Ω) resistor. A styrofoam material was placed inside the box with a hole to place the cuvette. The blue LEDs (430 nm) were situated in front of each other and a resistor (100 Ω) was placed in the 3 and 5 cm height holes. In this case, the objective is that LED 2 illuminates the cuvette and the amount of transmitted light at LED 1 is measured by the potential using a multimeter. Indeed, the wires from LED 1 were welded in a plug to couple to a multimeter. Then a power supply (5V; 1A), for example an old cell phone charger was connected to the colorimeter. It is important to notice the sustainable appeal in the construction of the colorimeter in this sense, since we were able to reuse cell chargers that are widely discarded and could be considered an electronic trash problem. The positive wire from the power supply was welded to the resistor (a), the second wire from resistor was welded in the positive pole of LED 2 (b) and the negative wire from energy supply was welded in the negative pole of LED 2 (5V; 1A), (Fig. 3).

Fig. 3:

Handmade colorimeter schema.

Determination of the molar absorptivity of the GFTD

The molar absorptivity of the product GFTD at pH 9 was determined using the colorimeter under conditions with excess of GFT, since under this condition the only product formed is GFTD. The absorbance was correlated to the potential (Volts) of LED 1, measured using a multimeter. The potential value was converted to absorbance using eq. 2 and the molar absorptivity (ε_GFTD) was calculated using the Lambert-Beer equation [26], considering that the concentration of GFTD is equal to the initial concentration of DNFB (totally consumed). This was repeated for different reactions and lead ε values with less than 10% error.

where A is the absorbance, V₀ is the potential obtained with 100% transmittance (solvent only) and V is the potential obtained for any given solution.

Detection with handmade colorimeter

The detection and quantification of GFT was carried out using the colorimeter as follows and under conditions with excess of DNFB. The reaction started with the addition of an aliquot of 30 μL of a GFT solution (1×10⁻³ mol L⁻¹) and 30 μL of a DNFB solution (0.1 mol L⁻¹) to a 300 μL cuvette (0.1 cm of optical path) containing 240 μL of a aqueous buffered solution (bicarbonate 0.1 mol L⁻¹), giving a final concentration of 1×10⁻⁴ mol L⁻¹ and 1×10⁻² mol L⁻¹ of GFT and DNFB, respectively. The reaction was carried out at 25°C and pH 9. The reaction progress was monitored by measuring the potential (Volts) of LED 1 and converting to absorbance using eq. 2, after 10 min. The concentration of the product GFTD was obtained considering its the molar absorptivity (ε_GFTD). Finally, it was possible to determine the concentration of GFT in the sample using eq. 3, obtained from combining a first order equation with the Lambert-Beer equation. The concentration of the product (GFTD) at infinity is equal to the concentration of GFT at the beginning of the reaction which is calculated considering the rate constant for this reaction (k_GFT) at the given time (10 min). Thus, the kinetic studies are fundamental for obtaining the rate constant. It should be noted that 10 min was the ideal condition for quantifying GFT, but varying values of times could also be used.

where [GFT]₀ is the initial concentration of GFT of the analyzed sample, A is the absorbance with 10 min of reaction, k_GFT is the rate constant for the reaction of GFT with DNFB (excess of DNFB) and t is the time of reaction, i.e. 10 min, but could be adapted to other times.

