Introduction

Potassium (K) is essential in various plant physiological processes, including photosynthesis, enzyme activation, and osmoregulation. These processes are vital for maintaining plant health and productivity1. Effective K management in agricultural soils requires understanding its behavior post-fertilization. The wet chemical extraction method, specifically the sequential extraction process by Tessier et al.2 categorizes K and other metals into five speciation groups: exchangeable, carbonate, iron and manganese oxides, organic matter, and residual fractions. These forms interact dynamically, influenced by soil type, moisture, and other nutrients. Upon K fertilizer application, K adsorption occurs as part of the soil’s dynamic equilibrium3. During adsorption, potassium ions (K+) from the fertilizer can be retained on soil particles or fixed in clay minerals, which affects their availability to plants. K adsorption is crucial in soil-plant systems, directly impacting fertilization effectiveness and long-term K availability for plant uptake.

Soil K adsorption depends on clay mineral composition, cation exchange capacity, organic carbon, and applied K levels. Minerals like quartz, illite, and calcite shape sorption sites through electrostatic interactions, ion exchange, and surface complexation. Plant uptake creates a K gradient, promoting diffusion toward roots and triggering K desorption to replenish the soil solution4,5. This release extends to non-exchangeable forms, as roots enhance interlayer K exchange in 2:1 layer silicate clays and secrete substances that accelerate mica weathering. Increased soil K, from fertilization or other sources, can lead to K fixation within silicate interlayers, reducing layer thickness and affecting K equilibrium6. While clay minerals are key to K adsorption, the extent of flocculation alters the soil’s physical structure, which in turn can either limit or facilitate K availability depending on the specific soil conditions.

Clay flocculation involves particle aggregation into larger clusters, influenced by interactions and polymers, though mechanisms remain unclear7. Understanding flocculation requires microstructural analysis, but traditional 2D methods like optical, scanning electron microscopy (SEM), and transmission electron microscopy (TEM) have limitations, including labor-intensive preparation and misrepresentation of particle morphology. In contrast, synchrotron-based transmission X-ray microscopy (TXM) offers a nondestructive, in situ alternative, providing deeper imaging without harmful treatments. TXM has been applied in soil science to study clay mineral changes during environmental swelling and shrinking8,9.

Previous studies mainly focused on K adsorption-desorption properties4,10. However, its transformation and the interaction with clay minerals following the adsorption process remained unclear. Understanding the transformation of K speciation in soil is crucial for comprehending the dynamics of soil K speciation and its behavior within the equilibrium process. This study addresses critical knowledge gaps in K adsorption processes, speciation transformations, and clay mineral microstructural changes following K adsorption across five distinct soil types under varying management practices. It is hypothesized that the characteristics and extent of K adsorption in soils affect its speciation transformations and clay mineral aggregation. These dynamics can be effectively studied using synchrotron-based TXM. Gaining deeper insights into these processes will advance our knowledge of K retention mechanisms and their impact on soil fertility and nutrient management. This work is notable as it is the first to report on K transformation in soil following its successful adsorption process and to assess the microstructural changes in clay minerals during the swelling and shrinking phases influenced by environmental factors.

Methods

Soil sampling and pretreatment

Five distinct soil types from various Taiwanese counties were selected, each with unique properties and management practices: non-fertilized soil(W), long-term K-fertilized soil (L), alkaline soil (A), red soil (R), and forest soil (F). After harvest, nine soil samples were randomly collected from agricultural and forest zones at a depth of 0–30 cm using a soil auger. Samples from each site were combined into composite samples, which were air-dried, ground, and sieved (100-mesh) to remove debris. Further details on sample sites, soil classification, and management practices are provided Supplementary Information (Text S1 and Table S1).

Determination of physicochemical properties

The basic physicochemical properties such as soil pH, EC, CEC, soil organic matter (SOM), total N, exchangeable Ca and Mg, soil texture, and clay mineral have been reported in our previous study11. The specific surface area (SSA), pore volume (cm3 g− 1), and average pore diameter (nm) were determined using the Brunauer-Emmett-Teller (N2-BET) adsorption method.

