LREEs vs HREEs

REE fractionation provides useful insight into the composition of the source magma (Christidis Reference Christidis1998; Kiipli et al. Reference Kiipli, Hints, Kallaste, Verš and Voolma2017; Hong et al. Reference Hong, Algeo, Fang, Zhao, Ji, Yin, Wang and Cheng2019), the mineralogical and geochemical characteristics of the altered volcanic ash, and the environmental conditions of the alteration process (Bau Reference Bau1991; MacRae et al. Reference MacRae, Nesbitt and Kronberg1992; Nath et al. Reference Nath, Bau, Rao and Rao1997; Chen et al. Reference Chen, Algeo, Zhao, Chen, Cao, Zhang and Li2015; Liao et al. Reference Liao, Hu, Cao, Wang, Yao, Wu and Wan2016). REE fractionation during the formation of diagenetic bentonites consists of two phases. In the first phase, some REEs, such as Eu and HREEs, are co-precipitated with plagioclases and/or incorporated into the structure of plagioclases in early-stage magma, causing their depletion in intermediate to felsic magma. The second phase of REE fractionation may occur after the deposition of volcanic ash in an aqueous environment where it undergoes alteration. The fractionation of REEs in the second phase is controlled by the pH, temperature, alkalinity, salinity, redox condition, and water/rock ratio in the aqueous environment (dos Muchangos Reference dos Muchangos2006; He et al. Reference He, Zhong, Xu and Li2014; Chen et al. Reference Chen, Algeo, Zhao, Chen, Cao, Zhang and Li2015; Hong et al. Reference Hong, Algeo, Fang, Zhao, Ji, Yin, Wang and Cheng2019). This will be discussed more in the next sections.

The ΣLREE/ΣHREE ratio in the parent rocks varies from 6.29 to 8.97 (mean 7.69, Table 2). Similarly, the ΣLREE/ΣHREE ratio in bentonites showed a somewhat comparable variation, ranging from 6.90 to 8.54 (mean 7.67), indicating the relatively elevated concentration of LREEs in both parent rocks and bentonites.

The chondrite-normalized REE patterns of the parent rocks (P/C) and bentonites (B/C) suggest the same trend of LREEs enrichment relative to HREEs (Fig. 3).

Fig. 3. Chondrite-normalized REE patterns of a, b parent rocks and c, d bentonites. Chondrite data are from McDonough & Sun (Reference McDonough and Sun1995).

Despite the small variations in the ΣLREE/ΣHREE, the average of the ratio is almost identical for both parent rocks and bentonites, suggesting relatively equal losses/gains, if any, in the concentration of REEs during the parent-rock alteration to bentonite. As stated, the source magma composition controls the degree of enrichment in LREEs compared to HREEs, so that the larger the silica contents of the magma, the greater the enrichment in LREEs. Modabberi et al. (Reference Modabberi, Namayandeh, Setti and López-Galindo2019) suggested that the initial source magma composition of the studied samples was intermediate to felsic in Groups 1 and 3, and felsic in Groups 2 and 4 (Table 1), thus explaining the depletion of HREEs and Eu in most of the samples.

Because the solubility of HREE complexes is greater than that of LREEs under alkaline pH where bentonite is formed (Christidis & Huff Reference Christidis and Huff2009), HREEs tend to be relatively dissolved in the second phase of REE fractionation (dos Muchangos Reference dos Muchangos2006), which provides an explanation for the depletion of HREEs from parent rocks to bentonites in the samples analyzed here.

