A Quick and Simple Chromatographic Method to Investigate the Adsorption of Glufosinate on a Subtropical Oxisol: Affinity Comparisons With Glyphosate

ABSTRACT

This work presents a fast and straightforward chromatographic method with fluorescence detection to quantify glufosinate in soil solutions. The method is based on the reaction between the primary amine group of glufosinate and o‐phthaldialdehyde with 2‐mercaptoethanol, producing a stable fluorescent derivative in under 2 min, which remains stable for more than 70 min (relative standard deviation [RSD] = 3.6%). The method does not require automated timing systems. Glyphosate, which lacks a primary amine, does not interfere, and potential interferences from aminomethyl phosphonic acid and ammonium are effectively circumvented by chromatographic separation, as their indole derivatives are well resolved from the glufosinate derivative. An isocratic chromatographic run using 20:80 methanol:piperazine‐N,N′‐bis(2‐ethanesulfonic acid) buffer at 1.0 mL/min lasts 5 min. The method showed linearity over the concentration range of 0.10 and 20 µmol/L, with detection and quantification limits of 0.015 and 0.050 µmol/L, respectively. Calibration curve reproducibility is high (slope RSD < 2.1%, n = 5). The method was effectively applied to study glufosinate adsorption in an Oxisol, revealing a maximum adsorption capacity of 2.5 ± 0.2 µmol/g, significantly lower than that reported for glyphosate.

Abbreviations

AMPA

aminomethyl phosphonic acid

FMOC

fluorenylmethyloxycarbonyl chloroformate

HILIC

hydrophilic interaction chromatography

HPLC

high‐performance liquid chromatography

LOD

limit of detection

LOQ

limit of quantification

MS

mass spectrometry

OPA‐2MCE

o‐phthaldialdehyde in the presence of 2‐mercaptoethanol

PIPES

piperazine‐N,N′‐bis(2‐ethanesulfonic acid)

PVDF

polyvinylidene fluoride

1. Introduction

Glufosinate ammonium is a broad‐spectrum, non‐selective, post‐emergence herbicide that inhibits the enzyme glutamine synthetase. It has been used to control a wide range of weeds in the cultivation of grains and cereals, such as corn, wheat, and soybeans. Glufosinate use has dramatically increased after the introduction and commercialization of genetically modified crops resistant to this herbicide. Thus, there are growing environmental concerns regarding glufosinate leaching and runoff, with consequent contamination of surface and groundwaters [1, 2, 3, 4].

Studying the transport of an herbicide through environmental compartments requires the determination of kinetic and equilibrium adsorption data, thus demanding several determinations of the free concentration of herbicide in contact with the soil. Therefore, analytical methods for these studies require the following features: high sampling throughput, high sensitivity, selectivity, and detectability, as well as a wide linear dynamic range, aiming to examine diverse levels of site occupation without the need for sample dilution or preconcentration [5].

Glufosinate, similarly to glyphosate (the currently most used herbicide around the world), is ionic, amphoteric, and highly soluble in water. These properties make quantification in soil extracts and water samples problematic [6]. Currently, most methods are based on gas or liquid chromatography coupled to mass spectrometers (LC‐MS) [7]. In LC, ion exchange has been extensively used [8]. Alternatively, pre‐column derivatizations have been used to produce fluorescent derivatives and enhance the retention on reversed‐phase stationary phases before the detection by mass spectrometry [9, 10]. Another strategy is the use of hydrophilic interaction chromatography (HILIC) columns to separate highly polar and hydrophilic compounds [11, 12]. Since separations occur in polar stationary phases with aqueous/organic mobile phases, this technique is often considered a normal‐phase separation in a reversed‐phase mode. In HILIC, the retention of polar compounds is greater than that of nonpolar compounds, different from reversed‐phase chromatography. Despite the advantages of not requiring derivatization, HILIC has some disadvantages, such as a lack of robustness [13, 14], resulting in retention time shifts, matrix effects, and a need for high dilution factors, which requires sensitive MS detection [15].