K sorption capacity

Triplicate 50 g soil samples were shaken (100 rpm, 25 °C) with 1 L KCl solutions (0-500 mg K L− 1) for 90 min. 5mL samples were taken at an interval of 10 min and filtered. Then, K concentration was measured using an atomic absorption spectrophotometer (AAS). K sorption was calculated from the difference between initial and final concentrations, determining the amount adsorbed per unit mass of soil (q) as follows:

$$q=\left(\frac{{C}_{o}-{C}_{t}}{W}\right) \times V$$
(1)

where q is adsorbed K (mg g− 1), Co and Ct are K concentrations (mg L− 1) at initial and at time t, respectively. W is soil weight (g), and V is solution volume (L).

K adsorption isotherms and kinetic models

The adsorption isotherms were analyzed using the Langmuir (Eq. 2) and Freundlich (Eq. 3) models12.

$$\hbox{Langmuir model:}\,\frac{{C}_{e}}{{q}_{e}}=\frac{1}{{q}_{max}{K}_{L}}+\frac{{C}_{e}}{{q}_{max}}$$
(2)

Plotting \(\frac{{C}_{e}}{{q}_{e}}\) versus Ce gives a straight line with slope = \(\frac{1}{{q}_{max}}\) and intercept = \(\:\frac{1}{{q}_{max}{K}_{L}}\), where Ce is equilibrium concentration of K in solution (mg L− 1), qe is the amount of K (mg L− 1) adsorbed at equilibrium, qmax is the maximum adsorption capacity (mg/g); KL is Langmuir constant related to the affinity of the binding sites (L mg− 1)13;

$$\hbox{Freundlich model:}\,log\,{q}_{e}=log\:{K}_{F}+\frac{1}{n}\text{log}\,{C}_{e}$$
(3)

Plotting log qe versus log Ce gives a linear line with slope = \(\frac{1}{n}\) and intercept = \(log\,{K}_{F}\); where KF is the Freundlich constant indicating adsorption capacity (mg g− 1)(L mg− 1)1/n]; n is empirical constant indicating adsorption intensity (dimensionless).

Similarly, the adsorption kinetics were evaluated by applying pseudo-first-order (Eq. 4) and pseudo-second-order (Eq. 5) models.

$$\hbox{Pseudo-first order:}\,({C}_{o}-{C}_{t})=a-{k}_{1}t$$
(4)
$$\hbox{Pseudo-second order:}\,\frac{t}{{q}_{t}}=\frac{1}{{k}_{2}{{q}_{e}}^{2}}+\frac{t}{{q}_{e}}$$
(5)

where a is a constant related to the initial adsorption rate; t is time (min); qt is the amount of K (mg g− 1) sorbed at time t (min). k1 and k2 (g mg− 1 min) are the rate constants for pseudo-first order and pseudo-second order equation, respectively.

K speciation in soil

Five distinct categories of soil samples were analyzed for K speciation both before and after the process of K adsorption, following the methodology outlined by Tessier et al.2. Table S2 presents detailed information regarding the procedure, including reagent concentrations, extraction durations, and specific conditions.

Zeta potential analysis

Soil zeta potential was conducted using a Zetasizer nano 317. The zeta potential measurements were measured as a function of pH mode. A 1 g of soil sample was weighed and placed into 100 mL of an aqueous solution containing 0.01M NaClO4 (1.0% w/v). Soil particles were mixed homogeneously with a magnetic stirrer. Samples of diluted suspension (0.1% w/v) were prepared from the clay fraction and inserted into the disposable measurement cell. The pH was adjusted in the range of 3.0–9.0 by dropwise addition of 0.01 M NaOH or HCl solution before each measurement. The zeta potential of the clay particles in deionized water (DI water) and KCl, alongside 0.01 M NaClO4, was measured in millivolts (mV) as a function of the suspension pH and time with continuous stirring to ensure sample homogeneity.

Clay suspensions and flocculation structure

Clay particles were separated from silt and sand using Stokes’ law. A 1 g clay sample was suspended in 100 mL of DI water and KCl solutions (0.025 M, 0.25 M, and 1 M), allowing soil flocs to form for SEM and TXM analysis. SEM was conducted using a JSM-6700 F microscope (JEOL, Tokyo, Japan). TXM was conducted at the BL01B1 end station of Taiwan’s NSRRC. The analysis involved drop-casting a solid suspension on a steel frame with Kapton tape. The sample was examined in 2D and 3D tomography (8–11 keV energy range, 50–60 nm resolution, Zernike-phase contrast). Exposure time for 2D imaging ranged from 15 s to 4 min. A 3D dataset was reconstructed from 151 sequential images, rotating from − 75° to + 75°.

