Adsorption–Desorption Behaviour of Imidacloprid, Thiamethoxam, and Clothianidin in Different Agricultural Soils

Abstract

This study evaluated the adsorption and desorption of imidacloprid (IMI), thiamethoxam (THM) and clothianidin (CLO) in an andisol (Freire soil) and an inceptisol (Chufquén soil) from southern Chile with different organic matter and clay contents. The soils had a slightly acidic pH and clay and clay-loam textures. The tests were carried out at 20 °C with CaCl₂ 0.01 M as the electrolyte. Kinetic experiments were performed and isotherms were fitted to the pseudo-second-order, Elovich, Weber–Morris, Freundlich and Langmuir models. The kinetics were best described by the pseudo-second-order model (R² > 0.99), indicating chemisorption; the rate was the highest for THM, although IMI and CLO achieved the highest retention capacities. The Chufquén samples, with lower organic matter but 52% clay, exhibited the highest K_f and q_m of up to 12.4 and 270 mg kg⁻¹, respectively, while the K_d (2.3–6.9 L kg⁻¹) and K_oc (24–167 L kg⁻¹) coefficients revealed a moderate leaching risk. THM was the most mobile compound due to its high solubility. Desorption was partially irreversible (H = 0.48–1.48), indicating persistence in soil. FTIR analysis confirmed the interaction with O-Al-O/O-O-Si-O groups without alterations in the mineral structure. In the soils examined in this study, the clay fraction and variable-charge minerals, rather than organic matter, were more closely associated with the adsorption behaviour of these NNIs.

1. Introduction

The extensive use of pesticides in agriculture has raised concerns about their environmental impact, particularly on agricultural soils where they can lead to long-term contamination [1]. Pesticides such as neonicotinoids (NNIs) have been widely used in agriculture due to their insecticidal properties [2]. Developed in the 1980s, NNIs are among the most widely used insecticides worldwide, accounting for about a quarter of all insecticide applications. Prominent NNIs include compounds such as imidacloprid (IMI), thiamethoxam (THM), clothianidin (CLO) and acetamiprid (ACE), with IMI being the first to be commercially available and still the most widely used today. These compounds are applied in a variety of ways, including as seed coatings and foliar sprays and directly to soil, on a range of crops, including corn, soybeans, oilseed rape, sunflowers, cereals, beets, potatoes, and pastures. Despite their widespread use, only 2% to 20% of the applied dose is typically absorbed by roots and translocated to various plant tissues, including flowers, pollen, and nectar [3,4,5]. This uptake helps to protect plants from insect pests at all stages of growth [6,7]. Conversely, a significant proportion of NNIs remain unabsorbed by plants and accumulate in agricultural soils, posing risks to soil health and biodiversity [8,9,10]. Given the scale of NNI use, their persistence in soils, their potential to leach into water sources and their systemic nature within plants, it is likely that most organisms in agricultural environments will be exposed to these compounds at some point [11].

NNIs are agonists of the nicotinic acetylcholine receptor in insects, causing nerve stimulation at low concentrations and paralysis or death at higher concentrations [12]. They bind more strongly to insect receptors than to vertebrate receptors, making them selectively toxic to insects [13]. IMI, a commonly used NNI, affects both target and non-target insects, including honeybees and predatory beetles. While NNIs pose risks to aquatic ecosystems, much attention has been paid to their effects on bees, particularly when applied to crops on which bees forage. Pollinators may be exposed to even higher levels from foliar sprays [14]. Although direct mortality is rarely observed in bees, there is strong evidence of significant sub-lethal effects, such as a reduced learning and foraging ability in honeybees and bumblebees [15,16,17,18]. These concerns led the European Union to ban the use of three NNIs (IMI, THM and CLO) in bee-pollinated crops from 2018 [18].

The fate of pesticides in soil is influenced by several processes, such as microbial and chemical degradation, adsorption–desorption, volatilisation, and leaching. In soils, NNIs can persist for long periods, which leads to their accumulation. Their half-lives vary, at 187 days for IMI, 121 days for THM and 545 days for CLO, while ACE has a shorter half-life of about 1.6 days [19]. The persistence of NNIs in soil is influenced by factors such as microbial communities, pH, temperature, and biodegradation rates [2]. Of concern is their potential for leaching into groundwater, which poses ecological and human health risks. NNI residues have been found in various environments, including plant nectar and pollen [20,21], wildflowers [22], fruits [23], honeybees [24] and water sources [25,26,27]. For example, Bonmatin et al. [28] found higher concentrations of IMI in soil than in water in Belize, while Zhang et al. [29] reported significant NNI concentrations in soil, water, and sediment samples in Guangzhou, China, with agricultural soils showing the highest levels.

