Quantitative clay mineralogy of a Vertic Planosol in southwestern Ethiopia: Impact on soil formation hypotheses

Geoderma 214–215 (2014) 184–196

Contents lists available at ScienceDirect

Geoderma journal homepage: www.elsevier.com/locate/geoderma

Quantitative clay mineralogy of a Vertic Planosol in southwestern Ethiopia: Impact on soil formation hypotheses Mathijs Dumon a,⁎, Alemayehu Regassa Tolossa a,b, Boris Capon c, Christophe Detavernier c, Eric Van Ranst a a b c

Department of Geology and Soil Science (WE13), Ghent University, Krijgslaan 281/S8, B-9000 Ghent, Belgium Department of Natural Resources Management, Jimma University College of Agriculture and Veterinary Medicine, Ethiopia Department of Solid State Sciences (WE04), Ghent University, Krijgslaan 281/S1, B-9000 Ghent, Belgium

a r t i c l e

i n f o

Article history: Received 30 May 2013 Received in revised form 2 September 2013 Accepted 15 September 2013 Available online 5 October 2013 Keywords: Planosol Ethiopia Quantitative Clay mineralogy Ferrolysis Fractionation

a b s t r a c t Planosols, characterised by a bleached, silt-textured surface horizon abruptly overlying a dense, clayey subsoil, are a very common soil type in Ethiopia. The origin of the abrupt textural change is still often debated in literature. One of the processes frequently put forward to explain the coarse textured material in the topsoil is ‘ferrolysis’: an oxidation–reduction sequence driven by bacterial decomposition of soil organic matter, resulting in the destruction of open 2:1 clay minerals. Recent studies of representative profiles of Vertic Planosols in south-western Ethiopia indicate that these soils are composed of a weathered volcanic ash layer deposited on top of a deflated vertic subsoil, which refutes the ferrolysis hypothesis. To strengthen the geogenetic origin of these profiles, a quantitative mineralogical analysis of the clay fraction was undertaken. Results of a sequential fractionation revealed a strong aggregation of clay particles in the bleached horizon, while the effect of aggregation was far more limited in the vertic horizon. This is believed to be related to the dispersed, impregnative nature of iron oxides in the bleached horizon, compared to the segregated nature of the sharp, nodular concretions found in the vertic horizon. The annealing XRD analysis revealed only minor changes in dehydroxylation temperatures of kaolinites and 2:1 minerals between untreated and DCB-treated samples, indicating that the pretreatment did not significantly alter the mineral lattices. Multi-specimen, full-profile fitting of XRD patterns revealed no large quantitative differences between sub-fractions of the bleached and vertic horizons, although a net increase of 1:1 layers over 2:1 layers towards the top of the profile can be observed in the bleached horizon. This could be interpreted as the result of neo-formation of kaolinite. The main mineralogical differences between the bleached and vertic horizons of the b2 μm fraction are mainly a result of the different proportions of sub-fractions. Interestingly, the b0.05 μm fraction seems to be dominated by a complex assemblage of kaolinite and smectite mixed-layer minerals. The obtained detailed view on the mineralogical composition of the clay fraction of a typical Vertic Planosol has provided new insights in the complex nature of these duplex soils, confirming ferrolysis not to be at the origin of the abrupt textural change. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Phyllosilicates are considered among the most interesting mineral phases in soils to understand and study, for a multitude of reasons (e.g. their (trans)formation(s), reactivity, abundance, properties, and function) (Hubert et al., 2012; Lado and Ben-Hur, 2004; Velde and Meunier, 2008). Phyllosilicates are also omnipresent in the clay fraction, which controls the physical and chemical properties of soils to a large extent (Agbenin and Tiessen, 1995; Boivin et al., 2004; Caner et al., 2010; Lado and Ben-Hur, 2004; Molina Ballesteros and Cantano Martin, 2002; Righi et al., 1999). Therefore, accurate, quantitative mineralogical

⁎ Corresponding author at: Laboratory of Soil Science (WE13), Faculty of Sciences, Ghent University, Krijgslaan 281/S8, B-9000 Ghent, Belgium. Tel.: +32 9 264 46 17. E-mail address: [email protected] (M. Dumon). 0016-7061/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.geoderma.2013.09.012

analysis of phyllosilicates is very useful when trying to understand soil formation processes. Because of their inherently small size, X-ray diffraction still is the primary tool for analysing phyllosilicates. However, quantifying them accurately is a difficult task as clay mineral assemblages, especially in soils, are often complex mixtures of several species, each having their own variability in terms of crystal size, morphologies and composition, and all too often, mixed-layer clay minerals are present as well (Caner et al., 2010; Righi et al., 1999). Therefore, in routine analysis, at best a qualitative interpretation can be made, and only the more crystalline, well-ordered clay phases present in soils are positively identified, while mixed-layer minerals are not detected that easily. This has mistakenly led to the conclusion that mixed-layer minerals are not that common in soils. However, recent studies applying a quantitative approach to analyse X-ray diffraction profiles for clay fractions in soils (Caner et al., 2010; Hubert et al., 2009, 2012; Righi et al., 1999) have

M. Dumon et al. / Geoderma 214–215 (2014) 184–196

proven that mixed-layer minerals are much more common in soils than originally thought. The Vertic Planosols in south-western Ethiopia are characterised by a bleached surface horizon with a stagnic colour pattern and periodic water stagnation abruptly overlying a dense, slowly permeable, heavy clay subsoil, and are in that sense typical examples for the Reference Soil Group of Planosols in World Reference Base (Driessen et al., 2001; Van Ranst et al., 2011). The origin of the abrupt textural change is still often debated in literature. Previously a process called ‘ferrolysis’ – an oxidation–reduction sequence driven by chemical energy derived from bacterial decomposition of soil organic matter able to destroy the lattices of open 2:1 clay minerals (Brinkman, 1970) – was held responsible for the formation of these soils. However, a more recent study has proved that these Vertic Planosols are more likely formed by the deposition of a (reworked) ash layer on top of a deflated Vertisol, which refutes the ferrolysis hypothesis (Van Ranst et al., 2011). The study of Van Ranst et al. (2011) focused primarily on chemical, physical and (micro-) morphological analysis and a qualitative mineralogical study using X-ray diffraction and optical microscopy. They concluded that the silt and sand fractions of the Vertic Planosol contain a large amount of amorphous material – identified as being a mixture of volcanic glass and silt-sized phytoliths – as well as quartz, feldspars and iron oxides. No significant differences in relative composition were observed when comparing sand and silt fractions of vertic and bleached horizons. On the other hand the clay fraction exhibits a striking difference in mineralogical composition: the bleached horizon dominated by kaolinite and illite, while the vertic horizon contains significantly more smectite, with lower amounts of kaolinite and illite compared to the bleached horizon. The observed difference in mineralogy is one of the reasons why the ferrolysis process was originally put forward as a ‘probable explanation’ for the formation of this soil. However, since the observations are still qualitative in nature, a quantitative mineralogical analysis of the phyllosilicates in the clay fraction was undertaken to further test the proposed geogenetic origin of these soils. Additionally, Cornelis et al. (2014) proposed that under the current pedological conditions the sources and sinks of silica in the Vertic Planosol of southwestern Ethiopia are partially controlled by neoformation of kaolinite. More detailed mineralogical insights may help to confirm this. Furthermore, this work has tested whether the routinely made dithionite–citrate– bicarbonate (DCB) and H2O2 pretreatments have any detrimental mineralogical effect.