Kinetics studies

Kinetic studies were first carried for the hydrolysis reaction of DNFB, since it can occur concomitantly with the reaction of GFT. Therefore, it is necessary to evaluate its contribution to the overall reaction rate. The hydrolysis reaction is slow at 25°C, hence solely the rate constant at pH 8 was calculated from a temperature profile (35, 45, 55°C) using the Arrhenius equation (see the Supporting Information, SI). For the alkaline hydrolysis, rate constants were obtained by varying the concentration of hydroxide. Typical UV-Vis spectra are presented in Fig. 5a, showing the characteristic consumption of DNFB around 230 nm and the formation of the phenolic product DNP with maximum wavelengths at 360 and 400 nm [30]. Pseudo-first order kinetic profiles were obtained at 230 and 400 nm, as shown in the inset of Fig. 5a. Figure 5b shows the hydroxide concentration profile for the alkaline hydrolysis of DNFB, which was fitted with eq. 4, that considers that reaction with water (k_H2O) and the alkaline hydrolysis (k_OH). The slope of the linear plot in Fig. 5b gave the value of k_OH=7.05 min⁻¹ L mol⁻¹. Since obtaining k_H2O from the linear coefficient is not so reliable, this value was obtained from the value of k_obs calculated at pH 8 (25°C) and using eq. 4. A value of k_H2O=1.82×10⁻⁵ min⁻¹ was obtained for the hydrolysis of DNFB. Hence, the hydrolysis reaction is very slow and would take nearly 4 months to complete (5 half lifetimes) at 25 °°C.

Fig. 5:

(a) Typical UV-Vis spectra obtained for the hydrolysis of DNFB (1×10⁻⁵ mol L⁻¹) with KOH 0.01 mol L⁻¹ and 25°C; the inset shows the kinetic profiles at 230 and 400 nm, fitted with a pseudo-first order equations. (b) Alkaline hydrolysis profile for DNFB at 25°C. Solid line corresponds to the fit with eq. 4.

In the following, the reaction of DNFB with GFT (excess of GFT) was studied and its typical kinetic profiles are shown in Fig. 6a. In contrast to the hydrolysis of DNFB, the reaction with GFT occurs readily (reaction time under 1 h) at 25°C, already suggesting that is the predominant reaction occurring. Figure 6a shows the known consumption of DNFB and the formation of a new product at 400 nm (distinct from the DNP profile), that was attributed to the product GFTD. The same profile was obtained in all the pH range ~6–11. The kinetic profile for GFTD formation at 400 nm also obeys a pseudo-order behavior, shown in the inset of Fig. 6a. This reaction forms the product GFTD which presents a yellow colour. Indeed, during the reactions which initially was colorless (only GFT and DNFB), the yellowish color appeared (picture shown in Fig. 6a). This color change enables a simple straightforward colorimetric detection of GFT, which can be done visually, in a short period of time and under mild conditions (weakly basic aqueous medium and room temperature). Even though the hydrolysis reaction of DNFB also leads to a colored product (DNP, yellow), the kinetic studies rule out the significant contribution of this reaction under the reaction conditions evaluated (reaction takes up to 4 months). In addition the spectrum profile for the products GFTD and DNP are different. Figure 6b presents the typical spectrum at the end of the reactions for the hydrolysis of DNFB (forms DNP) and the reaction of GFT with DNFB (forms GFTD). Clearly, it is possible to differentiate the profiles of the two products, reiterating that the hydrolysis reaction of DNFB does not contribute in the colorimetric reaction proposed with GFT.

Fig. 6:

(a) UV-Vis spectra obtained for the reaction of DNFB (1×10⁻⁵ mol L⁻¹) with GFT (5×10⁻² mol L⁻¹) at pH 9.5 and 25°C; the inset shows the kinetic profiles at 400 nm and the picture shows the color change of the reaction medium. (b) Reaction spectrum comparison of the products (after reaction completion) for the hydrolysis of DNFB (1×10⁻⁵ mol L⁻¹; 25°C) with KOH 0.01 mol L⁻¹ and the reaction of GFT (5×10⁻² mol L⁻¹) with DNFB (1×10⁻⁵ mol L⁻¹, 25°C).

Fig. 7:

pH rate profile for the reaction of GFT (5×10⁻² mol L⁻¹) with DNFB (1×10⁻⁵ mol L⁻¹), 25°C. Solid line corresponds to the fit with eq. 5.