Adsorption and desorption are fundamental processes that affect the behaviour of chemicals in soil. NNIs are highly soluble in water, with solubilities ranging from 184 to 570,000 mg L⁻¹, and typically exhibit low to moderate adsorption capacity. This characteristic increases their potential for leaching into the environment. According to Pietrzak et al. [2], the adsorption capacity of NNIs on soil particles is relatively low. They reported adsorption partition coefficients (K_d) ranging from 0.62 to 1.94 L kg⁻¹ for CLO, from 0.08 to 15.1 L kg⁻¹ for IMI, from 0.88 to 1.80 L kg⁻¹ for THM, and from 0.17 to 35.9 L kg⁻¹ for thiacloprid. Li et al. [30] studied the adsorption of seven NNIs in different soils characterised by a pH range from 5.6 to 7.2 and low organic carbon (OC) content ranging from 0.28 to 0.73%. Their results showed that the K_d values for all NNIs were below 2.0 L kg⁻¹ in all soils tested, indicating the low adsorption affinity and high mobility of these compounds. Similarly, Aseperi et al. [31] reported the adsorption capacity of four NNIs in five different soils, showing ranges of 0.17 to 11.26 µg g⁻¹ and 0.19 to 115.33 µg g⁻¹ at low (2.5 µg g⁻¹) and high (25 µg g⁻¹) concentrations, respectively. The soils studied had pH values between 7.1 and 8.8 and OC contents between 0.8% and 12.5%. These authors concluded that the OC content in the soil plays a crucial role in the adsorption of NNIs, with higher OC levels leading to higher adsorption rates [32,33].

Although adsorption–desorption govern the initial retention and mobility of NNIs in agricultural soils, they represent only the first phase of their environmental fate. Once adsorbed onto organic matter and mineral surfaces, NNIs undergo simultaneous abiotic (hydrolysis and photolysis) and biotic (microbial metabolism) transformations [8], with rates and pathways modulated by edaphic factors (pH, OC content and texture) and microbial community structure [2,8,34]. These processes yield metabolites of varying polarity and sorption affinity, some of which may persist in the soil profile or leach into groundwater, while others may be assimilated by non-target organisms [8]. Therefore, integrating sorption–desorption kinetics with degradation assays, metabolite profiling, and detailed soil characterisation is essential to developing a comprehensive fate model for NNIs in arable land.

Soil organic matter significantly affects pesticide adsorption, with high OC levels in agricultural soils being associated with increased crop yields, particularly for maize and wheat [35]. As a result, these soils are often subject to high pesticide applications. Data on the degradation and adsorption of NNIs in soils are critical to evaluating the fate and transport in the environmental matrices [30]. In this study, we use two soils, an andisol and an inceptisol, characterised by high but differentiated organic matter contents and acidic pHs, which support intense agricultural activity and are therefore subject to the application of NNIs. Studying the adsorption kinetics and adsorption–desorption behaviour of pesticides in soil is crucial because it determines how quickly and to what extent these chemicals bind to or release from soil particles, which directly affects their mobility, bioavailability, persistence and potential leaching into groundwater. In this first part, we present the adsorption kinetics and adsorption–desorption results, which provide a basis for understanding pesticide degradation and metabolic pathways in future studies.

2. Materials and Methods

2.1. Pesticides and Soils

The analytical standards of the insecticides IMI (C₉H₁₀ClN₅O₂; purity, 99.6%), THM (C₈H₁₀ClN₅O₃S; purity, 99.7%) and CLO (C₆H₈ClN₅O₂S; purity, 100%) (

Table 1. Chemical properties of neonicotinoid insecticides (chemical structure, molecular mass, solubility in water and octanol–water partition coefficient).

Pesticide	Chemical Structure	Molecular Mass (g mol−1)	Water Solubility (mg L−1)	Koc
IMI		255.66	610	249–336 *
THM		291.71	4100	56.2
CLO		249.7	327	123

Information from the Pesticide Properties Database (PPDB) [19]. Available at: https://sitem.herts.ac.uk/aeru/ppdb/en/index.htm, accessed on 15 September 2024. * National Pesticide Information Centre. Available at: https://npic.orst.edu/factsheets/archive/imidacloprid.html, accessed on 15 March 2025.

), manufactured by Sigma-Aldrich Production GmbH, Switzerland, were used in the experiments. All reagents used were of analytical or high-performance liquid chromatography (HPLC)-grade.

Two soil types were sampled in May 2023 from different agricultural sites in the Araucanía region of Chile and used in this study. The first one was an andisol of the Freire series (medial, mesic, typical placuands; 38°50′ S, 72°35′ W), and the second one was an inceptisol of the Chufquén series (mixed, mesic, fluventic humic dystrudepts; 38°22′ S, 72°37′ W) [36]. The soil samples were collected from the upper 0–20 cm layer, air-dried at room temperature (20 ± 2 °C) for at least five days, sieved through a 2 mm mesh and characterised at the Soil and Plant Laboratory of the Institute of Agroindustry of the University of La Frontera (see

Table 2. Soil chemical characteristics.

Parameter	Freire Soil	Chufquén Soil
pH (H2O)	6.18	6.03
pH (CaCl2)	5.26	5.16
OM (%)	16	7
OC (%)	9.3	4.1
ECEC (cmol+/kg)	1.24	14.80
Clay (%)	36.0	52.5
Silt (%)	42.2	34.3
Sand (%)	21.8	13.2
Texture class	Clay loam	Clay

OM: organic matter; OC: organic carbon; ECEC: effective cation exchange capacity at natural soil pH. Soil characterisation was carried out in accordance with the standards methods of the Chilean Soil Science Society [37].

).

2.2. Adsorption Kinetics

Duplicate samples of 1.0 g of soil were placed in 50 mL centrifuge tubes and mixed with 10 mL of 0.01 M CaCl₂ containing pesticide at a concentration of 1.0 mg L⁻¹. The tubes were shaken at 300 rpm with a horizontal shaker at 20 ± 1 °C for 15, 30 and 45 min, as well as for 1, 3, 6, 12, 24, and 48 h. After each contact time, samples were centrifuged at 9000 rpm for 10 min at 4 °C, and the supernatant was filtered through 0.22 μm PVDF membranes. The filtrates were stored at 4 °C until high-performance liquid chromatography (HPLC) analysis. This procedure was performed for the three pesticides under study.