The most common pre‐column derivatization is made with 9‐fluorenylmethyloxycarbonyl chloroformate (FMOC), which is the basis of the method recommended in the ISO 16308:2014 norm for water analysis. This reaction is slow, and literature reports that at least 2 h are necessary to reach satisfactory yields, although most articles report overnight reaction times [9, 16, 17, 18]. The excess reagent must be extracted before the chromatographic analysis using a toxic solvent such as dichloromethane [9]. This derivatization enables the simultaneous determination of glufosinate, glyphosate, and aminomethyl phosphonic acid (AMPA, the major glyphosate metabolite) because FMOC derivatizes both primary (glufosinate and AMPA) and secondary (glyphosate) amines. Derivatization with o‐phthaldialdehyde in the presence of 2‐mercaptoethanol (OPA‐2MCE) [7] is less used because, as a secondary amine, glyphosate does not react directly with OPA‐2MCE, so it is first oxidized to glycine by hypochlorite before mixing with the fluorogenic reagent [19].

The increased use of glufosinate in agriculture demands the development of analytical methods for its determination, independent of its occurrence together with glyphosate. Therefore, the OPA‐2MCE derivatization may be an alternative method since the reaction is fast, independent of additional derivatizations, and does not require the extraction of the reagent excess [10].

In this study, we demonstrate that the derivatization of glufosinate by the OPA‐2MCE reagent enables the development of a fast, selective, precise, and accurate chromatographic method to determine glufosinate in equilibrium with soils rich in iron and aluminum oxides. The method allowed us to compare the sorption properties of glufosinate with glyphosate, previously studied in soils collected in the state of Paraná [20], south of Brazil, one of the states with the largest agricultural production in Brazil.

2. Materials and Methods

2.1. Apparatus and Reagents

Chromatographic experiments used a Dionex Ultimate 3000 Dual Micro LC system (Dionex Softron GmbH, ThermoFisher Scientific, Germany) composed of dual micro DGP‐3600 RS pumps, an SRD‐3600 in‐line degasser, a WPS‐3000SL automatic sampler, and a TCC 3000SD column compartment. Detection was made using a Vanquish Core fluorescence detector (ThermoFisher), coupled to an 8 µL illuminated flow cell. The software Chromeleon 7.3 controlled the instrument, data acquisition, and data processing. Separation was achieved in a 50 × 4.6 mm Onyx C₁₈ monolithic column from Phenomenex coupled to a 10 × 4.6 mm guard column also from Phenomenex (Torrance, CA, USA).

Pestanal grade glufosinate‐ammonium was purchased from Merck (Merck KGaA, Darmstadt, Germany). All other reagents were of analytical grade and purchased from Merck. Methanol for LC, LiChrosolv, was supplied by Supelco (Merck KGaA, Darmstadt, Germany). Piperazine‐N,N′‐bis(2‐ethanesulfonic acid (PIPES) from Merck was used to buffer the chromatographic mobile phase, which consisted of 10:90 (v/v) methanol: 0.010 mol/L PIPES (pH 7.0). The derivatizing fluorogenic reagent was prepared by dissolving 30 mg of OPA in 1 mL of methanol and 50 µL of 2‐MCE, completing the volume to 50 mL with 0.040 mol/L borate buffer (pH 9.5). This OPA‐2MCE reagent has a 1:3 molar ratio of OPA to MCE and was prepared weekly and stored in the refrigerator in an amber flask while not in use [21, 22].

Batch adsorption/desorption experiments were made in a thermostatic orbital shaker from Marconi (Piracicaba, SP, Brazil), set at a stirring rate of 250 rpm, and 25.0 ± 0.5°C. All the adsorption/desorption experiments were done in polypropylene Falcon tubes (capacity of 15 mL), horizontally accommodated inside the shaker. Dispersions were centrifuged at 4000 × g for 15 min in a Sorvall ST16R refrigerated centrifuge (Thermo Fisher Scientific, Waltham, MA, USA) and filtered through 0.45‐µm polyvinylidene fluoride (PVDF) syringe filters (Sartorius Stedim Biotech GmbH, Germany).

