Lithium Sorption/Desorption in Some Alkaline Soils: Modeling of the Kinetics Behavior

Global concern over lithium (Li) in the environment has significantly increased due to its widespread uses. However, the literature review on Li kinetics in alkaline soils is scant. Hence, batch experiments were conducted to quantify and simulate the retention and release of Li in sandy and sandy clay loam alkaline soils. Sorption isotherms were fitted using Freundlich and Langmuir equations. Second-order two site (SOTS) and multi-reaction (MRM) models were applied to simulate sorption/desorption kinetic data. Results revealed the nonlinear behavior of Li sorption isotherms in both studied soils. The maximum sorption capacity (Smax) for the sandy clay loam soil (1268.6 mg kg−1) was three-fold higher than the sandy soil (368.9 mg kg−1) after 7 days of equilibration. The sandy soil and sandy clay loam soils exhibited distinctive kinetic Li sorption/desorption behavior. MRM and SOTS models described Li sorption/desorption kinetic well by considering the reversible and irreversible reactions in alkaline soils. Results indicate that sandy soil exhibits a time-dependent sorption behavior and a restricted desorption kinetic. In contrast, the sandy clay loam exhibited a limited time-dependent Li sorption, while the kinetic behavior was observed during the desorption reaction.

of leaching and the accumulation of Li through the evaporation process (Ammari et al., 2011;Schäfer, 2004). The average of Li in light sandy, loamy, and calcareous soils are 22, 53, and 56 mg kg −1 , respectively (Kabata-Pendias & Mukherjee, 2007). The elevated concentration of Li in the soil causes plant toxicity and crop yield reduction (Duff et al., 2014;Kalinowska et al., 2013;Shahzad et al., 2016). Moreover, the elevated level of Li caused the rise of soil salinity, pH, and the loss of some plant nutrients such as K and N from agricultural soil (Hayyat et al., 2021).
Understanding the reactivity of chemical elements in soils is essential to assess the fate and mobility in the environment (Elbana, 2022;Strawn, 2021). The particular reactivity of Li in comparison with other alkali elements can be ascribed to its polarization strength and solubility properties. Specifically, Li ion is an intensely hydrated element exhibiting a second hydration shell of water molecules (Mähler & Persson, 2012). Mostly, Li ions in soil solution exhibit weak retention into soil colloids, especially with decreasing soil pH and increasing ionic strength (Robinson et al., 2018). In surface-calcareous Jordan Valley soils, Ammari et al. (2011) stated that the water-soluble Li, measured in saturated paste extracts, varied from 0.95 to 2.68 mg L −1 . Rogóż (2010) found that Li was associated with various geochemical fractions in the following order: crystal lattices > bound to Fe-Mn oxides > water-soluble > adsorbed to soil colloids > bound to organic matter. Additionally, Pogge von Strandmann et al. (2021) quantified the distribution of Li in different soil orders, where 1 to 4% of Li is associated with carbonates, and 10 to 25% of Li bounded with oxide fractions, whereas a high fraction of Li (70 to 90%) associated with silicate minerals.
