Characterization of bentonites

XRD analysis

XRD analysis of the B-Clay, Y-Clay, G-Clay, and P-Clay (Figs S1–S4, see Supplementary material) and the results of Rietveld refinements are summarized in Table 1. All samples consisted mainly of dioctahedral smectite (montmorillonite), based on the positions of diagnostic diffraction peaks. The largest montmorillonite content was observed for the Y-Clay (98 wt.%). Slightly smaller montmorillonite contents were observed for the G-Clay (88 wt.%) and the B-Clay (83 wt.%), and the P-Clay contained much less montmorillonite (63 wt.%).

Table 1.

Quantitative mineralogical composition of bentonites

Accessory minerals included kaolinite (G-Clay), quartz (Y-Clay, B-Clay, P-Clay, and G-Clay), plagioclase (Y-Clay, B-Clay, and P-Clay), cristobalite (B-Clay and P-Clay), calcite (P-Clay and G-Clay), mordenite (G-Clay), and mica (B-Clay). The montmorillonite 001 reflection (d ₀₀₁) was 15.4 Å for the Y-Clay, 15.3 Å for the B-Clay and 15.6 Å for the G-Clay corresponding to a two H₂O-layer hydration state of alkaline-earth-smectite consistent with Ca²⁺ as the dominant interlayer cation (Chipera et al., Reference Chipera, Carey, Bish, Gilfrich, Noyan, Jenkins, Huang, Snyder, Smith, Zaitz and Predecki1997; Ferrage et al., Reference Ferrage, Tournassat, Rinnert and Lanson2005; Goo et al., Reference Goo, Kim, Kwon and Jo2023). The d ₀₀₁ value of the P-Clay was ~12.5 Å and is characteristic of Na-montmorillonite hydrated with one H₂O layer (Chipera et al., Reference Chipera, Carey, Bish, Gilfrich, Noyan, Jenkins, Huang, Snyder, Smith, Zaitz and Predecki1997; Ferrage et al., Reference Ferrage, Tournassat, Rinnert and Lanson2005; Goo et al., Reference Goo, Kim, Kwon and Jo2023).

Chemical composition

The results of the chemical analysis of the bentonites (Table 2) indicated large amounts of SiO₂ (49.24–68.00%) and Al₂O₃ (15.59–25.04%) consistent with the presence of major montmorillonite in all samples. CaO varied from 3.09% to 5.17% in the Ca-bentonites (B-Clay, Y-Clay, and G-Clay). CaO in the G-Clay (5.07%) and in the P-Clay (4.39%) originated from montmorillonite but also from calcite that was present as an accessory mineral. MgO varied from 1.31% to 3.19%, consistent with the largely dioctahedral montmorillonite in these samples (Bayram et al., Reference Bayram, Önal, Yılmaz and Sarıkaya2010; Lacerda et al., Reference Lacerda, Lacerda, Andrade, Fujiwara and Kloss2023; Shwan et al., Reference Shwan, Qadr, Kaufhold and Ufer2023). The amounts of Na₂O and K₂O in the B-Clay, Y-Clay, and G-Clay were 0.04–0.28 and 0.18–0.68%, respectively, whereas the P-Clay contained 1.24% of Na₂O and 1.39% of K₂O. The amount of Na₂O in Na-montmorillonites generally represented the amount of Na as an exchangeable cation (Lacerda et al., Reference Lacerda, Lacerda, Andrade, Fujiwara and Kloss2023), and the calculated Na content in P-Clay of 0.92% (40 meq 100 g^–1) was consistent with the amount of Na determined in the exchangeable positions – 39.0 meq 100 g^–1 (see ‘CEC and zeta potential’ section below). The loss on ignition (LOI) value originated from H₂O in the interlamellar space and the dehydroxylation of montmorillonite as well as from the de-carbonation of calcite in some samples (P-Clay and G-Clay).

Table 2.

Chemical compositions of the bentonites

* Shannon et al. (Reference Shannon, Ledoux, Rottinghaus, Shaw, Daković and Marković2017).

CEC and zeta potential

The amounts of the exchangeable cations, the total CEC values, and the zeta potential values for bentonites are presented in Table 3.

Table 3.

Amounts of exchangeable cations, CEC, and zeta (ξ) potential values for bentonites

The total CEC values of bentonites varied from 97 meq 100 g^–1 (G-Clay) to 120 meq 100 g^–1 (B-Clay). The dominant cation in exchangeable positions in all three natural bentonites was Ca²⁺, with lesser amounts of Mg²⁺ (8–26 meq 100 g^–1) and very small amounts of Na⁺ and K⁺. The P-Clay contains mainly Na⁺ (39 meq 100 g^–1) and Ca²⁺ (55 meq 100 g^–1) ions in the exchangeable cation positions demonstrating that the P-Clay was a Na/Ca-montmorillonite.

Various data on the electrokinetic properties of montmorillonites have been reported in the literature (Şans et al., Reference Şans, Güven, Esenli and Çelik2017; Missana et al., Reference Missana, Alonso, Fernández and García-Gutiérrez2018), but generally, the zeta potential of Na-montmorillonites was more negative than the zeta potential of Ca-montmorillonites. For the Ca-montmorillonites studied, the zeta potential varied from −21.1 mV for the B-Clay to −15.9 mV for the G-Clay. As expected, the smallest zeta potential value (−29.0 mV) was observed for the P-Clay, which was a Na/Ca-montmorillonite. These results for Ca-montmorillonites were consistent with the results of Şans et al. (Reference Şans, Güven, Esenli and Çelik2017). They studied nine different Ca-bentonites with relatively high smectite content in order to relate their zeta potential values to particle size, pH, CEC, smectite structural formula, and layer charge. They reported that negative values of the zeta potential of Ca-bentonites, with smectite contents of between 60 and 95%, varied from −9.5 to −17.4 mV in water at natural pH values. They showed that a negative zeta potential value decreased with increase in CEC for their studied bentonites, although these parameters were not highly correlated. Data for Ca-montmorillonites used in this study revealed an obvious correlation and negative values of zeta potential decreased and became more negative as the CEC values of bentonites increased (B-Clay>Y-Clay>G-Clay).