2.3. Adsorption–Desorption Experiments

Batch adsorption experiments were carried out by using the classic batch method described by the OECD [38] to obtain adsorption isotherms for the pesticides with the two types of soil. Each experiment involved the use of 50 mL centrifuge tubes containing 1.0 g of soil and 10 mL of a 0.01 mol L⁻¹ CaCl₂ aqueous solution with pesticides at six concentrations (Cs) ranging from 0.1 to 20 mg L⁻¹. The tubes were shaken horizontally in a shaker set to 300 rpm at a temperature of 20 ± 1 °C in the dark for 24 h. After this period, the tubes were centrifuged for 10 min at 4 °C; the supernatant was removed and stored at −20 °C until HPLC analysis.

After removing almost all the supernatant from the adsorption step, an equal volume of background solution was added to the tubes. They were then shaken again at 300 rpm and 20 ± 1 °C in the dark for 24 h. The tubes were then centrifuged and analysed as described for the adsorption experiment. A volume of 2 mL of the supernatants was filtered through 0.22 mm membranes and transferred to HPLC vials for further analysis. The assay was conducted for each pesticide in triplicate to ensure accuracy.

2.4. HPLC Analysis

The liquid samples were analysed by using a Shimadzu Prominence HPLC chromatograph LC-20AT (Shimadzu Scientific Instruments, Columbia, MD, USA), with a diode array detector (SPD-M20A), using a prontoSil column RP-C18 (250 mm length and 4.6 internal diameter, Bischoff Chromatography, Leonberg, Germany).. The mobile phase was acetonitrile and 0.1% phosphoric acid (40:60 v v⁻¹ for IMI and THM and 40:60 v v⁻¹ for CLO) injected at a flow rate of 1 mL min⁻¹. The column temperature was maintained at 40 ± 1 °C and the detector was set for data acquisition at 254 nm, 267 nm, and 269 nm for THM, CLO, and IMI, respectively.

Instrument calibration and quantification were performed using pure reference standards (0.05–50 mg L⁻¹) for each pesticide. The limits of detection (LOD) and quantification (LOQ) were determined according to the IUPAC Harmonized Guidelines for single-laboratory validation of analytical methods [39]. The LOD was calculated using the formula LOD = (Sy/S), where Sy is the standard deviation of the response and S is the slope of the calibration curve. The resulting LODs for IMI, THM, and CLO were 0.012, 0.097, and 0.025 mg L⁻¹, respectively. LOQs were calculated using the formula LOQ = 10(Sy/S), yielding values of 0.036, 0.294, and 0.075 mg L⁻¹ for IMI, THM, and CLO, respectively. The precision (relative standard deviation) was <5%, and the chromatographic response for the calibration curve was linear up to 50 mg L⁻¹ (R² = 0.996). Blank soil samples without pesticide were used to evaluate the soil matrix effect. No significant interference was recorded. Additionally, the recovery of the method was evaluated by spiking 1 g soil samples at two levels (1.0 and 5.0 mg/kg) in triplicate for each pesticide. Samples were quantified with HPLC, with mean recoveries over 90%.

2.5. Data Treatment

2.5.1. Kinetic Model

Pseudo-first-order kinetic reaction model: This model assumes that the rate of adsorption is directly proportional to the number of available adsorption sites, making it a good approximation for systems where surface processes dominate. Its formula is given in Equation (1).

$$\log (q_{\max }-q_t)=\log q_{\max }-{\displaystyle \frac{k_1}{2.303}}t$$

(1)

where q_max (mg kg⁻¹) is the maximum adsorbed amount, q_t is the adsorbed quantity (mg kg⁻¹) at time t, t (h) is the solid solution contact time and k₁ (h⁻¹) is the first-order rate constant.

Pseudo-second-order kinetic reaction model: This model assumes that the adsorption rate is proportional to the square of the number of available adsorption sites, making it suitable for systems where chemical interactions or electron sharing/transfer occur during pesticide adsorption. Its formula is expressed in Equation (2).

$${\displaystyle \frac{t}{q_t}}={\displaystyle \frac{1}{q_{\max }^2{k}_2}}+{\displaystyle \frac{t}{q_{\max }}}$$

(2)

where q_t and q_max are the same as in the previous model and k₂ (mg kg⁻¹ h⁻¹) is the reaction rate constant. The initial adsorption rate, h (mg kg⁻¹ h⁻¹), is defined as q²_max k₂ and can be obtained from the straight-line intercept.

Elovich equation: This model describes sorption kinetics in two phases: a rapid initial phase associated with the movement of the pesticide towards the most accessible parts of the soil and a slower second phase where particles diffuse into the soil micropores. Its linear form is expressed in Equation (3).

$$q_t={\displaystyle \frac{1}{Y}}\ln (XY)+{\displaystyle \frac{1}{Y}}\ln t$$

(3)

where q_t is the adsorbed quantity (mg kg⁻¹) at time t, and X and Y are empirical constants. The intercept (1/Y ln (X Y)) corresponds to the sorbed quantity in the fast phase, while the slope (1/Y) represents the duration of the second phase.