Ce and Eu anomalies

REEs occur exclusively in a trivalent oxidation state except for Ce and Eu, which are also found in tetravalent and divalent oxidation states, respectively. Because Ce and Eu are redox-sensitive, Ce⁴⁺ and Eu²⁺ may be fractionated relative to trivalent REEs, causing significant positive or negative anomalies under certain redox conditions (Elderfield Reference Elderfield1988; Bau Reference Bau1991; MacRae et al. Reference MacRae, Nesbitt and Kronberg1992; Nath et al. Reference Nath, Bau, Rao and Rao1997; Chen et al. Reference Chen, Algeo, Zhao, Chen, Cao, Zhang and Li2015). The variation in the anomalies of Ce and Eu can be used as a proxy to study paleo-sedimentary environments and processes involved in the formation of clay minerals (Christidis Reference Christidis1998; Hastie et al. Reference Hastie, Kerr, Pearce and Mitchell2007; Özdamar et al. Reference Özdamar, Ece, Uz, Boylu, Ercan and Yanik2014; Liao et al. Reference Liao, Hu, Cao, Wang, Yao, Wu and Wan2016; Caballero & de Cisneros Reference Caballero and de Cisneros2017; Kiipli et al. Reference Kiipli, Hints, Kallaste, Verš and Voolma2017; Hong et al. Reference Hong, Algeo, Fang, Zhao, Ji, Yin, Wang and Cheng2019). The parent rock-normalized REE patterns of the bentonites (B/P) lead us to expect a further fractionation of REEs during the second fractionation phase (Fig. 4). To achieve a better quantitative insight into the distribution of REEs in the parent rocks, δEu and δCe values were calculated using Eqs 1 and 3, respectively (Liao et al. Reference Liao, Hu, Cao, Wang, Yao, Wu and Wan2016):

(3) $\mathrm{\delta Ce}={\mathrm{Ce}}_N/{\left({\mathrm{La}}_N\times {Pr}_N\right)}^{1/2}$

where the subscript N denotes chondrite-normalized.

Fig. 4. Parent rock-normalized REE patterns of bentonites.

The values of δEu and δCe for P/C, B/C, and B/P (parent rock-normalized REE patterns of bentonites) were also calculated (Table 2). The δEu values for parent rocks ranged from 0.635 to 0.902 (mean 0.752), and from 0.291 to 0.750 (mean 0.509) for bentonites, suggesting a characteristic negative anomaly for Eu in both types of samples. Regarding δCe, with the exception of ST (parent rock and bentonite), all samples showed a very flat pattern, with a very weak negative Ce negative anomaly (1.01 > δCe > 0.953). When bentonites were normalized to parent rocks, Ce did not display any significant enrichment/depletion, showing a short range of δCe values from 0.925 to 1.17 (mean 1.01). Eu, conversely, exhibited a variable anomaly across the samples in B/P, showing a wide range of δEu values, from 0.333 to 0.831 (mean 0.590).

The δEu and δCe values of P/C were inherited chiefly from the source magma, as no advanced degree of weathering was observed in the parent rocks. In general, the δCe values were almost similar from P/C to B/C verified by the Ce values in B/P (~1.00), suggesting a relatively equal Ce loss/gain, if any, in comparison to La and Pr in the second phase of REE fractionation. ST is the only sample showing a different δCe value (0.691) in P/C. Conversely, throughout the second phase of REE fractionation, Eu was variously fractionated relative to Sm and Gd, as indicated by more pronounced negative Eu anomalies in B/C compared to P/C under increasingly oxidizing diagenetic conditions. Ce³⁺ was transformed to the relatively insoluble Ce⁴⁺ and, as such, it may be removed from pore waters, showing a positive anomaly through accumulation in the solid-phase sediments (Elderfield Reference Elderfield1988; Nath et al. Reference Nath, Bau, Rao and Rao1997). Chen et al. (Reference Chen, Algeo, Zhao, Chen, Cao, Zhang and Li2015) proposed that, in marine basins, the δCe value is <0.5 in oxic water (>1.5 in its solid phase sediment), ~0.6–0.9 in suboxic water (~1.1–1.4 in its solid phase sediment), and 0.9–1.0 in anoxic water (1.0–1.1 in its solid phase sediment). In other words, the larger the oxygen content in pore waters, the more Ce is removed from the water to enrich the sediment. The δCe values in B/P, hence, can be used as an indicator to evaluate the redox conditions of the local environment of the studied samples. The δCe average value was ~1.00 in all samples, suggesting a suboxic to anoxic aquatic environment for the transformation of volcanic ash to bentonite.