Results

Soil physicochemical properties

Table S3 presents the physicochemical characteristics of five soil samples (W, L, R, A, and F), highlighting notable differences in their chemical composition, texture, and mineralogy. The soil pH ranged from strongly acidic (4.30 in sample F) to alkaline (7.59 in sample A), indicating variable soil acidity across the sites. EC values were generally low, suggesting minimal salinity, with the highest value observed in samples A (0.294 dS m− 1) and L (0.292 dS m− 1). SOM content varied considerably, with sample F exhibiting the highest level (35.60 g kg− 1), followed by A (22.63 g kg− 1), while W and L had relatively lower SOM levels (< 10 g kg− 1). CEC was highest in F (24.18 cmolc kg− 1), reflecting a greater ability to retain nutrients, whereas W and L exhibited lower CEC values. Textural analysis classified the soils from sandy loam (W and L) to clay loam (R) and silty clay (A), with clay content ranging from 9.04% in W to 43.20% in A. These variations in soil properties indicate differing capacities for nutrient retention, water holding, and potential plant growth support among the sampled soils.

Soil electrochemistry and textural characteristics of soil samples

Figure 1 demonstrates the relationship between clay zeta potential and pH, revealing a consistent reduction in zeta potential with increasing pH. All soil samples exhibited negative zeta potential values across the pH range of 3–9, indicating no observable point of zero charge. The reduced negativity at low pH levels is attributed to the protonation of soil organic matter (SOM), which is rich in phenolic and carboxylic groups that contribute to surface charge. At higher pH levels, zeta potential became more negative due to diminished acid-base neutralization and increased hydroxyl adsorption. Beyond pH 6, forest soil displayed the most negative zeta potential, followed by non-fertilized soil, alkaline soil, long-term K-fertilized soil, and red soil, reflecting variations in SOM content and mineralogical composition. Sodium ions further reduced the negative zeta potential through K complex adsorption onto clay surfaces, emphasizing their influence on soil electrochemistry14,15.

Fig. 1
figure 1

Zeta potential of soil samples (1% w/v) in 0.01 M NaClO4. W = non-fertilized soil; L = long-term K-fertilized soil; A = alkaline soil; R = red soil, and F = forest soil.

Full size image

Table 1 summarizes the textural properties of the soils. The average pore diameters ranged from 17.39 to 23.83 nm, while pore volumes varied from 0.13 to 0.26 cm3 g− 1, categorizing these soils as microporous (5–30 nm). Fine-textured soils, particularly those rich in organic matter and clay, demonstrated higher surface areas compared to coarse soils. Forest and alkaline soils exhibited smaller pores and larger surface areas, which enhanced their capacity for K retention and water-holding potential. Adsorption followed an L-shaped curve, consistent with the Langmuir isotherm, indicating monolayer adsorption. Regression analysis identified soil EC, available phosphorus (Avai. P), exchangeable calcium (Exch. Ca), and CEC as critical factors influencing K adsorption rates, as outlined in Eqs. 6 and 7:

$${\text{K}}_{{\text{L}}} = - 55.47 - 803.70\,{\text{EC}} + 3.52\,{\text{Avai. P}} + 0.06\,{\text{Exch. Ca}} + 2.61\,{\text{CEC}}\,({\text{r}}^{2} = 0.93)$$
(6)
$${\text{K}}_{{\text{F}}} = - 0.03 - 0.79\,{\text{EC}} + 0.004\,{\text{Avai. P}} + 0.01\,{\text{CEC}}\,({\text{r}}^{2} = 0.80)$$
(7)

where KL and KF are K adsorption rates for Langmuir and Freundlich isotherm, respectively.

Table 1 Textural properties of the soil samples.
Full size table

K adsorption capacity

Distinct K adsorption capacities were observed among the soils, reflecting their physicochemical differences. Immersion in K solutions revealed that over 50% of the added K remained in the aqueous phase, indicating variability in adsorption efficiency. As shown in Fig. 2a, alkaline soil exhibited high adsorption at lower concentrations (25 mg K L− 1), likely due to its elevated SOM and clay content, which provide abundant adsorption sites. At concentrations above 50 mg K L− 1 (Fig. 2b–e), forest soil demonstrated the highest adsorption capacity (6.33–49.87%), followed by alkaline soil, long-term K-fertilized soil, non-fertilized soil, and red soil. Forest soil’s superior performance is attributed to its high CEC (24.18 cmolc kg− 1) and SOM content (35.60 g kg− 1), which promote sustained nutrient retention.