The sorption behavior of Li varies in different soils and for the same soil through various environmental conditions. For instance, based on batch and column experiments using a silt loam and a sandy loam, Akhtar et al. (2003) showed that the higher sorption capacity of Li is associated with the finer soil texture. Anghel et al. (2002) reported that Li-sorption and Lispecific exchange capacities correlated significantly to the tuff contents of smectite and zeolite. Moreover, Hindshaw et al. (2019) identified three bonding configurations for Li on Mg-rich smectite-clay minerals, including the inner-sphere, outer-sphere, and occupying the cavities in tetrahedral sites. Soils and clay minerals exhibit preferential sorption to Li at low initial concentrations (Anderson et al., 1989). Abbas et al. (2021) showed that sorption efficiency in removing Li using clay soil (pH of 7.6 and CaCO 3 of 29%) dropped from 93 to 59% by increasing the initial Li concentration from 5 to 40 mg L −1 in the aqueous solution. Such a behavior can be ascribed to the soil property of a fixed maximum Li sorption capacity. Otherwise, Park et al. (2015) reported that Li occupied an L-type sorption isotherm on manganese oxide, which indicates increasing the affinity with increasing the initial Li concentration. Steinhoefel et al. (2021) emphasized the high affinity of Feoxide, kaolinite, and the clay-interlayer sites to sorb Li during the chemical weathering of soil clays. However, Prodromou (2016) found that gibbsite exhibited lower Li sorption capacity than amorphous aluminum hydroxides, and both chemicals exhibited higher affinity to Li at pH 8 than at pH 6. 5. Moreover, Park et al. (2015) attributed the increase of synthesizedmanganese oxide affinity to sorb Li at alkaline pH to the rise of the electrostatic attractive force between Li and the oxide through a redox or an ion-exchange process via a spontaneous and exothermic reaction.
Modeling Li sorption isotherms and the kinetics of sorption/desorption processes is an essential tool to quantify Li chemical behavior in soils. Newman et al. (1991) found that Li retention on tuff (60% feldspar, 22% quartz, 15% cristobalite, and 1% smectites) was nonlinear reversible sorption with a slower desorption rate than the sorption one. Additionally, Anghel et al. (2002) showed that Li sorption isotherms on various tuff materials exhibited a nonlinear reaction, where a fraction of the sorbed Li exhibited a non-cation exchange reaction. Li and Liu (2020) successfully fitted Li sorption kinetic data to the pseudo-second order kinetic model with a rapid forward reaction and a relatively slower backward desorption. The soil, as a heterogeneous system, contains multiple sorption sites that exhibit a specific rate of reaction for each definite site. Besides, several sorption mechanisms may be involved simultaneously (Strawn, 2021). Considering two reaction sites to simulate chemical reactivity in the soil is recommended whenever initial rapid sorption is followed by a slow rate of reaction (Goldberg et al., 2007). The multi-reaction sites for modeling the sorption behavior of chemicals in soils provide better simulations than using a single reaction site (Amacher et al., 1986;Goldberg et al., 2007;Peng et al., 2018). For instance, multiple reaction approaches, such as the multi-reaction model (MRM) and the second-order two-site Model (SOTS), have been widely applied to simulate the kinetic behavior of various chemicals in soils. Selim and Zhang (2007) successfully simulated the kinetic behavior of As in different acidic soils using the multi-reaction and second-order models. Elbana and Selim (2019) simulated the retention of Cu and Pb in different acidic and alkaline soils based on a multi-reaction sites approach, where the models assume a fraction of the total sorption sites is reversible, whereas another fraction was consecutively or concurrently irreversible.
However, numerous researchers explored Li distribution and occurrence in the natural ecosystem (Bolan et al., 2021;Liu et al., 2020;Sharma et al., 2022), our literature review revealed that limited studies quantified and modeled Li sorption/desorption kinetics behavior in alkaline soils. Thus, the current research aims to quantify and model Li reactivity in alkaline soil. The specific objectives are (i) to assess hysteresis of Li sorption/desorption in two different alkaline soils and (ii) to model the kinetic of Li reactivity based on a multi-reaction/site approach.

Soil Characterization
Two surface soil samples (0-20 cm) were collected to investigate the reactivity of Li in coarse and fine-textured soils from irrigated farms, western Nile delta, Egypt. Soil pH and salinity were measured in 1:2.5 suspension (soil: distilled water). Particle-size distribution and soil texture were determined according to Soil Survey Staff (2022) using the pipette analysis (for fine texture soil) and the dry sieving method (for coarse texture soil). The total CaCO 3 equivalent was measured using pressure calcimeter apparatus (Loeppert & Suarez, 1996). Organic matter (OM) content was quantified according to the Walkley-Black method (Nelson & Sommers, 1996). The ammonium acetate method was applied to measure soil CEC (Estefan et al., 2013). Free iron oxides were extracted using the citrate-dithionite-bicarbonate method (Loeppert & Inskeep, 1996), and Fe concentration was measured using atomic absorption spectroscopy (PerkinElmer atomic absorption spectrometer (Analyst 400).