Thermal analysis

Most clay minerals are thermally characterized by three reactions: (1) processes of dehydration (25–300°C); (2) dehydroxylation (300–700°C); and (3) decomposition/recrystallization (>800°C) (Wolters and Emmerich, Reference Wolters and Emmerich2007; Rehman et al., Reference Rehman, Manan, Uzair, Khan, Ullah, Ahmad, Wazir, Qazi and Khan2021). The detailed octahedral structure of smectites cannot be determined by XRD analysis, as is possible for more ordered dioctahedral 2:1 phyllosilicates (e.g. illites) (Wolters and Emmerich, Reference Wolters and Emmerich2007). According to Drits et al. (Reference Drits, Besson and Muller1995), the dehydroxylation behavior of dioctahedral smectites is related to the distribution of cations and vacancies in the octahedral sheet. Depending on the disposition of the OH groups with respect to the vacant octahedral site, two types of configurations in the octahedral sheet of dioctahedral smectites exist: cis-vacant and trans-vacant. In the cis-vacant octahedra, OH groups are located on the same side, whereas in the trans-vacant octahedra, they are positioned on the opposite corners of the octahedral site (Drits and Zviagina, Reference Drits and Zviagina2009; Escamilla-Roa et al., Reference Escamilla-Roa, Hernández-Laguna and Sainz-Díaz2014). Thus, an essential characteristic of dioctahedral smectites is the nature of the octahedral vacancy. In general, the dehydroxylation peak temperature obtained by thermal analysis (DTA or differential scanning calorimetry – DSC) can be used for characterization of the structure of the octahedral sheet. cis-vacant smectites dehydroxylate at 650–700°C, whereas trans-vacant smectites dehydroxylate at 500–550°C. Most montmorillonites exhibit one endothermic dehydroxylation peak near 700°C, and only a few montmorillonites show one endothermic peak near 500°C (Grim and Kulbicki, Reference Grim and Kulbicki1961; Schultz, Reference Schultz1969), whereas other montmorillonites show an endothermic peak doublet between ~500°C and ~700°C (Drits et al., Reference Drits, Besson and Muller1995; Emmerich and Kahr, Reference Emmerich and Kahr2001; Wolters and Emmerich, Reference Wolters and Emmerich2007).

The DTA/TG curves of the bentonites investigated are presented in Fig. 2, and mass losses in different temperature regions are presented in Table 4. The DTA curves of the B-Clay, Y-Clay, and G-Clay showed dehydration in the first region, giving rise to an endothermic peak centered at 157°C (B-Clay), 166°C (Y-Clay), and 158°C (G-Clay) with a weak shoulder at 221°C (B-Clay), 230°C (Y-Clay), and 228°C (G-Clay) (Fig. 2a). This endothermic peak with a shoulder was characteristic of Ca-bentonites and corresponds to the evolution of adsorbed H₂O and H₂O coordinated to the interlayer cations. In this region, the TG curves of these bentonites (Fig. 2b) exhibited one dehydration step with a total mass loss of 14.90% (B-Clay), 18.85% (Y-Clay), and 15.00% (G-Clay). For the P-Clay, a single endothermic dehydration peak at 143°C was observed and is characteristic of Na-bentonites (Fig. 2a), with a mass loss from the TG curve of 12.86%. In the dehydroxylation temperature range (250–750°C), DTA curves of B-Clay and P-Clay showed one endothermic peak at 679°C and 700°C, respectively, confirming the predominance of cis-vacant octahedra. The endothermic peak for the G-Clay was at 554°C and is characteristic of a trans-vacant smectite. The DTA curve for the Y-Clay showed two endotherms at 527°C and 678°C, suggesting that this bentonite possesses both cis- and trans-vacant octahedra. Drits et al. (Reference Drits, Besson and Muller1995), Şans et al. (Reference Şans, Güven, Esenli and Çelik2017), and Zhao et al. (Reference Zhao, Wang, Hong, Algeo, Yin, Ji, Song, Abels and Christidis2021) reported that the lower dehydroxylation temperature of trans-vacant smectites is caused by the weaker bonds related to the larger number of tetrahedral Al³⁺ cations in the smectite structure. Thermal results in this study were consistent with the chemical analyses (see ‘Chemical composition’ section below) that confirmed larger amounts of Al₂O₃ in the trans-vacant smectite (G-Clay) and cis-/trans-vacant smectite (Y-Clay). The mass loss in the region 250–750°C related to dehydroxylation was 3.81% for the B-Clay and 3.51% for the P-Clay, whereas larger mass losses of 5.31% and 6.25% were observed for the Y-Clay and G-Clay, respectively. The third mass loss for the G-Clay (~1.35%) at temperatures above 750°C, with an endothermic peak at 802°C in the DTA curve, was due to the de-carbonation reaction of calcite present in bentonite (4%, XRD analysis). The P-Clay also contained 1% calcite (XRD analysis), but the endotherm originating from de-carbonation was not apparent in the DTA curve due to overlap with the dehydroxylation peak, which was slightly shifted towards higher temperatures (700°C). Decomposition and recrystallization of the bentonites occurred above 850°C (Wolters and Emmerich, Reference Wolters and Emmerich2007), and occasionally only one of the two reactions was observed in DTA curves. Endotherms were visible at 882°C and 920°C for the B-Clay and P-Clay, respectively, and the G-Clay showed an exotherm at 919°C. In contrast, the Y-Clay showed two peaks: an endotherm at 888°C and an exotherm at 947°C.

Figure 2.

Thermal curves of bentonites: (a) DTA and (b) TG.

Table 4.