At lower solution concentrations (25–100 mg K L− 1, Fig. 2a–c), the soils effectively adsorbed K due to the availability of vacant adsorption sites on the soil particles to bind with K ions, leading to a greater uptake. However, at higher concentrations (250 and 500 mg K L− 1, Fig. 2d, e), competition among K ions for limited binding sites reduced adsorption capacity, illustrating saturation effects16. Red soil retained the highest proportion of K in the aqueous phase (81.72–97.92%), reflecting its lower adsorption efficiency, likely due to reduced SOM content and limited active sites.

Fig. 2
figure 2

The percentage of K translocation in different soils that soaked in solution containing 25 mg K L− 1 (a), 50 mg K L− 1 (b), 100 mg K L− 1 (c), 250 mg K L− 1 (d), and 500 mg K L− 1 (e). W = non-fertilized soil; L = long-term K-fertilized soil; A = alkaline soil; R = red soil, and F = forest soil.

Full size image

Table 2 presents the sorption parameters of each soil sample, derived from the K adsorption isotherms shown in Fig. S1. The isotherm data fit the Langmuir model more closely than the Freundlich model, as indicated by the consistently higher r2 values (0.99 vs. 0.70–0.89). This demonstrated that the mechanism of K adsorption onto soil is more consistent with the Langmuir model than that of the Freundlich model, suggesting a monolayer adsorption mechanism. Among the soil samples, forest soil shows the highest maximum adsorption capacity (qmax) and the strongest affinity for K, indicating a greater number of available binding sites. In contrast, red soil showed the lowest qmax and KL, reflecting its comparatively limited adsorption capacity and lowest affinity for K binding under the Langmuir assumptions.

Table 2 Sorption isotherms describing the adsorption behavior of K onto the soil samples.
Full size table

K speciation adsorption transformation

Figure 3 illustrates the percentage of K speciation adsorption transformation in different soils. Increasing K concentration led to reduced adsorption capacity and lower transformation of K speciation. Mechanisms of K sorption included physical entrapment in soil interlayers, electrostatic attraction to negatively charged surfaces, ion exchange, binding to SOM functional groups, and precipitation with carbonate or phosphate. Exchangeable K, in particular, increased significantly in soils with higher SOM and clay content. For instance, exchangeable K percentages rose notably in non-fertilized soil (2.00–25.46%), long-term K-fertilized soil (2.69–35.89%), alkaline soil (4.59–53.37%), red soil (1.57–12.53%), and forest soil (5.72–33.50%) (Fig. 3a–e).

Fig. 3
figure 3

The percentage of K speciation adsorption transformation in non-fertilized soil (a), long-term K-fertilized soil (b), alkaline soil (c), red soil (d), and forest soil (e). Se = the amount of K speciation in soil after adsorption.

Full size image

Soil flocculation microstructure

SEM (Fig. 4a–d) and TXM (Fig. 4e–m) images revealed plate-like structures in soil suspensions, which aggregated into cellular networks in the presence of K solutions. Long-term K-fertilized soil exhibited the most pronounced flocculation, followed by alkaline soil, non-fertilized soil, red soil, and forest soil. Particle flocculation diameters presented in Table 3 increased with higher K concentrations (0.025–1 M KCl), suggesting that K ions play a pivotal role in altering soil aggregate structures. TXM analysis offered detailed insights, revealing that the void diameters within soil aggregates varied with K+ ion concentration. This observation aligns with known effects of ionic strength on soil structure and corresponding changes in hydraulic conductivity17,18.

Fig. 4
figure 4

SEM (a–d) and 2D TXM (e–h) images of 1% w/v soil flocculation in DI water, 0.025 M, 0.25 M, and 1 M KCl solutions. 3D TXM images of 1% w/v clay particle flocculation in 1 M KCl for non-fertilized soil (i), long-term K-fertilized soil (j), alkaline soil (k), red soil (l), and forest (m) at different angle. White area in 2D TXM images represent clay particle and black area belong to voids area.