Batch Experiments
Sorption/desorption batch experiments were conducted to quantify Li reactivity in the studied soils. For sorption experiments, a range of 30 to 400 mg L −1 Li using reagent-grade LiCl was applied. Initial concentrations were prepared in 5 mM KCl as a background solution to avoid the influence of different ionic strengths during sorption/desorption experiments. Specifically, in triplicates, 30 mL of each initial concentration (c i ) was added to a 3-g soil sample into a 50 mL centrifuge tube. Subsequently, the tubes were arranged on a reciprocal shaker at 150 rpm. An aliquot of 1 mL was sampled and diluted to 10 mL from each tube after 2 h, 4 h, 8 h, 24 h, and 7 days by centrifuging for 5 min at 1300 × g. For the desorption experiments, afterward the 7-day aliquot sampling, the remaining solution was replaced with a background solution (30 mL of 5 mM KCl) by weight. Aliquot samples for analysis were collected after 24, 48, 120, and 192 h. The solution was replaced each desorption cycle with 30 mL of the background solution. Moreover, in triplicates, 3-g soil sample with 30 mL of the background solution in absence of Li was treated as a blank during sorption/desorption steps. Concentrations of Li in the collected samples were measured using flame atomic absorption spectrophotometry (a Jenway PFP7 flame photometer, Cole-Parmer Ltd, Stone, Staffs, UK). A potassium ionization suppressant was applied to the sample and the standard solutions for controlling Li ionization during the measuring (Helmke & Sparks, 1996). Sorbed and released amount of Li was computed based on the mass balance calculations of the added initial concentration and the final Li concentration for each sorption/desorption step.

Modeling Li Sorption/Desorption Behavior
Relations between the sorbed Li on solid soil phase (S) in mg kg −1 and the soluble Li in soil solution (C) in mg L −1 were quantified using Freundlich Eq. (1) and Langmuir Eq. (2) sorption isotherms.
Where, K f (mg 1−b L b kg −1 ) and K l (L mg −1 ) are Freundlich and Langmuir coefficients, respectively; b is a dimensionless nonlinear Freundlich parameter; S max (mg kg −1 ) is the maximum Li sorption capacity on the soil. Nonlinear fittings of the observed sorption data to Freundlich and Langmuir equations were obtained using the Marquardt-Levenberg algorithm (SigmaPlot 13.0 software package, Systat Software, San Jose, CA).
Modeling the kinetic behavior of Li sorption/desorption was achieved using a multi-reaction model (MRM) and a second-order two-site (SOTS) model. The MRM and SOTS models are accessible as modules of the Chem-Transport software package (Selim, 2016). Modeling the kinetics of Li sorption/ desorption was achieved by considering two different scenarios. First, the sorption/desorption of Li can be reversible on S 1 site, followed by irreversible retention on S 2 site. Second, a fraction of Li sorption/desorption can be reversible on S 1 site when another fraction is simultaneously irreversible on S irr site Eq. (3) Eq. (4) Eq. (5) represent the MRMmathematical relations for the sorbed amount during the reaction time t in hours: where, k m1 , k m2 , k m3 , and k mirr are the associated rates of reactions in h −1 . The dimensionless reaction order n is estimated by the Freundlich model (value of b in Eq. (1). Symbols and are volumetric water content (cm 3 cm −3 ) and soil bulk density (g cm 3 ), respectively. The MRM Model is a multisite model that accounts for a nonlinear reversible sorption/desorption kinetic and a linear irreversible reaction (Selim & Amacher, 2001). Eq. (6) Eq. (7) Eq. (8) Eq. (9) represent the applied SOTS-mathematical relations: where, k s1 of units of (L mg −1 h −1 ). k s2 , k s3 , and k sirr are the associated rates of reactions (h −1 ) in the SOTS model. The φ represents the vacant sorption sites (mg kg −1 ). The SOTS model relies on two assumptions: (1) the existence of maximum sorption capacity, which is an intrinsic soil property; and (2) site availability and solute concentration in solution limit the rate of the reversible reaction on S 1 site (Elbana & Selim, 2012;Selim & Amacher, 1997). Here, we considered the value of S max estimated by Langmuir Eq.