Mass loss for bentonites over various temperature regions

FT-IR analysis

Additional structural information about the bentonites was obtained from FT-IR spectra (Fig. 3). The spectral absorption band at 3628 cm^–1 (Fig. 3a) for all bentonites was attributed to the stretching vibrations of structural OH groups related to various montmorillonite octahedral cations. The broad band at 3434 cm^–1 originated from OH-stretching vibrations of H₂O in the interlamellar space, together with the bending vibration of H₂O at 1638 cm^–1. An intense Si-O absorption band at 1039 cm^–1 was related to the Si–O stretching vibrations in the tetrahedral sheet of montmorillonite, and bands around 525, 469, and 425 cm^–1 were due to the Al-O-Si, Si-O-Mg, and Si-O-Si bending vibrations, respectively (Zhao et al., Reference Zhao, Wang, Hong, Algeo, Yin, Ji, Song, Abels and Christidis2021). An absorption band corresponding to the Al-Al-OH bending vibrations was observed around 915 cm^–1 (Fig. 3b) (Danková et al., Reference Danková, Mockovčiaková and Dolinská2014; Zhao et al., Reference Zhao, Wang, Hong, Algeo, Yin, Ji, Song, Abels and Christidis2021; Et-Tayea et al., Reference Et-Tayea, Harrati, Rachid, Nasri, Attou, Arkame, Manni, Mamouch, El Bouari, El Khazanti and Sadik2023; Lacerda et al., Reference Lacerda, Lacerda, Andrade, Fujiwara and Kloss2023). An absorption band visible for the B-Clay and Y-Clay at 842 cm^–1 was related to the Al-Mg-OH bending vibrations consistent with the isomorphic substitution of Al by Mg in the octahedral sheet that is characteristic of some dioctahedral smectites (Lacerda et al., Reference Lacerda, Lacerda, Andrade, Fujiwara and Kloss2023). This band was not visible in spectra of the P-Clay and G-Clay probably because it was overlapped with the band at 875 cm^–1 originating from calcite. The presence of calcite in these bentonites was also confirmed by the presence of a band at 1442 cm^–1 (Fig. 3a) (Rehman et al., Reference Rehman, Manan, Uzair, Khan, Ullah, Ahmad, Wazir, Qazi and Khan2021; Shwan et al., Reference Shwan, Qadr, Kaufhold and Ufer2023). The high-intensity band at 795 cm^–1 visible for the B-Clay and P-Clay was due to the Si-O stretching vibration, confirming the presence of cristobalite (Qiao et al., Reference Qiao, Liu, Zhang and Wu2019), and the low-intensity band around 796 cm^–1 for the G-Clay and Y-Clay confirmed the presence of quartz (Et-Tayea et al., Reference Et-Tayea, Harrati, Rachid, Nasri, Attou, Arkame, Manni, Mamouch, El Bouari, El Khazanti and Sadik2023). A band visible at 695 cm^–1 for the B-Clay, Y-Clay, and G-Clay was ascribed to the quartz impurity. A weak shoulder at ~3690 cm^–1 visible in FT-IR spectra of the G-Clay and Y-Clay originated from the presence of traces of kaolinite (Belousov et al., Reference Belousov, Chupalenkov, Christidis, Zakusina, Zakusin, Morozov, Chernov, Zaitseva, Tyupina and Krupskaya2021; Zhao et al., Reference Zhao, Wang, Hong, Algeo, Yin, Ji, Song, Abels and Christidis2021).

Figure 3.

FT-IR spectra of bentonites: (a) the 4000–400 cm^–1 region and (b) the 1000–400 cm^–1 region.

The bentonites (especially G-Clay and Y-Clay) contained Fe₂O₃, but bands related to Fe-Al-OH and Fe-Fe-OH vibrations (881 and 818 cm^–1) were absent from their FT-IR spectra, with the exception of the B-Clay where a small shoulder was visible at 881 cm^–1. The band at ~624 cm^–1 in the FT-IR spectra of the B-Clay, Y-Clay, and P-Clay was assigned to coupled Al-O and Si-O out-of-plane vibrations. The position of this band may be related to the structure of the octahedral sheet, specifically to the type of octahedral vacancy, and can be correlated to the dehydroxylation peak temperature on DTA curves. Belousov et al. (Reference Belousov, Chupalenkov, Christidis, Zakusina, Zakusin, Morozov, Chernov, Zaitseva, Tyupina and Krupskaya2021) reported FT-IR and DTA analysis of an industrial bentonite IB-6. They observed a FT-IR band near 622 cm^–1, which they attributed to the coupled Al-O and Si-O out-of-plane vibrations. The DTA curve of IB-6 showed an endothermic dehydroxylation peak at 667°C, indicating the predominance of cis-vacant octahedra. In another study by Musso et al. (Reference Musso, Pettinari, Pozo, Roca Jalil and González Gil-García2022), FT-IR spectra of selected (red and gray) bentonites showed a band near 620 cm^–1 due to coupled Al-O out-of-plane and Si-O vibrations, characteristic of montmorillonite. Their DTA curves showed endothermic dehydroxylation peaks at 620°C with a mass loss up to 4.0% in the range 200–720°C, indicating that these smectites were more similar to cis-vacant variants. Overall, the FT-IR results reported in this study were in accordance with the results of the XRD analysis (see ‘XRD analysis’ section below).

Textural properties

The N₂ adsorption-desorption isotherms of B-Clay, Y-Clay, G-Clay, and P-Clay (Figs S5 and S6; see Supplementary material) had a common shape similar to data in numerous publications (Kaufhold et al., Reference Kaufhold, Plötze, Klinkenberg and Dohrmann2013; Jabłońska et al., Reference Jabłońska, Busch, Kityk and Huber2019; Mannu et al., Reference Mannu, Vlachopoulou, Sireus, Mulas and Petretto2019). The adsorption portion of the isotherms consisted of a vertical, clearly visible, increase in the volume of adsorbed N₂ in the region p/p ₀≤0.05, with a transition to a moderate, almost-linear increase lasting up to p/p ₀=0.7, followed by a sharp increase in the adsorbed volume of N₂ in the region of high p/p ₀.