Eu is another highly redox-sensitive LREE, the trivalent oxidation state of which can be transformed to a more soluble divalent oxidation state under highly reducing conditions, causing the removal of Eu from the solid phase sediment (Elderfield Reference Elderfield1988; MacRae et al. Reference MacRae, Nesbitt and Kronberg1992; Chen et al. Reference Chen, Algeo, Zhao, Chen, Cao, Zhang and Li2015). The depletion of Eu from parent rocks to bentonites in the second phase of REE fractionation can be ascribed to the reducing conditions where volcanic ash underwent diagenetic alteration. This confirms the hypothesis that the local environment of bentonite formation was suboxic aquatic. The depletion of Eu, however, was not the same in all the bentonite deposits, increasing from Group 1 (δEu ~0.333) to Group 4 (δEu ~0.822) as a result of local variations in the redox potential of the diagenetic environments.

Unlike Ce, which can change valence at ambient temperature, high temperature (>200°C) is another critical factor needed for the transformation of Eu³⁺ to Eu²⁺ (Nath et al. Reference Nath, Bau, Rao and Rao1997). Bau (Reference Bau1991) calculated the Eu³⁺/Eu²⁺ redox equilibrium for temperatures up to 600°C, suggesting that even in low redox potential, Eu exists predominantly in a divalent oxidation state with increasing temperature. Based on field observations and petrographic evidence, two types of volcanic rocks were suggested for the study area: pyroclastic flow and volcaniclastic rocks (Modabberi et al. Reference Modabberi, Namayandeh, Setti and López-Galindo2019) which could provide the high temperature required for the transformation of Eu³⁺ to Eu²⁺. Christidis & Huff (Reference Christidis and Huff2009) reported that the alteration process is promoted at high temperatures and 300 to 850°C was suggested as a common temperature range for pyroclastic flows. Although this wide range of temperature is high enough to provide the required temperature for the reductive dissolution of Eu, it might also have a local impact on the amount of Eu³⁺ that can be transformed to Eu²⁺.

Trace elements distribution

Similar to REEs, the distribution of trace elements in bentonites is also controlled by their concentration in the parent rocks inherited from the source magma and the physiochemical conditions of the local environment, including the style of volcanic eruptions, pH, Eh, salinity, and temperature (Nesbitt Reference Nesbitt1979; Christidis Reference Christidis1998; Kiipli et al. Reference Kiipli, Hints, Kallaste, Verš and Voolma2017).

The parent rock-normalized trace element diagram was used to determine the distribution of trace elements in the bentonite samples (Fig. 5). Despite some similarities in the loss or gain of trace elements, one cannot, in general, conclude that they show the same behavior in all instances. Some relationship exists, however, among the large ion lithophile elements (LILE). Similarly, the high field strength elements (HFSE) show a significant positive correlation with each other. LILEs are characterized by large ionic radii and low valence (e.g. Ba, Rb, Sr, V, and Tl), whereas HFSEs are differentiated from other trace elements by small ionic radii and high valence states (e.g. Zr, Hf, Nb, Y) (Zielinski Reference Zielinski1985; Summa & Verosub Reference Summa and Verosub1992; Fanti Reference Fanti2009). In the samples analyzed here, the loss/gain of trace elements was relatively equal in HFSEs; unlike LILEs, their concentrations relative to each other did not show any significant changes.

Fig. 5. Parent rocks-normalized trace elements patterns of bentonites. Only the patterns of those elements are given that showed a concentration above the detection limit of the method.