2. Material and methods 2.1. Environmental setting The Gilgel-Gibe catchment in southwestern Ethiopia covers an area of about 5500 km2 and is located in the Oromiya region, Jimma zone about 260 km from Addis Ababa (latitude 7°22′72″–7°34′84″ N; longitude 37°21′05″–37°28′80″ E). The mountainous study area is characterised by steeply incising, V-shaped river valleys in the catchment flanks and fairly flat plains at the centre of the catchment and wider river valleys. The Borè site is located in the middle of this plain some 200 m from the Gibe river, some 30 m above its present alluvial plain. The mean annual rainfall and temperature are 2097 mm and 17.2 °C, respectively. The rainfall is mainly distributed between May and September. The study area is situated on the Ethiopian plateau (1096 to 3259 m a.s.l.), south-west of the rift valley. The outcropping geology of the catchment is dominated by Eocene, Paleocene and possibly Holocene magmatic materials (a.o. basalts, trachytes, rhyolites, tuffs, ignimbrites and ash deposits) and is inherently complex (Tadesse et al., 2003). Lower landscape positions are filled with alluvium and lacustrine sediments.

185

2.2. Materials The sampling site, located in the Borè valley, was carefully selected based on a terrain study using aerial photographs and on-site observations. In general, the Vertic Planosols in the catchment have an abrupt textural change at about 40 cm depth separating the bleached, silty topsoil from a buried, heavy clay Stagnic Vertisol, which exhibits gilgai and well developed slickensides. In this vertic material, wide cracks often form during the dry season in which bleached, topsoil material fall. Bulk samples were taken at fixed depth intervals (of every 5 cm down into the vertic material up to 75 cm below the soil surface, while taking care to sample just above and below the textural break). For a complete description of the soil profile, reference is made to Van Ranst et al. (2011). A subset of 5 samples at depth intervals of 10–15 cm, 20–25 cm, 30–35 cm, 40–45 cm and 70–75 cm was selected for this study. 2.3. Laboratory procedures Soil textural analysis was done on the fine-earth fraction (b 2 mm) of samples after removal of organic matter with H2O2. The sand fraction (2000–63 μm) was separated from the silt and clay fraction by wet sieving, and the silt fraction (63–2 μm) was separated from the clay fraction (b2 μm) by successive sedimentation using repeated syphoning of supernatant clay suspensions after dispersion of clay using Na2CO3. NaCl was used as the flocculating agent. The recovered clay fraction was thoroughly washed to remove excess Cl− (until testing negative with AgNO3), while centrifuging at 3500 rpm after each step. To better characterise the mineralogical composition of the clay fraction, an additional sequential fractionation was done on the b2 μm fraction in order to obtain four sub-fractions (2–0.2 μm, 0.2–0.1 μm, 0.1–0.05 μm and b 0.05 μm) according to the centrifugation method as described in Laird et al. (1991). This was performed on 1 g of the total b2 μm fraction without further pretreatment and on 1 g after treating with 5% H2O2 at 40 °C and DCB (Dahlgren, 1994) in order to remove any remaining organic carbon and iron oxides respectively. To obtain compositional limits (e.g. structural iron), the DCBextractable amounts of Fe, Al, Si and Mn were measured using atomic absorption spectroscopy (AAS), the total elemental composition of the untreated clay fraction (b2 μm) was determined after fusion with Li2CO3 and H3BO3 at 1000 °C and dissolution in HNO3−, and loss of ignition (LOI) was determined as the weight loss after ignition at 850 °C for half an hour. Concentrations were measured using inductively coupled plasma atomic emission spectroscopy (ICP-AES). After sequential fractionation, the total b2 μm fraction and the four sub-fractions for both the untreated and pretreated samples were saturated with Ca2 +. Excess electrolytes were removed by washing twice with deionized water and centrifugation after which they were transferred to a dialysis tube and placed in a beaker with distilled water. Dialysis was continued until no more Cl− could be detected using AgNO3, after which the samples were transferred to a beaker and dried. Oriented samples of all fractions were prepared by transferring a suspension using a pipette on glass slides. The suspension was prepared as such that the surface density of the sample on the glass slide is at least 10 mg/cm3. For each slide an X-ray diffraction pattern is recorded in air-dried and glycolated state on a Philips X'PERT SYSTEM with a PW 3710 based diffractometer equipped with a Cu tube anode, a secondary graphite beam monochromator and a proportional xenon filled detector. The incident beam was automatically collimated. The secondary beam side comprised a 0.1 mm receiving slit, a soller slit, and a 1° anti-scatter slit. The tube was operated at 40 kV and 30 mA, and the XRD data were collected in a θ, 2θ geometry from 3.00′ onwards, at a step of 0.020°2θ, and a count time of 10 s per step. In addition to regular X-ray diffraction, an annealing process was performed on oriented samples prepared in a similar way as described

186

M. Dumon et al. / Geoderma 214–215 (2014) 184–196

above, but using zero-background silicon wafers instead of regular glass slides. The samples were heated in an experimental heating chamber, mounted in a Bruker D8 Discover, dedicated for in-situ X-ray diffraction. Cu K alpha radiation was used as the X-ray source, while a linear Vantec detector monitored the oriented samples. The diffraction pattern was captured every 4 s over a range of 20°2θ during the annealing process, while the temperature was increased at a rate of 0.5 °C/s from room temperature to 900 °C. 2.4. X-ray diffraction profile fitting The multi-specimen, full-profile fitting technique used in this research involves the calculation of a complete XRD pattern using a structural model optimized for each clay species present. The mathematical formalism used for the calculation also allows for the calculation of mixed-layer minerals. The order or disorder in these mixed-layers can be described using Markovian statistics in which a key-concept is the Reichweite (e.g. R0, R1, see Drits and Tchoubar (1990) for details). The computer calculations are based on the algorithm developed initially by Drits and Sakharov (1976) and subsequent developments by Drits and Tchoubar (1990), Drits et al. (1997), Plançon (2002) and References therein. Due to the inherent variability of clay minerals, a single XRD pattern can usually be modelled in more than one way. Sakharov et al. (1999) have therefore proposed a multi-specimen approach, whereby at least two XRD patterns for different saturations or treatments of the same sample are modelled using structural models of which only relevant parameters (e.g. basal spacings and interlayer contents) are changed. Originally, these methods were developed for samples from a burial diagenesis context, however, more recently it has also been applied successfully for the analysis of clay fractions in unconsolidated sediments and soils (Caner et al., 2010; Hubert et al., 2009, 2012; Zeelmaekers et al., 2010). The computer program used for the calculations in this research is called PyXRD and was developed specifically with the multi-specimen method in mind. It allows the simultaneous calculation of multiple saturation states. A copy of the software can be requested from the author. After entering instrumental and experimental parameters such as beam divergences, goniometer radius and sample length, the general work-flow followed when modelling the X-ray diffraction patterns has similarities to a qualitative interpretation. Any rational series of peaks that might be present are first identified and the corresponding minerals are added in the model. After this initial step, sensitive parameters (e.g. average coherent scattering domain size (CSDS) and octahedral iron content) are adjusted to obtain a better fit. At this point, a careful comparison of the differences between the air-dry and glycolated patterns allows to identify the presence of swelling layers and to add the correct (mixed-layer) phase for them, after which sensitive parameters can be again adjusted. For the finer fractions (b 0.1 μm), rational series of reflections become less obvious as the number of mixed-layer phases increases. For these patterns, the information gained from the full-profile fitting of coarser fractions can be used as a starting point for the mixed-layers to be added. 3. Results