Weber–Morris model (intraparticle diffusion model): This model is widely used in the study of pesticide adsorption to evaluate the role of intraparticle diffusion in the adsorption process. It is particularly useful for systems where diffusion into the porous structure of an adsorbent is an important step in the adsorption mechanism. Its formula is expressed by Equation (4).

$$q_t=k_{\mathrm{int}}{t}^{1/2}+C$$

(4)

where q_t is the amount of pesticide adsorbed (mg kg⁻¹) at time t, C (mg kg⁻¹) is a constant related to the thickness of the boundary layer, and k_int (mg kg⁻¹ h^−1/2) is the intraparticle diffusion rate constant.

2.5.2. Adsorption and Desorption Isotherms

The adsorption isotherms were fitted by using the Freundlich model (Equation (5)).

$$q_e=K_f\ast C_e^{{\displaystyle \frac{1}{n}}}$$

(5)

where q_e is the amount of pesticide adsorbed per unit of soil mass (mg g⁻¹); Ce is the concentration of adsorbate on adsorbent at equilibrium (mg L⁻¹); K_f and K_f,d ((mg kg⁻¹)/(mg L⁻¹)^1/n) are the Freundlich affinity coefficient related to adsorption and desorption capacity, respectively; and 1/n and 1/n_d are the Freundlich intensity and nonlinearity constant of adsorption and desorption, respectively. The organic C distribution coefficient (K_oc) was calculated according to Equation (6).

$$K_{OC}={\displaystyle \frac{K_d}{\%OC}}\ast 100$$

(6)

where K_d is the distribution coefficient for a specific concentration within the concentration range of the adsorption isotherm and OC corresponds to soil organic carbon.

Finally, the desorption hysteresis index (H) was calculated by using Equation (7).

$$H={\displaystyle \frac{{1/n}_d}{1/n}}$$

(7)

where 1/n_d and 1/n are the Freundlich exponents for desorption and adsorption, respectively.

2.6. Fourier Transform Infrared Analysis

Samples of soil, pesticides and pesticide–soil complexes after adsorption were analysed with Fourier Transform infrared (FTIR) spectroscopy coupled to an ATR module (FT IR, Model Cary 630, Agilent Technologies, Santa Clara, CA, USA) equipped with Agilent Resolution Pro Software version 5.0.0.395. The samples were dehumidified in an oven and stored in a desiccator at room temperature before 1.0 mg of the sample was mixed with 250.0 mg of KBr (spectrometric grade). Pellets were made from 25.0 mg of the mixture, and samples were scanned from 4000 to 400 cm⁻¹.

3. Results

3.1. Adsorption Kinetics

The adsorption of IMI, THM, and CLO was evaluated in two agricultural soil samples over different contact times (Figure 1). The amounts of pesticides sorbed increased rapidly during the first hours of contact with the solid solution. In both soils, an apparent equilibrium was reached within 12 h for IMI and CLO, followed by a much slower progress that remained almost constant until the end of the experiment. However, equilibrium for THM began to become apparent after 3–6 h. In addition, both soils showed a higher sorption capacity for IMI and CLO, indicating higher affinity for these more hydrophobic compounds. Considering the contact time of 12 h, the adsorption reached 60%, 19% and 47% in the Freire soil and 47%, 22% and 47% in the Chufquén soil for IMI, THM, and CLO, respectively.

In order to understand the mechanisms involved in the adsorption process of the studied NNIs, the experimental data were analysed based on their fitting to several simple kinetic models: pseudo-first order, pseudo-second order, Elovich and Weber–Morris. The kinetic parameters are given in

Table 3. Adsorption kinetic parameters from linear fitting to pseudo-first-order, pseudo-second-order, Elovich and Weber–Morris models for imidacloprid (IMI), thiamethoxam (THM) and clothianidin (CLO) in two soils with different organic matter and clay contents.

		Soil
Model	Parameter	Freire			Chufquén
		IMI	THM	CLO	IMI	THM	CLO
Pseudo-first order	qmax (mg kg−1)	3.85	1.48	3.30	2.25	1.70	1.21
	K1 (h−1)	0.13	0.02	0.12	0.11	−0.01	0.04
	R2	0.879	0.514	0.961	0.914	0.466	0.93
Pseudo-second order	qmax (mgkg1)	7.18	2.21	5.88	5.07	2.02	3.56
	K2 (kg mg−1 h−1)	0.17	1.29	0.18	0.29	1.05	0.49
	R2	0.997	0.998	0.994	0.998	0.992	0.998
Elovich	α (mg kg−1 h−1)	301.41	40.34	1027.07	2019.27	22,505.75	5319.00
	β (kg mg−1)	1.36	3.74	2.06	2.42	7.49	3.89
	R2	0.955	0.826	0.897	0.972	0.385	0.904
Weber–Morris	Kint (mg kg−1 h1/2)	0.58	0.17	0.42	0.32	0.09	0.92
	C (mg kg−1)	3.77	1.26	3.19	3.18	1.51	2.19
	R2	0.873	0.603	0.9664	0.834	0.325	0.978

.