Losses or gains of elements

During the alteration of volcanic ash, variations in the concentration of each individual element in the residual materials are directly related to the concentration of other constituents (Nesbitt et al. Reference Nesbitt, Markovics and Price1980). Because of the dilution effect, the removal of major elements from volcanic ash and/or residual behavior of elements bring about drastic changes in the concentration of immobile materials (Kiipli et al. Reference Kiipli, Hints, Kallaste, Verš and Voolma2017). Residual behavior occurs when elements are essentially immobile so that their concentrations increase through the depletion of mobile elements from the environment (Christidis Reference Christidis1998; Kiipli et al. Reference Kiipli, Hints, Kallaste, Verš and Voolma2017; Hong et al. Reference Hong, Algeo, Fang, Zhao, Ji, Yin, Wang and Cheng2019). Study of the behavior of immobile elements during the devitrification of volcanic ash is imperative, as increases or decreases in their concentrations show the degree to which other elements are lost or gained (Nesbitt et al. Reference Nesbitt, Markovics and Price1980; Christidis Reference Christidis1998; Kiipli et al. Reference Kiipli, Hints, Kallaste, Verš and Voolma2017).

Instead of using the absolute concentration of elements in analyzing their mobility, Nesbitt (Reference Nesbitt1979) suggested use of the element’s ratio to an immobile element, as this ratio is not influenced by the dilution effect. Al has commonly been considered an immobile element during the chemical weathering, transport, diagenesis, and metamorphism of volcanic ash (Nesbitt Reference Nesbitt1979; Slack & Stevens Reference Slack and Stevens1994; Laviano & Mongelli Reference Laviano and Mongelli1996; Christidis Reference Christidis1998; Göncüoglu et al. Reference Göncüoglu, Günal-Türkmenoglu, Bozkaya, Ünlüce-Yücel, Okuyucu and Yilmaz2016; Kiipli et al. Reference Kiipli, Hints, Kallaste, Verš and Voolma2017; Hong et al. Reference Hong, Algeo, Fang, Zhao, Ji, Yin, Wang and Cheng2019) which makes it an appropriate choice for studying the percentage changes in the concentration of elements during the alteration process.

The percentage changes relative to parent rocks were calculated following Nesbitt et al. (Reference Nesbitt, Markovics and Price1980):

(4) $\%\text{change}=\left(\left(R_s–R_p\right)/R_p\right)\times 100$

where R _s and R _p are the concentrations of an element in a bentonite and its parent rock, respectively, each of which is normalized to the corresponding Al %. In addition, the absolute concentrations of Al in bentonites and parent rocks were used to measure the percentage change of Al itself to determine whether other elements were lost or gained (Nesbitt Reference Nesbitt1979):

(5) $\%\text{change}=\left(\left({\mathrm{Al}}_s–{\mathrm{Al}}_p\right)/{\mathrm{Al}}_p\right)\times 100$

where Al _s is the concentration of Al in a bentonite and Al _p is the concentration of Al in the corresponding parent rock.

The percentage changes in elements during alteration of volcanic ash to bentonite (Table 3) revealed that the average amount of Al lost was small (~12.5%) among all the deposits and this agrees well with the literature which has reported Al as an immobile to poorly mobile element (Nesbitt Reference Nesbitt1979; Slack & Stevens Reference Slack and Stevens1994; Laviano & Mongelli Reference Laviano and Mongelli1996; Christidis Reference Christidis1998; Göncüoglu et al. Reference Göncüoglu, Günal-Türkmenoglu, Bozkaya, Ünlüce-Yücel, Okuyucu and Yilmaz2016; Kiipli et al. Reference Kiipli, Hints, Kallaste, Verš and Voolma2017; Hong et al. Reference Hong, Algeo, Fang, Zhao, Ji, Yin, Wang and Cheng2019). Although the variation of Al concentration was not significant in most of the samples, the loss of Al was found to be large in a few samples such as SK (–34.1%). The changes observed in Al concentration may be the result of the dilution effect caused by losses or gains of other major elements such as Si, which is the predominant element in both parent rocks and bentonites. The most negative percentage change in Al concentration was observed in SK, where Si showed the most positive percentage change (59.42%). When the percentage changes of Al and Si were plotted against each other (Fig. 6), they exhibited an inversely proportional trend, with R² > 90, suggesting a significant inverse correlation.