Chromatographic separations were isocratic with a 20: 80 (v/v) methanol: 0.010 mol/L PIPES buffer at a 1.0 mL/min flow rate. Excitation wavelength was 340 nm, and fluorescence emission was measured at 430 nm. The injection volume was 10 µL in all experiments.

2.2. Samples

Four soils from the Londrina municipality in the state of Paraná (Brazil) and one soil from the municipality of Candido Motta in the State of São Paulo (Brazil) were studied. At the laboratory, the soils were dried in a vacuum oven at 50°C until constant weight, gently ground to pass through a 250‐µm sieve, and stored in a desiccator. Hanke et al. characterized the soils, and their main characteristics are given in Table S1 [23]. All selected soils were developed from basalt weathering, which resulted in high contents of iron oxides in the clay fraction (Table S1).

2.3. Selectivity/Matrix Effects

1.5 g of soil (Alfisol A, Ultisol A, Ultisol B, and Oxisol) was weighed and dispersed in 250 mL of 0.01 mol/L CaCl₂. The dispersions were incubated at 25°C under orbital agitation at 220 rpm for 20 h. After this period, the dispersions were distributed into six 50 mL Falcon tubes and centrifuged for 15 min at 4000 × g. After centrifugation, the supernatants were filtered using 0.45 µm PVDF membranes. Glufosinate solutions with concentrations of 0.5, 1.0, 2.5, 5.0, 10.0, 15, and 20 µmol/L were prepared in the 0.010 mol/L CaCl₂ soil extracts to evaluate the matrix effect.

2.4. Adsorption Kinetics

A stock 20 µmol/L glufosinate in 0.01 mol/L CaCl₂ (10.0 mL) was distributed into 15 mL Falcon tubes containing 100 mg (± 0.01 mg) of soil. The tubes were incubated at 25°C under orbital agitation at 220 rpm. After incubation times of 0.25, 0.50, 1, 2, 4, 8, 24, and 48 h (in duplicate), the tubes were centrifuged at 4000 × g for 15 min. The supernatant solution was filtered through a 0.45 µm PVDF membrane before derivatization and analysis by high‐performance LC (HPLC).

2.5. Adsorption Isotherm

100 mg (± 0.01 mg) of Oxisol was weighed into 15 mL Falcon tubes and dispersed in 10.0 mL of 1.0, 5.0, 10, 20, 30, and 40 µmol/L glufosinate solutions in 0.01 mol/L CaCl₂. The tubes were incubated at 25°C under orbital agitation at 220 rpm for 2 h. After incubation, the tubes were centrifuged at 4000 × g for 15 min, and 9.0 mL of the supernatant was aliquoted and filtered through a 0.45 µm PVDF membrane for derivatization and HPLC analyses.

2.6. Desorption Isotherm

The solid residue and the remaining 1.0 mL of the solution inside the tube used to construct the adsorption isotherm were dispersed in 9.0 mL of 0.01 mol/L CaCl₂. The tubes were incubated at 25°C under orbital agitation at 220 rpm for 2 h. After incubation, the tubes were centrifuged at 4000 × g for 15 min, and 5.0 mL of the supernatant was filtered through a 0.45 µm PVDF membrane for derivatization and HPLC analyses.

2.7. Data Treatment

The adsorbed amount of glufosinate (q) was computed as:

$$q=\frac{\left(C_i-C_t\right)V}{m}$$ (1)

where $c_i$ is the initial concentration of the herbicide (in µmol/L), ${\mathrm{c}}_t$ is the free concentration after a given contact time with the adsorbent, V is the volume of solution where the adsorbents were dispersed (10.0 mL), and m is the mass of adsorbent. As the equilibrium sets, the term ${\mathrm{c}}_t$ becomes ${\mathrm{c}}_e$, that is, the free concentration at equilibrium.