The first portion of the isotherm corresponded to adsorption of N₂ in the microporous region with simultaneous adsorption on the outer surfaces of particles, whereas the further slow growth was exclusively related to adsorption of N₂ in the monolayer. By increasing the relative pressure, N₂ was packed uniformly in the second adsorption layer, and with a further increase in p/p ₀, adsorption occurred in the following layers. Depending on the textural properties of the material, two resulting isotherm shapes were possible with further increase in p/p ₀. For mesoporous materials, capillary condensation occurred as a result of the complete filling of the available mesopore space. In the second case, for non-porous and macroporous materials, the conditions for multi-layer adsorption were achieved at high p/p ₀ values, and the delayed capillary condensation occurred on the surface of the particles. For mesoporous materials (type IV isotherm according to IUPAC nomenclature), the completion of mesopore filling was recognizable by the appearance of a plateau on the adsorption isotherm, as well as on its corresponding desorption branch, in the region of high relative pressures. The plateau on the isotherms is not visible in the majority of literature data for untreated bentonites (Sing and Williams, Reference Sing and Williams2004). In fact, a sharp increase in the amount of adsorbed N₂ is an indicator of N₂ condensation in the interparticle space, on the surface of particles, and/or condensation in macropores. Despite the existence of a hysteresis loop (clearly visible in most published works), which is also a common feature of the isotherms for the bentonites studied here, the isotherms can be classified as type II according to the IUPAC nomenclature (Thommes et al., Reference Thommes, Kaneko, Neimark, Olivier, Rodriguez-Reinoso, Rouquerol and Sing2015) or IIb according to Rouquerol et al. (Reference Rouquerol, Rouquerol and Sing1999). The isotherms of B-Clay and P-Clay were similar (IIb type) (Fig. S5; see Supplementary material), and resultant textural parameters (Table 5) were similar (especially values of V _mic, V _mes, and V _tot). The IIb isotherm form was attributed to adsorption on aggregates of plate-like particles which, therefore, possessed non-rigid groove- and/or slit-shaped pores (Rouquerol et al., Reference Rouquerol, Rouquerol and Sing1999).

Table 5.

Textural properties of bentonites

The SSA_N2BET of the studied samples decreased in the following order: Y-Clay>G-Clay>B-Clay>P-Clay, and the SSA_N2BET of the P-Clay was only slightly lower than that of the B-Clay (57 m² g^–1 vs 63 m² g^–1). As Na-bentonites usually show SSA_N2BET values ≤40 m² g^–1 (Kaufhold et al., Reference Kaufhold, Dohrmann, Klinkenberg, Siegesmund and Ufer2010), the P-Clay appears more similar to a Na/Ca-montmorillonite. The largest SSA_N2BET value was obtained for the Y-Clay (105 m² g^–1), and the SSA_N2BET of the G-Clay was slightly lower (88 m² g^–1). Such high values of SSA_N2BET were not unexpected for Ca-bentonites with a large montmorillonite content, as Ca²⁺/Mg²⁺ bentonites (<5% Na⁺) in particular show a broad distribution of SSA_N2BET values in the range 50–130 m² g^–1 (Kaufhold et al., Reference Kaufhold, Dohrmann, Klinkenberg, Siegesmund and Ufer2010; Kaufhold et al., Reference Kaufhold, Plötze, Klinkenberg and Dohrmann2013). Reported SSA_N2BET values were consistent with the results of chemical analyses, CEC, and zeta potential (see ‘Chemical composition’ and ‘CEC and zeta potential’ sections below). However, it is important to note that the shape of the isotherms and these SSA _N2BET values were not only a consequence of the content of montmorillonite, but also of the overall mineralogical and chemical composition and particle size of the bulk clay and even of the pre-treatment preparation procedures used before measurement (Noyan et al., Reference Noyan, Önal and Sarikaya2006). Therefore, the almost uniform mineralogical composition (98% montmorillonite) of the Y-Clay was an exception, evidenced by the unique hysteresis loop almost parallel to the p/p ₀ axis, and significantly wider than the loops of the remaining three samples, closely resembling the H4 type that occurs in slit-shaped micropore materials.

Adsorption of aflatoxin B1 by bentonites

Adsorption of AFB1 (%) versus amount of adsorbent, from 0.002 to 0.02% (w/v) for B-Clay, Y-Clay, G-Clay, and P-Clay, at pH 3 and 7, is presented in Table 6. Experiments performed according to the Commission Implementing Regulation (EU) showed that at pH 5 all studied bentonites were highly efficient at adsorbing AFB1 (adsorption index >90%) (Table 6).

Table 6.

Adsorption of AFB1 (C₀=4 ppm) by bentonites at pH 3, 5, and 7

* Shannon et al. (Reference Shannon, Ledoux, Rottinghaus, Shaw, Daković and Marković2017).

Data measured at pH 3 and 7 (Table 6) illustrated that the percentage of AFB1 adsorption by bentonites increased with an increasing amount of each adsorbent in suspension. At the largest amount (0.02% w/v), adsorption of AFB1 by all bentonites was very high (>90%) with no differences in their adsorption at pH 3 and 7. With a decrease in the amount of bentonite, the differences in AFB1 adsorption became more apparent. For the smallest amount (0.002% w/v), for the Y-Clay and G-Clay, adsorption of AFB1 was slightly higher at pH 3 (50.1% for Y-Clay and 45.6% for G-Clay) than at pH 7 (42.4% for Y-Clay and 30.9% for G-Clay). For the other two clays, AFB1 adsorption was practically independent of pH, but was much higher for the B-Clay (Ca-dominant, cis-vacant montmorillonite), 52.2% at pH 3 and 50.0% at pH 7, than for the P-Clay (Na/Ca form, cis-vacant montmorillonite), 31.1% at pH 3 and 28.9% at pH 7.

Adsorption of AFB1 by bentonites was further investigated through the evaluation of adsorption isotherms, measured in phosphate buffer, at pH 3 and 7, and results are presented in Fig. 4. Generally, the isotherms were obtained by plotting the amounts of solute adsorbed per unit weight of adsorbent (mg g^–1) against the equilibrium concentrations of the solute in the solution (mg L^–1). The results showed that at both pH values, non-linear adsorption isotherms were obtained for all bentonites, and adsorption of AFB1 increased as the initial toxin concentration increased, with the greater adsorption observed at pH 3. Non-linear adsorption isotherms suggested a complex AFB1 adsorption mechanism. The experimental data were fitted by common non-linear Langmuir (Eqn (1)) and Freundlich (Eqn (2)) models for the adsorption of AFB1 on all bentonites:

(1) $$ {q}_{\mathrm{e}}=\frac{q_{\mathrm{m}}{\mathrm{K}}_{\mathrm{L}}{c}_{\mathrm{e}}}{1+{\mathrm{K}}_{\mathrm{L}}{c}_{\mathrm{e}}} $$