Figure 7 presents the pH rate profile for the reaction of GFT with DNFB. The results show an increase in the rate constant with higher pH, evidencing that the most reactive species is formed at higher pH (totally deprotonated; pK_a3~10; GFT species in Fig. 4). Interestingly, even at pH 6.5–8.5 where there are very low amounts of the GFT species, the reaction is appreciably faster than the hydrolysis of DFNB, indicating that the monoprotonated species (GFT⁺ pK_a2~6 Fig. 4) also reacts with DNFB forming GFTD. Indeed, all spectra profiles are characteristic for GFTD. Hence, eq. 5 was used to fit the pH profile, which was deduced based on the schematic presented in Fig. 8 [24]. It considers the contribution of the hydrolysis (k_H2O, k_OH, agrees with Fig. 5) reaction and the reaction with the two species: GFT and GFT⁺, related to the second-order rate constants kGFT2, and kGFT+2, respectively. The terms χ_GFT and χ_GFT+ refer to their molar fraction (calculated using pK_as), while [GFT]_t is the total concentration of GFT. Fitting agreed with previous values of k_H2O and k_OH and (i) pK_as of 6.0 and 10.18, consistent with the literature [27]; (ii) rate constants of kGFT2=17.6 and kGFT+2=1.5×10−2 min−1 L mol−1. This evidences that the reaction with GFT is even more efficient that the alkaline hydrolysis of DNFB (kGFT2>>>kOH). Comparing the reaction of GFT with DNFB with its spontaneous reaction, rate enhancements (comparing second-order constants, kH2O2=3.28×10−7 min−1 L mol−1 ) of 5×10⁷ and 4×10⁴-fold are observed for GFT and GFT⁺ species, respectively. This means that the proposed reaction is highly effective and fast. Thus, the reaction completion (estimated as 5 half lifetimes at 0.05 M) reaches 4 min for GFT species and 77 h for GFT⁺. At pH 9, where there is a mixture of the species, the reaction takes 2 h, in contrast to 27 days for the spontaneous hydrolysis of DNFB under the same conditions. Hence, the hydrolysis should not contribute significantly in the reaction of DNFB with GFT, under the reaction conditions, guaranteeing the formation of one product: GFTD.

Fig. 8:

Reaction pathways possible in the reaction of DNFB with GFT.

Henceforth, the kinetic for the reaction of DNFB with GFT is presented herein for the first time, showing that the predominant product formed is the colored GFTD, which is compared to the hydrolysis of DNFB. The results show that the reaction of DNFB with GFT is highly effective, fast and pH-dependent, with two possible reactive species of GFT: totally deprotonated and monoprotonated. The first species is expected for the proposed nucleophilic substitution, while the second is more intriguing. Possibly, the hydrogen of the amine moiety of GFT⁺ can be abstracted by a general basic catalyst (water or the anionic sites of GFT⁺), concomitant to the nucleophilic substitution with DNFB. A similar explanation for the biprotonated species of GFT was discarded, since it would not fit the pH rate profile.

NMR spectroscopy analysis

The chemical structure of the product formed (GFTD) from the reaction of DNFB with GFT has already been mentioned in the literature as a possible product formed [22], but its chemical structure has not been confirmed by any techniques. Previous kinetic studies herein show the UV-Vis characterization but seeking deeper understanding, consecutive ¹H NMR spectra were acquired for the reaction of DNFB with GFT. Figure 9 presents the ¹H NMR spectra obtained in the aromatic and aliphatic region, where the typical signals for DNFB and GFT were attributed. Additional peaks were attributed to GFTD that will be discussed in the following. Table 1 details all attributions.

Fig. 9:

Successive ¹H NMR spectra for the reaction of DNFB (1×10⁻² mol L⁻¹) with GFT (1×10⁻² mol L⁻¹), in D₂O, pD 9, 25°C.