(2) as an input parameter for SOTS fitting. Both models (MRM and SOTS) consider sorption/desorption experimental data for all tested initial concentrations as one set to provide the best-fitted sorption/desorption kinetic parameters (Selim, 2016).

Results and Discussions
Some of the physical and chemical properties of the studied soils are given in Table 1. Soil-I and soil-II are characterized by non-saline alkaline soil with a pH of 8.01 and 8.05, respectively. Soil texture analyses indicate the variation in the particle size distribution of the studied soils. Specifically, soil-I is sandy soil with a sand fraction of 96.22%, whereas soil-II is classified as sandy clay loam containing 60.76% sand. Soil-I and soil-II contain 0.27 and 0.55 g/kg free iron oxides, respectively. Moreover, the studied soils contain 5.65% and 4.15% of CaCO 3 for soil-I and soil-II, respectively. Soil-II contains OM of 2.66%, which is five folds higher than soil-I. Consequently, soil-II exhibit higher CEC (38.7 cmol c /kg) than soil-I (2.1 cmol c /kg).

Sorption Isotherms
Simulation results of Li sorption isotherms for soil-I and soil-II using Freundlich and Langmuir models (Eqs. [1] and [2]) are shown in Table 2. The results indicate a distinguished sorption behavior for Li on each soil surface. Specifically, after 7 days of equilibration time, the sandy soil (soil-I) exhibits lower sorption capacity (S max of 368.9 mg kg −1 ) than the sandy clay loam soil (soil-II) with S max of 1268.6 mg kg −1 . Furthermore, results reveal that the higher soil CEC exhibited the higher sorption of Li (see Table 1). Several studies reported a positive association between the retention of Li on soil surfaces and the clay content (Robinson et al., 2018). Négrel and Millot (2019) showed that the preferential sorption of Li on Fe-Mn oxides controls the fractionation of Li in soils. In addition, Hoyer et al. (2015) showed that clay minerals with high CEC, such as bentonite and zeolite, exhibited higher Li sorption capacity than kaolin. Therefore, the higher affinity of the studied sandy clay loam can be ascribed to the high CEC, clay content, and free Fe oxides compared with the sandy soil (see Table 1). However, S max of soil-I is more than doubled the CEC of the soil, indicating that ion exchange is not the only sorption mechanism. In agreement with our results, Anghel et al. (2002) concluded that the non-cation exchange process governs a fraction of Li affinity onto devitrified tuff (quartz: 28-55%; feldspar: 16-60%; smectite of < 10%). Also, Prodromou (2016) found that ~ 20% of the sorbed Li on Al-hydroxides was physisorption, whereas ~ 80% was chemisorption. On the other hand, for soil-II, the CEC is large enough to retain Li at the maximum sorption capacity, where only half of the exchangeable sites would be occupied with Li. Sorption isotherm data for soil-I reveals that the sandy soil exhibited kinetic Li sorption behavior. Specifically, S max changed from 913.8 to 368.9 mg kg −1 by extending the equilibration time from 2 h to 7 days. On the other hand, soil-II exhibits a limited kinetic   (Sposito, 2016). Moreover, the Langmuir binding affinity (K l ) of Li sorption on sandy soil increased by about three folds by extending the equilibration time from 2 h to 7 days, whereas the sandy clay loam exhibits limited variation in K l values (Table 2). For the sandy soil, we can speculate the occurrence of nonspecific and specific Li sorption on soil particles. The nonspecific sorption is a quick reaction (after 2 h) that can be ascribed to Li retention on soil via a simple Coulomb force (physisorption). However, a