(2) $$ {q}_{\mathrm{e}}={\mathrm{K}}_{\mathrm{F}}{c_{\mathrm{e}}}^{\frac{1}{n}} $$

where c _e (mg L^–1) is the concentration of AFB1 in solution at equilibrium, q _e (mg g^–1) is the amount of AFB1 adsorbed by each bentonite at equilibrium, q _m (mg g^–1) is the maximum amount of AFB1 adsorbed by the bentonites, K_L $ \Big( $ L mg^–1) and $ {\mathrm{K}}_{\mathrm{F}} $ ( $ {\mathrm{L}}^n\;{\mathrm{mg}}^{1-n}{\mathrm{g}}^{-1} $ ) are the Langmuir and Freundlich constants, respectively, and 1/n is a heterogeneity factor (dimensionless). The Langmuir theoretical isotherm model predicts monolayer adsorption at specific active sites on a homogeneous adsorbent surface, meaning that no further adsorption occurs once an adsorbate is adsorbed at these sites in the adsorbent. The Freundlich model usually considers a multi-layer adsorption process on the heterogeneous surface or surface supporting sites of various affinities. This means that the adsorption process is not restricted to the formation of a single adsorbate layer on the adsorbent surface (Wang et al., Reference Wang, Miao, Sun and Zheng2018). Parameters of Langmuir and Freundlich isotherms are listed in Table 7.

Figure 4.

Adsorption isotherms for AFB1 adsorption by bentonites: Langmuir model (continuous line) and Freundlich model (dashed line).

Table 7.

Characteristic parameters of AFB1 adsorption isotherms

The R ² (coefficient of determination) shown in Table 7 suggested that the isotherms are empirically consistent with both the Langmuir and Freundlich models, with slightly larger values observed for the Langmuir model. The calculated maximum adsorption capacities (q _m) for AFB1 derived from Langmuir plots of the isotherms were from 0.24 mol kg^–1 (75.43 mg g^–1) at pH 3 and 0.20 mol kg^–1 (62.91 mg g^–1) at pH 7 for the P-Clay to 0.32 mol kg^–1 (100.25 mg g^–1) at pH 3 and 0.25 mol kg^–1 (78.58 mg g^–1) at pH 7 for the B-Clay (Table 7).

The value of the 1/n dimensionless parameter from the Freundlich model (Table 7) is usually used to determine the adsorption affinity of the adsorbate to adsorb on various adsorbents. Depending on the value 1/n, the isotherm can be of S-type (1/n>1), C-type (1/n=1), or L-type (0<1/n<1). An S-type isotherm indicates weak adsorption between the adsorbate and adsorbent, a C-type means that the adsorption is linear, and an L-type isotherm suggests that the interactions between adsorbate and adsorbent are strong (Wang et al., Reference Wang, Miao, Sun and Zheng2018). Calculated 1/n values in the range 0.266–0.317 at pH 3 and 0.240–0.337 at pH 7 were characteristic of an L-type isotherm, demonstrating strong binding of AFB1 to all bentonites at both pH values.

The Gibbs free energy (ΔG) can be estimated from the Langmuir adsorption constant (Milonjić, Reference Milonjić2007):

(3) $$ \Delta G=-{\mathrm{R}T\mathrm{lnK}}_{\mathrm{L}}\ast $$

(4) $$ {\mathrm{K}}_{\mathrm{L}}^{\ast }={\mathrm{K}}_{\mathrm{L}}M\left(\mathrm{AFB}1\right)C\left({\mathrm{H}}_2\mathrm{O}\right) $$

where T is the absolute temperature in K, and R is the universal gas constant (8.314 J mol^–1 K^–1). The dimensionless Langmuir sorption constant K_L* was derived from the Langmuir adsorption constant K_L (L mg^–1), where M ( $ \mathrm{AFB}1 $ ) is the molar mass of AFB1 (312,277 mg mol^–1) and C(H₂O) is the molar concentration of water (55.56 mol dm^–3). The results are presented in Table 8. Calculated ΔG values for all bentonites were negative and similar, indicating that the AFB1 adsorption process was spontaneous in nature at both pH values.

Table 8.

Gibbs free energy for AFB1 adsorption by bentonites

Non-linear isotherms and high adsorption of AFB1 by various bentonites were reported by many authors, but depending on the characteristics of the bentonites as well as experimental conditions described in these papers, AFB1 adsorption capacities differed significantly. For example, Gan et al. (Reference Gan, Hang, Huang and Deng2019) measured AFB1 adsorption isotherms for six Ca saturated smectite samples in water, following the procedure of Grant and Phillips (Reference Grant and Phillips1998) at 25°C for 24 h, showing substantial differences in AFB1 affinity and adsorption capacity. They noticed that compared with Na-saturation, Ca-saturation improved the AFB1 adsorption capacities of all six samples, confirming observations that Ca-bentonites were in general better AFB1 adsorbents than Na-bentonites. Marroquín-Cardona et al. (Reference Marroquín-Cardona, Deng, Taylor, Hallmark, Johnson and Phillips2009, Reference Marroquín-Cardona, Deng, Garcia-Mazcorro, Johnson, Mitchell, Tang, Wang, Harvey and Phillips2022) reported non-linear (Langmuir) isotherms for adsorption of AFB1 by Na-bentonite in ultra-pure water (UP-H₂O) by adjusting the pH to 2 or 6.5 (with HCl or NaOH). Adsorption of AFB1 was followed by various initial toxin concentrations ranging from 0.4 to 8 μg mL^–1 in 5 mL of solution and with 0.1 mg of adsorbent, with a contact time of 2 h. They reported favorable AFB1 adsorption at both pH values, with slightly less adsorption observed at pH 2. Vekiru et al. (Reference Vekiru, Fruhauf, Rodrigues, Ottner, Krska, Schatzmayr, Ledoux, Rottinghaus and Bermudez2015) performed isotherm experiments for AFB1 adsorption by seven bentonites: dioctahedral smectites (two cis-vacant Ca-smectites: R and M34; four cis-vacant Na-smectites: MB, B7, M5, and M32; and one trans-vacant Na-smectite: C2) in phosphate buffer (potassium di-hydrogen phosphate/sodium hydrogen phosphate-dihydrate) at pH 7 (Vekiru et al., Reference Vekiru, Fruhauf, Sahin, Ottner, Schatzmayr and Krska2007), using the procedure of Grant and Phillips (Reference Grant and Phillips1998). The largest AFB1 capacity was found for cis-vacant Na-smectite (MB, 0.44 mol kg^–1) and the smallest was for trans-vacant Na-smectite (C2, 0.26 mol kg^–1). D’Ascanio et al. (Reference D’Ascanio, Greco, Menicagli, Santovito, Catucci, Logrieco and Avantaggiato2019) reported AFB1 adsorption isotherms for 29 dioctahedral smectite (bentonite) samples, measured at 37°C, pH 7 (phosphate buffer: potassium di-hydrogen phosphate 2-hydrate/di-sodium hydrogen phosphate 12-hydrate) and at contact time of 90 min. They obtained non-linear (Langmuir) isotherms for all bentonites studied and found that Na-bentonites exhibited maximum AFB1 adsorbed amounts >100 μg mg^–1 (0.32 mol kg^–1), compared with much smaller amounts measured with Ca-bentonites.