For the pseudo-first-order model, the observed R² values ranged from 0.466 to 0.961, while the R² values for the pseudo-second-order model were higher, exceeding 0.992. Thus, the adsorption kinetics of IMI, THM, and CLO conform to the pseudo-second-order model, indicating that the adsorption processes in both Freire and Chufquén soils are mainly driven by chemisorption. According to this model, the q_max (mg kg⁻¹) values for the NNIs ranged from 2.21 to 7.18 in Freire soil and from 2.02 to 5.07 for the Chufquén soil; in both soils, the highest values were observed for IMI and CLO and the lowest for THM. The value of the kinetic rate constant (K²) was 1.29 kg mg⁻¹ h⁻¹ for THM, indicating very fast adsorption compared with IMI and CLO, and this is repeated in the Chufquen soil. Overall, the k² values for all three NNIs were generally higher in the Chufquén soil, indicating more favourable adsorption kinetics in this matrix.

We input our results into the Elovich and Weber–Morris models to gain a more complete understanding of the mechanisms (Table 3). Overall, the R² values exceeded 0.826, except for THM, which had an R² of 0.385. The values of the β constants ranged from 1.36 to 3.74 mg kg⁻¹ in the Freire soil and from 2.42 to 7.49 in the Chufquén soil, with the highest value being recorded for THM in both soils. The Weber–Morris model showed a similar trend in R², with values exceeding 0.873, except for THM in both the Freire and Chufquén soils, where its values were 0.603 and 0.325, respectively. This suggests that the model does not adequately describe the behaviour of THM. The calculated values for Kint ranged from 0.09 to 0.92, with higher values being recorded for IMI in the Freire soil and for CLO in the Chufquén soil. Furthermore, the calculated values of C (mg kg⁻¹) for three pesticides were higher than the corresponding Kint values.

3.2. Soil Adsorption–Desorption

The adsorption and desorption isotherm parameters of the three NNIs in the two agricultural soils are presented in

Table 4. Freundlich model of adsorption–desorption isotherms, partition coefficient K_d, organic carbon partition coefficient K_oc and hysteresis index H for imidacloprid (IMI), thiametoxam (THM) and clothianidin (CLO) in Freire and Chufquén soils.

		Soil
Process	Parameter	Freire			Chufquén
		IMI	THM	CLO	IMI	THM	CLO
Adsorption isotherm	Kf (mg1−1/n k−1 L1/n)	3.00 ± 0.05	4.46 ± 0.05	3.75 ± 0.02	8.03 ± 0.02	12.39 ± 0.05	7.72 ± 0.04
	1/n	0.63 ± 0.05	0.96 ± 0.02	1.26 ± 0.05	1.82 ± 0.05	0.221 ± 0.05	1.01 ± 0.02
	R2	0.931	0.9515	0.951	0.973	0.841	0.989
	Kd (L kg−1)	2.56	2.78	2.28	2.81	3.32	6.89
	KOC	28	127	24	75	81	167
Desorption isotherm	Kf,d (mg1−1/n k−1 L1/n)	3.78 ± 0.03	4.40 ± 0.06	8.70 ± 0.04	6.65 ± 0.05	27.13 ± 0.06	18.85 ± 0.02
	1/nd	0.92 ± 0.02	1.60 ± 0.03	0.92 ± 0.02	0.87 ± 0.03	1.43 ± 0.05	1.27 ± 0.02
	R2	0.963	0.961	0.994	0.999	0.824	0.965
	H	1.47	1.24	0.73	0.48	1.48	1.26

. The Freundlich equation fitted the adsorption data of the NNIs well, with the R² ranging from 0.841 to 0.989 and the lowest fit being observed for THM in the Chufquén soil. The K_f varied from 3.00 for IMI to 4.46 for THM in the Freire soil and from 7.72 for CLO to 12.39 for THM in the Chufquén soil, indicating an increase in adsorption capacity for the three pesticides when the amount of organic matter was the lowest and the amount of clay in the soil increased. Regarding the heterogeneity of adsorption, the 1/n values varied from 0.63 to 1.29 in the Freire soil and from 0.22 to 1.82 in the Chufquén soil. According to this parameter, IMI in Freire soil and THM in Chufquén soil have favourable adsorption, CLO in Chufquén soil presented an adsorption line, and THM and CLO in Freire soil and IMI in Chufquén soil present unfavourable adsorption.

The adsorption affinity between the solid and aqueous phases of the soil was higher in the Chufquén soil than in the Freire soil. K_d ranged from 2.28 to 2.78 in the Freire soil and from 2.81 to 6.89 in the Chufquén soil, with the order being IMI < THM < CLO. Regarding K_oc, the values ranged from 24 to 127 in the Freire soil and from 75 to 167 in the Chufquén soil, with the highest value for CLO observed in the Chufquén soil.

Similar to the case of the adsorption isotherms, the Freundlich model also provided a good fit for the desorption isotherms, with R² values ranging from 0.824 to 0.999. The desorption capacity (K_f,d) varied from 3.78 to 27.13, with the lowest value being observed for IMI in the Freire soil and the highest value for THM in the Chufquén soil (Table 4). The 1/n_d ranged from 0.87 to 1.60, indicating favourable desorption (1/n_d > 1) for THM in the Freire soil, as well as for THM and CLO in the Chufquén soil. Finally, the H index ranged from 0.48 to 1.48, indicating hysteresis for CLO and IMI in the Freire and Chufquén soils, respectively.