specific Li sorption can be attributed to the time-dependent sorption that is occurred by expanding the equilibration time to 7 days. The Langmuir simulations of Li sorption isotherms for soil-I and soil-II after 1 and 7 days are shown in Fig. 1. For soil-I, the difference between the sorbed Li after 1 and 7 days is less than 10% for the case of applying an initial concentration of less than 100 mg L −1 . The individual difference ranged between 37 and 50% when the initial Li concentration was higher than 100 mg L −1 . However, the associated difference is less than 10% for the sandy clay soil through the entire initial concentration range (30 to 400 mg L −1 ). Davey and Wheeler (1980) reported that 63 to 75% of the total sorbed Li was found as neither exchangeable nor soluble fractions after 5 days equilibration using an initial concentration of less than 2 mg L −1 . The isotherms of sandy clay loam soil showed that Li did not reach the maximum sorption capacity. In contrast, the sorbed amount of sandy soil tended to reach a plateau after 7 days (Fig. 1). The results confirm the necessity of long equilibration time (7 days) to study the sorption behavior of Li on light-textured soil such as sandy soil. In contrast, an equilibration time of 1 day is appropriate to obtain Li maximum sorption capacity for the sandy clay loam soil. The obtained Li S max values indicate that both soils have a low affinity for retaining Li compared to heavy metals retention on soils (Elbana, 2022;Elbana et al., 2018).
Moreover, Freundlich parameters of Li sorption isotherms are reported in Table 2. For soil-I, the estimated Freundlich capacity coefficient (K f ) of 46.9 mg 1−b L b kg −1 and adsorption intensity (b) of 0.49 are obtained after 2 h. Whereas, after 7 days of sorption, K f of 77.5 mg 1−b L b kg −1 and b of 0.27 were obtained. However, the Freundlich parameters for soil-II reveal a sorption intensity of 0.62 and a slight variation in capacity coefficient with K f of 19.2 mg 1−b L b kg −1 and 19.4 mg 1−b L b kg −1 after 2 h and 7 days of equilibration time, respectively (Table 2). Such result indicates the nonlinear behavior of sorption isotherms for both studied soils. Yousefi et al. (2014) estimated Freundlich parameters as K f of 28.4 and 15.4 mg 1−b mL b g −1 with b values of 1.00 and 0.74 for clay loam and sandy loam soils, respectively. In addition, Akhtar et al. (2003) reported a nonlinear Li sorption behavior on silt loam and sandy loam soils using a wide range of initial Li concentrations (0 to 1200 mg L −1 ). Moreover, Anghel et al. (2002) observed nonlinear sorption of Li onto tuff (b ranged between 0.64 and 0.92), where Li sorption was correlated to the contents of smectite and zeolite. Generally, the Li adsorption Fig. 1 Lithium sorption isotherms for soil-I and soil-II after 1 day and 7 days equilibrium time; solid and dashed lines represent Langmuir simulations for 1 day and 7 days, respectively isotherm in both studied soils is markedly nonlinear and fitted to Langmuir and Freundlich with R 2 of more than 0.91.

Kinetic Behavior of Li Sorption/Desorption
Sorption/desorption simulations of Li using MRM and SOTS models are shown in Figs. 2 and 3 for soil-I and soil-II, respectively. Specifically, the change of Li concentration in the liquid phase vs. time is fitted using MRM and SOTS models. Results reveal that the sorption of Li in both soils depends upon initial Li concentration. Similar behavior was reported for the sorption of Li on manganese oxide, where the sorbed amount was found to be dependent on sorbate and sorbent concentrations (Park et al., 2015).