The data above illustrated that it is difficult to compare results from this study with literature data because the majority of investigations were performed either in distilled water (with or without adjustment of pH) or using different salts/acids for preparation of buffers at acidic (acetate, citrate, or phosphate salts) and at neutral (mainly phosphate salts) pH values. Also, in literature studies, experiments were performed with different initial AFB1 concentrations, at different contact times or at different temperatures. However, similar to results obtained in this study, several studies showed that the differences in AFB1 adsorption by bentonites at different pH values were clearly visible, illustrating the importance of solution pH in toxin adsorption.

Despite the abundance of literature data on the adsorption of AFB1 by bentonites in aqueous solutions, there is still a lack of information on the effects of buffer-solution compositions on the montmorillonite surface structure and its properties and consequently on AFB1 adsorption. Clay minerals are well known to be common cation exchangers, and thus when montmorillonite is suspended in an electrolyte solution, a significant fraction of exchangeable cations in the montmorillonite interlayer or from the surface appears in the solution and may be replaced by hydrogen ions or other cations from a buffer solution (Nagy et al., Reference Nagy, Kónya and Urbin1997). For these reasons, the behavior of bentonites in buffer solutions (pH 3 and 7) with and without the presence of AFB1 was investigated. For comparison, the behavior of bentonites with or without AFB1 in water without adjustment of pH was also studied under the same conditions. After separation of phases, supernatants were collected and the amounts of exchangeable cations (Ca²⁺, Mg²⁺, and Na⁺) released from montmorillonites were determined. Possible dissolution of Si and Al from the tetrahedral and octahedral sheets was also measured. The amounts of cations released were compared with the amounts of exchangeable cations in each bentonite (Table 3) calculated in mol kg^–1 and with maximum experimentally adsorbed amounts of AFB1 (mol kg^–1). The results are presented in Fig. 5.

Figure 5.

Cations in bentonites: (a) amounts of exchangeable cations; (b) amounts of cations released in water and buffer solutions; and (c) amounts of cations released in water and buffer solutions in the presence of AFB1.

The results in Fig. 5b showed that negligible amounts of Ca²⁺, Mg²⁺, and Na⁺ were detected in the aqueous solution without buffer for the B-Clay, Y-Clay, and G-Clay. However, significant amounts of Ca²⁺ and Mg²⁺ were released from all of the bentonites studied into the buffer solutions at pH 3 or 7. This behavior in buffer solutions was in agreement with the results of Metz et al. (Reference Metz, Amram and Ganor2005) and was attributed mainly to cation exchange. In the phosphate buffer used in the present study, at pH 3, K⁺, H⁺, H₂PO₄^–, and H₃PO₄^– were the dominant species in solution (Tanyol et al., Reference Tanyol, Yonten and Demir2015), and the results suggested that large amounts of Ca²⁺ and Mg²⁺ from the montmorillonite interlayers and surfaces were replaced by K⁺ and H⁺ from the buffer solution. The larger amount of Ca²⁺ dissolved in the phosphate buffer at pH 3 than the amount in exchange sites for the G-Clay probably resulted from dissolution of accessory calcite that makes up 4% of this bentonite (see ‘XRD analysis’ section below) (Hofmann et al., Reference Hofmann, Bauer and Warr2004). At pH 7, K⁺, HPO₄^2–, and H₂PO₄^– ions were mainly present in the buffer solution (Tanyol et al., Reference Tanyol, Yonten and Demir2015), and the presence of K⁺ induced similar behavior as at pH 3, reflecting exchange of Ca²⁺ and Mg²⁺ from interlayer and surface sites with K⁺ ions. The behavior of divalent cations in the buffer at pH 3 and 7 was thus expected and could be explained by their properties shown in Table 9 (Marcus, Reference Marcus1991). Ca²⁺ has a larger ionic radius and a smaller hydrated radius than Mg²⁺, meaning that Ca²⁺ can be exchanged easily with K⁺ with the smallest hydrated radius in the buffer solution. Although an H⁺ ion has the smallest ionic radius, it also has a large hydrated radius and can replace some cations from the interlayer space of montmorillonites and at their external surfaces. It can also lead to protonation of edge groups forming -Si-OH groups at tetrahedral sheet edges or -Al-OH₂ groups at octahedral sheet edges at pH 3 (Wang et al., Reference Wang, Liu, Wu, Li, Chen and Teng2010).

Table 9.

Properties of cations

r, ionic radius; Δr, width of hydration shell; n, number of water molecules in this shell; ΔG, molar Gibbs energy of hydration; ^aMarcus (Reference Marcus1991).