Analysis of Fig. 9 shows the consumption of GFT and DNFB with the decay of the characteristic signals of these compounds in the aliphatic and aromatic region, respectively [23], [31] (2.97 and 3.72 ppm for GFT; 7.70, 8.66 and 9.09 ppm for DNFB). In addition, it is possible to observe the appearance of new signals corresponding to the formation of GFTD (3.97, 4.07, 7.42, 8.21 and 8.68 ppm). The multiplicity of the –CH₂– groups belonging to GFTD is the same of the –CH₂– moiety of GFT. Hence, corresponding to the doublet at 3.72 ppm and a singlet at 4.07 ppm for GFTD. These signals are upshifted relative to the –CH₂– of GFT due to the insertion of the aromatic nitro group which is highly electron withdrawing, which is expected. In the aromatic region, the characteristic signals of 2,4-substituted compounds were observed for the GFTD (a signals set with 2 duplets and 1 doublet of doublets). For DNFB, these signals are more complex due to the hydrogen-fluoride coupling [30]. It is important to note the small intensity of the signals referent to the hydrolysis reaction DNFB that lead to the product DNP (6.44, 8.12 and 8.92 ppm) [32], evidencing the small contribution of this parallel reaction. (See SI).

For ³¹P NMR analysis, a spectrum was acquired only at the end of the reaction that corroborates with ¹H NMR analysis. Figure 10 evidences only two signals due to GFT in 7.9 ppm and a signal upshifted in 12.4 ppm attributed to GFTD. This signal is a triplet with a coupling constant of J_PH 10 Hz, which agrees with the coupling constant of the doublet of doublet in the ¹H spectra (4.07 ppm) attributed to –CH₂– of GFTD (hydrogen-phosphorus coupling). NMR analyzes solidly confirm the structure of the GFTD product, supporting the proposed reaction proposed between DNFB and GFT. Table 1 shows the chemical shifts and multiplicity pattern for each compound signal.

Fig. 10:

³¹P NMR spectra for the reaction DNFB (1×10⁻² mol L⁻¹) with GFT (1×10⁻² mol L⁻¹), in D₂O, pD 9, 25°C, after total DNFB consumption.

Colorimetric detection of GFT with a handmade colorimeter

Seeking the detection and quantification of GFT using the handmade colorimeter, a kinetic study was carried out with excess of DNFB, that would be the optimal condition for field analysis (GFT<DNFB). UV-Vis spectra obtained for the reaction of DNFB with GFT (excess of DNFB) at pH 9 and 25°C are shown in Fig. 11. The results show the band characteristic of GFTD, overlapped with the band of DNP, indicating that under these conditions the formation of DNP is quite significant in the overall reaction with GFT. The kinetic profile obtained at 400 nm, given in the inset of Fig. 11, shows a distinct behavior from the case with excess of GFT (Fig. 6), indicating two concomitant pseudo-first order reactions which are attributed to the GFTD formation (faster reaction) and also to the hydrolysis product DNP (slower reaction). The data was fitted with eq. 1 for parallel reaction, which enabled obtaining the rate constants for the reaction of GFT with DNFB (k_GFT) and its hydrolysis (k_H2O). Thus, it was possible to kinetically separate the contribution of the parallel reactions, which differ significantly from each other. Values of k_GFT=1.6×10⁻² min⁻¹ and k_H2O=1.82×10⁻⁵ min⁻¹ were obtained, showing than even with two reactions occurring, the reaction with GFT is faster than the hydrolysis by nearly 10⁴-fold. Further the hydrolysis constant obtained agrees with the value obtained in the kinetic studies for the hydrolysis (Fig. 5). Although, the rate constant obtained with GFT (with excess of DNFB) cannot be correlated to the one obtained in the kinetic study for the analogous reaction with excess of GFT (Fig. 7). It is important to notice that the thorough understanding in distinguishing the kinetics of the proposed reaction under different conditions is fundamental in the elucidation of the reaction and also for the next steps regarding the detection of GFT with the portable analyzer. Indeed, we showed previously for the reaction of GFT with DNFB that solely the product GFTD is formed under excess of GFT and the pH-dependence was interpreted. In the case of excess of DNFB, which would be the conditions in real field/food analysis that have low amounts of GFT, two parallel reactions occur, predominating the reaction that leads to GFTD, in contrast to the DNP product. Under excess of DNFB, the pH was also varied (pH 8-9.5, see SI), but the competing reactions are difficult to discriminate since in pH<9, the overall reaction is too slow and for pH>9, the alkaline hydrolysis contributes more significantly. Thereby, the optimal conditions considered for the detection was at with excess of DNFB and pH 9, that will be shown with the handmade colorimeter.