Fig. 1. Results of sequential fractionation (A) for untreated samples and (B) for pretreated samples.

fractions between the bleached and vertic horizon has been observed. In general, for both pretreated and untreated samples, the coarser 2−0.2 μm fraction is more dominant in the bleached horizon compared to the vertic horizon. On average, in the bleached horizon the coarse 2–0.2 μm fraction accounts for 80% when untreated and for 48% after pretreatment. In the vertic horizon this fraction represents only 12% when untreated and a slightly lower 8.5% when pretreated. In both horizons, regardless of treatments, the intermediate size fractions (0.2−0.1 and 0.1–0.05 μm) represent a small part of the bulk b2 μm fraction. On average, the 0.2–0.1 μm fraction accounts for 4.1% in untreated and 5.3% in pretreated samples and the 0.1–0.05 μm fraction accounts for 5.2% in untreated and 6.8% in pretreated samples. The very fine b 0.05 μm fraction represents 13% before pretreatment which increases to 40% after pretreatment. In the vertic horizon the very fine b 0.05 μm fraction accounts for 74% before pretreatment and for a slightly higher 78% after pretreatment.

3.1. Sequential fractionation 3.2. Total elemental analysis and specific extractions The measured relative weights of the sequential fractionation of the b2 μm fraction for several intervals of the profile are presented in Fig. 1, after normalizing to 100%. It is worth mentioning here that the use of a dialysis method instead of regular washing allowed obtaining an almost perfect recovery rate: on average 99.4% (σ = 0.3). This is a significant improvement compared to previous attempts (e.g. 83% in Hubert et al. (2012)). Within each horizon the variations in relative proportions remain limited, however a marked difference in proportions of the different

Results from the total elemental analysis and DCB-extraction of the untreated Na-saturated b 2 μm samples are presented in Table 1. When comparing the DCB-extracted amounts in the b2 μm fraction with values obtained for the fine earth (b2 mm), it can be observed that Fed is concentrated in the finer fractions in the bleached horizon, especially just above the textural break (30–35 cm interval), while in the vertic horizon the Fed contents in the b2 μm fraction are lower compared to the b2 mm fraction.

M. Dumon et al. / Geoderma 214–215 (2014) 184–196 Table 1 Total elemental composition and loss on ignition (LOI) of untreated Na+-saturated clay fraction and DCB-extractable content (expressed as wt.% oxides) of the clay and fine earth fraction. Untreated b 2 μm (%) Hor.

E

E

E

Bi

Bi

Depth (cm)

10–15

20–25

30–35

45–50

70–75

SiO2 Al2O3 Fe2O3 MnO MgO CaO Na2O K2O TiO2 P2O5 LOI Total

55.24 17.04 5.97 0.03 0.52 0.13 1.61 1.44 2.35 0.11 13.97 98.41

63.94 14.40 5.62 0.03 0.46 0.11 1.10 1.80 3.01 0.08 9.39 99.94

63.51 11.85 9.68 0.03 0.45 0.13 1.31 1.89 3.07 0.07 8.07 100.07

49.95 19.60 12.08 0.12 0.94 0.12 2.56 1.15 1.13 0.04 12.70 100.40

45.84 16.96 9.47 0.02 0.92 0.03 7.39 1.12 1.02 0.03 17.03 99.83

DCB extractable b 2 μm (%) 0.60 SiO2 0.39 Al2O3 Fe2O3 1.04 MnO 0.01

0.57 0.28 1.35 0.00

0.86 0.31 4.62 0.01

1.22 0.41 1.59 0.08

1.31 0.37 1.13 0.10

DCB extractable b 2 mm (%) 2.59 SiO2 0.49 Al2O3 1.19 Fe2O3 MnO 0.05

1.75 0.45 1.23 0.01

2.63 0.74 2.20 0.01

1.22 0.58 4.68 2.38

3.66 0.39 3.03 0.99

3.3. In-situ XRD annealing Both the pretreated and untreated b2 μm fractions were analysed using high-temperature in-situ XRD analysis of oriented samples. The pretreated sample from the 10–15 cm interval did not produce an interpretable result due to severe cracking and curling upon heating, and was therefore excluded. An overview of observed reaction temperatures is given in Table 2. Only the inversion from α to β-quartz at ca. 575 °C (shift of the 0.426 nm peak to lower angle, most clearly visible in the samples from the bleached horizon) is excluded from the table. In the samples from the bleached horizon (Fig. 2) only one dehydration reaction could be identified, with a minimal effect on the X-ray diffraction patterns around 60 °C continuing up to 100 °C. For the samples of the vertic horizon (Fig. 2) an additional dehydration reaction, probably related to more strongly held interlayer-water being removed, could be identified starting at about 200 °C and continuing up to 250 °C. After dehydration of these samples, often a reflection intermediate between 1.0 and 0.73 nm or an increase of the low-angle shoulder of the kaolinite peak can be observed, tentatively interpreted as caused by the presence of some kaolinite–smectite interstratification. This interpretation is corroborated by the results from the full-profile fitting (see further).

Table 2 Temperatures at which reactions took place for untreated and pretreated samples from different depth intervals by using in-situ HT-XRD. Hor.

E E E Bi Bi E E Bi Bi

Interval (cm)

Treatment

10–15 20–25 30–35 45–50 70–75 20–25 30–35 45–50 70–75

Untreated Untreated Untreated Untreated Untreated Pretreated Pretreated Pretreated Pretreated