3.3. Fourier Transform Infrared Analysis (FTIR)

The FTIR spectra (Figure 2) show the corresponding assignments for the standard products IMI, THM, and CLO (a), and the Freire soil (b) and the Chufquén soil (c) without any treatment and after contact with the pesticide. The FTIR spectra of IMI, THM, and CLO show characteristic absorption bands corresponding to their functional groups. The spectra show a broad absorption band around 3300 cm⁻¹, attributed to the N-H stretching vibration; in the case of CLO, the two signals (3285 and 3330) correspond to the two stretching modes (symmetrical stretching and asymmetrical stretching) of the –NH group of nitroguanidine. The presence of aromatic and aliphatic C-H stretching vibrations was observed at around 3100 cm⁻¹ and 2900 cm⁻¹, respectively. In addition, a strong absorption band between 1350 and 1500 cm⁻¹ was assigned to the stretching vibrations of the nitro (NO₂) group, while the C=C stretching vibration of the aromatic ring appeared around 1600 cm⁻¹. There are differences among the spectra of the three pesticides. Although they share some similar functional groups, variations in the intensity and position of certain bands suggest differences in the structure and in the arrangement of the functional groups. The FTIR spectra of the Freire soil (andisol) before and after pesticide adsorption showed no significant changes in the position or intensity of the characteristic functional group bands (Figure 2b). The O-H stretching band (~3300 cm⁻¹) remained constant in all samples. Similarly, the C=C stretching band (~1600 cm⁻¹) showed no significant variations. Finally, in the fingerprint region (~1100–900 cm⁻¹), no significant shifts or changes in intensity were observed for the O-Al-O and C-H bands. The spectrum of untreated Chufquén soil (Figure 2c) showed characteristic absorption bands attributed to Si-OH stretching (3623 cm⁻¹), possible aliphatic C-H or CO₂ asymmetric stretching (2360 cm⁻¹), C=C stretching (1950 cm⁻¹), and O-Si-O and O-Al-O stretching (950 cm⁻¹). No significant peak shifts were observed during pesticide adsorption.

4. Discussion

Understanding the adsorption of pesticides in soils is essential to managing their environmental impact and ensuring sustainable agricultural practices [40]. Adsorption largely determines the mobility, bioavailability, and persistence of these compounds in soil, directly influencing their potential for leaching into groundwater or surface water, as well as their impact on non-target organisms such as pollinators or soil microorganisms [9,41]. Data obtained from adsorption studies are essential to informing transport and degradation models, as well as for environmental risk assessment required by national and international regulations [42]. In this context, the evaluation of adsorption kinetics and adsorption–desorption isotherms allows us to characterise the mechanisms involved and provides key information for understanding one part of the dynamics of these pesticides in different soils.

In this study, two soils were used, the Freire series (andisol) and the Chufquén series (inceptisol), which are soils derived from volcanic ash and sedimentary materials, respectively. These soils were chosen because they represent large areas of high agricultural importance. The main differences between them are the organic matter content and the texture class, with a predominance of clay in Chufquén soil.

Kinetic studies on atrazine [43] and diuron [44,45] carried out in an andisol showed a pattern similar to that observed in this study, characterised by a rapid initial phase followed by a slower second phase. Here, a similar trend was observed in the Chufquén soil, suggesting that both soils have a high adsorption capacity, probably due to their texture and organic matter content [46]. However, this behaviour was more pronounced in the Freire soil, possibly due to its higher organic matter content, greater specific surface area or more reactive mineralogical composition [47]. This trend was particularly evident for the pesticides IMI and CLO, which showed the highest adsorption percentages. In contrast, THM showed the rapid initial adsorption typical of soil adsorption processes, followed by a lower total adsorption capacity, which can be attributed to its tendency to remain in the soil solution due to its higher solubility (4100 mg L⁻¹) compared with the other NNIs.

After applying different kinetic models to the experimental data, our results were well fitted to the pseudo-second-order model for all NNIs in both soils, which is in agreement with other research results [31,48]. According to this model, the adsorption mechanism is controlled by chemical processes, probably involving strong interactions between the NNIs and soil components including mineral surface and water contact [49]. Fernandez-Bayo et al. [48] studied the adsorption process of IMI in different Spanish soils and reported q_max values (mg kg⁻¹) between 1.06 and 6.07, indicating that different adsorption processes could be involved, minimising the influence of soil organic matter content. Similarly to our study, Aseperi et al. [31] studied the behaviour of four NNIs in different soils and indicated that THM was the least adsorbed pesticide, with q_max values of 0.52 and 3.78. According to our results, the k₂ values indicate that the adsorption rate varies depending on both the pesticide and the soil type. THM showed the highest adsorption rate in both soils, and overall, the Chufquén soil allowed for faster adsorption of the studied compounds, possibly due to physicochemical differences that enhance pesticide–soil interactions. In this context, Chufquén soil, characterised by a large amount of clay, could favour an interaction by cation exchange, considering that IMI and other NNIs can be protonated at the soil surface in the -NH of their molecules, especially in soils with low pH [48]. The pH can modify the surface charge of the adsorbent and alter the ionization state of the adsorbate, thereby affecting the adsorption capacity. A study conducted by Yang et al. [49] reported a strong adsorption of IMI onto montmorillonite at pH 4.0. Montmorillonite is a clay that is common in soils derived from volcanic ash such as some andisols and soils with high contents of organic matter and bases, such as those used in this study.