Table 3 Percentage change of elements during the alteration of volcanic ash.

Fig. 6. The inverse correlation between the percentage changes of Al

lthough the concentration of Al is more or less controlled by the variation in the Si concentration, it may not be enough to explain the –34.1% decrease in the Al concentration of the SK sample. The initial volcanic ash from which bentonites formed was deposited in an aqueous environment which increased the water content of the samples during the hydrolysis reaction that converts the volcanic ash to bentonite. The loss on ignition (LOI) increased considerably in all the samples during alteration, but this value was significantly larger in the SK sample (~1909%) (Table 3). Increasing the water content of the samples caused the residual elements such as Al to be diluted, providing another explanation for the decrease in the Al content of the SK sample. The variation in the LOI content of the samples showed that the physiochemical condition during the formation of bentonites was locally different.

The smallest variation in Al concentration was observed in the SG sample (~0.4%). To evaluate the mobility of elements, several samples from this deposit (Chah-Golestan) were used to decrease the impact of local geochemical and lithological conditions. This will be discussed further below.

As stated, Al has been considered consistently as an immobile element and, thus, ∽–35% change, which is the most negative percentage change in Al concentration, can loosely be regarded as a plausible threshold to identify potential immobile elements, as the percentage changes in the Al concentration mostly result from the dilution effect. Similarly, elements showing a positive percentage (residual enrichment) were also proposed by Christidis (Reference Christidis1998) to be considered immobile. For the purpose of the present study, those elements displaying positive percentage changes, or negative changes smaller than –35% in more than half of the deposits (4 of 7), were hypothesized as potential immobile elements, including Si, Al, Na, Mg, Be, Ga, In, U, Y, Zn, Pr, Nd, Sm, Gd, Tb, Dy, Ti, Ba, Cs, Ge, Nb, Pb, Sc, Sn, Th, V, Zr, La, Ce, Eu, Er, Yb, Hf, and Ho. At the current stage of the study, in no case has the immobility of any of these elements been proved. This method was used solely to achieve an understanding of the potential immobile elements, and further analysis is needed to evaluate the mobility of the elements.

Statistical analysis

The results of bivariate correlation analyses (Tables 4 and 5), calculated using the percentage changes provided in Table 3, showed that besides the expected negative correlation with Al (r > –0.953), Si also displayed an inverse correlation with LREEs. A positive correlation (r > 0.900) existed between HFSEs and REEs. Both LREEs and HREEs have positive correlations with other elements of their group (r > 0.950), but the r values fall to <0.850 when LREEs are compared with HREEs. Eu exhibited a less significant correlation with other REEs (R < 0.800), while no significant correlation was found between Lu and HREEs.

Table 4 The bivariate correlation analysis of Si with Al, HFSEs, and REEs¹0.

1 The bivariate correlation of elements was calculated using the percentage changes provided in Tables 4 and 5.

** Correlation is significant at the 0.01 level

* Correlation is significant at the 0.05 level

Table 5 The bivariate correlation analysis of Si with Al and LILEs¹0.

1 The bivariate correlation of elements was calculated using the percentage changes provided in Tables 4 and 5.

** Correlation is significant at the 0.01 level

* Correlation is significant at the 0.05 level

Some unexpected elements (e.g. Si and Na) appeared among the potential immobile elements in the previous section, as they have been reported to be actively mobile (Caballero et al. Reference Caballero, Reyes, Delgado, Huertas and Linares1992; Christidis et al. Reference Christidis, Scott and Marcopoulos1995; Christidis & Huff Reference Christidis and Huff2009). The binary diagram was plotted using Chah-Golestan (SG) bentonite and parent rock samples to evaluate the mobility of each of the elements provided in the list of potential immobile elements (Fig. 7a and b). The largest deposit in the studied area was SG, where various layers of bentonite are exposed at the surface. These exposed layers provided numerous samples from this deposit (instead of using samples from different deposits) which were used in the binary diagrams to decrease the impact of local geochemical and lithological conditions, supposedly affecting the mobility of element. Overall, seven bentonites and two volcanic rocks from Chah-Golestan were used in the binary diagrams (see Table B in the SI). The SG2 sample is the parent rock of the samples SG-D, SG-E, SG-F, and SG-G where the gradual transition of SG2 to SG-G could be observed clearly (Fig. 1). The samples SG-A, SG-B, and SG-C (bentonite) and SG3 (volcanic rock) were also taken from other horizons of the Chah-Golestan deposit to increase the diversity of the samples in the binary diagrams and ensure that samples were taken from the different sections of the deposit.