Observed reaction temperatures (°C) Dehydration

Dehydroxylation 1

Onset dehydroxylation 2

60–100 60–100 60–100 60–(250) 60–250 60–100 60–170 60–170 60–250

475–550 475–550 475–550 450–525 450–525 425–525 330–400 350–500 350–500

630 630 630 630 640 650 660 650 640

187

In the medium temperature range (300 up to 700 °C) two changes can be observed, interpreted as being the result of dehydroxylation. The first dehydroxylation occurred in a temperature range of 450–550 °C for the untreated samples and is linked with the presence of kaolinite layers. In the pretreated samples this reaction appears to be starting at a lower temperature, the lowest being recorded in the bleached horizon, just above the textural break with the vertic horizon (depth interval of 30–35 cm). The low dehydroxylation temperatures for kaolinites also suggest their rather poor crystallinity. Probably the pretreatment has somewhat destabilized these kaolinites. It is possible that the pretreatment has destabilized these kaolinites somewhat. The second dehydroxylation is observed from 630 to 650 °C onwards and is linked with the presence of smectite and/or illite layers. 3.4. X-ray diffraction qualitative description Overviews of the XRD patterns for all fractions of the 20–25 cm and 70–75 cm depth intervals after pretreatment are given in Figs. 3 and 4. The X-ray diffraction pattern for the bulk b 2 μm fraction contains reflections that can be attributed to both clay minerals as well as non-clay minerals like quartz (e.g. reflections at 0.426, 0.335, 0.246, 0.229 and 0.213 nm) and feldspars (e.g. reflections at 0.65 and 0.325 nm). The clay minerals readily identified by their rational series of reflections are kaolinite (0.716, 0.358 and 0.238 nm) and illite (1.00, 0.500 and 0.333 nm). The broad, relatively weak reflection around 1.50 nm in air-dry (AD) conditions and around 1.80 nm after ethylene-glycol solvation (EG) are related to the presence of smectite layers. The presence of some mixed-layer minerals can be deduced when considering the change in intensity of the 1.00 nm and the 0.716 nm peak, and the broad reflection between these two peaks before and after glycolation. Based on these observations alone, both kaolinite–smectite and illite– smectite mixed layers might be present. Quartz, feldspars and discrete clay minerals (kaolinite and illite) are concentrated in the coarser sub-fractions (N0.1 μm), which is a common observation for clay fractions in soils (Hubert et al., 2012). On the other hand, X-ray diffraction patterns for the finest fractions (b0.1 μm) show broadened reflections and reflection maxima deviating from the positions expected for discrete minerals and an increase in the differences between AD and EG states. This indicates that the finer fractions contain a higher amount of mixed-layer minerals and swelling layers, and their mineralogy is more complex compared to the one of the coarser fractions. When comparing the diffraction patterns of untreated with those of pretreated samples, the differences appear to be small, but can be noticed nonetheless. For instance, in the coarsest 2–0.2 μm fraction, a clear decrease in the intensity for the smectite reflection can be observed when compared with the kaolinite reflection. Additionally, the orientation of clay minerals appears to be better after the pretreatment, considering the smaller hk0 bands. An additional Li+-saturation test has been applied on the b0.05 μm fraction of untreated and pretreated samples from the 70–75 cm interval to gain information on what type of smectites are present. Similarly a formamide treatment (Churchman et al., 1984) was applied on the pretreated b0.05 μm fraction from the 70–75 cm interval to gain information on what type of 1:1 layers are present. Results for both tests are presented in Fig. 5. It can be clearly observed that the Li+-saturation does not cause the irreversible loss of charge in the swelling layers, indicating that charge is present in both tetrahedral and octahedral sheets. Formamide intercalation did not shift the 00l basal spacing of the 1:1 minerals in the b 0.05 μm fraction. This is an indication that the 1:1 layers are truly kaolinite-like and not dehydrated halloysites. 3.5. Full profile fitting Only the coarse (2–0.2 μm) and very fine (b0.05 μm) fractions were selected for full profile fitting, because (i) the intermediate fractions

188

M. Dumon et al. / Geoderma 214–215 (2014) 184–196

Fig. 2. 2D representation of the in-situ XRD analysis for the pretreated and untreated samples from the 20–25 cm (bleached horizon) and 70–75 cm interval (vertic horizon). Spacings at the bottom are in nm.

Fig. 3. Observed X-ray diffraction patterns for the bulk, pretreated b 2 μm fraction and sub-fractions of the 20–25 cm interval of the bleached horizon. Black and grey traces are patterns for air-dry conditions and after glycolation respectively. Quartz peak at 0.334 nm is clipped for clarity. An * indicates hk0 bands, open triangles are accessory quartz and feldspar reflections and a diamond indicates goethite reflections. Relative mass (wt.%) is indicated at the right of each sub-fraction pattern. Spacings of important 00l reflections are in nm.

M. Dumon et al. / Geoderma 214–215 (2014) 184–196

189

Fig. 4. Observed X-ray diffraction patterns for the bulk, pretreated b2 μm fraction and sub-fractions of the 70–75 cm interval of the vertic horizon. Legend is identical to Fig. 3.

represent less than 20 wt.% of the total b2 μm fraction and (ii) their XRD patterns appear to be intermediate between those of the coarse (2–0.2 μm) and very fine (b0.05 μm) fractions (Figs. 3 and 4). Therefore, the additional information that can be obtained from modelling these fractions will be very limited. Modelling the 2–0.2 μm fractions proved to be a fairly easy task by following the workflow presented in the Methods section. However, the kaolinite peaks in this fraction are slightly asymmetric, with a low-angle ‘tail’. Commonly, this is interpreted as the result of a bi-modal CSDS distribution, and has been modelled using two kaolinite phases, well crystalline (WC) and poorly crystalline (PC), having different CSDS's (Hubert et al., 2009). However, for the b 0.05 μm fractions, the modelling proved not to be as easy, as rational series of reflections are absent in the XRD patterns. Therefore, it can be assumed that only mixed-layer minerals are present. Since not all reflections shift after glycolation (e.g. in the 10–12°2θ region), some non-swelling mixed layers must be present, composed of 2:1 illite-like layers and 1:1 kaolinite-like layers. Reflections that do shift after glycolation indicate smectite-containing mixed-layers. Based on these qualitative interpretations, it was chosen to add three types of (mixed-layer) phases: (i) R1 kaolinite/non-expanding 2:1 mixed layers with partial segregation (KI R1), (ii) R0 kaolinite/smectite mixed-layers (KSS R0) and (iii) R0 smectite-containing mixed-layers having different hydration and glycolation states (SSS R0). Initially only one phase was added for each of these mixed-layer types. Using this initial setup, the general shape of the observed pattern could be approximated, but either the positions or the shape of the peaks would not coincide very well. After this setback, an attempt was made to improve the fit by adding other types of mixed-layer minerals (e.g. containing Al-hydroxy interlayered smectite layers), but this proved unsuccessful (not shown). Considering the first attempt approximated the observed XRD patterns it was assumed that the variability within the samples was not yet adequately described. Therefore, three additional variations of the original mixed-layer types were added to the initial model. This proved to be a successful approach for both the untreated and pretreated samples. In addition, after further parameter refinement,

two of the additional mixed-layers in all b0.05 μm fractions of untreated samples (figures not shown) and one mixed-layer in the b0.05 μm fraction of pretreated samples (Figs. 8 and 9) from the vertic horizon could be eliminated. An overview of the phases and their parameters used in the model is presented in Tables 3 to 6. Fig. 10 gives an overview of the changes in mineralogy for the two fractions and their weighted average. Discrete kaolinite (sum of both kaolinite WC and kaolinite PC) and illite make up an important part of the coarser 2–0.2 μm fraction: in the bleached horizon on average 37 wt.% and 46 wt.% for untreated and pretreated samples respectively, and in the vertic horizon 57 wt.% and 58 wt.% for untreated and pretreated samples respectively. For untreated samples, discrete kaolinite contents for vertic and bleached horizons do not differ: 15 wt.% on average for both. However, for pretreated samples, kaolinite contents decrease from the bleached to the vertic horizon: from 27 to 17 wt.%. Illite contents are similar for untreated and pretreated samples but do increase from the bleached to the vertic horizon: from 20 to 41 wt.%. In the untreated samples, the mixed layer phases comprise an R0 kaolinite rich, poorly expandable kaolinite/smectite (KSSS R0), an R0 kaolinite poor, well-expandable kaolinite/smectite (KSS R0) and an R0 illite/smectite rich in illite. The mineral assemblage for the pretreated samples is identical, except for the absence of the R0 kaolinite rich, poorly expandable kaolinite/smectite mixed layer (KSSS R0). The latter mixed-layer is characterised by a broad, shoulder-like reflection between the kaolinite and illite first order reflections (7–13°2θ, figures not shown), remaining largely unaffected by glycolation, hence the term poorly expandable. This shoulder-like reflection is no longer present in the pretreated samples (Figs. 6 and 7), indicating the absence of that phase in those samples. Its contents decrease from ~40 wt.% in the bleached horizon to ~31 wt.% in the vertic horizon. The R0 kaolinite poor, well-expandable kaolinite/smectite (KSS R0) is characterised by a broad reflection in the 5–9°2θ range, with a long tail towards higher angles in air-dry state. It clearly shifts and splits in two reflection maxima upon glycolation (Fig. 6 and 7). Its contents decrease slightly from the bleached horizon (on average 15 and 28 wt.% for