2.7.1. Kinetics

The experimental data were fitted by Lagergren's 1st order, pseudo 2nd order, and intraparticle diffusion kinetic models, given by Equations (2)–(4), respectively:

$$q_t=q_e\left(1-ex{p}^{-k_{1}t}\right)$$ (2)

$$q_t=\frac{q_e^2{k}_{2}t}{q_e{k}_{2}t+1}$$ (3)

$$q_t=k_{id}{t}^{0.5}+C$$ (4)

where q_t is the amount of glufosinate (µmol) adsorbed per gram of adsorbent at time t, q_e is the fitted amount at the equilibrium (µmol/g), and k₁ , k₂ , and k_id are the Lagergren's 1st order [24], pseudo 2nd order [25], and intraparticle diffusion rate constants [26], respectively. The term C (Equation (4)) is related to the thickness of the boundary layer.

2.7.2. Isotherms

The values of q plotted as a function of ${\mathrm{c}}_e$ were fitted to the Langmuir [27] and the Freundlich equations (Equations (5) and (7), respectively) [28].

$$q=q_{max}\frac{K_L{c}_e}{1+K_L{c}_e}$$ (5)

where q_max is the maximum amount of glufosinate adsorbed per unit mass of soil (µmol/g), and K_L is the Langmuir adsorption constant (L/µmol).

From the K_{L ,} it is possible to calculate the dimensionless constant separation factor (R_L ) using Equation (6):

$$R_L=\frac{1}{1+K_L{C}_0}$$ (6)

where C₀ is the largest initial concentration of adsorbate used in the adsorption isotherm (40 µmol/L in the present study). If 0 < R_L < 1, the adsorption is considered favorable, if R_L > 1, the adsorption is unfavorable, and if R_L = 0, the adsorption is irreversible [29].

The Freundlich equation is represented by:

$$q=K_F{c}_e^{1/n}$$ (7)

where K_F is the Freundlich empirical constant related to adsorption capacity (µmol/g)/(µmol/L) ⁿ , and 1/n is the nonlinearity parameter associated with the energetic heterogeneity of the adsorption sites. All the fittings were made with the Origin 2020 64‐bit Academic software (OriginLab Corporation, Northampton, MA, USA) using the Leveneberg‐Marquardt iteration algorithm. The maximum number of iterations and tolerance were 400 and 1 × 10⁻⁹ as the software's default [30, 31].

3. Results and Discussion

3.1. Derivatization and Stability of the Indole Derivative

Molnár‐Perl and Hanczkó, along with Molnár‐Perl, reviewed the derivatization of amino acids and primary amines using the OPA‐2MCE reagent, detailing the optimized reaction conditions, reagent preparation, stability, and chromatographic behavior of OPA‐2MCE amino derivatives [21, 32]. The OPA‐2MCE derivatization method stands out for its simplicity, sensitivity, and selectivity. It reduces analytical effort, time, and cost, as OPA‐2MCE amino acid derivatives form within 1.5–7 min and remain stable for at least 109 min. However, caution is required for amino acids and amines containing the ‐CH₂NH₂ moiety, which are converted into other derivatives, leading to a significant signal reduction, even when summing the responses of all derivatives [21]. Automated methods with strict reaction time control were developed to address these stability issues [33, 34]. Unlike glycine (the glyphosate derivative that reacts with the OPA‐2MCE reagent) and AMPA, glufosinate lacks the ‐CH₂NH₂ moiety, and no stability issues were anticipated. This reaction characteristic was tested by derivatizing a 20 µmol/L glufosinate solution and performing sequential injections of the same solution after a 2‐min incubation. The interval between each injection was approximately 5 min, which is the total chromatogram time. Figure 1 confirms the acceptable stability of the derivative, as approximately 70 min elapsed between the first and last injections, with the fluorescence reading at the peak maximum showing a relative standard deviation (RSD) of 3.56% (inset of Figure 1).

FIGURE 1.

Chromatograms of a 20 µmol/L indole derivative of glufosinate taken after 2 min of incubation time, followed by sequential injections of the same solution. The inset shows the fluorescence readings at the peak maximum. Chromatographic conditions: Flow rate = 1 mL/min, Sample volume = 10 µL, Isocratic elution with a mobile phase composed of 20:80 (v/v) MeOH: 0.010 mol/L piperazine‐N,N′‐bis(2‐ethanesulfonic acid (PIPES) buffer (pH 7.0), Excitation at 340 nm, Emission at 430 nm.