Fig. 7. a Representative binary diagrams of Al vs potential immobile elements with R² values >0.800 for the bentonites and parent rocks of Chah-Golestan deposit. b The representative binary diagrams of Al vs potential immobile elements with R² values (a, b, c, d) between 0.500 and 0.800, and (e, f, g, h) < 0.500 for the bentonites and parent rocks of the Chah-Golestan deposit. The negative R² value for the Al vs Si line is because the intercept was set at zero, forcing the line to pass through the origin.

If two different chemical elements of several bentonites are immobile during the alteration of volcanic ash to bentonite, they should show a strong correlation, with the least squares line (R²) passing through the origin of the binary diagram. Two such elements should be considered immobile, as two elements with distinctly different chemical properties are unlikely to be of equal solubility and mobility (Christidis Reference Christidis1998; Kiipli et al. Reference Kiipli, Hints, Kallaste, Verš and Voolma2017).

Based on the correlation of each element with the least squares line, i.e. R² values (Fig. 8), the mobility of elements was loosely classified into three groups: immobile, poorly mobile, and mobile. Note that the following is merely a qualitative classification because choosing a definitive boundary delineating each group of elements may not be applicable at this stage of the study. The elements with R² values > 0.800 (80% accuracy) were thus proposed to be considered immobile, as Kiipli et al. (Reference Kiipli, Hints, Kallaste, Verš and Voolma2017) also found the R² values up to 0.73 to be relatively low. For the purpose of the present study, elements with R² values between of < 0.500 and 0.800 were also considered to be poorly mobile and those with values < 0.500 were considered to be mobile.

Fig. 8. Qualitative estimate of the mobility of elements during the alteration of volcanic ash to bentonite for the eastern Iranian bentonite deposits. The R² values were obtained from the binary diagrams of Al vs potential immobile elements. For the binary diagram with –R², the real values are not shown here, and all of them are represented as <0.

The highest R² values (0.800–1.00) were observed in Ti, Gd, Ga, Pr, Tb, Nd, Sm, Ce, and Nb, all of which were included in the list of potential immobile elements. Along with Al, they were probably immobile during bentonite formation. Regression lines pointing to the origin of the chart for Sc, Hf, U, Dy, In, Zr, La, and Eu show less correlation, with R² values between 0.500 and 0.800. These elements were also included in the list of potential immobile elements but, because of the lower R² values, they cannot be verified as being immobile, so they are classified in the poor-mobility group. In addition, the elements (Si, Na, Mg, Be, Y, Zn, Ba, Cs, Ge, Pb, Th, V, Er, and Yb) with R² < 0.500 were certainly mobile during bentonite formation; therefore, they are listed in the mobile elements group. Most of the elements classified here as immobile or of low mobility agree quite well with those reported in the literature on mobility of elements throughout volcanic ash transformation (Nesbitt Reference Nesbitt1979; Nesbitt et al. Reference Nesbitt, Markovics and Price1980; Zielinski Reference Zielinski1985; Christidis Reference Christidis1998; dos Muchangos Reference dos Muchangos2006; Özdamar et al. Reference Özdamar, Ece, Uz, Boylu, Ercan and Yanik2014; Kacmaz Reference Kacmaz2016; Liao et al. Reference Liao, Hu, Cao, Wang, Yao, Wu and Wan2016; Kiipli et al. Reference Kiipli, Hints, Kallaste, Verš and Voolma2017; Hong et al. Reference Hong, Algeo, Fang, Zhao, Ji, Yin, Wang and Cheng2019).