190

M. Dumon et al. / Geoderma 214–215 (2014) 184–196

Fig. 5. Top: Observed XRD patterns for the b0.05 μm fraction of the 70–75 cm interval for both untreated and pretreated samples, after Li+-saturation and heating for 24 h at 300 °C (bottom patterns) and after subsequent glycolation (top patterns). Bottom inset: observed XRD patterns of smear mounts using water (grey line) and formamide (black line) for the b0.05 μm fraction of pretreated samples from the 70−75 cm interval.

untreated and pretreated samples respectively) towards the vertic horizon (on average 10 and 20 wt.% for untreated and pretreated samples respectively). The R0 illite/smectite mixed layer (ISS R0) explains the low-angle shoulder of the illite peak and the minimal shift upon glycolation. Its contents also decrease from the bleached horizon (on average 7 and 26 wt.% for untreated and pretreated samples respectively) towards the vertic horizon (on average 3 and 23 wt.% for untreated and pretreated samples respectively). As explained, the b0.05 μm fraction contains 4 mixed-layer minerals in the untreated samples, while the pretreated samples show the presence of one or two additional mixed-layers. This higher amount of phases needed to model the b0.05 μm fraction of the pretreated compared to the untreated samples, can be interpreted as a result of a more variable mineralogy. This is in turn a result of the pretreatments, which break down micro-aggregates into finer particles, as is also shown by the increase in the amount of very fine particles (b 0.05 μm) and the decrease in the amount of coarse particles (2–0.2 μm) before and after pretreatment. Both pretreated and untreated samples contain an R0 smectite phase which can be considered ‘well-expandable’ (SSS WE), as shown by the broad reflections at about 1.50 nm, 0.53 nm

and 0.32 nm shifting to 1.70 nm, 0.93 nm, 0.58 nm, and 0.34 nm after glycolation (Figs. 8 and 9). The pretreated samples have an additional poorly-expandable smectite phase (SSS PE) which has somewhat broader reflections at about the same positions. Both untreated and pretreated b 0.05 μm fractions contain two kaolinite/smectite mixedlayers. One of these is kaolinite-poor and well-expandable (KSS R0), while the other is kaolinite rich and poorly expandable (KSSS R0). This difference in kaolinite content is reflected in their XRD profiles: the kaolinite-rich mixed-layer has a broad reflection with a maximum around 0.76 nm shifting slightly after glycolation, while the kaolinitepoor mixed-layer has a broad, smeared region of higher intensity between 5° and 14°2θ with 2 maxima at positions intermediate between those for pure kaolinite and smectite, which clearly shift after glycolation (Figs. 8 and 9). Two R1 kaolinite/non-expandables are present in the pretreated samples: one with slightly higher kaolinite content and higher segregation (KI R1 A), and one with slightly lower kaolinite content, lower segregation and a lower CSDS (KI R1 B). The latter is only present in the bleached horizon. Both phases have reflections in the same regions (between 7° and 13°2θ and at about 25°2θ, Figs. 8 and 9), but logically, for the more segregated

M. Dumon et al. / Geoderma 214–215 (2014) 184–196

191

Table 3 Phases and parameter values used for modelling XRD patterns of untreated 2–0.2 μm samples. Rp AD/EG: Rp-factor (residual error) of the air-dried and glycolated patterns respectively; N: mean coherent scattering domain size expressed as number of layers; FeVI: percentage of octahedral sites occupied by iron. Mixed-layer notations are as follows: K: kaolinite content, I: illite content, S: smectite content (2w: bi-hydrated, 1w: mono-hydrated, 0w: dehydrated, 2 g: two layers of glycol, 1 g: one layer of glycol), Nexp: non-expandable 2:1 layer. Untreated samples

Depth interval

10–15 cm

20–25 cm

30–35 cm

40–45 cm

70–75 cm

Horizon

E

E

E

Bi

Bi

Rp AD/EG (%)

6.4/5.3

6.9/6.8

8.1/7.7

8.8/9.2

8.8/8.7

N FeVI N N N I/S2w/S1w I/S2g/S1g FeVI I/S N K/S2w/S1w K/S2g/S1g FeVI K/S N K/S2w/S1w/S0w K/S2g/S1g/Nexp FeVI K/S

14 0% 6 20 8 90/9/1 90/8/2 25%/25% 15 40/21/39 40/27/33 0%/12.5% 18 55/9/11/25 55/9/22/14 0%/12.5%

14 0% 6 20 8 90/9/1 90/8/2 25%/25% 15 50/25/25 50/25/25 0%/12.5% 18 55/9/14/22 55/9/29/7 0%/12.5%

13 0% 7 20 9 90/9/1 90/9/1 25%/25% 15 45/33/22 45/39/16 0%/12.5% 18 55/9/14/22 55/9/29/7 0%/12.5%

13 0% 8 20 9 84/8/8 84/8/8 25%/25% 15 45/27/28 45/39/16 0%/12.5% 18 55/9/14/22 55/9/29/7 0/12.5%

13 0% 8 20 9 84/8/8 84/8/8 25%/25% 15 40/27/33 40/45/15 0%/12.5% 18 55/9/14/22 55/9/29/7 0%/12.5%

Phases Illite Kaolinite, poorly crystalline (PC) Kaolinite, well crystalline (WC) Illite–smectite (ISS) R0

Kaolinite–smectite (KSS) R0 ‘kaolinite-poor, well-expandable’

Kaolinite–smectite–Nexp (KSSS) R0 ‘kaolinite-rich, poorly-expandable’

phase (KI R1 A) the peaks are less broad and have two maxima related to the two different layer types, while these maxima are merged and broader for the more random, lower-CSDS phase. In the untreated samples a single kaolinite/non-expandable is present which has a relatively high CSDS, but lower kaolinite content and is not as segregated as its high-CSDS counterpart in the pretreated samples. As mentioned, the fit of the model for the b 0.05 μm fraction is not perfect. The reliability of the absolute weight percentages obtained is therefore questionable and a detailed discussion of the exact changes for each phase with depth would not be very useful. On the other hand, discussing general trends and observations qualitatively remains a valid approach. One of these observations is that the b0.05 μm fraction is dominated by kaolinite/smectite. Another observation is a constant decrease in the amount of kaolinite-bearing phases in both untreated and pretreated samples and the concomitant increase in the amount of smectite phases. When comparing the untreated with the pretreated samples for the b 0.05 μm fraction, a marked difference in mineralogy can be observed in the upper part (5–10 and 15–20 cm intervals) of the bleached horizon, due to the increase in the amount of kaolinite/ smectite and a decrease in the amount of pure smectite mixed-layers. For the lower part of the bleached horizon and the vertic horizon,

the mineralogical composition seems similar for both untreated and pretreated samples, albeit more complex for the pretreated samples. 4. Discussion Full-profile fitting using a multi-specimen approach has been applied successfully on soil materials in this research. However, it should be noted that it was difficult to obtain even a moderate fit for the b 0.05 μm fraction. This is a shortcoming of the model, probably because it failed to adequately describe all types of disorder present in the b0.05 μm fraction. Because of the importance of this fraction in terms of reactivity and explaining soil behaviour, improving the full-profile fitting approach of such fractions should be a future research topic. Since pretreating samples improved particle orientation and did not change the mineralogy of the samples, it could be concluded this has a positive effect. However, the in-situ high temperature XRD data led to a more nuanced conclusion. The difference in temperature between untreated and pretreated samples at which kaolinites start to dehydroxylate is small, but consistent. The weakening of the 0.5 and 0.35 nm reflections above 800 °C in pretreated samples compared to untreated samples from the vertic horizon, and the absence of these