3.2. Chromatography and Selectivity

Most OPA‐2MCE chromatographic methods run with mixtures of MeOH and phosphate buffer at a pH of around 7.0 [22, 35]. This buffer, however, was avoided because adsorption studies in soils are usually made in 0.010 mol/L CaCl_2, so precipitation of Ca₃(PO₄)₂ may clog the column. Therefore, we used a PIPES buffer, which is commonly used as a non‐complexing buffer in biochemical studies at pH 7.0, without any precipitation evidence.

Isocratic elution with a mobile phase composed of 20:80 (v/v) MeOH: 0.010 mol/L PIPES buffer (pH 7.0) enabled the separation of glufosinate derivative from AMPA and NH₃ derivatives, which are potential interferences (Figure 2a–f). The chromatographic peak of the glufosinate derivative appears at a retention time of 2.10 min (Figure 2e,f). In contrast, the peak corresponding to the AMPA derivative appears at 3.6 min (Figure 2d), implying a resolution of 2.1 relative to glufosinate. The OPA‐2MCE derivative of ammonium was strongly retained in the column, and its peak did not appear in a 5‐min time window corresponding to the total chromatographic run (Figure 2c). Glyphosate does not interfere with the determination of glufosinate because it does not react directly with the OPA‐2MCE reagent (Figure 2b). The possible complexation of Ca²⁺ by glufosinate did not affect the yield of the OPA‐2MCE reaction, provided that the peak areas and retention times were independent of the Ca²⁺ presence (Figure 2e, f).

FIGURE 2.

Chromatograms of (a) blank solution (0.010 mol/L CaCl₂), (b) 5.0 µmol/L glyphosate‐derivative in 0.010 mol/L CaCl₂, (c) 5.0 µmol/L NH₃ derivative in 0.010 mol/L CaCl₂, (d) 5.0 µmol/L AMPA‐derivative in 0.010 mol/L CaCl₂, (e) 5.0 µmol/L glufosinate‐derivative in 0.010 mol/L CaCl₂, and (f) 5.0 µmol/L glufosinate‐derivative in water. Chromatographic conditions: Flow rate = 1 mL/min, Sample volume = 10 µL, Isocratic elution with a mobile phase composed of 20:80 (v/v) MeOH: 0.010 mol/L piperazine‐N,N′‐bis(2‐ethanesulfonic acid (PIPES) buffer (pH 7.0), Excitation at 340 nm, Emission at 430 nm.

In addition to the potential interferences, 0.010 mol/L CaCl₂ soil extracts reveal that the chromatograms are free of peaks at the retention time of the glufosinate derivative. In the 0–5 min time window, peaks in the 3.0–3.5 min range may be assigned to AMPA derivatives (Figure 3).

FIGURE 3.

Chromatograms of 0.01 mol/L CaCl₂ extracts of soil samples. The dashed lines emphasize the time range expected for the glufosinate derivative.

3.3. Evaluation of Matrix Effects

Figure 4 compares the chromatograms obtained in pure 0.010 mol/L CaCl₂ with those obtained in the 0.010 mol/L CaCl₂ extracts of Alfisol and Oxisol, showing no significant matrix effect, as the slopes of the calibration curves in the presence and absence of the matrix did not exhibit any significant difference (Table 1). These results demonstrate that external calibration performed directly in 0.010 mol/L CaCl₂ is adequate for adsorption studies.

FIGURE 4.

Chromatograms and calibration curves (insets) were obtained in (a) pure 0.010 mol/L CaCl₂ and (b) 0.010 mol/L CaCl₂ extract from the Alfisol. Chromatographic conditions are the same as described in the Caption of Figure 1.

TABLE 1.

Slope and intercepts of the calibration curves obtained in pure 0.010 mol/L CaCl₂ with those obtained in the 0.010 mol/L CaCl₂ soil extracts.