The Elovich model provides a simple way of describing how adsorption evolves with time on surfaces that have a distribution of sites with different adsorption energies. High values of α indicate that adsorption is very rapid in the early moments, while high values of β imply that the adsorption rate decreases rapidly as the adsorption sites become occupied. In this context, the high α value of THM in the Chufquén soil suggests that adsorption is very intense in the first hours, while the opposite would occur in Freire soil. However, although adsorption can start very quickly, it slows down considerably when the solid becomes saturated or the most active sites are filled, which was observed for THM in the Chufquén soil. Similar to the pseudo-second-order model, the Elovich model indicates a contribution of chemisorption to our results. However, the poor fit for THM, a compound with relatively higher solubility than IMI and CLO, could be explained by the fact that the Elovich equation neglects the effects of simultaneous desorption. A similar trend was observed by Rodriguez-Liebana and Peña [50] in dimethenamid, a compound with a solubility higher than 1200 mg L⁻¹. In addition, Aseperi et al. [31] reported that their results were congruent with the two-phase principle of sorption mechanisms proposed by the Elovich model for the NNIs thiacloprid and THM in both soils characterised by different amounts of soil OC but similar amounts of clay.

The adsorption–desorption data were fitted to the Freundlich equation. The results indicate that the adsorption of the NNIs in both soils occurs on a heterogeneous surface, i.e., there is a range of adsorption energies and possibly the formation of several layers of adsorbate, as described for the NNI nitenpyram [51]. The adsorption capacity or affinity of IMI, THM, and CLO (K_f) varied with pesticide and soil properties, with the adsorption capacity observed in our soils being similar to or even greater than that reported for NNIs in other soils. Zhang et al. [8] reported K_f values ranging from 1.01 to 3.42 for IMI and from 0.99 to 3.39 for CLO in soils with OC % ranging from 0.1 to 3.0. Kumar et al. [52] reported lower K_f values of 0.002 for IMI in soils with low organic matter content (<0.95%), indicating a low adsorption capacity of the insecticide in the soils. Asperi et al. [31] reported K_f values of 5.2 to 7.51 for THM in soils with OC % ranging from 0.8 to 12.5, but the higher adsorption capacity was not related to soils with higher OC. In this context, the extent of adsorption would be influenced by the surface chemistry of the adsorbent. Contrary to what was reported by Aseperi et al. [31], THM was the most adsorbed in both soils, probably indicating that the risk of the pesticide contaminating groundwater would be minimised in soils with a high organic matter content. In addition, the increase in the adsorption capacity of the three NNIs in the Chufquén soil would suggest a complementary influence of clay on NNI adsorption, as clay has a greater surface area, offering a greater number of sites to which these NNIs could be bound. According to Fernandez-Bayo et al. [48], some NNIs, such as IMI, can form hydrogen bonds or ion–dipole interactions with water or Ca²⁺, K⁺ ions in the clay through the nitroguanidine structure, thus retarding the movement of these compounds [41]. A recent study by Yang et al. [49] reported that the adsorption capacity of several minerals such as montmorillonite and kaolinite for NNIs increased with increasing K⁺ ion concentration.

Regarding the adsorption intensity indicated by the Freundlich parameter (1/n), values < 1 represent favourable or normal adsorption; values ≈ 1 indicate almost linear adsorption, i.e., the adsorbed amount increases proportionally with the equilibrium concentration; and values > 1 indicate cooperative adsorption. According to our results, the strength of adsorption in the adsorption process of NNIs could be favourable or cooperative, which has been reported for NNIs in different soils [31]. Finally, both K_d and K_oc were calculated to understand the distribution of pesticides between the solid and aqueous phases of the soils studied and how the pesticides adsorb onto the soil due to the amount of OC, respectively. As the isotherms were not linear, K_d and K_oc were determined for a concentration of the pesticides [53]. According to our K_d values, IMI, THM, and CLO showed a moderate soil–water distribution, suggesting a moderate risk of leaching. However, previous studies have reported values lower than 4 L kg⁻¹ as an indication of “low” soil–water distribution for pesticides in volcanic soils [54], such as the andisol evaluated in this study. Therefore, the risk of leaching of the NNIs into the studied soils could be considered moderate to high. Our results are in agreement with other values reported in the literature, e.g., 0.90–16.20 L kg⁻¹ [8] and 0.96- 4.21 L kg⁻¹ [30], but are higher than others, e.g., 0.62–1.94 L kg⁻¹ for CLO, 0.59–2.03 L kg⁻¹ for THM [53] and <1.26 L kg⁻¹ for IMI [55], probably due to the low OC content in the soils studied.

The K_oc obtained by normalising the adsorption coefficient to the OC of the Freire and Chufquén soils varied between them. Li et al. [30] reported K_oc values of 404–548 L kg⁻¹ for IMI, 435–588 L kg⁻¹ for CLO, and 398–530 L kg⁻¹ for THM. Uthman et al. [55] reported K_oc values of 470 and 504 mg g⁻¹ for IMI in the 0–15 cm layer of an ultisol and an entisol, respectively, and Mörtl et al. [41] reported K_oc values of 106–160 mL g⁻¹ for CLO, and 54–106 mL g⁻¹ for THM. The latter values are closer to those observed in our study. Despite its lower OC content, the Chufquén soil had significantly higher K_oc values than the Freire samples. This apparent inconsistency is attributed to the high proportion of reactive clay minerals in Chufquén soil, which provide abundant adsorption sites for NNIs. As conventional K_oc normalises K_d exclusively to OC, the substantial contribution of the mineral fraction is not taken into account, artificially inflating K_oc in soils [56].