Table 4 Phases and parameter values used for modelling XRD patterns of pretreated 2–0.2 μm samples. Rp AD/EG: Rp-factor (residual error) of the air-dried and glycolated patterns; N: mean coherent scattering domain size expressed as number of layers; FeVI: percentage of octahedral sites occupied by iron; Mixed-layer notations as in Table 3. Pretreated samples

Depth interval

10–15 cm

20–25 cm

30–35 cm

40–45 cm

70–75 cm

Horizon

E

E

E

Bi

Bi

Rp AD/EG (%)

7.6/7.6

9.2/8.2

10.0/9.7

9.7/10.1

8.9/9.2

N FeVI N N N I/S2w/S1w I/S2g/S1g FeVII/S N K/S2w/S1w K/S2g/S1g FeVI K/S

20 0% 10 30 9 95/4/1 95/4/1 25%/25% 5 65/7/28 65/14/21 0%/0%

20 0% 10 30 9 95/4/1 95/4/1 25%/25% 5 65/7/28 65/14/21 0%/0%

20 0% 10 30 10 95/4/1 95/4/1 25%/25% 5 65/7/28 65/14/21 0%/0%

26 0% 10 30 10 94/3/3 94/5/1 25%/25% 5 50/22/28 50/40/10 0%/0%

26 0% 10 30 10 94/3/3 94/5/1 25%/25% 5 50/22/28 65/42/8 0%/0%

Phases Illite Kaolinite, poorly crystalline (PC) Kaolinite, well crystalline (WC) Illite–smectite (ISS) R0

Kaolinite–smectite (KSS) R0

192

M. Dumon et al. / Geoderma 214–215 (2014) 184–196

Table 5 Phases and parameter values used for modelling XRD patterns of untreated b0.05 μm samples. Rp AD/EG: Rp-factor (residual error) of the air-dried and glycolated patterns; N: mean coherent scattering domain size expressed as number of layers; SG1: degree of segregation (R1 mixed layers); FeVI: percentage of octahedral sites occupied by iron; δd001: variation in basal spacing; Mixed-layer notations as in Table 3. Untreated samples

Depth interval

10–15 cm

20–25 cm

30–35 cm

40–45 cm

70–75 cm

Horizon

E

E

E

Bi

Bi

Rp AD/EG (%)

6.9/8.6

6.9/7.2

8.7/9.9

9.0/10.9

8.6/10.5

Rp AD/EG (%)

6.9/8.6

6.9/7.2

8.7/9.9

9.0/10.9

8.6/10.5

N K/I SG1 FeVI K/I δd001 K/I N K/S2w/S1w K/S2g/S1g FeVI K/S δd001 K/S2/S1 N K/S2w/S1w/S0w K/S2g/S1g/Nexp FeVI K/S δd001 K/S2/S1/S0 N S2w/S1w/S0w S2g/S1g/Nexp FeVI δd001 S2/S1/S0

6 70/30 0.429 –/50% .005/.005 4 60/0/40 60/20/20 –/50% .005/.002/.002 6 65/10/4/21 65/10/4/21 –/12.5% .001/.001/.001/.001 4 47/21/32 70/12/18 20% .02/.02/.005

6 70/30 0.429 –/50% .005/.005 4 60/0/40 60/20/20 –/50% .005/.002/.002 6 65/10/4/21 65/10/4/21 –/12.5% .001/.001/.001/.001 4 47/21/32 70/12/18 30% .02/.02/.005

6 70/30 0.429 –/50% .005/.005 4 55/0/45 55/22/23 –/50% .005/.002/.002 6 70/8/4/18 70/8/4/18 –/12.5% .001/.001/.001/.001 4 50/15/35 75/15/10 30% .02/.02/.005

6 60/40 0.533 –/50% .0005/.005 4 50/0/50 50/25/25 –/50% .005/.002/.002 6 80/5/3/12 80/5/3/12 –/12.5% .001/.001/.001/.001 3 60/8/32 75/10/15 30% .02/.02/.005

6 60/40 0.533 –/50% .0005/.005 4 50/0/50 50/28/22 –/50% .005/.002/.002 6 80/5/3/12 80/5/3/12 10%/30% .001/.001/.001/.001 3 60/8/32 75/10/15 30% .02/.02/.005

Phases Kaolinite–illite (KI) R1

Kaolinite–smectite (KSS) R0 ‘kaolinite-poor, well-expandable’

Kaolinite–smectite–Nexp (KSSS) R0 ‘kaolinite-rich, poorly-expandable’

Well-expandable smectite (SSS WE) R0

Table 6 Phases and parameter values used for modelling XRD patterns of pretreated b0.05 μm samples. Rp AD/EG: Rp-factor (residual error) of the air-dried and glycolated patterns respectively; N: mean coherent scattering domain size expressed as number of layers; SG1: degree of segregation (R1 mixed layers); FeVI: percentage of octahedral sites occupied by iron; δd001: variation in basal spacing; Mixed-layer notations as in Table 3. Pretreated samples

Depth interval

10–15 cm

20–25 cm

30–35 cm

40–45 cm

70–75 cm

Horizon

E

E

E

Bi

Bi

Rp AD/EG (%)

7.6/8.6

7.5/7.7

7.4/8.3

8.1/9.8

8.0/9.8

N K/I SG1 FeVI K/I δd001 K/I N K/I SG1 FeVI K/I δd001 K/I N K/S2w/S1w K/S2g/S1g FeVI K/S δd001 K/S2/S1 N K/S2w/S1w/S0w K/S2g/S1g/Nexp FeVI K/S δd001 K/S2/S1/S0 N S2w/S1w/S0w S2g/S1g/Nexp FeVI δd001 S2/S1/S0 N S2w/S1w/S0w S2g/S1g/Nexp FeVI δd001 S2/S1/S0

6 80/20 0.625 –/50% .002/.002 4 68/32 0.412 –/50% 0.003/0.002 4 60/4/36 60/28/12 –/50% .005/.0005/.002 6 75/9/6/10 75/9/6/10 –/12.5% .0005/.001/.001/.002 4 50/3/47 60/0/40 25% .005/.005/0.004 4 70/18/12 70/18/12 – .003/.003/.001

6 80/20 0.625 –/50% .002/.002 4 68/32 0.412 –/50% 0.003/0.002 4 60/4/36 60/28/12 –/50% .005/.0005/.002 6 75/9/6/10 75/9/6/10 –/12.5% .0005/.001/.001/.002 4 50/3/47 60/0/40 25% .005/.005/.004 4 70/18/12 70/18/12 – .003/.003/.001

6 80/20 0.625 –/50% .002/.002 4 68/32 0.412 –/50% 0.003/0.002 4 60/4/36 60/28/12 –/50% .005/.0005/.002 6 75/9/6/10 75/9/6/10 –12.5% .0005/.001/.001/.002 4 50/3/47 60/0/40 25% .005/.005/.004 4 70/18/12 70/18/12 30% .005/.005/.004

6 75/25 0.733 –/50% .002/.002 – – – – – 4 50/5/45 50/25/25 –/50% .005/.0005/.002 6 70/11/7/12 70/11/7/12 –/12.5% .0005/.001/.001/.002 3 50/3/47 70/0/30 40% .005/.005/.004 4 75/10/15 90/5/5 30% .005/.005/.004