Major element mobility

The results demonstrate that even elements displaying strong residual enrichment such as Si (up to ~60%) and Na (up to ~80%) may not be immobile, as shown in the binary diagram analysis (Fig. 7b). The release of Si from volcanic ash to the aqueous environment is a critical factor in the formation of bentonite (Christidis & Huff Reference Christidis and Huff2009). Despite the alteration of volcanic ash to bentonite, the Si fraction that was apparently washed off the parent rocks was not leached entirely from the environment, even exhibiting strong residual enrichment in some of the deposits. The mineralogy of bentonite samples in eastern Iran showed that the bentonite samples are not composed entirely of smectite minerals (Na-montmorillonite), and they include up to 39% of opaline silica (i.e. opal-CT and opal-C) as a secondary phase (Table 1). Given this, a fraction of Si was indeed depleted from the volcanic ash but precipitated and subsequently crystallized to opaline silica in situ. Modabberi et al. (Reference Modabberi, Namayandeh, Setti and López-Galindo2019) reported that in the eastern Iranian bentonite zone, the water/rock ratio, which is one of the factors controlling the availability of Si in the system during alteration, was low. At higher water/rock ratios in an open system, the free Si released from the volcanic ash may not precipitate to any great extent, accumulating instead in the fluid phase and leaching out of the environment (Christidis & Huff Reference Christidis and Huff2009). Overall, Si was depleted from the volcanic ash, showing strong mobility, but fluid flow volume was not sufficient to leach it from the environment, i.e. any residual enrichment does not equal immobility. In all instances, the SiO₂/Al₂O₃ ratio is significant (up to 7), and thus, the variation in the concentration of Si has a considerable impact on that of the other elements in general and immobile elements in particular.

The behavior of alkalies during volcanic-ash transformation governs the formation of either smectites or zeolites; if the alkalies are not removed from the system, the (Na⁺+K⁺)/H⁺ ratio increases and, instead of smectites, zeolites are formed (Christidis et al. Reference Christidis, Scott and Marcopoulos1995; Yildiz & Kuscu Reference Yildiz and Kuscu2004). The absence of zeolites in the samples studied suggests that alkalies were leached out of the environment to some extent. The changes in K percentage showed substantial depletion in all deposits (up to 97%), while Na seems to be less mobile across the deposits and acts as a counter cation in the structure of Na-montmorillonite, which is the main mineralogical phase in the samples studied (Table 3).

The large (Mg²⁺)/(H⁺) ratio and leaching of alkali elements from the aqueous environment promotes the destabilization of the plagioclase structure and provides a suitable pH (~8) for the formation of bentonite instead of zeolite (Christidis Reference Christidis1998; Christidis & Huff Reference Christidis and Huff2009; Modabberi et al. Reference Modabberi, Namayandeh, Setti and López-Galindo2019). Christidis & Huff (Reference Christidis and Huff2009) suggested that in the transformation of acidic parent rocks with a small Mg content, the required Mg is supplied by the fluid phase. The concentration of Mg has increased significantly from the parent rocks to bentonites in all the samples studied with the exception of FKH (Table 3), suggesting the addition of Mg to the bentonite samples from the aqueous environment. Because the FKH sample is not acidic in composition, it did not need to take up Mg from the aqueous environment as its parent rock already had the largest concentration of Mg ( 2.60%) across the samples.