Matrix	Slope ×103	Intercept × 103	R 2
0.010 mol/L CaCl2 – day 1	8.51 ± 0.06	−1.0 ± 0.5	0.9997
0.010 mol/L CaCl2 – day 2	8.21 ± 0.07	0.4 ± 0.7	0.9996
0.010 mol/L CaCl2 – day 3	8.23 ± 0.14	−2 ± 1	0.998
0.010 mol/L CaCl2 extract of Alfisol	8.52 ± 0.07	0.5 ± 0.6	0.9996
0.010 mol/L CaCl2 extract of Oxisol	8.58 ± 0.03	3 ± 3	0.9999

3.4. Linearity, Stability, and Detectability

The linear relation between peak area or peak height was observed for concentrations ranging from 0.10 to 20 µmol/L, with R ² > 0.995. The mean slopes of the calibration curves using OPA‐2MCE reagents prepared on five different days resulted in (8.41 ± 0.17) × 10⁻³ L/µmol, implying an RSD of only 2.1%, demonstrating that the method has excellent reproducibility, even without using automation to prepare the fluorescent derivative of glufosinate (Table 1).

The limits of detection (LOD) and quantification (LOQ) were computed as $LOD=\frac{3SD}{m}$ and $LOQ=\frac{10SD}{m}$, where $SD$ is the standard deviation of peak areas of a 0.10 µmol/L glufosinate solution (10 injections, Figure S1) and $m$ is the slope of the calibration curve. The computed LOD and LOQ were 0.015 and 0.050 µmol/L, respectively. These values correspond to 2.7 and 9.0 µg/L. The Minnesota Department of Agriculture started monitoring glufosinate in 2022, verifying that it was detected in 0.7% of groundwater samples and 1.1% of surface water samples in the United States. The highest concentrations detected in groundwater and surface water were 4.5 and 3.2 µg/L, respectively. The Minnesota Department of Health has established a drinking water reference value for glufosinate of 5 µg/L [36]. The lowest Environmental Protection Agency Office of Pesticide Products aquatic life benchmark for glufosinate is 72 µg/L for non‐vascular plants [36]. Therefore, even without including a preconcentration step, the proposed method provides sufficient sensitivity to detect environmentally relevant concentrations and meets the limits established by environmental regulatory agencies.

3.5. Adsorption Kinetics

Negligible adsorption was observed for four of the five soils studied over contact times ranging from 15 min to 48 h, except for the Oxisol. Among the studied samples, the Oxisol was the one with the lowest pH and the highest clay, iron, and aluminum oxide contents (Table S1) [23]. Comparison with previously reported glyphosate adsorption data in the same soil samples suggests that glufosinate exhibits significantly higher geochemical mobility. In the previous study, approximately 80–95% of the initial glyphosate was adsorbed onto all the studied soils within 24 h of contact [20].

For the adsorption of glufosinate onto Oxisol, the adsorption equilibrium was reached within 120 min of contact time, removing approximately 60% of the initial concentration, with no evidence of desorption for up to 48 h (Figure 4). Compared to the pseudo‐first‐order rate law (Equation (2)), the data fitted better to the pseudo‐second‐order kinetic model (Equation (3)), yielding q_e = 0.294 ± 0.003 µmol/g and k₂ = 1.0 ± 0.1 g/(min µmol) with an R² = 0.997 and χ² = 2.7 × 10⁻⁵. The fitting results were independent of the numerical approach (linear or nonlinear regression analyses) used to fit the experimental data [31].

The intraparticle diffusion model was tested; however, the plot of q_t versus t^0.5 (Equation (4)) was not linear and did not pass through the origin, indicating that diffusion is not the dominant mechanism controlling sorption kinetics. The C > 0 value (Equation (4)) further confirms that sorption is rapid and occurs on the external surface of soil particles.

The adsorption of glyphosate in the same soil sample resulted in q_e = 12.7 µmol/g and k₂ = 0.0107 g/(µmol min), with R² = 0.9999 [20]. While the Oxisol exhibited a significantly higher adsorption capacity for glyphosate compared to glufosinate, the adsorption of glufosinate was faster. This is consistent with the kinetic curve, which showed that equilibrium for glyphosate was reached only after contact times exceeding 8 h [20].