For the P-Clay (Na/Ca-montmorillonite), a similar trend for the release of Ca and Mg in buffers (pH 3 and 7) and in water was observed as for the B-Clay, Y-Clay, and G-Clay. However, for the P-Clay, a large amount of Na was detected in the water as well as in the buffer solution at pH 3 and 7 compared with its amount in exchange sites. Similar behavior of Na-bentonite in water was noted by Marroquín-Cardona et al. (Reference Marroquín-Cardona, Deng, Garcia-Mazcorro, Johnson, Mitchell, Tang, Wang, Harvey and Phillips2022), who reported a large amount of Na released in the water, 0.32 mol kg^–1, compared with the amount in exchange sites (0.59 mol kg^–1), whereas only small amounts of Ca and Mg were found in water solution (0.03 and 0.02 mol kg^–1) compared with their amounts in the exchange sites (0.33 and 0.20 mol kg^–1). In this study, the large amount of Na in the distilled water experiment suggested that the commercial P-Clay had been Na saturated but was insufficiently washed (chloride anions were detected by energy dispersive X-ray spectroscopy (EDS) analysis, results not shown). The large amount of Na released into solution at pH 3 and 7 confirmed its exchange with cations from the buffer solutions. The results for all bentonites confirmed that exchange was marked by the notable substitution of preferably divalent Ca²⁺ and Mg²⁺ with K⁺ and/or H⁺ derived from buffers. Additionally, the presence of negligible amounts of Al and Si in buffer solutions (pH 3 and 7) for all bentonites indicated that dissolution of montmorillonites or accessory mineral phases (aluminosilicates and amorphous SiO₂ phases) did not occur (Hofmann et al., Reference Hofmann, Bauer and Warr2004; Metz et al., Reference Metz, Amram and Ganor2005).

Compared with observed behavior of bentonites in buffer solutions at pH 3 and 7 (Fig. 5b), only slightly smaller amounts of Ca²⁺ and Mg²⁺ released were observed in the presence of AFB1 (Fig. 5c). These results suggested that when AFB1 was present in buffer solutions, AFB1 adsorption occurred simultaneously with ion exchange of native cations from the interlayer of montmorillonites and at their surfaces with ions from buffers. Negligible amounts of Al and Si were detected in supernatants of all bentonites in the presence of AFB1.

The dominant interlayer cations in the B-Clay, Y-Clay, and G-Clay are Ca²⁺ and Mg²⁺; thus, comparison of the amounts of divalent cations that were not released into the buffer solution in the presence of AFB1 with the amount of adsorbed AFB1 by each bentonite supported the suggestion that these divalent cations were the active sites at which toxin was adsorbed. Therefore, although significant amounts of divalent cations were released into the buffer solutions being exchanged with K⁺ (pH 3 and 7) and H⁺ (pH 3), the montmorillonite in bentonites still contained large amounts of divalent active sites available for AFB1 adsorption. Additionally, because Ca²⁺ and Mg²⁺ ions were replaced simultaneously with K⁺ and/or H⁺ from buffer solutions during AFB1 adsorption by bentonites, with an increase in the initial AFB1 concentration (isotherm study) and saturation of divalent active sites, the adsorbed monovalent cations also became active sites for AFB1 adsorption. Less and similar adsorption of AFB1 by all bentonites at pH 7, observed at greater initial AFB1 concentrations (Fig. 4b), may be explained by a lower proton concentration in buffer solution. In addition, similar to the behavior of bentonites in water, negligible amounts of Ca²⁺, Mg²⁺, and Na⁺ were detected in the aqueous solution in the presence of AFB1 for the B-Clay, Y-Clay, and G-Clay and maximum adsorbed amounts of AFB1 were similar to the amounts of toxin adsorbed in both buffers (Fig. 5a). Thus, because AFB1 adsorption by these bentonites in buffer solutions occurred simultaneously with ion exchange, both divalent and monovalent cations were the sites at which AFB1 was adsorbed. In contrast, in pure water solution where ion exchange did not occur, the dominant divalent cations in interlayers and at surfaces were involved in the AFB1 adsorption process.

The non-linear isotherms reported in this study revealed a complex AFB1 adsorption mechanism that probably results from a combination of several different interactions between the AFB1 molecule and montmorillonite. Deng et al. (Reference Deng, Velázquez, Billes and Dixon2010) proposed that the AFB1 molecule occupied the interlayer space of montmorillonite together with exchangeable cations and H₂O molecules, and that toxin adsorption occurred via H-bonding between the carbonyl oxygens and the hydration-shell of interlayer cations. In the case of H-bonding through H₂O bridges in solution, the cations would not affect the AFB1/montmorillonite bond strength. Deng et al. (Reference Deng, Velázquez, Billes and Dixon2010) also reported that the size and valence of cations as well as the presence of unpaired electrons in d-orbitals may influence the adsorption of AFB1. Daković et al. (Reference Daković, Matijašević, Rottinghaus, Ledoux, Butkeraitis and Sekulić2008, Reference Daković, Kragović, Rottinghaus, Ledoux, Butkeraitis, Vojislavljević, Zarić and Stamenić2012) studied the <5 μm fraction of the G-Clay as well as its Cu- and Zn-modified forms as AFB1 adsorbents in buffer solution at pH 3. In both studies, adsorption of AFB1 was greater by the Cu- and Zn-modified <5 μm fraction of the G-Clay than by the starting material. Daković et al. (Reference Daković, Matijašević, Rottinghaus, Ledoux, Butkeraitis and Sekulić2008) reported that in the Cu-modified <5 μm fraction of the G-Clay, the primary interlayer cations were Ca and Cu, and the presence of Cu enhanced AFB1 adsorption. For the Zn-modified <5 μm fraction of the G-Clay, greater AFB1 adsorption appeared to result in the formation of chelation complexes between Ca, Mg, and Zn and AFB1 molecules. The authors interpreted adsorption in terms of the radii of the exchangeable cations, increasing in the order: Mg<Zn<Ca. Although Zn has a larger ionic radius than Mg, it can interact more strongly with AFB1 than Mg because of the presence of d-electrons (Daković et al., Reference Daković, Kragović, Rottinghaus, Ledoux, Butkeraitis, Vojislavljević, Zarić and Stamenić2012). All of these observations showed that the properties of exchangeable cations, particularly their size, charge, and the presence/absence of d-electrons, play a major role in AFB1 adsorption by montmorillonite at pH 3 and 7.