Trace and rare earth element mobility

All the LILEs were listed in the mobile group of the classification provided. The alkaline trace elements such as Rb and alkaline earth metals (e.g. Ba and Sr), known mostly as LILE, are highly mobile in aqueous environments (Zielinski Reference Zielinski1985; Khan Reference Khan1990; Hong et al. Reference Hong, Algeo, Fang, Zhao, Ji, Yin, Wang and Cheng2019). They display pronounced positive correlations (r > 0.950) with K (i.e. alkaline metal) and Ca (i.e. alkaline earth metal) which are geochemically similar major elements (Tables 4 and 5). Biotite, feldspar, and K-feldspar host Ba and Rb in volcanic ash, rapidly releasing these mobile elements during alteration (Renock et al. Reference Renock, Landis and Sharma2016). Although LILEs show a positive correlation with K, their contents are not controlled significantly by K-bearing minerals but by sorption on the surface of bentonite (Nesbitt et al. Reference Nesbitt, Markovics and Price1980; Hodson Reference Hodson2002).

HFSEs are commonly considered less soluble and relatively immobile elements, preserving the characteristics of the source materials (Merriman & Roberts Reference Merriman and Roberts1990; Huff et al. Reference Huff, Merriman, Morgan and Roberts1993; Pellenard et al. Reference Pellenard, Deconinck, Huff, Thierry, Marchand, Fortwengler and Trouiller2003; Calarge et al. Reference Calarge, Meunier, Lanson and Formoso2006). Zr, Hf, and Nb are listed in the immobile to poorly mobile groups, while some other HFSEs, such as Y, Ta, and Th, showed some degree of mobility (Fig. 8). During the diagenetic alteration of volcanic ash, authigenic clay minerals are formed in situ, and, therefore, their immobile contents may not be fractionated and may be mostly inherited from the source magma composition. The physiochemical conditions (pH, Eh, salinity) of the local environment are also important factors controlling the mobility of trace elements and possibly causing some redistribution of immobile elements (Hong et al. Reference Hong, Algeo, Fang, Zhao, Ji, Yin, Wang and Cheng2019). For instance, while Zr and Y are known as immobile elements in the literature (Nesbitt Reference Nesbitt1979; Christidis Reference Christidis1998; Kiipli et al. Reference Kiipli, Hints, Kallaste, Verš and Voolma2017), they were found to be poorly mobile and mobile, respectively, in these samples (Fig. 8). Zr and Y are incorporated into the structure of early-crystallized heavy minerals such as zircon, garnet, and monazite. These minerals are very sensitive to physical sorting in aqueous environments, and hence, susceptible to being leached as colloidal phases thereby removing Zr and Y from the open system (Summa & Verosub Reference Summa and Verosub1992; Wray Reference Wray1995; Hong et al. Reference Hong, Algeo, Fang, Zhao, Ji, Yin, Wang and Cheng2019).

Because of similar geochemical characteristics, REEs are also classified as HFSEs (Zielinski Reference Zielinski1985; Summa & Verosub Reference Summa and Verosub1992; Fanti Reference Fanti2009; He et al. Reference He, Zhong, Xu and Li2014), showing a strong positive correlation with their equivalent trace elements such as Zr and Hf (Tables 4 and 5). All of the LREEs and Gd, Tb, and Dy from the HREEs were found to be either immobile or poorly mobile (Fig. 8). HREEs show a degree of mobility indicating that the heavier the element, the more mobile it is during volcanic ash conversion. LREEs have been proposed to be less soluble and retain their concentration in an alkaline solution, while HREEs favor making complexes in the aqueous phase, causing their removal from the solid phase-sediments (Summa & Verosub Reference Summa and Verosub1992; dos Muchangos Reference dos Muchangos2006). The various dissolution behaviors shown by REEs may explain the greater mobility of some of the heavier HREEs such as Er compared to lighter HREEs such as Gd and Tb. Summa & Verosub (Reference Summa and Verosub1992) stated that the degree to which HREEs accumulate in the fluid phase is controlled by the water/rock ratio such that their mobility increases in dense solutions (i.e. low water/rock ratio). Modabberi et al. (Reference Modabberi, Namayandeh, Setti and López-Galindo2019) proposed a shallow aqueous environment with a low water/rock ratio, such as a lagoon, as the aqueous environment in which the eastern Iranian bentonites were formed, which explains the greater mobility of HREEs.