3.6. Adsorption Isotherms

Adsorption isotherms were constructed using initial concentrations from 1.0 to 40 µmol/L glufosinate in 0.010 mol/L CaCl₂ at 25.0 ± 0.5°C and a contact time of 120 min, provided that such time is enough for the system to approach the equilibrium (Figure 5). The adsorption isotherms were classified as L‐type, meaning that a limited amount of energetically heterogeneous adsorption sites controls the adsorption (Figure 6). Both the Langmuir (Equation (5)) and the Freundlich (Equation (7)) models fitted the adsorption experimental data with R ² > 0.99 (Table 2). At low initial concentrations, glufosinate occupies the sites with stronger affinity. As these sites are saturated, the adsorption becomes less effective since the herbicide has to interact with the sites of lower affinity. The 1/n < 0.61 reflects the heterogeneity of the adsorption sites. The Langmuir separation factor (R_L , Equation (6)) was 0.29, indicating the adsorption was energetically favorable.

FIGURE 5.

Adsorption kinetics of glufosinate onto Oxisol from a 5 µmol/L solution in 0.010 mol/L CaCl₂ in 100 mg of Oxisol dispersed in 10 mL of solution fitted by pseudo‐1st and pseudo‐2nd order kinetic models using nonlinear regression analysis. The inset shows the linearized graphs for the same kinetic models. Temperature = 25 ± 0.5°C.

FIGURE 6.

Adsorption curves of glufosinate on a sample of Oxisol at 25.0 ± 0.5°C fitted by Langmuir and the Freundlich models adopting 120 min of contact time. The inset shows the desorption curves measured after 120 min of stirring with a fresh 0.010 mol/L CaCl₂ solution.

TABLE 2.

Parameters fitted by Langmuir and the Freundlich models applied to the experimental data obtained in the adsorption of glufosinate onto the Oxisol.

Model	Parameter	Adsorption	Desorption
Langmuir	q max (µmol/g)	2.6 ± 0.2	1.5 ± 0.3
	K L (L/g)	0.06 ± 0.01	0.15 ± 0.05
	R 2	0.995	0.98
	χ 2	0.0021	0.0026
Freundlich	K F (µmol/g)/(µmol/L) n	0.221 ± 0.002	0.22 ± 0.03
	1/n	0.61 ± 0.01	0.62 ± 0.08
	R 2	0.991	0.97
	χ 2	0.0014	0.0033

Consistent with the heterogeneity and finite number of adsorption sites on the Oxisol, the adsorption percentage decreased from 70% to 39% as the initial concentration (c_i ) increased from 1.0 to 40.0 µmol/L. Desorption increased from 22% to 42% over the same concentration range, which is also consistent with the stronger affinity of adsorption sites occupied at lower initial concentrations, leading to a stronger retention of sorbed molecules.

The fitted Langmuir parameters for the adsorption of glyphosate on the same soil, considering the initial concentrations between 0 and 40 µmol/L to be comparable with the conditions used for glufosinate, resulted in q_max = 12.4 ± 0.8 µmol/g and K_L = 0.50 ± 0.1 L/g (R² = 0.92). The maximum adsorption capacity of Oxisol is approximately 5‐fold higher for glyphosate compared to glufosinate. The K_L ​ value, which is related to the Gibbs free energy of adsorption, was about 8‐fold higher for glyphosate (Figure S2). This suggests that glufosinate is unable to reach the same adsorption sites as glyphosate (or the interaction mechanism differs from one herbicide to the other) and that these sites are readily available for the interaction, consistent with the faster attainment of adsorption equilibrium for glufosinate compared to glyphosate.

3.7. Molecular Modeling

For the molecular modeling, the interaction prediction of glyphosate and glufosinate with the main components of Oxisol, Al₂O₃, and Fe₂O₃, was performed. All molecules were initially optimized using the semi‐empirical MNDO calculation method, followed by density functional theory calculations to obtain more precise results and the B3LYP functional, elucidating the interaction of the herbicides with the mineral surface. 6–31G* basis set was employed, using the GAMESS (General Atomic and Molecular Electronic Structure System) software. The Oxisol structure was obtained with the assistance of the CHARMM‐GUI software [37, 38, 39, 40].