For the sandy soil (soil-I), fluctuations of solute concentration were observed during the sorption time and a significant increase of Li concentration in the solution was observed after 7 days (Fig. 2). The increase in solute concentration was associated with a decrease in the sorbed Li after 7 days of sorption. However, by replacing the solution with 5 mM of KCl for desorption steps, the concentration of Li significantly dropped after the first step (24 h of desorption), followed by a limited kinetic of Li release particularly after 120 h of desorption time. Specifically, the sorption/desorption results reveal that sandy soil exhibit a time-dependent sorption behavior and a limited desorption kinetic. Based on the statistical parameters of the best fitting shown in Table 3, the SOTS model provides a lower standard error of the estimated parameters and root mean square error (RMSE) than the MRM model. The obtained best SOTS simulations indicate the importance of considering reversible and irreversible sites to model Li behavior in this sandy soil. The best simulation is obtained by considering the sorption/desorption of Li as a reversible on S 1 site followed by consecutive irreversible retention on S 2 site (scenario-I, Table 3). The best-fitted kinetic parameters of 4.2371 ± 0.1195 L −1 mg −1 h −1 , 0.0027 ± 0.0003 h −1 , and 0.00013 ± 0.00002 h −1 for k s1 , k s2 , and k s3 , respectively, are estimated. The obtained parameters confirm the kinetic behavior of Li sorption onto the particles of sandy soil. Moreover, MRM and SOTS models provide accurate predictions of Li concentrations when a high Li concentration is initially applied (Ci > 100 mg L −1 ) to soil-I. However, both models underestimate Li concentration in the solution at low initial concentrations, particularly during the sorption time (Fig. 2).
For the sandy clay loam soil (soil-II), a slight variation of solute concentration was observed during the sorption phase, especially after 24 h (Fig. 3). However, a continuous gradual decrease of Li concentration was observed during desorption time. Soil-II exhibited a limited time-dependent Li sorption, whereas kinetic behavior was observed during the desorption reaction. Generally, both models, MRM and SOTS, showed good simulations of the change in Li concentration during the sorption/desorption time. Based on the statistical parameters of the best fitting shown in Table 3, the SOTS model provided a lower standard error of the estimated parameters than the MRM model. The SOTS model assumes that an intrinsic soil sorption capacity controls the sorption process and that the rate of reaction is solute concentration-dependent (Selim, 2016). Li and Liu (2020) suggested the inner-sphere complex (chemisorption) followed by lattice occupation as a dominant mechanism for Li sorption into kaolinite at alkaline pH. The modeling results indicate the minimal influence of the simultaneously irreversible Li retention on this sandy clay loam soil. Therefore, the best fit of sorption/ Fig. 3 Observed-lithium in solution vs. time during sorption/desorption experiment for the soil-II; solid lines (left side) and dashed lines (right side) represent second-order two-site (SOTS) and multi-reaction model (MRM) simulations, respectively Table 3 MRM and SOTS models parameters of fitting Li sorption/desorption data ± their standard errors * Model scenario I represents the case of Li sorption/desorption as reversible on S 1 site followed by irreversible retention on S 2 site. Model scenario II represents the case of Li sorption/desorption as reversible on S 1 site when it is simultaneously irreversible on S irr site desorption data was obtained by considering reversible sorption on S 1 site followed by irreversible retention on S 2 site (scenario-I, Table 3). The best-fitted kinetic parameters of 0.0116 ± 0.0005 L −1 mg −1 h −1 , 2.2421 ± 0.0800 h −1 , and 0.00057 ± 0.00029 h −1 for k s1 , k s2 , and k s3 , respectively, are estimated.