The cation properties presented in Table 9 (Marcus, Reference Marcus1991) suggested that because Mg²⁺ has the largest hydrated radius, it was expected to be the primary adsorption site in the bentonites studied for AFB1 molecules, followed by Ca²⁺. When these adsorption sites were fully occupied with AFB1 molecules, AFB1 adsorption continued via interactions with monovalent cations (Na⁺ and K⁺). In addition, protonated edge groups existing at pH 3 in tetrahedral or octahedral sheets, as well as protons at the surfaces or in the interlayers of montmorillonites, created new active sites that enhance AFB1 adsorption. Overall, irrespective of whether Na- or Ca-bentonites are better AFB1 adsorbents, the data from the present study suggest that all exchangeable cations present in the interlayer and at the surface as well as edge groups of montmorillonite were involved in AFB1 adsorption. The affinity of exchangeable cations toward AFB1 increased in the following order: K⁺<Na⁺<Ca²⁺<Mg²⁺.

In addition to the effects of exchangeable cation on AFB1 adsorption, a variety of other factors (mineralogical, structural, chemical, and physico-chemical characteristics of smectite) influenced adsorption. Generally, greater AFB1 adsorption was achieved with Ca-bentonites (B-Clay: cis-vacant; Y-Clay: cis/trans-vacant; and G-Clay: trans-vacant), and the AFB1 adsorption capacity increased with an increase in the CEC of these montmorillonites, which contrasts with the conclusions of Deng et al. (Reference Deng, Liu, Velázquez and Dixon2012). The P-Clay (cis-vacant montmorillonite containing Na⁺ and Ca²⁺ as primary cations in the interlayer and at the surface) exhibited the lowest AFB1 adsorption capacity at both pH values probably due to its having the smallest montmorillonite content (63%). However, at pH 3, the AFB1 capacity of this clay was similar to that of the G-Clay (trans-vacant, 88% of montmorillonite) in which Ca²⁺ was the dominant interlayer cation. Although literature data would suggest that the trans-vacant G-Clay would have a lower AFB1 adsorption capacity (Vekiru et al., Reference Vekiru, Fruhauf, Rodrigues, Ottner, Krska, Schatzmayr, Ledoux, Rottinghaus and Bermudez2015), the G-Clay exhibited a much greater AFB1 capacity at pH 7 than the P-Clay, probably due to its larger montmorillonite content. In fact, the G-Clay possessed similar AFB1 adsorption capacities at both pH values. The other two Ca-bentonites (B-Clay and Y-Clay) showed much higher AFB1 capacities at pH 3 but similar capacities at pH 7 as the G-Clay. The Y-Clay had the largest montmorillonite content (98%), but this clay was of the cis/trans vacant type, which may cause slightly lower adsorption in comparison with the B-Clay, which was cis-vacant with a montmorillonite content of 83%. Enhanced AFB1 adsorption by cis-montmorillonites at pH 3 may be related to the position of octahedral OH groups that were on the same side of the octahedral site, meaning they were probably more available for interactions with protons from buffer solution, thereby creating protonated sites that can interact with AFB1. The opposing OH groups in trans-vacant montmorillonite (G-Clay) reduced the number and availability of sites for proton adsorption, thereby decreasing adsorption of AFB1 at pH 3. Thus, for the G-Clay at pH 3, only the exchangeable cations were available for AFB1 adsorption and its adsorption capacity was similar to that at pH 7. In addition, although it was expected that bentonites with a larger SSA_N2BET value would exhibit greater AFB1 adsorption, comparison between the SSA_N2BET values (Table 5) and AFB1 adsorption capacities (Table 6) at pH 3 and 7 showed that this was not the case. At pH 3, the AFB1 adsorption capacity of the Y-Clay (cis/trans-vacant, 98% of Ca-montmorillonite) was less than that of the B-Clay (cis-vacant, 83% Ca-montmorillonite), whereas its SSA_N2BET was ~1.64 times greater than that of B-Clay. At pH 7, AFB1 adsorption capacities of these two bentonites were comparable but also did not correspond to the ratio of their specific surface areas. The lack of correlation between changes in AFB1 adsorption capacities and SSA_N2BET was also evident for the G-Clay (trans-vacant, 88% Ca-montmorillonite) and P-Clay (cis-vacant, 63% Na/Ca montmorillonite). In summary, results confirmed that mineralogical, structural, chemical, and physico-chemical characteristics of smectite, all together, influenced AFB1 adsorption. In buffer solutions, ion exchange and AFB1 adsorption onto montmorillonites occurred simultaneously and exchangeable cations in the interlayer and at the surface, as well as protons adsorbed at the edges and in the interlayer under acidic conditions played an important role in AFB1 adsorption. In addition, the absence of plateaus in the measured AFB1 isotherms (Fig. 4) suggests that all exchangeable cations in montmorillonites were not fully saturated with AFB1 molecules. A summary of the adsorption of AFB1 onto bentonite(s) in buffer solutions is shown schematically in Fig. 6.

Figure 6.

Schematic representation of AFB1 adsorption onto montmorillonite in buffers at pH 3 and 7. The structure of montmorillonite was generated in the VESTA program (Momma and Izumi, Reference Momma and Izumi2011) using parameters from Viani et al. (Reference Viani, Gualtieri and Artioli2002).

Finally, although all studied Ca-bentonites showed significant in vitro AFB1 adsorption in buffer solutions (Table 6) and fulfil the requirements of EU Regulation (in which an AFB1 binding capacity over 90% from a single in vitro AFB1 experiment is proposed for bentonite selection as an additive), an in vivo safety evaluation for application as AFB1 binding agents is also essential before inclusion in diets. So far, from the Ca-bentonites studied, only the G-Clay (raw and <5 μm fraction) was tested under in vivo conditions (Shannon et al., Reference Shannon, Ledoux, Rottinghaus, Shaw, Daković and Marković2017). The addition of both adsorbents to the basal diet did not cause negative effects in any of the measured response variables, indicating that the G-Clay and concentrated G-Clay are safe to feed broiler chicks and will not negatively impact performance. Despite the fact that the <5 μm fraction of the G-Clay showed a greater AFB1 adsorption in vitro than the raw material (93.39 vs 79.30 mg g^–1), no large differences were observed between the two adsorbents with respect to reducing or preventing the toxic effects of AFB1, demonstrating that raw G-Clay was a very good AFB1 binding agent. In any case, although in vitro investigations are essential for understanding the mechanism of AFB1 adsorptions by bentonites, in vivo trials are necessary for their final selection as animal feed additives.