Fig. 11:

UV-Vis spectra obtained for the reaction of DNFB (1×10⁻² mol L⁻¹) with GFT (5×10⁻⁴ mol L⁻¹) at pH 9 and 25°C; the inset shows the kinetic profile at 400 nm.

Aiming towards obtaining the handmade colorimeter for the detection and quantification of GFT, firstly the reaction of DNFB with GFT (excess of DNFB) was carried out using the proposed handmade colorimeter, in order to compare with the commercial equipment. Figure 12 shows the kinetic profile obtained using the handmade colorimeter with the blue led (maximum wavelength 430 nm) [25] and the commercial UV-Vis spectrophotometer. They present very similar behaviors and the fits using eq. 1 give similar values, validating the proposed portable colorimeter, even for complex kinetics, such as parallel reactions. These results prove that the colorimeter can successfully substitute the commercial equipment, and further be applied for the GFT quantification as a simple, compact, portable and cheap apparatus.

Fig. 12:

Kinetic profiles obtained with the commercial UV-Vis spectrophotometer and the handmade colorimeter, for the reaction of DNFB (1×10⁻² mol L⁻¹) with GFT (5×10⁻⁴ mol L⁻¹) at pH 9, 25°C.

Figure 13 shows a photograph of the handmade colorimeter used in the detection of the GFT. It’s simple, portable, cheap and easy to use. Moreover, the power supply is accomplished by recycling a problematic electronic trash: a cellphone charger. Indeed, cellphone chargers are widely discarded and can be sustainably reused in the proposed apparatus, not only cheapening the analyzer but also comprising an environmentally friendly appeal. The detection is done by inserting the cuvette containing the reaction medium and measuring the potential variation with the multimeter (converting to absorbance with eq. 2). For that initially, the molar absorptivity of GFTD (ε_GFTD) was obtained, as described in the methodology giving a value of 5.4×10⁴ L mol⁻¹cm⁻¹. In the following the reaction between DNFB (1×10⁻² mol L⁻¹) and GFT (1×10⁻⁴ mol L⁻¹) at pH 9 was carried out for 10 min. For such short time, the contribution of the hydrolysis reaction is very low (less than 1%), hence in the beginning of the reaction the absorbance is only related to GFTD product. Then the absorbance was measured with 10 min of reaction and the concentration of initial GFT in the reaction medium was calculated using eq. 3. It is not necessary that the reaction completes, since the previous kinetic studies allows to confidently obtain the distinct rate constants. Therefore, using kinetic integrated equations (eq. 1), it is possible to determine the concentration of GFT as any given time (eq. 3).

Fig. 13:

Photograph of the proposed handmade colorimeter used in the detection and quantification of GFT.

Accordingly, the handmade colorimeter showed to be effective for detecting GFT, since the product GFTD could easily be measured by the reaction proposed. Further, for validating its quantification with eq. 3, we repeated the reaction in triplicate and the results are presented in Table 2. The results show a successful quantification of GFTY with the analyzer proposed with a mean error of 3% of the initial concentration of GFT.

It should be noted that even the report on a previous version of the colorimeter [25] had not been used for the analysis of agrochemicals, as proposed herein. The obtained values of concentration of the GFT in the solutions are very close to the nominal value prepared, having a low relative error. These results show that the colorimeter is capable of performing quantitative GFT analyzes under these conditions at short times. We believe that the proposed apparatus can be applied in the field more easily and can systematically monitor the use of GFT in food samples, avoiding its indiscriminate use. Further, the portable analyzer can surely be extended to other agrochemicals and even chemical warfare, which have analogous chemical structures. This does not restrict the detection just using DNFB and can be adapted relying on other reactions that produce colored products that can be detected by different LED detectors.