Hysteresis of Li Sorption/Desorption
The discrepancies of Li sorption/desorption isotherms in the studied sandy and sandy clay loam soils are shown in Fig. 4. Specifically, the observed sorbed Li after 7 days of the sorption cycle and at the end of the desorption experiment vs. Li concentration in solution are plotted. The solid and dashed curves in Fig. 4 represent SOTS simulations for sorption and desorption kinetic data, respectively, based on the best-fitting parameters in Table 3. The SOTS model matches the trend of sorption/desorption isotherm for soil-I. However, the model underestimated the Li concentration in the solution and consequently overestimated the sorbed amount, as shown in Fig. 4. On the other hand, well simulations of Li sorption/desorption data were obtained for soil-II. For the sandy soil (soil-I), at the end of the desorption experiment, soil released less than 7% of the total sorbed Li at a 7 days equilibration time. The nonsingularity of the sorption and desorption isotherms indicates irreversible retention in soil constituents (Elbana & Selim, 2019;Sander et al., 2005). It is worth mentioning that the sandy soil exhibited a significant Li release behavior after 24 h sorption cycle, especially for c i > 100 mg L −1 when the soil released more than 50% of the sorbed Li (Fig. 1). Such a decrease in the sorbed Li indicates nonequilibrium retention or a metastable state due to a weakness of sorption. Generally, the total sorbed Li at the end of desorption experiment, as a percentage of the estimated S max after 7-days sorption, (Table 2) varied between 25.7% (c i = 27.3 mg L −1 ) and 83.5% (c i = 312.0 mg L −1 ). Such nonsingularity of Li sorption/desorption isotherm indicates the incidence of an irreversible fraction of Li sorption on soil-I (Fig. 4).
The sandy clay loam (soil-II) exhibited lower sorption/desorption hysteresis than soil-I, where a higher Li was released during the desorption phase (Fig. 4). Specifically, at the end of the desorption experiment, the released Li varied between 24.2 and 60.5% of the total sorbed Li at 7 days sorption cycle. Results in Fig. 4 revealed that the higher the initial concentration, the higher hysteresis of Li sorption and release. As a result of the high S max (1268.6 mg kg −1 ) of this soil, the observed maximum sorbed Li at the end of the desorption experiment is 340.1 mg kg −1 representing only 26.8% of the estimated S max . The high desorption of Li implies substantial mobility and availability of Li in the studied agricultural soils.

Conclusion
A batch technique was applied to quantify and describe Li sorption/desorption behavior in sandy and sandy clay loam soils. Freundlich and Langmuir isotherm models were successfully fitted to sorption Fig. 4 Lithium sorption/desorption hysteresis in sandy and sandy clay loam soils; black circles represent observed sorption isotherm data after 7 days and the error bars are the standard deviations of the measurements; solid and dashed curves represent simulations of sorption and desorption data, respectively using second-order two-site model (SOTS) data with an R 2 of more than 0.91. Estimated Freundlich adsorption coefficients (K f ) were 77.5 ± 26.8 and 19.4 ± 8.4 mg 1−b L b kg −1 for the sandy and sandy clay loam soils, with reaction orders of 0.27 and 0.62, respectively. Additionally, the sandy clay loam soil exhibited the highest affinity for Li, with a S max of 1268.6 ± 392.7 mg kg −1 after 7 days of equilibration. Modeling of sorption/desorption kinetics was defined using the multi-reaction model (MRM) and the second-order two-site (SOTS) model. For both soils, the SOTS model provided the best simulation of Li sorption/desorption. Such results indicate that sorption capacity control Li retention on the studied soils and the rate of reaction is solute concentration dependent. The sorption/desorption results reveal that soils exhibit a time-and concentration-dependent kinetic behavior with noticeable hysteresis, indicating the occurrence of Li irreversibility in both studied soils. Therefore, multiplereaction models are recommended for simulating Li behavior in alkaline soils, which defines kinetic reversible and irreversible retention processes. Such a precise prediction of the change in Li concentration with time is essential for quantifying Li fate in the agricultural soil and environment.
Funding Open access funding provided by The Science, Technology & Innovation Funding Authority (STDF) in cooperation with The Egyptian Knowledge Bank (EKB).

Data Availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Conflict of Interest The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.