6 75/25 0.733 –/50% .002/.002 – – – – – 4 50/5/45 50/25/25 –/50% .005/.0005/.002 6 70/11/7/12 70/11/7/12 –/12.5% .0005/.001/.001/.002 3 50/3/47 70/0/30 40% .005/.005/.004 4 75/10/15 90/5/5 30% .005/.005/.004

Phases Kaolinite–illite (KI) R1 A

Kaolinite–illite (KI) R1 B

Kaolinite–smectite (KSS) R0 ‘kaolinite-poor, well-expandable’

Kaolinite–smectite–Nexp (KSSS) R0 ‘kaolinite-rich, poorly-expandable’

Poorly-expandable smectite (SSS PE) R0

Well-expandable smectite (SSS WE) R0

M. Dumon et al. / Geoderma 214–215 (2014) 184–196

193

Fig. 6. Observed (thick black line), calculated (grey line) and residual patterns for the air-dry (AD, left) and glycolated (EG, right) states of the 2–0.2 μm fraction from pretreated samples for the 20–25 cm interval and for each separate phase. Symbols as in Fig. 3.

differences in the bleached horizon indicate that fine smectite crystals, dominantly present in the vertic horizon, are affected more intensely by the pretreatment, while coarser illites, more abundant in the bleached horizon, remain mostly unaffected. All these observations indicate that the pretreatment can slightly destabilize kaolinites and smectites, but large mineralogical differences were not detected. The large range in dehydroxylation temperatures of the 2:1 phyllosilicates observed by in-situ XRD indicates a large compositional variability. This is not surprising, as vertic soil materials are commonly thought to have been formed by a combination of in-situ crystallisation of smectites from soil solutions and by down-slope transport of colloids into the alluvial or lacustrine lowlands (Driessen et al., 2001). These processes are probably still taking place, and, combined with biological activity and active churning by the vertic horizon, are slowly changing the texture of the silty bleached horizon to a more clayey texture, until the difference between both horizons becomes negligible. The pretreatment with H2O2 and DCB significantly alters the textural composition of the clay fraction of the bleached horizon, resulting in an increase of the very fine particle size. This can be attributed to the breakdown of micro-aggregates during the removal of organic matter and iron, as was also observed by other workers (e.g. Barberis et al., 1991). However, the effect of the pretreatment on aggregate disintegration of the vertic material is limited. This indicates that the aggregation within the clay fraction of the vertic horizon by iron and organic matter is much more limited. The observed differences in DCB-extractable iron for ) and clay fractions (Feclay fine earth (Fecoarse d d ) indicate that in the bleached horizon free iron oxides occur mainly in the fine fractions N Fecoarse ), while in the vertic horizon they are concentrated in (Feclay d d b Fecoarse ). Micro-morphology (Van Ranst the coarser fractions (Feclay d d et al., 2011) indicated rounded iron–manganese nodules with sharp boundaries in the vertic horizon and more impregnative, diffuse nodules occurring in the bleached horizon. This indicates that differences in environmental and textural conditions of both horizons have

influenced micro-aggregation and (macro-)structure respectively: the combination of churning and lower biological activity in the vertic horizon leads to poorer micro-aggregation but a strong, angular-blocky structure, while alternating wet and dry conditions, related changes in oxidation potential and higher biological activity in the bleached horizon contribute to a strong micro-aggregation and a looser, granular structure. The quantification of the clay mineralogy of sub-fractions has shown that changes with depth within each sub-fraction are limited and the mineralogical assemblages are almost identical (Fig. 9). The observed differences in mineralogy of the total b 2 μm fraction between the bleached and vertic horizons are therefore mainly attributed to the different proportions of the sub-fractions. However, a change with depth within the bleached horizon can still be observed. This is the result of an increase with depth of the illite content and the concomitant decrease of the kaolinite/smectite contents in the 2–0.2 μm fraction, and the decrease with depth of the segregated kaolinite/non-expandables (KI R1) contents and the concomitant increase of the smectite content within the b0.05 μm fraction. The end result is an increase of 1:1 layers over 2:1 layers in both fractions towards the top of the profile. Despite the similar mineralogical assemblages of the bleached and vertic horizon's subfractions, these observations contradict the ferrolysis process, often considered responsible for the destruction of open 2:1 phyllosilicates and the formation of Planosols. Smectites are still present in the bleached horizon, and ferrolysis is therefore highly improbable. It is also interesting to note the similarities of the sub-fractions considering the proposed geogenetic origin of the bleached horizon. This is an indication that the compositions of the parent materials of the vertic horizon are not that different from the composition of the bleached horizon's parent material. Since both are volcanic, this indicates that the source of volcanism is similar for both. However, the overall changes with depth within the bleached horizon indicate that some other process is influencing the mineralogical

194

M. Dumon et al. / Geoderma 214–215 (2014) 184–196

Fig. 7. Observed (thick black line), calculated (grey line) and residual patterns for the air-dry and glycolated states of the b0.05 μm fraction from pretreated samples for the 20–25 cm interval and for each separate phase. Symbols as in Fig. 3.

Fig. 8. Observed (thick black line), calculated (grey line) and residual patterns for the air-dry and glycolated states of the 2–0.2 μm fraction from pretreated samples for the 70–75 cm interval and for each separate phase. Symbols as in Fig. 3.

M. Dumon et al. / Geoderma 214–215 (2014) 184–196

195

Fig. 9. Observed (thick black line), calculated (grey line) and residual patterns for the air-dry and glycolated states of the b0.05 μm fraction from pretreated samples for the 70–75 cm interval and for each separate phase. Symbols as in Fig. 3.

Fig. 10. Overview of the obtained fractions for each phase in each sub-fractions (top four) and the weighted averages for groups of minerals for the bulk b2 μm fraction (bottom two).

196

M. Dumon et al. / Geoderma 214–215 (2014) 184–196

composition of this horizon. Possible explanations include physical processes, such as selective, downward migration of fines, or chemical processes, such as differences in pedo-environment, influencing the type of minerals formed. It has also been suggested that with increased drainage, kaolinite/smectite mixed layers are progressively transformed into kaolinites and iron oxides by dissolution and recrystallisation (Righi et al., 1999; Vingiani et al., 2004). Even though the Vertic Planosols occur in very level landscape positions, the bleached horizon has a good lateral drainage. Only at local depression (e.g. induced by the gilgai relief of the underlying vertic horizon) drainage and thus kaolinite formation, could be impeded. Considering the above, the previously postulated neo-formation of kaolinite at the top of the profile (Cornelis et al., 2014) seems indeed very probable and, despite not knowing the exact formation mechanism at this point, explains the observed mineralogical changes with depth in the bleached horizon. The full-profile fitting allowed identification of the large quantities of kaolinite/smectite and kaolinite/non-expandable mixed layers minerals, especially in the very fine b0.05 μm fraction. The occurrence of mixedlayers of 1:1 and 2:1 layer silicates in soils is not widely reported, but might be present more commonly than currently appreciated (Wilson, 1999). There have been a number of publications on the mechanisms involved in the smectite to kaolinite transformation (Cuadros and Dudek, 2006; Ryan and Huertas, 2009; Wilson, 1999), and the evolution has been well illustrated in the so-called tropical ‘red–black soilscapes’ (Bühmann and Grubb, 1991; Vingiani et al., 2004), a term which also applies to this study area. It seems that the association of red–black soilscapes with kaolinite/smectite is worth further investigation. 5. Conclusions This research has demonstrated that, although time-consuming, multi-specimen full-profile fitting can be successfully applied to soil samples. It has also shown that (i) pretreating clay fractions with DCB and H2O2 can slightly destabilize kaolinites and smectites, but large mineralogical differences are not observed and (ii) the current model cannot completely account for the disorder present in very small crystallites. Further research would be needed here. The Vertic Planosol in the Gilgel-Gibe catchment in south-western Ethiopia is characterised by a complex clay mineralogy. The bleached horizon is characterised by a high micro-aggregation by iron oxides and organic matter and a mineralogy composed of a mixture of coarser kaolinites and illites and finer kaolinite/smectite mixed-layers, while the vertic horizon is less micro-aggregated and dominated by large amounts of fine clay composed of smectites and kaolinite/smectite. Indications for the neo-formation of kaolinite towards the top of the bleached horizon have also been found. These findings are in good agreement with micro-morphology and indicate that environmental conditions strongly influence micro-aggregation. Despite the similarities in mineralogical assemblages in each subfraction, the presence of significant amounts of open 2:1 clay minerals contradicts the ferrolysis process as being at the origin of the formation of these soils and further strengthens the geogenetic origin of the bleached horizon. Acknowledgements This study was partly made possible by the ‘Soil Fertility Project’ within the Institutional Co-operation Programme between the Flemish Interuniversity Council (VLIR) and the Jimma University in Ethiopia.