The electrostatic interaction energies obtained for glyphosate and glufosinate were ‐83.47 and ‐73.51 kJ/mol, respectively (Figure 7), with the most intense electrostatic interaction occurring with glyphosate. Due to its high content of iron and aluminum oxides, the Oxisol exhibits positive charges in an acidic medium [41]. Glufosinate contains the phosphinate, carboxylate, and amino groups, whereas glyphosate contains the phosphonate, carboxylate, and amino groups, which interact with the charged surfaces of the iron and aluminum oxides, as well as the surface hydroxyl groups (‐FeOH).

FIGURE 7.

Electrostatic potential map of glyphosate and glufosinate. Warm colors represent surfaces with more negative charge.

The pH plays a major role in adsorption, as in an acidic medium, the soil surfaces are positively charged. Therefore, the carboxylate (‐COO⁻) and phosphinate (‐PO₂H) groups of glufosinate can interact electrostatically with these surfaces, and the amino group (‐NH₂) forms hydrogen bonds with the oxygens present in hematite (FeOH₂ ⁺).

Estimated adsorption energies for glyphosate and glufosinate were ‐119.4 and ‐43.7 kJ/mol, respectively. Glyphosate exhibited similar interactions to glufosinate, but with a stronger adsorption force due to the presence of the phosphonate (‐PO₃H₂) group, which is more reactive than the phosphinate. The phosphonate group forms stronger electrostatic interactions with the positive sites of hematite, establishing partial covalent bonds with Fe³⁺ [42, 43]. Figure 8 shows the interaction distances; the interaction between glufosinate and hematite exhibited fluctuations in the order of 0.6 Å, indicating a weak and dynamic interaction [44].

FIGURE 8.

Interaction of glufosinate and glyphosate with hematite, the main component of the studied Oxisol.

The adsorption of glufosinate, although less energetic than that of glyphosate, occurs over a shorter timescale. This behavior is consistent with its less negative electrostatic interaction energy (approx. ‐73.51 kJ/mol). Consequently, its adsorption is weaker, more reversible, and more dynamic, leading to lower overall affinity for soil minerals. A key structural factor is the phosphinate group, which, unlike the phosphonate group in glyphosate, offers fewer coordination sites, resulting in less specific and faster‐dissociating interactions. Furthermore, the higher water solubility of glufosinate significantly enhances its mobility in solution and the environment [37, 44].

4. Concluding Remarks

A fast and robust reversed‐phase chromatographic method was developed to determine glufosinate in soil solutions using the reaction with OPA‐2MCE to produce a fluorescent indole derivative. The limits of detection and quantification of 0.015 and 0.050 µmol/L (2.7 and 9.1 µg/L), respectively, were consistent with environmental concentrations in impacted samples and regulatory levels. The proposed method was used to investigate the adsorption of glufosinate in five soil samples. Significant adsorption was observed only for an Oxisol that, among the studied samples, was the one with the lowest pH and highest clay, iron, and aluminum oxides contents. Compared to glyphosate, glufosinate exhibited lower adsorption capacity and adsorption energy, as confirmed by molecular modeling. These results suggest that glufosinate has a greater mobility than glyphosate and presents a greater potential for contamination of surface and groundwater.

Author Contributions

Tatiana Valeria Furquim: conceptualization, investigation, writing – original draft, methodology, validation, and formal analysis. Luan Viana: investigation, writing – original draft, validation, and software. Jorge C. Masini: conceptualization, funding acquisition, writing – review and editing, visualization, data curation, and resources.

Funding

This work was funded by grants 306674/2021‐1 from the National Council for Scientific and Technological Development (CNPq) and 2024/16458‐0 from the São Paulo Research Foundation (FAPESP). The present work was carried out with the support of the Institute of Chemistry and its Analytical Center—Code CAIQUSP/100.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: jssc70491‐sup‐0001‐SuppMat.docx.

Data Availability Statement

Most of the data presented in this work is included in the main manuscript and the Supporting Information document, and additional data are available from the corresponding author upon reasonable request.