The K_f,d values and hysteresis coefficients (>1) observed in both soils indicate that the NNIs are desorbed and that this process is partially irreversible. Studies have reported K_f,d values of 0.14 to 0.62 L kg⁻¹ for THM and 0.23 to 1.05 L kg⁻¹ for CLO, where part of the NNIs was found to be irreversible [53]. Zhang et al. [8] reported K_f,d values of 1.20 to 3.76 L kg⁻¹ for IMI and 1.32 to 5.05 L kg⁻¹ for CLO, and the HI values were > 0, indicating that all NNIs exhibited some degree of desorption hysteresis. However, the H value indicates that in our soils, the desorption of the NNIs studied would be more difficult than their adsorption, presenting partial irreversibility and possibly contributing to higher persistence, mainly due to the binding to organic matter and mineral particles [57], it reflects the potential for these compounds to persist in the solid phase, which could influence their bioavailability and subsequent degradation. The variability in 1/nd and the excellent R² fits suggest multiple binding sites and entrapment mechanisms, leading to less favourable desorption at low concentrations for IMI in both soils, and positive desorption at low concentrations for THM. These results emphasise the need to assess the adsorption–desorption behaviour to fully understand the mobility and environmental fate of these pesticides.

NNIs are compounds with relative stability and resistance to degradation [49]. Therefore, understanding their fate and degradation potential within the soil compartment is essential to elucidate the pathways of residual pesticides and their transformation metabolites at application sites [58]. The persistence of NNIs in soil poses a threat to the soil microbiome, affecting its functionality. A review by Briceño et al. [10] reported that NNIs can lead to either a decrease or increase in the population, diversity, and specific groups of microorganisms that play key roles in soil fertility, as well as impact microbial activity. On the other hand, the structure and abundance of microbial communities exposed to NNIs may influence the different degradation rates and selective degradation pathways [8]. The environmental behaviour of NNIs in soils is varied. Therefore, studying the adsorption and degradation in soil is essential to understand their environmental fate. Together, these processes help assess ecological risk and inform sustainable management strategies. The adsorption results will provide key insights into the retention behaviour of NNIs in the studied soils. This understanding will support future research on their degradation dynamics and transformation pathways.

Finally, FTIR analysis is a valuable technique for the study of the changes in functional groups during the interaction between adsorbates and adsorbents [51]. Zhang et al. [8] studied the interaction of IMI, CLO and thiacloprid and observed that after adsorption on soil, the bands associated with N-H, C-H, C=N, and -NO₂, among others, disappeared, suggesting the involvement of the chlorine, nitrogen, and sulphur atoms present in the molecules studied. In our case, after NNI adsorption, the soil spectra were similar, suggesting that the adsorption process of IMI, CLO, and THM did not disrupt the basic structure of the soils [51]. Andisols are rich in organic matter, have a high specific surface area, and contain short-range ordered minerals such as allophane [59]. The hydroxyl groups of Si, Al, and Fe, together with organic functional groups such as enolic, phenolic, and carboxylic groups, are key players in determining chemical adsorption capacity and interaction in andisols and inceptisols, with a higher concentration of O-Al-O and O-Si-O groups originating from Al-Si in allophane/imogolite in the latter. These functional groups could provide adsorption sites for specific interaction with NNI pesticides, probably through intermolecular hydrogen bonding and/or chemisorption [8,51] or by cation exchange, considering that part of the pesticide can protonate (-NH⁺) and adsorb on negative sites of the clay surface, especially in acidic soils with variable charge.

5. Conclusions

Overall, our results demonstrate that the interactions of IMI, THM, and CLO in an andisol (Freire soil) and an inceptisol (Chufquén soil) from southern Chile are mainly governed by chemisorption processes following pseudo-second-order kinetics. Despite the higher organic matter content of the andisol, both the adsorption rate and capacity for IMI, THM, and CLO were greater in the Chufquén soil, indicating that clay fraction and reactive mineral surfaces play a more decisive role than organic carbon in retaining these three compounds. The Freundlich isotherm parameters (Kd = 2.3–6.9 L kg⁻¹; Koc = 24–167 L kg⁻¹) suggest a moderate-to-high leaching potential, especially for THM, given its high solubility. A desorption experiment on IMI, THM, and CLO showed hysteresis indices of >1 and high K_f,d values, which indicate a partial and irreversible release of the pesticides and their persistence in the soil profile. FTIR analysis confirmed that the adsorption of these three pesticides occurs without alterations in the basic soil structure, suggesting the involvement of O-Al-O/O-O-Si-O groups, hydrogen bonds, and possible cation exchange with negatively charged sites. These results emphasise that for IMI, THM, and CLO, mineralogical factors, particularly clay content and mineral type, should be explicitly incorporated into environmental fate assessments and risk management strategies in volcanic and sedimentary agricultural soils. Future research should systematically evaluate the influence of soil pH and ionic strength on IMI, THM, and CLO adsorption–desorption using controlled batch and column experiments. Furthermore, carefully studying the degradation of these pesticides in these soils, taking into account metabolic transformation, will provide a more advanced perspective in order to assess the environmental risks of these insecticides and develop sustainable pesticide management strategies.