References Agbenin, J.O., Tiessen, H., 1995. Soil properties and their variations on two contiguous hill slopes in Northeast Brazil. Catena 24, 147–161. Barberis, E., Marsan, F.A., Arduino, E., 1991. Aggregation of soil particles by iron oxides in various size fractions of soil B horizons. J. Soil Sci. 42, 535–542. Boivin, P., Saejiew, A., Gruberger, O., Arunin, S., 2004. Formation of soils with contrasting textures by translocation of clays rather than ferrolysis in flooded rice fields in Northeast Thailand. Eur. J. Soil Sci. 55, 713–724. Brinkman, R., 1970. Ferrolysis, a hydromorphic soil forming process. Geoderma 3, 199–206. Bühmann, C., Grubb, P.L.C., 1991. A kaolin–smectite interstratification sequence from a red and black complex. Clay Miner. 26, 343–358. Caner, L., Hubert, F., Moni, C., Chenu, C., 2010. Impact of clay mineralogy on stabilisation of organic matter in the clay fraction of a Neo-Luvisol and a Cambisol. 19th World Congress of Soil Science: Soil Solutions for a Changing World, pp. 73–76 (Brisbane, Australia). Churchman, G.J., Whitton, J.S., Claridge, G.G.C., Theng, B.K.G., 1984. Intercalation method using formamide for differentiating halloysite from kaolinite. Clays Clay Miner. 32, 241–248. Cornelis, J.-T., Dumon, M., Tolossa, A.R., Delvaux, B., Deckers, J., Van Ranst, E., 2014. The effect of pedological conditions on the sources and sinks of silicon in the Vertic Planosols in south-western Ethiopia. CATENA 112, 131–138. Cuadros, J., Dudek, T., 2006. FTIR investigation of the evolution of the octahedral sheet of kaolinite–smectite with progressive kaolinization. Clays Clay Miner. 54, 1–11. Dahlgren, R.A., 1994. Quantification of allophane and immogolite. Quantitative Methods in Soil Mineralogy.Soil Science Society of America, Madison, USA 430–451. Driessen, P., Deckers, J., Spaargaren, O., Nachtergaele, F., 2001. Lecture Notes on the Major Soils of the World (Rome, Italy). Drits, V.A., Sakharov, B.A., 1976. X-ray structure analysis of mixed-layer minerals. Dokl. Akad, Nauk SSSR, Moscow. Drits, V.A., Tchoubar, C., 1990. X-Ray Diffraction by Disordered Lamellar Structures: Theory and Applications to Microdivided Silicates and Carbons. Springer-Verlag, Berlin, Germany. Drits, V., Środoń, J., Eberl, D.D., 1997. XRD measurement of mean crystallite thickness of illite and illite/smectite: reappraisal of the Kubler index and the Scherrer equation. Clays Clay Miner. 45, 461–475. Hubert, F., Caner, L., Meunier, A., Lanson, B., 2009. Advances in characterization of soil clay mineralogy using X-ray diffraction: from decomposition to profile fitting. Eur. J. Soil Sci. 60, 1093–1105. Hubert, F., Caner, L., Meunier, A., Ferrage, E., 2012. Unraveling complex b2 μm clay mineralogy from soils using X-ray diffraction profile modelling on particle-size sub-fractions: implications for soil pedogenesis and reactivity. Am. Miner. 97, 384–398. Lado, M., Ben-Hur, M., 2004. Soil mineralogy effects on seal formation, runoff and soil loss. Appl. Clay Sci. 24, 209–224. Laird, D.A., Barak, P., Nater, E.A., Dowdy, R.H., 1991. Chemistry of smectitic and illitic phases in interstratified soil smectite. Soil Sci. Soc. Am. J. 55, 1499–1504. Molina Ballesteros, E., Cantano Martin, M., 2002. Study of weathering processes developed on old piedmont surfaces in western Spain: new contributions to the interpretation of the “Rana” profiles. Geomorphology 42, 279–292. Plançon, A., 2002. New modelling of X-ray diffraction by disordered lamellar structures such as phyllosilicates. Am. Miner. 87, 1672–1677. Righi, D., Terribile, F., Petit, S., 1999. Pedogenic formation of kaolinite–smectite mixed layers in a soil toposequence developed from basaltic parent material in Sardinia (Italy). Clays Clay Miner. 47, 505–514. Ryan, P.C., Huertas, F.J., 2009. The temporal evolution of pedogenic Fe–smectite to Fe–kaolin via interstratified kaolin–smectite in a moist tropical soil chronosequence. Geoderma 151, 1–15. Sakharov, B.A., Lindgreen, H., Salyn, A., Drits, V.A., 1999. Determination of illite–smectite structures using multispecimen X-ray diffraction profile fitting. Clays Clay Miner. 47, 555–566. Tadesse, S., Milesi, J.P., Deschamps, Y., 2003. Geology and mineral potential of Ethiopia: a note on geology and mineral map of Ethiopia. J. Afr. Earth Sci. 36, 273–313. Van Ranst, E., Dumon, M., Tolossa, A.R., Cornelis, J.-T., Stoops, G., Vandenberghe, R.E., Deckers, J., 2011. Revisiting ferrolysis processes in the formation of Planosols for rationalizing the soils with stagnic properties in WRB. Geoderma 163, 265–274. Velde, B.B., Meunier, A., 2008. The Origin of Clay Minerals in Soils and Weathered Rocks. Springer-Verlag, Berlin, Germany. Vingiani, S., Righi, D., Petit, S., Terribile, F., 2004. Mixed-layer kaolinite–smectite minerals in a red–black soil sequence from basalt in Sardinia (Italy). Clays Clay Miner. 52, 473–483. Wilson, M.J., 1999. The origin and formation of clay minerals in soils: past, present and future perspectives. Clay Miner. 34, 7. Zeelmaekers, E., Honty, M., Derkowski, A., De Craen, M., Vandenberghe, N., Van Geet, M., 2010. A new and improved methodology for qualitative and quantitative mineralogical analysis of Boom clay. Clays in Natural & Engineered Barriers for Radioactive Waste Confinement, 4th International Meeting. Nantes, France, pp. 73–74.