Full size image
Table 3 Particle flocculation diameter (nm) after suspension with various solutions.
Full size table

Discussion

The findings underscore fundamental mechanisms governing soil nutrient adsorption and transformation, providing critical insights into sustainable soil management strategies. SOM, pH, clay content, and types of clay minerals were identified as key factors contributing to the development of negative surface charge in the soils, consequently influencing zeta potential, K adsorption capacity, and soil structural stability (Table S3). Forest and alkaline soils, characterized by higher SOM and CEC, demonstrated superior nutrient retention and transformation capacities compared to other soil types. This emphasizes the need for soil management practices tailored to specific soil properties to optimize nutrient availability and retention. Soil electrochemistry properties are congruent with earlier findings19,20 that highlight the predominance of monolayer adsorption and chemisorption in nutrient retention processes. The more negative zeta potential at higher pH levels exemplifies the role of SOM and hydroxyl adsorption in improving soil adsorption properties. Additionally, the regression analyses validated the critical roles of soil EC, available P, and CEC in modulating adsorption efficiency, in agreement with Yudina et al.21. These attributes collectively reduce nutrient leaching and enhance soil resilience, particularly under intensive agricultural systems where nutrient losses are prevalent. Regarding the roles of organic matter in K adsorption, the marked increase in exchangeable K percentages in soils with elevated SOM content (Section “K speciation adsorption transformation”) underscores the transformative role of SOM in nutrient availability. This highlights the importance of maintaining and enriching SOM through organic amendments, as also suggested by Kabala and Jedrzejewski22. The notable flocculation observed in K-fertilized soils (Section “Soil flocculation microstructure”) further underscores the influence of K ions on soil aggregation and structural cohesion. This structural improvement, as supported by Geoffroy et al.23 enhances water infiltration and retention, which are critical for maintaining soil health under varying climatic conditions.

Moreover, studies by Guo et al.24 and Qi et al.25 elucidate mechanisms underlying K retention, including electrostatic interactions, ion exchange, and surface complexation. These mechanisms are reinforced by SOM functional groups, particularly carboxyl and hydroxyl groups, which enhance the affinity of soils for K ions. This highlights the dual role of SOM in not only nutrient retention but also in fostering a microenvironment conducive to nutrient cycling. Additional support for this was found in the work of Paola et al.26 who demonstrated that fixed K ions, often stabilized by clay minerals, significantly reduce leaching losses, ensuring long-term availability for plants. Similarly, Ewis et al.27 emphasized that clay-SOM complexes are essential for buffering nutrient availability, particularly under fluctuating environmental conditions. The significance of these interactions extends to their implications for managing soils in arid and semi-arid regions, where nutrient losses are typically exacerbated by low organic content and high mineralization rates. Additionally, Yang et al.28 highlighted the role of soil mineralogy in enhancing K adsorption, particularly in soils dominated by kaolinite and illite minerals, which exhibit high adsorption efficiency under acidic to neutral pH conditions. Furthermore, research by Dor et al.29 revealed that clay particles’ pseudo-hexagonal structure significantly influences their ability to retain nutrients, making these insights particularly relevant for designing soil amendments targeting specific mineralogical compositions.

Future research should prioritize exploring the long-term consequences of K adsorption and speciation dynamics on soil fertility and plant productivity. Understanding how K retention interacts with microbial activity, particularly in soils with varying SOM content, could offer deeper insights into the nutrient cycling process. Investigating the interplay between SOM enrichment and clay mineral interactions may yield actionable strategies for optimizing soil amendment practices. Furthermore, leveraging advanced imaging technologies such as TXM, integrated with field studies, will provide a nuanced understanding of nutrient retention and release dynamics. These insights could drive the development of site-specific soil management frameworks, fostering innovation in sustainable agricultural practices globally.

Conclusions

This study demonstrates that K adsorption in soils aligns well with the Langmuir isotherm model, indicating a predominant monolayer adsorption mechanism. Adsorption capacity declined at higher K concentrations, likely due to sorption site saturation. Soils with higher SOM and clay content facilitated greater K transformation into exchangeable forms. Additionally, K adsorption was found to influence soil flocculation behavior, with long-term K fertilization promoting greater aggregation and increased particle size. These structural alterations, visualized via SEM and synchrotron-based TXM, may affect soil porosity and hydraulic conductivity. Nonetheless, it is essential to note that these findings are based on controlled laboratory conditions and five representative soil types, which may not fully capture the complexity of field-scale agroecosystems. Future studies should incorporate field experiments, long-term monitoring, and biological components such as plant uptake and microbial activity to validate the observed trends further and support the development of effective, sustainable K fertilization strategies.